USVI Public Records

Document text (beginning — full text at the TXT link above)

United States Environmental Protection Agency Office of Solid Waste and Emergency Response. Washington, DC 20460 9355.4-23 EPA/540/R-96/D1& PB96-063505 April 1996 Superfund •, ERA Soil Screening Guidance: User's Guide TUT 008 103 8 *65067* 65067 EPA/540/R-96/018 April 1996 Soil Screening Guidance: User's Guide Office of Emergency and Remedial Response U.S. Environmental Protection Agency Washington, DC 20460 TUT OO8 1O39 ACKNOWLEDGMENTS The development of this guidance was a team effort led by the staff of the Office of Emergency and Remedial Response. David Cooper served as Team Leader for the overall effort MadeneBezg coordinated the series of Outreach meetings with interested parties outside the Agency. SheiriClar^ JanineDinan and lirenHeniiiiig were u^princ^ authors. Paul White of EPA's Office of Research and Development provided tremendous support in me devdopinent of statistical approaches to site Exceptional technical assistance was provided by several contractors. Robert Truesdate of Research Triangle Institute (RT1) led their team effort in the development of the Technical BadcgronndDocninett under EPA Contnict68-Wl-0021. Craig Mann of Environmental Quality Management, Inc. (EC3 provided expert support mmotelhigiiuialatm EPA Contract 68-D3-0035. Dr. Smita Siddhanti of Booz-AUen & Hamilton, Inc. provided technical support for the final production of the User's Guide and Technical Background Document under EPA Contract 68-W1-0005. In addition, the authors would like to thank an EPA, State, public and peer reviewers whose careful review and thoughtful comments contributed to the quality of this document TUT OOS 1040 DISCLAIMER Notice: The Soil Screening Guidance is based on policies set out in the Preamble to the Final Rule of the National Oil and Hazardous Substances Pollution Contingency Plan (NCP), which was published on March 8,1990 (55 Federal Register 8666). . This guidance document sets forth recomtnmrtBd approaches based on EPA's best thinking to date with respect to soil screening. This document does not establish biiidingniles. Aheniatrve approaches for screening may be fouid to be more appropriate at specific sites (e.g., where site circumstances do not match the underlying •«V""J*"MH-. conditions and models of the guidance). The decision whether to use an alternative approach and a description of any such approach should be placed inthe Administrative Record for the site. Accordingly, if comments are received at individual sites questioning the use of the approaches recommended in mis guidance, the comments should be considered and an explanation provided for the selected approach. The Soil Screening Guidance: Technical BadcgioundDocumett(TBD)maybehe]pAilmRsponduigto such i The policies set out in both the Soil Screening Guidance: User's Guide and the supporting TBD are intended solely as guidance to the U.S. Environmental Protection Agency (EPA) personnel; they are not final EPA actions and do not constitute mlemaking These policies are not intended, nor can they be rehed upon, to create any rights enforceable by any party in litigation with the United States government. EPA officials may decide to follow the guidance provided in this document, or to act at variance with the guidance, based on an o^aiysis of specific site circumstances. EPA also reserves the right to change the guidance at any time without public notice. u TUT cos io4i TABLE OF CONTENTS 1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . i........................................ 1 1.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Hole of Sofl Screening Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Scope of Sofl Screening Guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.0 SOIL SCREENING PROCESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ; . . . . . . . . . . . . . . . . . . . . . . 6 2.1 Step 1: Developing a Conceptual Site Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.1 Collect Existing Site Data ............................................ 7 2.1.2 Organize and Analyze Existing Site Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.3 Constract a Preliminary Diagram of the CSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.4 Perform Site Reconnaissance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 Step 2: Comparing CSM to SSL Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2.1 • Identity Pathways Present at the Site Addressed by Guidance . . . . . . . . . . . ... ; . . . . . . . 9 2.2.2 Identify Additional Pathways Present at the Site Not Addressed by Guidance . . . . . . . . . . 10 2.2.3 Compare Available Data to Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3 Step 3: Defining Data Collection Needs for Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.3.1 Stratify the Site Based on Existing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.3.2 Develop Sampling and Analysis Plan for Surface Soil . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3.3 Develop Sampling and Analysis Plan for Subsurface Soils . . . . . . . . . . . . . . . . . . . . . . . 16 2.3.4 Develop Sampling and Analysis Plan to Determine Soil Characteristics . . . . . . . . . . . . . 19 2.3.5 Determine Analytical Methods and Establish QA/QC Protocols . . . . . . . . . . . . . . . . . . . 20 2.4 Step 4: Sampling and Analyzing She Soils & DQA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4.1 Delineate Area and Depth of Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.4.2 Perform DQA Using Sample Results . . . . . . . . . . . . . . . . . . . . . , . . . . : . . . . . . . . . 22 2.4.3 Revise the CSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.5 Step 5: Calculating She-specific SSLs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.5.1 SSL Equations-Surface Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ! . . . . 23 2.5.2 SSL Equations-Subsurface Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.5.3 Address Exposure to Multiple Chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.6 Step 6: Comparing Site Sofl Contaminant Concentrations to Calculated SSLs . . . . . . . . . . . 35 2.7 Step 7: Addressing Areas Identified for Further Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 ATTACHMENTS A. Conceptual Site Model Summary . . . . , . . ' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-l B. Soil Screening DQOs for Surface Soils and Subsurface Soils . . . . . . . . . . . . . . . . . . B-l C. Chemical Properties for SSL Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C-l D. Regulatory and Human Health Benchmarks Used for SSL Development . . . . . . . . . . . D-l 111 TUT COS 1O42 LIST OF EXHIBITS Exhibit 1 Conceptual Risk Management Spectramf or Contaminated Soil ..................... 2 Exhibit 2 Exposure Pathways Addressed by SSLs ................'..................... 4 Exhibits Key Attributes of the User's Guide ........................................ 4 Exhibit 4 . Soil Screening Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Exhibits Data Quality Objectives Process . . . . . . . ; . . . : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Exhibit 6 Defining Study Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Exhibit 7 Designing Sampling and Analysis Plan for £ujfac£ Soils . . . . . . . . . . . . . . . . . . . . . . . . . 13 Exhibits TVcigning Sampling and Analysis Plan for SiAsmfaee Soils . . . . . . . . . . . i . . . . . . . . . . 15 Exhibit 9 U.S. Department of Agriculture Soil Texture Classification . . . . . . . . . . . . . . . . . . . . . . . .18 Exhibit 10 Site-Specific Parameters for Calculating Subsurface SSLs .......................... 25 Exhibit 11 CyC Values by Source Area, City, and Climatic Zone . . . . . . . . . . . . ' . . . . . . . . . . . . . . . 27 Exhibit 12 Simplifying Assumptioiis for SSL Migration to Ground Water Pathway . . . . . . . . . . . . . . 29 Exhibit 13 SSL Chemical with Non-carcinogen Toxic Effects on Specific Target Organ/Systems . . . . . . 34 rv TUT 008 1043 LIST OF ACRONYMS ARAR ASTM CERCLA CLP CSM CV DAF DNAPL DQA DQO EA EPA HBL HEAST HELP HHEM HQ IRIS ISC2 MCL MCLG NAPL NOAEL NPL .A/SI PCB PEF PRO Q/C QA/QC QL RAGS RCRA RfC RID RI RI/FS RME ROD SAB SAP SPLP SSL TBD TCLP USDA VF VOC Applicable or Relevant and Appropriate Requ American Society for Testing and Materials Comprehensive Environmental Response, Compensation and Liability Act Qmtract Laboratory Program Conceptual Site Model Coefficient of Variation Dilution Attenuation Factor Dense Nonaqueous Phase Liquid Data Quality Assessment Data Quality Objective Exposure Area Environmental Protection Agency Health Based Limit Health Effects Assessment Summary Table Hydtological Evaluation of Landfill Performance Human Health Evaluation Manual • Hazard Quotient Integrated Risk Information System Industrial Source Complex Model Maximum Contaminant Level Maximum Contaminant Level Goal Nonaqueous Phase Liquid No-Observed-Adverae-Effect Level National Priorities List National Technical Information Service Office of Emergency and Remedial Response Preliminary Assessment/Site Inspection Polychlorinated Biphenyl Paniculate Emission Factor Preliminary Remediation Goal Site-Specific Dispersion Model Quality Assurance/Quality Control Quantitation Limit . Risk Assessment Guidance for Superfund Resource Conservation and Recovery Act Reference Concentration Reference Dose Remedial Investigation Remedial Investigation/Feasibility Study Reasonable Maximum Exposure Record of Decision Science Advisory Board Sampling atwf Analysis Plan Synthetic Precipitation Leaching Procedure Soil Screening Level Technical Background Document Toxitity Characteristic Leaching Procedure U.S. Department of Agriculture Volatilization Factor Volatile Organic Compound f'UT OOS 1O44 1.0 INTRODUCTION 1.1 Purpose The Soil Screening Guidance is a tool that the U.S. Environmental Protection Agency (EPA) developed to help standardize and accelerate the evaluation and cleanup of contaminated soils at sites 09 the National Priorities List (NPL) with future residential land use.* This guidance provides a methodology for environmental science/engineering professionals to calculate risk-based, site-specific, soil screening- levels (SSLs) for contaminants in soil that may be used to identify areas needing further investigation at NPL sites. SSLs are not national cleanup standards. SSLs alone do not trigger the need for response actions or define "unacceptable" levels of contaminants in soil. In this guidance, "screening" refers to the process of identifying and defining areas, contaminants, and conditions, at a particular site that do not require further Federal attention. Generally, at sites where contaminant concentrations fall below SSLs, no f ^ h e r action or study is warranted under the comprehensive Environmental Response, Compensation and Liability Act (CERCLA). (Some States have developed screening numbers that are more stringent than the generic SSLs presented here; therefore, further study may be warranted under State programs.) Generally, where contaminant concentrations equal or exceed SSLs, further study or investigation, but not necessarily cleanup, is warranted. SSLs are risk-based concentrations derived from equations combining exposure information assumptions with EPA toxicity data. This User's Guide focuses on the application of a simple site- specific approach by providing a step-by-step methodology to calculate site-specific SSLs and is part of a larger framework that includes both generic and more detailed approaches to calculating screening levels. The Technical Background Document (TBD) (EPA, 1996), provides more information about these other approaches. Generic SSLs for the most common contaminants found at NPL sites are included in the TBD. Generic SSLs are calculated from the same equations presented in this .x—«iidance, but are based on a number of default assumptions chosen to be protective of human health for most site conditions. Generic SSLs can be used in place of site-specific screening levels; however, in general, they are expected to be more conservative than site-specific levels. The site manager should weigh the cost of collecting the data necessary to develop site-specific SSLs with the potential for deriving a higher SSL that provides an appropriate level of protection. The framework presented in the TBD also includes more detailed modeling approaches for developing screening levels that take into account more complex she conditions than the simple she-specific methodology emphasized in this guidance. More detailed approaches may be appropriate when site conditions (e.g., a thick vadose zone) are different from those assumed in the simple site-specific methodology presented here. The technical details supporting the methodology used in this guidance are provided in the TBD. SSLs developed in accordance with this guidance are based on future residential land use assumptions and related exposure scenarios. Using this guidance for sites where residential land use assumptions do not apply could result in overly conservative screening levels; however, EPA recognizes that some parties responsible for sites with non-residential land use might still find benefit in using the SSLs as a tool to conduct a conservative initial screening. SSLs developed in accordance with this guidance could also be used for Resource Conservation and. Recovery Act (RCRA) corrective action sites as "action levels," since the RCRA corrective action program currently views the role of action levels as generally fulfilling the same purpose as soil screening levels.* In addition, States may use this guidance in their voluntary cleanup programs, to the extent they deem appropriate. When applying SSLs to RCRA corrective action sites or for sites under State voluntary cleanup programs, users of this guidance should recognize, as stated above, that SSLs are based on residential land use assumptions. Where these assumptions do not apply, other approaches 1 Note that the Superfuad program define* "toil" at having • particle size under 2mm, while the RCRA program allows for panicles wider 9mm m size. 2 Further information on the role of action toveb in the RCRA corrective action program is available in an Advance Notice of Proposed ' Rvlemakmg (signed April 1996). TUT OOS 1O45 for determining the need for further study might be ** appropriate'. 1.2 Role of Soil Screening Levels In identifying and managing risks at sites, EPA considers a spectrum of contaminant concentrations. The level of concern associated with those concentrations depends on the likelihood of exposure to soil contamination at levels of potential concern to human health or to ecological receptors. Exhibit 1 illustrates the spectrum of soil contamination encountered at Superfund sites and the conceptual range of risk management responses. At one end are levels of contamination mat clearly warrant a response action; at the other end are levels that are below regulatory concern. Screening levels identify the lower bound of the spectrum — levels below which EPA believes mere is no concern under CERCLA, provided conditions associated with the SSLs are met. Appropriate cleanup goals for a particular site may fall anywhere within this range depending on site-specific conditions. No further study Sfta-spadfic Raaponta wajranled under dMnup action daarly CERdA QoaJnaval • waivartad T T *2aro* Scraaning Raapon concantralicf) taval ' lava) Exhibit 1. Conceptual Risk Spectrum for Contamim • ——— l^ ee VeiyNQh Management ited Soil EPA anticipates the use of SSLs as a tool to facilitate prompt identification of contaminants and exposure areas of concern during both remedial actions and some removal actions under CERCLA. However, the application of mis or any screening methodology is not mandatory at sites being addressed under CERCLA or RCRA. The framework leaves discretion to the site manager and technical experts (e.g., risk assessors, hydrogeologists) to determine whether a screening approach is appropriate for the she and, if screening is to be used, the proper method of implementation. If comments are received at individual sites questioning the use of the approaches recommended in this guidance, the comments should be considered and an explanation provided as part of the site's Record of Decision (ROD). The decision to use a screening approach should be made early in the process of investigation at the site. EPA developed the Soil Screening Guidance to be consistent with and to enhance the current Superfund investigation process and anticipates its primary use during the early stages of a remedial investigation (RI) at NPL sites. It does not replace the Remedial Investigation/Feasibility Study (RI/FS) or risk assessment, but use of screening levels can focus the RI and risk assessment on aspects of the she that are more likely to be a concern under CERCLA. By screening out areas of sites, potential chemicals of concern, or exposure pathways from further investigation, she managers and technical experts can limit the scope of the remedial investigation or risk assessment. SSLs can save resources by helping to determine which areas do not require additional Federal attention early in the process. Furthermore, data gathered during the soil screening process can be used in later Superfund phases, such as the baseline risk assessment, feasibility study, treatability study, and remedial design. This guidance may also be appropriate for use by the removal program when demarcation of soils above residential risk-based numbers coincides with the purpose and scope of the removal action. The process presented in this guidance to develop and apply simple, she-specific soil screening levels is likely to be most useful where it is difficult to determine whether areas of soil are contaminated to an extent that warrants further investigation or response (e.g., whether areas of soil at an NPL she require further investigation under CERCLA through an RI/FS). As noted above, the screening levels have been developed assuming residential land use. Although some of the models and methods presented in this guidance could be modified to address exposures under other land uses, EPA has not yet standardised assumptions for those other uses. Applying site-specific screening levels involves developing a conceptual site model (CSM), collecting a tew easily obtained site-specific soil parameters (such as the dry bulk density and percent TUT COS IO46 moisture), and sampling to measure contaminant levels in surface and subsur^ce soils. Often, much of >4he information needed to develop the CSM can be srived from previous site investigations [e.g., the jhreliminary Assessment/Site Inspection (PA/SI)] and, if properly planned, SSL sampling can be accomplished in one mobilization. An important part of this. guidance is a recommended sampling approach that balances the need for more data to reduce uncertainty with the need to limit data collection costs. Where data are limited such mat use of the "maximum test" (Max test) presented here is not appropriate, the guidance provides direction on the use of other conservative estimates of contaminant concentrations for comparison with the SSLs. This guidance provides the information needed to calculate SSLs for 110 chemicals. Sufficient information may not be available to develop soil screening levels for additional chemicals. These chemicals should not be screened out, but should be addressed in the baseline hsk assessment for the site. The Risk Assessment Guidance for Sitperfund (RAGS), Volume 1: Human Health Evaluation Manual (HHEM), Part A, Interim Final. (U.S. EPA, 1989a) provides guidance on conducting baseline /•""""Sk assessments for NPL sites. In addition, the \ feline risk assessment should address the chemicals, exposure pathways, and areas at the site that are not screened out. Although SSLs are "risk-based," they do. not eliminate the need to conduct a site-specific risk assessment. SSLs are concentrations of contaminants in soil that are designed to be protective of exposures in a residential setting. A site-specific risk assessment is an evaluation of the risk posed by exposure to site contaminants in various media. To calculate SSLs, the exposure equations and pathway models are run in reverse to backcalculate an "acceptable level" of a contaminant in soil. For the ingestion, dermal, and inhalation pathways, toxicity criteria are used to define an acceptable level of contamination in soil, based on a one-in-a-million (10-6) individual excess cancer risk for carcinogens and a hazard quotient (HQ) of 1 for non-carcinogens. SSLs are backcalculated for migration to ground water pathways using ground water concentration limits [nonzero maximum contaminant level goals (MCLGs), maximum contaminant levels (MCLs), or health-based limits (HBLs) (1(H cancer risk or a HQ '.of 1) where MCLs are not available]. SSLs can be used as Preliminary Remediation Goals (PRGs) provided appropriate conditions are met (i.e., conditions found at a specific she are similar to conditions assumed in developing the SSLs). The concept of calculating risk-based contaminant levels in soils for use as PRGs (or "draft" cleanup levels). was introduced in the RAGS HHEM. Part B, Development of Risk-Based Preliminary Remediation Goals.' (U.S. EPA, 199Ic). The models, equations, and assumptions presented in the Soil Screening Guidance to address inhalation exposures supersede those described in RAGS HHEM, Part B, for residential soils. In addition, this guidance presents .methodologies to address the leaching of contaminants through soil to an underlying potable aquifer. This pathway should be addressed in the development of PRGs. PRGs may then be used as the basis for developing final cleanup levels based on the nine-criteria analysis described in the National Contingency Plan [Section 300.430 (3)(2)(I)(A)]. The directive entitled Role of the Baseline Risk Assessment in Super/and Remedy Selection Decisions (U.S. EPA, 199Id) discusses the modification of PRGs to generate cleanup levels. The SSLs should only be used as. cleanup levels when a site-specific nine- criteria evaluation of the SSLs as PRGs- for soils indicates that a selected remedy achieving the SSLs is protective, complies with Applicable or Relevant and Appropriate Requirements (ARARs), and appropriately balances the other criteria, including cost. 1.3 Scope of Soil Screening Guidance In a residential setting, potential pathways Of exposure to contaminants in soil are as follows (see Exhibit 2): • Direct ingestion • Inhalation of volatiles and fugitive dusts TUT OO8 1O47 Ingestion of contaminated ground water caused by migration of chemicals through soil to an underlying potable aquifer Dermal absorption Ingestion of homegrown produce that has been contaminated via plant uptake Migration of volatiles into basements. Direct Ingestion of Ground Water and Soil Blowing Dust Volatilization Also Addressed: • Plant Uptake • Dermal Absorption Exhibit 2. Exposure Pathways Addressed by SSLs. : The Soil Screening Guidance addresses each of these pathways to the greatest extent practical. The first three pathways — direct ingestion, inhalation of volatiles and fugitive dusts, and ingestion of potable ground water — are the most common routes of human exposure to contaminants in the residential setting. These pathways have generally accepted methods, models, and assumptions that lend themselves to a standardized approach. The additional pathways of exposure' to soil contaminants, dermal absorption, plant uptake, and migration of volatiles into basements, may also contribute to the risk to human health from exposure to specific contaminants in a residential setting. This guidance addresses these pathways to a limited extent based on available empirical data. (See Step 5 and the TBD for further discussion). The Soil Screening Guidance addresses the human exposure pathways listed previously and will be appropriate for most residential settings. The presence of additional pathways or unusual site conditions does not preclude. the use of SSLs in areas of the site that are currently residential or likely to be residential in the future. However, the risks associated with additional pathways or conditions (e.g., fish consumption, raising of livestock, a heavy truck traffic on unpaved roads) should be considered in the RI/FS to determine whether SSLs are adequately protective. An ecological assessment should also be performed as part of the RI/FS to evaluate potential risks to ecological receptors. The Soil Screening Guidance should. not be used for areas with radioactive contaminants. Exhibit 3 provides key attributes of the Soil Screening Guidance: User's Guide. Exhibit 3: Key Attributes of the User's Guide • Standardized equations are presented to address human exposure pathways in a residential setting consistent with Superfund's concept of'Reasonable Maximum Exposure* (RME). • Source size (area and depth) can be considered on a she-specific basis using mass-limit models. • Parameters are identified for which site- specific information is needed to develop SSLs. • Default values are provided to calculate generic SSLs when site-specific information is not available. • SSLs are based on a 10* risk for carcinogens or a hazard quotient of 1 for noncarcinogens. SSLs for migration to ground water are based on (in order of preference): nonzero maximum contaminant level goals (MCLGs), maximum contaminant levels (MCLs), or the aforementioned risk- based targets. TUT oos 1048 2.0 SOIL SCREENING PROCESS The soil screening process (Exhibit 4) is a step-by- step approach mat involves: • Developing a conceptual site model (CSM) • Comparing the CSM to the SSL scenario • Defining data collection needs • Sampling and analyzing soils at site • Calculating site-specific SSLs • Comparing site soil contaminant concentrations to calculated SSLs • Determining which areas of the site require further study. It is important to follow this process to implement the Soil Screening Guidance properly. The remainder of this guidance discusses each activity in detail. 2.1 Step 1! Developing a Conceptual Site Model The CSM is a three-dimensional "picture" of site conditions that illustrates contaminant distributions, release mechanisms, exposure pathways and migration routes, and potential receptors. The CSM documents current site conditions and is supported by maps, cross sections, and site diagrams that illustrate human and environmental exposure through contaminant release and migration to potential receptors. Developing an accurate CSM is critical to proper implementation of the Soil Screening Guidance. As a key component of the RI/FS and EPA's Data Quality Objectives (DQO) process, the CSM should be updated and revised as investigations produce new information about a site. Data Quality Objectives for Superfund: Interim Final Guidance (U.S. EPA, 1993a) and Guidance for Conducting Remedial Investigations and Feasibility Studies under CERCLA (U.S. EPA, 1989c) provide a general discussion about the development and use of the CSM during RIs. Developing the CSM involves several steps, discus^d in the following subsections. 2.1.1 Collect Existing Site Data. The initial design of the CSM is based on existing site data compiled during previous studies. These data may include site sampling data, historical records, aerial photographs, maps, and State soil surveys, as well as information on local and regional conditions relevant to contaminant migration and potential receptors. Data sources include Superfund site assessment documents (i.e., the PA/SI), documentation of removal actions, and records of other site characterizations or actions. Published information on local and regional climate, soils, hydrogeology, and ecology may be useful. In addition, information on Hie population and land use at and surrounding the site will be important to identify potential exposure pathways and receptors. The RI/FS guidance (U.S. EPA, 1989c) discusses collection of existing data during RI scoping, including an extensive list of potential data sources. 2.1.2 Oroar^jffi flpd Analyze Existing Site Pjift. One of the most important aspects of the CSM development process is to identify and characterize all potential exposure pathways and receptors at the site by-considering site conditions, relevant exposure scenarios, and the properties of contaminants present in site soils. Attachment A, the Conceptual Site Model Summary, provides four forms for organizing site data for soil screening purposes. The CSM summary organizes site data according to general site information, soil contaminant source characteristics, exposure pathways and receptors. Note: If a CSM has already been developed for the site in question, use the summary forms in Attachment A to ensure that it is adequate. 2.1.3 Construct a Preliminary Diagram of the CSM. Once the existing site data have been organized and a basic understanding of the site has been attained, draw a preliminary "sketch" of the site conditions, highlighting source areas, potential exposure pathways, and receptors. TUT 1049 Exhibit 4 Soil SrMnln Step One: Develop Conceptual Site Model • Collect existing site data (historical records, aerial photographs, maps, PA/SI data, available background information. State soil surveys, etc.) • Organize and analyze existing site data - Identify known sources of contamination - Identify affected media - Identify potential migration routes, exposure pathways, and receptors • Construct a preliminary diagram of the CSM • Perform site reconnaissance . - Confirm and/or modify C S M ' 1 - Identify remaining data gaps . | Step Two: Compare Soil Component of CSM to Soil Screening Scenario • Confirm oat future residential land use is a reasonable assumption for the site • Identify pathways present at the site mat are addressed by me guidance • Identify additional pathways present at the site not addressed by the guidance • Compare pathway-specific generic SSLs with available concentration data • .Estimate whether background levels exceed generic SSLs Step Three: Define Data Collection Needs for Soils to Determine Which Site Areas Exceed SSLs • Develop hypothesis about distribution of soil contamination (i.e., which areas of the site have soil contamination that exceed appropriate SSLs?) • Develop sampling and analysis plan for determining soil contaminant concentrations - Sampling strategy for surf ace soils (incloses defining study boundaries', developing a decision rule, specifying limits on decision errors, and optimizing the design) - Sampling strategy for subsurface soils (includes defining study boundaries, developing a decision rule, specifying limits on. decision errors, and optimizing the design) - Sampling to measure soil characteristics (bulk density, moisture content, organic carbon content, porosity, pH) • Determine appropriate field methods and establish QA/QC protocols Step Four: Sample and Analyze Soils at Site • Identify contaminants . I • Delineate area and depth of sources ' • Determine soil characteristics • Revise CSM, as appropriate Step Five: Derive Site-specific SSLs, If needed • Identify SSL equations for relevant pathways • Identify chemical of concern for dermal exposure and plant uptake • Obtain site-specific input parameters from CSM summary • Replace variables in SSL equations with site-specific data gathered in Step 4 Calculate SSLs - Account for exposure to multiple contaminants Step Six: Compare Site Soil Contaminant Concentrations to Calculated SSLs • For surface soils, screen out exposure areas where all composite samples do not exceed SSLs by a factor of 2 • For subsurface soils, screen out source areas where the highest average soil core concentration does not exceed the SSLs • Evaluate whether background levels exceed SSLs Step Seven: Decide How to Address Areas Identified for Farther Stady • Consider likelihood that additional areas can be screened out with more data • Integrate soil data with other media in the baseline risk assessment to estimate cumulative risk at the site • Determine the need for action • Use SSLs as PRGs TUT OOS 1OSO Ultimately, when site investigations are complete, thr sketch will be refined into a three-dimensional diagram that summarizes the data. Also, a brief ——summary of the contamination problem should Accompany the CSM. Attachment A provides an example of a complete CSM summary. . 2.1.4 Perform Site Reconnaissance. At this point, a site visit would be useful because conditions at the site may have changed since the PA/SI was performed (e.g., removal actions may have been taken). During site reconnaissance, update site sketches/topographic maps with the locations of buildings, source areas, wells, and sensitive environments. Anecdotal information from nearby residents or she workers may reveal undocumented disposal practices and thus previously unknown areas of contamination that may affect the current CSM interpretation. Based on the new information gained from site reconnaissance, update the CSM as appropriate. Identify any remaining data gaps in the CSM so that these data needs can be incorporated into the .Sampling and Analysis Plan (SAP). 2.2 Sifip_2: Comparing CSM to SSL Scenario .e Soil Screening Guidance is likely to be appropriate for sites where residential land use is reasonably anticipated. However, the CSM may include other sources and exposure pathways that are not covered by this guidance. Compare the CSM with the assumptions and limitations inherent in the SSLs to determine whether additional or more detailed assessments are needed for any exposure pathways or chemicals. Early identification of areas or conditions where SSLs are not applicable is important so that other characterization and response efforts can be considered when planning the sampling strategy. 2.2.1 Identify Pathways Present at the Site Addressed by Guidance. The following are potential pathways of exposure to soil contaminants in a residential setting and are addressed by this guidance document: • Direct ingestion • Inhalation of volatiles and fugitive dusts • Ingestion of contaminated ground water caused by migration of chemicals through soil to an under- lying potable aquifer • Dermal absorption • Ingestion of homegrown produce that has been contaminated via plant uptake • Migration of volatiles into basements. This guidance quantitatively addresses the ingestion, inhalation, and migration to ground water pathways and also addresses, more qualitatively, the potential for dermal absorption and plant uptake based on limited empirical data. Whether some or all of the pathways are relevant at the site depends upon the contaminants and conditions at the site. For surface soils under the residential land use assumption, routinely consider the direct ingestion route in the soil screening decision. Inhalation of fugitive dusts and dermal absorption can be of concern for certain chemicals and site conditions. For subsurface soils, risks from inhalation of volatile contaminants and migration of soil contaminants to an underlying aquifer are potential concerns for this scenario. The inhalation pathway may be eliminated from further analysis if the presence of volatile contaminants are not suspected in the subsurface soils. Likewise, consideration of the ground water pathway may be eliminated if ground water beneath or adjacent to the site is not a potential source of drinking water. Coordinate this decision on a she-specific basis with State or local authorities responsible for ground water use and classification. The rationale for excluding this exposure pathway should be consistent with EPA ground water policy (U.S. EPA, 1988a, 1990a, 1992a, 1992c, and 1993b). The potential for plant uptake of contaminants should be addressed for bom surface and subsurface soils. . In addition to the more common pathways of exposure in a residential setting, concerns have been raised regarding the potential for migration of TUT COS 1051 volatile organic compounds (VOCs) from subsurface soils into basements. Tb» Johnson and Ettinger model (1991) was developed to address this pathway, and an analysis of the potential use of this model for soil screening is provided in the TBD (U.S. EPA, 19%). The analysis suggests that the use of the model is limited due to its sensitivity to a number of parameters such as distance from the source to the building, building ventilation rate and the number and size of cracks in the basement wall. Such data are difficult to obtain for a current use scenario, and extremely uncertain for any future use scenario. Thus, instead of relying exclusively on the model, data from a comprehensive soil-gas survey .are recommended to address the potential for migration of VOCs in the subsurface. Soil-gas data and site-specific information on soil permeability can be used to replace default parameters in the Johnson and Ettinger model to obtain a more reliable estimate for the impact of this pathway on site risk. Identify Additional Pathways Present at the Site Not Addressed by Guidance. The presence of additional pathways does not preclude the use of SSLs in site areas that are currently residential or likely to be residential in the future. However, the risks associated with these additional pathways should also be considered in the RI/FS to determine whether SSLs are adequately pro- tective. Where the following conditions exist, a more detailed site-specific .study should be performed: • The site is adjacent to bodies of surface water where the potential for contamination of surface water by overland flow or release of contaminated ground water into surface water through seeps should be considered. • There are potential terrestrial or aquatic ecological concerns. • There are other likely human exposure p a t h w a y s that were not considered in development of the SSLs (e.g., local fish consumption, raising of beef, dairy, or other livestock). • There are unusual site conditions such as the presence of nonaqueous phase liquids (NAPLs), large areas of contamination, unusually high fugitive dust levels due to soil being tilled for agricultural use, or heavy traffic on unpavr roads. .| • There are certain subsurface site conditions' such as leant, fractured rock aquifers, or contamination extending below the water table, that result in the screening models not being sufficiently conservative. 2.23 Compare Available Data to Background. EPA may be concerned with two types of background at sites: naturally occurring and anthropogenic. Natural background is usually limited to metals; whereas, anthropogenic (i.e., man-made) background can include both organic and inorganic contaminants. A comparison of available data (e.g., State soil surveys) on local background concentrations with generic SSLs may indicate whether background concentrations at the site are elevated. Although background concentrations exceeding generic SSLs do not necessarily indicate mat a health threat exists, further investigation may be necessary. Generally, EPA does not cleanup below natural background levels; however, where anthropogenic background levels exceed SSLs and EPA ha determined that a response action is necessary am. feasible, EPA's goal will be to develop a comprehensive response to address area soils. This will often require coordination with different authorities that have jurisdiction over other sources of contamination in tile area (such as a regional air board or RCRA program). This will help avoid response actions that create "clean islands" amid widespread contamination. To determine the need for a response action, the site investigation should include gathering site- specific background data for any potential chemicals of concern and their speciation, because contaminant solubility in water and bioavailability (absorption into an organism) are important considerations for the risk assessment. Speciation of compounds such as metals and congener-specific analysis of similar organic chemicals [e.g., dioxins, polychlorinated biphenyls (PCBs)] can sometimes provide improved estimates of exposure and subsequent toxicity of chemically related compounds. While water solubility is not often a TUT O08 1-052 good predictor of uptake of a toxicant into the blood of an exposed receptor for physiological reasons, relative bioavailability and toxicity can /^•sometimes be estimated through analytical peciation of related compounds. For example, various forms of metals are more or less toxic and can behave as quite disparate compounds in terms of exposure and risk. Inorganic forms of metals are not likely to cross biological membranes as easily or may not bioaccumulate as readily as organometallics. Different valences of metals can produce dramatically different toxicities (e.g., chromium). Different matrices can render metals • more or less bioaccessible (e.g., lead in auto emissions from leaded gas vs. lead in mine wastes). Similarly, the position and number of halogens on complex organic molecules can affect uptake and toxicity (e.g., dioxins). When applying these concepts to a screening analysis, the risk assessor should establish a credible rationale based on relevant literature and site data mat supports actual differences in uptake and/or toxicity, since one cannot predict bioavailability from simple solubility studies. More likely, such an in-depth evaluation .of chemical speciation and bioavailability would be conducted as part of a more detailed site-specific risk assessment. 2.3 Step 3: Defining Data . Collection Needs for Soils Once the CSM has been developed and the site manager has determined that the Soil Screening Guidance is appropriate to use at a site, an SAP should be developed. Attachment A, the Conceptual Site Model Summary, lists the data needed to apply the Soil Screening Guidance. The summary will help identify data gaps in the CSM that require collection of site-specific data. The soil SAP is likely to contain different sampling strategies that address: • Surface soil • Subsurface soil • Soil characteristics To develop sampling strategies that will properly assess she contamination, EPA recommends that site managers consult with the technical experts in their Region, including risk assessors, toxicologists, chemists and hydrogeologists. These experts can assist the site manager to use the DQO process to satisfy Superfund program objectives. The DQO process is a systematic planning process developed by EPA to ensure that sufficient data are collected" to support EPA decision making. A full discussion of the DQO process is provided in Data Quality Objectives for Superfund: Interim Final Guidance (U.S. EPA, 1993a) and the Guidance for the Data Quality Objectives Process (U.S. EPA, 1994a). Most key elements of the DQO process have already been incorporated as part of this Soil Screening Guidance (see Exhibits 5 through 8 and Attachment B). The remaining elements involve identifying the site-specific information needed to calculate SSLs. For example, the dry bulk density and the fraction of organic carbon content will need to be collected for the subsurface soil investigation. The following sections present an overview of the sampling strategies needed to use the Soil Screening Guidance. For a more detailed discussion, see the supporting TBD. 2.3.1 Stratify the Si At this point in the soil screening process, existing data can be used to stratify the site into three types of areas requiring different levels of investigation: • Areas unlikely to be contaminated • Areas known to be highly contaminated • Areas that may be contaminated and cannot be ruled out. Areas that are unlikely to be contaminated generally will not require further investigation if historical site use information or other site data, which are reasonably complete and accurate, confirm this assumption. These may be areas of the site that were completely undisturbed by hazardous-waste- generating activities. TUT OO8 1O53 Exhibit 5: Data Quality Objectives Process 1. State the Problem . *• • Summarize the contamination problem that will require new environmental data, and identify the resources available 19 resolve the problem. * 2. Identify the Decision Identify the decision that requires new environmental data to. address the contamination problem. 3. Identify Inputs to the Decision Identify the information needed to support the decision, and specify which inputs require new environmental measurements. 4. Define the Study Boundaries Specify the spatial and temporal aspects of the environmental media that the data must represent to support the decision 5. Develop a Decision Rule Develop a logical *tf... men..." statement that defines the conditions that would cause the decision maker to choose among alternative actions. 6. Specify Limits on Decision Errors Specify the decision maker's acceptable limits on decision errors, which are used to estabfish performance goals for limiting uncertainty in the data. 7. Optimize the Design for Obtaining Data Identify the most resource-effective sampling and analysis design for generating data that are expected to satisfy the DQOs. Subsuri»c*Som Expanded iaExhft* TUT 008 1054 Exhibit 6: Defining the Study Boundaries 1. Define Geographic Area of the Investigation Study Boundaries 2. Define Population of Interest Surface Soil (usually top 2 centimetare) Subsurface Soil Water Table led Zone) 3. Stratify the Site Area Unlikely to be Contaminated Area of Suspected Contamination Area of Known Contamination (possible source) 4. Define Scale of Decision Making for Surface or Subsurface Soils exposure [Subsurface Soils] Contaminant Source Back to Exhibit 5, Stop *t ^Develop a Decision Ruto* j 11 TUT OOS 1O55 A crude estimate of the degree of soil contamination c- be made for other areas of the site by comparing site concentrations to the generic SSLs in Appendix A of the TBD. Generic SSLs have been calculated for 110 chemicals using default values in the SSL equations, resulting in conservative values that will be protective for the majority of site conditions. The pathway-specific generic SSLs can be compared with available concentration data from previous site investigations or removal actions to help divide die site into areas with similar levels of soil contamination and develop appropriate sampling strategies. The surface soil sampling strategy discussed in this document is most appropriate for those areas mat may be contaminated and can not be designated as uncontaminated. Areas which are known to be contaminated (based on existing data) will be investigated and characterized in the RI/FS. 2.3.2 Develop Sampling and Analysis Plan for Surface Soil. The surface soil sampling strategy is designed to collect the data needed to evaluate exposures via direct ingestion, dermal absorption, and inhalation of fugitive dusts. As explained in the Supplemental Guidance to RAGS: Calculating the Concentration Term (U.S. EPA, 1992d), an individual is assumed to move randomly across an exposure area (EA) over time, spending equivalent amounts of time in each location. Thus, the concentration contacted over time is best represented by the spatially averaged concentration over the EA. Ideally, the surface soil sampling strategy would determine the true population mean of contaminant concentrations in an EA. Because determination of the "true" mean would require extensive sampling at high costs, the maximum contaminant concentration from composite samples is used as a conservative estimate of the mean. This Max test strategy compares the results of composite samples with the SSLs. Another, more complex strategy called the Chen test is presented in Pan 4 of the TBD. The User's Guide uses the Max test rather than the Chen test because the Max test is based on a statistical null hypothesis that is more appropriate for NPL sites (i.e., the EA requires further investigation). Although the Chen test is not well suited for screening decisions at NPL sites, h may be useful in a non-NPL, voluntary cleanup context. The depth over which surface soils are sampled should reflect the type of exposures expected at the site. The Urban Soil Lead Abatement Demonstration Project (U.S. EPA 1993d) defined the top 2 centimeters as the depth of soil where direct contact predominantly occurs. The decision to sample soils below 2 centimeters depends on the likelihood of deeper soils being disturbed and brought to tiie surface (e.g., from gardening, landscaping or construction activities). Note that the size, shape, and orientation of sampling volume (i.e., "support") for heterogenous media have a significant effect on reported measurement values. For instance, particle size has ? "irying affect on the transport and fate of contaminants in the environment and on the potential receptors. Comparison of data from methods mat are based on different supports can be difficult. Defining the sampling support is important in the early stages of site characterization. This may be accomplished through the DQO process with existing knowledge of. the site, contamination, and identification of the exposure pathways that need to be characterized. Refer to Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies (U.S. EPA, 1992e) for more information about soil sampling support. The SAP developed for surface soils should specify sampling and analytical procedures as well as the development of QA/QC procedures. To identify the appropriate analytical procedures, the screening levels must be known. If data are not available to calculate she-specific SSLs (Section 2.5.1), men the generic SSLs in Appendix A of the TBD should be used. The following strategy can be used for surface soils to estimate the mean concentration of semivolatiles, inorganics, and pesticides in an exposure area. Volatiles are not included in the estimations because they are not expected to remain at the surface for an extended period of time. 12 TUT OOQ 1036 Exhibit 7: Designing a Sampling and Analysis Plan for Surface Soils ^ubdivide Site <nto EAs 2. Divide EA Into a Grid 3. Organize Surface Sampling Program for EA For surface soils, the individual unit for decision making is an 'EA,'or exposure area, h measures 0.5 acre in area or 01 o. » V 01 6. 01 OK O* This step defines the number of specimens (N) that will make up one composite sample. Placement of sample locations on the-grid was developed using a default sample size of 6 (which is based on acceptable error rates for a CV of 2.5) and a stratified random sampling pattern. If the EA CV is suspected to be greater than 2.5, use the table below to select an adequate sample size or refer to the TBD for other sample design options. Probability of Decision Error at 0.5 SSL and 2 SSL Using Max Test SampteSiz«b 6 7 8 9 CV=2.5« *>£' Eao d CVs3.0 EOS. E2.0 CV=3.5 GU iiO CV=4.0 EOJ5 E*o C * 4qKdm«fls per composite* &21 025 025 028 0.08 0.05 0.04 0.03 028 0.31 036 0.36 0.11 0.08 0.05 0.04 031 OJ6 0.42 0.44 0.11. 0.09 0.07 O07 0.35 0.41 041 0.48 0.16 0.15 0.09 0.08 The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA, ^including measurement error. e Sample size (N) = number of compos tie samples • d Eo.5 = Probability of requiring further investigatton when the EA mean is 0.5 SSL (E2.0 e Probability of not requiring further investigation when the EA mean is 2.0 SSL C & number of specimens per composite sample, when each composite consists of points from a stratified random or systemic grid sample from across the entire EA. NOTE: All decision error rates are based on 1,000 simulations that assume that each composite is representative of the entire EA, half the EA has concentrations betow the limh of detection, and half the EA Jv*concentrations that follow a gamma distribution (a conservative distributional assumption). 13 TUT 008 1057 • Divide areas to be sampled in the screening process into 0.5-acre exposure areas, the size of a suburban residential lot. If the site is currently residential, the exposure area should be the actual residential lot size. The exposure areas should not be laid out in such a way that they unnecessarily combine areas of high and low levels of contamination. The orientation and exact location of the EA, relative to the distribution of the contaminant in the soil, can lead to instances where sampling the EA may have contaminant concentration results above the mean, and in other instances, results below the mean. Try to avoid straddling contaminant "distribution units" within the 0.5-acre EA. • Composite surface soil samples" Because the objective of surface soil screening is to estimate the mean contaminant concentration, the physical "averaging" that occurs during compositing is consistent with the intended use of the data. Compositing allows sampling of a larger number of locations while controlling analytical costs, since several individual samples are physically mixed (homogenized) and one or more subsamples are drawn from the mixture and submitted for analysis. • Strive to achieve a false negative error rate of 5 percent (i.e., in only 5 percent of the cases, soil . contamination is assumed to be below the screening level when it is really above the screening level). EPA also strives to achieve a 20 percent false positive error rate (i.e., in only 20 percent of the cases, Soil contamination is assumed to be above the screening level when it is really below the screening level). These error rate goals influence the number of samples to be collected in each exposure area. For this guidance, EPA has defined the "gray region" as one-half to 2 times the SSL. Refer to Section 2.6 for further discussion. • The default sample size chosen for this guidance (see Exhibit 7) provides adequate coverage for a coefficient of variation (CV) based upon 250 percent variability in contaminant values (CV=2.5). (If a CV larger than 2.5 is expected, use an appropriate sample size from the table in Exhibit 7 of the User's Guide, or tables in the TBD.) •. Take six composite samples, for each exposure area, with each composite sample made up of four individual samples. Exhibit 7 shows other sample sizes' needed to achieve the decision error rates for other CVs. Collect the composites randomly across the EA and through the top 2 centimeters of soil, which are of greatest concern for incidental ragestion of soil, dermal 'contact, and inhalation of fugitive dust. • Analyze the six samples per exposure area to determine the contaminants present and their concentrations . For further information on compositing across or within EA sectors, developing a random sampling strategy, and determining sample sizes that control decision error rates, refer to the TBD. Note that the Max test requires a Data Quality Assessment (DQA) test following sampling and analysis (Section 2.4.2) to ensure that the DQOs (i.e., decision error rate goals) are achieved. If DQOs are not met, additional sampling may be required. Develop Sampling and Analysis Plan for Subsurface Soils. The subsurface and surface soil sampling strategies differ because the exposure mechanisms differ. Exposure to surface contaminants occurs randomly as individuals move around a residential lot. The surface soil sampling strategy reflects this type of random exposure. In general, exposure to subsurface contamination occurs when chemicals migrate up to the surface or down to an underlying aquifer. Thus, subsurface sampling focuses on collecting the data required for modeling the volatilization and migration to ground water pathways. Measurements of soil characteristics and estimates of the area and depth of contamination .and the average contaminant concentration in each source area are needed to supply the data necessary to calculate the inhalation and migration to ground water SSLs. 14 TUT 008 lose Exhibit 8: Designing a Sampling and Analysis Plan for Subsurface soils Delineate Source Area .Contaminant Source Soil Borings Choose Subsurface Soil Sampling Locations Design Subsurface Sampling and Analysis Plan Lab/Field Analysis for soil parameters" Soil Boring . (depth below ground surface in feet) For screening purposes, EPA recommends drilling 2 to 3 borings per source area in areas of highest suspected concentrations. Soil sampling should not extend past water table or saturated zone. Lab Analysis for soil contaminants 3k depicts a continuous boring with 2 foot segments. For information on other methods such as interval sampling and depth .»dd analysis, please refer to 2.3.3 of the User's Guide or 4.2 of the TBD. Soil Texture, Dry Bulk Density, Soil Organic Carbon, pH. Retain samples for possble discrete contaminant sampling. 15 "UT OO8 1O59 Source areas are the decision units for subsurface soils.. A source 'area is defined by the horizontal extent, and vertical extent or depth of contamination. For this purpose, "contamination" is defined by either the Superfund's Contract Laboratory Program (CLP) practical quantitation limits (QLs) for each contaminant, or the SSL. Sites with multiple sources should develop separate SSLs for each source. The SAP developed for subsurface soils should specify sampling and analytical procedures as well as the development of QA/QC procedures. To identify the appropriate procedures, the SSLs must be known. If data are not available to calculate site* specific SSLs (Section 2.5.2), then the generic SSLs in Appendix A of the TBD should be used. The primary goal of the subsurface sampling strategy is to estimate the mean contaminant concentration and average soil characteristics within the source area. As with the surface soil sampling strategy, the subsurface soil sampling strategy follows the DQO process (see Exhibits 5, 6, and 8). The decision rule is based on comparing the mean contaminant concentration within each contaminant source with source-specific SSLs. Current investigative techniques and statistical methods cannot accurately determine the mean concentration of subsurface soils within a contaminated source without a costly and intensive sampling program that is well beyond the Jevel of effort generally appropriate for screening. Thus, conservative assumptions should be used to develop hypotheses on likely contaminant distributions. This guidance bases the decision to investigate a source area further on the highest mean soil boring contaminant concentration within the source, reflecting the conservative assumption that the highest mean subsurface soil boring concentration among a set of borings taken from the source area represents the mean of the entire source area. Similarly, estimates of contaminant depths should be conservative. The investigation should include the maximum depth of contamination encountered within the source without going below the water table. For each source, the guidance recommends taking 2 or 3 soil borings located in the areas suspected of having the highest contaminant concentrations within the source. These subsurface soil sampling locations are based primarily on knowledge of likely surface soil contamination patterns (see Exhibit 6) and subsurface conditions. However, buried sources may not be discernible at the surface. Information on past practices at the site included in the CSM can help identify subsurface source areas. For sites contaminated with VOCs, the subsurface sampling strategy should include soil gas surveys as well as soil matrix sampling. VOCs are commonly found in vapor phase in the '"rsatiiratgd zone, and soil matrix samples may yield results that are deceptively low. Soil gas data are needed to help locate sources, define source size, to place soil boring locations within a source, and can also be used in conjunction with modeling to address VOC transport 'in the vadose zone for both the volatilization and migration to ground water iways. Take soil cores from the soil boring using either split spoon sampling or other appropriate sampling methods. Description and Sampling of Contaminated Soils: A Field Pocket Guide (U.S. EPA, 1991f£ and Subsurface Characterization and Monitoring "-Techniques: A Desk Reference Guide, Vol. l&tt (U.S. EPA, 1993e), can be consulted for information on appropriate subsurface sampling methods. Sampling should begin at the ground surface and continue until either no contamination is encountered or the water table is reached. Subsurface sampling intervals can be adjusted at a site to accommodate site-specific infor- mation on subsurface contaminant distributions and geological conditions (e.g., thick vadose zones in the West). The concept of "sampling support" introduced in Section 2.3.2 also applies to subsurface sampling. For example, sample splits and subsampling . should be performed according to Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies (U.S. EPA, 1992e). 16 TUT COS 1060 If each subsurface soil core segment represents the """•^e subsurface'soil interval (e.g., 2 feet), then the £rage concentration from the surface to die depth of contamination is the simple arithmetic average of contaminant concentrations measured for core samples representative of each of the 2-foot segments from the surface to the depth of contamination. However, if the sample intervals are not all of the same length (e.g., some are 2 feet while others are 1 foot or 6 inches), then the calculation of the average concentration in the total core must account for the different lengths of the segments. If Cj is the concentration measure in a core sample, representative of a core interval or segment of length lj, and the n-th segment is considered to be the last segment sampled in the core (i.e., the n-th segment is at the depth of contamination), then the average concentration in the core from the surface to the depth of contamination should be calculated as the following depth-weighted average (c). Alternatively, the average boring concentration can be determined by adding the total contaminant masses together (from the sample results) for all sample segments to get the total contaminant mass for the boring. The total contaminant mass is then divided by the total dry weight of the core (as determined by the dry bulk density measurements) to estimate average soil boring concentration. For the leach test option, collect discrete samples along a soil boring from within the zone of contamination and composite them to produce a sample representative of the average soil boring concentration. Take care to split each discrete sample before analysis so that information on contaminant distributions with depth will not be lost. A leach test may be conducted on each soil core. Finally, the soil investigation for the migration to ground water pathway should not be conducted independently of ground water investigations. Contaminated ground water may indicate the presence of a nearby source area that would leach contaminants from soil into aquifer systems. 2£A Develop Sampling and Analysis Plan to Dctarrnine Soil Characteristics The soil parameters necessary for SSL calculations are soil texture, dry bulk density, soil organic carbon, and pH Some can be measured in the field, while others require laboratory measurement. Although laboratory measurements of these parameters cannot be obtained under Superfund's CLP, independent soil testing laboratories across the country can perform these tests at a relatively low cost. To appropriately apply the volatilization and migration-to-ground water models^ average or typical soil properties should be used for a source in the SSL equations (see Step 5). Take samples for measuring soil parameters with samples for measuring contaminant concentrations. If possible, consider splitting single samples for contaminant and soil parameter measurements. Many soil testing laboratories can handle and test contaminated samples. However, if testing contaminated samples for soil parameters is a problem, samples may be obtained from clean areas of the site as long as they represent the same soil texture and are taken from approximately' the same depth as the contaminant concentration samples. Soil Texture. Soil texture class (e.g., loam, sand, silt loam) is necessary to estimate average soil moisture conditions and to apply the Hydrological Evaluation of Landfill Performance (HELP) model to estimate infiltration rates (see Attachment A). The appropriate texture classification is determined by a particle size analysis and the U.S. Department of Agriculture (USDA) soil textural triangle shown in Exhibit 9. This classification system is based on the USDA soil particle size classification. The particle size analysis method in Gee and Bauder (1986) can provide this particle size distribution. Other methods are appropriate as long as they provide the same particle size breakpoints for sand/silt (0.05 mm) and silt/clay (0.002 mm). Field methods are an alternative for determining soil 17 I"UT OOS 1O61 textual class; Exhibit 9 presents an example from Brady(1990). ' Dry Bulk Density. Dry soil bulk density (pb) is used to calculate total soil porosity and can be determined for any soil horizon by weighing a thin- wailed tube soil sample (e.g., Shelby tube) of known volume and subtracting the tube weight [American Society for Testing and Materials (ASTM) D 2937]. Determine moisture content (ASTM 2216) on a subsample of the tube sample to adjust field bulk density to dry bulk density. The other methods (e.g., ASTM D 1556, D 2167, D 2922) are generally applicable only to surface soil horizons and are not appropriate for subsurface characterization. ASTM soil testing methods are readily available in the Annual Book of ASTM Standards, Volume 4.08, Soil and Rock; Building Stones, available from ASTM, 100 Barr Harbor Drive, West Conshohocken, PA, 19428. Organic Carbon and pH. Soil organic carbon is measured by burning off soil carbon in a controlled- temperature oven (Nelson and Sommers, 1982). ..This parameter is used to determine soil-water partition coefficients from the organic carbon soil- water partition coefficient, KOC. Soil pH is used to select site-specific partition coefficients for metals (Table C-4, Attachment C) and ionizing organics (Table C-2, Attachment C). This simple measurement is made with a pH meter in a soil/water slurry (McLean, 1982) and may be measured in the field using a portable pH meter. 2.3.5 Determine Analytical Methods and Establish QA/QC Protocols. Assemble a list of feasible sampling and analytical methods during this step. Verify that a CLP method and a field method for analyzing the samples exist and that the analytical method QL or field method QL is appropriate for (i.e., is below) the site-specific or generic SSL. Sampler's Guide to the Contract Laboratory Program (U.S. EPA, 1990b) and User's Guide to the Contract Laboratory Program (U.S. EPA, 1991e) contain further .information on CLP methods. Field methods, such as soil gas . surveys, immunoassay, or X-ray fluorescence, can be used if the field method quantitation limit is below the SSL. EPA recommends the use of field methods where applicable and appropriate. However, at least 10 percent of both the 'discrete samples and the composites should be split and sent to a CLP laboratory for confirmatory analysis. (Quality Assurance for Superfund Environmental Data Collection Activities, U.S. EPA, 1993c). Because a great amount of variability and bias can exist in the collection, subsampling, and analysis of soil samples, some effort should be made to characterize this variability and bias. A Rationale for the Assessment of Errors in the Sampling of Soils (U.S. EPA, 1990c) outlines an approach that advocates the use of a suite of QA/QC samples to assess variability and bias. Field duplicates and splits are some of the best indicators of overall variability in the sampling and analytical processes. Field methods will be useful in defining the study boundaries, (i.e., area and depth of contamination) during both site reconnaissance and sampling. The 'design and capabilities of field portable instrumentation are rapidly evolving. Documents describing the standard operating procedures for field instruments are available though the National Technical Information Service (NTIS). Regardless of whether surface or subsurface soils are sampled, the Superfund quality assurance program guidance (U.S. EPA, 1993c) should be consulted. Standard limits on the precision and bias of sampling and analytical operations conducted during sampling do apply and should be followed to give consistent and defensible results. 2.4 Step 4: Sampling and Analyzing Site Soils & DQA Once the sampling strategies have been developed and implemented, the samples should be analyzed according to the analytical laboratory and field methods specified in the SAP. Results of the anal- yses should identify the concentrations of potential contaminants of concern for which site-specific SSLs will be calculated 18 TUT OO8 1O&2 Exhibit 9: U.S. Department of Agriculture soil texture classification. 100 X b -%> Percan Pe>rce>nt Sand criteria Used with the Field Method for Determining Soil Texture Classes (Source: Brady, 1990) Criterion Sand Sandy loam Loam Silt loam Clay loam Clay 1. Individual grains Yes Yes visMetoeye 2. Stability of dry Donotform Do not form dods 3. Stability of wet Sonw Few No dods 4. Stability of Unstable Does not Sightly stable Dots not form Eaeiy Moderately Hardand broken - eesty broken ataMe ModmMy Stable Very stable etabte Doasnottom Broken appearance Thin, wl break No Very hard andetabto Very etabte Very long, •enbte wet aoi rubbed Dslwstn thumb and fingers 0.002 Particle Size, mm 0.05 ' 0.100.25 0.5 1.0 2.0 U.S. Department of Agriculture Ciay Silt VeryRne j Finej Medj Coarse j Very Come Sand Qrevel Source: USOA. 19 TUT OO8 1063 2.4.1 Delineate Area and Depth of Source. Both spatial area and depth data, as well as soil characteristic data, -re needed to calculate she- specific SSLs for the inhalation of volatiles and migration to ground water pathways in the subsurface. Site information from the CSM or soil gas surveys can be used to estimate the arcal extent of the sources. 2.4.2 Perform DQA Using Sample Results. After sampling has been completed, a DQA should be conducted if all composite samples are less than 2 times the SSL. This is necessary to determine if the original CV estimate (2.5), and hence the number of samples collected (6), was adequate for screening surface soils. To conduct the DQA for a composite sample whose mean is below 2 SSL, first calculate the sample CV for the EA in question from the sample mean (x), the number of specimens per composite sample (C), and sample standard deviation (s) as follows: CV Use the sample size table in Exhibit 7 to check, for this CV, whether the sample size is adequate to meet the DQOs for the sampling effort. If sampling DQOs are not met, supplementary sampling may be needed to achieve DQOs. However, for EAs with small sample means (e.g., all composites are less than the SSL), the sample CV calculated using the. equation above may not be a reliable estimate of the population CV (i.e., as x approaches zero, the sample CV will approach infinity). To protect against unnecessary additional sampling in such cases, compare all composites against the formula SSL A/C . If the maximum composite sample concentration is below the value given by the equation, then the sample size may be assumed to be adequate and no further DQA is necessary In other words, EPA believes that the default sample size will adequately support walk- away decisions when all composites are well below the SSL. The TBD describes the development of this formula and provides additional information on implementing the DQA process. 2.4.3 Ravise the CSM. Because these analyses reveal new infbrmr1 on about the site, update the CSM accordingly. This revision could includ< identification of she areas that exceed the generic SSLs. 2.5 Step_£: Calculating Site- specific SSLs With the soil properties data collected in Step 4 of the screening process, she-specific soil screening levels can now be calculated using the equations presented in this section. For a description of how these equations were developed, as well as background on their assumptions and limitations, consult the TBD. All SSL equations were developed to be consistent with RME in the residential setting. The Superfund program estimates the RME for chronic exposures on a she-specific basis by combining an average exposure-point concentration with reasonably con- servative values for intake and duration (U.S. EPA, 1989a; RAGS HHEM, Supplemental Guidance: Standard Default Exposure Factors, U.S. EPA, 1991a). Thus, all site-specific parameters (soil, aquifer, and meteorologic parameters) used to calculate.«SSLs should reflect average or typical sh conditions in order to calculate average exposure concentrations at the she. Equations for calculating SSLs are presented for surface and subsurface soils in the following sections. For each equation, site-specific input parameters are highlighted in bold and default values are provided for use when site- specific data are not available. Although these defaults are not worst case, they are conservative. At most sites, higher, but soil protective SSLs can be calculated using site-specific data. The TBD describes development of these default values and presents generic SSLs calculated using the default values. Attachment D provides toxicity criteria for 110 chemicals commonly found at NPL sites. These criteria were obtained from Integrated Risk Information System (IRIS) (U.S. EPA, 1995b) or Health Effects Assessment Summary Tables (HEAST) (U.S. EPA, 1995a), which are regularly 20 TOT updated. Prior .to calculating SSLs at a site, check all relevant chemical-specific values in /""^ttachnient D against values from IRIS or ' JEAST. Only the most current Values should be used to calculate SSLs. Where toxicity values have been updated, the generic SSLs should also be recalculated with current toxicity information. 2J5.1 SSL Equations—Surface Soils. Exposure pathways addressed in the process for screening surface soils include direct ingestion, dermal contact, and inhalation of fugitive dusts. Direct Ingestion. The Soil Screening Guidance addresses chronic exposure to noncarcinogens and carcinogens through direct ingestion of contaminated soil in a residential setting. The approach for calculating noncarcinogenic SSLs presented in this guidance leads to screening levels that are approximately 3 times more conservative than PRGs calculated based on the approach presented in RAGS HHEM, Part B (i.e., using a 30- year, time-weighted average soil ingestiou rate for comparison to chronic toxicity criteria). Because a number of studies have shown that inadvertent >*M;esticm of soil is common among children age 6 j younger (Calabrese et al., 1989; Davis et al., 1990: Van Wijnen et al., 1990), several commenters suggested that screening values should be based tm this increased exposure during childhood. However, other commenters believe that comparing a six-year exposure to a chronic reference dose (RfD) is unnecessarily conservative. In their analysis of this issue, the Science Advisory Board (SAB) stated that, for most chemicals, the approach of combining the higher six-year exposure for children with chronic toxicity criteria is overly protective (U.S. EPA, 1993 f). However, they noted that the approach may be appropriate for chemicals with chronic RfDs based on toxic endpoints that are specific to children (e.g., fluoride and nitrates) or where the dose-response curve is steep [i.e., the difference between the no-observed-adverse-effect level (NOAEL) and the adverse effect level is small]. Thus for the purposes of screening, Office of Emergency Remedial Response (OERR) opted to base the generic SSLs for noncarcinogenic contaminants on the more conservative "childhood only" exposure (Equation 1). The issue of whether to maintain this more conservative approach throughout the Baseline Risk Assessmert and establishing remediation goals will depend on how the specific chemical's toxicology relates to the issues raised by the SAB. Equation 1: Screening Level Equation for Irigestion -of Noncarcinogenic Contaminant* in Residential Soil Screwing Lave) THQxBWxATx365CrVr- • 1/RfD0 x 104 ktttng * EF x ED x IR Parameter/Definition (units) THQAarget hazard quotient (unitieM) BW/body weight (kg) AT/averaging time (yr) RfDj/orml reference dose (mg/kg-d) .• EF/exposure frequency (oVyr) ED/exposure duration (yr) IR/soil ingestion rate (mg/d) Default 1 15 *• chemical-specific (Attachment D) 350 6 200 •For noncarcinogens, averaging time equals to exposure duration. For carcinogens, bom the magnitude and duration of exposure are important. Duration is critical because the toxicity criteria are based on "lifetime average daily dose." Therefore, the total dose received, whether it be over 5 years or 50 years, is averaged over a lifetime of 70 years. To be protective of exposures to carcinogens in the residential setting, Superfund focuses on exposures to individuals who may live in the same residence for a high-end period of time (e.g., 30 years) because exposure to soil is higher during childhood and decreases with age. Equation 2 uses a time-weighted average soil ingestion rate for children and adults. The derivation of this time-weighted average is presented in U.S. EPA, 1991c. Default values are used for all input parameters in the direct ingestion equations. The amount of data required to derive site-specific values for these parameters (e.g., soil ingestion rates, chemical- specific unavailability) makes their collection and use impracticable for screening. Therefore, site- specific data are not generally available for this 21 TUT OOS 1O65 exposure route. The generic ingestion SSLs p-?sented in Appendix A of the TBD are recommended for all NPL sites. Equation 2: Screening Level Equation for Ingestion of Carcinogenic Contaminants in Residential Soil Screening Level = TR x AT x 365 <*y (mg/kg) SF0x IC^IqB^rngxEFxIF^j^ Parameter/Definition (units) TFVtarget cancer risk (unttless) AT/averaging time (yr) SF0/oral slope factor (mg/kg-d)-1 EF/exposure frequency (d/yr) IF^oifeq /age-adjusted soil ingestion factor (mg-yr/kg-d) Default 10* 70 chemical-specific (Attachment D) 350 114 Dermal Contact. Contaminant absorption through dermal contact may contribute risk to human health in a residential setting. However, incorporation of dermal exposures into the soil screening process is limited by the amount of data available to quantify dermal absorption from soil for specific chemicals. Previous EPA studies suggest that absorption via the dermal route must be greater than 10 percent to equal or exceed the ingestion exposure (assuming 100 percent absorption of a chemical via ingestion; Dermal Exposure Assessment: Principles and Applications, U.S. EPA, 1992b). Of the 110 compounds evaluated, available data show greater than 10 percent dermal absorption for pentachlorophen.ol (Wester et al., 1993). Therefore, pentachlorophenol is the only chemical for which the Soil Screening Guidance directly considers dermal exposure. The ingestion SSL for pentachlorophenol should be divided in half to account for the assumption that exposure via the dermal route is equivalent to the ingestion route. Preliminary studies show that certain semivolatile compounds (e.g., benzo(o)pyrene) may also be of concern for this exposure route. As adequate dermal absorption data are developed for such chemicals, the ingestion SSLs may need to be adjusted. The Agency will provide updates on this issue as appropriate. Inhalation of Fugitive Dusts. Inhalation of fugitive dusts is a consideration for semivolatile organics and metals in surface soils. However, generic fugitive dust SSLs for semivolatile organics are several orders of magnitude nigher than the corresponding generic ingestion SSLs. EPA believes that since the .ingestion route should always be considered in screening decisions for surface soils, and ingestion SSLs appear to be adequately protective for inhalation exposures to fugitive dusts for organic compounds, the fugitive dust exposure route need not be routinely considered for organic chemicals in surface soils. ' Likewise, the ingestion SSLs are significantly more conservative than most of the generic fugitive dust SSLs. As a result, fugitive dust SSLs need not be calculated for most metals. However, chromium is an exception. For chromium, the generic fugitive dust SSL is below the ingestion SSL. This is due to the carcinogenicity of hexavajent chromium, Cr+*, through the inhalation. exposure route. For most sites, fugitive dust SSLs calculated using the conservative defaults will be adequately protective. However, if site conditions mat will result in higher fugitive dust emissions than the defaults (e.g., dry, dusty soils; high average annual windspeeds; vegetative cover less than 50 percent) are likely, consider calculating a si$e-specific fugitive dust SSL. Equations 3 and 4 are used to calculate fugitive dust SSLs for carcinogens and noncarcinogens. These equations require calculation of a paniculate emission factor (PEF, Equation 5) mat relates the concentration of contaminant in soil to the concentration of dust particles in air. This PEF represents an annual average emission rate based on wind erosion that should be compared with chronic health criteria. It is not appropriate for evaluating the potential for more acute exposures. Both the' emissions portion and the dispersion portion of the PEF equation have been updated since the first publication of RAGS HHEM, Part B, in 1991. As in Pan B, the emissions pan of the PEF equation is based on the "unlimited reservoir" model developed to estimate paniculate emissions due to wind erosion (Cowherd et al., 1985). Additional information on the update of the PEF equation is provided in thr TBD. Cowherd et al. (1985) present methods for site-specific measurement of the 22 TUT 008 1066 parameters necessary to calculate a PEF. A site- specific dispersion model 'O/Q is then selected as bribed in the section on calculating SSLs for the . -atile inhalaticm pathway later in this document. Equation 3: Screening Level Equation for , Inhalation of Carcinogenic Fugitive Dusts from Residential Soil Screening Level (mg/kg) TRxATx365dVr PEF Parameter/Definition (units) Default TrVtarget cancer risk (unitless) AT/averaging time (yr) URF/inhalation unit risk factor EF/exposure frequency (d/yr) ED/exposure duration (yr) PEF/particulate emission factor (m'/kg) 70 chemical-specific (Attachment D) 350 30 1.32 x 10* (Equation 5) [Equation 4: Screening Level Equation for Inhalation of Noneareinogenic Fugitive Dusts from Residential Soil screening Level « THQ x AT x 365 tifyr (mo/kg) EFx EDx f 1 x 1 1 RfC PEF Parameter/Definition (units) THQ/target hazard quotient (unitless) AT/averaging time (yr) EF/exposure frequency (d/yr) ED/exposure duration (yr) RfC/inhalation reference concentration (mg/m3) PEF/particulate emission factor <m»/kg) Default 1 30 350 30 chemical-specific (Attachment D) 1.32 x 10* (Equation 5) 2.5.2 SSL Equations—Subsurface Soils. The Soil Screening Guidance addresses two exposure pathways for subsurface soils: inhalation of volatiles and ingestion of ground water contaminated by the migration of contaminants through soil to an under- lying potable aquifer. Because the equations developed to calculate SSLs for these pathways assume an infinite source, they can violate mass- balance considerations, especially for small sources. Equation 5: Derivation of the Paniculate Emiaaion Factor PEF(rrtVkg)«QCx 3,600 s/h 0.036 x(1-V)x(LWUJ3xF(x) Parameter/Definition (units) PEF/particulate emission factor (mVkg) Q/C/invarse of mean cone, at canter of a 0.5»aere-aquare source (g/m2-s par kg/in*) Wfraction of vegetative cover (unitless) Um/mean annual windspeed (m/a) Ut /equivalent threshold value of wlndapeed at 7 m (m/a) F(x)/function dependent on Um/Ut derived using Cowherd at al. (1085) (unhless) » Default 1.32x10* 90.80 0.5 (50%) 4.69 11.32 0.194 To address mis concern, the guidance also includes equations for calculating mass-limit SSLs for each of these pathways when the size (i.e., area and depth) of the contaminated soil source is known or can be estimated with confidence. Attachment D provides the toxicity criteria and regulatory benchmarks for 110 chemicals commonly found at NPL sites. These criteria were obtained from IRIS (U.S. EPA, 1995b), HEAST (U.S. EPA, 199Sa), and Drinking Water Regulations and Health Advisories (U.S. EPA, 1995c), which are regularly updated. Prior to calculating SSLs at a site, all relevant chemical-specific values in Attachment D should be checked against the most recent version of their sources to ensure that they are up to date. Toxicity data are not available for all chemicals for the inhalation exposure route. At the request of commenters, EPA has looked into methods for extrapolating inhalation toxicity values from oral toxicity data. The TBD presents the results of this analysis along with information on current EPA 23 "UT 008 1O67 practices for conducting such route-to-route extrapolations. Chemical properties necessary to calculate SSLs for the inhalation and migration to ground water path- ways include solubility, air and water diffusivities, Henry's law constant, and soil/water partition coeffi- cients. Attachment C provides values for 110 chemicals commonly found at NPL sites. Site-specific parameters necessary to calculate SSLs for subsurface soils are listed on Exhibit 10, along with recommended sources and measurement methods. In addition to the soil parameters described in Step 3, other site-specific input parameters include soil moisture, infiltration "rate, aquifer parameters, and meteorologic data. Guidance for collecting or estimating these other parameters at a site is provided on Exhibit 10 and in Attachment A. Inhalation of Volatiles. Equations 6 and 7 are used to calculate SSLs for the inhalation of carcinogenic and noncarcinogenic volatile contaminants .To use these equations to calculate inhalation SSLs, a volatilization factor (VF) must be calculated. The VF equation can be broken into two separate models: a model to estimate the emissions and a dispersion model (reduced to the term Q/C) that simulates the dispersion of contaminants in ambient air. In addition, a soil saturation limit (CHt) must be calculated to ensure that the VF model is applicable to soil contaminant conditions at a site. Volatilization Factor (VF). The soil-to-air VF (Equation 8) is used to define the relationship between the concentration of the contaminant in soil and the flux of the volatilized contaminant to air. The Soil Screening Guidance replaces the Hwang and Falco (1986) model used as the basis for the RAGS HHEM, Part B, VF equation with the simplified equation developed by Jury et al, (1984). • The Jury model calculates the maximum flux of a contaminant from contaminated soil and considers soil moisture conditions in calculating a VF. The models are similar in their assumptions of an infinite contaminant source and vapor phase diffusion as the only transport mechanism (i.e., no transport takes place via nonvapor-phase diffusion and-there is no mass flow due to capillary action). In some situations, information about the size of the source is available and SSLs can be calculated using the mass-limit approach. Equation 6: Screening Level Equation for Inhalation of Carcinogenic Volatile Contaminants in Residential Soil Screening 'Level (mg'kg) TRxATx36Sdyr URF x 1.000 ugfrng x EFxEOi VF. Parameter/Definition (unlta) Default TRAarget cancer risk (unities*) AT/averaging time (yr) URF/inhalation unit risk factor (ug/m3)-i EF/exposure frequency (oVyr) ED/exposura duration (yr) VF/eoll-to-air volatilization factor (mVkg) 10* 70 chemical-specific (Attachment D) 350 30 chemical-specific (Equation 6) Equation 7: Screening Level Equation for Inhalation of Noncarcinogenic Volatile Contaminants in Residential Soil Screening Level = THQ x AT x 365 d/yr (mg/kg| EFx EDx [_]_ x_i_] RfC VF Parameter/Definition (units) THQ/targat hazard quotient (unitless) AT/averaging time (yr) EF/exposure frequency (d/yr) ED/exposure duration (yr) RfC/inhaiation reference concentration (mg/m3) VF/soil-to-air volatilization factor (m»/kg) Default 1 30 350 30 chemical-specific (Attachment D) chemical-specific (Equation 8) Other than initial soil concentration, air-filled soil porosity is the most significant soil parameter affecting the final steady-state flux of volatile contaminants from soil (U.S. EPA, 1980). In other words, the higher the air-filled soil porosity, the greater the emission flux of volatile constituents. 24 TUT 008 Exhibit 10. Site-specific Parameters for Calculating Subsurface SSLs SSLPatnvay Mgrationto Pajamaiar fcinalaaofl 0rouito waiar Sourca Characteristics Sourca tength <L) • Souroadapth • • Soi Characterir tics Soil texture O O • Dn/soibukdansitycp,,) • • Sol moisture content (w) O O SoB organic carbon (LJ • • SolpH 0 0 Saluratadriy*auic conductivity O O (K.) Avg. soil moistura content (BJ • ' • Mataorotogical Data Airdwparsion (actor (Q/C) • Hydrogaotogic Characteristics (DAF) Hydrogsotogic satting O •^ lntBtmtkxVrecharg»(l) - • Hydraulic conductivity (K) • Hydraulic graoSant(i) . • • Aquifar «vcknass(d) • • indicates parameters used in the SSL equations. O Indicates parameten/assuBiptions needed to estimate SSL eouahon t Dstaaouroa Samping data ^^^^^J.^ J^ m^ stamfjsng OBBI Samping date Labmaasuramant nalo maaauramant Labmaaaurarnant Lib maasuramant Fiaid maasuflsmant Look-up Look-up Cafcyated Q/Ctabia(TabteS) ConcapiuaJ site •MI nrlail friOQM HOP modal; Ragionalastimatas Raid maasuramant nagional asDmatas Raid maasuramant Ragional astimate* Raid maasuramant Ragional astinates wnmeters. Uattwd Maasura total area of contaminated soi Maasura tength of sourca paraM to ground water tow Maaauia dapvi of oofftfaminalon orusa Panlola ate anaJysts (Qaa & Baudar, 1966) and USDAc»us«oatorv.usadtoas*natee*&! Al sols: ASTM D 2937; ahaJtow softs: ASTM D 1586, ASTMD 2167, ASTMD 2822 A£rjMD22t6;usadtoaaarnatedrysoibul( NMMH And Sonvrwn (1982) Md-a^1982);usadtosalactpH-spacificKoc AttKtvTiifit A; uMdto dlciataMiQiiy Attschmant A; usad to calculate 6,, . AttKhmantA Satect valua eorraapondng to sourca ana, camatic ana, and city with oondrsons similar to site aL(19B7)torastBTiatenofparamatersbalow (saaAttechmantA) HELP (Schreadar at aL. 19B4) may ba usad «or srte-spacinc inHtration astimates; recharge asimatas also may ba takan from Alter at al. (1967) or may ba animated from knowtedaa of AquHar teats (La., pump tests, slug tests) praterrad; astimates abo may ba takan tarn Alar at al. (1967) or Maws* at a]. (1990) or may hydrogaotogk; condfeons Maasurad on map of site's water tabi* Nawai at at. (iwuior may oa aiimateu nwn Sravapacinc maasuramant (i.a., from soi boring (mm Nawal at al. (1990) or may ba astimated from knoKHsdgs of local hyrtogsotogic eoncitions 25 TUT COS 1O69 Equation 8: Derivation of the Volatilization Factor QfC x (3.14 x DA x T) i« x where Parameter/Definition (unite) VF/volatUization factor (ma/kg) DA /apparent diffushrity (cmz/s) Q/C/inveree of the mean cone, at the center of a 0.5-acre-square aource (g/m2-e per kg/m3) T/exposure interval (s) Pb/dry eoil bulk densHy (g/cm3) 6a /air-filled soil porosity a/total soil porosity 6w/water-filled eoil poroaity ps /soil particle density (g/cm3) D|/drffusivity in air (cm2/s) H'/dimensionless Henry's law constant Dw /diKusivity in water (crftZ/s) Kd /soil-water partition coefficient (cm3/g) = KOC foe (organics) KOC /soil organic carbon partition coefficient (crrfl/g) [oc/f raction organic carbon in eoil (gig) Default 68.81 9.5 x 10» 1.5 n-6w 0.15 2.65 chemical-specific* chemical-specific*. chemical-specific* chemical-specific* chemical-specific* 0.006 (0.6%) •See Attachment C. Among the soil parameters used in Equation 8, annual average water-filled soil porosity (6W) has the most significant effect on air-filled soil porosity (6J and hence volatile contaminant emissions. Sensitivity analyses have shown that soil bulk density (pb) has too limited a range for surface soils (generally between 1.3 and 1.7 g/cm 3) to affect results with nearly the significance of soil moisture content (U.S. EPA, 1996). Dispersion Model (Q/C). The box model in RAGS HHEM, Part B has been replaced with a Q/C term derived from the modeling exercise using the AREA- ST model incorporated into EPA's Industrial Source Complex Model (ISC2) platform. The AREA-ST model was run with a full year of meteorological data for 29 U.S. locations selected to be representative of a range of meteorologic conditions across the Nation (EQ, 1993). The results of these modeling runs are presented in Exhibit 11 for square area sources of 0.5 Jo 30 acres in size. When developing a she-specific VF for the inhalation pathway, place the site into a climatic zone (see Attachment B). Then select a Q/C value from Exhibit 11 that best represents a site's size and meteorological conditions. Soil Saturation Limit (Cut). The soil saturation limit (Equation 9) is the contaminant concentration at which soil pore air and pore water are saturated with the chemical and the adsorptive limits of the soil particles have been reached. Above this concentration, the contaminant may be present in free phase. Cut concentrations represent an upper limit to the applicability of the SSL VF model because a basic principle of the model (Henry's law) does not apply when contaminants are present in free phase. VF-based inhalation SSLs are reliable only if they are at or below CfBt. Equation 9 is used to calculate the soil saturation limit for each organic chemical in site soils. As an update to RAGS HHEM, Part B, this equation takes into account the -amount of contaminant that is in title vapor phase in the pore spaces of the soil in addition to the amount dissolved in the soil's pore water and sorbed to soil particles. €,« values should be calculated using the same site-specific soil characteristics used to calculate SSLs (e.g., bulk density, average water content, and organic carbon content). Because VF-based SSLs are not accurate for soil concentrations above C»«, these SSLs should be compared to Citt concentrations before they are used for sou screening. 26 TUT (->OS Exhibit 11. Q/C Values by Source Araa, City, and Climatic Zone Q/C (gVm2-S Zone 1 Seattle SaJem Zone II Fresno . Los Angeles San Francisco Zone III LasVegas Phoenix Afeuquerque Zone IV Boise Winnemucca Salt Lake City Casper Denver Zone V Bismark Minneapolis Lincoln Zone VI Little Rock Houston Atlanta Charleston Raleigh-Durham Zone VII Chicago Cleveland Huntingdon Harrisburg Zone VIII Portland Hartford Philadelphia Zone IX Miami 0.5 Acre 82.72 73.44 62.00 68.81 89.51 95.55 64.04 84.18 69.41 69.23 78.09 100.13 75.59 83.39 90.80 81.64 73.63 79.25 77.08 74.89 77.26 97.78 83.22 53.89 81.90 74.23 71.35 90.24 85.61 1 Acre 72.62 64.42 54>37 60.24 78.51 83.87 56.07 73.82 60.88 • 60.67 68.47 87.87 66.27 73.07 79.68 71.47 64.51 69.47 67.56 65.65 67.75 85.81 73.06 47.24 71.87 65.01 62.55 79.14 74.97 2 Acre 64.38 57.09 . 48.16 53.30 69.55 • 74.38 49.59 65.40 53.94 53.72 60.66 77.91 58.68 64.71 70.64 63.22 57.10 61.53 59.83 58.13 60.01 76.08 64.78 41.83 63.72 57.52 55.40 70.14 66.33 per kg/m3) 5 Acre 55.66 49.33 41.57 45.93 60:03 64.32 42.72 56.47 46.57 46.35 52.37 67.34 50.64 55.82 61.03 54.47 49.23 53.11 51.62 50.17 51.78 65.75 55.99 36.10 55.07 49.57 47.83 60.59 57.17 10 Acre 50.09 44.37 37.36 41.24. 53.95 57.90 38.35 50.77 41.87 41.65 47.08 60.59 45.52 50.16 54.90 48.89 44.19 47.74 46.37 45.08 46.51 59.16 50.38 32.43 49.56 44.49 43.00 54.50 51.33 30 Acre 42.86 37.94 31.90 . 35.15 46.03 49.56 32.68 43.37 35.75 35.55 40.20 51.80 38.87 42.79 46.92 41.65 37.64 40.76 39.54 38.48 39.64 50.60 43.08 27.67 42.40 37.88 36.73 46.59 43.74 27 TUT COS 1O71 Equation 9: Derivation *f the Soil Saturation Limit -H-e.) ft. Parameter/Definition (unite) Cut/soil saturation concentration (mg/kg) S/soiubilrty in water (mg/L-water) pb/dry aoil bulk density (kg/L) K<| /soil-water partition coefficient (L/kg) organic carbon/water partition coefficient (L/kg) foe/fraetion organic carbon in •oil (g/g) 6w/water-filled »oil porosity H'/dimensionless Henry's law constant ea /air-filled soil porosity n/total soil porosity ps /soil particle density (kg/L) Default chemical-specific* 1.6 KOC x foe (chemical- specific*]! chemical-specific* 0.006 (0.6%) 0.15 chemical-specific* n-6w 1-(Pb/P.) 2.65 *See Attachment C. CItt values represent chemical-physical limits in soil and are not risk based. However, since they represent the concentration at which soil pore air is saturated with a contaminant, volatile emissions reach their maximum at C§lt. In other words, at CM the emission flux from soil to air for a chemical reaches a plateau. Volatile emissions will not increase above mis level no matter how much more chemical is added to the soil. Chemicals with VF- based SSLs above C(at are not likely to present a significant volatile inhalation risk at any soil concentration. To illustrate this point, the TDB presents an analysis of the inhalation risk levels at CMt for a number of chemicals commonly found at Superfund sites whose generic SSLs (calculated using the default parameters shown in Equation 9) are above €„,. The analysis indicates that these Cut values are all well below the screening risk targets of a 1(H cancer risk or an HQ of 1. Although the inhalation risks appear to be negligible, C,«t does indicate a potential for nonaqueous phase liquid (NAPL) to be present in soil and a possible risk to ground water. Thus, EPA believes that farther investigation is warranted. Table C-3 (Attachment C) provides the physical state, liquid or solid, of various compounds at ambient soil temperature. When an inhalation SSL exceeds C,at for compounds that are liquid at ambient soil temperature, the SSL is set at CMt. Where soil concentrations exceed a Cut-based SSL, site managers should refer to EPA's guidance, Estimating the Potential for Occurrence ofDNAPL at Superfimd Sites (U.S. EPA, 1992c) for further information on determining the likelihood of dense nonaqueous. phase liquid (DNAPL) in the subsurface. Note that free-phase contaminants may be present at concentrations below C(at if multiple organic contaminants are present. The DNAPL guidance (U.S. EPA, 1992c) also provides tools for evaluating the potential for such multiple component mixtures in soil. For organic compounds that are solid at ambient soi temperature, concentrations above CMt do not pose a significant inhalation risk or a potential for NAPL occurrence. Thus, soil screening decisions should be based on the appropriate SSL for other site pathways (e.g., migration to ground water, direct ingestion). Migration to Ground Water SSLs. The Soil Screening Guidance uses a simple linear equilibrium soil/water partition equation or a leach test to estimate contaminant release in soil leachate. It also uses a simple water-balance equation to calculate a dilution factor to account for reduction of soil leachate concentration from mixing in an aquifer. The methodology for developing SSLs forme migra- tion to ground water pathway was designed for use during the early stages of a site evaluation when information about subsurface conditions may be limited. Hence, the methodology is based on rather conservative, simplified assumptions about the release and transport of contaminants in the 28 TUT 008 1072 subsurface (Exhibit 12). These'assumptions are s jtnherent in the SSL equations and should be reviewed »r consistency with the conceptual site model (see step 2) to determine the applicability of SSLs to the migration to ground water pathway. Exhibit 12: Simplifying Assumptions for the SSL Migration to Ground Water Pathway • Infinite source (i.e., steady-state concentrations are maintained over the exposure period) • Uniformly distributed contamination from the surface to the top of the aquifer • No contaminant attenuation (i.e., adsorption, biodegradation, chemical degradation) in soil • Instantaneous and linear equiltorium soil/water partitioning • Unconfined, unconsolidated aquifer with homogeneous and isotropic hydrologic properties • Receptor well at the downgradient edge of the source and screened within the plume • No contaminant attenuation in the aquifer • No NAPLs present (if NAPLs are present, the SSLs do not apply). To calculate SSLs for the migration to ground water pathway, multiply the acceptable ground water concentration by the dilution factor to obtain a target soil leachate concentration. For example, if the dilution factor is 10 and the acceptable ground water concentration is 0.05 mg/L, the target soil/water leachate concentration would be 0.5 mg/L. Next, the partition equation is used to calculate the total soil concentration (i.e., SSL) corresponding to this soil leachate concentration. Alternatively, if a leach test is used, compare the target soil leachate concentration to extract concentrations from the teach tests. Equation 10: Sell Screening Level Partitioning; Equation for Migration to Ground Water Screening Level in Sol (mo/kg) ft Parameter/Definition (units) soil leachate concentration (mg/L) rV*o8-water partition coefficient (LAg) Koc/soH organic carbon/water partition coefficient (LAg) /fraction organic carbon in soil (gig) ew/water-filled soil porosity 6,/air-fyied soil pb/dry soil bulk density (kg/L) n/soil porosity particle density (kg/i) H'/dmensionless Henry's law constant Default nonzero MCLG. MCL.orHBL«x dilution factor chemical-specific* x foe (organic*) chemical-specific^ 0.002 (0.2%) 0.3 1.8 1-(pb/Ps) 2.65 chemical-specific*1 (assume to be zero for inorganic con- taminants except mercury) •Chemical-specific (see Attachment D). »See Attachment C. Soil/Water Partition Equation. The soil/water partition equation (Equation 10) relates concentrations of contaminants adsorbed to soil organic carbon to soil leachate concentrations in the zone of contamination. It calculates SSLs corresponding to target soil leachate contaminant concentrations (C*). An adjustment has been added to the equation to relate sorbed concentration in soil to the measured total soil concentration. This adjustment assumes that soil-water, solids, and gas are conserved during sampling. If soil gas is lost during sampling, 6. should be assumed to be zero. Likewise, for inorganic contaminants except 29 TUl OOl 1O73 mercury, there is no significant vapor pressure and H' may be assumed to be zero. waste disposal scenarios. Consult these documents for further information. The use of the soil/water partition equation to calculate SSLs assumes an infinite source of contaminants extending to the top of the aquifer. More detailed models may be used to calculate higher SSLs that are still protective in some situations. For example, contaminants at sites with. shallow sources, thick unsaturated zones, degradable contaminants, or unsaturated zone characteristics (e.g., clay layers) may attenuate before they reach ground water. The TBD provides information on the use of unsaturated zone models for soil screening. The decision to use such models should be based on balancing the additional investigative and modeling costs required to apply the more complex models against the cost savings that will result from higher SSLs. Leach Test A leach test may be used instead of the soil/water partition equation. In some instances, a leach test may be more useful than the partitioning method, depending on the constituents of concern and the possible presence of RCRA wastes. If this option is chosen, soil parameters are not needed for this pathway. However, a dilution factor must still '•be calculated. This guidance suggests using the EPA Synthetic Precipitation Leaching Procedure (SPLP, EPA SW-846 Method 1312, U.S. EPA, 1994d). The SPLP was developed to model an acid rain leaching environment and is generally appropriate for a contaminated soil scenario. Like most leach tests, the SPLP may not be appropriate for all situations (e.g., soils contaminated with oily constituents may not yield suitable results). Therefore, apply the SPLP with discretion. EPA is aware that many leach tests are available for application at hazardous waste sites, some of which may be appropriate in specific situations (e.g., the Toxicity Characteristic Leaching Procedure (TCLP) models leaching in a municipal landfill environment). It is beyond the scope of this document to discuss in detail leaching procedures and the appropriateness of their use. Stabilization/Solidification of CERCLA and RCRA Wastes (U.S. EPA, 1989b) and the EPA SAB's review of leaching tests (U.S. EPA, 1991b) discuss the application of various leach tests to various See Step 3 for guidance on collecting subsurface soil samples mat can be used for leach tests. To ensure adequate precision of leach test results, leach tests should be conducted in triplicate. Dilution Factor Model. As soil leachate moves through soil and ground water, contaminant concentrations are attenuated by adsorption and degradation. In the aquifer, dilution by clean ground water further reduces concentrations before contaminants reach receptor points (i.e., drinking water wells). This reduction in concentration can be expressed by a dilution attenuation factor (DAF), defined as the ratio of soil leachate concentration to receptor point concentration. The lowest possible DAF is 1, corresponding to the situation where there is no dilution or attenuation of a Contaminant (i.e., when the concentration in the receptor well is equal to the soil leachate concentration). On the other hand, high DAF values correspond to a large reduction in contaminant concentration from the contaminated soil to the receptor well. The Soil Screening Guidance addresses only one of these dilution-attenuation processes: contaminant dilution in ground water. A simple mixing zone equation derived from a water-balance relationship (Equation 11) is used to calculate a site-specific dilution factor. Mixing-zone depth is estimated from Equation 12, which relates it to aquifer thickness along with the other parameters from Equation 11. Mixing zone depth should not exceed aquifer thickness (i.e., use aquifer thickness as the upper limit for mixing zone depth). Because of the uncertainty resulting from the wide variability in subsurface conditions that affect contaminant migration in ground water, defaults are not provided for the dilution model equations. Instead, a default DAF. of 20 has been selected as protective for contaminated soil sources up to 0.5 acre in size. Analyses using the mass-limit models described below suggest mat a DAF of 20 may be protective of larger sources as well; however, this hypothesis should be evaluated on a site-specific basis. A discussion of the basis for the default DAF and a description of the mass-limit analysis is found in the TBD. However, since migration to ground 30 TUT COS .1074 water SSLs are most sensitive to the DAF, site- nfic dilution factors should be calculated. .^specific Equation 11: Derivation of Dilution Factor dilution f actor =1+ Kid «- Parameter/Definition (unit*) dilution factor (unities*) K/aqulfer hydraulic conductivity (m/yr) I/hydraulic gradient (m/m) I/infiltration rate (m/yr) d/mixing zone depth (m) L/aource length parallel to ground water flow (m) Default 20 (O.S-acre aource) Equation 12: Estimation of Mixing Zone Depth d * (0.0112 L2)0-5 + d. {1 - Parameter/Definition (units) d/mixing zone depth (m) L/source length parallel to ground water flow (m) JKinfiltration rate (m/yr) f 'aquifer hydraulic conductivity (m/yr) I i/hydraulic gradient (m/m) da/aquifer thickness (m) Mass-Limit SSLs. Use of infinite source models to estimate volatilization and migration to ground water can violate mass balance considerations, especially for small sources. To address mis concern, the Soil Screening Guidance includes models for calculating mass-limit SSLs for each of these pathways (Equations 13 and 14) that provide a lower limit to SSLs when the area and depth (i.e., volume) of the source are known or can be estimated reliably. A mass-limit SSL represents the level of contaminant in the subsurface that is still protective when the entire volume of contamination either volatilizes or leaches over the 30-year exposure duration and the level of contaminant at the receptor does not exceed the health-based limit. To use mass-limit SSLs, determine the area and depth of the source, calculate both standard and mass-limit SSLs, compare mem for each chemical of concern and select the higher of the two "dues. Analyze the inhalation and migration to ground . water pathways separately. Equation 13: Mass-Limit Volatilization Factor VF » OK? x [T x (3.15 x 1Q7afrr) ] (pbxd.x10«gWg) Parameter/Definition (units) d, /average source depth. (m) T/exposure interval(s) Q/C/invers* of mean cone, at center of a square source (g/m2-s per kg/m3) pb/dry soil bulk density (kg/L or Mg/m>) Default site-specific 9.5 x 10" 68.81 1.5 Equation 14: Mass-Limit Soil Screening Level for Migration to Ground Water Screening Level in Soil = • (CMxIx-ED) (npfcB) ftxd, Parameter/Definition (units) Cg/target soil toachate concentration (mg/L)- ds/depth of source (m) I/infiltration rate (m/yr) ED/exposure duration (yr) pb/dry soil bulk density (kg/L) Default (nonzero MCLG, MCI_orHBL)«x dilution factor site-specific 0.18 70 1.8 •Chemical-specific, see Attachment D. Note that Equations 13 and 14 require a site-specific determination of the average depth of contamination in the source. Step 3 provides guidance for conducting subsurface sampling to determine source depth. Where the actual average depth of contamination is uncertain, a conservative estimate should be used (e.g., the maximum possible depth in the unsaturated zone). At many sites, the average water table depth may be used unless there is reason to believe that contamination extends below the water table. In this case SSLs do not apply and 31 TUT COS 1075 further investigation of the source in question is n^-ied. Plant Uptake. Consumption of garden fruits and vegetables grown in contaminated residential soils can result in a risk to human health. This exposure pathway applies to both surface and subsurface soils. The TBD includes an evaluation of the soil-plant- human pathway along with a discussion of the site- specific factors that influence plant uptake and plant contamination concentration. Generic screening levels are calculated for arsenic, cadmium, mercury, nickel, selenium, and zinc based on empirical data on the uptake (i.e., bioconcentration) of these inorganics into plants. In addition, levels of inorganics that have been reported to cause phytotoxicity (Will and Suter, 1994) are presented. Organic compounds are not addressed due to lack of empirical data. The empirical data indicate that site-specific factors such as soil type, pH, plant type, and chemical form strongly influence the uptake of metals into plants. Where site conditions allow for the mobility and bioavailability of metals, the results of our generic analysis suggest that the soil-plant-human pathway may be of particular concern for sites with soils contaminated with cadmium and arsenic. However, the phytotoxicity of certain metals may limit the amount that can be bioconcentrated in plant tissues. The data on phytotoxicity suggest that, with the exception of arsenic, metal concentrations in soil that are considered toxic to plants are well below the levels that may impact human health through the soil-plant-human pathway. This implies that phytotoxic effects may prevent completion of this pathway for these metals. However, like plant uptake, phytotoxicity is also greatly influenced by the site-specific factors mentioned above. Thus, it is necessary to evaluate on a site-specific basis, the potential bioavailability of certain inorganics for the soil-plant-human pathway and the potential for phytotoxic effects in order to assess possible human health and ecological impacts through plant uptake. 2.5.3 Address Exposure to Multiple Chemicals. The SSLs generally correspond to a 10-6 risk level for carcinogens and a hazard quotient of 1 for noncarcinogens. This 'target" hazard quotient is used to calculate a soil concentration below which it is unlikely that sensitive populations will experience adverse health effects. The potential for additive effects has not been "built in" to the SSLs through apportionment. For carcinogens, EPA believes mat setting a 10-6 risk level for individual chemicals and pathways generally will lead to cumulative site risks within the 10-* to 10-6 n^ range for the combinations of chemicals typically found at NPL sites. For noncarcinogens, mere is no widely accepted risk range, and EPA recognizes that cumulative risks from noncarcinogenic contaminants at a site could exceed the target hazard quotient. However, EPA also recognizes that noncancer risks should be added only for those chemicals with the same toxic endpoint or mechanism of action. Ideally, chemicals would be grouped according to their exact mechanism of action, and effect-specific toxicity criteria would be available for chemicals exhibiting multiple effects. Instead, .data are often limited to gross lexicological effects in an organ (e.g., increased liver weight) or an entire organ system (e.g., neurotoxicity), and RfDs/reference concentrations (RfCs) are available for just one of the several possible endpoints of toxicity for a chemical. Given the currently available criteria, noncarcinogenic contaminants should be grouped according to the critical effect listed as the basis for the RfD/RfC. If more man one chemical detected at a site affects the same target organ/system, SSLs for those chemicals should be divided by the number of chemicals present in the group. Exhibit 13 lists several chemicals with noncarcinogenic affects in the same target organ/system. However, the list is limited, and a lexicologist should be consulted prior to using SSLs on a site-specific basis. If additive risks are being considered in developing site-specific SSLs for subsurface soils, recognize that, for certain chemicals, SSLs may be based on a "ceiling limit" concentration (Cstt) instead of toxicity. Because they are not risk-based, CHt*based SSLs should not be modified to account for additivhy. 32 TUT 008 1076 2.6 Step 6: Comparing Site Soil Contminant Concentrations to Calculated SSLs Now that the she-specific SSLs have been calculated for the potential contaminants of concern, compare them with the site contaminant concentrations. At this point, it is reasonable to review the CSM with the actual site data to confirm its accuracy and the overall applicability of the Soil Screening Guidance. In theory*; an exposure area would be screened from further investigation when the true mean of the population of contaminant concentrations falls below the established screening level. However, EPA recognizes that data obtained from sampling and analysis are never perfectly representative and accurate, and that the cost of trying to achieve perfect results would 'be quite high. Consequently, EPA acknowledges that some uncertainty in data must be tolerated, and focuses on controlling the uncertainty which affects decisions based on those data. Thus, in the Soil Screening Guidance, EPA has developed an approach for surface soils to minimize the chance of incorrectly deciding to: Screen out areas when the correct decision would be to investigate further (Type I error); or Decide to investigate further when the correct decision would be to screen out the area (Type II error). The approach sets limits on the probabilities of making such decision errors, and acknowledges that there is a range (i.e., gray region) of contaminant levels around the screening level where the variability in the data will make it difficult to determine whether the exposure area average concentration is actually above or below the screening level. The Type I and Type II decision error rates have been set at 5 percent and 20 percent, respectively, and the gray region has been set between one-half and two times the SSL. By specifying the upper edge of the gray region as twice the SSL, it is possible that exposure areas with mean contaminant concentration values slightly above the SSL may be screened from further study. Commenters have expressed concern that this is not adequately protective for SSLs based on noncarcinogenic effects. However, EPA believes that the approaches taken in this guidance to address chronic exposure to noncarcinogens are conservative enough for the majority of site contaminants (i.e., comparison of the 6 year "childhood only" exposure to the chronic RID); and, use of maximum composite concentrations provide high coverage of the true population mean (i.e., there is high probability mat the value equals or exceeds the true population mean). Thus, for surface soils, the contaminant concentrations in each composite sample from an exposure area are compared to two times the SSL. Under the Soil Screening Guidance DQOs, areas are screened out from further study when contaminant concentrations in all of the composite samples are less than two times the SSLs. Use of this decision rule (comparing contaminant concentrations to twice the SSL) is appropriate only when the quantity and quality of data are comparable to the levels discussed in this guidance, and the toxicity of the chemical has been evaluated against the criteria presented in Section 2.5.1. For existing data sets mat may be more limited than those discussed in this guidance, the 95 percent upper-confidence limit on the arithmetic mean of contaminant concentrations in surface soils (i.e., the Land method as described in the Supplemental Guidance to RAGS: Calculating the Concentration Term (U.S. EPA, 1992d) should be used for comparison to the SSLs. The TBD discusses the strengths and weaknesses of using the Land method for making screening decisions. 33 TUT COS 1O77 Exhibit 13: SSL Chemicals with Noncarcinogeriic Toxic Effects on Specific Target Orr"»n/Sy«tem -•get Organ/System Effect Kidney . Acetone 1,1-Dichloroethane Cadmium Chlorobenzene Di-n-octyl phthalate Endosulfan Ethylbenzene Ruoranthene Nitrobenzene Pyrene Toluene 2,4,5-Trichlorophenol Vinyl acetate Liver Acenaphthene Acetone Butyl benzyl phthalate . Chlorobenzene Di-n-octyl phthalate Endrin Ethylbenzene Flouranthene . Nitrobenzene Styrene Toluene 2,4,5-Trichlorophenol Central Nervous System Butanol Cyanide (amenable) 2,4 Dimethylphenol Endrin 2-Methylphenol Mercury Styrene Xylenes Adrenal Gland Nitrobenzene 1,2,4-Trichlorobenzene Increased weight; nephrotoxicity Kidney damage Significant proteinuria Kidney effects Kidney effects Glomerulonephrosis Kidney toxicity Nephropathy Renal and adrenal lesions Kidney effects Changes in kidney weights Pathology Altered kidney weight Hepatotoxicity • Increased weight Increased liver-to-body weight and liveMo-brain weight ratios Histopathology Increased weight; increased SGOT and SGPT activity Mild histotogfcal lesions in liver Liver toxicity Increased Kver weight Lesions Liver effects Changes in Kver weights Pathology Hypoactivrty and ataxia Weight loss, myeftn degeneration Prostatration and ataxia Occasional convulsions Neurotoxicity Hand tremor, memory disturbances Neurotoxicity Hyperactivfty Adrenal lesions Increased adrenal weights; vacuolization in cortex 34 TUT 008 1O78 C" Exhibit 13: (continued) Target Organ/System Effect r Circulatory Syetem Antimony Barium tr*n*-1,2-Dichloroethene ew-1,2-Dtchloroethyl»ne 2.4-Dimethyiph«nol Fluoranthene Fluorene Nitrobenzene Styrene Zinc Reproductive Syetem Barium Carbon disuHide 2-Chlorophonol Methoxychlor Phenol Respiratory Syetem 1,2-Dichloropropane Hexachtorocyciopentadiene Methyl bromide' Vinyl acetate Gastrointestinal System Hexachiorocyclopentadiene Methyl bromide Immune System 2,4-Dichlorophenol p-Chloroaniline Altered blood chemistry and myocardial effects Increased Wood pressure • • Increased alkalne phosphatase level Decreased hematocrit and hemoglobin Altered blood chemistry Hematotogfc changes Decreased BBC and hemoglobin I ierikatulogic changes Red blood cell effects Decrease in erythrocyte superoxide dismutase (ESOO) Fetotoxicity Fetal toxicity and malformations Reproductive effects Excessive loss of litters. Reduced fetal body weight in rats Hyperplasia of the nasal mucosa Squamous metaplasia Lesions on the olfactory epithelium of the nasal cavity Nasal epithelial lesions Stomach lesions Epithelial hyperplasia of the forestomach Altered immune function Nonneopiastic lesions of splenic capsule Source: U.S. EPA. 1995b. r 35 TUT COS 1079 In this guidance^ fewer samples are collected for subsurface soils than for surface soils; therefore, different decision rules apply. Since subsurface soils are not characterized as well, there is less confidence that the concentrations measured are representative of the entire source. Thus, a more conservative approach to screening is warranted. Because it may not be protective to allow for comparison to values above the SSL, mean contaminant concentrations from each soil boring taken in a source area are compared with the calculated SSLs. Source areas with any mean soil boring contaminant concentration greater than the SSLs generally warrant further consideration. On the other hand, where.the mean soil boring contaminant concentrations within a source are all less than the SSLs, that source area is generally screened out. 2.7 Step 7: Addressing Areas Identified for Further Study The chemicals, exposure pathways, and areas that have been identified for further study become a subject of the RI/FS. The results of the baseline risk assessment conducted as part of the RI/FS will establish the basis for taking remedial action. The threshold for taking action differs from the criteria used for screening. As outlined in Role of the Baseline Risk Assessment in Superfund Remedy Selection Decisions (U.S. EPA, 1991d), remedial action at NPL sites is generally warranted where cumulative risks for current or future land use exceed IxlO- 4 for carcinogens or a HQ of 1 for noncarcinogens. The data collected for soil screening are useful in the RI and baseline risk assessment. However, additional data will probably need to be collected during future site investigations. Once the decision has been made to initiate remedial action, the SSLs can then serve as preliminary remediation goals. This process is referenced in Section 1.2 of this document. FOR FURTHER INFORMATION More detailed discussions of the technical background and assumptions supporting the development of the Soil Screening Guidance are presented in the Soil Screening Guidance: Technical Background Document (U.S. EPA, 1996). For additional copies of this guidance document, the Technical Background Document, or other EPA documents, call the National Technical Information Service (NT1S) at (703) 487-4650 or 1-800-553- NTIS (6847). 36 TUT OO8 1O80 REFERENCES Aller, L., T. Bennett, J.H. Lehr, R.J. Petty, and G. Hackett. 1987. DRASTIC: A Standardized System for Evaluating Ground Water Pollution Potential Using Hydrogeologic Settings. Prepared for U.S. EPA, Office of Research and • Development, Ada, OK. National Water Well Association, Dublin, OH. EPA-600/2-87-035. ' American Society for Testing Materials. 1995. Annual Book of ASTM Standards. Volume 4.08. Soil and Rock. Building Stones. Philadelphia, PA. Brady, N.C. 1990. The Nature and Properties of £0//j.Macmillan Publishing Company, New York, NY. Calabrese, E.I., H. Pastides, R. Bames, et al. 1989. How much soil do young children ingest: an epidemiologic study. Petroleum Contaminated Soils, Vol. 2. E.J. Calabrese and P.T. Kostecki, eds. pp. 363-417. Lewis Publishers, Chelsea, MI. wherd, C., G. Muleski, P. Engelhart, and D. Gillette. 1985. Rapid Assessment of Exposure to Paniculate Emissions from Surface Contamination. Prepared for U.S. EPA, Office of Health and Environmental Assessment, Washington, DC. EPA/600/8-85/002. NTIS PB85-192219 7AS. Davis, S., P. Waller, R. Buschom, J. Ballou, and P. White. 1990 Quantitative estimates of soil ingestion in normal children between the ages of 2 and 7 years: population-based estimates using Al, Si, and Ti as soil tracer elements. Arch. Environ. Health 45:112-122. Dragun, J. 1988. The Soil Chemistry of Hazardous Materials. Hazardous Materials Control Research Institute, Silver Spring, MD. EQ (Environmental. Quality Management). 1993. Evaluation of the Dispersion Equations in the Risk Assessment Guidance for Svperfund (RAGS): Volume I— Human Health Evaluation Manual (Part B, Development of Preliminary ^^Remediation Goals),. Contract No. 68-02- D120. Prepared for Office of Emergency and Remedial Response, Washington, DC. Gee, G.W., and J.W. Bander. 1986. Particle size analysis. A. Klute (ed.), Methods of Soil 'Analysis. Part 1. Physical and Mineralogical Methods. 2nd Edition, 9(1):383-411, American Society of Agronomy, Madison, WI. Hwang, S. T., and J. W. Falco. 1986. Estimation of Multimedia Exposure Related to Hazardous Waste Facilities. Y. Cohen, Ed. Plenum Publishing. Johnson, P.C. and RA. Ettinger. 1991. Heuristic model for predicting the intrusion rate of contaminant vapors into buildings. Environmental Science and Technology 25(8): 1445-1452. Jury, W.A., W.J. Farmer, and W.F. Spencer. 1984. Behavior assessment model for trace organics in soil: II. Chemical classification and .parameter sensitivity. J. Environ. Qua I. 13(4):567-572. McLean, E.O. 1982. Soil pH and lime requirement. In: A.L. Page (ed.), Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. 2nd Edition, 9(2): 199-224, American Society of Agronomy, Madison, WI. 'Nelson, D.W., and L.E. Sommers. 1982. Total carbon, organic carbon, and organic matter. In: A.L. Page (ed.), Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. 2nd Edition, 9(2):539-579, American Society of Agronomy, Madison, WI. » . Newell, C.J., L.P. Hopkins, and P.B. Bedient. 1990. A hydrogeologic database for ground water modeling. Ground Water 28(5):703-714. Schroeder, P.R., A.C. Gibson, and M.D. Smolen. 1984. Hydrological Evaluation of Landfill Performance (HELP) Model; Volume 2: Documentation for Version I. EPA/530-SW- 84-010. Office of Research and Development, 37 COS 1OS1 U.S. EPA, Cincinnati, OH. 100832. NTIS PB85- U.S. EPA. 1980. Land D^posal of Hexachloro- benzene Wastes Controlling Vapor Movement in Soil. EPA-600/2-80-119 Office of Research and Development, Cincinnati, OH. NTIS PB80-216575. U.S. EPA. 1988a. Guidance on Remedial Actions for Contaminated Ground Water at Superfund Sites. Office of Emergency and Remedial Response, Washington, DC. Directive 9283.1- 2. EPA/540/G-88/003. NTIS PB89- 184618/CCE U.S. EPA. 1988b. Superfund Exposure^Assessment Manual. OSWER Directive 9285.5-l.Office of Emergency and Remedial Response, Washington, DC. EPA/540/1-88/001. NTIS PB89-135859. U.S. EPA. 1989a. Risk Assessment Guidance for Superfund (RAGS): Volume I: Human Health Evaluation Manual (HHEM), Part A, Interim Final. Office of Emergency and Remedial Response, Washington, DC. EPA/540/1 - 89/002. NTIS PB90-155581/CCE. U.S. EPA. 1989b. Stabilization/Solidification of CERCLA and RCRA Wastes. EPA/625/6- 89/022. U.S. EPA. 1989c. Guidance'for Conducting Remedial Investigations and Feasibility Studies under CERCLA. Office of Emergency and Remedial Response, Washington, DC. EPA/540/G-89/004. OSWER Directive 9355.3- 01. NTIS PB89-184626. U.S. EPA. 1990a. Suggested ROD Language for Various Ground Water Remediation Options. Office of Emergency and Remedial Response, Washington, DC. Directive 9283.1-03. NTIS PB91-921325/CCE. U.S. EPA. 1990b. Sampler's Guide to the Contact Laboratory Program. Office of Emergency and Remedial Response, Washington, DC. NTIS PB91-92133OCDH. U.S. EPA 1990c. A Rationale for the Assessment of Errors in the Sampling of Soils. Environmental Monitoring Systems Laboratory, Office of Research and Development, Las Vegas, NV. EP A/600/4- 90/013. NTIS PB90-242306. U.S. EPA. 1991a. Human Health Evaluation Manual (HHEM), Supplemental Guidance: Standard Default Exposure Factors. Office of Emergency and Remedial Response, Washington, DC. Publication 9285.6-03. NTIS PB91-921314. U.S. EPA. 1991b. Leachability Phenomena. Recommendations and Rationale for Analysis of Contaminant Release by the Environmental Engineering Committee. Science Advisory Board, Washington, DC. EPA-SAB-EEC-92- 003. U.S. EPA. 1991c. Risk Assessment Guidance for Superfund (RAGS), Volume 1: Human Health Evaluation Manual (HHEM). Part B, Development of Risk-Based Preliminary Remediation Goals. Office of Emergency and Remedial Response, Washington, DC. Publication 9285.7-01B. NTIS PB92-963333. U.S. EPA. 1991d. Role of the Baseline Risk Assessment in Superfund Remedy Selection Decisions. Office of Emergency and Remedial Response, Washington, DC. Publication 9355.0-30. jmS PB91-9213597CCE. U.S. EPA. 1991e. User's Guide to the Contract Laboratory Program. Office of Emergency and Remedial Response, Washington, DC. NTIS PB91-921278CDH. U.S. EPA, 1991f. Description and Sampling of Contaminated Soils: A Field Pocket • Guide . Office of Environmental Research Information, Cincinnati, OH. EP A/625/12- 91/002. U.S. EPA. 1992a, Considerations in Ground- Water Remediation at Superfund Sites and RCRA Facilities— Update. Office of Emergency and Remedial Response, Washington. DC. Directive 9283.1-06. NTIS PB91- 238584/CCE. U.S. EPA 1992b. Dermal Exposure Assessment: Principles and Applications. Interim Report. Office of Research and Development, Cincinnati, OH. EPA/600/8-91/011B. 38 008 1082 U.S. EPA. 1992c. Estimating Potential for Occurrence of DNAPL at Superfund Sites, • Office of Emergency and Remedial Response, Washington, DC. Publication 9355.4-07FS. NTIS PB92-963338. U.S. EPA. 1992d. Supplemental Guidance to RAGS; Calculating the Concentration Term. Volume 1, Number 1, Office of Emergency and Remedial Response, Washington, DC. NTIS PE92-963373 U-S. EPA, 1992e. Preparation of Soil Sampling Protocols: Sampling Techniques' and Strategies. Office of Research and Development, Washington, DC. EPA/600/R- 92/128. U.S. EPA. I993a. Data Quality Objectives for Super fund: Interim Final Guidance. Publication 9255.9-01. Office of Emergency and Remedial Response, Washington, DC. EPA 540-R-93-071. NTIS PB94-963203. U.S. EPA. 1993b. Guidance for Evaluating Technical Impracticability of Ground-Water - Restoration: EPA/540-R-93-080. Office of Emergency and Remedial Response, Washington, DC. Directive 9234.2-25. J. EPA. 1993c. Quality Assurance for Superfund Environmental Data Collection Activities. Quick Reference Fact Sheet. Office of Emergency and Remedial Response, Washington, DC. NTIS PB93-963273. U.S. EPA 1993d. The Urban Soil Lead Abatement Demonstration Project. Vol I: Integrated Report Review Draft. National Center for Environmental Publications and Information. EPA 600/AP93Q01/A. NTIS PB93-222-651. U.S. EPA, 1993e. Subsurface Characterization and Monitoring Techniques: A Desk Reference Guide. Vol. I & 11 . Environmental Monitoring Systems Laboratory, Office of Research and Development, Las Vegas, NV. EPA/625/R- 93/003a. U.S. EPA, 1993f. Risk Assessment Guidance for Superfitnd (RAGS), Human Health Evaluation ." Manual (HHEM). Science Advisory Board Review of the Office of Solid Waste and Emergency Response draft. Washington, DC. EPA-SAB-EHC-93-007. U.S. EPA. 1994a. Guidance for the Data Quality Objectives Process. Quality Assurance Management Staff, Office of Research and Development, Washington, DC. EPA QA/G-4. U.S. EPA. 1994c. Methods for Evaluating the Attainment of Cleanup Standards—Volume 3: Reference-Based Standards for Soils and Solid Media. Environmental Statistics and Information Division, Office of Policy, Planning, and Evaluation, Washington,' DC. EPA 230-R-94-004. U.S. EPA. 1994d. test Methods for Evaluating Solid Waste. Physical/Chemical Methods (SW- 846), Third Edition, Revision 2. Washington, DC. U.S. EPA. 1995a. Health Effects Assessment Summary Tables (HEAST). Annual Update, FYI993! Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Office of Research and Development, Cincinnati, OH. U.S. EPA. 1995b. Integrated Risk Information System (IRIS). Cincinnati, OH. U.S. EPA. 1995c. Drinking Water Regulations and Health Advisories. Office of Water, Washington, DC. U.S. EPA. 1996. Soil Screening Guidance: Technical Background Document. Office of Emergency and Remedial Response, Washington, DC. EPA/540/R95/128. Van Wijnen, J.H., P. Clausing, and B. Brunekieef. 1990. Estimated soil ingestion by children. Environ Research 51:147-162. Wester et al. 1993. Percutaneous absorption of • pentachlorophenol from .soil. Fundamentals of Applied Toxicology, 20. Will, M.E. and G.W. Suter n. 1994. Toxicological Benchmarks for Screening Potential Contaminants 'of Concern for Effects on Terrestrial Plants, 1994 Revision. ES/ER/TM- g/Rl. Prepared for the U.S. Department of Energy by the Environmental Sciences Division of Oak Ridge National Laboratory. 39 TUT 008 1O83 Attachment A Conceptual Site Model Summary TIJT 008 1084 Attachment A * . , Conceptual Site Model Summary Step 1 of the Soil Screening Guidance: User's Guide describes the development of a conceptual site model (CSM) to support the application of soil screening levels (SSLs) at a she. The CSM summary forms at the end of mis attachment contain the information necessary to: • Determine the applicability of SSLs to the site SSLs. v ' By identifying data gaps, these summary forms will help focus data collection and evaluation on the site-specific development and application of SSLs. The site investigator should use the summary forms during the SSL sampling effort to, collect site-specific data and continually update the CSM with new information as appropriate. The CSM summary forms indicate the information required for determining the applicability of the soil screening process to the she. Forms addressing source characteristics may be photocopied if more than one source is present at a she. A site map showing contaminated soil sources and exposure areas (EAs) should be attached to the summary. If available, additional pages of other maps, summaries of analytical results, or more detailed descriptions of the she may be attached to the summary. Form 1. General Site Information The information included in this form is identical to the first page of the Site Inspection (SI) Data Summary form (page B-3 in Guidance for Performing Site Inspections Under CERCLA, U.S. EPA, 1992). However, the form should be updated to reflect any she activities conducted since the SI was completed. Form 2. Site Characteristics Form 2 indicates the information necessary to address the migration to ground water pathway and identify subsurface conditions that may limit the applicability of subsurface SSLs. A hydrogeologic setting is defined as a unit with common hydrogeologic characteristics and therefore common vulnerability to contamination. Each setting provides a composite description of the hydrogeologic factors that control ground water movement and recharge. These factors can be used to make generalizations in the CSM about ground water conditions. After placing the she into one of Heath's ground water regions (Heath, 1984), consider geologic and geomorphic features of the she and select a generic hydrogeologic setting from Aller et al. (1987) that is most similar to the she. If existing site information is not sufficient to definitively place the she in a setting, h should be possible to narrow the choice to two or three settings that will reduce the range of values necessary to develop SSLs. A copy of the setting diagram from Aller et al. (1987) should be attached to the CSM checklist to provide a general picture of subsurface she conditions. Ground Water Flow Direction. The direction of ground water flow in the uppermost aquifer underlying each source is needed to determine source length parallel to that flow. If ground water flow direction is unknown or uncertain, assume h is parallel to the longest source dimension. A-l TUT OO8 1085 Aquifer Parameters. Aquifer parameters ne^ed to estimate a she-specific dilution factor include hydraulic conductivity (K), hydraulic gradient (i), and aquifer thickness (d J. She-measured values for these parameters are the preferred alternative. Existing site documentation should be reviewed for in situ measurements of aquifer conductivity (i.e., from pump test data), water table maps that can be used to estimate hydraulic gradient, and boring .logs that indicate the thickness of the uppermost aquifer. Detailed information on conducting and interpreting aquifer tests can be found in Nielsen (1991). If she-measured values are not available, hydrogeologic knowledge of regional geologic conditions or measured values in the literature-may be sources of reasonable estimates. Values from a similar site in the same region and hydrogeologic setting also may be used, but must be carefully reviewed to ensure thaUhe subsurface conceptual models for the two sites show reasonable agreement For all of these options, h is critical mat the estimates and sources be reviewed by an experienced hydrogeologist knowledgeable of regional hydrogeologic conditions. A third option is to obtain parameter estimates for the site's hydrogeologic setting from Aller et al. (1987) or from the American Petroleum Institute's (API's) hydrogeologic database (HGDB) (Newell et al., 1989, 1990). Aller et al. (1987) present ranges of values for K and i by hydrogeologic setting, The HGDB contains measured values for these parameters and .aquifer depth for a number of sites in each hydrogeologic setting. If HGDB data are used, the median value presented for each setting should be used unless she-specific conditions indicate otherwise. Aquifer parameter values from these sources also can serve as a check of the validity of she-measured values or estimates obtained from other sources. If outside sources such as Aller et al. (1987) are used to characterize site hydrogeologic conditions, the appropriate references and diagrams should be attached to the CSM checklist. Infiltration Rate. Infiltration rate is used to calculate SSLs for subsurface soils (see Step 5). The simplest way to estimate infiltration rate (I) is to assume that infiltration is equal to recharge and obtain recharge estimates for the she's hydrogeologic setting from Aller et al. (1987). When using the Aller et al. (1987) estimates the user should recognize that these are estimates of average recharge conditions throughout the setting and she-specific values may differ to some extent. For example, areas within the setting whh steeper than average slopes will tend to have lower infiltration rates and areas whh flatter man average slopes will tend to have higher infiltration man average. An alternative is to use infiltration rates determined for a better-characterized site in the same hydrogeologic setting and whh similar meteorological conditions as the she in question. A third alternative is use the HELP model. Although HELP was originally written for hydrologic evaluation of landfills (Schroeder et al., 1984), inputs to the HELP program can be modified to estimate infiltration in undisturbed soils in natural settings. The most recent version of HELP and the most recent user's guide and documentation can be obtained by sending an address and two double- sided, high-density, DOS-formatted disks to: attn. Eunice Burk U.S EPA 5995 Center Hill Ave. Cincinnati, OH 45224 (513) 569-7871. A-2 TUT 008 Meteorologic Parameters. Select a site-specific Q/C value from in the guidance for the volatilization actor (VF) equation or paniculate emission actor (PEF) equation to place the site in a climatic zone (Figure A-1). Several site-specific parameters are required to calculate a PEF if fugitive dusts are of concern at the she (see Step 5 for surface soils). The threshold windspeed at 7 meters above ground surface (U^) is calculated from source area roughness height and the mode soil aggregate size as described in Cowherd et al. (1985). Mode soil aggregate size refers to the mode diameter of aggregated, soil particles measured under field conditions. Other she-specific variables necessary for calculating the PEF include fraction vegetative cover (V) and the mean annual windspeed (UB). Fraction vegetative cover is estimated by visual observations of the surface of known or suspected source areas at the she. Mean annual windspeed can be obtained from the National Weather Service surface station nearest to the she. Form 3. Exposure Pathways and Receptors Form 3 includes information necessary to determine the applicability of the Soil Screening Guidance to a site (see Step 2 of the User's Guide). This form summarizes the she information necessary to identify and characterize potential exposure pathways and receptors at the site, such as site conditions, relevant exposure scenarios, and the properties of soil contaminants listed on Form 4. Table A-1 provides an example of exposure pathways that are not addressed by the guidance, but have relevance to CSM development. Table A-1. Example Identification of Exposure Pathways Not Addressed by SSLs Receptor*/ Exposure Pathways Contaminant Characteristic* Site Condition* Human /Direct Pathways ingestion (acute exposure) inhalation • fugitive dusts (acute exposure) acute health effects (e.g., cyanide, phenol) acute health effects residential setting high fugitive dusts (e.g., from soil tillage, heavy traffic on dirt roads; construction) Human / Indirect Pathways consumption of meat or dairy products •fish consumption bioaccumulation, biomagnification biomagnification nearby meat or dairy production nearby surface waters with recreational or subsistence fishing Ecological Pathways aquatic terrestrial aquatic toxicity toxicity to terrestrial organisms (e.g., DOT, Hg) nearby surface waters or wetlands sensitive species on or near Site A-3 TUT DOS 1087 Figure A-1. U.S. climatic zones A-4 TUT OO8 1O83 Form 4. Soil Contaminant Source Characteristics This fonn prompts the investigator to provide information on source characteristics, including soil contaminant levels and the physical and chemical parameters of site soils needed to calculate SSLs. One form should be completed for each contaminated soil source. Initially, the form should be filled out to the greatest extent possible with existing site information collected during CSM development (see Step 1 of the User's Guide). The forms should be updated after the SSL sampling effort is . complete. Measurement of contaminant levels and the soil parameters listed on this form is described in Step 3 of mis guidance. Average soil moisture content (6W) defines the fraction of total soil porosity mat is filled by water and air. These parameters are necessary for determining the volatilization factor (VF) and the soil saturation limit (C,*) and to apply the soil/water partition equation. It is important that the moisture content used to calculate these parameters represent the annual average soil moisture conditions. Moisture content measurements on .discrete soil samples should not be used because they are affected by preceding rainfall events and thus may not represent average conditions. Volumetric average soil water content may be estimated by the following relationship developed by Clapp and Homberger (1978) and presented in the Superfund Exposure Assessment Manual (U.S. EPA, 1988): where n = total soil porosity I - infiltration rate (m/yr) K, — saturated hydraulic conductivity (m/yr) b — soil-specific exponential parameter (unitless). Total soil porosity (n) is estimated from dry soil bulk density <pb) as follows: n=l-(pb/p.) where pf = soil particle density = 2.65 kg/L. Values for K, and the exponential term l/(2b+3) are shown in Table. A-2 by soil texture class (soil class determination is discussed under Step 3). Site-specific values for infiltration rate (I) may be estimated using the HELP model or may be assumed to be equivalent to recharge (see Form 2). TUT' COS .1089 Table A-2. Paramatar Estimates for Calculating Average Soil Moisture Content (6W) Soil text"-* K.(m/yr) 1/(2b+3) Sand Loamy sand Sandy loam Silt loam Loam Sandy day loam SiK clay loam Clay loam Sandy clay Silt clay Clay 1,830 540 230 120 60 40 ' 13 20 10 8 5 0.090. 0.085 0.080 0.074 0.073 0.058 0.054 0.050 0.042 0.042 0.039 Source: U.S. ERA, 1988. Worksheets The worksheets following Forms 1 through 4 provide a convenient means of assembling chemical- specific parameters necessary to calculate SSLs for the contaminants of concern (Worksheet 1), existing site data on contaminant concentrations collected during CSM development or. the SSL sampling effort (Worksheet 2), and SSLs calculated for EAs (Worksheet 3) or contaminant sources (Worksheet 4) of concern at the site. CSM Diagram The CSM diagram is a product of CSM development that represents the linkages among contaminant sources, release mechanisms, exposure pathways and routes, and receptors to summarize the current understanding of the soil contamination problem (see Step 1 of the guidance). An example SSL CSM diagram, Figure A-2 (U.S. EPA, 1989), and a she sketch, Figure A-3 (U.S. EPA, 1987) are provided following the Worksheets. A-6 TUT COS 1090 References Aller, L., T. Bennett, J.H. Lehr, R.J. Petty, and G. Hackett. 1987. DRASTIC: A Standardized System for Evaluating Ground Water Pollution Potential Using Hydrogeologic &#mgs. Prepared for U.S. EPA Office of Research and Deve1' pment, Ada, OK. National Water Well Association, Dublin, OH. EPA-600/2-87-035. Clapp, R.B., and G.M. Homberger. 1978. Empirical equations for some soil hydraulic properties. Water Resources Research, 14:601-604. Cowherd, C, G. Muleski, P. Engelhart, and D. Gillette. 1985. Rapid Assessment of Exposure to Particulate Emissions from Surface Contamination. Prepared for Office of Health and Environmental Assessment, U.S. EPA, Washington, DC. NTIS PB85-192219 7AS. EPA/60Q/8-85/002. . Heath, R.C. 1984. Ground-Water Regions of the United States. USGS Water Supply Paper 2242. U.S. Geological Survey, Reston, VA. Newell, C.J., L.P. Hopkins, and P.B. Bedient 1989. Hydrogeologic Database for Ground Water Modeling. API Publication No. 4476. American Petroleum Institute, Washington, DC. Newell, C.J., L.P. Hopkins, and P.B. Bedient. 1990. A hydrogeologic database for ground water- modeling. Ground Water, 28(5):703-7-14. Nielsen, D.M. (ed.). 1991. Practical Handbook of Ground-Water Monitoring. Lewis Publishers, Chelsea, MI. ' ' Schroeder, P.R., A.C. Gibson, and M.D. Smolen. 1984. Hydrological Evaluation of Landfill Performance (HELP) Model; Volume 2: Documentation for Version 1. NTIS PB85-100832. Office of Research and Development, U.S. EPA, Cincinnati, OH. EPA/530-SW-84-010. U.S. EPA. 1987. Data Quality Objectives for Remedial Response Activities. Example Scenario: RI/FS Activities at a Site with Contaminated Soil and Groundwater. Office of Emergency and Remedial Response, Washington, DC. NTIS PB88-13188. U.S. EPA 1988, Superjund Exposure Assessment Manual. OSWER Directive 9285.5-1. Office of Emergency and Remedial Response, Washington, DC. EPA/540/1-88/001. NTIS PB89- 135859. U.S. EPA. 1989. Guidance for Conducting Remedial Investigations and Feasibility Studies under CERCLA. EPA/540/G-89/004. OSWER Directive 9355.3-01. Office of Emergency and Remedial Response, Washington, DC. NTIS PB89-184626. U.S. EPA. 1992. Guidance for Performing Site Inspections Under CERCLA. EPA/540-R-92-0021. Office of Emergency and Remedial Response, Washington, DC. NTIS PB92-963375. A-7 TUT OOS .1091 Soil Scr*«ning Guidance Conceptual Site Modal Summary Forms Form 1: Genaral Site Information ERA Regie- _______________\ Site Name. Date. Contractor Name and Address: State Contact: 1. CERCUSIDNo. V Address • County____ .City State ZpCode. Congressional District. 2. Owner Name Owner Address City_____ State .Operator Name _ .Operator Address. .City ______ State 3. Type of ownership (check all that apply): D Private O Federal Agency ____ Other _______________ O State O County Ref. __ D Municipal 4. Approximate size ol property acres Ref. 5. Latitude o_I_ . _" Longitude _ o_ I_ . _" Ref. 6. Site status D Active D Inactive D Unknown Ref. 7. Years of operation From_ To O Unknown Ref. 8. Previous investigations Type Agency/State/Contractor Date Ref. Ref. Ref. Ref. Ref. Ref. Ref. = reference(s) on information source A-8 TUT 008 j.092 Soil Screening Guidance Conceptual Site Modal Summary Forms Form 2: Site Characteristics Site Name —————— Hydreoeolooie Characteristics (migration to ground water pathway) is ground water of concern at the site? D yes D no (if no, move to infiltration RM« below). Heath region _________________ Hydrogeologie eettlng ______ (attach setting diagram) Check setting characteristics that apply: D karst D fractured rock O solution limestone Describe the stratigraphy and hydrogeologic characteristics of the. site. (Attach available maps and cross-sections.) Ref._ Identify and describe nearby sites in similar settings that have already been characterized. Ref._ Aquifer Parameters Unit Typical Min. Max. Reference or Source hydraulic conductivity (K) hydraulic gradient (i) thickness (d ) nVy nYm m General direction of ground water flow across the site (e.g., NNE, SW): (attach map.) Ref. '_____________ Infiltration rate . mfyr Method Meteorological Characteristics (inhalation pathway) climate-logical zone: ________________ (zone*, city) Q/C ___ tract, vegetative cover (V) _______;______ (unrtless) Reference mean annual windspeed (U_) ____________ m/s Reference _(g/m2-sperkg>m3> equivalent threshold value of windspeed at 7 m (Ut) fraction dependent on U,,/Ut ___________ Comments: __________________ .m/s .(unities*) A-9 TUT OOS 1093 Soil Screening Guidance Conceptual Sits Modal Summary Forms Form 3: Exposure Pathways and Racaptera Site Name • ________ . Land Use Conditions Current site use: Surrounding tend use: Future land use: _ residential _ residential _residential ._ industrial _ industrial _ industrial _commercial _commercial __ commercial _ agricultural _ agricultural __ agricultural _recreational __ recreational _recreational _other . _other _other Size of exposure areas (in acres) _________ Contaminant Release Mechanisms (cheek all that apply): Source #__ D leaching D volatilization O fugitive dusts O erosion/runoff D uptake by plants Source #_ D leaching O volatilization D fugitive dusts D erosion/runoff O uptake by plants Source #_ D leaching D volatilization O fugitive dusts D erosion/runoff D uptake by plants (describe rationale for pot including any of the above release mechanisms) Media affected (or potentially affected) by soil contamination. Source # _ Dair D ground water D surface water O sediments O wetlands Source* _ Dair D ground water D surface water D sediments D wetlands Source # __ Dair D ground water D surface water D sediments D wetlands Check rf present on-sfte or on surrounding land (attach map showing locations) D wetlands D surface water D subsistence fishing D recreational fishing D dairy/beef production Check SSL exposure psthwsys applicable at site; describe basis for ujal including any pathway 1 D mgestion D inhalation D migration to ground water D dermal D soil-plant-human Check Potential for: • D Acute Effects (describe) D Other Human Exposure Pathways (describe) ' D Ecological concerns (describe) J A-10 ' • ^ « * Soil Screening Guidance) Conceptual Sit* Model Summary Forms Form 4: Soil Contaminant Source Characteristics Source No.: ___ Name:_____________________________ Type: ——————————_______________ Location:_______________________:______ Waste type: __________________________ Description (describe history of contamination, other information) Site Name (e.g., drum storage area) (e.g., spBI, dump, wood treater) (site map) (e.g., solvents, waste oil) Describe past/current remedial or removal actions Source depth: Source area: acres m(O measures O estimated) Ref. m2 (O measures O estimated) Ref. Source length parallel to ground water flow. ____ m (if uncertain, use longest source-dimension) Contaminant types (check all that apply): O volatile organics Q other organics Q metals D other inorganics Soil Contaminants Present (list): '________________________________,______ (attach Worksheet #1) Describe previous soil analyses, (attach available results and map showing sample locations) (attach Worksheet #2) Are NAPLs suspected? D Yes O No Reason. Averaoa Soil Characteristics average water content (6W)_ fraction organic carbon (f^.) . dry bulk density (pb) ___ pH _____ .9/9 .(kg/L) *•<• Ref. Ref. A-ll TUT ooe 1095 Workahaat 1. Contaminant-apaclfic properties Site Name Regulatory and Human Health Benchmarks1 i Contaminant CAS# MCLG, MCL,or HBL(mgA) Sources (no.) • . . RID (mgfcg/-d) . Chemical Properties2 Contaminant CAS* Sources (no.) Koc3 (L^g) Kd4 (mg/kg/-dr1 URF H5 (crr&s) (cnr^/s) FVC (mg/m3) S5 (mgO.) 1. ARachriMnt 0 2. Attaehmwit C 3. For organic compounds 4. For metal* and Inorganic compounds 5. Not applicable to rmtafc except mercury A-12 TUT OOS 1096 Worksheet 2. Contaminant concentrations by aourca at* Name. r SAUIM fe , t Contaminant " CAS* average -x Source #: Contaminant CAS* •- average standard deviation standard deviation number of samples • number of samples mwwnum • minimum maximum •. maximum variance . variance A-13 TUT Worksheet 3. Surface SSLs by Exposure Area (EA) StoName EA f: SSL type: D she-specific D generic (default) . Contaminant • CAS# Soil Screening Level ingestion , other (plant uptake; fugitive dust) • EA t: SSL type: D she-specific D generic (default) Contaminant CAS# - Soil Screening Level ingestion ; other (plant uptake; fugitive dust) A-14 TUT 008 1098 Source *: Contaminant SSL . type: D aite-apeclf ie D generic (default) CAS* •• Soil Screening Level inhalation of volatile* • f • migration to ground water Source t:. SSL type: O site-specific O generic (default) Contaminant CAS* Soil Screening Level . inhalation of volatiles migration to ground water _ A-15 TUT 008 1099 I i • - • J 1 i t I 1 i • t • 1 • 1 i •: • • • i 1 - • • - 1 i Figure A-2. Example conceptual ette modal diagram for contaminated aoil (adapted, from U.S. EPA, 1989). A-16 TUT 003 1.100 /"""""x WOODED AREA REGIONAL GEOLOGIC CROSS SECTION WOODED AREA RLL MATERIAL DEPRESSION (DIRECT CONTACT) . (GROUND WATER) CUSTRINE DEPOSITS GLACIAL TILL HALE BEDROCK ^^>y^>y>;i->^vC^ POTENTIAL CONTAMINANT MIGRATION PATHWAY SITE CROSS SECTION Figure A-3. Exampto Site Sketch (adapted from U.S. EPA, 1987) A-17 TUT COS 1101 Attachment B Soil Screening DQOs for Surface Soils and Subsurface Soils TUT 008 Soil Semiring DQOs for Surface Soils Using th« Max Tsst DQO Process Steps Soil Serening inputs/Output* State the Problem Identify scoping team Develop conceptual site model (CSM) Define exposure scenarios Specify available resources Write brief summary of contamination problem Site manager and technical experts (e.g., toxteologtsts, risk assessors, statisticians) CSM development (described in Step 1) Direct ingestion and inhalation of fugitive participates in a residential setting; dermal contact and plant uptake for certain contaminants Sampling and analysis budget, scheduling constraints, and available personnel Summary of the surface soil contamination problem to be investigated at the site Identify the Decision identify decision Identify alternative actions Do mean soH concentrations for particular contaminants (e.g., contaminants of potential concern) exceed appropriate screening levels? Eliminate area from further study under CERCLA or Plan and conduct further investigation Identify Inputs to the Decision Identify inputs Define basis for screening Identify analytical methods Ingestion and particulate inhalation SSLs for specified contaminants Measurements of surface soil contaminant concentration Soil Screening Guidance Feasible analytical methods (both field and laboratory) consistent with program- level requirements______________. ; ______ Define the Study Boundaries Define geographic areas of field investigation Define population of interest Divide site into strata Define scale of decision making Define temporal boundaries of study Identify practical constraints The entire NPL site, (which may include areas beyond facility boundaries), except for any areas with clear evidence that no contamination has occurred Surface soils (usually the top 2 centimeters, but may be deeper where activities could redistribute subsurface soils to the surface) Strata may be defined so that contaminant concentrations are likely to be relatively homogeneous within each stratum based on the CSM and field measurements - Exposure areas .(EAs) no larger than 0.5 acre each (based on residential land use) Temporal constraints on scheduling field visits Potential impediments to sample collection, such as access, health, and safety issues Develop a Decision Rule Specify parameter of interest Specify screening level Specify 'if..., then..." decision rule True mean" (u) individual contaminant concentration in each EA. However, since the determination of the "true mean" would require the collection and analysis of many samples, another sample statistic, the maximum composite concentration, or "Max Test" is used. Screening levels calculated using available parameters and site data, (or generic SSLs if site data are unavailable) Ideally, if the "true mean" EA concentration exceeds the screening level, then investigate the EA further. If the true mean" is less than the screening level, then no further investigation of the EA is required under CERCLA B-l TUT Soil ScrMning DQOs for Surface Soils Using th* Max T«st (continued) DQO Process Steps Soil Screening Inputs/Outputs Specify Limits on Decision Errors* Define baseline condition (null hypothesis) Define the gray region** Define Type I and Type II decision errors Identify consequences Assign acceptable probabilities of Type I and Type II decision errors Define QA/QC goals The EA needs further investigation From 0.5 SSL to 2 SSL Type I error Do not investigate further ("walk away from*) an EA whose true mean" exceeds the screening level of 2 SSL Type II error Investigate further when an EA's true, mean' falls below the screening level of 0.5 SSL Type I error potential public health consequences Type II error unnecessary expenditure of resources to investigate further Goals: Type 1:0.05 (5%) probability of not investigating further when true mean' of theEAis2SSL Type II: 0.20 (20%) probability of investigating further when "true mean' of the EA is 0.5 SSL CLP precision and bias requirements 10% CLP analyses for field methods_____:_______ Optimize the Design Determine how to best estimate true mean* Determine expected variability of EA surface soil contaminant concentrations Design sampling strategy by evaluating costs and performance of alternatives Develop planning documents for the field investigation_______________________ Samples composited across the EA as physical estimates of EA mean Gc )• Use maximum composite concentration as a conservative estimate of the true EAmean. A conservatively large expected coefficient of variation (CV) from prior data for the site, field measurements, or data from other comparable sites and expert judgment A minimum default CV of 2.5 should be used when information is insufficient to estimate the CV. Lowest cost sampling design option (i.e., compositing scheme and number of composites) that will achieve acceptable decision error rates Sampling and Analysis Plan (SAP) Quality Assurance Project Plan (QAPjP)________________________ Since the DQO process controls the degree to which uncertainty in data affects the outcome of decisions that are based on that data, specifying limits on decision errors will allow the decision maker to control the probability of making an incorrect decision when using the DQOs. . The gray region represents the area where the consequences of decision errors are minor, (and uncertainty in sampling data makes decisions too close to call). B-2 TUT OOS Soil Screening DQOs for Subsurface Soils OQO Process Steps Soil Screening Inputs/Outputs State the Problem Identify scoping team Develop conceptual site model (CSM) Define exposure scenarios Specify available resources Write brief summary of contamination problem «. Site manager and technical experts (e.g., lexicologists, risk assessors, hydrogeologists, statisticians). CSM development (described in Step 1>. Inhalation of volatiles and migration of contaminants from sol to potable ground water (and plant uptake for certain contaminants). Sampling and analysis budget, scheduling constraints, and available personnel. Summary of the subsurface sol contamination problem to be investigated at the site. Identify the Decieion identify decision Identify alternative actions Do mean soil concentrations for particular contaminants (e.g., contaminants of potential concern) exceed appropriate SSLs? Eliminate area from further action or study under CERCLA or Plan and conduct further investigation. Identify Inputs to the Decision identify decision' Define basis for screening , Jdentrfy analytical methods Volatile inhalation and migration to ground water SSLs for specified contaminants Measurements of subsurface soil contaminant concentration Soil Screening Guidance Feasible analytical methods (both field and laboratory) consistent with " program-level requirements. Specify the Study Boundaries Define geographic areas of field investigation Define population of interest Define scale of decision making Subdivide site into decision units Define temporal boundaries of study Identify (list) practical constraints The entire NPL site (which may include areas beyond facility boundaries), except for any areas with clear evidence that no contamination has occurred. Subsurface soils Sources (areas of contiguous soil contamination, defined by the area and depth of contamination or to the water table, whichever is more shallow). Individual sources delineated (area and depth) using existing information or field measurements (several nearby sources may be combined into a single source). Temporal constraints on scheduling field visits. Potential impediments tp sample collection, such as access, health, and safety issues. / B-3 TUT OO8 11O5 Soil Screening DQOs for Subsurface Soils (continued) Develop-a Decision Rule Specify parameter of interest Specify screening level Specify If..., then..." decision rule Mean soil contaminant concentration in a source (i.e., discrete contaminant concentrations averaged within each boring). SSLs calculated using avaflable parameters and site data<or generic SSLs if site data are unavailable). If the mean soil concentration exceeds the SSL, then investigate the source further. If mean soil concentration in a source is less than the SSL, then no further investigation is required under CERCLA. Specify Limits on Decision Errors DefineQA/QC goals CLP precision and bias requirements 10% CLP analyses for field methods Optimize the Design Determine how to estimate mean concentration in a source Define subsurface sampling strategy by evaluating costs and site-specific conditions Develop planning documents for the field investigation For each source, the highest mean soil boring concentration (i.e., depth- weighted average of discrete contaminant concentrations within a boring). Number of soil borings per source area; number of sampling intervals with depth. Sampling and Analysis Plan (SAP) Quality Assurance Project Plan (QAPjP) B-4 TUT 008 1106 Attachment C Chemical Properties for SSL Development TUT 008 1.1O7 Attachment C Chemice' Properties This attachment provides the chemical properties necessary to calculate inhalation and migration to ground water SSLs (see Section 2.5.2) for 110 chemicals commonly found at Superfimd shes. The Technical Background Document for Soil Screening Guidance describes the derivation and sources for these property values. • Table C-l provides soil organic carbon - water partition coefficients (K^.), air and water difrusivities (T\» and J\w), water solubilities (S), and dimensionless Henry's law constants (H'). • Table C-2 provides pH-specific K^ values for organic contaminants that ionize under natural pH conditions. She-specific soil pH measurements (see Section 2.3.5.) can be used to select appropriate KOC values for these chemicals. Where she-specific soil pH values are not available, values corresponding to a pH or 6.8 should be used (note that the KoC values for these chemicals in Table C-l are for a pH of 6.8). • Table C-3 provides the physical state (liquid or solid) for organic contaminants. A contaminant's liquid or solid state is needed to apply and interpret soil saturation limit (Cut) results (see Section 2.5.2, p.23). • Table C-4 provides pH-specific soil-water partition coefficients (K*) for metals. She-specific soil pH measurements (see Section 2.3.5) can be used to select appropriate K<j values for these metals. Where she-specific soil pH values are not available, values corresponding to a pH of 6.8 should be used. Except for air and water difrusivities, the chemical properties necessary to calculate SSLs for additional chemicals may be found hi the Superfund Chemical Data Matrix (SCDM). Additional air and water difrusivities may be obtained from the CHEMDAT8 and WATERS models, both of which can be downloaded off EPA's SCRAM electronic bulletin board system. Accessing information is OAQPS SCRAM BBS (919)541-5742 (24 hr/d, 7 d/wk except Monday AM) Line Settings: 8 bits, no parity, 1 stop bit Terminal emulation: VT100 or ANSI System Operator: (919)541-5384 (normal business hours EST) C-l TUT 008 1108 Table C-1. Chemical-Specific Properties used in SSL Calculations CAS No. 83-32-9 67-64-1 309-00-2 120-12-7 56-55-3 71-43-2 205-99-2 207-06-9 65-85-0 50-32-8 111-44-4 117-81-7 75-27-4 75-25-2 71-36-3 85-68-7 86-74-8 75-15-0 56-23-5 57-74-9 106-47-8 108-90-7 124-48-1 67-66-3 9S-.57-8 216-01-9 72-54-8 72-55-9 50-29-3 53-70-3 84-74-2 95-50-1 106-46-7 91-94-1 75-34-3 107-06-2 75-35-4 156-59-2 156-60-5 120-83-2 78-87-5 542-75-6 60-57-1 84-66-2 105-67-9 Compound Acenaphthene Acetone Aidrin •Anthracene Benz(a)anthracene Benzene Benzo(6)fluoranthene Benzo(*)fluoranthene Benzole acid Benzo(«)pyrene Bw(2-chloroethyl)ether Bi»(2-ethylhexyl)phthaJate Bromodichlorpmethane Bromoform Butanol Butyl benzyl phthalate Carbazote Carbon disulfide Carbon tetrachloride Chlordane p-Chtoroaniline • Chlorobenzene Chlorodibromomethane Chloroform 2-Chtorophenol Chryeene ODD DDE DOT Dfeenz(a,/>)anthracene Di-n-butyl phthalate 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 3,3-DichtorobenzkJine 1,1-Dichloroethane 1 ,2-Dichtoroethane 1 ,1-Dichloroethylene c/»-1 ,2-Dichtoroethytene trmns-1 ,2-Dichloroethylene 2,4-Dichiorophenol 1 ,2-Dichloropropane 1 ,3-Dich!oropropene Dwldnn Diethylphthalate 2,4-Dimethylphenol Koc Ou D,,. (L/kg) (cm*/*) (cm*/.) 7.08E+03 5.75E-01 2.45E+06 2.95E-MM 3.98E+05 5.89E+01 T.23E+06 1.23E+06 6.00E-01 1.02E+06 1.55E+01 1.51E+07 5.50E-f01 8.71 E+01 6.92E+00 5.75E-MM 3.39E-MJ3 4.57E+01 1.74E+02 1.20E+05 6.61 E+01 2.19E+02 6.3lE-f01 3.98E+01 3.68E+02 3.98E-f05 1.00E+06 4.47E-f06 2.63E-1-06 3.80E+06 3.39E-K04 6.17E+02 6.17E+02 7^4E+02 3.16E-MJ1 1.74E+01 5.89E+01 3.55E-KJ1 5.25E+01 ' 1.47E+02 4.37E+01 4.57E*01 2.14E-MM 2.88E+02 2.09E+02 4^1E-02 1^4E-01 1^2E-02 3.24E-02 5.10E-02 8.80E-02 2J26E-02 2^6E-02 5^6E-02 430E-02 6.92E-02 3.51 E-02 2.98E-02 1.49E-02 8.00E-02 1.74E-02 3.90E-02 1.04E-01 7.80E-02 1.18E-02 4.83E-02 7.30E-02 1.96E-02 1.04E-01 5.01 E-02 2.48E-02 1.69 E-02 1.44E-02 1.37E-02 2.02E-02 4^8E-02 6.90E-02 6.90E-02 1.94E-02 7.42E-02 1.04E-01 9.00E-02 7.36E-02 7.07E-02 3.46E-02 7.82E-02 6^6E-02 1^5E-02 2.56E-02 5.84E-02 7.69E-06 1.14E-05 4.86E-06 7.74E-06 9.00E-06 8.80E-06 5S6E-06 5^6E-06 7^7E-06 9.00E-06 7^3E-06 3.66E-06 1.06E-05 1.03E-05 9^0E-06 4.83E-06 7.03E-06 1.00E-05 8.80E-06 4.37E-06 1.01E-05 8.70E-06 1.05E-05 1.00E-05 9.46E-06 6^1E-06 4.76E-06 5.87E-06 4.95E-06 5.18E-06 7.86E-O6 7.90E-06 7.90E-06 6.74E-06 1.05E-05 9.90E-06 1.04E-05 1.13E-05 1.19E-05 8.77E-06 8.73E-06 1.00E-05 4.74E-06 6.3SE-06 8.69E-06 S H* (mg/L) (dimeneionleee) 4^4E+00 1.00E+06 1.80E-01 4.34E-02 9.40E-03 1.75E-f03 1.50E-03 8.00E-04 3^0E*03 1.62E-03 1^2Ef04 3.40E-01 6.74E+03 3.10E+03 7.40E+04 2.69E+00 7.48E+00 1.19E+03 7.93E-t-02 5.60E-02 5.30E+03 4.72E*02 Z60E+03 7.92E+03 2^0E+04 1.60E-03 9.00E-02 1J20E-01 2.50E-02 - 2.49E-03 1.12E+01 1.56E^02 7.38E+01 3.11E^OO 5.06E+03 8.52E+03 2^5E+03 3^0E+03 6.30E+03 4.50E+03 Z80E+03 2.80E403 1.85E-01 1.08E+03 7.87E+03 6.36E-03 1.59E-03 6.97E-03 2.67E-03 1.37E-04 2.28E-01 4.55E-03 3.40E-05 6.31 E-05 4.63E-05 7.38E-04 4.18E-06 6.56E-02 2.19E-02 3.61 E-04 5.17E-05 6.26E-07 1.24E+00 1.25E+00 1.99E-03 1.36E-05 1.52E-01 3.21 E-02 1.50E-01 1.60E-02 3.B8E-03 1.64E-04 8.61 E-04 3.32E-04 6.03E-07 3.85E-08. 7.79E-02 9.96E-02 1.64E-07 2.30E-01 4.01 E-02 1.07E-^00 1.67E-01 3.85E-01 1.30E-04 1.15E-01 7^6E-01 6.19E-04 1.85 E-05 8.20E-05 C-2 TUT Table C-1 (continued) CAS No. 51-28-5 121-14-2 606-20-2 117-84-0 115-29-7 72-20-8 100-41-4 206-44-0 $6-73-7 76-44-8 1024-57-3 118-74-1 87-68-3 319-84-6 319-85-7 58-89-9 77-47-4 67-72-1 193-39-5 78-59-1 7439-97-6 72-43-5 74-83-9 75-09-2 95-48-7 91-20-3 98-95-3 86-30-6 621-64-7 1336-36-3 87-86-5 108-95-2 129-00-0 100-42-5 79-34-5 127-18-4 108-88-3 8001-35-2 120-82-1 71-55-6 79-00-5 79-01-6 95-95-4 88-06-2 Compound 2,4-Dinrtrophenol 2.4-Dinitrotoluene 2,6-Dinitrotohiene Di-n-octyl phthalate Endosulfan Endrin Ethybenzene Fluoranthene Fkiorene Heptachtor Heptachlor apoxide Hexachlorobenzene Hexachloro-1 ,3-butadiene a-HCH (a-BHC) B-HCH (B-BHC) -fHCH (Undana) Hexachlorocycloperttadiene Haxachloroathane lndeno(1 ,2,3-ed)pyrene Isophorone Mercury Mathoxychlor Methyl bromide Mathytene chloride 2-Mathylphanol Naphthalene Nitrobenzene AANitrosodiphenylamine M-NHrosodi-n-propytamine PCBs Pentachlorophenol Phenol Pyrene Styrene 1 ,1 ,2,2-Tetrachloroethane Tetrachloroethylene Toluene Toxaphene 1 ,2,4-Trichlorobenzane 1 ,1 ,1 -Trichloroethane 1 ,1 ,2-Trichloroethane Trichloroethylene 2,4,5-Trichlorophenol 2,4,6-Trichlorophenol Kec (L/lcB) 1.00E-02 9.55E+01. 6.92E+01 8.32E+07 2.14E+03 1.23E+04 3.63E+02 1.07E+05 1.38E+04 1.41E+06 8.32E+04 5.50E+04 5.37E+04 1.23E+03 1.26E+03 1.07E+03 2.00E+05 1.78E+03 3.47E+06 . 4.6BE+01 — 9.77E+04 1.05E+01 1.17E+01 9.12E+01 2.00E+03 6.46E+01 1.29E+03 2.40E+01 3.09E+05 5.92E+02 2.88E+01 1.05E-f05 7.76E+02 9.33E+01 1.55E+02 1.82E+02 . 2.57E-f05 1.78E+03 1.10E+02 5.01 E+01 1.66E+02 1.60E+03 3.81E-t-02 DM (cm?/.) 2.73E-02 2.03E-01 327E-02 1.51E-02 1.15E-02 1^5E-02 7.50E-02 3.02E-02 3.63E-02 1.12E-02 1.32E-02 5.42E-02 5.61 E-02 1.42E-02 1.42E-02 1.42E-02 1.61 E-02 2.50E-03 1.90E-02 623E-02 3.07E-02 1.56E-02 7^8E-02 1.01E-01 7.40E-02 5.90E-02 7.60E-02 3.12E-02 5.45E-02 — 5.60E-02 8J20E-02 2.72E-02 7.10E-02 7.10E-02 7J20E-02 8.70E-02 1.16E-02 3.00E-02 7.80E-02 7.80E-02 7.90E-02 2.91 E-02 3.18E-02 Dl. (cm2/«) 9.06E-06 7.06E-06 72BE-Q6 3.58E-06 4^5EX)6 4.74E-06 7.80E-06 6.35E-06 7.88E-06 5.69E-06 4.23E-06 5.91 E-06 6.16E-06 7.34E-06 734E-06 7.34E-06 7^1 E-06 6.BOE-06 5.66E-06 6.76E-06 6.30E-06 4.46E-06 1^1E-05 1.17E-05 8.30E-06 7.50E-06 8.60E-06 6.35E-06 8.17E-06 ~ 6.10E-06 9.10E-06 724E-06 8.00E-06 7:90E-06 8J20E-06 8.60E-06 4^4E-06 8J23E-06 8.80E-06 8.80E-O6 9.10E-06 7.03E-06 6^5E-06 S H' (mg/L) (dimanalonlaaa) 2.79E+03 2.70E+02 1.82E*02 2.00E-02 5.10E-01 2.50E-01 1.69E+02 2.06E-01 1.98E+00 1.80E-01 2.00E-01 650E+00 3^3E+00 2.00E+00 2.40E-01 6.80E+00 1.BOE+00 S.OOE-t-01 2.20E-05 1^0E+04 ~ ' . 4.50E-02 1.52E+04 1.30E+04 2.60E+04 3.10E+01 2.09E+03 ' 3^i1E+01 9.B9E+03 7.00E-01 1.95E+03 828E+04 1.35E-01 3.10E+02 2.97E+03 2.00E-f02 5^6E+02 7.40E-01 • 3.00E+02 1.33E403 4.42E+03 1.10E+03 1^0E+03 B.OOE-t-02 1.82E-05 3.80E-06 3.06E-05 2.74E-03 4.59E-04 3.08E-04 3.23E-01 6.60E-04 2.61 E-03 6.07E+01 3.ME-04 5.41 E-02 3.34E-01 4.35E-04 3.05E-05 5.74E-04 1.11E+00 1.59E-01 6.56E-05 2.72E-04 4.67E-01 6.48E-04 2.56E-01 8.98E-02 4.92E-05 1.98E-02 9.84E-04 2.05E-04 9.23E-05 — 1.00E-06 1.63E-05 4.51 E-04 1.13E-01 1.41 E-02 7.54E-01 2.72E-01 2.46E-04 5.82E-02 7.05E-01 3.74E-02 4.22E-01 1.78E-04 3.19E-04 C-3 TUT COS iilO Table C-1 (continued) CAS No. 108-05-4 75-01-4 .108-38-3 05-47-6 106-42-3 Compound Vinyl acetate Vinyl chloride m-Xylene o-Xybne pOCytene ..••_(i5r«> 5J25E+00 1.66E+01 4.07E+02 3.63E+02 3.89E+02 (em'/E • 1.06&01 7.00E-02 8.70E-02 7.60E-02 °l,* 8 H' /emi/e) (mg/L) (dim«neionle*») 9^0E-06 7.80E-06 1.00E-05 B.44E-06 &OOE+04 2.76E+03 1.61E+02 1.78E+02 &10E-02 t.11E+00 3.01 E-01 2.13E-01 3.14E-01 - Soloi toonftwtornriio S H' Htnnrt>tawcoMtrt (HLC * 41) (28 -Q. C-4 TUT OO8 111! Table C-2. Koe Values for ionizing Organics as a Function of pH pH 4.9 5.0 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 8.0 ** ————— •- DawBOIC AoU 5.54E+00 4.64E+00 3.88E+00 3.25E+00 2.72E+00 229E+00 V1.94E+00 1.65E+00 1.42E+OQ 1.24E+00 1.09E+00 9.69E-01 8.75E-01 7.99E-01 7.36E-01 6.89E-01 6.51 E-01 6.20E-01 5.95E-01 5.76E-01 5.60E-01 5.47E-01 . 5.38E-01 5.32E-01 5.25E-01 5.19E-01 B.16E-01 5.13E-01 5.09E-01 5.06E-01 5.06E-01 5.06E-01 2- CfllOfO* 3.98E+02 3.98E+02 3.98E+02 3.98E+02 3.98E+02 3.98E+02 3.97E+02 3.97E+02 3.97E+02 3.97E+02 3.97E+02 3.96E+02 3.96E+02 3.96E+02 3.95E+02 3.94E+02 3.93E+02 3.92E+02 3.90E+02 3.88E+02 3.86E+02 3.83E+02 3.79E+02 3.75E+02 3.69E+02 3.62E+02 3.54E+02 3.44E+02 3.33E+02 3.19E+02 3.04E+02 2.86E+02 2«4'PiCnlolO* piwnoi 1.59E+02 1.59E+02 1.59E+02 1.59E+02 1.59E+02 1.58E+02 1.58E+02 1.58E+02 1.58E+02 1.5BE+02 1.57E+02 1.57E+02 1.57E+02 1.56E+02 1.S5E+02 1.54E+02 1.53E+02 1.52E+02 1.50E+02 1.47E+02 1.45E+02 1.41 E+02 1.38E+02 1.33E+02 1.28E+02 1.21 E+02 1.14E+02 1.07E+02 8.84E+01 8.97E+01 8.07E+01 7.17E+01 Dteltg- 1 2.94E-02 2.55E-02 &23E-02 1.9BE-02 1.78E-02 1.62E-02 1.SOE-02 1.40E-02 1.32E-02 1.25E-02 1.20E-02 1.16E-02 1.13E-02 1.10E-02 1.08E-02- 1.06E-02 1.05E-02 1.04E-02 1.03E-02 1.02E-02 1.02E-02 1.02E-02 1.02E-02 1.01 E-02 1.01 E-02 1.01 E-02 1.01 E-02 1.01 E-02 1.00E-02 1.00E-02 1.00E-02 1.00E-02 * ———— *--•-•- —— rawnapDfwofi*" 9.05E+03 7.96E+03 6.93E+03 5.97E+03 5.10E+03 4.32E+03 3.65E+03 3.07E+03 2.58E+03 2.18E+03 1.84E+03 1.56E+03 1.33E+03 1.15E+03 9.98E+02 8.77E+02 7.81 E+02 7.03E+02 6.40E+02 5.92E+02 5.52E+02 5.21 E+02 4.96E+02 4.76E+02 4.61 E+02 4.47E+02 4.37E+02 4.29E+02 4^3E.+02 4.18E+02 4.14E+02 4.10E+02 f^MMOTD. 1.73E+04 1.72E+04 1.70E+04 .1.67E+04 1.65E+04 1.61 E+04 1.57E+04 1.52E+04 1.47E+04 1.40E+04 1^2E+04 1JI4E+04 1.1SE+04 1.05E+04 4.51E+03 8.48E+03 7.47E+03 6.49E+03 5.5BE+03 4.74E+03 3.99E+03 3.33E+03 2.76E+03 2^BE+03 1.87E+03 1.53E+03 1^5E+03 1.02E+03 8.31 E+02 6.79E+02 5.56E+82 4.58E+02 pfMDOl 4.45E+03 4.15E+03 "3.83E+03 3.49E+03 3.14E+03 2.79E+Q3 2.45E+03 2.13E+03 1.83E+03 1.56E+03 1.32E+03 1.11E+03 9^7E+02 7.75E+02 6.47E+02 5.42E+02 4.55E+02 3.B4E+02 3^7E+02 2.80E+02 2.42E+02 2.13E+02 1.B8E+02 1.69E+02 1.53E+02 1.41 E+02 1.31 E+02 1^3E+02 1.17E+02 1.13E+02 1.08E+02 1.0SE+02 yg*»~ 2.37E+03 2.36E+03 2.38E+03 2.35E+03 2.34E+03 2.33E+03 '2.32E+03 ^31E+03 2£9E+03 Z27E+03 2^4E+03 2J21E+03 2.17E+03 2.12E+03 2.06E+03 1.99E+03 1.91E+03 1.82E+03 1.71E+03 1.60E+03 1.47E+03 1.34E+03 1.21E+03 1.07E+03 9.43E+02 8.19E+02 7.03E+02 5.99E+02 5.07E+02 4.26E+02 3.57E+02 2.98E+02 •TrieMofCK 1.04E+03 1.03E+03 1.02E+03 1.01E+03 9.99E+02 9.62E+02 9.62E+02 9.38E+02 9.10E+02 a77E+02 8.39E+02 7.96E+02 7.48E+02 6.97E+02 6.44E+02 5.B9E+02 5.33E+02 4.80E+02 4.29E+02 3.81 E+02 3.38E+02 3.00E+02 2.67E+02 2.39E+02 2.15E+02 1.95E+02 1.78E+02 1.64E+02 1.53E+02 1.44E+02 1.37E+02 1.31 E+02 -C-5 TUT 008 1112 Table C-3. Physical State of Organic SSL Chemicals compounds liquid at son temperatures Compounds aoKd at aoil temperatures GAS No. Chemical Making Point CASNo. Chemical Melting Point fC) 67-64-1 Acetone -94.8 71-43-2 Benzene 5.5 117-61-7 Bis(2-ethylhexyl)phthalate -55 111-44-4 Bis{2-chloroethyl)ether -51.9 75-27-4 Bromoofchioromathane -57 75-25-2 Bromofonm & 71-36-3 Butanol ' -89.6 85-68-7 Butyl benzyl phthalate -35 75-15-0 Carbon dwulfide -115 56-23-5 Carbon tetrachloride -23 106-90-7 Chforobenzene -45.2 124-48-1 Chtorodibromomethane -20 67-66-3 Chloroform -63.6 95-57-8 2-Chforophenol 9.6 64-74-2 Di-n-butyl phthalate -35 95-50-1 1,2-Dichlorobenzene -16.7 75-34-3 1,1-Oichloroethane -96.9 107-06-2 1.2-Dichloroethane -35.5 75-35-4 1,1-Oichlbroethylene -122.5 156-59-2 e/s-1,2-Dichloroethylene -80 156-60-5 frans-1,2-Dichforoethylene -49.8 78-87-5 1,2-Dichloropropane -70 542-75-6 1,3-Oichlofopropene NA 84-66-2 DietnylphthaJate -40.5 117-84-0 Di-n-octyl phtnalate -30 100-41-4 Ethylbenzene -94.9 87-68-3 Hexachloro-1,3-butadiene -21 77-47-4 Hexachlorocyclopentadiene -9 78-59-1 Isophorone -8.1 74-83-9 Methyl bromide -93.7 75-09-2 Methytone chloride . -95.1 98-95-3 Nitrobenzene 5.7 100-42-5 Styrene -31 79-34-5 1,1,2,2-Tetrachloroethane -43.8 127-18-4 Tetrachloroethylene -22.3 108-88-3 Toluene -94.9 120-82-1 1,2,4-Trichlorobenzene 17 71-55-6 1,1,1-Trichloroethane -30.4 79-00-5 1,1,2-Trichloroethane -36.6 79-01-6 Trichloroethylene -84.7 108-05-4 Vinyl acetate -93.2 75-01-4 Vinyl chloride -153.7 • 108-38-3 m-Xylene -47.8 95-47-6 o-Xylene -25.2 106-42-3 p-Xylene 13.2 83-32-9 Acenaphthene 93.4 309-00-2 AMrin 104 120-12-7 Anthracene 215 56-55-3 Benz(a)anthracene ' 64 50-32-8 Benzo(a)pyrene 176.5 20549-2 Benzo(*)fluoranthene 168 207-08-9 Benzo(*)fluoranthene 217 65-85-0 Benzoicacid 122.4 86-74-8 Carbazole 246.2 57-74-9 Chtordane T06 106-47-8 p-Chbroanlne 72JS 218-01-9 Chryaena 258.2 72-54-8 000 109.5 72-55-8 DOE 89 50-29-3 DOT 108.5 53-70-3 Dfcanzo(a,A)anthracene 269.5 106-46-7 1,4-Dichlorobanzane 52.7 91-94-1 3,3-Dfchtorobenzidine 132.5 120-83-2 2,4-Dichlorophenol 45 60-57-1 Dteldrin 175.5 105-67-9 Z4-Dimathy1phahol 24.5 51-28-5 2,4-Dwitrophenol 115-116 121-14-2 2,4-Dinrtrotokiene 71 . 606-20-2 2,6-Dinitrotoluene 66 72-20-8 Endrin 200 206-44-0 Fluoranthene 107.6 86-73-7 Fluorene 114.8 76-44-6 Heptachlor 95.5 1024-57-3 Heptachlor apoxide 160 118-74-1 Hexachlorobenzene 231.8 319^84-6 o-HCH(o-BHC) 160 319-85-7 B44CH(6-BHC} 315 58-89-9 yHCH (Lmdane) 112.5 67-72-1 Hexachloroethane 187 193-39-5 lne»eno(1,2,3-eo)pyrene 161.5 72-43-5 Methoxychlor 67 95-48-7 2-Methylphenol 29.8 621-64-7 AANitroaodi-n-propyiamme NA 86-30-6 AWitroeooSpnenyJamine 66.5 91-20-3 Naphthalene 80.2 87-86-5 Pentachlorophenol 174 108-95-2 Phenol 40.9 129-00-0 Pyrene 151.2 8001-35-2 Toxaphene 65-90 95-95-4 2,4,5-Trichlorophenol 69 88-06-2 2.4,6-Trichlorophenol 69 115-29-7 Endoaullfan 106 NA-Not«v*J«W». C-6 oos 1113 •Aj$A|p8dsej ooo'I PUB '6*6 'S* e*& uinipBUBA puB 'epiueAo 'Auoiu|iUB aoj sen|BA fy ojueBjoui )uepuedep-Hd uou 30+36*8 20+30'* 20+3 re 20+3*"2 20+36' (. 20+39-1. 20+36*1 20+31*1. 10+38*6 (.0^16*8 (.0+38*4 10+38*9 lO+32'9 (.0+38-8 lO+3»-S 10+3 rs (.0+34** (.0+3*** 10+32** 10+36*6 10+39*6 (.0+3**6 W+32'6 W+30'6 1.0+38*2 10+39*2 (.0+38-2 W+36'2 10+3 r2 fO+36-l. 10+39' I 1.0+39- 1 MZ (.0+39*6 iO+3*'6 (.0+31*6 (.0+36*8 (.0+34*8 (.6+38*8 10+32*8 (.0+30*8 10+38*1 10+39*4 l.O+3**4 (.0+36-4 10*31*4 (.0+38*9 (.0+34*9 10+39*9 I.O+3**9 10+32*9 10+31*9 10+36*8 10+39*8 10+39'S 10+38*8 lO+3*-S (.0+32'S 10+3 rs (.0+30*8 10+39'* (.0+34-* |.0+39t 10+38** 10+3*** 11 00+32*2 00+3*'2 00+38*2 00+34*2 00+36*2 00+3 re 00+36*6 00+38'e 00+39*6 00+3 It 00+36** 00+34** 00+30*8 00+36*8 00+34*8 00+3 1' 9 00+38*9 OD+30'4 00+38'4 00+30*8 00+39-8 00+32*6 00+39*6 (.0+31*1 . (.0+3 n. (.0+32*1. (.0+36*1. Ik0+3*'l (.0+38*1 10+39' I (.O+34'l. 10+39' I •S 20+31* I (.0+36*8 10+36*4 (.0+36*8 10+38'* (.0+36*6 1-0+3 re (.0+38*2 10+10*2 10+39' I 10+36*1. (.0+30*1 00+36*8 00+39*9 00+36*8 00+32'* 00+3**6 00+34*2 00+31*2 00+34' I 00+36*1. 00+3 rt. >0-3**8 (.0-34*9 (.0-36*8 10-32'* (.0-36'e •tO-39'2 1.0-3 r2 (.0-39*1. (.0-36*1. (.0-30*1 BV 60+36* i 60+3** I 20+36'6 20+30*4 . 20+36** 20+38*6 20+38*2 20+38*1 20+3** I 20+3 r I (.0+38-8 |.0+3**4 (.0+38*9. (.0+38*8 I.O+3**S 10+30*8 (.0+34-* 10+38'* 10+32'* 10+30** 10+38*6 10+39*6 io+3*-e (.0+32*6 10+30*6 (.0+39*2 10+39*2 lO+3*'2 lO+32'2 l.0+30'2 10+39' 1. I.O+39'I. IN 20+30*2 20+36'l 20+36-1 20+38' I 20+34't 20+39'!. 20+39' (. 20+36-1. 20+32'l (.0+36-6 (.0+32-9 (.0+39-9 10+32'S . I.O+30-* 10+30'e (.Q+32'2 (.0+39*1.. (.0+3 n 00+38*4 00+3 i*S 00+38'e 00+36*2 00+39' I 00+30* I (.0-36*9 10-39* (.0-30'e (.0-30*2 10-3** I 20-30*6 20-30'9 20-30** BH (.0+3** 1. (.0+3*- 1 (.0+3*-*. 10+3S' I (.0+38*1. (.0+39- (. (.0+39* (. 10+39* I 10+34*1 10+34*1. (.0+38*1. (.0+38*1. 10+36' i (.0+36-1. l.0+30'2 t.0+30'2 (.0+31*2 10+32*2 I.O+32'2 (.0+36'2 10+36*2 lO+3*'2 (.0+38-2 io+38'2 (.0+39-2 10+34*2 10+34*2 10+38*2 I.O+36'2 10+30'e 1.0+3 re 10+3 re (9+) JO 90+36'* 90+36'* 90+36'* 90+32'* 90+31'* ^90+36*6 90+34*6 90+3**6 90+3 re 90+38*2 90+38*2 90+31*2 90+39*1 90+39*1. 90+32' I SO+36'6 80+34*4 80+38*8 80+32'* SO+30'6 SO+30'2 80+36'*. *0+34'8 *0+3S'S *0+3S'6 *0+31'2 *0+36'l. eo+sre 60+36'* eo+30'e 60+36' 1. 60+32'^ (e+) JO 60+36'* 60+36*2 60+36*1. 60+36*1. 20+34*8 20+36'S 20+30** 20+382 20+30*2 20+38' 1. 20+3 r I (.0+31*6 (.0+39*4 I.O+3*'9 I.O+34'S (-0+32*8 1.0+38'* I.O+3*'* (.0+32** (.0+30** (.0+34*6 10+38*6 10+36*6 1.0+3 re 10+36*2 10+34*2 10+3S'2 10+36*2 (.0+3 1'2 (.0+36' I 10+34* i '10+38- 1 PO 80+30* (. *0+36'8 *0+39** *0+30*6 *0+30*fS *0+36**i 60+39*8 60+34'S 60+38*6 60+38*2 60+34*1. eo+sri 20+36'4 20+38'S 20+36'C 20+38-2 20+3 1 '2 20+39' i 20+32*1 |.0+36'6 (.0+32*8 I.O+36'9 (.0+30*9 (.0+36*8 (.0+34** 10+32** (.0+38'e (.0+38*8 10*3 re 10+38*2 |.0+39'2 lO+36'2 •8 10+32 S 10+30 S I.O+36-* 10+34** I.O+39'* (.0+39** (.0+38** (.0+3*-* (.0+3*-* (.0+36'* (.0+32** 10+32** 10+3 It 10+30'* (.0+36*6 (.0+34*6 (.0+39*6 (.0+38'e (.0+36*6 (.0+31*6 1.0+30*6 l.O+38'2 (.0+39-2 lO+3*'2 J.0+32'2 (.0+31*2 10+36* 1. (.0+341- (.0+38*1. W+3**l 10+32* IT 10+31' I •8 (.0+3 re 10*3 re 1.0+3 re 10*31*6 io+3t*e io+3o*e W+30'6 io+3o*e (.0+30*6 10+36*2 W+36? 10+36 2 (.0+36*2 (.0+36*2 1.0+382 10+38 2 (.0+39*2 10+382 10+38*2 (.0+34*2 (.0+34*2 10+34*2 lO+34'2 lO+34'2 (.0+39-2 lO+39'2 I.O+39'2 t.0+39'2 (.0+39*2 (.0+33*2 (.0+38*2 (.0+38*2 •V 0'8 6L 00 8*4 ' § 4'4 • 9'4 t; S'4- *'4 e*4 2'4 r4 0'4 6'9 9'9 4'9 9'9 S9 *9 17 6'9 ° 29 . 1*9 0*8 6*8 9*8 4*8 9*8 S'S *'S 6'S 28 -rs O'S 6'* Hd •Hd jo uo|jounj e SB (6)(n) san|BA ^ Attachment D Regulatory and Human Health Benchmarks Used for SSL Development TUT 008 1115 Attachment D Regulatory and Human Health Benchmarks for SSL Development This attachment provides regulatory and human health benchmarks necessary to calculate SSLs for 110 chemicals commonly found at National Priority List (NPL) sites. The sources of these values (shown in the following table) are regularly updated by EPA. Prior to calculating SSLs at a site, check all relevant chemical-specific values in this attachment against the most recent version of their sources to ensure that they are up-to-date. D-l TUT oos Attachment D. Regulatory and Human Health Benchmarks Used for SSL Development CAS ftumlul ii...,. 83-32-9 AcamphtMna 67-64-1 Ac*tofw(2-Propanant) 309-00-2 AMrin 120-12-7 Arrihmeana 7440-36-0 Anftmony 7440-38-2 Artanlc 7440-39-3 Barium 56-55-3 Bantfc )antta««m 71-43-2 Banzana 205-99-2 BariM(D)lluoran1hana 207-06-9 Banw(*)lluomnthaoa 65-854) Btmtofcadd 50-32-8 Banzo(a Jpyrane 7440-41-7 Bwyium 117-81-7 Blt(2-a>iyt»»iyl)pM>alala 75-27-4 BmnpuTcHuromeftane 75-25-2 Bfomolorm (Iribnmimnalhana) 71-36-3 Butanol 8548-7 Bu*lbaruytphtMMa 7440-43-9 Cadmium 88-74-8 Carbatoto 75-15-0 Carton dtouMda 56-23-5 Carbon WracHoride 57-74-9 Chtofdana 106-47-8 p-CHoroanHna 108-90-7 CKotobaniana 124-48-1 CHoredbromomatiane 67-66-3 Chtorotorm 95-57-8 2-ChtefCphanol Maximum Contaminant Laval Goal (mg(L) MCLQ (PMCLO) Ral.* 6.0E-03 3 2.0E+00 3 4.06-03 3 5.06-03 3 1.06-01 3 6.06-02 3 Maximum Contaminant Laval (mcH) MCL(PMCL) R((. 6.0E-03 3 5.06-02 3 2.0E+00 3 5.06-03 3 2.06-04 3 4.06-03 3 6.06-03 3 1.0E-01 * 3 1.06-01 * 3 5.06-03 3 5.0643 3 2.06-03 3 1.06-01 3 1.0E-01 ' 3 1.06-01' 3 WalarHaaMiBaaad LknHa HBL* Bmt* 2E+00 RJD 4E400 RIO 5E-08 SF. 1E+01 RID 16-04 SF. 16-04 SF. 1E-03 SF. 1E+02 RID 66-05 SF. 46400 RID 76400 RIO • 46-03 SF. 4E+00 RID IE-01 RID 26-01 RID Canear Stepa Factor Care. 0 82 1.76401 1 0 A 1.56400 1 82 7.3641 4 A 2.96-02 1 82 7.36-01 4 B2 7.36-02 4 82 7.36400 82 4.36400 82 1.16400 B2 1.46-02 82 6.26-02 82 7.9643 D C 82 2.0642 2 82 1,3641 1 82 1.36400 1 D C 8.4E42 1 82 8.1643 1 UnMRtok Factor bare. 0 82 4.9643 1 D A 4.3643 1 82 A 8.3648 1 82 82 82 82 2.4643 1 82 3.3644 1 82 82 82 1.1648 1 D C 81 1.8643 1 82 1.SE45 1 82 3.7E44 1 D C 82 2.3645 1 RafarancwDoM "• Baf.« "V3E41 1 1.0641 1 3.0645 1 3.0641 1 4.0644 1 3.0644 1 7.0642 1 4.06400 1 6.0643 i 2.0642 1 2.0642 1 10642 1.0641 2.0641 1,0643" 1.0641 7.0E44 8.0645 4.0643 2.0642 2.0642 1.0E42 1 5.0643 1 Ralaranea__ Coneanli allufi "« •»-.• 5.0644 2 t 7.0641 1 a 2.0642 2 • Ptopoted MCI . 0.08 mg/i. DrinUng Hfrfcr flsjufcifon* tnd HetHi AWsww*. U.S. EPA (1995). " Cadmium RID 1* baaed on dtotan; axpoture. o )ro CD Attachment D (continued) Number Chemical Name 7440-47-3 Chromium 18065-83-1 Chromium (tlQ 18540-29-9 Chromium(VI) 218-01-9 Chtyiana 57-12-5 Cyar*to(*mantM«) 72-54-8 ODD 72-5S-9 DOE 50-29-3 DOT 53*70-3 Dlbtftz^A jMttvitcvnB • . 84-74-2 Dl-n -butyl pWhatata • 95-50-1 1.2'OfcMorobanzww 106-46-7 1,4-DfcMorabaiuana •1-94-1 3.3-OfcNorobaniMr* 75-34-3 1,1-WcNoroatiana 107-08-2 1.2-DfcMoreatMna 75-354 1,1-DlcHoroatiytene • CA-Cft 9 aj» 4 4_nL4AtMuM*i^M^ I9V-W« CW *9ff'UiU9HfVUmifimw 158-80-S *WM .1 ,2-OfcMoNMrihylana 120-83-2 2.4-OMiteipphanol 78-87-5 1.2-0°lcHareprapaM S42-75-8 1>OlcHoniprapana 60-57-1 OMdrin AM AB 1 fftLafiauaVvttlftftataatfcBV ' •i wi c UMinyiiininavnir 105-67-9 z.4-Dlma«ylphirwl 51-28-5 Z.4-DWttt)pf»»nol 121-14-2 2,4-DHkotoliiana** 606-20-2 2,6-DHfrotoluam" 117-844> Dl-n-oelylphthitala 115-29-7 Endotullan 72-20-8 Endrin Matknum ContambiaHt Level Goal (mgd) MCtQ (PMCLQ) Rat.' I.OE-01 3 (2.0E-01) 3 8.0E-01 3 7.5E-02 3 7.0E-03 3 7.0E-02 3 1.0E-01 3 •' 2.0E-03 3 Maximum °rt*(m5l5L"'* MCL(PHCL) B-. I.OE-01 3 1.0E4)1 3 ' (2.0E-01) 3 e.oE-oi 3 7.5E-02 3 5.0E-03 3 7.0E-03 3 7.0E-02 3 I.OE-01 3 5.0E-03 3 2.0E-03 3 "'tlmHa * (mgrt.) HBL» **** 4Et01 RID IE-02 SF. . 4E-04 SF. 3E-04 SF. 3E-04 SF. 1E-05 SF. 4E+00 RIO 2E-04 SF. 4E+00 RIO IE-01 RIO 5E-04 SF. 5E-OB SF. 3E+01 RID 7E-01 RIO 4E-02 RID IE-04 SF. IE-04 SF. 7E-01 .HID 2E-01 •• • RID c",^r" Claaa* * **' A A • B2 7.3E-03 4 D 02 2.4E-01 1 02 3.4E-01 1 B2 3.4E-01 1 B2 7.3E+00 4 0 D 82 2.4E-02 2 82 4.5E-01 1 c. .. . 82 6. IE-02 1 C 6.0E-01 1 0 82 8.8E-02 2 82 UE-Ot 2 82 1.8E+01 1 D 82 6.8E-01 t 02 8.8E-01 1 0 UiHlMak Factor Carfe Claw* "^ "** A 1.2E-02 1 A UE-02 1 0 82 82 82 8.7C-OS 1 02 0 D 82 82 C 82 2J64S . 1 C S.OE-05 1 D 82 82 3.7E-05 2 82 4.8E-03 1 0 . 0 RafaraneaDoaa » W.' 5.0E-03 1 1.0E*00 1 5.0E-03 1 2.0E-02 1 S.OE-04 1 I.OE-01 1 t.OE-02 1 > ., I.OE-01 7 t.OE-03 1 1.0E-02 2 2.0E-02 1 3.0E-03 1 3.-OE44 1 5.0E-OS I IOE-01 1 2.0E-02 1 2.0E-03 1 2.0E-03 1 1.0E-03 2 2.0E-02. 2 6.0E43 2 3.0E-04 1 "S" we 2.0C-61 2 8.0E-01 1 8.0E-01 2 4.0E-03 t 2.0E-02 1 oo 03 !-* !— j-- 1 ce • MCL tor total chromhimJ* based on Cr (VI) toxldly, " Cancer Slope Factor Is for 2,4-. 2,6-Dlnilnitolu«ne mixture. Attachment D (continued) C AS I^Ljaailnal Mann Utttv&tMf vlnPnliwm nvntv 100-41-4 Elhytannrw 206-44-0 FhionmlhwM 86-73-7 Fkioran* 76-44-8 MptocHor 1024-57-3 HaptacNoropwkfe 118-74-1 HmtcHorobanMiw 87-68-3 H«MKHoro-1,3-taut«»«ra 319-84-8 a-HCH(a-BHC) 319-85-7 B-HCHW-BHC) 58-80-0 r-HCH(Llntana) 77-47-4 Hmachtorocydoponladlan* 67-72-1 HoMcHoreotMM 103-30-5 Nfena(1.2.3-crf)p)mnB 78-50-1 tophoran* 7430-07-6 Mwcwy 72-43-5 Metoxydilor 74-83-0 Matfiyibiwnkto VC 4M 1 M^dM^AI^ jj^tflfc^ 75-00-2 Mwnyiini CMMM* 95-48-7 2-Mrtirt»n»nol(o-crMoQ 91-20-3 NapNMww 744042-0 WCM 98-95-3 Nlliobtom* 86-304 W -Nfcwodlpntnylafnina . 621-64-7 N -NNrotodl-n -propyliinlM 67-06-5 PtrtMHwopnMiol 108-05-2 PNnel 120404 Pjwrn 7782-40-2 Sttonlum 7440-22-4 Slwr 100-42-5 Stynm 70-34-5 1.1^-TMrMhhMMtMm • Mmhiwm coRunwiMtt Lwtti Goal (mgl) MCLQ (PMCLO) R«f.* 7.0E41 3 1.0E43 3 2.06-04 3 5.0E42 3 2.06-03 3 4.0E42 3 S.OE42 3 1.06-01 3 H«nL>HHH Hmnmuin • CofitAffwiMrt Lvml (mgVL) «CL(PMCL) R-. 7.0E41 3 4.06-04 3 2.0E44 3 t.OE-03 3 2.0E44 3 S.OE42 3 2.0E43 3 4.0E-02 3 5.06-03 3 1.0E-03 3 S.OE42 3 1.0E41 3 Water Health BaMd LbnH* (mgA.) HBL» Ba-« 1E+00 RID 1E+00 RID 1E43 SF. 1E45 SF. SE-05 SF. 6E43 SF. IE-04 SF. 0E-02 SF. SE42 RID 2E+00 RD 1E+00 RID 1E41 HA* 2E-02 RID 26-02 SF. 1E45 SF. 2E+01 RID 1E+00 RID 2E-01 RID 4E44 SF. CMtCGT 9|0|M*^ Factor (mgfliBHir' care. Cb»* *' "*' D D D B2 4.SE+00 1 . 82 0.1E+00 1 B2 1.66*00 1 C 7.8E42 1 B2 6.36+00 1 C 1.BE+00 1 B2 1.3E*00 2 D C 1.4E42 1 B2 7.3E41 4 C 9.5E44 1 D . D 0 82 7.5E-03 1 C D A D 82 4.0E43 1 82 7.0E400 1 82 1.2E41 1 D 0 0 D G 2.0E41 1 UnftRM Factor (MgNV cam. cb_. "^ R«f.' D 0 82 1.3E43 82 2.6E43 M M MZJM ^.OCHn C 2.2E-OS 82 1.8E43 C S.3E44 C D C 4.0E46 1 82 C D D 0 82 4.7E47 1 C D A 2.4E44 1 0 • 82 82 82 D 0 0 D C S.8E45 1 R*t«r*nc« DM* (mgfktMl) *° Bal.' 1.0E41 4.0E42 4.0E42 S.OE44 1.3E45 8.0E-04 2.0E-04 2 3.0E44 1 7.0E-03 1 1.06-03 1 2.06-01 1 3.0E44 2 S.OE-03 1 1.4E43 1 6.0E-02 1 S.OE-02 1 4.0E42 6 2.0E42 1 5.06-04 1 3.0E42 1 6.0E41 1 3.0E42 1 S.OE43 1 S.OE43 1 2.06-01 1 'Rafarenc* ConeMitratton (m^) "^ HA' 1.0E+00 1 7.0645 2 3.06-04 2 S.OE43 1 A nCaon 9 «.UCTUU € 2.0E43 2 1.06*00 1 • HMhh «(M«oiy tor nlckrt (Ma l» cur»n»V wninSd); ERA Offic* ol Sctanw «nd T«choolog)r, 7/KV05. oi. o 63 i-4- i— "0 Attachment D (continued) Number Chemteal Name 127-18-4 Tatrachtoroettiylene 7440-28-0 Thallium 108-88-3 Toluene 8001-35-2 Toxaphana 120-82-1 1,2.4-TitehteroDenzane 71-SS-6 1.1.1-TrleHoroetwn* 79-00-5 1.1,2-Trtehloroetiane 79-01-6 Trichtoroatiytana 95-95-4 2.4,5-Trlchtorophanol 88-06-2 2.4,6-Trichtorophanol 7440-82-2 Vanadhm 108-05-4 Vinyl acalata 7541-4 Vinyl cHorlda (ehtoroatharw) 108-38-3 m -Xytena 95-47-6 e-Xytona 108-42-3 p-Xytena 7440-88-6 line Maximum Contaminant Laval Goal (mgIL) MCtQ (PMCLQ) Raf.* S.OE-04 3 LOEtOO 3 7.0E-02 3 2.0E-01 3 3.0E-03 3 zaro 3 1.0E*01 3 * 1.0E*01 3' 1.0E401 3* Maximum Contaminant Laval (mgl) MCL(PMCL) R-. 5.0E-03 3 2.0E-03 3 I.OEtOO 3 3.0E-03 3 7.0E-02 3 2.0E-01 3 5.0E-03 3 S.OE-03 3 2.0E-03 3 1.0E+01 3* 1.0E+01 3 * 1.0E+01 3* WatarHaaMiBaaad LlmHa (moA) HBL* B<-* 4E+00 RIO BE-03 SF. 3E-01 RIO 4E+01 RID 1E+01 RID Canoar Stop* Factor (rr^k»<«»' care. SF. a-t • Claaa* Hai. D B2 1.1E+00 1 D 0 C 5.7E-02 1 1.1E-02 5 B2 1.1E-02 1 • . A 1.9E400 2 D D D D Unit Hlek Factor (mwr care. •URF B-J • i«^^t Raj. 5.8E-07 5 D B2 3.2E-04 1 0 0 C 1.8E-05 1 ; 1.7E-08 S B2 3.1E-06 1 A 8.4E-05 2 D D D 0. Ralaranoa Doaa (moAo>d) 9m Rat.' 1.0E-02 1 2.0E-01 1 1.0E-02 1 4.0E-03 1 1.0E-01 1 ' 7.0E-03 2 1.0E+00 1 2.0E+00 2 2.0E+00 2 2.0E400 1 " 3.0E-01 1 ••nMWvAOv C—— — ^«Jr*ilj>a» micwfiiTmvfi (mgAtr) "^ M.* 1 ————————— 4.0E-01 1 • 2.0E-01 2 'l.OE+00 S 9 2.0E-01 1 .; ~f • MCI tor total xytonaa |1330-20-7) to 10 mot. " RIO tor total xytonat ta 2 mgflqhday. • Ralerencet: 1 «IRIS. U.S. EPA (1995) 2.HEAST.U.S..EPA(1995) 3. U.S. EPA (1995) 4.0HEA.U.S. EPA(1993) 5 • httorlm toxWty criteria, provided by Supartund Haalti Rltk TacHneal Support Cantor. Environmental Criteria A«te»ament Offlca (ECAO). Onclnnall, OH (1994) 8«ECAO.US.EPA(1994I) 7. ECAO. U.S. EPA(1994h) • Health Bated LlmHf ealculatad tor 30-year exposure duration. 10' rl»k or hazard quotient»1. CataQOrizaion of ovara.1 walynl ol aviaanoa tor human cwd QroupA: human cardnogan QroupB: probaMa human cardnogan B1: IMMavldanMfr nlcltyt tin. ••MiMb»*at««B* ^ajk^«M*^ fcwM«« ak*«lW«B«l •huX^M ••wfl MavAJ^^MtAfai11 — J-l-^——— __ IK, •unmvni •viiivncv vuiii WHIM mnw •no RVKIK|UWIV wuoiiuv or •__ J— 4^» ^^UBB aaaa^«^MBak*LMaVk « no on* ironi vpniMnKiiopjic Group C: pottlM* humin cvdnootn Group D: Group E: evhtenc« ol »oiK.Mdnog«n>cHy lor hutmn» UNITED STATES ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, D.C. 20460 OFFICE OF SOLID tttSTE AND EMERGENCY ..... RESPONSE MAY.| 7 1996 MEMORANDUM SUBJECT: Final Soil FROM: Elliott P. Assistant TO: Director, 'office of Site Remediation and Restoration Region I Director, Emergency and Remedial Response Division Region II Director, Hazardous Waste Management Division Regions III, IX Director, Waste Management Divisicr. Region IV Director, Superfund Division Regions V, VI, VII Assistant Regional Administrator, Office of Ecosystems Protection and Remediation Region VIII Director, Environmental Cleanup Office Region X This memorandum transmits the Soil Screening Guidance and < its supporting Technical Background Document for your immediate *\use as a tool at Superfund National Priorities List (NPL) sites. This guidance provides a simple method for calculating site- specific screening levels for chemicals commonly found at these sites. BZiifiTSo TUT 00 3 .1.121 BACKGROUND These screening levels can help EPA rapidly identify those areas at residential sites that do not warrant further remedial investigation or action at the national level. This guidance may not be appropriate for all sites and its use is not required. However, where a screening approach is useful and the assumptions and condition are appropriate/ this guidance should be applied. As part of our Administrator's charge in 1991 to accelerate the rate of cleanup at NPL sites, the Agency has been working to develop tools both to standardize and streamline decision making for contaminated soils. As part of this effort, the Agency aggressively sought input from a wide range of stakeholders, including other Federal and State Agencies, industry groups, environmental advocacy groups, auditors, insurers, lenders, and independent technical experts. Most agreed that while there was indeed a great need to standardize the current soil cleanup process, it was also important to allow appropriate consideration of site-specific information in order to ensure cost-effective decisions. Development of this guidance has been a resource intensive effort. In addition to a great deal of technical analysis/peer review, it also required a number of very difficult policy choices. This final guidance blends a wide range of "standard" approaches and assumptions with the flexibility to design a decision-making process based on site-specific conditions. Use of this guidance may result in greater national consistency, overall programmatic cost savings, and .a data collection/decision process that will be substantially more transparent to involved stakeholders. DI8CU88IOM The Soil Screening Guidance encourages collection of site- specific data that can make a significant difference in a Superfund soil cleanup strategy. Likewise, it is designed to help site managers avoid collecting unnecessary data from areas of a site, or on exposure pathways that are highly unlikely to .present * substantial human health threat. The guidance is intended to be used early in the RI/FS process whenever residential land use is likely. However, this guidance does not •replace the remedial investigation or baseline risk assessment Nihere screening indicates that further investigation is warranted. - 2 - TUT OO8 1122 The guidance is not designed to address commercial - industrial land use, nor to address potential ecological effects of soil contaminants. At specific sites, consideration of other exposures, land uses, or receptors may be very important; however, other guidance and analysis will be necessary for those purposes. This guidance may be useful in the context of RCRA Corrective Actions, economic redevelopment projects (brownfields), and voluntary cleanups, where residential land use is anticipated and human health exposures are of concern. However, misuse of this guidance may result in actions that are either unnecessarily expensive, or not adequately protective of human health and ecological receptors. Because of the widespread interest in this guidance beyond the boundaries of our own program, we have attempted to document both the issues and their resolution in enough detail so that interested readers might fully understand why the guidance is designed as it is. As a result, a great deal-of discussion on current site characterization and risk assessment methods has been presented. Accordingly, and in an effort to minimize complexity, we have issued this guidance in two parts: (1) Soil Screening Guidance: User's Guide (OSWER Directive 9355.4-23, PB96-9635Q5) designed for the RPM, OSC, or Regional manager who seeks to understand the basic concepts, approaches, and assumptions in the soil screening decision framework; and, (2) Soil Screening Guidance: Technical Background Document (TBD) (OSWER Directive 9355.4-17A, PB96-963502) that provides a comprehensive analysis of the technical or policy issues and cnoices. Please encourage your technical staff to read this TBD carefully. It contains update information, not found in other Superfund documents. In addition, a very brief nontechnical fact sheet, titled Soil Screening Guidance: Fact Sheet iu*>#ER Directive 9533.1- 14FSA, PB96-963501) is available and presents a general overview of the soil screening process. Finally/ an overview of our response to comments received during the public coaaent period is also available; this document is titled Soil Screening Guidance: v . Response to Comments (OSWER Directive 9355.4-22, PB96-963506) . Readers will notice that this guidance often relies on ^equations and assumptions derived froa the Risk Assessment "Guidance for Superfund (RAGS)/ Human Health Evaluation Manual (HHEM), Part B. In many cases/ this guidance updates the RAGS/ HHEM/ Part B models to reflect improvements in Superfund risk assessment guidance or clarifications in program policies. - 3 - TUT OOS 1123 It is clear from concerns raised during outreach sessions that there is potential for mis-use of this guidance. For this reason, I would like to highlight several important messages: • A foil Screening Level (SST ) derived according to this guidance is NOT a "cleanup standard," and only in limited situations should it be used as a "cleanup goal." An SSL is a concentration in soil which represents a level of contamination above which there is sufficient concern to warrant further site-specific s t udy . Concentrations in soil above these levels do NOT automatically designate a site as dirty/ nor gnat ion. trigger a response act • Generally/ where concentrations fall below the screening level/ no further study or response action would be warranted under CERCLA except when certain site conditions described in the guidance are present. • SSLs do not supersede existing Federal or State ARARs that may affect selection of cleanup levels for soils. IMPLEMENTATION Effective immediately, the attached guidance and its technical support document should be used where it can better focus our limited financial resources at Super fund NPL sites by screening soils from further study or action. If you or your staff have any questions, please contact your OERR Regional Accelerated Response Center. In addition, where Region-specific screening tools already exist that use tables of risk-based numbers or similar formulae or risk-based assumptions/ it is strong A y recommended that these tools be modified as appropriate to be consistent with this guidance. When applying a soil screening approach for NPL sites, the models, equations and assumptions presented in this guidance should be used to screen the need for further site-specific study at residential properties. Also/ as stated above/ the models used in this guidance relevant to residential land use supersede those used in the Risk Assessment Guidance for Superfund, HHEM — * Part B. The HHEM Part B models should no longer be used for the ""•development of risk-based preliminary remediation goals in residential settings. - 4 - TUT OO8 1124 Due to limitations on our printing budget, we are sending only two copies of this guidance to each Regional Office. I am asking that your Region provide managers and staff with copies as needed. Additional bound copies will be made available by Headquarters in the future when funding becomes available. We are also pursuing ••.•ays to make the guidance available in an electronic format. The public can purchase the documents mentioned above by calling the National Technical Information Service (NTIS) at 703- 487-4650 and requesting these documents by name or number. For further information, you may call David Cooper of OERR at 703-603-8763 or at «cooper.davide6epamail.epa.gov». Attachments - 5 - TUT 008 1125 cc: Stephen Luftig, OERR Michael Shapiro, OSW Gerald Clifford, OSRE Barry Breen, FFEO James Woolford, FFRRO Lisa Lund, OUST Tim Fields, OSWER William Farland, ORD Ramona Trayato, ORIA Larry Starfield, OGC Barry Johnson, ATSDR Regional Toxics Integration Coordinators Ground Water Forum, Chairperson cc (without attachments): Superfund Branch Chiefs, Regions I-X Superfund Section Chiefs, Regions I-X - 6 - TUT 008 1126 Unled States Environmental Protection Acency Office of Solid Waste and Emergency Response Washington, DC 20460 8355.4-17A EPA/54Q/R-95/128 PB96-963502 May 1996 Supertund SERA Soil Screening Guidance: Technical Background Document TUT O08 1127 Publication 9355.4-17A May 1996 Soil Screening Guidance: Technical Background Document Office of Emergency and Remedial Response U.S. Environmental Protection Agency Washington, DC 20460 TUT 008 1128 DISCLAIMER Notice: The Soil Screening Guidance is based on policies set out in the Preamble to the Final Rule of the National Oil and Hazardous Substances Pollution Contingency Plan (NCP), which was published on March 8, 1990 (55 Federal Register 8666). This guidance document sets forth recommended approaches based on EPA's best thinking to date with respect to soil screening. Alternative approaches for screening may be found to be more appropriate at specific sites (e.g., where site circumstances do not match the underlying assumptions, conditions, and models of the guidance). The decision whether to use an alternative approach and a description of any such approach should be placed in the Administrative Record for the site. The policies set out in both the Soil Screening Guidance: User's Guide and the supporting Soil Screening Guidance: Technical Background Document are intended solely as guidance to the U.S. Environmental Protection Agency (EPA) personnel; they are not final EPA actions and do not constitute rukmalting. These policies are not intended, nor can they be relied upon, to create any rights enforceable by any party in litigation with the United States government EPA officials may decide to follow the guidance provided in this document, or to act at variance with the guidance, based on an analysis of specific site circumstances. EPA also reserves the right to change the guidance at any time without public notice. u TUT O08 TABLE OF CONTENTS Section Page Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii List of Tables ............................................................ . vi List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii List of Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Pidace ..............................................."................. .ix Acknowledgments ......................................................... x Part 1: Introduction 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Purpose of SSLs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Scope of Soil Screening Guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3.1 Exposure Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3.2 Exposure Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3.3 Risk Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3.4 SSL Model Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4 Organiratifm of the Tkvjimgrrt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Part 2: Development of Pathway-Specific Soil Screening Lavals 2.1 Human Health Basis ................................. :.................... 9 2.1.1 Additive Risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1.2 Apportionment and Fractionation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.1.3 Acute Exposures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.1.4 Route-to-Route Extrapolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2 Direct Ingestkm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3 Dermal Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4 Inhalation of Volatiles and Fugitive DttStS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.4.1 Screening Level Equations for Direct Inhalation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.4.2 Volatilization Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.3 Dispersion Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.4.4 Soil Saturation Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 2.4.5 Paniculate Emission Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.5 Migration to Ground Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.5.1 Development of Soil/Water Partition Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.5.2 •Organic Compounds—Partition Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.5.3 Inorganics (Metals)—Partition Theory ................................... 40 2.5.4 Assumptions for Soil/Water Partition Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . : . . . 40 2.5.5 Diluncm/Attenuation Factor Developmenl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.5.6 Default Dilution-Attenuation Factor ..................................... 46 2.5.7 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 2.6 Mass-Limit Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 2.6.2 Migration to Ground Water Mass-Limn Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 2.6.3 Inhalation Mass-Limit Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 2.7 Plant Uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.8 intrusion of Volatiles into Basements: Johnson and EttmgerModel . . . . . . . . . . . . . . . . . . . . . . 62 iii TUT 008 1130 TABLE OF CONTENTS (continued) Section , Page Part 3: Modals for Detailed Awessment 3.1 Inhalation of Volatile*: Detailed Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 3.1.1 Finite Source Volatilization Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 3.1.2 Air Dispersion Models .............................................. 66 3.2 Migration to Ground Water Pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.2.1 Saturated Zone Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 3.2.2 UnsaturatedZone Models .....;...................................... 68 Part 4: Measuring Contaminant Concentrations in Soil 4.1 4.2 4.3 Sampling Surface Sofls , . . . . . , , . . . , . . . . . . . . . . . . . . . 4.1.1 State the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2 Identify the Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413 Identify Inputs to the Decision . , , . . ... . , . . . . . . . . . 4 1 4 TVfjn* th* Stnrfy RnvndariM 4.1.5 Develop a Decision Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.6 Specify Limits on Decision Errors for the Max Test . . . . . . . . . . . . . . . . . . . 4.1.7 Optimize the Design for the Max Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.8 Using the DQA Process. Analyzing Max Test Data . . . . . . . . . . . . . . ... . . . 4.1.9 Specify Limits on Decision Errors for Chen Test . . . . . . . . . . . . . . . . . . . . . 4". 1.10 Optimize the Design Using the Chen Test . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.11 Using the DQA Process: Analyzing Chen Test Data . . . . . . . . . . . . . . . . . . . 41 12 Special rnnsirtrratinm for M^plf rnnt^Tninanl<: 4.1.13 Quality Assurance/Quality Control Requirements . . . . . . . . . . . . . . . . . . . . . 4 1 1 4 Final Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.15 Reporting . . . : . , , . , , . , . . . , , . . , . . . . . . , . . . , . . , . . . , . . . . . Sampling Subsurface Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . : . . . . . . . . 4.2. 1 State the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Identify the Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 Identify Inputs to the Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.24 Define t h e Study Boundaries . . . . . . . . 4.2.5 Develop a Decision Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.6 Specify Limits on Decision Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.7 Optimize the Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.8 Analyzing the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.9 Reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B?CT<:fnrthi. Suffare So'1 Sanipling Strategies Technical Analyses Performed 4 7 1 1QQ4.T>faft Guidance Sampling Strategy 4.3.2 Test of Proportion Exceeding a Threshold . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.3 Relative Performance of Land, Max, and Chen Tests . . . . . . . . . . . . . . . . . . . 4.3.4 Treatment of Observations Below the Limit of Quantitation . . . . . . . . . . . . . . 4.3.5 Multiple Hypothesis Testing Considerations . . . . . . . . . . . . . . . . . . . . . . . . 4.3.6 Investigation of Compositing Within EA Sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 . . . . . . . 82 . . . . . . . 82 . . . . . . . 84 . . . . . . . 84 ....... 85 . . . . . . . . 86 . . . . . . . . 87 . . . . . . . 96 . . . . . . . . 99 . . . . . . . 100 . . . . . . . 107 . . . . . . . 107 . . . . . . . 107 . . . . . . . 109 . . . . . . . 109 . . . . . . . 110 . . . . . . . 110 . . . . . . . 110 . . . . . . . 110 . . . . . . . 114 . . . . . . . 114 . . . . . . . 114 . . . . . . . 115 . . . . . . . 116 . . . . . . . 116 . . . . . . . 117 . . . . . . . 117 . . . . . . . 119 . . . . . . . . 121 . . . . . . . . 127 . . . . . . . . 127 . . . . . . . . 129 IV TUT TABLE OF CONTENTS (continued) ' Section Page Part 5: Chemical-Specific Parameters 5.1 Solubility, Henry's Law Constant, and K«w . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 5.2 Air (Dj^) and Water (Pi,w)Diflfasivitics . . . . . . . . . . . . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . . . . . 133 5.3 Soil Organic Carbon/Water Partition Coefficients (K«e) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 5.3.1 KOC for Nomorrizing Organic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 5.3.2 KOC for Ionizing Organic Compounds ................................... 145 5.4 Soil-Water Distribution Coefficients (K<|) for Inorganic Constituents . . . . . . . . . . . . . . . . . . . . . . 149 5.4.1 Modeling Scope and Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.4.2 Input Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 5.4.3 Assumptions a"d Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 5.4.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 5.4.5 Analysis of Peer-Review Comments . . . . . . . . . . . . . . . . : . . . . . . . . . . . . . . . . . . . . 160 Part 6: References References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ' . . . . . . . . . . . . . . . . . - • 161 Appendices . A Generic SSLs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-l fi Route-tO-Route Extrapolation nf Inhalation Renehmatfcs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-l C Limited Validation of the Jury Infinite Source and Jury Finite Source Models (EQ, 1995) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ; . C-l D Revisions to VF and PEF Equations (EQ, 1994b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D-l E Determination of Ground Water Dilution Attenuation Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . • E-l F Dilution Factor Modeling Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F-l G Background Discussion for Soil-Plant-Human Exposure Pathway . . . . . . . . . . . . . . . . . . . . . . . . . G-l H Evaluation of the Effect on the Draft SSLs of the Johnson and Ettinger Model (EQ, 1994a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H-l I SSL Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-1 J Piazza Road Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1 K Soil Organic Carbon (Kec) / Water (K,*,) Partition Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . K-l L KOC Values for Ionizing Organics as a Function of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L-l M Response to Peer-Review Comments on MINTCQA2 Model Results ...................... M-l TUT ooe 1137 LIST OF TABLES 'Table 1. Regulatory and Human Health Benchmarks Used for SSL Development . . . . . . . . . . . . . . . . . . 10 Table 2. SSL Chemicals withNoncatcinogemc Effects on Specific Target Organ/System . . . . . . . . . . . 15 Table 3. Q/C Values by Source Area, City, and Climatic Zone . . . . . . . . . . . . ' . . . . . . . . . . . . . . . . 27 Table 3-A. Risk Levels Calculated at Csat for Contaminants that have SSLinh Values Greater than C^ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Table 4. Physical Slate of Organic SSL Chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Table 5. Variation of DAF with Size of Scarce Aiea for SSL EPACMTP Modeling Effort .......... 48 Table 6. Recharge Estimates for DNAPL Site Hydrogeologic Regions ......................... 50 Table 7. SSL Dilation Factor Model Results DNAPL and HGDB Sites . . . . . . . . . . . . . . . . . . . . . . 51 Table 8. Sensitivity Analysis for SSL Partition Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Table 9. Sensitivity Analysis for SSL Dilution Factor Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Table 10. Input Parameters Required for RTTZ M o d e l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Table 11. Input Parameters Required for VD* Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Table 12. Input Parameters Required for CMLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Table 13. Input Parameters Required for HYDRUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Table 14. Input Parameters Required for SUMMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Table 15. Input Parameters Required for MULTIMED ...........................:........ 71 Table 16. Input Parameters Required for VLEACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Table 17. Input Parameters Required for SESOIL (Monthly Option) . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Table 18. Input Parameters Required for PRZM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 Table 19. Input Parameters Required for VADOFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Table 20. Characteristics of Unsaturated Zone Models Evaluated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Table 21. Sampling Soil Screening DQOs for Surface Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Table 22. Sampling Soil Screening DQOs for Surface Soils under the Max Test . . . . . . . . . . . . . . . . . . . 86 Table 23. Probability of Decision Error tat 0.5 SSL and 2 SSL Using Max Test . . . . . . . . . . . . . . : . . . , 93 Table 24. Sampling Soil Screening DQOs for Surface Soils under Chen Test . . . . . . . . . . . . . . . . . . . . . 99 Table 25. Minimum Sample Size for Chen Test at 10 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Table 26. Minimum Sample Size for Chen Test at 20 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Table 27. Minimum Sample Size for Chen Test at 40 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Table 28. Minimum Sample Size for Chen Test at 10 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given the Expected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Table 29. Minimum Sample Size for Chen Test at 20 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Table 30. Minimum Sample Size for Chen Test at 40 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, GtvenExpected CV for Concentrations Across the EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Table 31. Soil Screening DQOs for Subsurface Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Table 32. Comparison of Error Rates for Max Test, Chen Test (at .20 and .10 Significance Levels), and Original Land Test, Using 8 Composites of 6 Samples Each, for Gamma Contamination Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 VI TUT U.33 LIST OF TABLES (continued) Table 33. Enor Rates of Max Test and Chen Test at ^ (C20) and. 1 (CIO) Significance Level for CV « 2,25,3,3.5, C - # of Specimens per Composite, N - # of Composite Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Table 54. Probability of "Walking Away" from an EA When Comparing Two Chemicals to SSLs ....... 127 Table 35. Means and CVs for Dioxin Concentrations for 7 Piazza Road Exposure Areas . . . . . . . . . . . . . . 129 Table 36. Chemical-Specific Properties Used in SSL Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Table 37. Air Diffusiviry (D^ and Water Diffusivity (Dj,w) Values for SSL Chnmicals (25°Q . . . . . . . . . . 135 Table 38. Summary Statistics for Measured KOC Vames: Nbnionizing Orgamcs . . . . . . . . . . . . . . . . . . . . 139 Table 39. Comparison of Measured and Calculated KOC Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Table 40. Degree of lomzation (Fraction of Neutral Species, F) as a Function of pH . . . . . . . . . . . . . . . . . . 145 Table 41. Soil Organic Carbon/Water Partition Coefficients and pKa Vames for Ionizing Organic Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Table 42. Predicted Soil Organic CarbonWaterPartm'on Coefficients (KwJAg) as a Function of pH: Ionizing Organics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Table 43. Summary of Collected K<j Values Reported in Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Table 44. Summary of Geochemical Parameters Used in SSL MINTEQ Modeling E f f o r t . . . . . . . . . . . . . 151 Table 45. Background Pore-Water Chemistry Assumed for SSL MINTEQ Modeling Effort . . . . . . . . . . . . 152 Table 46. Estimated Inorganic K<t Values for SSL Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 vu TUT OO8 LIST OF FIGURES Figure 1. Conceptual Risk Management Spectrum for Contaminated Soi . . . . . . . . . . . . . . . . . . . . . . . 2 Figure 2. Exposure Pathways Addressed by SSLs. ...................................... 4 Figure 3. Migration to ground water pathway—-EPACMTP modeling effort. . . . . . . . . . . . . . . . . . . . . 46 Figure 4. The Data Quality Objectives process.......................................... 80 Figure 5. Design performance goal diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Figure 6. Systematic (square grid points) sample with systematic compositing scheme (6 composite samples consisting of 4 specimens). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Figure?. Systematic (square grid points) sample with random compositing scheme (6 composite samples consisting of 4 specimens). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Figure *'•. Stratified random sample with random compositing scheme (6 composite samples consisting of 4 specimens). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Figure 9. U.S. Department of Agriculture soil texture classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure 10. Empirical pH-dependent adsorption relationship: arsenic (+3), chromium (-H>), selenium, thallium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Figure 11. Metal Kj as a function of pH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 LIST OF HIGHLIGHTS Highlight 1. Key Attributes of the Soil Screening Guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Highlight 2: Simplifying Assumptions for the Migration to Ground Water Pathway . . . . . . . . . . . . . . . . . 34 Highlight 3: Procedure for Compositing of Specimens from a Grid Sample Using a Systematic Scheme (Figure 6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Highlight 4. Procedure for Compositing of Specimens from a Grid Sample Using a Random Scheme (Figure 7) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Highlight 5: Procedure for Compositing of Specimens from a Stratified Random Sample Using a Random Scheme (Figure 8) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Highlight 6: Directions for Data Quality Assessment for the Max Test . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Highlight?: Directionsfor the Chen Test Using Simple Random Sample Scheme . . . . . . . . . . . . . . . . . . 106 Vlll TUT 008 1135 PREFACE This document provides the technical background for the development of methodologies described in the Soil Screening Guidance: User's Guide (EPA/540/R-96/018), along with additional information useful for soil screening. Together, these documents define the framework and methodology for developing Soil Screening Levels (SSLs) for chemicals commonly found at Superfund sites. This document is an updated version of the background document developed in support of the December 30,1994, draft Soil Screening Guidance. The methodologies described in this document and the guidance have been revised in response to public comment and extensive peer review. The revisions, along with other technical analyses conducted to address the comments, are described herein. This background document is presented in five parts. Part 1 describes the sofl screening process and its application and implementation at Superfund sites. Part 2 describes the methodology used to develop SSLs, including the assumptions and theories used. Part 3 provides information on more detafled models that may be used to develop site-specific SSLs. Part 4 addresses sampling schemes for manning soil contaminant levels during the soil screening process. Pan 5 provides technical background on the determination of chemical-specific properties for calculating SSLs. IX TUT 008 1136 ACKNOWLEDGMENTS This technical background document was prepared by Research Triangle Institute (RTI) under EPA Contract 68- Wl-0021, Work Assignment D2-24, for the Office of Emergency and Remedial Response (OERR), U.S. Environmental Protection Agency (EPA). Janine Dinan and Loren Henning of EPA, the EPA Work Assignment Managers for this effort, guided the effort and are also principal EPA authors of the document along with Sherri Clark of EPA. Robert Truesdale is the RTI Work Assignment Leader and principal RTI author of the document. Craig Mann of Environmental Quality Management, Inc. (EQ), conducted the Fueling effort for the inhalation pathway and provided background information on that effort Dr. Zubair Sateem of EPA's Office of Solid Waste conducted the EPACMTP modeling effort and provided the discussion on the use of this model for generic DAF development. The. authors would like to thank all EPA, State, public, and peer reviewers whose careful review and thoughtful comments greatly contributed to the quality of this document. Technical support for the final document production was provided by Dr. Smita Siddhanti of Booz*Alien & Hamilton. UT 000 1 1 ~r "" -*~ ••'_ •„,» Part 1: INTRODUCTION This document provides the technical background for the Soil Screening Guidance. Hie Soil Screening Guidance is a tool that the U.S. Environmental Protection Agency (EPA) developed to help standardize and accelerate the evaluation and cleanup of contaminated soils at sites on the National Priorities last (NPL) with anticipated future residential land use scenarios.! This guidance provides a methodology for environmental science/engineering professionals to calculate risk-based, site- specific, soil screening levels (SSLs), for contaminants in soil mat may be used to identify areas needing further investigation at NPL sites. SSLs are not national cleanup standards. SSLs alone do not trigger the need for response actions or define "unacceptable" levels of contaminants in soil. "Screening," for the purposes of this guidance, refers to the process of identifying and defining areas, contaminants, and conditions at a particular site that do not require further Federal attention. Generally, at sites where contaminant concentrations fall below SSLs, no further action or study is warranted under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA). (Some States have developed screening numbers or methodologies that may be rrore stringent than SSLs; therefore further study may be warranted under State programs.) Where contaminant concentrations equal or exceed the SSLs, further study or investigation, but not necessarily cleanup, is warranted. The Soil Screening Guidance provides a framework for screening contaminated soils that encompasses both simple and more detailed approaches for calculating site-specific SSLs, and generic SSLs for use where site-specific data are limited. The Soil Screening Guidance: User's Guide (U.S.. EPA, 1996) focuses on the application of vie simple site-specific approach by providing a step-by- step methodology to calculate site-specific SSLs and plan the sampling necessary to apply them. This Technical Background Document describes the development and technical basis of the methodology presented in the User's Guide. It includes detailed modeling approaches for developing screening levels that can take into account more complex site conditions than the simple site- specific methodology emphasized in the User's Guide. It also provides generic SSLs for the most common contaminants found at NPL sites. 1.1 Background The Soil Screening Guidance is the result of technical analyses and coordination with numerous stakeholders. The effort began in 1991 when the EPA Administrator charged the. Office of Solid Waste and Emergency Response (OSWER) with conducting a 30-day study to outline options for accelerating the rate of cleanups at NPL sites. One of the specific proposals of the study was for OSWER to "examine the means to develop standards or guidelines for contaminated soils." Over the past 4 years, several drafts of the guidance and the accompanying technical background document have had widespread reviews both within and outside EPA. In the Spring of 1995, final drafts were released for public comment and external scientific peer review. Many reviewers' comments contributed significantly to the development of this flexible tool that uses site-specific data in a methodology that can be applied consistently across the nation. 1. Note that the Superfund program defines "soil" as having a particle size under 2 millimeters, while the RCRA piogiaiu allows for particles under 9 millimeters in size. 1 TUT' OO8 :( 1 "i. ••** • 1.2 Purpose of SSLs In identifying and managing risks at sites, EPA considers a spectrum of contaminant concentrations. The level of concern associated with those concentrations depends on the likelihood of exposure to soil contamination at levels of potential concern to human health or to ecological receptors. Figure 1 illustrates the spectrum of soil contamination encountered at Superfund sites and the conceptual range of risk management. At one end are levels of contamination that clearly warrant a response action; at the other end are levels that are below regulatory concern. Appropriate cleanup goals for a particular site may fall anywhere within this range depending on she-specific conditions. Screening levels identify the lower bound of the spectrum — levels below which there is no concern under CERCLA, provided conditions associated with the SSLs are met. No further study warranted under CERCLA Site-specific cleanup goal/level Response action clearly warranted •Zero" concentration Screening level Response level Very high concentration Figure 1. Conceptual Risk Management Spectrum for Contaminated Soil Although the application of SSLs during site investigations is not mandatory at sites being addressed by CERCLA or RCRA, EPA recommends the use of SSLs as a tool to facilitate prompt identification of contaminants and exposure areas of concern. EPA developed the Soil Screening Guidance to be consistent with and to enhance the current Superfund investigation process and anticipates its primary use during the early stages of a remedial investigation (RI) at NPL sites. It does not replace the Remedial Investigation/Feasibility Study (RI/FS) or risk assessment, but use of screening levels can focus the RI and risk assessment on aspects of the site that are more likely to be a concern under CERCLA. By screening out areas of sites, potential chemicals of concern, or exposure pathways from further investigation, she managers and technical experts can limit the scope of the remedial investigation or risk assessment. SSLs can save resources by helping to determine which areas do not require additional Federal attention early in the process. Furthermore, data gathered during the soil screening process can be used in later Superfund phases, such as the baseline risk assessment, feasibility study, treatabilhy study, and remedial design. This guidance may also be appropriate for use by the removal program when demarcation of soils above residential risk-based numbers coincides with the purpose and scope of the removal action. EPA created the Soil Screening Guidance to be consistent with and to enhance current Superfund processes. The process presented in mis guidance to develop and apply simple, she-specific soil screening levels is likely to be most useful where h is difficult to determine whether areas of soil are contaminated to an extent that warrants further investigation or response (e.g., whether areas of soil at an NPL site require further investigation under CERCLA through an RI/FS). The screening levels have been developed assuming future residential land use assumptions and related exposure scenarios. Although some of the models and methods presented in this guidance could be modified to address exposures TUT 008 1139 under other land uses, EPA has not yet standardized assumptions for those other uses. Using this guidance for sites where residential land use assumptions do not apply could result in overly conservative screening levels. However, EPA recognizes that some parties responsible for sites with non-residential land use might still benefit from using SSLs as a tool to conduct conservative initial screening. EPA created the Soil Screening Guidance: User's Guide (U.S. EPA, 1996) to be easy to use: it provides a simple step-by-step methodology for calculating SSLs thai are specific to the user^s site. Applying she-specific screening levels involves developing a conceptual site model (CSM), collecting a few easily obtained site-specific soil parameters (such as the dry bulk density and percent soil moisture), and sampling soil to measure contaminant levels in surface and subsurface soils. Often, much of the information needed to develop the CSM can be derived from previous site investigations (e.g., the preliminary assessment/site inspection [PA/SI]) and, if properly planned, SSL sampling can be accomplished in one mobilization. SSLs can be used as Preliminary Remediation Goals (PRGs) provided appropriate conditions are met (i.e., conditions found at a specific site are similar to conditions assumed in developing the SSLs). The concept of calculating risk-based soil levels for use as PRGs (or "draft" cleanup levels) was introduced in the Risk Assessment Guidance for Superfund (RAGS), Volume I,. Human Health Evaluation Manual (HHEM), Part B (U.S. EPA, 1991b). PRGs are risk-based values that provide a reference point for establishing she-specific cleanup levels. The models, equations, and assumptions presented in the Soil Screening Guidance and described herein to address inhalation exposures supersede those described in RAGS HHEM, Part B, for residential soils. In addition, this guidance presents methodologies to address the leaching of contaminants through soil to an underlying potable aquifer. This pathway should be addressed in the development of PRGs. EPA emphasizes that SSLs are not cleanup standards. SSLs should not be used as she-specific cleanup levels unless a she-specific nine-criteria evaluation using SSLs as PRGs for soils indicates that a selected remedy achieving the SSLs is protective, compliant with applicable or relevant and appropriate requirements (ARARs), and appropriately balances the other criteria, including cost. PRGs may then be converted into final cleanup levels based on the nine-criteria analysis described in the National Contingency Plan (NCP; Section 300.430 (3)(2)(A)). The directive entitled Role of the Baseline Risk Assessment in Superfund Remedy Selection Decisions (U.S. EPA, 1991c) discusses the modification of PRGs to generate cleanup levels. The generic SSLs provided in Appendix A are calculated from the same equations used in the simple site-specific methodology, but are based on a number of default assumptions chosen to be protective of human health for most site conditions. Generic SSLs can be used in place of site-specific screening levels; however, they are expected to be generally more conservative than she-specific levels. The she manager should weigh the cost of collecting the data necessary to develop she-specific SSLs with the potential for deriving a higher SSL that provides an appropriate level of protection. 1.3 Scope of Soil Screening Guidance The Soil Screening Guidance incorporates readily obtainable site data into simple, standardized equations to derive she-specific screening levels for selected contaminants and exposure pathways. Key attributes of the Soil Screening Guidance are given in Highlight 1. TUT COS 1.14O Highlight 1: toy Attributes of the Soil Serening Guidance • Standardized equations are presented to address human exposure pathways in a residential setting consistent with Superfund's concept of 'Reasonable Maximum Exposure" (RME). • Source size (area and depth) can be considered on a she-specific basis using mass-limit models. • Parameters are identified for which site-specific information is needed to develop she-specific SSLs. Default values are provided to calculate generic SSLs where site-specific information is not available. • SSLs are generally based on a 10* risk for carcinogens, or a hazard quotient of 1 for noncarcinogens; SSLs for migration to ground water are based on (in order of preference): nonzero maximum contaminant level goals (MCLGs), maximum contaminant levels (MCLs), or the aforementioned risk-based targets. 1.3.1 Exposure Pathways. In a residential setting, potential pathways of exposure to contaminants in soil are as follows (see Figure 2): • Direct ingestion • Inhalation of volatiles and fugitive dusts • Ingestion of contaminated ground water caused by migration of chemicals through soil to an underlying potable aquifer • Dermal absorption . • Ingestion of homegrown produce that has been contaminated via plant uptake • Migration of volatiles into basements The Soil Screening Guidance addresses each o: these pathways to the greatest extent practical. The first three pathways — direct ingestion, inhalation of volatiles and fugitive dusts, and ingestion of potable grounc water, are the most common routes of human exposure to contaminants in the residential setting. These pathways have generally accepted methods, models, and assumptions that lend themselves to a standardized approach. The additional pathways of exposure to soil contaminants, dermal absorption, plant uptake, and migration of volatiles into basements, may also contribute to the risk to human health from exposure to specific contaminants in a residential setting. This guidance addresses these pathways to a limited extent based on available empirical data (see Part 2 for further discussion). Direct Ingestion of Ground Water and Soil Blowing. Dust Volatilization Also Addressed: • Plant Uptake • Dermal Absorption Figure 2. Exposure Pathways Addressed by SSLs. TUT oos 1141 The Soil Screening Guidance addresses the human exposure pathways listed previously and will be appropriate for most residential settings. The presence of additional pathways or unusual site conditions does not preclude the use of SSLs in areas of the site that are currently residential or likely to be residential in the future. However, the risks associated with these additional pathways or conditions (e.g., fish consumption, raising of .livestock, heavy truck traffic on unpaved roads) should be considered in the remedial investigation/feasibility study (RI/FS) to determine whether SSLs are adequately protective. An ecological assessment should also be performed as part of the RI/FS to evaluate poten- tial risks to ecological receptors. The Soil Screening Guidance should not be used for areas with radioactive contaminants. 1 .3.2 Exposure Assumptions. SSLs are risk-based concentrations derived from equations combining exposure assumptions with EPA toxicity data. The models and assumptions used to calculate SSLs were developed to be consistent win Superfund's concept of "reasonable maximum exposure" (RME) in the residential setting. The Superfund program's method to estimate the RME for chronic exposures on a site-specific basis is to combine an average exposure point concentration with reasonably conservative values for intake and duration in the exposure calculations (U.S. EPA, 1989b; U.S. EPA, 199 la). The default intake and duration assumptions presented in U.S. EPA (199 la) were chosen to represent individuals living in a small town or other nontransient community. (Exposure to members of a more transient community is assumed to be shorter and thus associated with lower risk.) Exposure point concentrations are either measured at the site (e.g., ground water concentrations at a receptor well) or estimated using exposure models with site-specific model inputs. An average concentration term is used in most assessments where the focus is on estimating long-term, chronic exposures. Where the potential for acute toxicity is of concern, exposure estimates based on maximum concentrations may be more appropriate. The resulting site-specific estimate of RME is then compared with a chemical-specific toxicity criterion such as a reference dose (RfD) or a reference concentration (RfC). EPA recommends using criteria from the Integrated Risk Information System (IRIS) (U.S. EPA, 199Sb) and Health Effects Assessment Summary Tables (HEAST) (U.S. EPA, 1995d), although values from other sources may be used in appropriate cases. SSLs are concentrations of contaminants in soil that are designed to be protective of exposures in a residential setting. A site-specific risk assessment is an evaluation of the risk posed by exposure to site contaminants in various media. To calculate SSLs, the exposure equations and pathway models are run in reverse to backcalculate an "acceptable level" of a contaminant in soil corresponding to a specific level of risk. 1.3.3 Risk Level. For the ingestion, dermal, and inhalation pathways, toxicity criteria are used to define an acceptable level of contamination in soil, based on a one-in-a-milUon (10-6) individual excess cancer risk for carcinogens and a hazard quotient (HQ) of 1 for non-carcinogens. SSLs are backcalculated for migration to ground water pathways using ground water concentration limits [nonzero maximum contaminant level goals (MCLGs), maximum contaminant levels (MCLs), or health-based limits (HBLs) (10-* cancer risk or a HQ of 1) where MCLs are not available]. The potential for additive effects has not been "built in" to the SSLs through apportionment. For carcinogens, EPA believes mat setting a 10-* risk level for individual chemicals and pathways will generally lead to cumulative risks within the risk range (1(M to 10-6) for the combinations of TUT OOa 1142 chemicals typically found at Superfund sites. For noncarcinogens, additive risks should be considered only for those chemicals with the same toxic endpoint or mechanism of action (see Section 2.1). 1.3.4 SSL Model Assumptions. The models used to calculate inhalation and migration to ground water SSLs were designed for use at an early stage of site investigation when site •information may be limited. Because of this constraint, they incorporate a number of simplifying assumptions. The models assume that the source is infinite. Although the assumption is highly conservative, a finite source model cannot be applied unless there are accurate data regarding source size and volume. EPA believes h to be unlikely that such data will be available from the limited subsurface sampling that is done to apply SSLs. However, EPA also recognizes mat infinite source models can violate mass balance (i.e., can release more contaminants than are present) for certain contaminants and site conditions (e.g., small sources). To address this problem, mis guidance includes simple models that provide a mass-based limit for the inhalation and migration to ground water SSLs (see Section 2.6). A site-specific estimate of source depth and area are required to calculate SSLs using these models. The infinite source assumption leads to several other simplifying assumptions. Fractionation of contaminant mass between the inhalation and migration to ground water pathways cannot be addressed with infinite source models. For the migration to ground water pathway, an infinite source overrides adsorption in the unsaturated zone or in the aquifer. The models also assume that contamination is evenly distributed throughout the source (i.e., homogeneous) and mat no biological or cher: cal degradation occurs in the soil or in the aquifer. Again, models capable of addressing heterogeneities or degradation processes require collection of she-specific data that is well beyond the scope of the Soil Screening Guidance. Although' the Soil Screening Guidance encourages.the use of site-specific data to calculate SSLs, conservative default parameters are provided for use where site-specific data are not available. These defaults are described in Part 2 of this document. Appendix A provides an example set of "generic" SSLs for 110 chemicals that are calculated using these defaults. Because they are designed to be protective of most site conditions across the nation, they are conservative. A default 0.5 acre source area is used to calculate the generic SSLs. A 30 acre source size was used in the December 1994 guidance. EPA received an overwhelming number of comments that suggest that most contaminated soil sources addressed under the Superfund program are 0.5 acres or smaller. Because of the infinite source assumption, generic SSLs based on a 0.5 acre source size can be protective of larger sources as well (see Appendix A). However, this hypothesis should be examined on a case-by-case basis before applying the generic SSLs to sources larger than 0.5 acre. 1.4 Organization of the Document Pan 2 of this document describes the development of the simple equations used to calculate SSLs. It describes and supports the assumptions behind these equations and presents the results of analyses conducted to develop the SSL methodology. Some of the more sensitive parameters are identified for which she-specific data are likely to have a significant impact. Default values are provided along with their sources and limitations. Part 3 presents information on other, more complex models that can be used to calculate inhalation and migration to ground water SSLs when more extensive she data are available or can be obtained. "TUT 008 U43 Some of these models can consider a finite source and fractionation between exposure pathways. They also can model more complex site conditions than the simple SSL equations, including conditions that can lead to higher, yet still protective, SSLs (e.g., thick umatnratftd zones, biological and chemical degradation, layered soils). .Part 4 provides the technical background for the development of the soil sampling design methodology for SSL application. It addresses methods for surface soil, including a test based on a maximum soil composite sample and the Chen method, which allows decision errors to be controlled. Part 4 also provides simulation results that measure the performance of these methods and sample size tables for different contaminant distributions and compositing schemes. Step-by-step guidance is provided for developing sample designs using each statistical procedure. Part 5 describes the selection and development of the chemical properties used to calculate SSLs. TUT OOS 1144 Part 2: DEVELOPMENT OF PATHWAY-SPECIFIC SOIL SCREENING LEVELS This part of the Technical Background Document describes the methods used to calculate SSLs for residential exposure pathways, along with their technical basis and limitations associated with their use. Simple, standardized equations have been developed for three common exposure pathways at Superfund sites: \ ' • Ingestion of soil (Section 22) • Inhalation of volatiles and fugitive dust (Section 2.4) • Ingestion of contaminated ground water caused by migration of contaminants through soil to an underlying potable aquifer (Section 2.5). The equations were developed under the following constraints: • They should be consistent with current Superfund risk assessment methodologies and guidance. • ' To be appropriate for early-stage application, they should be simple and easy to apply. • They should allow the use of site-specific data where they are readily available or can be easily obtained. • The process of developing and applying SSLs should generate information that can be used and built upon as a site evaluation progresses. The equations for the inhalation and migration to ground water pathways include easily obtained site- specific input parameters. Conservative default values have been developed for use where site-specific data are not available. Generic SSLs, calculated for 110 chemicals using these default values, are presented in Appendix A. The generic SSLs are conservative, since the default values are designed to be protective at most sites across the country. The inhalation and migration to ground water pathway equations assume an infinite source. As pointed out by several commenters to the December 1994 draft Soil Screening Guidance (U.S. EPA, 1994h), SSLs developed using these models may violate mass-balance for certain contaminants and site conditions (e.g., small sources). To address this concern, EPA has incorporated simple mass-limit models for these pathways aysmning that the entire volume of contamination either volatilizes or leaches over the duration of exposure and that the level of contaminant at the receptor does not exceed the health-based limit (Section 2.6). Because they require a site-specific estimate of source depth, these models cannot be used to calculate generic SSLs. Dermal adsorption, consumption of garden vegetables grown .in contaminated soil, and migration of voiatiles into basements also may contribute significantly to the risk to human health from exposure to soil contaminants in a residential setting. These pathways have been incorporated into the Soil Screening Guidance to the greatest extent practical. TUT 008 1145 Although methods for quantifying dermal exposures are available, their use for calculating SSLs is limited by the amount of data available on dermal absorption of specific chemicals (Section 2.3). Screening equations have been developed to estimate human exposure from the uptake of soil contaminants by garden plants (Section 2.7). As with dermal absorption, the number of chemicals for which adequate empirical data on plant uptake are limited. An approach to address migration of volatiles into basements is presented in Section 2.8, and limitations of the approach are discussed. 'Section 2.1 describes the human health basis of the Soil Screening Guidance and provides the human toxicity and health benchmarks necessary to calculate SSLs. The selection and development of the chemical properties required to calculate SSLs are described in Part 5 of this document. » 2.1 Human Health Basis Table 1 lists the regulatory and human health benchmarks necessary to calculate SSLs for 110 chemicals including: • • Ingestion SSLs: oral cancer slope factors (SF0) and noncancer reference doses (RfDs) • Inhalation SSLs: inhalation unit risk factors (URFs) and reference concentrations (RfCs) • Migration to ground water SSLs: drinking water standards (MCLGs and MCLs) and drinking water health-based levels (HBLs). The human health benchmarks in Table 1 were obtained from IRIS (U.S. EPA, 199Sb) or HEAST (U.S. EPA, 1995d) unless otherwise indicated. MCLGs and MCLs were obtained from U.S. EPA (1995a). Each of these references is updated regularly. Prior to calculating SSLs, the values in Table 1 should be checked against the most recent version of these sources to ensure that they are up-to-date. 2.1.1 Additive Risk. For soil ingestion and inhalation of volatiles and fugitive dusts, SSLs correspond to a 10*6 risk level for carcinogens and a hazard quotient of 1 for noncarcinogens. For carcinogens, EPA believes that setting a 10-6 risk level for individual chemicals and pathways generally will lead to cumulative risks within the 1CH to 1(H range for the combinations of chemicals typically found at Superfund sites. Whereas the carcinogenic risks of multiple chemicals are simply added together, the issue of additive risk is much more complex for noncarcinogens because of the theory that a threshold exists for noncancer effects. This threshold level, below which adverse effects are not expected to occur, is the basis for EPA's RfD and RfC. Since adverse effects are not expected to occur at the RfD or RfC and the SSLs were derived by setting the potential exposure dose equal to the RfD or RfC (i.e., an HQ equal to 1), it is difficult to address the risk of exposure to multiple chemicals at levels where the individual chemicals alone would not be expected to cause any harmful effect. However, problems may arise when multiple chemicals produce related toxic effects. EPA believes, and the Science Advisory Board (SAB) agrees (U.S. EPA, 1993e), that HQs should be added only for those chemicals with the same toxic endpoint and/or mechanism of action. 008 Table 1. Regulatory and Human Health Benchmarks Used for SSL Development t Numb*r Cn*mtealName " 63-32-§ Acenaphlhene 67-64-1 Acetone (2-Propanone) 309-00-2 Aldrln 120-12-7 Anthracene 7440-36-0 Antimony 7440-38-2 Arsenic 7440-39-3 Batlu- 56-55-3 Benz(a )anthracene 71-43-2 Benzene . 205-99-2 Benzo(b )fluoranlhene . 207-06-9 Benzo(k)lluoranth*ne 65-65-0 Benzole add 50-32-8 Benzo(« )pyr*na 7440-41-7 Beiyium 111.44-4 Bls(2-chteroelhyl)ether . 117-81-7 Bl8(2-etnylhexyl)ph»ial«te 75-27-4 Bromodtehtemmetisne 75-25-2 Bromcfnrm (Mbromomethane) 71-38-3 But9f& 85-68-7 Butyl benzyl phtwlate 7440-43-9 Cadmium 86-74-8 Cerbazoto 75-15-0 Carbon ditulfid* 56-23-5 Carbon WracHorlde 57-74-9 CHordarw 106-47-8 p -CHoreanMna 106-90-7 CNorobanzane 124-48-1 CNorodlbromometiarw 67-66-3 Chloroform 95-57-8 2-CMorophenol Maximum Goal MCLQ (PMCLG) Rat. * 6.0E-03 3 2.0E+OP 3 4.0E-03 3 5.0E-03 3 1.0E-01 3 8.0E-02 3 Maximum Contaminant Lav*) (mg«-) MCL(PMCL) R- . 6.0E-03 3 5.0E-02 3 2.0E+00 3 5.0E-03 3 2.0E-04 3 4.0E-03 3 60E-03 3 1.0E-01 ' 3 1.0E-01 * 3 S.OE-03 3 5.0E-03 3 2.0E-03 3 1.0E-01 3 1.0E-01 ' 3 1.0E-01 * 3 LbnHa (mgn.) HBLk BM>* ~H+i55 —— RUT" 4E*00 RID 5E-06 SF. 1E+01 RID IE-04 SF. IE-04 SF. IE-03 SF. 1E+02 RID BE-05 SF. 4E+00 RID 7E+00 RID 4E-03 SF. 4E*00 RID 1E-01 RIO • 2E-01 RID Cancer Slop* Factor <msA»Hl)' Care. 0 B2 1.7E+01 1 D A 1.5E+00 1 B2 7.3E-01 4 A 2.9E-02 1 82 7.3E-01 4 82 7.3E-02 4 B2 7.3E+00 1 B2 4.3E+00 1 B2 1.1E+00 1 B2 1.4E-02 1 82 6.2E-02 1 82 7.9E-03 1 D C 82 2.0E-02 2 82 1.3E-01 1 82 1.3E+00 1 D C 8.4E-02 1 B2 6.1E-03 1 UnNRtek Factor (W^')1 care. URF n_, . Cb*a* D B2 4.9E-03 1 D A 4.3E-03 1 82 A B.3E-06 1 82 B2 82 B2 2.4E-03 1 82 3.3E-04 1 82 82 82 1.1E-06 1 D C 81 1.8E-03 1 82 1.5E-05 1 82 3.7E-04 1 D C 82 2.3E-05 1 Relcranc* Do** W0 Hal.' 6.0E-02 1 1.0E-01 1 3.0E-05 3.0E-01 4.0E-04 3.0E-04 7.0E-02 4.0E*00 1 S.OE-03 1 2.0E-02 1 2.0E-02 2.06-02 1.0E-01 2.0E-01 1.0E-03" 1.0E-01 7.0E-04 6.0E-05 4.0E-03 2.0E-02 2.0E-02 1.0E-02 5.0E-03 1 R*f«r*nc« («»*"') •« «-.- S.OE-04 2 7.0E-01 1 2.0E-02 2 • Proposed MCL = 0.08 mat. Drinking Water Regulations and Heal* Jdwsones . U.S. EPA (1995). •• Cadmium RID is based on dietary exposure. H- -fa •-•4 Table 1 (continued) ...9*?... ChemteelName nun MI 7440-47-3 Chromium 16065-83-1 Chromium (III) 18540-29-9 Chromium (VI) 218-01-9 Chryeane 57-12-5 Cyanide (amenable) 72-54-8 ODD 72-55-9 DDE 50-29-3 DOT 53-70-3 D*eni(«.h (anthracene 64-74-2 Dl-n -butyl phthalale 95-50-1 1.2-DlcNorobentene 106-46-7 1.4-DicNoroberttene 91-94-1 3.3-DtertorobeniWne 75-34-3 1.1-Dlenloroetwne 107-08-2 1,2-Dlchloroe*iane 75-35-4 1.1-Dtehloroeftytene iuiiKa.9 tt» 1 ~i flh le^MnelliHlene 19ff-39-c C«» *I»C *l™r»"wH^niy^B^ 15640-5 Iran* .1.2-Dtehteroethylene 120-83-2 2.4-DW*>?ophenol 78-87-5 1.2-DteHoropropane 542-75-8 1,3-DfcNoropflJpene 60-57-1 DMdrin 8446-2 DtolhylpMheWe 10547-9 2.4-Dknelhylpnanol 51-28-5 2.4-DWtrophenol 121-14-2 2.4-DHIrololuene" 608-20-2 2,8-DHltitoluene" 1 17-84-0 Dl-n -oclyl phthatale 115-29-7 Endosullan 72-20-8 Endrln . Maximum Contaminant Level Qoal (moA) MCLQ (PMCLO) Ret* 1.0E-01 3 (2.0E-01) 3 8.0E-01 3 7.5E-02 3 70E-03 3 7.0E-02 3 1.0E-01 3 2.0E-03 3 _ I _ ,,_, mmum Contaminant Level (moA) MCLfPMCL) R(| . 1.0E-01 3 1.0E-01 3* (2.0E-01) 3 8.0E-01 3 7.5E-02 3 S.OE-03 3 7.0E-03 3 7.0E-02 3 1.0E-01 3 5.0E-03 3 2.0E-03 3 Water HeaMiBaeed LbnMe (mg/L] MBL* Ba*to 4E»01 RID IE-02 SF. 4E-04 SF. 3E-04 SF. 3E-04 SF. 1E-05 SF. 4E+00 RID 2E-04 SF. 4E+00 RID IE-01 RID 5E-04 . SF. SE-06 SF. 3E»01 RID 7E-01 RID 4E-02 RID IE-04 SF. IE-04 SF. 7E-01 RIO 2E-01 RID Cancer Slope Feeler (mgrk^d)1 Care, Che.* *' •"** A A 82 7.3E-03 4 D B2 2.4E-01 1 82 3.4E-01 1 82 3.4E-01 1 82 7.3E+00 4 D D 82 2.4E-02 2 82 45E-01 1 C 82 9.1E-02 1 C 80E-01 1 D 82 8.8E-02 2 82 1.8E-01 2 82 1.6E+01 1 D 82 6.8E-01 1 82 6.8E-01 1 0 UtiRRtoh Feeler to»oVr' Care. c|tw. "» R-.- A 1.2E-02 1 . A 1.2E-02 1 D 82 82 82 9.7E-OS 1 82 0 D 82 82 C 82 26E-05 1 C 5.0E-05 1 0 82 82 3.7E-05 2 82 4.6E-03 1 D D Reference Doea (mgrtio^l) 1110 Bel.* S.OE-03 1 1.0E+00 1 5.0E-03 1 2.0E-02 1 S.OE-04 1 1.0E-01 1 9.0E-02 1 1.0E-01 7 9.0E-03 1 1.0E-02 2 2.0E-02 1 3.0E-03 1 3.0E-04 S.OE-OJS 8.0E-01 2.0E-02 2.0E-03 2.0E4)3 1.0E-03 . 2 2.0E-02 2 6.0E-03 2 3.0E-04 1 Reference Cone entf alien (fno/nT) "^ Rel.* 2.0E-01 2 8.0E-01 1 5.0E-01 2 4.0E-03 1 2.0E-02 1 • MCL lor total chromium Is based on Cr (VI) toxteHy. 1 Cancer Steps Factor Is lor 2,4-, 2,6-Olnltrotoluerw mixture CO 8OO i ^ 2 >3 -* -^ n «k 9 0 K « fO O <B fO 0> <O 4k <O » . _ 9 9 £ ' P < ? ' T r 9 £ 9 r ¥ > u u ^ o -» -I OD •» 3 - - •* f - S <3 S 8 3 S S 828 2 3 t 2 t S <l.i»^.ui — i . W - U 4 > i > — i ) « - i j 6 i . _-* eg to to ,5 2 3 «£ ^ ?. a * 2 i * i-X 1 robenzene M 2- i • !. i . «!? 11.1. ? i • ' ^* i <•> i ? x |: i!; 3- i ^3 v I s ft ft 2 S m rfl ^ s s o 6 TO ft TO 6 6 6 - t o u 6 6 ro u 3 & TO TO TO 6 6 6 tt n§ i * M MHM "* M M ** m M TO TO ^" 2* ^ ^ ^ ^ 7^ r 2 2 .82 3 3 3 2 8 3 CO 9 3)3} -CO CO 3 X 33 3 3] ." a aa ."".""a-^aa a <O -i O> in (rt tn 3 2 3 8 3 31 3J a a o o o o o g g g o > o o g o o o o g o o g o g o g S g t 3 0 o . • !^ m fn ni * -H -» - f c ^ J W ^ l ^ f D oI s6 TO TO TO TO T^ rn ^i o a a o o g g g o > o o g o o a n g o a o c ) g o g g g oo ui —. ro 4> M — u is ro a A u m ffi in m m m o o o o o o 4k U UI 4> S W i tocntnunw < n M * t » d » ^ w u r o « * - ^ 4 u M Q D - * u t » j k TO 1^ ft ft IW ft TO (n ft ft ft TO ^^ ft ft ft ft ft TO TO TO TO TO TO ffl 1 ' toi U CA U 6 6 u * to -*. ro Om6 ft il Table 1 (continued) numwv (JMfl-4 . Telmchtoroelnylene 7440-28-0 1>iaMum 108-88-3 Toluene 8001-35-2 Tomphena 120-82-1 1,2,4-Trlchlorobertfene 71-SS-8 1,1,1-Trfcbtoroeftane 79-00-5 1,1,2-TricHoroetiana 79-01-6 Trichtoroatiylena 95-95-4 2.4.5-Trichterophenol 88-08-2 2,4,6-TricHorophenol < 7440-82-2 Vanadium 108-05-4 Vinyl acetate 75-01-4 Vinyl chtorlde(chtoroelhene) 108-38-3 m -Xylana ' 95-474 o -Xytene 106-42-3 p-Xylene 7440-66-6 Zinc Maximum Contaminant Level Goal (m?A) MCLG (PMCLO) Ral.* 5.0E-04 3 1.0E400 3 7.0E-02 3 2.0E-01 3 3.0E-03 3 nro 3 1.0E+01 3* 1.0E401 3 ' 1.0E*01 3* Maximum Contaminant Level (mo/l) MCL(PMCL) R(( . S.6E-63 ——— r~ 2.0E-03 3 10E*00 3 3.0E-03 3 7.0E-02 3 2.0E-01 3 5.0E-03 3 5.0E-03 3 2.0E-03 3 1.0E*01 3* 1.0E+01 3' 1.0E+01 3 ' Water Health Baaed LlmHe (mglL) HBL» B-«« 4E400 RIO 8E-03 SF. 3E-01 RID 4E»01 RID 1E+01 RIO Cancer Slop* Factor (rncAo-dr1 care. Claaa* *' B--' -iHaa —— ST5F5? —— 5— D B2 1.1E+00 1 D D C 5.7E-02 1 1.1E-02 5 B2 1.1E-02 1 A 1.9E+00 2 L 0 0 D UnKmak Factor (Mgnu1)' care. URF B— • Cba** S.8C-67 5 0 B2 3.2E-04 1 D D C 1.8E-05 1 1.7E-08 5 B2 . 3.1E-06 1 A B.4E-OS 2 D 0 0 D Reference Ooaa (mg/keH) RID R(f • i.6C^2 \ 2.0E-01 1 1.0E-02 1 4.0E-03 1 1.0E-01 1 7.0E-03 2 1.0E+00 1 2.0E+00 2 2.0E+00 2 2.0E+00 1 " 3.0E-01 1 net efeitoe * Concentfatton • (mcym') mc B««.' 4.0E-01 1 2.0E-01 2 t.OE+00 5 2.0E-01 1 ' MCL tor total xytorm (1330-20-71 to 10 mpA. '• RID lor total xyfana* to 2 mgflqrday. * References: 1. IRIS, U.8. EPA (1W5b) 2»HEAST,U.8.EPA(199Sd) 3«U.S.EPA(1995a) 4»OHEA,U.S.EPA(1993c) S = Interim toxfcHy criteria provided by Superlund Healt) fltok TecNncal Support Center. Enviranmtntal Criteria Assessment Olfica (ECAO). Ondnnatt. OH (1994) 6»ECAO.U.S.EPA(1994g) 7 « ECAO. US. EPA (19941) * Health Based Limits calculated lor 30-year exposure duration. 10 * risk or hazard quotient = 1. c Catagorizalon ol overall weight ol evidence tor human cardnogenlcHy: Group A: human carcinogen ' . QraupB: probable human carcinogen B1: flrrttedevMenMlromepldariMoglcituolat Of. eufflderw evidence from anNnal •tUOfoft Wu fnedBOJUBte* evidence Of Group C: pOMlbto human cardnognn Group 0: nolc(aa»Hteblesstoh«aHhcarcinoojen*ctty Group E: evidence of noncardnogenlcfty lor human* Additivity of the SSLs for noncarcinogenic chemicals is further complicated by the fact that not all SSLs are based on toxicity. Some SSLs are determined instead by a "ceiling limit" concentration (Cut) above which these chemicals may occur as ncnaqueous phase liquids (NAPLs) in soil (see Section 2.4.4). Therefore, the potential for additive effects must be carefully evaluated at even- site by considering the total Hazard Index (HI) for chemicals with RfDs or RfCs based on the same endpoint of toxicity (i.e., has the same critical effect as defined by the Reference Dose Methodology), .excluding chemicals with SSLs based on C»«. Table 2 lists several SSL chemicals with RfDs/RfCs. grouping those chemicals whose RfDs or RfCs are based on toxic effects in the same target organ or system. However, this list is limited, and a lexicologist should be consulted prior to addressing additive risks at a specific site. 2.1.2 Apportionment and Fractionation. EPA also has evaluated the SSLs for noncarcinogens in light of two related issues: apportionment and fractionation. Apportionment is typically used as the percentage of a regulatory health-based level that is allocated to the source/pathway being regulated fe.g., 20 percent of the RfD for the migration to ground water pathway). Apportioning risk assumes that the applied dose from the source, in this case contaminated soils, is only one portion of the total applied dose received by the receptor. In the Superfund program, EPA has traditionally focused on quantifying exposures to a receptor that are clearly sue-.slated and has not included exposures from other sources such as commercially available household products or workplace exposures. Depending on the assumptions concerning other source contributions, apportionment among pathways and sources at a site may result in more conservative regulatory levels (e.g., levels that are below an HQ of 1). Depending on site conditions, this may be appropriate on a site-specific basis. In contrast to apportionment, fractionation of risk may lead to less conservative regulatory levels because it assumes that some fraction of the contaminant does not reach the receptor due to partitioning into another medium. For example, if only one-fifth of the source is assumed to be available to the ground water pathway, and the remaining four-fifths is assumed to be released to air or remain in the soil, an SSL for the migration to ground water pathway could be set at five times the HQ of 1 due to the decrease in exposure (since only one-fifth of the possible contaminant is available to the pathway). However, the data collected to apply SSLs generally will not support the finite source models necessary for partitioning contaminants between pathways. 2.1.3 Acute Exposures. The exposure assumptions used to develop SSLs are representative of a chronic exposure scenario and do not account for situations where high-level exposures may lead to acute toxicity. For example, in some cases, children may ingest large amounts of soil (e.g., 3 to 5 grams) in a single event. This behavior, known as pica, may result in relatively high short-term exposures to contaminants in soils. Such exposures may be of concern for contaminants that primarily exhibit acute health effects. Review of clinical reports on contaminants addressed in this guidance suggests that acute effects of cyanide and phenol may be of concern in children exhibiting pica behavior. If soils containing cyanide and phenol are present at a site, the protectiveness of the chronic ingestion SSLs for these chemicals should be reconsidered. Although the Soil Screening Guidance instructs site managers to consider the potential for acute exposures on a site-specific basis, there are two major impediments to developing acute SSLs First, although data are available on chronic exposures (i.e., RfDs, RfCs, cancer slope factors), there is a paucity of data relating the potential for acute effects for most Superfund chemicals. Specifically, there is no scale to evaluate the severity of acute effects (e.g., eye irritation vs. dermatitis), no consensus on how TO incorporate the body's recovery mechanisms following acute exposures, and no toxicity benchmarks to apply for short-term exposures (e.g., a 7-day RfD for a critical endpoint). 14 TUT OO8 1151 Table 2. SSL Chemicals with Noncarcinogenic Effects on Specific Target Organ/System Target Organ/System________ Effect _____________________________ Kidney Acetone 1,1-Dichioroethane Cadmium Chtorobenzene Di-n-octyl phthalate Endosulfan Ethylbenzene Fiuoranthene Nitrobenzene Pyrene Toluene 2,4,5-Trichlorophenol Vinyl acetate Liver Acenaphthene Acetone Butyl benzyl phthalate Chlorobenzene Di-n-octyl phthalate Endrin Flouranthene Nitrobenzene Styrene Toluene 2,4,5-Trichlorophenol Central Nervous System Butanol Cyanide (amenable) 2,4 Dimethytphenol Endrin 2-Methylphenol Mercury Styrene Xytenes Adrenal Gland Nitrobenzene 1,2,4-Trichlorobenzene Increased weight; nephrotoxicrty Kidney damage Significant proteinuria Kidney effects Kidney effects Glomerulonephrosis Kidney toxicity Nephropathy Renal and adrenal lesions Kidney effects Changes in kidney weights Pathology Altered kidney weight Hepatotcxicity Increased weight Increased liver-to-body weight and liver-to-brain weight ratios Histopathology Increased weight; increased SGOT and SGPT activity Mild histological lesions in liver Increased liver weight Lesions Liver effects Changes in liver weights Pathology Hypoactivity and ataxia Weight loss, myelin degeneration Prostatration and ataxia Occasional convulsions Neurotoxicity Hand tremor, memory disturbances Neurotoxicity Hyperactivity Adrenal lesions increased adrenal weights; vacuolization in cortex ______ 15 oon 1152 Table 2: (continued) Target Organ/System Effect Circulatory System Antimony Barium frans-1,2-Dichloroethene c/s- .k-Dichloroethylene 2,4-Dimethyiphenol Fluoranthene Fluorene Nitrobenzene Styrene Zinc Reproductive System Barium Carbon disuH ide 2-Chlorophenol Methoxychlor Phenol Respiratory System 1,2-Dichloropropane Hexachiorocyclopentadiene Methyl bromide Vinyl acetate Gastrointestinal System Hexachiorocyclopentadiene Methyl bromide Immune System 2,4-Dichlorophenol p-Chloroaniline Altered blood chemistry and myocardial effects Increased blood pressure Increased alkaline phosphatase level Decreased hematocrit and hemoglobin Altered blood chemistry Hematotogic changes Decreased RBC and hemoglobin Hematologic changes Red blood cell effects Decrease in erythrocyte superoxide dismutase (ESOD) Fetotoxicity Fetal toxicity and malformations Reproductive effects Excessive loss of litters Reduced fetal body weight in rats Hyperplasia of the nasal mucosa Squamous metaplasia Lesions on the olfactory epithelium of the nasal cavity Nasal epithelial lesions Stomach lesions Epithelial hyperplasia of the forestomach Altered immune function Nonneoplastic lesions of splenic capsule__________ Source: U.S. EPA, 1995b, U.S. EPA, 1995d. Second, the inclusion of acute SSLs would require the development of acute exposure scenarios that would be acceptable and applicable nationally. Simply put, the methodology and data necessary to address acute exposures in a standard manner analogous to that for chronic exposures have not been developed. 2.1.4 Route-to-Route Extrapolation. For a number of the contaminants commonly found at Superfund sites, inhalation benchmarks .for toxicity are not available from IRIS or HEAST (see Table 1). Given that many of these chemicals exhibit systemic toxicity, EPA recognizes that the lack of such benchmarks could result in an underestimation of risk from contaminants in soil through the inhalation pathway. As pointed out by commenters to the December 1994 draft Soil Screening Guidance, ingestion SSLs tend to be higher than inhalation SSLs for most volatile chemicals with both inhalation and ingestion benchmarks. This suggests that ingestion SSLs may not be adequately protective for inhalation exposure to chemicals without inhalation benchmarks. 16 TUT OO8 i i s: However, with the exception of vinyl chloride (which is gaseous at ambient temperatures), migration to ground water SSLs are significantly lower man inhalation SSLs for volatile organic chemicals (see ,—. the generic SSLs presented in Appendix A). Thus, at sites where ground water is of concern, migration to ground water SSLs generally will be protective from the standpoint of inhalation risk. However, if the ground water pathway is not of concern at a she, the use of SSLs for soil ingestion may not be adequately protective for the inhalation pathway. 'To address this concern, OERR evaluated potential approaches for deriving inhalation benchmarks using route-to-route extrapolation from oral benchmarks (e.g., RfC^ from RfDowi) EPA evaluated a number of issues concerning route-to-route extrapolation, including: the potential reactivity of airborne toxicants (e.g., portal-of-entry effects), the phannacoltinetic behavior of toxicants for different routes of exposure (e.g., absorption by the girt versus absorption by the lung), and the significance of physicochemical properties in determining dose (e.g., vapor pressure, solubility). During this process, OERR consulted with staff in the EPA Office of Research and Development (ORD) to identify the most appropriate techniques for route-to-route extrapolation. Appendix B describes this analysis and hs results. As part of this analysis, inhalation benchmarks were derived using simple route-to-route extrapolation for SO contaminants lacking inhalation benchmarks. A review of SSLs calculated from these extrapolated benchmarks indicated that for 36 of the SO contaminants, inhalation SSLs exceed the soil saturation concentration (Cttf), often by several orders of magnitude. Because maximum volatile emissions occur at Ciat (see Section 2.4.4), these 36 contaminants are not likely to pose significant risks through the inhalation pathway at any soil concentration and the lack of inhalation benchmarks is not likely to underestimate risks. All of the 14 remaining contaminants with extrapolated inhalation SSLs below C^ have inhalation SSLs above generic SSLs for the migration to ground water pathway (dilution attenuation factor [DAF] of 20). This suggests mat migration to ground water SSLs will be adequately protective of volatile inhalation risks at sites where ground .s**^, water is of concern. At sites where ground water is not of concern (e.g., .where ground water beneath or adjacent to the she is not a potential source of drinking water), the Appendix B analysis suggests that for certain contaminants, ingestion SSLs may not be protective of inhalation risks for contaminants lacking inhalation benchmarks. The analysis indicates that the extrapolated inhalation SSL values are below SSL values based on direct ingestion for the following chemicals: acetone, bromodichloromethane, chiorodibromomethane, c;jr-l,2-dichloroethylene, and /ra«5-l,2-dichloroethylene. This supports the possibility that the SSLs based on direct ingestion for the listed chemicals may not be adequately protective of inhalation exposures. However, because this analysis is based on simplified route-to- route extrapolation methods, a more rigorous evaluation of route-to-route extrapolation methods may be warranted, especially at sites where ground water is not of concern. Based on these results, EPA reached the following conclusions regarding the route-to-route extrapolation of inhalation benchmarks for the development of inhalation SSLs. First, it is reasonable to assume that, for some volatile contaminants, the lack of inhalation benchmarks may underestimate risks due to inhalation of volatile contaminants at a she. However, the analysis in Appendix B suggests that mis issue is only of concern for sites where the exposure potential for the inhalation pathway approaches that for ingestion of ground water or at sites where the migration to ground water pathway is not of concern. Second, the extrapolated inhalation SSL values are not intended to be used as generic SSLs for site investigations; the extrapolated inhalation SSLs are useful in determining the potential for inhalation risks but should not be misused as SSLs. The extrapolated inhalation benchmarks, used to calculate extrapolated inhalation SSLs, simply provide an estimate of the air concentration 17 1.154 required to produce an inhaled dose equivalent to the dose received via oral administration, and lack the scientific rigor required by EPA for route-to-route extrapolation. Route-to-route extrapolation methods must account for a relationship between physicochemical properties, absorption and distribution of toxicants, the significance of portal-of-entry effects, and the potential differences in metabolic pathways associated with the intensity and duration of inhalation exposures. However, methods required to develop sufficiently rigorous inhalation benchmarks have only recently been developed by the ORD. EPA's ORD has made available a guidance document mat addresses many of the issues critical to the development of inhalation benchmarks. The document, entitled Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994d), presents methods for applying inhalation dosimetry to derive inhalation reference concentrations and represents the current state-of-the-science at EPA with respect to inhalation benchmark development. The fundamentals of inhalation dosimetry are presented with respect to the toxicokinetic behavior of contaminants and the physicochemical properties of chemical contaminants. Thus, at sites where the migration to ground water pathway is not of concern and a site manager determines that the inhalation pathway may be significant for contaminants lacking inhalation benchmarks, route-to-route extrapolation may be performed using EPA-approved methods on a case-by-case basis. Chemical-specific route-to-route extrapolations should be accompanied by a complete discussion of the data, underlying assumptions, and imcertainties identified in the extrapolation process. Extrapolation methods should be consistent with the EPA guidance presented in Methods for Derivation of Inhalation Reference Concentrations and Applications of Inhalation Dosimetry (U.S. EPA, 1994d). If a route-to-route extrapolation is found not to be appropriate based on the ORD guidance, the information on extrapolated SSLs may be included as part of the uncertainty analysis of the baseline risk assessment for the she. 2.2 Direct Ingestion Calculation of SSLs for direct ingestion of soil is based on the methodology presented for residential land use in RAGS HHEM, Part 8 (U.S. EPA, 199Ib). Briefly, this methodology backcalculates a soil concentration level from a target risk (for carcinogens) or hazard quotient (for noncarcinogens). A number of studies have shown that inadvertent ingestion of soil is common among children 6 years old and younger (Calabrese et al., 1989; Davis et al., 1990; Van Wijnen et al., 1990). Therefore, the approach uses an age-adjusted soil ingestion factor that takes into account the difference in daily soil ingestion rates, body weights, and exposure duration for children from 1 to 6 years old and others from 7 to 31 years old. The higher intake rate of soil by children and their lower body weights lead to a lower, or more conservative, risk-based concentration compared to an adult-only assumption. RAGS HHEM, Pan 6 uses this age-adjusted approach for both noncarcinogens and carcinogens. For noncarcinogens, the definition of an RfD has led to debates concerning the comparison of less- than-lifetime estimates of exposure to the RfD. Specifically, it is often asked whether the comparison of a 6-year exposure, estimated for children via soil ingestion, to the chronic RfD is unnecessarily conservative. In their analysis of the issue, the SAB indicates that, for most chemicals, the approach of combining the higher 6-year exposure for children with chronic toxicity criteria is overly protective (U.S. EPA, 1993e). However, they noted that there are instances when the chronic RfD may be based on endpoints of toxicity that are specific to children (e.g., fluoride and nitrates) or when the dose- response curve is steep (i.e., the dosage difference between the no-observed-adverse-effects level [NOAEL] and an adverse effects level is small). Thus, for the purposes of screening, OERR opted to base the generic SSLs for noncarcinogenic contaminants on the more conservative "childhood only" TUT 008 exposure (Equation 1). The issue of whether to maintain this more conservative approach throughout the baseline risk assessment and establishing remediation goals will depend on how the toxicology of the chemical relates to the issues raised by the SAB. Screening Level Equation for Ingestion of Noncarcinogenic Contaminants in Residential Soil (Source: RAGS HHEM, Part B; U.S. EPA, 1991b) Screening Level (mg /kg) THQ x BW x AT x 365 d/yr 1 /RfD0 x 10"* kg/mg x EF x ED x IR 0) Parameter/Definition (units) THQ/target hazard quotient (unrtless) BW/body weight (kg) AT/averaging time (yr) RfD0/oral reference dose (mg/kg-d) EF/exposure frequency (d/yr) ED/exposure duration (yr) IR/soil ingestion rate (mg/d)_____ Default 1 15 6« chemical-specific 350 6 200 For noncarcinogens, averaging time is equal to exposure duration. Unlike RAGS HHEM, Part B, SSLs are calculated only for 6-year childhood exposure. For carcinogens, both the magnitude and duration of exposure are important. Duration is critical because the toxicity criteria are based on "lifetime average daily dose." Therefore, the total dose received, whether it be over 5 years or SO years, is averaged over a lifetime of 70 years. To be protective of exposures to carcinogens in the residential setting, RAGS HHEM, Part B (U.S. EPA, 1991b) and EPA focus on exposures to individuals who may live in the same residence for a "high- end" period of time (e.g., 30 years). As mentioned above, exposure to soil is higher during childhood and decreases with age. Thus, Equation 2 uses the RAGS HHEM, Part B time-weighted average soil ingestion rate for children and adults; the derivation of this factor is shown in Equation 3. Screening Level Equation for Ingestion of Carcinogenic Contaminants in Residential Soil (Source: RAGS HHEM, Part B; U.S. EPA, 1991b) Screening Level (mg /kg) = TR x AT x 365 d/yr (2) SF0 x 10"6 kg/mg x EF x IFtoil/adj 19 TUT 008 H56 Parameter/Definition (units) TR/target cancer risk (unftless) AT/averaging time (yr) SF0 /oral slope factor (mg/kg-d)-i EF/exposure frequency (d/yr) '^soii/adj /age-adjusted soil ingestion factor (mg-yr/kg-d) Default 10-6 70 chemical-specific 350 114 Equation for Age-Adjusted Soil Ingestion Factor, IFMii/adi TF " « (mg-yr/Hg-d) "Soil/«tel -6 X "^nel-6 -31 /«te7-31 BW^el.6 BW (3) mge7-31 Parameter/Definition (units) IFsoii/adj /age-adjusted soil ingestion factor (mg-yr/kg-d) IR«oii/agei.« /ingestion rate of soil age 1-6 (mg/d) EDagsi-6 /exposure duration during ages 1-6 (yr) IR$oii/age7-3i /ingestion rate of soil age 7-31 (mg/d) EDag^-st /exposure duration during ages 7-31 (yr) BWagei-6 /average body weight from ages 1-6 (kg) i /average body weight from ages 7-31 (kg) Default 114 200 6 100 24 15 70 Source: RAGS HHEM, Part B (U.S. EPA, 1991b). Because of the impracticability of developing she-specific input parameters (e.g., soil ingestion rates, chemical-specific bioavailability) for direct soil ingestion, SSLs are calculated using the defaults listed in Equations 1, 2, and 3. Appendix A lists these generic SSLs for direct ingestion of soil. 2.3 Dermal Absorption Incorporation of dermal exposures into the Soil Screening Guidance is limited by the amount of data available to quantify dermal absorption from soil for specific chemicals. EPA's ORD evaluated the available data on absorption of chemicals from soil in the document Dermal Exposure Assessment: Principles and Applications (U.S. EPA, 1992b). This document also presents calculations comparing the potential dose of a chemical in soil from oral routes with that from dermal routes of exposure. These calculations suggest that, assuming 100 percent absorption of a chemical via ingestion, absorption via the dermal route must be greater than 10 percent to equal or exceed the ingestion exposure. Of the 110 compounds evaluated, available data are adequate to show greater than 10 percent dermal absorption only for pentachlorophenol (Wester et al., 1993). Therefore, the ingestion SSL for pentachlorophenol is adjusted to account for this additional exposure (i.e.. the ingestion SSL has been divided in hai to account for increased exposure via the dermal route). Limited data suggest that dermal absorption of other semivolatile organic chemicals (e.g., benzo(a)pyrene) from soil may exceed 10 percent (Wester et al., 1990) but EPA believes that 20 TUT 008 1157 further investigation is needed. As adequate dennal absorption data are developed for such chemicals the ingestion SSLs may need to be adjusted. EPA will provide updates on this issue as appropriate. 2.4 Inhalation of Volatiles and Fugitive Dusts EPA toxicity data indicate that risks from exposure to some chemicals via inhalation far outweigh the risks via ingestion; therefore, the SSLs have been designed to address this pathway as well. Hie models and assumptions used to calculate SSLs for inhalation of volatiles axe updates of risk assessment methods presented in RAGS HHEM, Part B (U.S. EPA, 1991b). RAGS HHEM, Part B evaluated the contribution to risk from the inhalation and ingestion pathways simultaneously. Because toxicity criteria for oral, exposures are presented as administered doses (in mg/kg-d) and criteria for inhalation exposures are presented as concentrations in air (in ug/m3), conversion of air concentrations was required to estimate an administered dose comparable to the oral route. However, EPA's ORD now believes that, due to portal-of-entry effects and differences in absorption in the gut versus the lungs, the conversion from concentration in air to internal dose is not always appropriate and suggests evaluating these exposure routes separately. The models and assumptions used to calculate SSLs for the inhalation pathway are presented in Equations 4 through 12, along with the default parameter values used to calculate the generic SSLs presented in Appendix A. Particular attention is given to the volatilization factor (VF), saturation limit (CMt), and the dispersion portion of the VF and paniculate emission factor (PEF) equations, all of which have been revised since originally presented in RAGS HHEM, Part B. The available chemical-specific human health benchmarks used in these equations are presented in Section 2.1. Part 5 presents the chemical properties required by these equations, along with the rationale for their selection and development. 2.4.1 Screening Level Equations for Direct Inhalation. Equations 4 and 5 are used to calculate SSLs for the inhalation of carcinogenic and noncarcinogenic contaminants, respectively. Each equation addresses volatile compounds and fugitive dusts separately for developing screening levels based on inhalation risk for subsurface soils and surface soils. Separate VF-based and PEF-based equations were developed because the SSL sampling strategy addresses surface and subsurface soils separately. Inhalation risk from fugitive dusts results from particle entrainment from the soil surface; thus contaminant concentrations in the surface soil horizon (e.g., the top 2 centimeters) are of primary concern for this pathway. The entire column of contaminated soil can contribute to volatile emissions at a site. However, the top 2 centimeters are likely to be depleted of volatile contaminants at most sites. Thus, contaminant concentrations in subsurface soils, which are measured using core samples, are of primary concern for quantifying the risk from volatile emissions. 21 TUT COS 1158 Screening Level Equation for Inhalation of Carcinogenic Contaminants in Residential Soil Volatile Screening Level (mg/kg) TR x AT x 365d/yr URF x 1,000 u.g/mg x EF x ED x [ 1 iw (4) TR x AT x 36Sd/yr Paniculate Screening Level ("«/kg) URF x 1,000 Mg/Qg x EF x ED x f 1 Parameter/Definition (units) TR/target cancer risk (unrttess) AT/averaging time (yr) URF/inhalation unit risk factor (ug/m3)-i EF/exposure frequency (d/yr) ED/exposure duration (yr) VF/soil-to-air volatilization factor (ntf/kg) PEF/particulate emission factor (m3/kg) Default 10-6 70 chemical-specific 350 30 chemical-specific 1.32x109 Source: RAGS HHEM. Part B (U.S. ERA. 1991b). Screening Level Equation.for Inhalation of Noncarcinogenic Contaminants in Residential Soil Volatile Screening Level (OS/kg) THQ x AT x 365 d/yr F X ^ X V RfC X (5) Paniculate Screening Level (mg/kg) THQ x AT x 365 d/yr RfC 22 TUT 008 1159 Parameter/Definition (units) THQAarget hazard quotient (unitless) AT/averaging time (yr) EF/exposure frequency (oVyr) ED/exposure duration (yr) RfC/inhalation reference concentration (mg/m3) VF/soil-to-air volatilization factor (nWkg) PEF/particulate emission factor (irfi/kg) (Equation 10) Default 1 30 350 30 chemical-specific chemical-specific 1.32 x Source: RAGS HHEM, Part B (U.S. ERA, 1991b). To calculate inhalation SSLs, the volatilization factor and paniculate emission factor must be calculated. The derivations of VF and PEF have been updated since RAGS HHEM, Part B was published and are discussed fully in Sections 2.4.2 and 2.4.5, respectively. The VF and PEF equations can be broken into two separate models: models to estimate the emissions of volatiles and dusts, and a dispersion model (reduced to the term Q/C) that simulates the dispersion of contaminants in the atmosphere. 2.4.2 Volatilization Factor. The soil-to-air VF is used to define the relationship between the concentration of the contaminant in soil and the flux of the volatilized contaminant to air. VF is calculated from Equation 6 using chemical-specific properties (see Pan 5) and either site-measured or default values for soil moisture, dry bulk density, and fraction of organic carbon in soil. The User's Guide (U.S. EPA, 1996) describes how to develop she measured values for these parameters. Derivation of Volatilization Factor .1/2 (6) (3.14 x DA x T)1 . . . VF(m3/kg) = Q/C x —————-——— x 10-4(m2/cm2) ( 2 x p b x D A ) where :°'3 DJ/n2] e 23 TUT OOS 1160 Parameter/Definition (units) Default Source VF/volatilization factor (m3/kg) DA /apparent diffusivity (cm2/s) Q/C/inverse of the mean cone, at center of square source (g/m2-s per kg/rr>3) t/exposure interval (s) Pb/dry soil bulk density (g/crn3) 0a/air-filled soil porosity (L^/Uoa) n/total soil 6w/water-filted soil os /soil particle density (g/cm3) i /diffusivity in air (cm2/s) H'/dimensionless Henry's law constant Dw /diffusivity in water (cm2/s) Kd /soil-water partition coefficient (cm3/g) = KOC foe KQC /soil organic carbon-water partition coefficient (cnvVg) foc/organic carbon content of soil (g/g) _________ 68.81 9.5 x 10» 1.6 0.28 0.43^ 0.15 2.65 chemical-specific chemical-specific chemical-specific chemical-specific chemical-specific 0.006 (0.6%) Table 3 (for 0.5-acre source in Los Angeles, CA) U.S. EPA (1991 b) U.S. EPA (1991b) EQ, 1994 U.S. EPA (1991 b) see Part 5 see Part 5 see Part 5 see Part 5 see Part 5 Carseletal. (1988) The W. equation presented in Equation 6 is based on the volatilization model developed by Jury et al. (1984) for infinite sources and is theoretically consistent with the Jury et al. (1990) finite source volatilization model (see Section 3.1). This equation represents a change in the fundamental volatilization model used to derive the VF equation used in RAGS HHEM, Part B and in the December 1994 draft Soil Screening Guidance (U.S. .EPA, 1994h). The VF equation presented in RAGS HHEM, Part 8 is based on tie volatilization model developed by Hwang and Falco (1986) for dry soils. During the reevaluation of RAGS HHEM, Part B, EPA sponsored a study (see the December 1994 draft Technical Background Document, U.S. EPA, 1994i) to validate the VF equation by comparing the modeled results with data from (1) a bench-scale pesticide study (Fanner and Letey, 1974) and (2) a pilot-scale study measuring the rate of loss of benzene, toluene, xylenes, and ethylbenzene from soils using an isolation flux chamber (Radian. 1989). The results of the study verified the need to modify the W equation in Part B to take into account the decrease in the rate of flux due to the effect of soil moisture content on effective diffusivity (Dei). In the December 1994 version of this background document (U.S. EPA, 1994i), the Hwang and Falco model was modified to account for the influence of soil moisture on the effective diffusivity using the Millington and Quirk (1961) equation. However, inconsistencies were discovered in the modified Hwang and Falco equations. Additionally, even a correctly modified Hwang and Falco model does not consider the influence of the liquid phase on the local equilibrium partitioning. Consequently, EPA evaluated the Jury model for its ability to predict emissions measured in pilot-scale volatilization studies (Appendix C; EQ, 1995). The infinite source Jury model emission rate predictions were consistently within a factor of 2 of the emission rates measured in the pilot-scale volatilization studies. Because the Jury model predicts well the available measured soil contaminant volatilization rates, eliminates the inconsistencies of the modified Hwang and Falco model, and considers the 24 TUT oos -U61 influence of the liquid phase on the local equilibrium partitioning, it was selected to replace the <--•- modified Hwang and Falco model for the derivation of the VF equation. Defaults. Other than initial soil concentration, air-filled soil porosity is the most significant soil parameter affecting the final steady-state flux of volatile contaminants from soil (U.S. EPA, 1980). In other words, the higher the air-filled soil porosity, the greater the emission flux of volatile constituents. Air-filled soil porosity is calculated as: 6. - n - 6W (7) where 6, - air-filled soil n - total soil porosity (Lpore/LMu) 6W =• water-filled soil porosity and n=l-(0b/p.) (8) where . Pb - dry soil bulk density (g/cm3) p, - soil particle density (g/cm3). Of these parameters, water-filled soil porosity (6W) has the most significant effect on air-filled soil porosity and hence volatile contaminant emissions. Sensitivity analyses have shown that soil bulk density (Pb) has too limited a range for surface soils (generally between 1.3 and 1.7 g/cm3) to affect results with nearly the significance of soil moisture conditions. Therefore, a default bulk density of 1.50 g/cm3, the mode of the range given for U.S. soils in the Superfund Exposure Assessment Manual (U.S. EPA, 1988), was chosen to calculate generic SSLs. This value is also consistent with the mean porosity (0.43) for loam soil presented in Carsel and Parrish (1988). The default value of 6W (0.15) corresponds to an average annual soil water content of 10 weight percent. This value was chosen as a conservative compromise between that required to achieve a monomolecular layer of water on soil particles (approximately 2 to 5 weight percent) and that required to reduce the air-filled porosity to zero (approximately 29 weight percent). In this manner, nonpolar or weakly polar contaminants are desorbed readily from the soil organic carbon as water competes for sorption sites. At the same time, a soil moisture content of 10 percent yields a relatively conservative air-filled porosity (0.28 or 28 percent by volume). A water-filled soil porosity (6W) of 0.15 lies about halfway between the mean wilting point (0.09) and mean field capacity (0.20) reported for Class B soils by Carsel et al. (1988). Class B soils are soils with moderate hydrologic characteristics whose average characteristics are well represented by a loam soil type. The default value of p, (2.65 g/cm3) was taken from U.S. EPA (1988) as the particle density for most soil mineral material. The default value for foc (0.006 or 0.6 percent) is the mean value for the top 0.3 m of Class B soils from Carsel et al. (1988). • .25 TUT 008 1162 2.4.3 Dispersion Model. The box model in RAGS HHEM Part B has been replaced with a Q/C term derived from a modeling exercise using meteorologic data from 29 locations across the United States. The dispersion model used in the Part B guidance is based on the assumption that emissions into a hypothetical box will be distributed uniformly throughout the box. To arrive at the volume within .the box, it is necessary to assign values to the length, width, and height of the box. The length (LS) was the length of a side of a contaminated she with a default value of 45 m; the width was based on the windspeed in the mixing zone (V) with a default value of 2.25 m (based on a windspeed of 2.25 m/s); and the height was the diffusion height (DH) with a default value of 2 m. However, the assumptions and mathematical treatment of dispersion used in the box model may not be applicable to a broad range of site types and meteorology and do not utilize state-of-the-art techniques developed for regulatory dispersion modeling. EPA was very concerned about the defensibility of the box model and^sought a more defensible dispersion model that could be used as a replacement to the Part B guidance and had the following characteristics: • Dispersion modeling from a ground-level area source • Onsite receptor • A long-term/annual average exposure point concentration • Algorithms for calculating the exposure point concentration for area sources of different sizes and shapes. To identify such a model, EPA held discussions with the EPA Office of Air Quality Planning and Standards (OAQPS) concerning recent efforts to develop a new algorithm for estimating ambient air concentrations from low or ground-level, nonbuoyant sources of emissions. The new algorithm is incorporated into the Industrial Source Complex Model (ISC2) platform in both a short-term mode (AREA-ST) and a long-term mode (AREA-LT). Both models employ a double numerical integration over the source in the upwind and crosswind directions. Wind tunnel tests have shown that the new alp. nthm performs well with onsite and near-field receptors. In addition, subdivision of the source is not required . r these receptors. Because the new algorithm provides better concentration estimates for onsite and for near-field receptors, a revised dispersion analysis was performed for both volatile and paniculate matter contaminants (Appendix D; EQ, 1994). The AREA-ST model was run for 0.5-acre and 30-acre square sources with a full year of meteorologic data for 29 U.S locations selected to be representative of the national range of meteorologic conditions (EQ, 1993). Additional modeling runs were conducted to address a range of square area sources from 0.5 to 30 acres in size (Table 3). The Q/C values in Table 3 for 0.5- and 30-acre sources differ slightly from the values in Appendix D due to differences in rounding conventions used in the final model runs. To calculate site-specific SSLs, select a Q/C value from Table 3 mat best represents a site's size and meteorologic condition. To d-- Mop a reasonably conservative default Q/C for calculating generic SSLs, a default site (Los Ang: CA) was chosen that best approximated the 90th percentile of the 29 normalized concur.orations (kg/m3 per g/m^-s). The inverse of this concentration results in a default VF Q/C value of 68.81 g/m^-s per kg/m* for a 0.5-acre site. 26 TUT 008 116" Table 3. Q/C Values by Source Area, City, and Climatic Zone Q/C (g/m2-s par kg/m3) Zone 1 Seattle SaJern Zone II Fresno Los Angeles SanFlanctsoo Zone III LasVegas Phoenix Albuquerque Zone IV Boise Wmnemucca Salt Lake City Casper Denver Zone V Bismark Minneapolis Lincoln Zone VI Little Rock Houston Atlanta Charleston Raleigh-Durham Zone VII Chicago Cleveland Huntington Harrisburg Zone VIII Portland Hartford Phiadelphia Zone IX Marri 0.5 Acre 82.72 73.44 62.00 68.81. 89.51 95.55 64.04 84.18 69.41 69.23 78.09 100.13 75.59 83.39 90.80 81.64 73.63 79.25 77.08 74.89 77.26 97.78 83.22 53.89 81.90 74.23 71.35 90.24 85.61 1 Acre 72.62 64.42 54.37 60.24 78.51 83.87 56.07 73.82 60.88 60.67 68.47 87.87 66.27 73.07 79.68 71.47 64.51 69.47 67.56 65.65 67.75 85.81 73.06 47.24 71.87 65.01 62.55 79.14 74.97 2 Acre 64.38 57.09 48.16 53.30 69.55 74.38 49.59 65.40 53.94 53.72 60.66 77.91 58.68 64.71 70.64 63.22 57.10 61.53 59.83 58.13 60.01 76.08 64.78 41.83 63.72 57.52 55.40 70.14 66.33 5 Acre 55.66 49.33 41.57 45.93 60.03 64.32 42.72 56.47 46.57 46.35 52.37 67.34 50.64 55.82 61.03 54.47 49.23 53.11 51.62 50.17 51.78 65.75 55.99 36.10 55.07 49.57 47.83 60.59 . 57.17 10 Acre 50.09 44.37 37.36 41.24 53.95 57.90 38.35 50.77 41.87 41.65 47.08 60.59 45.52 50.16 54.90 48.89 44.19 47.74 46.37 45.08 46.51 59.16 50.38 32.43 49.56 44.49 43.00 54.50 51.33 30 Acre 42.86 37.94 31.90 35.15 46.03 49.56 32.68 43.37 35.75 35.55 40.20 51.80 38.87 42.79 46.92 41.65 37.64 40.76 39.54 38.48 39.64 50.60 43.08 27.67 42.40 37.88 36.73 46.59 43.74 27 TUT COS 1164 2.4.4 Soil Saturation Limit. The soil saturation concentration (CMt) corresponds to the contaminant concentration in soil at which the absorptive limits of the soil particles, the solubility limits of the soil pore water, and saturation of soil pore air have been reached. Above this concentration, the soil contaminant may be present in free phase, i.e.. nonaqueous phase liquids (NAPLs) for contaminants that are liquid at ambient soil temperatures and pure solid phases for compounds that are solid at ambient soil temperatures. Derivation of the Soil Saturation Limit ew + H'e.) (9) Parameter/Definition (units) Cut/soil saturation concentration (mg/kg) S/solubility in water (mg/L-water) Pt/dry soil bul- density (kg/L) Kd/soil-water partition coefficient (L/kg) Koc/soil organic carbon/water partition coefficient (L/kg) foc/fraction organic carbon of soil (g/g) Gy/water-fiDed soil porosity (LwateA-soii) H'/dimensionless Henry's law constant H/Henn/s law constant (atm-ma/mol) Ba/air-filled soil porosity (Laii/Uoil) nAotal soil porosity (Lp<Wl-»oii) ps/soil partide density (kg/L) Default - chemical-specific 1.5 Kocxfoc(organics) chemical-specific 0.006 (0.6%) 0.15 H x 41 , where 41 is a conversion factor chemical-specific 0.28 043 2.65 Source see Part 5 U.S. EPA, 1991b see Part 5 Carseletal., 1988 EQ, 1994 U.S. EPA, 1991b see Part 5 n-e* 1 - Pb/Ps U.S. EPA, 1991b Equation 9 is used to calculate Csat for each site contaminant. As an update to RAGS HHEM, Part 8, this equation takes into account the amount of contaminant that is in the vapor phase in the pore spaces of the soil in addition to the amount dissolved in the soil's pore water and sorbed to soil particles. Chemical-specific CU1 concentrations must be compared with each volatile inhalation SSL because a basic principle of the SSL volatilization model (Henry's law) is not applicable when free-phase contaminants are present (i.e., the model cannot predict an accurate VF or SSL above CM). Thus, the VF-based inhalation SSLs are applicable only if the soil concentration is at or below CMt. When calculating volatile inhalation SSLs, CMt values also should be calculated using the same site-specific soil characteristics used to calculate SSLs (i.e., bulk density, average water content, and organic carbon content). At Csd the emission flux from soil to air for a chemical reaches a plateau. Volatile emissions will not increase above this level no matter how much more chemical 2S added to the soil. Table 3-A shows that for compounds with generic volatile inhalation SSLs greater than CMt, the risks at CMt are significantly below the screening risk of 1 x 10-* and an HQ of 1. Since C^ corresponds to maximum 28 TUT O08 1165 volatile emissions, the inhalation route is not likely to be of concern for those chemicals with SSLs exceeding Cut concentrations. Table 3-A. Risk Levels Calculated at CMt for Contaminants that have Values Greater than C.at URF Chemical name (ng/m3)-i DOT 9.7E-05 1 ,2-Dichlorobenzene — 1 ,4-Dichlorobenzene — Ethylbenzene — P-HCH (P-BHC) 5.3E-04 Styrene — Toluene — 1,2,4-Trichlorobenzene — 1 ,1 ,1 -Trichloroethane — RfC (mg/m3) — 2.0E-01 8.0E-01 1.0E+00 — 1 .OE+00 4.0E-01 2.0E-01 1 .OE+00 VF .(mVkg) 3.0E+07 1.5E+04 1.3E+04 5.4E+03 1.3E+06 1.3E+04 4.0E+03 4.3E+04 2.2E-03 CMt Carcinogenic (mg/kg) Risk 4.0E+02 5.2E-07 6.0E+02 - 2.8E+02 — 4.0E+02 — 2.0E+00 3.4E-07 1 .5E+03 — 6.5E+02 — 3.2E+03 — 1 .2E+03 - Non- Carcinogenic Risk — 0.2 0.03 0.07 — 0.1 0.4 0.4 0.5 Table 4 provides the physical state (i.e. liquid or solid) for various compounds at ambient soil temperature. When the inhalation SSL exceeds Cut for liquid compounds, the SSL is set at CMt. This is because, for compounds that are liquid at ambient soil temperature, concentrations above Csat indicate a potential for free liquid phase contamination to be present, and the possible presence of NAPLs. ERA. believes that further investigation is warranted when free nonaqueous phase liquids may be present in soils at a she. Table 4. Physical State of Organic SSL Chemicals Compounds liquid at soil temperatures CAS No. Chemical 67-64-1 Acetone 71-43-2 Benzene 1 17-81-7 Bis(2-ethylhexyl)phthalate 111.44-4 Bis(2-chloroethyl)etner 75-27-4 Bromodichloromethane 75-25-2 Bromofomn 71-36-3 Butanol 85-68-7 Butyl benzyl phthalate 75-15-0 Carbon disulfide 56-23-5 Carbon tetrachloride 108-90-7 Chlorobenzene • 124-48-1 Chlorodibromomethane 67-66-3 Chloroform Melting Point CO -94.8 5.5 -55 -51.9 -57 8 -89.8 -35 -115 -23 -45.2 -20 -63.6 Compounds eolid at aoil temperatures CAS No. Chemical 83-32-9 Acenaphthene 309-00-2 Aldrin 120-12-7 Anthracene 56-55-3 Benz(a)anthracene 50-32-8 Benzo(a)pyrene 205-99-2 Benzo(b)fluoranthene 207-08-9 Benzo(*)fluoranthene 65-85-0 Benzole acid 86-74-8 Carbazole 57-74-9 Chterdane 106-47-8 p-CWoroaniline 218-01-9 Chrysene 72-54-8 ODD Melting Point CC) 93.4 104 215 84 176.5 168 217 122.4 246.2 106 72.5 258.2 109.5 29 TUT Table 4. (continued) Compounds liquid at toil temperatures Melting Point CC) CAS No. Chemical Compound* solid at eeil temperatures Melting CAS No. Chemical Point CC) 95-57-8 2-Chloropheno! 9.8 72-55-9 64-74-2 Di-n-butyl phthalate -35 50-29-3 95-50-1 1,2-Dichlorobenzene -16.7 53-70-3 75-34-3 1,1-Dichloroethane -96.9 106-46-7 107-06-2 1,2-Dtehloroethane -35.5 91-94-1 75-35-4 1,1-Dtehloroethylene -122.5 120-83-2 156-59-2 c/s-1,2-Dichloroethylene -80 60-57-1 15f 50-5 trans-1,2-Diehloroethylene - -49.8 105-67-9 7fe-J -5 1,2-Dtehloropropane -70 51-28-5 542-75-6 1,3-Dichloroprop*ne NA 121-14-2 84-66-2 Diethylphthalate -40.5 606-20-2 117-64-0 Di-n-octyl phthalate -30 72-20-8 100-41-4 Ethylbenzene -94.9 206-44-0 87-68-3 Hexachioro-1,3-butadiene -21 86-73-7 77.47.4 Hexachlorocyclopentadiene -9 76-44-8 78-59-1 Isophorone -8.1 1024-57-3 74-83-9 Methyl bromide -93.7 118-74-1 75-09-2 Methylene chloride -95.1 319-84-6 98-95-3 Nitrobenzene 5.7 319-85-7 100-42-5 .Styrene -31 58-89-9 79-34-5 1,1,2.2-Tetrachloroethane -43.8 67-72-1 127-18-4 Tetrachloroethylene -22.3 193-39-5 108-88-3 Toluene -94.9 72-43-5 120-82-1 1,2,4-Trichlorobenzene 17 95-48-7 71-55-6 1,1,1-Trichloroethane -30.4 621-64-7 79-00-5 1,1,2-Trichloroethane -36.6 86-30-6 79-01-6 Trichloroethylene -64.7 91-20-3 108-05-4 Vinyl acetate -93.2 87-86-5 75-01-4 Vinyl chloride -153.7 106-95-2 108-38-3 n>Xylene -47.8 129-00-0 95-47-6 o-Xylene -25.2 8001-35-2 106-42-3 p-Xylene 13.2 95-95-4 88-06-2 115-29-7 DOE 89 DOT 108.5 Dtoenzo(a,h)anthracene 269.5 1,4-Dichlorobenzene 52.7 3,3-Dichlorobenzidine 132.5 2,4-Dichlorophenol 45 Dietdrin 175.5 2,4-Dimethylphenol 24.5 2,4-Dinitrophenol 115-116 2,4-Dinitrotoluene 71 2,6-Dinrtrotoluene 66 Endrin 200 Ruoranthene 107.8 Ruorene 114.8 Heptachlor 95.5 Heptachlor epoxide 160 Hexachlorobenzene 231.8 o-HCH(o-BHC) 160 B-HCH(B-BHC) 315 yHCH (Lindane) 112.5 Hexachloroethane 187 lndeno(1,2.3-cd)pyrene 161.5 Methoxychlor 87 2-Methylpheno! 29.8 M-Nitrosodi-n-propylamine NA W-Nitrosodiphenylamine 66.5 Naphthalene 80.2 Pentachlorophenol 174 Phenol 40.9 Pyrene 151.2 Toxaphene 65-90 2.4.5-Trichlorophenol ' 69 2.4.6-Thchiorophenol 69 EndosulKan 106 NA.htotavaiiabto. 30 TUT 008 1167 When free phase liquid contaminants are suspected. Estimating the Potential for Occurrence of DNAPL at Superfund Sites (U.S. EPA, 1992c) provides information on determining the likelihood of dense nonaqueous phase liquid (DNAPL) occurrence in the subsurface. Free-phase contaminants may also be present at concentrations lower than CMJ if multiple component mixtures are present. The DNAPL guidance (U.S. EPA, 1992c) also addresses the likelihood of free-phase contaminants when multiple contaminants are present at a site. For compounds that are solid at ambient soil temperatures (e.g.. DOT), Table 3-A indicates mat the inhalation risks are well below the screening targets (i.e., these chemicals do not appear to be of concern for the inhalation pathway). Thus, when inhalation SSLs are above C^ for solid compounds, soil screening decisions should be based on the appropriate SSLs for other pathways of concern at the she (e.g., migration to ground water, ingestion). 2.4.5 Particulate Emission Factor. The paniculate emission factor relates the concentra- tion of contaminant in soil with the concentration of dust particles in the air. This guidance addresses dust generated from open sources, which is termed, "fugitive" because it is not discharged into the atmosphere in a confined flow stream. Other sources of fugitive dusts that may lead to higher emissions due to mechanical disturbances include unpaved roads, tilled agricultural soils, and heavy construction operations. Both the emissions portion and the dispersion portion of the PEF equation have been updated since RAGS HHEM, Part B. As in Part B, the emissions pan of the PEF equation is based on the "unlimited reservoir" model from Cowherd et al. (1985) developed to estimate paniculate emissions due to wind erosion. The unlimited reservoir model is most sensitive to the threshold friction velocity, which is a function of the mode of the size distribution of surface soil aggregates. This parameter has the greatest effect on the emissions and resulting concentration. For this reason, a conservative mode soil aggregate size of 500 urn was selected as the default value for calculating generic SSLs. The mode soil aggregate size determines how much wind is needed before dust is generated at a site. A mode soil aggregate size of SOO um yields an uncorrected threshold friction velocity of 0.5 ro/s. This means that the windspeed must be at least 0.5 m/s before any fugitive dusts are generated However, the threshold friction velocity should be corrected to account for the presence of nonerodible elements. In Cowherd et al. (1985), nonerodible elements are described as . . . clumps of grass or stones (larger than about 1 cm in diameter) on the surface (that will) consume pan of the shear stress of the wind which otherwise would be transferred to erodible soil. Cowherd et al. describe a study by Marshall (1971) that used wind tunnel studies to quantify the increase in the threshold friction velocity for different kinds of nonerodible elements. His results are presented in Cowherd et al. as a graph showing the rate of corrected to uncorrected threshold friction velocity vs. Lc, where Lc is a measure of nonerodible elements vs. bare, loose soil. Thus, the ratio of corrected to uncorrected threshold friction velocity is directly related to the amount of nonerodible elements in surface soils. Using a ratio of corrected to uncorrected threshold friction velocity of 1, or no correction, is roughly equivalent to modeling "coal dust on a concrete pad," whereas using a correction factor of 2 corresponds to a windspeed of 19 m/s at a height of 10 m. This means that about a 43-mph wind would be required to produce any paniculate emissions. Given mat the 29 meteorologic data sets used in this modeling effort showed few windspeeds at, or greater than, 19 m/s, EPA felt that it was necessary to choose a default correction ratio between 1 and 2. A value of 1.25 was selected as a 31 008 1168 reasonable number that would be at the more conservative end of the range. This equates to a corrected threshold friction velocity of 0.625 m/s and an equivalent windspeed of 11.3 m/s at a height of 7 meters. As with the VF model, Q/C values are needed to calculate the PEF (Equation 10); use the QC value in Table 3 that best represents a site's size and meteorologic conditions (i.e., the same value used to calculate the VF; see Section 2.4.2). Cowherd et al. (1985) describe how to obtain site-specific ' estimates of V, Um, Ut, and F(x). Unlike volatile contaminants, meteorologic conditions (i.e., the intensity and frequency of wind) affect both the dispersion and emissions of paniculate matter. For this reason, a separate default Q/C value was derived for paniculate matter [nominally 10 urn and less (PM10)] emissions for the generic SSLs. The PEF equation was used to calculate annual average concentrations for each of 29 sites across the country. To develop a reasonably conservative default Q/C for calculating generic SSLs, a default site (Minneapolis, MN) was selected that best approximated the 90th percentile concent r~-ion. The resuiis produced a revised default PEF Q/C value of 90.80 g/ntf-s per kg/m3 for a 0.5-acre site (see Appendix D; EQ, 1994). The generic PEF derived using the default values in Equation 10 is 1.32 x 109 mVkg, which corresponds to a receptor point concentration of approximately 0.76 U£/m3. This represents an annual average emission rate based on wind erosion that should be compared with chronic health criteria; it is not appropriate for evaluating the potential for more acute exposures. Derivation of the Participate Emission Factor PEF(m3/kg) = Q/C x 3,600s/n 0.036 x (1-V) x (Uffi/U,r x F(x) (10) Parameter/Definition (units) PEF/particulate emission factor (nWkg) Q/C/inverse of mean cone, at center of square source (g/m2-s per kg/m3) V/fraction of vegetative cover (unttless) Um/mean annual windspeed (nVs) U/equivalent threshold value of windspeed at 7 m (nVs) F(x)/function dependent on Um/Ut derived using Cowherd et al. (1985) (unrttess) Default 1.32x109 90.80 0.5(50%) 4.69 11.32 0.194 Source Table 3 (for 0.5-acre sourc • in Minneapolis. MN) U.S. EPA. 1991b EQ. 1994 U.S. EPA. 1991b U.S. EPA. 1991b 2.5 Migration to Ground Water The methodology for calculating SSLs for the migration to ground water pathway was developed to identify chemical concentrations in soil that have the potential to contaminate ground water. 32 . TUT 008 Migration of contaminants from soil to ground water can be envisioned as a two-stage process: (1) release of contaminant in soil leachate and (2) transport of the contaminant through the underlying soil and aquifer to a receptor well. The SSL methodology considers both of these fate and transport mechanisms. The methodology incorporates a standard linear equilibrium soil/water partition equation to estimate contaminant release in soil leachate (see Sections 2.5.1 through 2.5.4) and a simple water-balance equation that calculates a dilution factor to account for dilution of soil leachate in an aquifer (see Section 2.5.5). The dilution factor represents the reduction in soil leachate contaminant concentrations by mixing in the aquifer, expressed as the ratio of leachate concentration to the concentration in ground water at the receptor point (i.e., drinking water well). Because the infinite source assumption can result in mass-balance violations for soluble contaminants and small sources, mass-limit models are provided that limit the amount of contaminant migrating from soil to ground water to the total amount of contaminant present in the source (see Section 2.6) SSLs are backcalculated from acceptable ground water concentrations (i.e., nonzero MCLGs, MCLs, or HBLs; see Section 2.1). First, the acceptable ground water concentration is multiplied by a dilution factor to obtain a target leachate concentration. For example, if the dilution factor is 10 and the acceptable ground water concentration is 0.05 mg/L, the target soil leachate concentration would be 0.5 mg/L. The partition equation is then used to calculate the total soil concentration (i.e., SSL) corresponding to this soil leachate concentration. The methodology for calculating SSLs for the migration to ground water pathway was developed under the following constraints: • Because of the large nationwide variability in ground water vulnerability, the methodology should be flexible, allowing adjustments for site-specific conditions if adequate information is available. . To be appropriate for early-stage application, the methodology needs to be simple, requiring a minimum of she-specific data. The methodology should be consistent with current understanding of subsurface processes. • The process of developing and applying SSLs should generate information that can be used and built upon as a site evaluation progresses. Flexibility is* achieved by using readily obtainable site-specific data in standardized equations, conservative default input parameters are also provided for use when site-specific data are not available. In addition, more complex unsaturated zone fate-and-transport models have been identified that can be used to calculate SSLs when more detailed site-specific information is available or can be obtained (see Part 3). These models can extend the applicability of SSLs to subsurface conditions that are not adequately addressed by the simple equations (e.g., deep water tables; clay layers or other unsaturated zone characteristics that can attenuate contaminants before they reach ground water). The SSL methodology was designed for use during the early stages of a site evaluation when information about subsurface conditions may be limited. Because of this constraint, the methodology is based on conservative, simplifying assumptions about the release and transport of contaminants in the subsurface (see Highlight 2). 33 TUT COS 117O Highlight 2: Simplifying Assumptions for the migration to Ground Water Pathway • The source is infinite (i.e., steady-state concentrations will be maintained in ground water over the exposure period of interest). • Contaminants are uniformly distributed throughout the zone of contamination. • Soil contamination extends from the surface to the water table (i.e., adsorption sites are filled in the unsaturated zone beneath the area of contamination). • There is no chemical or biological degradation in the unsaturated zone. • Equilibrium sowwater partitioning is instantaneous and linear in the contaminated soil. • The receptor well is at the edge of the source (i.e., there is no dilution from recharge downgradient of the site) and is screened within the plume. • The aquifer is unconsoiidated and unconftned (surficial). • Aquifer properties are homogeneous and isotropic. • There is no attenuation (i.e., adsorption or degradation) of contaminants in the aquifer. • NAPLs are not present at the site. Although simplified, the SSL methodology described in this section is theoretically and operationally consistent with the more sophisticated investigation and modeling efforts that are conducted to develop soil cleanup goals and cleanup levels for protection of ground water at Superfund sites. SSLs developed using this methodology can be viewed as evolving risk-based levels that can be refined as more site information becomes available. The early use of the methodology at a site will help focus further subsurface investigations on areas of true concern with respect to ground water quality and will provide information on soil characteristics, aquifer characteristics, and chemical properties that can be built upon as a she evaluation progresses. 2.5.1 Development of Soil/Water Partition Equation. The methodology used to estimate contaminant release in soil leachate is based on the Freundlich equation, which was developed to model sorption from liquids to solids. The basic Freundlich equation applied to the soil/water system is. K =C /Cn (11) d s w where K<j = Freundlich soil/water partition coefficient (L/kg) Cs = concentration sorbed on soil (mg/kg) Cw - solution concentration (mg/L) n - Freundlich exponent (dimensionless). 34 TUT 008 Assuming that adsorption is linear with respect to concentration (n-1)* and rearranging to backcalculate a sorbed concentration (Cs): (12) For SSL calculation. Q* is the target soil leachate concentration. Adjusting Sorbed Soil Concentrations to Total Concentrations. To develop a screening level for comparison with contaminated soil samples, the sorbed concentration derived above (Cs) must be related to the total concentration measured in a soil sample (Ct). In a soil sample, contaminants can be associated with the solid soil materials, the soil water, and the soil air as follows (Feenstra et al., 1991): where Furthermore, and (13) Mt = total contaminant mass in sample (mg) MS = contaminant mass sorbed on soil materials (mg) Mw = contaminant mass in soil water (mg) M, = contaminant mass in soil air (mg). (14) (15) (16) (17) where pb = dry soil bulk density (kg/L) V^ = sample volume (L) 6W = water-filled porosity C» = concentration on soil pore air 6. = air-filled soil porosity (Lair/L»oii) For contaminated soils (with concentrations below Cut), C, may be determined from Cw and the dimensionless Henry's law constant (H') using the following relationship: (18) The linear assumption will tend to overestimate sorption and underestimate desorption for most organics at higher concentrations (i.e., above 1C-5 M for organics) (Piwoni and Banerjee, 1989). 35 TUT 008 1172 thus Substituting into Equation 13: (19) (20) or C = C - C *"§ *-t *~w (21) Substituting into Equation 12 and rearranging: Soil-Water Partition Equation for Migration to Ground Water Pathway: Inorganic Contaminants C, = Cw Kd e,H' (22) Parameter/Definition (units) Default Source CVscreening level in soil (mg/kg) Cw/target soil leachate concentration (mg/L) K<ysoit-water partition coefficient (L/kg) Bw/water-filled soil porosity (Lw*t»iA*oii) Oa/air-filled soil porosity (Uj/Uoil) nAotal soil porosity (Lpon/Lsod) pb/dry soil bulk density (kg/L) Ps/soi! particle density (kg/L) H'/dimensionless Henry's law constant H/Henn/s law constant (stm-rr^/mol) (nonzero MCLG, MCL, orHBL)x20DAF chemical-specific 0.3 (30%) 0.13 0.43 1.5 2.65 H x 41, where 41 is a conversion factor chemical-specific Table 1 (nonzero MCLG, MCL); Section 2.5.6 (DAF for 0.5-acre source) see Part 5 U.S. EPA/ORD n-6* 1-Pb/Ps U.S. EPA, I991b U.S. EPA, 1991b U.S. EPA, 1991b see Part 5 36 TUT UUt: 117: Equation 22 is used to calculate SSLs (total soil concentrations, Q) corresponding to soil leachate concentrations (Cw) equal to the target contaminant soil leachate concentration. The equation assumes that soil water, solids, and gas are conserved during sampling. If soil gas is lost during sampling, 6, should be assumed to be zero. Likewise, for inorganic contaminants except mercury, there is no significant vapor pressure and H' may be assumed to be zero. -The User's Guide (U.S. EPA, 1996) describes how to develop site-specific estimates of the soil parameters needed to calculate SSLs. Default soil parameter values for the partition equation are the same as those used for the VF equation (see Section 2.4.2) except for average water-filled soil porosity (6W). A conservative value (0.15) was used in the VF equation because the model is most sensitive to this parameter. Because migration to ground water SSLs are not particularly sensitive to soil water content (see Section 2.5.7), a value that is more typical of subsurface conditions (0.30) was used. This value is between the mean field capacity (0.20) of Class B soils (Carsel et al., 1988) and the saturated volumetric water content for loam (0.43). Kd varies by chemical and soil type. Because of different influences on Kd values, derivations of Kj values for organic compounds and metals were treated separately in the SSL methodology. 2.5.2 Organic Compounds—Partition Theory. Past research has demonstrated that, for hydrophobic organic chemicals, soil organic matter is the dominant sorbing component in soil and that KJ is linear with respect to soil organic carbon content (OC) as long as OC is above a critical level (Dragun, 1988). Thus, Kd can be normalized with respect to soil organic carbon to KOC. a chemical-specific partitioning coefficient that is independent of soil type, as follows: (23) where KOC = organic carbon partition coefficient (L/kg) foc = fraction of organic carbon in soil (mg/mg) Substituting into Equation 22: Soil-Water Partition Equation for Migration to Ground Water Pathway: Organic Contaminants e* 37 TUT 008 1174 Parameter/Definition (units) Default Source CVscreening level in soil mg/kg) Cy/target leachate concentration (mg/L) il organic carbon-water partition coefficient (L/kg) Worganic carbon content of soil (kg/kg) 6^/water-iilied soil porosity (Lwatt/Uoil) 6a/atr-filled soil porosity (Laj/L»oil) nAotal soil porosity (Lpom/Uoa) ivary soil bulk density (kg/L) pg/soil particle density (kg/L) -. H'/dimensionless Henry's law constant H/Henry's law constant (atm-ms/mol) (nonzero MCLG, MCL, orHBL)x20DAF chemical-specific 0.002 (0.2%) 0.3 (30%) 0.13 0.43 1.5 2.65 Hx41, where 41 is a conversion factor chemical-specific Table 1 (MCL, nonzero MCLG); Section 2.5.6 (DAF for a 0.5-acre source) see Part 5 Carseletal., 1988 U.S. EPA/ORD n-6* 1-pt/P. U.S. EPA, 1991b U.S. EPA, 1991 b U.S. EPA, 1991b see Part 5 Pan 5 of this document provides KQC values for organic chemicals and describes their development. The critical organic carbon content, foc* , represents OC below which sorption to mineral surfaces begins to be significant. This level is likely to be variable and to depend on both the properties of the soil and or the chemical sorbate (Curtis et al., 1986). Attempts to quantitatively relate foc* to such properties have been made (see McCarty et al., 1981), but at this time there is no reliable method for estimating foc* for specific chemicals and soils. Nevertheless, research has demonstrated that, for volatile nalogenated hydrocarbons, foc* is about 0.001, or 0.1 percent OC, for many low-carbon soils and aquifer materials (Piwoni and Banerjee, 1989; Schwarzenbach and Westall, 1981). If soil OC is below mis critical level, Equation 24 should be used with caution. This is especially true if soils contain significant quantities of fine-grained minerals with high sorptive properties (e.g., clays). If sorption to minerals is significant, Equation 24 will underpredict sorption and overpredict contaminant concentrations in soil pore water. However, this foe* level is by no means the case for all soils; Abdul et al. (1987) found that, for certain organic compounds and aquifer materials, sorption was linear and could be adequately modeled down to f^ = 0.0003 by considering K^ alone. For soils with significant inorganic and organic sorption (i.e., soils with foc < 0.001), the following equation has been developed (McCarty et al., 1981; Karickhoff, 1984): (25) where = soil inorganic partition coefficient - fraction of inorganic material = 1. 38 OOS 1175 Although this equation is considered conceptually valid, K jc values are not available for the subject _,«.. chemicals. Attempts to estimate K;0 values by relating sorption on low-carbon materials to properties such as clay-size fraction, clay mineralogy, surface area, or iron-oxide content have not revealed any consistent correlations, and semiquantitative methods are probably years away (Piwoni and Banerjee. 1989). However, Piwoni and Banerjee developed the following empirical correlation (by linear regression,, r? = 0.85) that can be used to estimate Kj values for hydrophobic .organic •chemicals from K<,wfor low-carbon soils: log Kd = 1.01 log Kow - 0.36 (26) where = octanol/water partition coefficient. The authors indicate that this equation should provide a Kj estimate that is within a factor of 2 or 3 of the actual value for nonpolar sorbates with log KOW < 3.7. This Kj estimate can be used in Equation 22 for soils with foc values less than 0.001. If sorption to inorganics is not considered for low-carbon soils where it is significant, Equation 24 will underpredict sorption and overpredict contaminant concentrations in soil pore water (i.e., it will provide a conservative estimate). The use of fixed Koc values in Equation 24 is valid only for hydrophobic, nonionizing organic chemicals. Several of the organic chemicals of concern ionize in the soil environment, existing in both neutral and ionized forms within the normal soil pH range. The relative amounts of the ionized and neutral species are a function of pH. Because the sorptive properties of these two forms differ, it is important to consider the relative amounts of the neutral and ionized species when determining Koc values at a particular pH. Lee et al. (1990) developed a theoretically based algorithm, developed from thermodynamic equilibrium equations, and demonstrated that the equation adequately predicts laboratory-measured KOC values for pentachlorophenol (PCP) and other ionizing organic acids as a function of pH. The equation assumes that sorbent organic carbon determines the extent of sorption for both the ionized and neutral species and predicts the overall sorption of a weak organic acid (K^p ) as follows 4> n) (27) where KOC.II, KOC,; - sorption coefficients for the neutral and ionized species (L/kg) On = (1 + lOpH-pK.)-! pKa = acid dissociation constant. This equation was used to develop Koc values for ionizing organic acids as a function of pH, as described in Part 5. The User's Guide (U.S. EPA, 19%) provides guidance on conducting she-specific measurements of soil pH for estimating KOC values for ionizing organic compounds. Because a national distribution of soil pH values is not available, a median U.S. ground water pH (6.8) from the STORET database (U.S. EPA, 1992a) is used as a default soil pH value that is representative of subsurface pH conditions. 39 TUT 008 ii/ 6 2.5.3 Inorganics (Metals)—Partition Theory. Equation 22 is used to estimate SSL? for metals for the migration to ground water pathway. The derivation of Kd values is much more complicated for metals than for organic compounds. Unlike organic compounds, for which K«i va; es are largely controlled by a single parameter (soil organic carbon), Kd values for metals ore significantly affected by a variety of soil conditions. The most significant parameters are pH, oxidation-reduction conditions, iron oxide content soil organic matter content, cation exchange 'capacity, and major ion chemistry. The number of significant influencing parameters, their variability in the field, and differences in experimental methods result in a wide range of K«i values for individual metals reported in the literature (over 5 orders of magnitude). Thus, it is much more difficult to derive generic K<j values for metals than for organics. The Kd values used to generate SSLs for Ag, Ba, Be, Cd, Cr-*, Cu, Hg, Ni, and Zn were developed using an equilibrium geochemical speciation model (MINTEQ2). The values for As, Cr6+, Se, and Th were taken from empirical, pH-dependent adsorption relationships developed by EPA/ORD. Metal Kd values for SSL application are presented in Part 5, along with a description of their development and limitations. As with the ionizing organics, K<j values are selected as a function of she-specific soil pH, and metal Kd values corresponding to a pH of 6.8 are used as defaults where site-specific pH measurements are not available. 2.5.4 Assumptions for Soil/Water Partition Theory. The following assumptions are implicit in the SSL partitioning methodology. These assumptions and their implications for SSL accuracy should be read and understood before using this methodology to calculate SSLs. 1 ' There is no contaminant loss due to volatilization or degradation. The source is considered to be infinite; i.e., these processes do not reduce soil leachate concentrations over time. This is a conservative assumption, especially for smaller sites. 2. Adsorption is linear with concentration'. The methodology assumes that adsorption is independent of concentration (i.e., the Freundlich exponent = 1). This has been reported to be true for various halogenated hydrocarbons, polynuclear aromatic hydrocarbons, benzene, and chlorinated benzenes. In addition, this assumption is valid at low concentrations (e.g., at levels close to the MCL) for most chemicals. As concentrations increase, however, the adsorption isotherm can depart from the linear. Studies on trichloroethane (TCE) and chlorobenzene indicate that departure from linear is in the nonconservative direction, with adsorbed concentrations being lower than predicted by a linear isotherm. However, adequate information is not available to establish nonlinear adsorption isotherms for the chemicals of interest. Furthermore, since the SSLs are derived at relatively low target soil leachate concentrations, departures from the linear at high concentrations do not significantly influence the accuracy of the results. . 3. The system is at equilibrium with respect to adsorption. This ignores adsorption/desorption kinetics by assuming that the soil and pore water concentrations are at equilibrium levels. In other words, the pore-water residence time is assumed to be longer than the time it takes for the system to reach equilibrium conditions. This assumption is conservative. If equilibrium conditions are not met, the concentration in the pore water will be less than that predicted by the methodology. The kinetics of adsorption are not adequately understood for a sufficient number of chemicals and site conditions to consider equilibrium kinetics in the methodology. 40 1177 4. Adsorption is reversible. The methodology assumes that desorption processes operate in the same way as adsorption processes, since most of the K<,c values are measured by adsorption experiments rather than by desorption experiments. In actuality, desorption is slower to some degree than adsorption and, in some cases, organics can be irreversibly bound to the soil matrix. In general, the significance of this effect increases with KOW. This assumption is conservative. Slower desorption rates and irreversible sorption will result in lower pore-water concentrations man that predicted by the methodology. Again, the level of knowledge on desorption processes is not sufficient to consider desorption kinetics and degree of reversibility for all of the subject chemicals. 2.5.5 Dilution/Attenuation Factor Development. As contaminants in soil leachate move through soil and ground water, they are subjected to physical, chemical, and biological processes that tend to reduce the eventual contaminant concentration at the receptor point (i.e., drinking water well). These processes include adsorption onto soil and aquifer media, chemical transformation (e.g., hydrolysis, precipitation), biological degradation, and dilution due to mixing of the leachate with ambient ground water. The reduction in concentration can be expressed succinctly by a DAF, which is defined as the ratio of contaminant concentration in soil leachate to the concentration in ground water at the receptor point. When calculating SSLs, a DAF is used to backcalculate the target soil leachate concentration from an acceptable ground water concentration (e.g., MCLG). For example, if the acceptable grout") water concentration is 0.05 mg/L and the DAF is 10, the target leachate concentration would be 0.5 mg/L. The SSL methodology addresses only one of these dilution-attenuation processes: contaminant dilution in ground water. A simple equation derived from a geohydrologic water-balance relationship has been developed for the methodology, as described in the following subsection. The ratio factor calculated by this equation is referred to as a dilution factor rather than a DAF because it does not consider processes that attenuate contaminants in the subsurface (i.e., adsorption and degradation processes). This simplifying assumption was necessary for several reasons. First, the infinite source assumption results in all subsurface adsorption sites being eventually filled and no longer available to attenuate contaminants. Second, soil contamination extends to the water table, eliminating attenuation processes in the vmsaturated zone. Additionally, the receptor well is assumed to be at the edge of the source, minimizing the opportunity for attenuation in the aquifer. Finally, chemical-specific biological and chemical degradation rates are not known for many of the SSL chemicals; where they are available they are usually based on laboratory studies under simplified, controlled conditions. Because natural subsurface conditions such as pH, redox conditions, soil mineralogy, and available nutrients have been shown to markedly affect natural chemical and biological degradation rates, and because the national variability in these properties is significant and has not been characterized, EPA does not believe that it is possible at this time to incorporate these degradation processes into the simple site-specific methodology for national application. If adsorption or degradation processes are expected to significantly attenuate contaminant concentrations at a site (e.g., for sites with deep water tables or soil conditions that will attenuate contaminants), the site manager is encouraged to consider the option of using more sophisticated fate and transport models. Many of these models can consider adsorption and degradation processes and can model transient conditions necessary to consider a finite source size. Part 3 of this document presents information on the selection and use of such models for SSL application. 41 TUT oo The dilution factor model assumes that the aquifer is unconfined and unconsolidated and has homogeneous and isotropic properties. Unconfined (surficial) aquifers are common across the country, are vulnerable to contamination, and can be used as drinking water sources by local residents. Dilution model results may not be applicable to fractured rock or karst aquifer types. The site manager should consider use of more appropriate models to calculate a dilution factor (or DAF) for such settings. ' In addition, the simple dilution model does not consider facilitated transport This ignores processes such as colloidal transport, transport via solvents other than water (e.g., NAPLs), and transport via dissolved organic matter (DOM). These processes have greater impact as Kow (and hence, KOC) increases. However, the transport via solvents other than water is operative only if certain site- specific conditions are present. Transport by DOM and colloids has been shown to be potentially significant under certain conditions in laboratory and field studies. Although much research is in progress on these processes, the current state of knowledge is not adequate to allow for their consideration in SSL calculations. ^ If there is the potential for the presence of NAPLs in soils at the site or she area in question, SSLs should not be used for this area (i.e., further investigation is required). The Cut equation (Equation 9) presented in Section 2.4.4 can be used to estimate the contaminant concentration at which the presence of pure-phase NAPLs may be suspected for contaminants that are liquid at soil temperature. If NAPLs are suspected in site soils, refer to U.S. EPA (1992c) for additional guidance on how to estimate the potential for DNAPL occurrence in the subsurface. Dilution Model Development. EPA evaluated four simple water balance models to adjust SSLs for dilution in the aquifer. Although written in different terms, all four options reviewed can be expressed as the same simple water balance equation to calculate a dilution factor, as follows. Option 1 (ASTM): dilution factor - (1 + U^ d/IL) (28) where Ugw = Darcy ground water velocity (m/yr) d = mixing zone depth (m) I = infiltration rate (m/yr) L = length of source parallel to flow (m). For Darcy velocity. Ugw = Ki (29) where K = aquifer hydraulic conductivity (m/yr) i = hydraulic gradient (m/m). Thus dilution factor = 1 + (Kid/IL) (30) 42 TUT 008 1179 Option 2 (EPA Ground Water Forum): dilution factor = (Qp + QA)/Qp (31) where Qp = percolation flow rate (mVyr) QA - aquifer flow rate (m3/yr) For percolation flow rate: QP = IA (32) where A = facility area (m*) = WL. For aquifer flow rate: QA = WdKi (33) where W = width of source perpendicular to flow (m) d - mixing zone depth (m). Thus dilution fector - (1A + WdKi)/IWL = l+(Kid/IL) (34) Option 3 (Summers Model): C« - (Qp Cp)/(Qp + QA) (35) where Cw = ground water contaminant concentration (mg/L) Cp - soil leachate concentration (mg/L) given that Cw = Cp/dilution factor 43 TUT ....„ 1180 I/dilution factor = Qp/(Qp + QA) or dilution factor = (Qp + QA)/Qp (see Option 2) .Option 4 (EPA ORD/RSKERL): dilution factor = (Qp + QA)/QP = RX/RL (36) where • . R = recharge rate (m/yr) = infiltration rate (I, m/yr) X = distance from receptor well to ground water divide (m) (Nott that the intermediate equation is the same as Option 2.) This option is a longer-term option that is not considered further in this analysis because valid X values are not currently available either nationally or for specific sites. EPA is considering developing regional estimates for these parameters. Dilution Model Input Parameters. As shown, all three options for calculating contaminant dilution in ground water can be expressed as the same equation: Ground Water Dilution Factor dilution factor = 1 + (Kid/lL) (37) Parameter/Definition (units) K/aqu'rfer hydraulic conductivity (nvyr) ^hydraulic gradient (nVm) d/mixing zone depth (m) I/infiltration rate (nvyr) I/source length parallel to ground water flow (m) Mixing Zone Depth (d). Because of its dependence on the other variables, mixing zone depth is estimated with the method used for the MULTIMED model (Sharp-Hansen et al., 1990). The MULTIMED estimation method was selected to be consistent with that used by EPA's Office of Solid Waste for the EPA Composite Model for Landfills (EPACML). The equation for estimating mixing zone depth (d) is as follows: d = (20vL)o 5 * d, {1 - exp[(-LI)/(V1ned.)l} (38) 44 TUT 008 H81 where Ov = vertical dispersivity (m/m) Vs = horizontal seepage velocity (m/yr) ne = effective aquifer porosity (Lpore/Laquifer) d» = aquifer depth (m). The first term, (2OvL)° J, estimates the depth of mixing due to vertical dispersivity (dav) along the length of ground water travel. Defining the point of compliance with ground water standards at the downgradient edge of the source, this travel distance becomes the length of tile source parallel to flow L. Vertical dispersivity can be estimated by the following relationship (Gelhar and Axness, 1981): Ov = 0.056 ctL (39) where (XL = longitudinal dispersivity- 0.1 ^ xr - horizontal distance to receptor (m). Because the potential receptor is assumed to have a ™ell at the edge of the facility, x, = L and Ov = 0.0056 L (40) Thus (41) The second term, d, {1 - exp[(-LI) / (Vtned»)]}, estimates the depth of mixing due to the downward velocity of infiltrating water, djv In this equation, the following substitution may be made: (42) so dlv = d, {l-exp[(-LIV(Kida)]} (43) Thus, mixing zone depth is calculated as follows: (44) Estimation of Mixing Zon« D*pth d = (0.0112 U)0-3 + d, { 1 - exp[(-U)/(KidJ]} (45 45 TUT 008 -U82 Parameter/Definition (units) d/mixing zone depth (m) L/source length parallel to ground water flow (m) I/infiltration rate (nVyr) K/aquifer hydraulic conductivity (nVyr) da/aquifer thickness (m) Incorporation of this equation for mixing zone depth into the SSL dilution equation results in five parameters that must be estimated to calculate dilution: source length (L), infiltration rate (I), aquifer hydraulic conductivity (K), aquifer hydraulic gradient (i), and aquifer thickness (d,). Aquifer thickness also serves as a limit for mixing zone depth. The User's Guide (U.S. EPA, 1996) describes how to develop site-specific estimates for these parameters. Parameter definitions and defaults used to develop generic SSLs are as follows: • Source Length (L) is the length of the source (i.e., area of contaminated soil) parallel to ground water flow and affects the flux of contaminant released in soil leachate (IL) as well as the depth of mixing in the aquifer. Hie default option for this parameter assumes a square, 0.5-acre contaminant source. This default was changed from 30 acres in response to comments to be more representative of actual contaminated soil sources (see Section 1.3.4). Increasing source area (and thereby area) may result in a lower dilution factor. Appendix A includes an analysis of the conservatism associated with the 0.5-acre source size. • Infiltration Rate (I). Infiltration rate times the source area determines the amount of contaminant (in soil leachate) that enters the aquifer over time. Thus, increasing infiltration decreases the dilution factor. Two options can be used to generate infiltration rate estimates for SSL calculation. The first assumes that infiltration rate is equivalent to recharge. This is generally true for uncontrolled contaminated soil sites but would be conservative for capped sites (infiltration < recharge) and nonconservative for sites with an additional source of infiltration, such as surface impoundments (infiltration > recharge). Recharge estimates for this option can be obtained from Aller et al. (1987) by hydrogeologic setting, as described in Section 2.5.6. The second option is to use the HELP model to estimate infiltration, as was done for OSW's EPACML and EPA's Composite Model for Leachate Migration with Transformation Products (EPACMTP) modeling efforts. The Soil Screening Guidance (U.S. EPA, 1995c) provides information on obtaining and using the HELP model to estimate site-specific infiltration rates. • Aquifer Parameters. Aquifer parameters needed for the dilution factor model include hydraulic conductivity (K, m/yr), hydraulic gradient (i, m/m), and aquifer thickness (d,, m). The User's Guide (U.S. EPA, 1996) describes how to develop aquifer parameter estimates for calculating a site-specific dilution factor. 2.5.6 Default Dilution-Attenuation Factor. EPA has selected a default DAF of 20 to account for contaminant dilution and attenuation during transport through the saturated zone to a compliance point (i.e., receptor well); At most sites, this adjustment will more accurately reflect a contaminant's threat to ground water resources than assuming a DAF of 1 (i.e., no dilution or attenuation). EPA selected a DAF of 20 using a "weight of evidence" approach. This approach 46 TUT ^ 1183 considers results from OSW's EPACMTP model as well as results from applying the SSL dilution model described in Section 2.5.5 to 300 ground water sites across the country. The default DAF of 20 represents an adjustment from the DAF of 10 presented in the December 1994 draft Soil Screening Guidance (U.S. EPA, 1994h) to reflect a change in default source size from 30 acres to 0.05 acre. A DAF of 20 is protective for sources up to 0.5 acre in size. Analyses presented in Appendix A indicate that it can be protective of larger sources as well. However, mis hypothesis should be examined on a case-by-case basis before applying a DAF of 20 to sources larger than 0.5 acre. EPACMTP Modeling Effort. One model considered during selection of the default DAF is described in Background Document for EPA's Composite Model for Leachate Migration with Transformation Products (U.S. EPA, 1993a). EPACMTP has a three-dimensional module to simulate ground water flow that can account for mounding under waste sites. The model also has a three- dimensional transport module and both linear and nonlinear adsorption in the unsaturated and saturated zones and can simulate chain decay, thus allowing the simulation of the formation and the fate and transport of daughter (transformation) products of degrading chemicals. The model can also be used to simulate a finite source scenario. EPACMTP is comprised of three main interconnected modules: An unsaturated zone flow and contaminant fate and transport module A saturated zone ground water flow and contaminant fate and transport module • A Monte Carlo driver module, which generates model parameters from nationwide probability distributions. The unsaturated and saturated zone modules simulate the migration of contaminants from initial release from the soil to a downgradient receptor well. More information on the EPACMTP model is provided in Appendix E. EPA has extensively verified both the unsaturated and saturated zone modules of the EPACMTP against other available analytical and numerical models to ensure accuracy and efficiency. Both the unsaturated zone and the saturated zone modules of the EPACMTP have been reviewed by the EPA Science Advisory Board and found to be suitable for generic applications such as the derivation of nationwide DAFs. EPACMTP Model Inputs (SSL Application). For nationwide Monte Carlo model applications, the input to the model is in the form of probability distributions of each of the model input parameters. The output from the model consists of the probability distribution of DAF values, representing the likelihood that the DAF will not be less than a certain value. For instance, a 90th percentile DAF of 10 means that the DAF will be 10 or higher in at least 90 percent of the cases. For each model input parameter, a probability distribution is provided, describing the nationwide likelihood that the parameter has a certain value. The parameters are divided into four main groups: • Source-specific parameters, e.g., area of the waste unit, infiltration rate • Chemical-specific parameters, e.g., hydrolysis'constants, organic carbon partition coefficient 47 TUT OOS 1184 • Unsaturated zone-specific parameters, e.g., depth to water table, soil hydraulic conductivity • Saturated zone-specific parameters, e.g., saturated zone thickness, ambient ground water flow rate, location of nearest receptor well. Probability distributions for each parameter used in the model have been derived from nationwide 'surveys of waste sites, such as EPA's landfill survey (53 FR 28692). During the Monte Carlo simulation, values for each model parameter are randomly drawn from their respective probability distributions. In the calculation of the DAFs for generic SSLs, she data from over 1,300 municipal landfill sites in OSWs Subtitle D Landfill Survey were used to define parameter ranges and distributions. Each combination of randomly drawn parameter values represents one out of a practically infinite universe of possible waste sites. The fate and transport modules are executed for the specific set of model parameters, yielding a corresponding DAF Value. This procedure is repeated, typically on the order of several thousand times, to ensure that the entire universe of possible parameter combinations (waste sites) is adequately sampled. In the derivation of DAFs for generic SSLs, the model simulations were repeated 15,000 times for each scenario investigated. At the conclusion of the analysis, a cumulative frequency distribution of DAF values was constructed and plotted. • X(<tat>nc*1rom«ouro»tDW*H)*Oft • Y (transvMM vmll location) » Monte Cario within 1/2 wtdth of *3uro> • Z <w«ll InteJw potnt b»tow water tab*)« Monte Carto. rang* IS •* 300 it • RakttaH - Monte Carto • SoMtyp** Monte Carto • Dapth to aquifer • Monte Carto • Assumes Mnlte «oun» term Figure 3. Migration to ground water pathway—EPACMTP modeling •ffort. EPA assumed an infinite waste source of fixed area for the generic SSL modeling scenario. EPA chose this relatively conservative assumption because of limited information on the nationwide distribution of the volumes of contaminated soil sources. For the SSL modeling scenario, EPA performed a number of sensitivity analyses consisting of fixing one parameter at a time to determine the parameters that have the greatest impact on DAFs. The results of the sensitivity analyses indicate that the climate (net precipitation), soil types, and size of the contaminated area have the greatest effect on the DAFs. The EPA feels that the size of the contaminated area lends itself most readily to practical application to SSLs. • To calculate DAFs for the SSL scenario, the receptor point was taken to be a domestic drinking water well located on the downgradient edge of the contaminated area. The location of the intake point (receptor well screen) was assumed to vary between 15 and 300 feet below the water table (these 48 TUT •1185 values are based on empirical data reflecting a national sample distribution of depth of residential drinking water wells). The location of the intake point allows for mixing within the aquifer. EPA believes that this is a reasonable assumption because there will always be some dilution attributed to the pumping of water for residential use from an aquifer. The horizontal placement of the well was assumed to vary uniformly along the center of the downgradient edge of the source within a width of one-half of the width of the source. Degradation and retardation of contaminants were not considered in this analysis. Figure 3 is a schematic showing aspects of the subsurface SSL conceptual model used in the EPACMTP modeling effort. Appendix E is the background document prepared by EPA/OSW for this modeling effort EPACMTP Model Results. The results of the EPACMTP analyses indicate a DAF of about 170 for a 0.5-acre source at the 90th percentile protection level (Table 5). If a 95th percentile protection level is used, a DAF of 7 is protective for a 0.5-acre source. Table 5. Variation of DAF with Size of Source Area for SSL EPACMTP Modeling Effort Area (acres) 0.02 0.04 0.11 0.23 0.50 0.69 1.1 1.6 1.8 3.4 4.6 11.5 23 30 46 69 85th 1.42E+07 9.19E+05 5.54E+04 1.16E+04 2.50E+03 1 .43E+03 668 417 350 159 115 41 21 16 12 8.7 DAF 90th 2. 09 E +05 2.83E+04 2.74E+03 644 170 120 60 38 33 18 13 5.5 3.5 3.0 2.4 2.0 95th 946 211 44 15 7:0 4.5 3.1 2.5 2.3 1-7 1.6 1.2 1.2 1.1 1.1 1.1 Dilution Factor Modeling Effort. To gain further information on the national range and distribution of DAF values, EPA also applied the simple SSL water balance dilution model to ground water sites included in two large surveys of hydrogeologic site investigations. These were American Petroleum Institute's (API's) hydrogeologic database (HGDB) and EPA's database of conditions at Superfund sites contaminated with DNAPL. The HGDB contains the results of a survey sponsored by API and the National Water Well Association (NWWA) to determine the national variability in simple hydrogeologic parameters (Newell et al., 1989). The survey was conducted to validate EPA's use of the EPACML model as a screening tool for the land disposal of hazardous wastes. The survey involved more than 400 ground 49 TUT OOS 1186 water professionals who submitted data on aquifer characteristics from field investigations at actual waste sites and other ground water projects. The information was compiled in HGDB. which is available from API and is included in OASIS, an EPA-sponsored ground water decision support system. Newell et al. (1990) also present these data as "national average" conditions and by hydrogeologic settings based on those defined by Aller et al. (1987) for the DRASTIC modeling effort. Aller et al. (1987) defined these settings within the overall framework defined by Heath's ground water regions (Heath, 1984). The HGDB estimates of hydraulic conductivity and hydraulic gradient show reasonable agreement with those in Aller et al. (1987), which serves as another source of estimates for these parameters. The SSL dilution factor model (including the associated mixing zone depth model) requires estimates for five parameters: d, = aquifer thickness (m) L = length of source parallel to flow (m) I = infiltration rate (m/yr) K = aquifer hydraulic conductivity (m/yr) i = hydraulic gradient (m/m). Dilution factors were calculated by individual HGDB or DNAPL site to retain as much she-correlated parameter information as possible. The HGDB contains estimates of aquifer thickness (da), aquifer hydraulic conductivity (K), and aquifer hydraulic gradient (i) for 272 ground water sites. The aquifer hydraulic conductivity estimates were examined for these sites, and sites with reported values less than 5 x 10- 5 cm/s were culled from the database because formations with lower hydraulic conductivity values are not likely to be used as drinking water sources. In addition, sites in fractured rock or solution limestone settings were removed because the dilution factor model does not adequately address such aquifers. This resulted in 208 sites remaining in the HGDB. The DNAPL site database contains 92 site estimates of seepage velocity (v), which can be related to hydraulic conductivity and hydraulic gradient by the following relationship: V=Ki/n e (46) where ne = effective porosity. Effective porosity (ne) was assumed to be 0.35, which is representative of sand and gravel aquifers (the most prevalent aquifer type in the HGDB). Thus, for the DNAPL sites, 0.35xv was substituted for Ki in the dilution factor equation. Estimates of the other parameters required for the modeling effort are described below. Site-specific values were used where available. Because the modeling effort uses a number of site-specific modeling results to determine a nationwide distribution of dilution factors, typical values were used to estimate parameters for sites without site-specific estimates. Source Length (L). The contaminant source (i.e., area of soil contamination) was assumed to be square. This assumption may be conservative for sites with their longer dimensions perpendicular to ground water flow or nonconservatrve for sites with their longer dimensions parallel to ground water flow. The source length was calculated as the square root of the source area for the source sizes in question. To cover a range of contaminated soil source area sizes, five source sizes were modeled: 0.5 acre, 10 acres, 30 acres, 60 acres, and 100 acres. 50 TUT OO8 1187 Infiltration Rate (I). Infiltration rate estimates were not available in either database. Recharge estimates for individual hydrogeologic settings from Aller et al. (1987) were used as infiltration estimates (i.e.. it was assumed that infiltration = recharge). Because of differences in database contents, it was necessary to use different approaches to obtaining recharge/infiltration estimates for the HGDB and DNAPL sites. . The HGDB places each of its sites in one of the hydrogeologic settings defined by Aller et al. (1987). A recharge estimate for each HGDB site was simply extracted for the appropriate setting from Aller et al. The median of the recharge range presented was used (Table 6). The DNAPL database does Hot contain sufficient hydrogeologic information to place each site into the Aller et al. settings. Instead,-each of the 92 DNAPL sites was placed in one of Heath's ground water regions. The sites were found to lie within five hydrogeologic regions: nonglaciated central, glaciated central, piedmont/blue ridge, northeast and superior uplands, and Atlantic/Gulf coastal plain. Recharge was estimated for each region by averaging the median recharge value from all hydrogeologic settings except for those with steep slopes. The appropriate Heath region recharge estimate was then used for each DNAPL site in the dilution factor calculations. Aquifer Parameters. All aquifer parameters heeded'for the SSL dilution model are included in the HGDB. Because hydraulic conductivity and gradient are included in the seepage velocity estimates in the DNAPL site database, only aquifer thickness was unknown for these sites. Aquifer thickness for all DNAPL sites was set at 9.1 m, which is the median value for the "national average" condition in the HGDB (Newell et al., 1990). Dilution Modeling Results. Table 7 presents summary statistics for the 92 DNAPL sites, the 208 HGDB sites, and all 300 sites. One can see that the HGDB sites generally have lower dilution factors than the DNAPL sites, although the absolute range in values is greater in the HGDB. However, the available information for these sites is insufficient to fully explain the differences in these data sets. The wide range of dilution factors for these sites reflects the nationwide variability in hydrogeologic conditions affecting this parameter. The large difference between the average and geometric mean statistics indicates a distribution skewed toward the lower dilution factor values. The geometric mean represents a better estimate of the central tendency of such skewed distributions. Appendix F presents the dilution modeling inputs and results for the HGDB and DNAPL sites, tabulated by individual site. Selection Of the Default DAF. The default DAF was selected considering the evidence of the national DAF and dilution factor estimates described above. A DAF of 10 was selected in the December 1994 draft Soil Screening Guidance to be protective of a 30-acre source size. The EPACMTP model results showed a DAF of 3 for 30 acres at the 90th percentile The SSL dilution model results have geometric mean dilution factors for a 30-acre source of 10 and 7 for DNAPL sites and HGDB sites, respectively. In a weight of evidence approach, more weight was given to the results of the DNAPL sites because they are representative of the kind of sites to which SSLs are likely to be applied. Considering the conservative assumptions in the SSL dilution factor model (see Section 2.5.5), and the conservatism inherent in the soil partition methodology (see Section 2.5.4), EPA believes (1) that these results support the use of a DAF.of 10 for a 30-acre source, and (2) that this DAF will protect human health from exposure through this pathway at most Superfund sites across the Nation 51 TUT OOS Table 6. Recharge Estimates for DNAPL Site Hydrogeologic Regions Hydrogeologic tatting Recharge (m/yr) Mln. Max. Avg.Hydrogeologic setting Recharge (m/yr) Mln. Max. Avg. H C H H- H- CO -0 Nonglaelated Central (Region 8) Alluvial Mountain Valleys After. SS/LS/Sh.. thin Soil Alter. SS/LS/Sh., Deep Regollth Solution Limestone* Alluvium w/ Overbank Deposits Alluvium w/o Overbank Deposits Braided River Deposits Triassic Basins Swamp/Marsh Met./lg. Domes & Fault Blocks Unconsol./Semiconsol. Aquifers 0.10 0.10 0.10 0.25 0.18 0.18 0.10 0.10 0.10 0.00 0.00 0.18 0.18 0.18 0.38 0.25 0.25 0.18 0.18 0.18 0.05 0.05 Overall Average: Glaciated Central (Region 7) Glacial THI Over Bedded Rock Glacial THI Over Outwash Glacial Tffl Over Sol. Limestone Glacial Tilt Over Sandstone Glacial (fit Over Shale Outwash Outwash Over Bedded Rock* Outwash Over Solution Limestone' Moraine . Buried Valley > Alluvium w/Overbank Deposits Alluvium w/o Overbank Deposits' Glacial Lake Deposits Thin TM Over Bedded Rock- Beaches, B. Ridges/Dunes' Swamp/Marsh 0.10 0.10 0.10 0.10 0.10 0.18 0.25 0.25 0.18 0.18 0.10 0.25 0.10 0.18 0.25 0.10 0.18 0.18 0.18 0.18 0.18 0.25 0.38 0.38 0.25 0.25 0.18 0.38 0.18 0.25 0.38 0.18 0.14 0.14 0.14 0.32 0.22 0.22 0.14 0.14 0.14 0.03 0.03 0.15 Overall Average: 0.14 0.14 0.14 0.14 0.14 0.22 0.32 0.32 0.22 0.22 0.14 0.32 0.14 0.22 0.32 0.14 0.20 Piedmont/Blue Ridge (Region 8) Alluvial Mountain Valleys Regolfth River Alluvium Mountain Crests Swamp/Marsh 0.18 0.10 0.18 0.00 0.10 0.25 0.18 0.25 0.05 0.18 Overall Average: Northeast A Superior Uplands (Region 9) Alluvial Mountain VaHeys Tin Over Crystalline Bedrock Glacial Till Over Outwash Outwash' Moraine Alluvium w/ Overbank Deposits Alluvium w/o Overbank Deposits' Swamp/Marsh Bedrock Uplands Glacial Lake/Marine Deposits Beaches, B. Ridges. Dunes* 0.18 0.18 0.18 0.25 0.18 0.18 0.25 0.10 0.10 0.10 0.25 0.25 0.25 0.25 0.38 0.25 0.25 0.38 0.18 0.18 0.18 0.38 Overall Average: Atlantic/Gulf Coastal Plain (Region 10) Regional Aquifers Un/Semfconsol. Surftelal Aquifer* Alluvium w/ Overbank Deposits Alluvium w/o Overbank Deposits* Swamp* 0.22 0.14 0.22 0.03 0.14 0.15 0.22 0.22 0.22 0.32 0.22 0.22 0.32 0.14 0.14 0.14 0.32 0.22 0.00 0.05 0.03 0.25 0.38 0.32 0.18 0.25 0.22 0.25 0.38 0.32 0.25 0.38 0.32 Overall Average: 0.24 Source: Attar »t«/. (1997); hydrogoohglc regbru torn Heato (1994). '0.25 into 0.38m (9.8 in to IStyused as recharge n>ng« for 25+m setting values from Aloretal. (1987). Table 7. SSL Dilution Factor Model Results: DNAPL and HGDB Sites Source area (acres) DNAPL Sites (92) Geomean Average 10th percentile 25th percentile Median 75th percentile 90th percentile HGDB sites (208) Geomean Average 10th percentile 25th percentile Median 75th percentile 90th percentile All 300 sites Geomean Average 10th percentile 25th percentile Median 75th percentile 90th percentile 0.5 34 321 3 8 30 140 336 16 958 2 3 10 56 240 20 763 2 4 15 70 292 10 15 138 2 4 13 60 144 10 829 1 2 6 30 134 11 617 1 2 8 35 144 30 W 80 1 3 8 35 84 7 561 1 1 5 19 90 8 414 .1 2 5 23 88 100 6 44 1 2 5 20 46 5 371 1 1 3 12 51 6 271 1 1 4 13 49 600 4 19 1 1 3 9 20 3 159 1 1 2 5 21 3 116 1 1 2 6 21 DNAPL = DNAPL Site Survey (EPA/OERR). HGDB = Hydrogeologic database (API). To adjust the 30-acre DAF for a 0.5-acre source, EPA considered the geomean 0.5-acre dilution factors for the DNAPL sites (34), HGDB sites (16), and all 300 sites (20). A default DAF of 20 was selected as a conservative value for a 0.5-acre source size. This value also reflects the ratio between 0.5-acre and 30-acre geomean and median dilution factors calculated for the HGDB sites (2.2 and 2.0, respectively). The HGDB data reflect the influence of source size on actual dilution factors more accurately than the DNAPL site data because the HGDB includes site-specific estimates of aquifer thickness. As shown in the following section, aquifer thickness has a strong influence on the effect of source size on the dilution factor since it provides an upper limit on mixing zone depth. Increasing source area increases infiltration, which lowers the dilution factor, but also increases mixing zone depth, which increases the dilution factor. For an infinitely thick aquifer, these effects tend to cancel each other, resulting in similar dilution factors for 0.5 and 30 acres. Thin aquifers limit mixing depth for larger sources; thus the added infiltration predominates and lowers the dilution factors for the larger source. Since the DNAPL dilution factor 53 TUT 008 j.190 analyses use a fixed aquifer depth, they tend to overestimate the reduction in dilution factors that result from a smaller source. 2.5.7 Sensitivity Analysis. A sensitivity analysis was conducted to examine the effects of site-specific parameters on migration to ground water SSLs. Both the partition equation and the dilution factor model were considered in this analysis. Because an adequate database of national .distributions of these parameters was not available, a nominal range method was used to conduct the analysis. In this analysis, independent parameters were selected and each was taken to maximum and minimum values while keeping all other parameters at their nominal, or default, values. Overall, SSLs are most sensitive to changes in the dilution factor. As shown in Table 7, -die 10th to 90th percentile dilution factors vary from 2 to 292 for the 300 DNAPL and HGDB sites. Much of this variability can be attributed to the wide range of aquifer hydraulic conductivity across the Nation. In contrast, the most sensitive parameter in the partition equation (foc) only affects the SSL by a factor of 1.5. Partition Equation. The partition equation requires the following she-specific inputs: fraction organic carbon, average annual soil moisture content, and soil bulk density. Although volumetric soil moisture content is somewhat dependent on bulk density (in terms of the porosity available to be filled with water), calculations were conducted to ensure mat the parameter ranges selected do not result in impossible combinations of these parameters. Because the effects of the soil parameters on the .~SLs are highly dependent on chemical properties, the analysis was conducted on four organic chemicals spanning the range of these properties: chloroform, trichloroethylene, naphthalene, and benzo(a)pyrene. The range used for soil moisture conditions was 0.02 to 0.43 L water/L soil. The lower end of this range represents a likely residual moisture content value for sand, as might be found in the drier regions of the United States. The higher value (0.43) represents full saturation conditions for a loam soil. The range of bulk density (1.25 to 1.75) was obtained from the Patriot soils database, which contains bulk density measurements for over 20,000 soil series across the United States. Establishing a range for subsurface organic carbon content (foc) was more difficult, hi spite of an extensive literature review and contacts with soil scientists, very little information was found on the distribution of this parameter with depth in U.S. soils. The range used was 0.001 to 0.003 g carbon / g soil. The lower limit represents the critical organic carbon content below which the partition equation is no longer applicable. The upper limit was obtained from EPA's Environmental Research Laboratory in Ada, Oklahoma, as an expert opinion. Generally, soil organic carbon content falls off rapidly with depth. Since the typical value used as an SSL default for surface soils is 0.006, and 0.002 is used for subsurface soils, this limited range is consistent with the other default assumptions used in the Soil Screening Guidance. The results of the partition equation sensitivity analysis are shown in Table 8. For volatile chemicals, the model is somewhat sensitive to water content, with up to 54 and 19 percent change in SSLs for chloroform and trichloroethylene, respectively. The model is less sensitive to bulk density, with a high percent change of 18 for chloroform and 14 for trichloroethylene. Organic carbon content has the greatest effect on SSLs for all chemicals except cb> ,reform. As expected, the effect of foc increases with increasing K^. The greatest effect was seen for benzo(a)pyrene whose SSL showed a 50 percent increase at an f^ of 0.03. An foc of 0.005 will increase the benzo(a)pyrene SSL by 150 percent. 54 TUT 008 1191 Table 8. Sensitivity Analysis for SSL Partition Equation Ui r'O m K) Chloroform SSL Percent Parameter assignments (mg/kg) change Ail default parameter values Less conservative parameter value Organic carbon Bulk density Soil moisture More conservative parameter value Organic carbon Bulk density Soil moisture 0.59 — 0.67 14 0.69 18 0.74 26 0.51 -14 0.51 -13 0.27 -54 Trlchloroethylene Naphthalene SSL (mg/kg) 0.057 0.074 0.065 0.062 0.040 0.051 0.046 Benzo(tf)pyrene Percent SSL Percent SSL Percent change (mg/kg) change (mg/kg) change — 84 — 29 124 48 14 85 1 9 86 2 -29 44 -48 -10 83 -1 -19 80 -4 8 — 12 50 8 . 0 8 0 4 -50 8 0 8 0 Conservatism t Chemical-specific parameters Koc H' c« Input parameters Fraction org. carbon (g>g) Bulk density (kg/L) Average soil moisture (L/L) • n = 0.53; q, = 0.23. «> n = 0.34; qa = 0.04. - Chloroform 398E+01 1.50E-01 ,2.0c Less 0.003 1.25« 0.43 Nominal More 0.002 0.001 1.50 1.75b 0.30 0.02 Trlchloroethylene Naphthalene 1.66E+02 2.00E+03 4.22E-01 1.98E-02 0.1C 20d Benzo(a)pyrene 1 .02E+06 4.63E-05 0.004C cMCLx20DAF. <!HBL(HQ=1)x20DAF. Dilution Factor. Site-specific parameters for the dilution factor model include aquifer hydraulic conductivity (K), hydraulic gradient (i), infiltration rate (1), aquifer thickness (d), and source length parallel to ground water flow (L). Because they are sorru what dependent, hydraulic conductivity and hydraulic gradient were treated together as Darcy velocity (K x i). The parameter ranges used for the dilution factor analysis represent the 10th and 90th percentile values taken from the HGDB and DNAPL site databases, with the geometric mean serving as the nominal value, as shown in Table 9. Source length was varied by assuming square sources of 0.5 to 30 acres in size. Bounding estimates were conducted for each of these source sizes. The results in Table 9 show that Darcy velocity has the greatest effect on the dilution factor, with a range of dilution factors from 1.2 to 85 for a 30-acre source and 2.1 to 263 for a 0.5-acre source. Infiltration rate has the next highest effect, followed by source size and aquifer thickness. Note mat aquifer thickness has a profound effect on the influence of source size on the dilution factor. Thick aquifers show no source size effect because the increase in infiltration flux from a larger source is balanced by the increase in mixing zone depth, which increases dilution in the aquifer. For very thin aquifers, the mixing zone depth is limited by the aquifer thickness and the increased infiltration flux predominates, decreasing the dilution factor for larger sources. 2.6 Mass-Limit Model Development This section describes the development of models to solve the mass-balance violations inherent in the infinite source models used to calculate SSLs for the inhalation and migration to ground water exposure pathways. The models developed are not finite source models per se, but are designed for use with the current infinite source models to provide a lower, mass-based limit for SSLs for the migration to ground water and inhalation exposure pathways for volatile and leachable contaminants. For each pathway, the mass-limit model calculates a soil concentration that corresponds to the release of all contaminants present within the source, at a constant health-based concentration, over the duration of exposure. These mass-based concentration limits are used as a minimum concentration for each SSL; below this concentration, a receptor point concentration time-averaged over the exposure period cannot exceed the health-based concentration on which it is based. 2.6.1 Mass Balance Issues. Infinite source models are subject to mass balance violations under certain conditions. Depending on a compound's volatility and solubility and the size of the source, modeled volatilization or leaching rates can result in a source being depleted in a shorter time than the exposure duration (or the flux over a 30- or 70-year duration would release a greater mass of contaminants than are present). Several commenters to the December 1994 draft Soil Screening' Guidance expressed concern that it is unrealistic for total emissions over the duration of exposure to exceed the total mass of contaminants in a source. Using the soil saturation concentration (CMt) and a 5- to 10-meter contaminant depth, one commentor calculated that mass balance would be violated by the SSL volatilization model for 25 percent of the SSL chemicals. Short of finite source modeling, the limitations of which in soil screening are discussed in the draft Technical Background Document for Soil Screening Guidance (U.S. EPA, 1994i), there were two options identified for addressing mass-balance violations within the soil screening process: • Shorten the exposure duration to a value that would reflect mass limitations given the volatilization rate calculated using the current method 56 TUT 008 1193 Table 9. Sensitivity Analysis for SSL Dilution Factor Model Dilution Factor Source area Parameter assignments 30-acre 0.5-acre Mixing depth (m) Ratio of 0.5- acre/30-acre 30-acre 0.5 acre All central parameters Less conservative Darcy velocity Aquifer thickness Infiltration rate More conservative Darcy velocity Aquifer thickness Infiltration rate 5.2 '85 15 39 1.2 2.1 3.2 15 263 15 118 2.1 9.1 8.7 2.9 3.1 1.0 3.0 1.8 4.3 2.7 12 12 40 12 12 3.0 12 5.1 4.8 5.1 4.8 12 3.0 5.5 Conservatism input parameters Darcy velocity (DV, rrVyr) Aquifer thickness (da, m) Infiltration rate (rrvyr) Less 442 46 0.02 Nominal 22 12 0.18 More 0.8 3 0.35 Parameter sources Percentile 10th 25th 50th 75th 90th Average: Geomean: DV« <m/yr) 0.8 4 22 121 442 800 22 dab (m) 3.0 5.5 11 23 46 28 12 • 300 DNAPL & HGDB sites. t> 208 HGDB sites. 57 TUT 1194 • Change the volatilization rate to a value corresponding to the uniform release of the total mass of contaminants over the period of exposure. The latter approach was taken in the draft Risk-Based Corrective Action (RBCA) screening methodology developed by the American Society for Testing and Materials (ASTM) (ASTM, 1994) As stated on page B6 of the RBCA guidance (B.6.6.6): In the event that the time-averaged flux exceeds that which would occur if all chemicals initially present in the surficial soil zone volatilized during the exposure period, then the volatilization factor is determined from a mass balance assuming that all chemical initially present in the surficial soil zone volatilizes during the exposure period. This was selected over the exposure duration option because it is reasonably conservative for screening purposes (obviously, Inore contaminant cannot possibly volatilize from the soil) and it avoided the uncertainties associated with applying the current models to estimate source depletion rates. In summary, the mass-limit approach offers the following advantages: • It corrects the possible mass-balance violation in the infinite-source SSLs. • It does not require development of a finite source model to calculate SSLs. • It is appropriate for screening, being based on the conservative assumption that all of the contaminant present leaches or volatilizes over the period of exposure. • It is easy to develop and implement, requiring only very simple algebraic equations and input parameters that are, with the exception of source depth, already used to calculate SSLs. The derivation of these models is described below. It should be noted that the American Industrial Health Council (AJHC) independently developed identical models to solve the mass-balance violation as part of their public comments on the Soil Screening Guidance. 2.6.2 Migration to Ground Water Mass-Limit Model. For the migration to ground water pathway, the mass of contaminant leached from a contaminant source over a fixed exposure duration (ED) period can be calculated as (47) where MI = mass of contaminant leached (g) Cv, = leachate contaminant concentration (mg/L or.g/m3) 1 = infiltration rate (iri/yr) AS = source area (m2) ED - exposure duration (yr). 58 TUT OOS 1195 The total mass of contaminants present in a source can be expressed as where Q pb = C txp bxA sxd s = total mass of contaminant present (g) = total soil contaminant concentration (mg/kg or g/Mg, dry = dry soil bulk density (kg/L or Mg/m3) = source area (m2) = source depth (m). (48) To avoid a mass balance violation, the mass of contaminant leached cannot exceed the total mass of contaminants present (i.e., M| cannot exceed MT). Therefore, the maximum possible contaminant mass that can be leached from a source (assuming no volatilization or degradation) is MT and the upper limit for MI is M, = MT or Rearranging to solve for the total soil concentration (Q) corresponding to this situation (i.e. maximum possible leaching) Mass-Limit Model for Migration to Ground Water Pathway (49) Parameter/Definition (units) Default Cyscreening level in soil (mg/kg) Cw/target soil teachate concentration (mg/L) l/infittration rate (m/yr) ED/exposure duration (yr) Pt/dry soil bulk density (kg/L) dj/average source depth (m) (nonzero MCLG, MCL, or HBL) x 20 DAF site-specific 70 1.5 site-specif!? 59 TijT COS This soil concentration (Ct) represents a lower limit for soil screening levels calculated for the migration to ground water pathway. It represents the soil concentration corresponding to complete release of soil contaminants over the ED time period at a constant soil leachate concentration (Cw). Below this Q, the soil leachate concentration averaged over the ED time period cannot exceed C*,. 2.6.3 Inhalation Mass-Limit Model. The volatilization factor (VF) is basically the ratio of the total soil contaminant concentration to the air contaminant concentration. VF can be calculated as VF « (Q/C) x (Cf>/J,«v«) x 10-JO m*kg/cm2mg (50) where VF == volatilization factor (m3/kg) Q/C - inverse concentration factor for air dispersion (g/imz-s per kg/m3) CT° = total soil contaminant concentration at t=0 (mg/kg or g/Mg, dry basis) js»ve = average rate of contaminant flux from the soil to the air (g/cm2-s). The total amount of contaminant contained within a finite source can be written as Mt = CT° x pt, x A, x a1* (51) where M, - total mass of contaminant within the source (g) 'Cf0 = total soil contaminant concentration at t=0 (mg/kg or g/Mg, dry basis) Pb = soil dry bulk density (kg/L = Mg/m3) AS = area of source (m2) dj = depth of source (m). If all of the contaminant contained within a finite source is volatilized over a given averaging time period, the average volatilization flux can be calculated as J$«« * M,/[(AS x 10* cm2/m2) x (T x 3.15E7 s/yr)] (52) where T — exposure period (yr). Substituting Equation 51 for Mt in Equation 52 yields • js.ve = (Ci* x pb x d,) / (10* cm2/m2 x T x 3.15E7 s/yr) (53) Rearranging Equation 53 yields 60 . C-r°/Js«ve = (1Q4 cn»2/m2 x T x 3.15E7 s/yr)/(pb x d,) (54) Substituting Equation 54 into Equation SO yields Mass-Limit Model for Inhalation of Votatiles VF = (Q/C) x [(T x 3.15E7 s/yr)/(pb x d. x H* g/Mg)] (55) Parameter/Definition (units) VF/volatilization factor (m3/kg) Q/C/inverse of mean cone, at center of source (g/rrl-s per kg/m3) T/exposure interval (yr) Pb/dry soil bulk density (kg/L) dj/average source depth (m)_________ . Default Tables 30 1.5 site-specific If the VF calculated using an infinite source volatilization model for a given contaminant is less than the VF calculated using Equation 55, then the assumption of an infinite source may be too conservative for that specific contaminant at that source. Consequently, VF, as calculated in Equation 55, could be considered a minimum value for VF. 2.7 Plant Uptake Commentors have raised concerns that the ingestion of contaminated produce from homegrown gardens may be a significant exposure pathway. EPA evaluated empirical data on plant uptake, particularly the data presented in the Technical Support Document for Land Application of Sewage Sludge, often referred to as the "Sludge Rule" (U.S. EPA, 1992d). EPA found that empirical plant uptake-response slopes were available for selected metals but that available data were insufficient to estimate plant uptake of organics. In an effort to obtain additional empirical data, EPA has jointly funded research with the State of California on plant uptake of organic contaminants. These studies support ongoing revisions to the indirect, multimedia exposure model CalTOX. The Sludge Rule identified six metals of concern with empirical plant uptake data: arsenic, cadmium, mercury, nickel, selenium, and zinc. Plant uptake-response slopes were given for seven plant categories such as grains and cereals, leafy vegetables, root vegetables, and garden fruits. EPA evaluated the study conditions (e.g., soil pH, application matrix) and methods (e.g., geometric mean, default values) used to calculate the plant uptake-response slopes for each plant category and determined that the geometric mean slopes were generally appropriate for calculating SSLs for the soil-plant-human exposure pathway. However, the geometric mean of empirical uptake-response slopes from the Sludge Rule must be interpreted with caution for several reasons. First, the dynamics of sludge-bound metals may differ from the dynamics of metals at contaminated sites. For example, the empirical data were derived 61 TUT 008 .1198 from a variety of studies at different soil conditions using different forms of the metal (i.e.. salt vs. nonsalt). In studies where the application matrix was sludge, the adsorption power of sludge in the presence of calcium ions may have reduced the amount of metal that is bioavailable to plants and. therefore, plant uptake may be greater in non-sludge-amended soils. In addition to these confounding conditions, default values of 0.001 were assigned for plant uptake in studies where the measured value was below 0.001. A default value was needed to calculate the geometric mean uptake-response slope values. Moreover, considerable study-to-study variability is shown in the plant uptake-response slope values (up to 3 orders of magnitude for certain plant/metal combinations). This variability could result from varying soil characteristics or experimental conditions, but models have not been developed to relate changes in plant uptake to such conditions. Thus, the geometric mean values represent "typical" values from the experiments; actual values at specific sites could show marked variation depending on soil composition, chemistry, and/or plant type. OERR has used the information in the Sludge Rule to identify six metals (arsenic, cadmium, mercury, nickel, selenium, and zinc) of potential concern through the soil-plant-human exposure pathway for consideration on a site-specific basis. The fact that these metals have been identified should not be misinterpreted to mean that other contaminants are not of potential concern for this pathway. Other EPA offices are looking at empirical data and models for estimating plant uptake of organic contaminants from soils and OERR will incorporate plant uptake of organics once these efforts are reviewed and finalized. Methods for evaluating the soil-plant-human pathway are presented in Appendix G. Generic screening levels are calculated based on the uptake factors (i.e., bioconcentration factors [Br]) presented in the Sludge Rule. Generic plant SSLs are compared with generic SSLs based on direct ingestion as well as levels of inorganics in soil that have been reported to cause phytotoxicity (Will and Suter, 1994). Although site-specific factors such as soil type, pH, plant type, and chemical form will determine the significance of this pathway, the results of our analysis suggest that the soil-plant- human pathway may be of particular concern for sites with soils contaminated with arsenic or cadmium. Likewise, the potential for phytotoxicity will be greatly influenced by site-specific factors; however, the data presented by Will and Suter (1994) suggest that, with the exception of arsenic, the levels of inorganics that are considered toxic to plants are well below the levels that may impact human health via the soil-plant-human pathway. 2.8 intrusion of Volatiles into Basements: Johnson and Ettinger Model Concern about the potential impact of contaminated soil on indoor air quality prompted EPA to consider the Johnson and Ettinger (1991) model, a heuristic model for estimating the intrusion rate of contaminant vapors from soil into buildings. The model is a closed-form analytical solution for both convective and diffusive transport of vapor-phase contaminants into enclosed structures located above the contaminated soil. The model may be solved for both steady-state (i.e., infinite source) or quasi-steady-state (i.e., finite source) conditions. The model incorporates a number of key assumptions, including no leaching of contaminant to ground water, no sinks in the building, and well- mixed air volume within the building. To evaluate the effects of using the Johnson and Ettinger model on SSLs for volatile organic contaminants, EPA contracted Environmental Quality Management, Inc. (EQ), to construct a case example to estimate a high-end exposure point concentration for residential land use (Appendix H; EQ and Pechan, 1994). The case example models a contaminant source relatively close or directly beneath a building where the soil beneath the building is very permeable and the building is 62 TUT OOS 1.199 underpressurized, tending to pull contaminants into the basement Where possible and appropriate, values of model variables were taken directly from Johnson and Ettinger (1991). Using both steady- """" state and quasi-steady-state formulations, building air concentrations of each of 42 volatile SSL chemicals were calculated. The inverses of these concentrations were substituted into the inhalation SSL equations (Equations 4 or 5) as an indoor volatilization factor (VFfodoor) to calculate carcinogenic or noncarcmogenic SSLs based on migration of contaminants into basements (i.e.. "indoor inhalation" SSLs). Results showed a difference of up to 2 orders of magnitude between the steady-state and quasi-steady- state results for the indoor inhalation SSLs. Infinite source indoor inhalation SSLs were less than the corresponding "outdoor" inhalation SSLs by as much as 3 orders of magnitude for highly volatile constituents. For low-volatility constituents, the difference was considerably less, with no difference in the indoor and outdoor SSLs in some cases. The EQ study also indicated that the most important input parameters affecting long-term building concentration (and thus the SSL) are building ventilation rate, distance from the source (i.e., source-building separation), soil permeability to vapor flow, and source depth. For lower-permeability soils, the number and size of cracks in the basement walls may be more significant, although this was not a significant variable for the permeable soils considered in the study. - ' EPA decided against using the Johnson and Ettinger model to calculate generic SSLs due to the sensitivity of the model to parameters that do not lend themselves to standardisation on a national basis (e.g., source depth, the number and size of cracks in basement walls). In addition, the only formal validation study identified by EPA compares model results with measured radon concentrations from a highly permeable soil. Although these results compare favorably, it is not clear how applicable they are to less permeable soils and compounds not already present in soil as a gas (as radon is). /""""*• The model can be applied on a she-specific basis in conjunction with the results of a soil gas survey. Where land use is currently residential, a soil gas survey can be used to measure the vapor phase concentrations at the foundation of buildings, thereby eliminating the need to model partitioning of contaminants, migration from the source to the basement, and soil permeability. For future use scenarios, although some site-specific data are available, the difficulties are similar to those encountered with generic application of the model. Predictions must be made regarding the distance from the source to the basement and the permeability of the soil, basement floor, and walls. EQ's report models the potential impact of placing a structure directly above the source: Depending on the permeability of the surrounding soils, the results suggest that the level of residual contamination would have to be extremely low to allow for such a scenario. Distance from the source can have a dramatic impact on the results and should be considered in more detailed investigations involving future residential use scenarios. 63 TUT COS 12OO Part 3: MODELS FOR DETAILED ASSESSMENT The Soil Screening Guidance addresses the inhalation and migration to ground water exposure pathways with simple equations that require a small number of easily obtained soil parameters, meteorologic conditions, and hydrogeologic parameters. These equations incorporate a number of conservative simplifying assumptions—an infinite source, no fractionatioh between pathways, no biological or chemical degradation, no adsorption—conditions that can be addressed with more complicated models. Applying such models will more accurately define the risk of exposure via the inhalation or the migration to ground water pathway and, depending on site conditions, can lead to higher SSLs that are still protective. However, input data requirements and modeling costs make this option more expensive to implement than the SSL equations. This part of the Technical Background Document presents information on the selection and use of more complex fate and transport models for calculating SSLs. Generally, the decision to use these models will involve balancing costs: if the models and assumptions used to develop simple site- specific SSLs are overly conservative with respect to she conditions (e.g., a thick unsaturatcd : one), the additional cost and time required to apply these models may be offset by the potentiu cost savings associated with higher, but still protective, SSLs. Sections 3.1 and 3.2 include information on equations and models that can accommodate finite contaminant sources and fractionate contaminants between pathways (e.g., VLEACH and EMSOFT) and predict the subsequent impact on either ambient air or ground water. However, when using a finite source model, the site manager should recognize the uncertainties inherent in site-specific estimates of subsurface contaminant distributions and use conservative estimates of source size and concentrations to allow for such uncertainties. In addition, model predictions should be validated against actual site conditions to the extent possible. 3.1 Inhalation of Volatiles: Detailed Models Developing SSLs for the inhalation of volatiles involves calculating a site-specific volatilization factor (VF) and dispersion factor (Q/C). This section provides a brief description of finite source volatilization models with potential applicability to SSL development and information on site- specific application of the AREA-ST dispersion model for estimating the Q/C values needed to calculate both VF and PEF. It should not be viewed as an official endorsement of these models (other volatilization models may be available with applicability to SSL development). 3.1.1 Finite Source Volatilization Models. To identify suitable models for addressing a finite contaminant source, EPA contracted Environmental Quality Management, Inc (EQ), to conduct a preliminary evaluation of a number of soil volatilization models, including volatilization models developed by Hwang and Falco (1986), as modified by EQ (1992), and by Jury et al. (1983, 1984, and 1990) and VLEACH, a multipathway model developed primarily to assess exposure through the ground water pathway. Study results (EQ and Pechan, 1994) show reasonable agreement (\vnhin a factor of 2) between emission predictions using the modified Hwang and Falco or Jury models, but consistently lower predictions from VLEACH. However, Shan and Stephens (1995) discovered an error in the VLEACH calculation of the apparent diffusivity, which has been subsequently corrected. The corrected VLEACH model, version 2.2, appears to provide emission estimates similar to the Jury and the modified Hwang and Falco models. The revised VLEACH (v.2.2) 64 TUT O08 12O1 program is available from the Center for Subsurface Modeling Support (CSMOS) at EPA's Environmental Research Laboratory in ; Ada, OldialioiM (WWW.EPA.GOV/ADA/ CSMOS.HTML), and is discussed further in Section 3.2. For certain contaminant conditions, Jury et al. (1990) present a simplified equation (Jury's Equation Bl) for estimating the flux of a contaminant from a finite source of contaminated soil. The following assumptions were used to dehve this simplified flux equation: • Uniform soil properties (e.g., homogeneous average soil water content, bulk density, porosity, and fraction organic carbon) • Instantaneous linear equilibrium adsorption • Linear equilibrium liquid-vapor partitioning (Henry's law) • Uniform initial contaminant incorporation at t=0 • Chemicals in a dissolved form only (i.e., soil contaminant concentrations are below • No boundary layer thickness at ground level (no stagnant air layer) • No water evaporation or 'r^hing • No chemical reactions, biodegradation, or photolysis d, » (4DAt)1/2 (ramifications of mis are discussed below). Under these assumptions, the Jury et al. (1990) simplified finite source model is J* •?- C0(DA/Jtt)i/2[l-exp(KiI2/4DAt)] (56) where Js = contaminant flux at ground surface (g/cm2-s) C0 = uniform contaminant concentration at t=0 (g/cm3) DA = apparent diflfusivity (cmVs) « = 3.14 t = time (s) dj = depth of uniform soil contamination at t=0 (cm), and DA-Ke.i^DjH' + ewi^DwynZWpbKd + ew + e.H') (57 where 6. = air-filled soil porosity (L^/Liou) = n - 6W n = total soil porosity (LpoK/Ljmi) = 1 - (Pb/Pi) 6W - water-filled soil porosity (Lw.m/L.oii) - Pi, = soil dry bulk density (g/cm3) ps = soil particle density (g/cm3) w = average soil moisture content (g/g) 65 TUT 008 1202 pw = water density (g/cm3) Di = diffusivity in air (cm2/s) H' = dimensionless Henry's law constant = 41 x HLC HLC = Henry's law constant (atm-m3/mol) Dw = diftusivity in water (cm2/s) Kd = soil-water partition coefficient (cm3/g) = KOC foc KOC = soil organic carbon partition coefficient (cm3/g) foe ~ organic carbon content of soil (g/g). To estimate the average contaminant flux over 30 years, the time-dependent contaminant flux must be solved for various times and the results averaged. A simple computer program or spreadsheet can be used to calculate the instantaneous flux of contaminants at set intervals and numerically integrate the results to estimate the average contaminant flux. However, the time-step interval must be small enough (e.g., 1-day intervals) to ensure that the cumulative loss through volatilization is less than the total initial mass. Inadequate time steps can lead to mass-balance violations. To address this problem, EPA/ORD's National Center for Environmental Assessment has developed a computer modeling program, EMSOFT. The computer program provides an average emission flux over time by using an analytical solution to the integral, thereby eliminating the problem of establishing adequate time steps for numerical integration. In addition, the EMSOFT model can account for water convection (i.e., leaching), and the impact of a soil-air boundary layer on the flux of contaminants with low Henry's law constants. EMSOFT will be available through EPA's National Center for Environmental Assessment (NCEA) in Washington, DC. Once the average contaminant flux is calculated, VF is calculated as: VF = (Q/C) x (C0/pb) x (l/Js«ve) x 1(H m2/cm2 (58) where VF = volatilization factor (m3/kg) Q/C = inverse concentration factor for air dispersion (g/m^s per kg/m3) Q - uniform contaminant concentration at t=0 (g/cm3) pb = soil dry bulk density (g/cm3) jstve = average rate of contaminant flux (g/cm2-s). 3.1.2 Air Dispersion Models. The inverse concentration factor for air dispersion, Q/C. is used in the determination of both VF and PEF. For a detailed site-specific assessment of the inhalation pathway, a site-specific Q/C can be determined using the Industrial Source Complex Model platform in the short-term mode (ISCST3). Only a very brief overview of the application, assumptions, and input requirements for the model as used to determine Q/C is provided in this section. This model is the final regulatory version of the ISCST3 model. The 1SCST3 model FORTRAN code, executable versions, sample input and output files, description, and documentation can be downloaded from the "Other Models" section of the Office of Air Quality Planning and Standards (OAQPS) Support Center for Regulatory Air Models bulletin board system (SCRAM BBS). To access information, call: 66 TUT °°s 1203 OAQPS SCRAM BBS (919) 541-5742 (24 hours/day, 7 days/week except Monday AM) 1,200-9,600, 14,400 baud Line Settings: 8 bits, no parity, 1 stop bit Terminal Emulation: VT100 or ANSI System Operator (919) 541-5384 (normal business hours ESI). The user registers in the first call and then has full access to the BBS. The ISCST3 model will output an air concentration (in ug/m3) when the concentration model option is selected (e.g., CO MODELOPT DFAULT CONC rural/urban). The surface area of the contaminated soil source must be determined. For the ISCST3 model, the source location of an area source is defined by the coordinates of the southwest comer of the square (e.g., SO LOCATION sourcename AREA ->/2length -Vjwidth height=0). For the source parameter input line, the contaminant's area emission rate (in units of g/m2-s) must be entered. The Irea emission rate is the site-specific average emission flux rate, as calculated in Equation 56, converted to units of g/m2-s (i.e., Aremis = Js»ve x 104 cm^/m2). Alternatively, an area emission rate of 1 g/m2-* can be assumed. A grid or circular series of receptor sites should be used in and around the area source to identify the point of maximum contaminant air concentration. Hourly meteorologic data (*.MET files) for the nearest city (i.e., airport) of similar terrain and the preprocessor PCRAMMET also can be downloaded from the SCRAM BBS. The ISCST3 model output concentration is then used to calculate Q/C as Q/C - (J$«ve x 1(H cm2/m2)/(C^ x 10-9 kg/ug) (59 where Q/C = inverse concentration factor for air dispersion (g/m^-s per kg/m3) js«ve = average rate of contaminant flux (g/cm^s) C^ - ISC output maximum contaminant air concentration (ug/m3). Note: If an area emission rate of 1 g/m2-s is assumed, men (Js >ve * 104 cm2/m2) = 1, and Equation 59 simplifies to simply the inverse of the maximum contaminant air concentration (in kg/m3). 3.2 Migration to Ground Water Pathway For the migration to ground water pathway, the SSL equations assume an infinite source, contamination extending to the water table, and no attenuation due to degradation or adsorption in the unsaturated zone. At sites with small sources, deep water tables, confining layers in the unsaturated zone that can block contaminant transport, or contaminants that degrade through biological or chemical mechanisms, more complex models mat can address such she conditions can be used to calculate higher SSLs that still will be protective of ground water quality. This section provides information on the use of such models in the soil screening process to calculate a dilution- attenuation factor (Section 3.2.1) and to estimate contaminant release in leachate and transport through the unsaturated zone (Section 3.2.2). 67 TUT COS 1.204 3.2.1 Saturated Zone Models. EPA has developed guidance for the selection and application of saturated zone transport and fate models and for interpretation of model applications. The user is referred to Ground Water Modeling Compendium, Second Edition 1994 (U.S. EPA. 1994b) and Framework for Assessing Ground Water Modeling Applications (U.S. EPA, 1994a) for further information. More complex saturated zone models can be used to calculate a dilution-attenuation factor (DAF) that, unlike the SSL dilution model, can consider attenuation in the aquifer. Some can handle a finite source through a transient mode that requires a time-stepped concentration from a finite-source unsaturated zone model (see Section 3.2.2). In general, to calculate a DAF using such models, the contaminant concentration at the water table under the source (Cw) is set to unity (e.g., 1 mg/L). The DAF is the reciprocal of the predicted concentration at the receptor point (Cu>) as follows: DAF=CW/CW.= I/CRP (60 3.2.2 Unsaturated Zone Models. In an effort to provide useful information for model application, EPA's ORD laboratories in Ada, Oklahoma, and Athens, Georgia, conducted an evaluation of nine unsaturated zone fate and transport models (Criscenti et al., 1994; Nofziger et al.. 1994). The results of this effort are summarize:! here. The models reviewed are only a subset of the potentially appropriate models available to the public and are not meant to be construed as having received EPA approval. Other models also may be applicable to SSL development, depending on site- specific circumstances. Each of the unsaturated zone models selected for evaluation are capable, to varying degrees, of simulating the transport and transformation of chemicals in the subsurface. Even the most unique site conditions can be simulated by either a single model or a combination of models. However, the intended uses and the required input parameters of these models vary. The models evaluated include: • RTTZ (Regulatory and Investigative Treatment Zone model) • VIP (Vadose zone Interactive Process model) CMLS (Chemical Movement in Layered Soils model) HYDRUS SUMMERS (named after author) MULTIMED (MULTIMEDia exposure assessment model) VLEACH (Vadose zone LEACHing model) SESOIL (SEasonal SOIL compartment model) PRZM-2 (Pesticide Root Zone Model). RITZ, VIP, CMLS, and HYDRUS were evaluated by Nofziger et al. (1994). SUMMERS. MULTIMED, VLEACH, SESOIL, and PRZM-2 were evaluated by Criscenti et al. (1994). These documents should be consulted for further information on model application and use. The applications, assumptions, and input requirements for the nine models evaluated are described in this section. The model descriptions include model solution method (i.e., analytical, numerical), the 68 •"LIT cos 120; purpose of the model, and descriptions of the methods used by the model to simulate water/contaminant transport and contaminant transformation. Each description is accompanied by a table of required input parameters. Input parameters discussed include soil properties, chemical properties, meteorologic data, and other site information. In addition, certain input control parameters may be required such as time stepping, grid discretization information, and output format. Information on determining general applicability of the models to subsurface conditions is provided, followed by an assessment of each model's potential applicability to the soil screening process. RITZ. Information on the RTTZ model was obtained primarily from Nofziger et al. (1994). RTTZ is a steady-state analytical model used to simulate the transport and fate rf chemicals mixed with oily wastes (sludge) and disposed of by land treatment. RITZ simulates two layers of the soil column with uniform properties. The soil layers consist of: (1) the upper plow zone where the oily waste is applied and (2) the treatment zone. The bottom of the treatment zone is the water table. It is assumed in the model that the oily waste is completely mixed in and does not migrate out of the plow zone, which represents the contaminant source at an initial time. RITZ also assumes an infinite source (i.e., a continuous flux at constant concentration). The flux of water is assumed to be constant with time and depth and the Clapp-Hornberger constant is used in defining the soil water content resulting from a specified recharge rate. Sorption, vapor transport, volatilization, and biochemical degradation are also considered (van der Heijde, 1994). Partitioning between phases is instantaneous, linear, and reversible. Input parameters required for the RITZ model are presented in Table 10. Biochemical degradation of the oil and contaminant is considered to be a first-order process, and dispersion in the water phase is ignored. Table 10. Input Parameters Required for RITZ Model Soil properties Site characteristics Pollutant properties OH properties Percent organic carbon Plow zone depth Bulk density Saturated water content Saturated hydraulic conductivity Clapp-Hornberger constant Treatment zone depth Recharge rate (constant) Evaporation rate (constant) Air temperature (constant) Relative humidity (constant) Sludge application rate Diffusion coefficient (water vapor in oil) Concentration in sludge Koc Henry's law constant Degradation half-life (constant) Diffusion coefficient (in air) Concentration of oil in sludge Density of oil Degradation half-life of oil VIP. Information on the VIP model was obtained from Nofziger et al. (1994). The VIP model is a one-dimensional, numerical (finite-difference) fate and transport model also designed for simulating the movement of compounds in the unsaturated zone resulting from land application of oily wastes. Like the RITZ model, VIP considers dual soil zones (a plow zone and a treatment zone) and considers 69 fUT 008 1206 the source to be infinite. VIP differs from RTTZ in that it solves the governing differential equations numerically, which allows variability in the flux of water and chemicals over time. Advection and hydrodynamic dispersion are the primary transport mechanisms for the contaminant in water (van der Heijde. 1994). Instead of assuming instantaneous, linear equilibrium between all phases. VIP considers the partitioning rates between the air, oil, soil, water, and vapor-phase transport. Contaminant transformation processes include hydrolysis, volatilization, and sorption. Oxygen- limited degradation and diffusion of the contaminant in the air phases are also considered. Sorption is instantaneous as described for the RTTZ model. The input parameters required for the VIP model are presented in Table 11. Table 11. Input Parameters Required for VIP Model Soil properties Porosity Bulk density Site characteristics Plow zone depth Treatment zone depth Pollutant properties Concentration in sludge Oil-water partition coefficient8 Oxygen properties Oft-air partition coefficient* Water-air partition coefficient8 Oil properties Density of oil Degradation rate constant of oil Saturated hydraulic conductivity Clapp-Homberger constant Mean daily recharge rate Temperature (each layer) Sludge application rate Sludge density Application period and frequency in period Weight fraction water in sludge Weight fraction oil in waste Air-water partition coefficient* Soil-water partition coefficient* Degradation constant in oil8 Degradation constant in water8 Dispersion coefficient Adsorption-desorption rate constant (water/oil) Adsorption-desorption rate constant (water/soil) Adsorption-desorption rate constant (water/air) Oxygen half-saturation constant in air phase * Oxygen half-saturation constant in oil phase a Oxygen half-saturation constant in water phase8 Oxygen naff-saturation constant (oil degradation) Stoichiometric ratio of oxygen to pollutant consumed Stoichiometric ratio of oxygen to oil consumed Oxygen transfer rate coefficient between oil and air phases Oxygen transfer rate coefficient between water and air phases • Parameters required tor plow zone and treatment zone. CMLS. Information on CMLS was obtained from Nofziger et al. (1994). CMLS is an analytical model developed as a management tool to describe the fate and transport of pesticides in layered soils and to estimate the amount of chemical at a certain position at a certain time. The model allows designation of up to 20 soil layers with uniform soil and chjmical properties defined for each layer. 70 008 12O7 Water in the soil system is "pushed ahead" of new water (recharge) entering the system. The water content is reduced to the field capacity after each infiltration event, and water is removed from the root zone in proportion to the available water stored in that layer (Nofziger et al., 1994). CMLS assumes movement of the chemical in liquid phase only and allows a finite source. Chemical partitioning between the soil and the water is assumed to be linear, instantaneous, and reversible. Volatilization is not considered. Dispersion and diffusion of the chemical is ignored and degradation is .defined as a first-order process. The input parameters required for the CMLS model are presented in Table 12. Table 12. Input Parameters Required for CMLS Soil properties_____ Site characteristics Chemical properties Depth of bottom of soil layers Daily infiltration or precipitation Degradation half-life (each soil layer) Organic carbon content Daily evapotranspifation Amount applied Bulk density — Depth of application Saturated water content — Date of application Field capacity — Permanent wilting point — HYDRUS. Information on the HYDRUS model was obtained from Nofziger et al. (1994). HYDRUS is a finite-element model for one-dimensional solute fate and transport simulations. The boundary conditions for flow, as well as soil and chemical properties, can therefore vary with time. A finite source also can be modeled. Soil parameters are described by the van Genuchten parameters. The model also considers root uptake and hysteresis in the water movement properties. Solute transport and transformation incorporates molecular diffusion, hydrodynamic dispersion, linear or nonlinear equilibrium partitioning (sorption), and first-order decay (van der Heijde, 1994). Volatilization is not considered. The input parameters required by HYDRUS are presented in Table 13. SUMMERS. Information oh the SUMMERS model was obtained from Criscenti et al (1994). SUMMERS is a one-dimensional analytical model that simulates one-dimensional, nondispersive transport in a single layer of soil from an infinite source. It was developed to determine the contaminant concentrations in 'soil that would result in ground water contamination above specified levels for evaluating geothermal energy sites. The model is similar to the SSL equations in that it assumes steady-state water movement and equilibrium partitioning of the contaminant in the unsaturated zone and performs a mass-balance calculation of mixing in an underlying aquifer. For the saturated zone, the model assumes a constant flux from the surface source and instantaneous, complete mixing in the aquifer. The mixing depth is therefore defined by the thickness of the aquifer. The model does not account for volatilization. The input parameters required for SUMMERS are listed in Table 14. 71 TUT OO8 1208 Table 13. Input Parameters Required for HYDRUS Soil properties Depth of soil layers Saturated water content Saturated hydraulic conductivity Bulk density Retention parameters Residual water content Site characteristics Uniform or stepwise rainfall intensity Contaminant concentrations in soil — — — *~ Pollutant properties Molecular diffusion coefficient Dispersivity Decay coefficient (dissolved) Decay coefficient (adsorbed) FreundKch isotherm coefficients w Root uptake parameters Power function in stress- response function Pressure head where transpiration is reduced by 50% Root density as a function of depth •» —— ^" Table 14. Input Parameters Required for SUMMERS . _____________Parameters required______. __________ Target concentration in ground water Thickness of aquifer • Volumetric infiltration rate into aquifer Width of pond/spill perpendicular to flow Downward porewater velocity Initial (background) concentration Ground water seepage velocity Equilibrium partition coefficient Void fraction Darcy velocity in aquifer Horizontal area of pond or spill___________Volumetric ground water flow rate____ MULTIMED. Information on the MULTIMED model was obtained from Chscenti et al. (1994) and Salhotra et al. (1990). MULTIMED was developed as a multimedia fate and transport model to simulate contaminant migration from a waste disposal unit. For this review, only the fate and transport of pollutants from the soil to migration to ground water pathway was considered in detail. In MULTIMED, infiltration of waste into the unsaturated or saturated zones can be simulated t tg a landfill module or by direct infiltration to the unsaturated or saturated zones. Flow in the unsaturated zone and for the landfill module is simulated by a one-dimensional, semianalytical module. Transport in the unsaturated zone considers the effects of dispersion, sorption, volatilization, biodegradation, and first-order chemical decay. The saturated transport module is also one-dimensional, but considers three-dimensional dispersion, linear adsorption, first-order decay, and dilution due to recharge. Mixing in the underlying saturated zone is based on the vertical dispersivtty specified, the length of the disposal facility parallel to the flow direction, the thickness of the saturated zone, the ground water velocity, and the infiltration rate. The saturated zone module can simulate steady-state and transient ground water flow and thus can consider a finite, source assumption through a leachate "pulse duration." The parameters required for the unsaturated and saturated zone transport in MULTIMED are presented in Table 15. 72 TUT 008 1209 Table 15. Input Parameters Required for MULTIMED Unsaturated zone parameters Saturated hydraulic conductivity Porosity Air entry pressure head Depth of unsatureted zone Residual water content Number of porous materials Number of layers Alpha coefficient van Genuchten exponent Thickness of each layer Longitudinal dispersivity Percent organic matter Soil bulk density Biological decay coefficient Acid, base, and neutral - hydrolysis rates Reference temperature Normalized distribution coefficient Air diffusion coefficient Reference temperature for air diffusion Molecular weight Infiltration rate Area of waste disposal unit Duration of pulse Source decay constant Initial concentration at landfill Particle diameter Saturated zone parameters Recharge rate First-order decay coefficient Biodegradation coefficient Aquifer thickness Hydraulic gradient Longitudinal dispersivity Transverse dispersivity Vertical dispersivity Temperature of aquifer pH Organic carbon content Well distance from site Angle off-center of well Well Vertical distance VLEACH. Information on the VLEACH model was obtained from Criscenti et al. (1994). VLEACH is a one-dimensional, finite difference model developed to simulate the transport of contaminants displaying bnear partitioning behavior through the vadose zone to the water table by aqueous advection and diffusion. Multiple layers can be modeled and are expressed as polygons with different soil properties and recharge rates. Water flow is assumed -to be steady state. Linear equilibrium partitioning is used to determine chemical concentrations between the aqueous, gaseous, and adsorbed phases (sorption and volatilization), and a finite source can be considered. Chemical or biological degradation is not considered. The input parameters required for VLEACH are presented in Table 16. Table 16. Input Parameters Required for VLEACH Soil properties Chemical characteristics Site properties Dry bulk density Total porosity Volumetric water content Fractional organic carbon. Koc Henry's law constant Aqueous solubility Free air diffusion coefficient Recharge rate Contaminant concentrations in recharge Depth to ground water Dimensions of 'polygons'' 73 TUT SESOIL. Information on the SESOIL model was obtained from Criscemi et al. (1994). SESO1L is a one-dimensional, finite difference flow and transport model developed for evaluating the movement of contaminants through the vadose zone. The model contains three components: (1) hydrologic cycle. (2) sediment cycle, and (3) pollutant fate cycle. The model estimates the rate of vertical solute transport and transformation from the land surface to the water table. Up to four layers can be simulated by the model and each layer can be subdivided into 10 compartments with uniform soil characteristics. Hydrologic data can be included using either monthly or annual data options. Solute transport is simulated for ground water and surface runoff including eroded sediment. Pollutant fate considers equilibrium partitioning to soil and air phases (sorption and diffusion), volatilization from the surface layer, first-order chemical degradation, biodegradation, cation exchange, hydrolysis, and metal complexation and allows for a stationary free phase. The required input parameters for SESOIL are presented in Table 17 for the monthly option. Table 17. input Parameters Required for SESOIL (Monthly Option) Climate data Soil data Chamical data Application data Mean air temperature8 Mean cloud cover tract ion8 Mean relative humidity3 Short wave albedo tract k>na Total precipitation Mean storm duration Number of storm events Number of layers and sublayers Thickness of layers pH of each layer Bulk density Intrinsic permeability Pore disconnectedness index Effective porosity Organic carbon content Cation exchange capacity Freundlich exponent Sift, sand, and clay fractions Soil loss ratio Solubility in water Air diffusion coefficient Henry's law constant Organic carbon adsorption ratio Soil adsorption coefficient Molecular weight Valence Hydrolysis constants (acid, base, neutral) Biodegradation rates (liquid, solid) Ligand stability constant Moles ligand per mole compound Molecular weight of ligand Ligand mass Application area She latitude Spill index Pollutant load Mass removed or transformed Index of volatile diffusion Index of transport in surface runoff Ratio pollutant cone, in rain to solubility Washload area Average slope and slope length Erodibility factor Practice factor Manning coefficient * SESOIL uses these parameters to calculate evapotranspiration if an evapotranspiration value is not specified 74 TU'T 008 1211 PRZM-2. Information on PRZM-2 was obtained from Criscenti et al. (1994). PR2M-2 is a combination of two models developed to simulate the one-dimensional movement of chemicals in the unsaturated and saturated zones. The first model, PRZM, is a finite difference model that simulates water flow and detailed pesticide fate and transformation in the unsaturated zone. The second model. VADOPT, is a one-dimensional finite element model with more detailed water movement simulation capabilities. The coupling of these models results in a detailed representation of contaminant transport and transformation in the unsaturated zone. PRZM has been used predominantly for evaluation of pesticide leaching in the root zone. PRZM uses detailed meteorologic and surface hydrology data for the hydrologic simulations. Runoff, erosion, plant uptake, leaching, decay, foliar washoff, and volatilization are considered in the surface hydrologic and chemical transport components. Chemical transport and fate in the subsurface is simulated by advection, dispersion, molecular diffusion, first-order chemical decay, biodegradation, daughter compound progeny, and soil sorption. The input parameters required for PRZM are presented in Table 18. VADOPT can be run independently of PRZM and output from the PRZM model can be used to set the boundary conditions for VADOPT. The lower boundaries could also be specified as a constant pressure head or zero velocity. Transport simulations consider advection and diffusion with sorption and first-order decay. The input requirements for VADOPT are presented in Table 19. Considerations for Unsaturated Zone Model Selection. The accuracy of a model in a site-specific application depends on simplifications and assumptions implicit in the model and their relationship to site-specific conditions. Additional error may be introduced from assumptions made when deriving input parameters. Although each of the nine models evaluated has been tested and validated for simulation of water and contaminant movement in the unsaturated zone, they are different in purpose and complexity, with certain models designed to simulate very specific scenarios. A model should be selected to accommodate a site-specific scenario as closely as possible. For example, if contaminant volatilization is of concern, the model should consider volatilization and vapor phase transport. After a model is determined to be appropriate for a site, contaminant(s), and conditions to be modeled, the site-specific information available (or potentially available) should be compared to the input requirements for the model to ensure mat adequate inputs can be developed. The unsaturated zone models addressed in this study use either analytical, semianalytical: or numerical solution methods. Analytical models represent the simplest models, requiring the least number of input parameters. They use a closed-form solution for the pertinent equations. In analytical models, certain assumptions have to be made with respect to the geometry of the system and external stresses. For this reason, there are few analytical flow models (van der Heijde. 1994). Analytical solutions are common, however, for fate and transport problems by solution of convection-dispersion equations. Analytical models require the assumption of uniform flow conditions, both spatially and temporally. Semianalytical models approximate complex analytical solutions using numerical techniques (van der Heijde, 1994). Transient or steady-state conditions can be approximated using a semianalytical model. However, spatial variability in soil or aquifer conditions cannot be accommodated. Numerical models use approximations of pertinent partial differential equations usually by finite- difference or finite-element methods. The resolution of the area and time of simulation is defined by the modeler. Numerical models may be used when simulating time-dependent scenarios, spatially variable soil conditions, and unsteady flow (van der Heijde, 1994). 75 TUT 008 1212 Table 18. Input Parameters Required for PRZM Daily climate data Pan evaporation and pan factor Temperature Precipitation Monthly daylight hours Windspeed Solar radiation Snowmelt factor Minimum evaporation extraction depth Erosion data Topographic factor/soil credibility Average duration of rainfall Field area Practice factor Crop data Surface condition of crop Maximum dry weight of crop after harvest Maximum interception storage Maximum rooting depth Emergence, maturation, and harvest dates Maximum canopy coverage Pesticide data Application quantity Number of applications (50 maximum) Number of chemicals (3 maximum) Application dates Foliar extraction coefficient Diffusion coefficient in air Initial concentration levels Incorporation depth Enthalpy of vaporization Parent/daughter transform rates Plant uptake factor KdandKoc Aqueous, sorbed, vapor decay rates Foliar decay rates Henry's law constant Soil data Compartment thicknesses Soil drainage parameter Wilting point Runoff curve numbers Hydrodynamic dispersion Percent organic carbon Core depth Bulk density Field capacity Number and thickness of horizons Initial soil water content Soil temperature Heat capacity per unit volume Thermal conductivity of horizon Biodegradation and Albedo Avgerage monthly bottom boundary temperature • irrigation parameters Reflectivity of soil surface Initial horizon temperature (not presented) Height of windspeed measurement Sand and day content 76 TUT O08 1213 Table 19. Input Parameters Required for VADOFT Pesticide data Soil data Number of chemicals Number of soil horizons Relative permeability vs. saturation Aqueous decay rate Horizon thicknesses Pressure head vs. saturation Initial concentration Saturated hydraulic conductivity Residual water phase saturation Longitudinal dispersivity Effective porosity Brooks and Corey n Retardation coefficient Air entry pressure head van Genuchten alpha Molecular diffusion — — Cone, flux at first node Input flux or head at first node (if independent of PRZM) (if independent of PRZM)_______________________' ___________ In certain cases, input parameters to be used in a model are not definitively known. Some models allow some input parameters to be expressed as probability distributions rather than a single value, referred to as Monte Carlo simulations. This method can provide an estimate of the uncertainty of the model output (i.e., percent probability that a contaminant will be greater than a certain concentration at a depth), but requires knowledge of the parameter distributions. Alternatively, a bounding approach can be used to estimate the effects of likely parameter ranges on model results where there is uncertainty in input parameter values. Model Applicability to SSLs. The unsaturated models evaluated herein can provide inputs necessary for soil screening by calculating leachate concentrations at the water table or by calculating infiltration rates. In the former application, they produce results comparable to the leach test option. As with the leach test, the leachate concentration from the model is divided by the dilution factor to obtain an estimated ground water concentration at the receptor well. This receptor point concentration is then compared with the acceptable ground water concentration to determine if a site's soils exceed SSLs. Table 20 summarizes characteristics and capabilities of the models evaluated for this study. All nine of the models can calculate contaminant concentrations in leachate that has infiltrated down to the water table from the vadose zone, although CMLS requires a separate calculation to estimate leachate concentration. If there is reliable site data indicating significant degradation in soil, several of the models can consider biological and/or chemical degradation processes. The models also can address contaminant adsorption; those that can model layered soils can be especially useful in settings where low-permeability clay layers may attenuate contaminants through adsorption. Finally, several of the models can address a finite source if the size of the source is accurately known. The average annual infiltration rate at a site is difficult to measure in the field yet is required for estimating a dilution factor or DAF. Four of the models evaluated, CMLS, HYDRUS, SESOIL, and PRZM, can calculate infiltration rates given either daily or monthly rainfall data. . Two models, VLEACH and SESOIL, address volatilization from the soil surface along with leachate emissions and therefore may be useful for SSL development for the volatilization and migration to ground water pathways. The volatile emission portion of VLEACH is discussed in Section 3.1. 77 TUT 008 1214 Table 20. Characteristics of Unsaturated Zone Models Evaluated Model RITZ VIP CMLS HYDRUS MULTIMED SUMMERS PRZM-2 SESOIL VLEACH Type "5 • . • Semlanalytical • 3 . • • • • Fate Finite source . • • • • Partitioning with oil phase •. and Transport Processes Considered Volatilization • • • • • • Vapor phase transport . • i Hydrodynamic dispersion • • • • g • • • • t • • • • . • • NonequHferlum partitioning . • Hydrolysis (Ist-order decay) • • • • Btodegradation • Layered soils • • • Root zone uptake • • i * OttV.T Saturated zone included • • «• s 4?w i.' €• • • Water balance calculations • • • • Table 20 addresses only unsaturated zone fate and transport model components, although two models (MULTIMED and SUMMERS) have saturated zone flow and transport capabilities. The fol'. 'wing text highlights some of the differences between the models, outlines their advantage and disadvantages, and describes appropriate scenarios for model application. RITZ. RITZ was designed to model land treatment units and is appropriate for sites where oily wastes are present (it includes sorption on an immobile oil phase as well as onto soil panicles). Sorption, degradation, volatilization, and first-order decay processes are considered hi the subrurface simulations. The most significant drawback for the model is the limit on the number of soil layers. Optimally, RITZ would be recommended for modeling .chemical migration in a uniform unsaturated zone as a result of land application. Although the oil phase can be omitted for simulations of scenarios without oily materials, the RITZ model's focus on oily waste degradation in land treatment units limits its utility for soil screening (SSLs are not applicable when soils contain a separate oil phase). VIP. VIP also is appropriate for sites where release of oily wastes has occurred. Some of the limitations described in RITZ also apply to the VIP model. VIP could be used as a followup model to RITZ since variable chemical and water fluxes can be simulated. In this case, significant add. onal 78 TUT oos input parameters are required to simulate transient partitioning between the air. soil, water, and oil phases. Like RTTZ, VIP's focus on land treatment of oily waste limits its application to SSLs. CMLS. CMLS differs from RITZ and VIP in mat it allows designation of up to 20 soil layers with different properties It does not consider nonaqueous phase liquids, dispersion, diffusion, or vapor phase transport, but a finite source can be modeled. CMLS estimates the location of the peak concentration of contaminants through a layered soil system. A limitation of the CMLS model for SSL application is that it does not calculate leachate concentrations. Instead, it calculates the amount of chemical at a certain depth at a certain time. The user must estimate the concentration based on the amount of chemical present and the total flux of water in the system (Nofziger et-al., 1994). The model is typically used to estimate the time for a chemical entering the unsaturated zone to reach a certain depth. HYDRUS. Like CMLS, the HYDRUS model can also simulate chemical movement in layered soils and can consider a finite source, but also includes dispersion and diffusion as well as sorption and first- order decay. In addition, HYDRUS outputs the chemical concentration in the soil water as a function of time and depth along with the amount of chemical remaining in the soil. The model considers root zone uptake, but other models such as PRZM should be used if the comprehensive effects of plant uptake are to be considered in the simulations. Because it can estimate infiltration from rainfall contaminant concentrations, HYDRUS may be useful in SSL applications. SUMMERS. The SUMMERS model is a relatively simple model designed to simulate leaching in the unsaturated zone and is essentially identical to the SSL migration to ground water equations in assumptions and limitations. It is appropriate for use as an initial screening model where site data are limited and where volatilization is not of concern. However, since attenuation processes such as biodegradation, first-order decay, volatilization, or other attenuation processes (other than sorption) are not considered, it is a quite conservative model. Since volatilization is not considered, it cannot be used to simulate migration of volatile compounds to the atmosphere. Because of its similarities to the SSL migration to ground water equations, the SUMMERS model is not suitable for a more detailed assessment of site conditions. MULTIMED. MULTIMED simulates simple vertical water movement in the unsaturated zone. Since an initial soil concentration cannot be specified, either the soil/water partition equation or a leaching test (SPLP) must be used to estimate soil leachate contaminant concentrations. MULTIMED is appropriate for simulating contaminant migration in soil and can be used to model vadose zone attenuation of leachate concentrations derived from a partition equation (see Section 2.5.1). In addition, since it links the output from the unsaturated zone transport module with a saturated zone module, it can be used to determine the concentration of a contaminant in a well located downgradient from a contaminant source. MULTIMED is appropriate for early-stage site simulations because the input parameters required are typically available and uncertainty analyses can be performed using Monte Carlo simulations for those parameters for which reliable values are not known. VLEACH. In VLEACH, biological or chemical degradation is not considered. It therefore provides conservative estimates of contaminant migration in soil. This model may be appropriate as an initial screening tool for sites for which there is little information available. VLEACH can estimate volatile emissions (see Section 3.1) and can consider a finite source. It is therefore potentially applicable to both subsurface pathways addressed by the soil screening process. 79 TUT OOS 1216 SESOIL. SESOIL was designed as a screening tool, but it is actually more complex than some of the models described. Some of the input data would be cumbersome to obtain, especially for use as an initial screening tool. It is applicable for simulating spill sites since it allows consideration of surface transport by erosion and runoff and can utilize detailed meteorologic information to estimate infiltration. In the soil zone, several fate and transport options are available such as metal complexation, hydrolysis, cation exchange, and degradation. This model is especially applicable to sites where significant subsurface and meteorologic information is available. Although the moucl does consider volatilization from surface soils, the available documentation (Criscenti et al., 1994) is not clear as to whether it produces an output of volatile flux to the atmosphere. PRZM-2. PRZM-2 is a relatively detailed model as a result of the coupling .of the two models PRZM and VADOFT. Although PRZM is predominantly used as a pesticide leaching model, it could also be used for simulation of transport of other chemicals. Because detailed meteorology and surface application parameters can be included, it is appropriate for simulation of surface spills or land disposal scenarios. In addition, uncertainty analyses can be performed based on Monte Carlo simulations. Numerous subsurface fate and transport options exist in PRZM. Water movement is somewhat simplified in PRZM, and it may not be applicable for low-permeability soils (Criscenti et al., 1994). However, water flow simulation is more detailed in the VADOFT module of the PRZM-2 program. The combination of these programs makes PRZM-2 a relatively complex model. This model is especially applicable to sites for which significant site and meteorologic data are available. SO TUT OOS 1217 Part 4: MEASURING CONTAMINANT CONCENTRATIONS IN SOIL Hie Soil Screening Guidance includes a sampling strategy for implementing the soil screening process. Section 4.1 presents the sampling approach for surface soils. This approach provides a simple decision rule based on comparing the maximum contaminant concentrations of composite samples with surface soil screening levels (the Max test) to determine whether further investigation is needed for a particular exposure area (EA). In addition, this section presents a more complex strategy (the Chen test) that allows the user to design a site-specific quantitative sampling strategy by varying decision error limits and soil contaminant variability to optimize the number of samples and composites. Section 42 provides a subsurface soil sampling strategy for developing SSLs and applying the screening procedure for the volatilization and migration to ground water exposure pathways. Section 4.3 describes the technical details behind the development of the SSL sampling strategy, including analyses and response to public and peer-review comments received on the December 1994 draft guidance. The sampling strategy for the soil screening process is designed to achieve the following objectives: • Estimate mean concentrations of contaminants of concern for comparison with SSLs • Fill in the data gaps in the conceptual site model necessary to develop SSLs. The soils of interest for the first objective differ according to the exposure pathway being addressed. For the direct ingestion, dermal, and fugitive dust pathways, EPA is concerned about surface soils The sampling goal is to determine average contaminant concentrations of surface soils in exposure areas of concern. For inhalation of volatiles, migration to ground water and, in some cases, plant uptake, subsurface soils are the primary concern. For these pathways, the average contaminant concentration through each source is the parameter of interest. The second objective (filling in the data gaps) applies primarily to the inhalation and migration to ground water pathways. For these pathways, the source area and depth as well as average soil properties within the source are needed to calculate the pathway-specific SSLs. Therefore, the sampling strategy needs to address collection of these site-specific data. Because of the difference in objectives, the sampling strategies for the ingestion pathway and for the inhalation and migration to ground water pathways are addressed separately. If both surface and subsurface soils are a concern, then surface soils should be sampled first because the results of surface soil analyses may help delineate source areas to target for subsurface sampling. At some sites, a third sampling objective may be appropriate. As discussed in the Soil Screening Guidance, SSLs may not be useful at sites where background contaminant levels are above the SSLs. Where sampling information suggests that background contaminant concentrations may be a concern, background sampling may be necessary. Methods for Evaluating the Attainment of Cleanup Standards - Volume 3: Reference-Based Standards for Soil and Solid Media (U.S. EPA, 1994e) provides further information on sampling soils to determine background conditions at a site. 81 TIT DOS In order to accurately represent contaminant distributions at a site. EPA used the Data Quality Objectives (DQO) process (Figure 4) to develop a sampling strategy that will satisfy Superfund program objectives. The DQO process is a systematic data collection planning process developed by EPA to ensure that the right type, quality, and quantity of data are collected to support EPA decision making. As shown in Sections 4.1.1 through 4.1.6, most of the key outputs of the DQO process already have been developed as part of the Soil Screening Guidance. The DQO activities addressed in this section are described in detail in the Data Quality Objectives for Superfund: Interim Final Guidance (U.S. EPA, 1993b) and the Guidance for the Data Quality Objectives Process (U.S. EPA: 1994c). Refer to these documents for more information on how to complete each DQO activity or how to develop other, site-specific sampling strategies. State the Problem * Identify the Decision * Identify Inputs to the Decision * Define the Study Boundaries * Develop a Decision Rule * Specify Limits on Decision Errors - Optimize the Design for Obtaining __________Data__________ 4.1 Sampling Surface Soils A sampling strategy for surface soils is presented in this section, organized by the steps of the DQO process. The first five steps of mis process, from defining the problem through developing the basic decision rule, are summarized in Table 21, and are described in detail in the first five subsections. The details of the two remaining steps of the DQO process, specifying limits on decision errors and optimizing the design, have been developed separately for two alternative hypothesis testing procedures (the Max test and the Chen method) and are presented in four (4.1.6, 4.1.7, 4.1.9, and 4.1.10) subsections. In addition, a data quality assessment (DQA) follows the DQO process step for optimizing the design. The DQA ensures that site-specific error limits are achieved. Sections 4.1.8 and 4.1.11 describe the DQA for the Max and Chen tests, .respectively. The technical details behind the development of the surface soil sampling design strategy are explained in Section 4.3. 4.1.1 State the Problem. In screening, the problem is to identify the contaminants and exposure areas (EAs) that do not pose significant risk to human health so that future investigations can be focused oh the areas and contaminants of concern at a site. Figure 4. The Data Quality Objectives process. The main site-specific activities involved in this first step of the DQO process include identifying the data collection planning team (including technical experts and key stakeholders) and specifying the available resources. The list of technical experts and stakeholders should contain all key personnel who are involved with applying the Soil Screening Guidance at the site. Other activities in this step include developing the conceptual site model (CSM), identifying exposure scenarios, and preparing a summary description of the surface soil contamination problem The User's Guide (U.S. EPA, 1996) describes these activities in with more detail. 4.1.2 Identify the Decision. The decision is to determine whether the mean surface soil concentrations exceed surface soil screening levels for specific contaminants within EAs. If so, the EA must be investigated further. If not, no further action is necessary under CERCLA for the specific contaminants in the surface soils of those EAs. 82 TUT DOS 1219 Table 21. Sampling Soil Screening DQOs for Surface Soils POO Process Steps Soil Screening Inputs/Output* State the Problem identify scoping team Develop conceptual site model (CSM) Define exposure scenarios Specify available resources VVrite brief summary of contamination problem_____________ Site manager and technical experts (e.g., toxicologists, risk assessors, statisticians, soil scientists) CSM development (described in Step 1 of the User's Guide, U.S. EPA, 1996) Direct ingestion and inhalation of fugitive particulates in a residential setting; dermal contact and plant uptake for certain contaminants Sampling and analysis budget, scheduling constraints, and available personnel Summary of the surface soH contamination problem to be investigated at the site Identify the Decision Identify decision Identify alternative actions Do mean soil concentrations for particular contaminants (e.g.. contaminants of potential concern) exceed appropriate screening levels? Eliminate area from further study under CERCLA or Plan and conduct further investigation___________________ Identify Inputs to the Decieion Identify inputs Define basis for screening Identify analytical methods Ingestion and paniculate inhalation SSLs for specified contaminants Measurements of surface soil contaminant concentration Soil Screening Guidance Feasfcle analytical methods (both field and laboratory) consistent with program-level requirements____________________ Define the Study Boundaries Define geographic areas of field investigation Define population of interest Divide site into strata Define scale of decision making Define temporal boundaries of study Identify practical constraints The entire NPL site (which may include areas beyond facility boundaries), except for any areas with clear evidence that no contamination has occurred Surface soils (usually the top 2 centimeters, but may be deeper where activities could redistribute subsurface soils to the surface) Strata may be defined so that contaminant concentrations are likely to be relatively homogeneous within each stratum based on the CSM and field measurements Exposure areas (EAs) no larger than 0.5 acre each (based on residential land use) Temporal constraints on scheduling field visits Potential impediments to sample collection, such as access, health, and safety issues______________________________' Develop a Decision Rule Specify parameter of interest Specify screening level Specify "if..., then..." decision rule True mean" (u) individual contaminant concentration in each EA. (since the determination of the "true mean* would require the collection and analysis of many samples, the "Wax Test" uses another sample statistic, the maximum composite concentration). Screening levels calculated using available parameters and site data (or generic SSLs If site data are unavailable). If the "true mean" EA concentration exceeds the screening level, then investigate the EA further. If the "true mean" is less than the screening level, then no further investigation of the EA is required under CERCLA. 83 TtJT 008 1220 4.1.3 Identify Inputs to the Decision. This step of the DQO process requires identifying the inputs to the decision process, including the basis for further investigation and the applicable analytical methods. The inputs for deciding whether to investigate further are the ingestion, dermal, and fugitive dust inhalation SSLs calculated for the site contaminants as described in Part 2 of this document, and the surface soil concentration measurements for those same contaminants. Therefore, the remaining task is to identify Contract Laboratory Program (CLP) methods and/or field methods for which the quantitation limits (QLs) are less than the SSLs. EPA recommends the use of field methods, such as soil gas surveys, immunoassays, or X-ray fluorescence, where applicable and appropriate as long as quantitation limits are below the SSLs. At least 10 percent of field samples should be split and sent to a CLP laboratory for confirmatory analysis (U.S. EPA, 1993d). 4.1.4 Define the Study Boundaries. This step of the DQO process defines the sample population of interest, subdivides the she into appropriate exposure areas, and specifies temporal or practical constraints on the data collection. The description of the population of interest must include the surface soil depth. Sampling Depth. When measuring soil contamination levels at the surface for the ingestion and inhalation pathways, the top 2 centimeters is usually considered surface soil, as defined by Urban Soil Lead Abatement Project (U.S. EPA 1993f). However, additional sampling beyond this depth may be appropriate for surface soils under a future residential use scenario in areas where major soil disturbances can reasonably be expected as a result of landscaping, gardening, or construction activities. In this situation, contaminants that were at depth can be moved to the surface. Thus, it is important to be cognizant of local residential construction practices when determining the depth of surface soil sampling and to weigh the likelihood of that area being developed. Subdividing the Site. This step involves dividing the site into areas or strata depending on the likelihood of contamination and identifying areas with similar contaminant patterns. These divisions can be based on process knowledge, operational units, historical records, and/or prior sampling. Partitioning the site into such areas and strata can lead to a more efficient sampling 'design for the entire site. For example, the site manager may have documentation that large areas of the site are unlikely to have been used for waste disposal activities. These areas would be expected to exhibit relatively low variability and the sampling design could involve a relatively small number of samples. The greatest intensity' of sampling effort would be expected to focus on areas of the site where there is greater uncertainty or greater variability associated with contamination patterns. When relatively large variability' in contaminant concentrations is expected, more samples are required to determine with confidence whether the EA should be screened out or investigated further. Initially, the site may be partitioned into three types of areas: 1. Areas that are not likely to be contaminated 2. Areas that are known to be highly contaminated 3. Areas that are suspected to be contaminated and cannot be ruled out. Areas that are not likely to be contaminated generally will not require further investigation if this assumption is based on historical site use information or other she data that are reasonably complete and accurate. (However, the site manager may also want take a few samples to confirm this assumption). These may be parts of the she that are within the legal boundaries of the property but 84 TUT 008 1221 were completely undisturbed by hazardous-waste-generating activities. All other areas need investigation. Areas that are known to be highly contaminated (i.e., sources) are targeted for subsurface sampling. The information collected on source area and depth is used to calculate she-specific SSLs for the inhalation and migration to ground water pathways (see Section 4.2 for more information). Areas that are suspected to be contaminated (and cannot be ruled out for screening) are the primary subjects of the surface soil investigation. If a geostatistician is available, a geostatistical model may be used to characterize these areas (e.g., kriging model). However, guidance for this type of design is beyond the scope of the current guidance (see Chapter 10 of U.S. EPA, 1989a). Defining Exposure Areas. After the she has been partitioned into relatively homogeneous areas, each region that is targeted for surface soil sampling is then subdivided into EAs. An EA is defined as that geographical area in which an individual may be exposed to contamination over time. Because the SSLs were developed for a residential scenario, EPA assumes the EA is a suburban residential lot corresponding to 0.5 acre. For soil screening purposes, each EA should be 0.5 acre or less. To the extent possible, EAs should be constructed as square or rectangular areas that can be subdivided into squares to facilitate compositing and grid sampling. If the she is currently residential, then the EA should be the actual residential lot size. The exposure areas should not be laid out in such a way that they unnecessarily combine areas of high and low levels of contamination. The orientation and exact location of the EA, relative to the distribution of the contaminant in the soil, can lead to instances where sampling of the EA may lead to results above the mean, and other instances, to results below the mean. Try to avoid straddling contaminant "distribution units" within the 0.5 acre EA. The sampling strategy for surface soils allows investigators to determine mean soil contaminant concentration across an EA of interest. An arithmetic mean concentration for an EA best represents the exposure to site contaminants over a long period of time. For risk assessment purposes, an individual is assumed to move randomly across an EA over time, spending equivalent amounts of time in each location. Since reliable information about specific patterns of nonrandom activity for future use scenarios is not available, random exposure appears to be the most reasonable assumption for a residential exposure scenario. Therefore, spatially averaged surface soil concentrations are .used to estimate mean exposure concentrations. Because all the EAs within a given stratum should exhibit similar contaminant concentrations, one site-specific sampling design can be developed for all EAs within that stratum. As discussed above, some strata may have relatively low variability and other strata may have relatively high variability. Consequently, a different sampling design may be necessary for each stratum, based upon the stratum-specific estimate of the contaminant variability. 4.1.5 Develop a Decision Rule. Ideally, the decision rule for surface soils is: If the mean contaminant concentration within an EA exceeds the screening level, then investigate that EA further. This "screening level" is the actual numerical value used to compare against the she contamination data. It may be identical to the SSL, or h may be a multiple of the SSL (e.g., 2 SSL) for a hypothesis test designed to achieve specified decision error rates in a specified region above and below the SSL. In addition, another sample statistic (e.g., the maximum concentration) may be used as an estimate of the mean for comparison with the "screening level." 85 TUT COS 1222 4.1.6 Specify Limits on Decision Errors for the Max Test. Sampling data will be used to support a decision about whether an EA requires further investigation. Because of variability in contaminant concentrations within an EA, practical constraints on sample sizes, and sampling or measurement error, the data collected may be inaccurate or nonrepresentative and may mislead the decision maker into making an incorrect decision. A decision error occurs when sampling data mislead the decision maker into choosing a course of action that is different from or less desirable than the course of action that would have been chosen with perfect information (i.e., with no constraints on sample size and no measurement error). EPA recognizes that data obtained from sampling and analysis are never perfectly representative and accurate, and that the costs of trying to achieve near-perfect results can outweigh the benefits. Consequently, EPA acknowledges mat uncertainty in data must be tolerated to some degree. The DQO process controls the degree to which uncertainty in data affects the outcomes of decisions that are based on those data. This step of the DQO process allows the decision maker to set limits on the probabilities of n^ing an incorrect decision. The DQO process utilizes hypothesis tests to control decision errors. When performing a hypothesis test, a presumed or baseline condition, referred to as the "null hypothesis" (H0), is established. This baseline condition is presumed to be true unless the data conclusively demonstrate otherwise, which is called "rejecting the null hypothesis" in favor of an alternative hypothesis. For the Soil Screening Guidance, the baseline condition, or HO, is that the site needs further investigation. When the hypothesis test is performed, two possible decision errors may occur 1 . Decide not to investigate an EA further (i.e., "walk away") when the correct decision (with complete and perfect information) would be to "investigate further" 2 . Decide to investigate further when the correct decision would be to "walk away." Since the site is on the NPL, she areas are presumed to need further investigation. Therefore, the data must provide clear evidence that it would be acceptable to "walk away." This presumption provides the basis for classifying the two types of decision errors. The "incorrectly walk away" decision error is designated as the Type I decision error because one has incorrectly rejected the baseline condition (null hypothesis). Correspondingly, the "unnecessarily investigate further" decision error is designated as the Type II decision error. To complete the specification of limits on decision errors, Type I and Type II decision error probability limits must be defined in relation to the SSL. First a "gray region" is specified with respect to the mean contaminant concentration within an EA. The gray region represents the range of contaminant levels near the SSL, where uncertainty in the data (i.e., the variability) can make the decision "too close to call." In other words, when the average of the data values is very close to the SSL, it would be too expensive to generate a data set of sufficient size and precision to resolve what the correct determination should be. (i.e., Does the average concentration fall "above" or "below" the SSL?) The Soil Screening Guidance establishes a default range for the width and location of the "gray region", from one-half the SSL (0.5 SSL) to two tunes the SSL (2 SSL). By specifying the upper edge of the gray region as twice the SSL, it is possible mat exposure areas with mean values slightly higher than the SSL may be screened from further study. However, EPA believes that the exposure scenario 86 TUT 008 and assumptions used to derive SSLs are sufficiently conservative to be protective in such cases. On the lower side of the gray region, the consequences of decision errors at one-half the SSL are primarily financial. If the lower edge of the gray region were to be moved closer to the SSL. then more exposure areas that were truly below the SSL would be screened out, but more money would be spent on sampling to make this determination. If the lower edge of the gray region were to be moved closer to zero, then less money could be spent on sampling, but fewer EAs that were truly below the SSLs would be screened out, leading-to unnecessary investigation of EAs. Hie Superfund program chose the gray region to be one-half to two times the SSL after investigating several different ranges. This range for the gray region represents a balance between the costs of collecting and analyzing soil samples and making incorrect decisions. While it is desirable to estimate exactly the exposure area mean, the number of samples required are much more than project managers are generally willing to collect in a "screening" effort. Although some exposure areas will have contaminant concentrations that are between the SSL and twice the SSL and will be screened out, human health will still be protected given the conservative assumptions used to derive the SSLs. The Soil Screening Guidance establishes the following goals for Type I and Type II decision error rates: Prob ("walk away" when the true EA mean is 2 SSL) = 0.05 Prob ("investigate further" when the true EA mean is 0.5 SSL) = 0.20. This means that there should be no more than a 5 percent chance mat the site manager will "walk away" from an EA where the true mean concentration is 2 SSL or more. In addition, there should be no more than a 20 percent chance that the site manager will unnecessarily investigate an EA when the mean is 0.5 SSL or less. These decision error limits are general goals for the soil screening process. Consistent with the DQO process, these goals may be adjusted on a site-specific basis by considering the available resources (i.e., time and budget), the importance of screening surface soil relative to other potential exposure pathways, consequences of potential decision errors, and consistency with other relevant EPA guidance and programs. Table 22 summarizes this step of the DQO process for the Max test, specifying limits on the decision error rates, and the final step of the DQO process for the Max test, optimizing the design. Figure 5 illustrates the gray region for the decision error goals: a Type I decision error rate of 0.05 (5 percent) at 2 SSL and a Type II decision error rate of 0.20 (20 percent) at 0.5 SSL 4.1.7 Optimize the Design for the Max Test. This section provides instructions for developing an optimum sampling strategy for screening surface soils. It discusses compositing, the selection of sampling points for composited and uncomposited surface soil sampling, and the recommended procedures for determining the sample sizes necessary to achieve specified limits on decision errors using the Max test. 87 TUT 008 1224 Table 22. Sampling Soil Screening DQOs for Surface Soils under the Max Test DQO Process Steps Soil Screening inputs/Outputs Specify Limits on Decision Errors* Define baseline condition (null hypothesis) 'Define the gray region** Define Type I and Type II decision errors Identify consequences Assign acceptable probabilities of Type I and Type II decision errors Define QA/QC goals The EA needs further investigation From 0.5 SSL to 2 SSL. Type I error. Do not investigate further ("walk away from*) an EA whose true mean exceeds the screening level of 2 SSL Type II error. Investigate further when an EA's true mean falls below the screening level of 0.5 SSL Type I error, potential public health consequences Type II error, unnecessary expenditure of resources to investigate further Goals: Type 1:0.05 (5%) probability of not investigating further when true mean" of the EA is 2 SSL Type II: 0.20 (20%) probability of investigating further when "true mean" of the EA is 0.5 SSL CLP precision and bias requirements 10% CLP analyses for field methods____________________ Optimize the Design Determine how to best estimate true mean" Determine expected variability of EA surface soil contaminant concentrations Design sampling strategy by evaluating costs and performance ol alternatives Develop planning documents for the field investigation __ ___________ Samples composited across the EA estimate the EA mean (x). Use maximum composite concentration as a conservative estimate of the true EA mean. A conservatively large expected coefficient of variation (CV) from prior data for the site, field measurements, or data from other comparable sites and expert judgment A minimum default CV of 2.5 should be used when information is insufficient to estimate the CV. Lowest cost sampling design option (i.e., compositing scheme and number of composites) that will achieve acceptable decision error rates Sampling and Analysis Plan (SAP) Quality Assurance Project Plan (QAPjP)___________________ Since the DQO process controls the degree to which uncertainty in data affects the outcome of decisions that are based on that data, specifying limits on decision errors will allow the decision maker to control the probability of making an incorrect decision when using the DQOs. The gray region represents the area where the consequences of decision errors are minor (and uncertainty in sampling data makes decisions too close to call). 88 TUT O08 1225 1.0 Probability of Deciding that the Mean Exceeds the Screening Level Tolerable Type I Decision Error' Rates Decision Enw Rates (Relatively Large Decision Error Rates are Considered Tolerable.) cue 0.5 x SSL SSL 2 x SSL True Mean Contaminant Concentration Figure 5. Design performance goal diagram. Note that the size, shape, and orientation of sampling volume (i.e., "support") for heterogenous media have a significant effect on reported measurement values. For instance, particle size has a varying affect on the transport and fate of contaminants in the environment and on the potential receptors. Because comparison of data from methods that are based on different supports can be difficult, defining the sampling support is important in the early stages of site characterization. This may be accomplished through the DQO process with existing knowledge of the site, contamination, and identification of the exposure pathways that need to be characterized. Refer to Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies (U.S. EPA, 1992f) for more information about soil sampling support. The SAP developed for surface soils should specify sampling and analytical procedures as well as the development of QA/QC procedures. To identify the appropriate analytical procedures, the screening levels must be known. If data are not available to calculate site-specific SSLs, then the generic SSLs in Appendix A should be used. Compositing. Because the objective of surface soil screening is to ensure that the mean contaminant concentration does not exceed the screening level, the physical "averaging", that occurs during compositing is consistent with the intended use of the data. Compositing' allows a larger number of locations to be sampled while controlling analytical costs because several discrete samples are physically mixed (homogenized) and one or more subsamples are drawn from the mixture and submitted for analysis. If the individual samples in each composite are taken across the EA, each composite represents an estimate of the EA mean. A practical constraint to compositing in some situations is the heterogeneity of the soil matrix. The 89 TUT 008 1226 efficiency and effectiveness of the mixing process may be hindered when soil particle sizes van' widely or when the soil matrix contains foreign objects, organic matter, viscous fluids, or stick} material. Soil samples should not be composited if matrix interference among contaminants is likely (e.g., when the presence c one contaminant biases analytical results for another). Before individual specimen* are composited for chemical analysis, the site manager should consider homogenizing and splitting each specimen. By compositing one portion of each specimen with the other specimens and storing one portion for potential future analysis, the spatial integrity of each specimen is maintained. If the concentration of a contaminant in a composite sample is high, the splits of the individual specimens from which it was composed can be analyzed discretely to determine which individual specimens) have high concentrations of the contaminant. This will permit the site manager to determine which portion within an EA is contaminated without making a repeat visit to the site. Sample Pattern. The Max test should only be applied using composite samples that are representative of the entire EA. However, the Chen test (see Section 4.1.9) can be applied with individual, uncomposited samples There are several options for developing a sampling pattern for compositing that produce samples that should be representative. If individual, uncomposited samples will be analyzed for contaminant concentrations, the N sample points can be selected using either (1) simple random sampling (SRS), (2) stratified SRS, or (3) systematic grid sampling (square or rectangular grid) with a random starting point (SyGS/rs). Step-by-step procedures for selecting SRS and SyGS/rs samples are provided in Chapter 5 of the U.S. EPA (1989a) and Chapter 5 of U.S. EPA (1994e). If stratified random sampling is used, the sampling rate must be the same in every sector, or stratum of the EA. Hence, the number of sampling points assigned to a stratum must be directly proportional to the surface area of the stratum. Systematic grid sampling with a random starting point is generally preferred because it ensures that the sample points will be dispersed across the entire EA. However, if the boundaries of the EA are irregular (e.g., around the perimeter of the site or .the boundaries of a stratum within which the EAs were defined), the number of grid sample points that fall within the EA depends on the random starting point selected. Therefore, for these irregularly shaped EAs, SRS or stratified SRS is recommended. Moreover, if a systematic trend of contamination is suspected across the EA (e.g., a strip of higher contamination), then SRS or stratified SRS is recommended again. In this case, grid sampling would be likely to result in either over- or under representation of the strip of higher contaminant levels, depending on the random starting point. For composite sampling, the sampling pattern used to locate the discrete sample specimens that form each composite sample (N) is important. The composite samples should be formed in a manner that is consistent with the assumptions underlying the sample size calculations. In particular, each composite sample should provide an unbiased estimate of the mean contaminant concentration over the entire EA. One way to construct a valid composite of C specimens is to divide the EA into C sectors, or strata, of equal area and select one point at- random from each sector. If sectors (strata) are of unequal sizes, the simple average is no longer representative of the EA as a whole. Five valid sampling patterns and compositing schemes for selecting N composite samples that each consist of C specimens are listed below: ' . 1. Select an SRS consisting of C points and composite all specimens associated with these points into a sample. Repeat this process N times, discarding any points that were used in a previous sample. 90 TUT O08 1227 2. Select an SyGS/rs of C points and composite all specimens associated with the points in this sample. Repeat this process N times, using a new randomly selected starting point each time. 3. Select a single SyGS/rs of CxN points and use the systematic compositing scheme that is described in Highlight 3 to form N composites, as illustrated in Figure 6. 4. Select a single SyGS/rs of CxN points and use the random compositing scheme that is described in Highlight 4 to form N composites, as illustrated in Figure 7. 5. Select a stratified random sample of CxN points and use a random compositing scheme, as described in Highlight 5, to form N composites, as illustrated in Figure 8. Methods 1, 2, and 5 are the most statistically defensible, with method 5 used as the default method in the Soil Screening Guidance. However, given the practical limits of implementing these methods, either method 3 or 4 is generally recommended for EAs with regular boundaries (e.g., square or rectangular). As noted above, if the boundaries of the EA are irregular, SyGS/rs sampling may not result in exactly CxN sample points. Therefore, for EAs with irregular boundaries, method 5 is recommended. Alternatively, a combination of methods 4 and 5 can be used for EAs that can be partitioned into C sectors of equal area of which K have regular boundaries and the remaining C - K have irregular boundaries. Additionally, compositing within sectors to indicate whether one sector of the EA exceeds SSLs is an option that may also be considered. See Section 4.3.6 for a full discussion. Sample Size. This section presents procedures to determine sample size requirements for the Max test that achieve the she-specific decision error limits discussed in Section 4.1.6. The Max test is based on the maximum concentration observed in N composite samples that each consist of C individual specimens. The individual specimens are selected so that each of the N composite samples is representative of the site as a whole, as discussed above. Hence, this section addresses determining the sample size pair, C and N, that achieves the site-specific decision error limits. Directions for performing the Max test in a manner that is consistent with DQOs established for a site are presented later in this section. Table 23 presents the probabilities of Type I errors at 2 SSL and Type II errors at 0.5 SSL (the boundary points of the gray region discussed in Section 4.1.6) for several sample size options when the variability for concentrations of individual measurements across the EA ranges from 100 percent to 400 percent (CV - 1.0 to 4.0). Two choices for the number, C, of specimens per composite are shown in this table: 4 and 6. Fewer than four specimens per composite is not considered sufficient for the Max test. Fewer than four specimens per composite does not achieve the decision error limit goals for the level of variability generally encountered at CERCLA sites. More than six specimens may be more than can be effectively homogenized into a composite sample. The number, N, of composite samples shown in Table 23 ranges from 4 to 9. Fewer than four samples is not considered sufficient because, considering decision error rates from simulation results (Section 4.3), the Max text should be based on at least four independent estimates of the EA mean. More than nine composite samples per EA is generally unlikely for screening surface soils at Superfund sites. However, additional sample size options can be determined from the simulation results reported in Appendix I. 91 TUT 008 1228 Highlight 3: Procedure for Compositing of Specimens from a Grid Sample Using a Systematic Scheme (Figure 6) 1. Lay out a square or triangular grid sample over the EA, using a random start. Step-by-step procedures can be found in Chapter 5 of U.S. ERA (1989a). The number of points in the grid should be equal to CxN, where C is the desired number of specimens per composite and N is the desired number of composites. 2. Divide;the EA into C sectors (strata) of equal area and shape such that each sector contains the same number of sample points. The number of sectors (C) should be equal to the number of specimens in each composite (since one specimen per area will be used in each composite) and the number of points within each sector, N, should equal the desired number of composite samples. 3. Label the points within one sector in any arbitrary fashion from 1 to N. Use the same scheme for each of the other sectors. 4. Form composite number 1 by compositing specimens with the '1' label, form composite number 2 by compositing specimens with the *2' label, etc. This leads to N composite samples that are subjected to chemical analysis. • 1 «2 •3 f>4 •5 f>6 • 1 «2 •3 «4 •5 • f>6 • 1 «2 •4 •3 *4 •5 «6 • 1 «2 •3 -f>4 •5 f>6 Figure 6. Systematic (square grid point*) sample with systematic compositing scheme (6 composite samples consisting of 4 specimens). 92 TUT 1229 Highlight 4: Procedure for Compositing of Specimens from a Grid Sample Using a Random Scheme (Figure 7) 1. Lay out a square or triangular grid sample over the EA, using a random start. Step-by-step procedures can be found in Chapter 5 of U.S. EPA (1989a). The number of points in the grid should be equal to CxN, where C is the desired number of specimens per composite and N is the desired number of composites. 2. Divide the EA into C sectors (strata) of equal area and shape such that each sector contains the same number of sample points. The number of sectors (C) should be equal to the number of specimens in each composite (since one specimen per area will be used in each composite) and the number of points within each sector, N, should equal the desired number of composite samples. 3. Use a random number table or random number generator to establish a set of labels for theN points within each sector. This is done by first labeling the points in a sector in an arbitrary fashion (say, points A, B, C,...) and associating the first random number with point A, the second with point B, etc. Then rank the points in the sector according to the set of random numbers and relabel each point with its rank. Repeat this process for each sector. 4. Form composite number 1 by compositing specimens with the '1' label, form composite number 2 by compositing specimens with the *2' label, etc. This leads to N composite samples that are subjected to chemical analysis. •3 «2 • 1 *>4 •5 t)6 •6 •S • 1 «4 •3 «2 I] 05 •6 t)3 •2 *4 •4 t)6 •3 «5 •2 •! Figure 7. Systematic (square grid points) sample wtth random compositing scheme (6 composite samples consisting of 4 specimens). 93 TUT OOS 1230 Highlight S: Procedure for Compositing of Spocinwns from m Stntlfi0d Random Samp/* Using a Random Schame (Figure 8) 1. Divide the EA into C sectors (strata) of equal area, where C is equal to the number of specimens to be in each composite (since one specimen per stratum will be used in each composite). 2. Within each stratum, choose N random locations, where N is the desired number of composites. Step-by-step procedures for choosing random locations can be found in Chapter 5 of U.S. EPA(1989a). 3. Use a random number table or random number generator to establish a set of labels for the N points within each sector. This is done by first labeling the points in a sector in an arbitrary fashion (say, points A,TJ, C,...) and associating the first random number with point A, the second with point B, etc. Then rank the points in the sector according to the set of random numbers and relabel each point with its rank. Repeat this process for each sector. 4. Form composite number 1 by compositing specimens with the 'V label, form composite number 2 by compositing specimens with the '2' label, etc. This leads to N composite samples that are subjected to chemical analysis. •1 •2 t>4 •5 •6 •4 96 •3 • 1 «5 •2 •6 «3 •2 •4 •! •5 •5 •4 •3 •6 • 1 Figure 8. Stratified random sample with random compositing scheme (6 composite samples consisting of 4 specimens). ; 94 TUT COS 1231 Table 23. Probability of Decision Error at 0.5 SSL and 2 SSL Using Max Test Sample Size* 4 5 6 7 8 9 4 5 8 7 8 9 CV=1.0« EO.S" Ea.o" CVsl.S EOS ^2.0 CV=2.0 EO.S E*.0 CV=2.5 Eo.i E*.0 CV=3.0 EO.S E*.0 CV=3.5 EO.S E2.o CV=4.0 EO.S E2.o C = 4 specimens per composite' <.01 <.01 <.01 <.01 <.01 <.01 0.08 0.05 0.03 0.01 0.01 0.01 0.02 0.02 0.02 0.03 0.03 0.05 0.11 0.06 0.04 0.02 0.01 0.01 0.09 0.11 Q.11 0.12 0.16 0.16 0.13 0.10 0.06 0.04 0.02 0.01 0.14 0.15 0.21 0.25 0.25 0.28 0.19 0.10 0.08 0.05 0.04 0.03 0.19 0.26 0.28 0.31 0.36 0.36 0.20 0.17 0.11 0.08 0.05 0.04 0.24 0.26 0.31 0.38 0.42 0.44 0.26 0.18 0.11 0.09 0.07 0.07 0.25 0.31 0.35 0.41 0.41 0.48 0.30 0.25 0.16 0.15 o.or 0.08 C « 8 specimens per composite <.01 <.01 <.01 <.01 <.01 <.01 0.08 0.05 0.03 0.01 0.01 0.01 <.01 <.01 0.01 0.01 0.01 0.01 0.11 0.06 0.04 0.02 0.01 0.01 0.03 0.04 0.06 0.06 0.06 0.06 0.12 0.09 0.04 0.02 0.02 0.01 0.08 0.11 0.14 0.14 0.15 0.18 0.16 0.09 0.08 0.04 0.02 0.02 0.15 0.17 0.19 0.23 0.25 0.28 0.17 0.13 0.09 0.06 0.03 0.03 0.29 0.22 0.25 0.29 0.30 0.34 0.20 0.15 0.09 0.08 0.04 0.03 0.23 0.25 0.29 0.37 0.40 0.39 0.27 0.20 0.12 0.08 0.06 0.04 Ift H C O O 03 • The CV is the coefficient of variation for Individual, uncomposfted measurements across the entire EA, including measurement error.. b Sample size (N) 3 number of composite samples. c EO 5 s Probability of requiring further Investigation when the EA mean Is 0.5 SSL d Ego * Probability of not requiring further Investigation when the EA mean Is 2.0 SSL. • C * number of specimens per composite sample, where each composite consists of points from a stratified random or systematic grid sample from across the entire EA. . NOTE: AH decision error rates are based on 1,000 simulations that assume that each composite Is representative of the entire EA, that half the EA has concentrations below the quantttatlon limit (I.e., SSL/100), and half the EA has concentrations that follow a gamma distribution (a conservative distributional assumption). The error rates shown in Table 23 are based on the simulations presented in Appendix 1. These simulations are based on the following assumptions 1. Each of the N composite samples is based on C specimens selected to be representative of the EA as a whole, as specified above (C = number of sectors or strata). 2. One-half the EA has concentrations below the quantisation limit (which is assumed to be SSL/100). 3. One-half the EA has concentrations that follow a gamma distribution (see Section 4.3 for additional discussion). 4. Each chemical analysis is subject to a 20 percent measurement error. The error rates presented in Table 23 are based on the above assumptions which make them robust for most potential distributions of soil contaminant concentrations. Distribution assumptions 2 and 3 were used because they were found in the simulations to produce high error rates relative to other potential contaminant distributions (see Section 4.3). If the proportion of the site below the quantitation limit (QL) is less than half or if the distribution of the concentration measurements is some other distribution skewed to the right (e.g., Icgnormal), rather man gamma, men the error rates achieved are likely to be no worse than those cited in Table 23. Although the actual contaminant distribution may be different from those cited above as the basis for Table 23, only extensive investigations will usually generate sufficient data to determine the actual distribution for each EA. Using Table 23 to determine the sample size pair (C and N) needed to achieve satisfactory error rates with the Max test requires an a priori estimate of the coefficient of variation for measurements of the contaminant of interest across the EA. The coefficient of variation (CV) is the ratio of the standard deviation of contaminant concentrations for individual, uncomposited specimens divided by the EA mean concentration. As discussed in Section 4.1.4, the EAs should be constructed within strata expected to have relatively homogeneous concentrations so that an estimate of the CV for a stratum may be applicable for all EAs in that stratum. The site manager should use a conservatively large estimate of the CV for determining sample size requirements because additional sampling will be needed if the data suggest that the true CV is greater than that used to determine the sample sizes. Potential sources of information for estimating the EA or stratum means, variances, and CVs include the following (in descending order of desirability): Data from a pilot study conducted at the site • Prior sampling data from the site Data from similar sites Professional judgment. • • For more information on estimating variability, see Section 6.3.1 of U.S. EPA (1989a). 4.1.8 Using the DQA Process: Analyzing Max Test Data. This section provides guidance for analyzing the data for the Max test. The hypothesis test for the Max test is very simple to implement, which is one reason that the Max test is attractive as a surface soil screening test. If Xj, x2, ..., XN represent concentration measurements for N composite samples that each consist of C specimens selected so that each 96 TUT O08 1233 composite is representative of the EA as a whole (as described in Section 4.1.7), the Max test is implemented as follows: If Max (x,, x2. .... XN) * 2 SSL, then investigate the EA further; If Max (x,. x2, .... XN) < 2 SSL. and the data quality assessment (DQA) indicates that the sample size was adequate, then no further investigation is necessary. In addition, the step-by-step procedures presented in Highlight 6 must be implemented to ensure that the site-specific error Limits, as discussed in Section 4.1.6, are achieved. If the EA mean is below 2 SSL, the DQA process may be used to determine if the sample size was sufficiently large to justify the decision to not investigate further. To use Table 23 to check whether the sample size is adequate, an estimate of the CV is needed for each EA'. The first four steps of Highlight 6, the DQA process for the Max test, present a process for the computation of a sample CV for an EA based on the N composite samples that each consist of C specimens. However, the sample CV can be quite large when all the measurements are very small (e.g.. well below the SSL) because CV approaches infinity as the EA sample mean (x) approaches zero. Thus, when the composite concentration values for an EA are all near zero, the sample CV may be questionable and therefore unreliable for determining if the original sample size was sufficient (i.e., it could lead to further sampling when the EA mean is well below 2 SSL). To protect against unnecessary additional sampling in such cases, compare all composites agrinst the equation given in Step 5 of Highlight 6. If the maximum composite sample concentration is below the value given by the equation, then the sample size may be assumed to be adequate and no further DQA is necessary. To develop Step 5, EPA decided that if there were no compositing (C=l) and all the observations (based on a sample size appropriate for a CV of 2.5) were less than the SSL, then one can reasonably assume that the EA mean was not greater than 2 SSL. Likewise, because the standard error for the mean of C specimens, as represented by the composite sample, is proportional to l//cf, the comparable condition for composite observations is that one can reasonably assume that the EA mean was not greater than 2 SSL when all composite observations were less than SSL/fc . If this is the case for an EA sample set, the sample size can be assumed to be adequate and no further DQA is needed Otherwise (when at lease one composite observation is not this small), use Table 23 with the sample CV for the EA to determine whether a sufficient number of samples were taken to achieve DQOs * In addition to being simple to implement, the Max test is recommended because it provides good control over the Type I error rates at 2 SSL with small sample'sizes. It also does not need any assumptions regarding observations below the QL. Moreover, the Max test error rates at 2 SSL are fairly robust against alternative assumptions regarding the distribution of surface soil concentrations in the EA. The simulations in Appendix I show that these error rates are rather stable for lognormal or Weibull contaminant concentration distributions and for different assumptions about portions of the site with contaminant concentrations below the QL. 97 TUT 008 1234 Highlight 6: Directions for Data Quality Assessment for the Max Test Let x-i, x2,..., XN represent contaminant concentration measurements for N composite samples tha each consist of C specimens selected so that each composite is representative of the EA as a whoie. The following describes the steps required to ensure that the Max test achieves the DQOs established for the site. STEP 1: The site manager determines the Type I error rate to be achieved at 2 SSL and the Type II error rate to be achieved at 0.5 SSL, as described in Section 4.1.6. STEP 2: Calculate the sample mean x- \^ ~,\ N STEP 3: Calculate the sample standard deviation s SB STEP 4: Calculate the sample estimate of the coefficient of variation, CV, for individual concentration measurements from across the EA. NOTE: This is a conservation approximation of the CV for individual measurements. SSL STEP 5. If Max (x , x_...., XN) < .— , then no further data quality assessment is needed and the EA vc needs no further investigation. Otherwise proceed to Step 6. STEP 6: Use the value of the sample*CV calculated in Step 4 as the true CV of concentrations to determine which column of Table 23 is applicable for determining sample size requirements. Using the error limits established in Step 1, determine the sample size requirements from this table. If the required sample size is greater than that implemented, further investigation of the EA is necessary. The further investigation may consist of selecting a supplemental sample and repeating the Max test with the larger, combined sample. A limitation of the Max test is that it does not provide as good control over the Type II error rates at 0.5 SSL as it does for Type I error rates at 2 SSL. In fact, for a fixed number, C, of specimens per composite, the Type n error rate increases as the number of composite samples, N, increases. As the sample size increases, the likelihood of observing an unusual sample with the maximum exceeding 2 SSL increases. However, the Type II error rate can be decreased by increasing the number of 98 TUT OOS 1235 specimens per composite. This unusual performance of the Max test as a hypothesis testing procedure occurs because the rejection region is fixed below 2 SSL and thus does not depend on the sample size (as it does for typical hypothesis testing procedures). 4.1.9 Specify Limits on Decision Errors for Chen Test. Although the Max test is adequate and appropriate for selecting a sample size for site screening, there are other alternate methods of screening surface soils. One such alternate method is the Chen test. In general, the Chen test differs from the Max test in its basic assumption about site contamination and the purpose of soil sampling. Because of this variation, these two methods have different null hypotheses and different decision error types. There are two formulations of the statistical hypothesis test concerning the true (but unknown) mean contaminant concentration, |i, that achieve the Soil Screening Guidance decision error rate goals specified in Section 4.1.6. They are: . 1. Test the null hypothesis, H0: u £ 2 SSL, versus the alternative hypothesis, H]: n < 2 SSL, at the 5 percent significance level using a sample size chosen to achieve a Type II error rate of 20 percent at 0.5 SSL. 2. Test the null hypothesis, HO: \i £ 0.5 SSL, versus the alternative hypothesis, HI: \i > 0.5 SSL, at the 20 percent significance level using a sample size chosen to achieve a Type D error rate of 5 percent at 2 SSL. The first formulation of the problem (which is commonly used in the Superfund program) has the advantage that the error rate that has potential public health consequences is controlled directly via the significance level of the test. The error rate that has primarily cost consequences can be reduced ^, mcreasiijg the sample size above the minimum requirement. However, EPA has identified a new test procedure, the Chen test (Chen, 1995), which requires the second formulation but is less sensitive to assumptions regarding the distribution of the contaminant measurements than the Land procedure used in the December 1994 draft Technical Background Document (see Section 4.3). This section provides guidance regarding application of the Chen test and is, therefore, based on the second formulation of the hypothesis test. A disadvantage of the second formulation is its performance when the true EA mean is between 0.5 SSL and the SSL. In this case, as the sample size increases, the test indicates the decision to investigate further, even though the mean is less than the SSL. In fact, no test procedure with feasible sample sizes performs well when the true EA mean is in the "gray region" between 0.5 SSL and 2 SSL (see Section 4.3). Whenever large sample sizes are feasible, one should modify the problem statement and test the null hypothesis, HO: u £ SSL, instead of HO: U £ 0.5 SSL. One would then develop appropriate DQOs for this modified hypothesis test (e.g., significance level of 20 percent at the SSL and 5 percent probability of decision error at 2 SSL). When the true mean of an EA is compared with the screening level, there are two possible decision errors that may occur: (1) decide not to investigate an EA further (i.e., "walk away") when the correct decision would be to "investigate further"; and (2) decide to investigate further when the correct decision would be to "walk away." For the Chen test, the "incorrectly walk away" decision error is designated as the Type n decision error because it occurs when we incorrectly accept the null hypothesis. Correspondingly, the "unnecessarily investigate further" decision error is designated as the Type I decision error because it occurs when we incorrectly' reject the null hypothesis. 99 TUT COS 1236 As discussed in Section 4.1.6, the Soil Screening Guidance specifies a default gray region for decision errors from 0.5 SSL to 2 SSL and sets the following goals for Type I and Type II error rates: Prob ("investigate further" when the true EA mean is 0.5 SSL) = 0.20 Prob ("walk away" when the true EA mean is 2 SSL) = 0.05. Table 24 summarizes this step of the DQO process for the Chen test, specifying limits on the decision error rates, and the final step of the DQO process, optimizing the design. 4.1.10 Optimize the Design Using the Chen Test. This section includes guidance on developing an optimum sampling strategy for screening surface soils. It discusses compositing, the selection of sampling points for composited and uncomposited surface soil sampling, and the recommended procedures for determining the sample sizes necessary to achieve specified limits on decision errors using the Chen test. Note that the size, shape, and orientation of sampling volume (i.e., "support") for heterogenous media have a significant effect on reported measurement values. For instance, particle size has a varying affect on the transport and fate of contaminants in the environment and on the potential receptors. Because comparison of data from methods that are based on different supports can be difficult, defining the sampling support is important in the early stages of site characterization. This may be accomplished through the DQO process with existing knowledge of the site, contamination, and identification of the exposure pathways that need to be characterized. Refer to Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies (U.S. EPA, 1992f) for more information about soil sampling support. The SAP developed for surface soils should specify sampling and analytical procedures as well as the development of QA/QC procedures. To identify the appropriate analytical procedures, the screening levels must be known. If data are not available to calculate site-specific SSLs, then the generic SSLs in Appendix A should be used. Compositing. Because the objective of surface soil screening is to ensure that the mean contaminant concentration does not exceed the screening level, the physical "averaging" that occurs during compositing is consistent with the intended use of the data. Compositing allows a larger number of locations to be sampled while controlling analytical costs because several discrete samples are physically mixed (homogenized) and one or more subsamples are drawn from the mixture and submitted for analysis. If the individual samples in each composite are taken across the EA, each composite represents an estimate of the EA mean. A practical constraint to compositing in some situations is the heterogeneity of the soil matrix. The efficiency and effectiveness of the mixing process may be hindered when soil particle sizes vary widely or when the soil matrix contains foreign objects, organic matter, viscous fluids, or sticky material. Soil samples should not be composited if matrix interference among contaminants is likely (e.g., when the presence of one contaminant biases analytical results for another). 100 TUT OO8 1237 Table 24. Sampling Soil Screening DQOs for Surface Soils under Chen Test POO Process Steps Soil Screening Inputs/Outputs Specify Limits on Decision Errors Define baseline condition (null hypothesis) Define gray region Define Type I and Type II decision errors Identify consequences Assign acceptable probabilities of Type I and Type II decision errors EA needs no further investigation From 0.5 SSL to 2 SSL Type I error Investigate further when an EA's true mean concentration is below 0.5 SSL Type II error Do notmvestigate further ("walk away from*) when an EA true mean concentration is above 2 SSL Type I error unnecessary expenditure of resources to investigate further Type II error, potential public health consequences Goals: Type 1:0.20 (20%) probability of investigating further when EA mean is 0.5 SSL Type II: 0.05 (5%) probability of not investigating further when EA mean is 2 SSL Optimize the Design Determine expected variability of EA surface soil contaminant concentrations Design sampling strategy by evaluating costs and performance of alternatives A conservatively large expected coefficient of variation (CV) from prior data for the site, field measurements, or data from other comparable sites and expert judgment Lowest cost sampling design option (i.e., compositing scheme and number of composites) that will achieve acceptable decision error rates Develop planning documents for the field investigation Sampling and Analysis Ran (SAP) Quality Assurance Project Plan (QAPjP) Before individual specimens are composited for chemical analysis, the site manager should consider homogenizing and splitting each specimen. By compositing one portion of each specimen with the other specimens and storing one portion for potential future analysis, the spatial integrity of each specimen is maintained. If the concentration in a composite is high, the splits of the individual specimens of which it was composed can be analyzed subsequently to determine which individual specimen(s) have high concentrations. This will permit the site manager to determine which portion within an EA is contaminated without making a repeat visit to the site. Sample Pattern. The Chen test can be applied using composite samples that are representative of the entire EA or with individual uncomposited samples. Systematic grid sampling (SyGS) generally is preferred because it ensures that the sample points will be dispersed across the entire EA. However, if die boundaries of the EA are irregular (e.g., around the perimeter of the site or the boundaries of a stratum within which the EAs were defined), the number of grid sample points that fall within the EA depends on the random starting point selected. Therefore, for these irregularly shaped EAs, SRS or stratified SRS is recommended. Moreover, if a systematic trend of contamination is suspected across the EA (e.g., a strip of higher contamination), 101 TUT OO8 1238 then SRS or stratified SRS is recommended again. In this case, grid sampling would be likely to result in either over- or under representation of the strip of higher contaminant levels, depending on the random starting point. For composite sampling, the sampling pattern used to locate the C discrete sample specimens that form each composite sample is important. The composite samples must be formed in a manner that is consistent with the assumptions underlying the sample size calculations. In particular, each composite sample must provide an unbiased estimate of the mean contaminant concentration over the entire EA. One way to construct a valid composite of C specimens is to divide the EA into C sectors, or strata, of equal area and select one point at random from each sector. If sectors (strata) are of unequal sizes, the simple average is no longer representative of the EA as a whole. Valid sampling patterns and compositing schemes for selecting N composite samples mat each consist of C specimens include the following: 1. Select an .SRS consisting of C points and composite all specimens associated with these points into a sample. Repeat this process N times, discarding any points that were used in a previous sample. 2. Select an SyGS/rs of C points and composite all specimens associated with the points in this sample. Repeat this process N times, using a new randomly selected starting point each time. 3. Select a single SyGS/rs of CN points and use the systematic compositing scheme that is described in Highlight 3 to form N composites, as illustrated in Figure 6. 4. Select a single SyGS/rs of CxN points and use the random compositing scheme that is described in Highlight 4 to form N composites, as illustrated in Figure 7. 5. Select a stratified random sample of CxN points and use a random compositing scheme, as described in Highlight 5, to form N composites, as illustrated in Figure 8. Methods 1, 2, and 5 are the most statistically defensible, with method 5 used as the default method in the Soil Screening Guidance. However, given the practical limits of implementing these methods, either method 3 or 4 is generally recommended for EAs with regular boundaries (e.g., square or rectangular). As noted above, if the boundaries of the EA are irregular, SyGS/rs sampling may not result in exactly CxN sample points. Therefore, for EAs with irregular boundaries, method 5 is recommended. Alternatively, a combination of methods 4 and 5 can be used for EAs that can be partitioned into C sectors of equal area of which K have regular boundaries and the remaining C - K have irregular boundaries. Sample Size. This section provides procedures to.determine sample size requirements for the Chen test that achieve the site-specific decision error limits discussed in Section 4.1.6. The Chen test is an upper-tail test for the mean of positively skewed distributions, like the lognormal (Chen. 1995). It is based on the mean concentration observed in a simple random sample, or equivalent design, selected from a distribution with a long right-hand tail. The Chen procedure is a hypothesis testing procedure that is robust among the family of right- skewed distributions (see Section 4.3). That is, decision error rates for a given sample size are relatively insensitive to the particular right-skewed distribution that generated the data. This 102 TUT 008 1239 robustness is important in the context of surface soil screening because the number of surface soil ,„.-, samples will usually not be sufficient to determine the distribution of the concentration measurements. The procedures presented above for selecting composited or uncomposited simple random or systematic grid samples can all be used to generate samples for application of the Chen test. The Chen procedure is based on a simple random sample, or one that can be analyzed as if it were an SRS. Directions for performing the Chen test in a manner mat is consistent with the DQOs that have been established for a site are presented later. Tables 25 through 30 provide the sample sizes required for the den test performed at the 10, 20, or 40 percent levels of significance (probability of Type I error at 0.5 SSL) and achieve, at most, a 5 or 10 percent probability of (Type II) error at 2 SSL. The Type II error rates at 2 SSL are based on the simulations presented in Appendix I. These simulations are based on the following assumptions: 1. Each of the N composite samples is based on C specimens selected to be representative of the EA as a whole, as specified above. 2. One-half the EA has concentrations below the quantitation limit (which is assumed to be SSL/100). 3. One-half the EA has concentrations ihat follow a gamma distribution. 4. Measurements below the QL are replaced by 0.5 QL for computation of the Chen test statistic. 5. Each chemical analysis is subject to a 20 percent measurement error. Distributional assumptions 2 and 3 were used as the basis for the Type II error rates at 2 SSL (shown in Tables 25 through 30) because they were found in the simulations to produce high error rates relative to other potential contaminant distributions. If the proportion of the site below the QL is less than half or if the distribution of the concentration measurements is some other right-skewed distribution (e.g., lognormal), rather man gamma, then the Type II error rates achieved are likely to be no worse than those cited in Tables 25 through 30. No sample sizes, N, less than four are shown in these tables (irrespective of the number of specimens per composite) because consideration of the simulation results presented in Section 4.3 has led to a program-level decision that at least four separate analyses are required to adequately characterize the mean of an EA. No sample sizes in excess of nine are presented because of a program-level decision that more than nine samples per exposure area is generally unlikely for screening surface soils at Superfund sites. However, additional sample size options can be determined from the simulations reported in Appendix I. When using Tables 25 through 30 to determine the sample size pair (C and N) needed to achieve satisfactory error rates with the Chen test, investigators must have an a priori estimate of the CV for measurements of the contaminant of interest across the EA. As previously discussed for the Max test, the site manager should use a conservatively large estimate of the CV for determining sample size requirements because additional sampling will be required if the data suggest that the true CV is greater than that used to determine the sample sizes. 103 TUT OOS 124O Table 25. Minimum Sample Size for Chen Test at 10 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA Number of specimens per composite** 2 3 4 5 6 Coefficient of variation (CV)- 1.0 7 5 4 4 4 1.5 9 7 6 5 4 2.0 >9 9 8 6 5 2.5 >9 >9 >9 8 . 7 3.0 >9 >9 >9 >9 9 •The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA and includes measurement error. bEach composite consists of points from a stratified random or systematic grid sample across the entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume that each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, and that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). Table 26. Minimum Sample Size for Chen Test at 20 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA Number of specimens per composite1" 1 2 3 4 5 6 Coefficient of 1.0 9 5 4 4 4 4 1.5 >9 7 5 4 4 4 2.0 >9 >9 7 6 4 4 variation (CV)* 2.5 >9 >9 9 7 6 5 3.0 >9- >9 >9 >9 8 8 3.;' >9 >9 >9 >9 >9 9 •The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA and includes measurement error. *>Each composite consists of points from a stratified random or systematic grid sample across the entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume that each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, and that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). 104 Table 27. Minimum Sample Size for Chen Test at 40 Percent Level of Significance to Achieve a 5 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA Number of specimens per composite** 1 2 3 4 5 6 Coefficient of variation (CV)- 1.0 5 4 4 ' 4 4 4 1.5 9 4 4 4 4 4 2.0 >9 8 5 4 4 4 2.6 >9 9 7 5 5 4 3.0 >9 >9 >9 8 6 5 3.5 >9 >9 >9 >9 9 8 4.0 >9 >9 >9 >9 >9 9 •The CV is the coefficient of variation tor indnridual, uncomposited measurements across the entire EA and includes measurement error. b£ach composite consists of points from a stratified random or systematic grid sample across trie entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume thai each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, and that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). Table 28. Minimum Sample Size for Chen Test at 10 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given the Expected CV for Concentrations Across the EA Number of specimens per composite*1 2 3 4 5 6 Coefficient of 1.0 6 4 4 4 4 1.5 7 5 4 4 4 2.0 >9 7 6 5 4 variation (CV)- 2.5 >9 >9 7 6 5 3.0 >9 >9 >9 8 7 3.5 >9 >9 >9 >9 9 •The CV is the coefficient of variation for indmdual, uncomposited measurements across the entire EA and includes measurement error. *>Each composite consists of points from a* stratified random or systematic grid sample across the entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume that each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, and that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). 105 TUT —— 1242 Table 29. Minimum Sample Size for Chen Test at 20 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA Number of specimens per composite^ 1 2 3 4 5 6 Coefficient of variation (CV)» 1.0 7 4 4 4 -, 4 4 1.5 9 5 4 4 4 4 2.0 >9 8 5 4 4 4 2.5 >9 >9 8 5 5 4 3.0 >9 >9 >9 8 6 5 3.5 >9 >9 >9 >9 8 7 4.0 >9 >9 >9 >9 >9 9 •The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA and includes measurement error. bEach i - - uosKe consists of points from a stratified random or systematic grid sample across the entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume that each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, and that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). Table 30. Minimum Sample Size for Chen Test at 40 Percent Level of Significance to Achieve a 10 Percent Chance of "Walking Away" When EA Mean is 2.0 SSL, Given Expected CV for Concentrations Across the EA * Number of ___________Coefficient of variation (CV)*___________ per^osi'teb 1-0 1.5 2.0 2.5 3.0 3.5 4,0 1 2 3 4 5 6 4 4 4 4 4 4 7 4 4 4 4 4 9 5 4 4 - 4 4 >9 8 5 4 4 4 >9 9 7 5 5 4 >9 >9 9 7 6 5 >9 >9 >9 >9 8 6 •The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA and includes measurement error. bEach composite consists of points from a stratified random or systematic grid sample across the entire EA. NOTE: Sample sizes are based on 1,000 simulations that assume that each composite is representative of the entire EA, that half the EA has concentrations below the limit of detection, tfid that half the EA has concentrations following a gamma distribution (a conservative distributional assumption). 106 TUT 008 1243 Given an a priori estimate of the CV of concentration measurements in the EA, the site manager can use Table 26 to determine a sample size option mat achieves the decision error goals for surface soil screening presented in Section 4.1.6 (i.e., not more than 20 percent chance of error at 0.5 SSL and not more than 5 percent at 2 SSL). For example, suppose that the she manager expects that the maximum true CV for concentration measurements in an EA is 2. Then Table 26 shows that six composite samples, each consisting of four specimens, will be sufficient to achieve the decision error limit goals. 4.1.11 Using the DQA Process: Analyzing Chen Test Data. Step-by-step instructions for using the Chen test to analyze data from both discrete random samples and pseudo- random samples (e.g., composite samples constructed as described previously) are provided in Highlight 7. This method for analyzing the data is a robust procedure for an upper-tailed test for the mean of a positively skewed distribution. As explained by Chen (1995), this procedure is a robust generalization of the familiar Student's t-test; it further generalizes a method developed by Johnson (1978) for asymmetric distributions. The only assumption necessary for valid application of the Chen procedure is that the sample be a random sample from a right-skewed distribution. This robustness within the broad family of right- skewed distributions is appropriate for screening surface soil because the distribution of concentrations within an EA may depart from the common assumption of lognormality. Computation of the Chen test statistic, as shown in Highlight 7, requires mat concentration values be available for all N individual or composite samples analyzed for the contaminant of interest. If an analytical test result is reported below the quantitation limit, it should be used in the computations. For results below detection, substitute one-half the QL. A disadvantage of the Chen procedure is that the hypothesis, "the EA needs no further investigation," must be treated as the alternative hypothesis, rather than as the null hypothesis. As a result, the Type I error rate at 0.5 SSL is controlled via the significance level of the test, rather than the error rate at 2 SSL, which may have public health consequences. Hence, if the sample sizes (C and N) are based on an assumed CV that is too small, the desired error rate at 2 SSL is likely not to be achieved. Therefore, it is important to perform the data quality assurance check specified in Steps 6 through 8 of Highlight 7 to ensure that the desired error rate at 2 SSL is achieved. Moreover, it is important that the site manager base the initial EA sample sizes on a conservatively large estimate of the CV so that this process will not result in the need for additional sampling. 4.1.12 Special Considerations for Multiple Contaminants, if the surface soil samples collected for an EA will be tested for multiple contaminants, be aware mat the expected CVs for the different contaminants may not all be identical. A conservative approach is to base the sample sizes for all contaminants on the largest expected CV. 4.1.13 Quality Assurance/Quality Control Requirements. Regardless of the sampling approach used, the Superfund quality assurance program guidance must be followed to ensure that measurement error rates are documented and within acceptable limits (U.S. EPA, 1993d). 107 TUT ooe 1244 Highlight 7: Directions for the Chin Test Using Simple Random Sample Scheme Let Xi, X2,..., XN, represent concentration measurements for N random sampling points or N pseudo- random sampling points (i.e., from a design that can be analyzed as if ft were a simple random sample). The following describes the steps for a one-sample test for H0: \i £ 0.5 SSL at the 100a% significance level that is designed to achieve a 10013% chance of incorrectly accepting HO when u * 2 SSL FN 1 1 STEP1: Calculate the sample mean x = £ x,| TT h.i I N - STEP 2: Calculate the sample standard deviation S = l"~~ STEP 3: Calculate the sample skewness N ?.("'-")' b = N (N-l) (N-2)s3 STEP. 4: Calculate the Chen test statistic, t2, as follows: b _ x -0.5 SSL l~ s / t2 = t+a(l + 2t2) +4a2 (t-f2t3) STEP 5: Compare t2 to Za, the 100(1 - a) percerrtite of the standard normal probability distribution. If t2 > za, the null hypothesis is rejected, and the EA needs further investigation. If \2 £ za. there is insufficient evidence to reject the null hypothesis. Proceed to Step 6 to determine if the sample size is sufficient to achieve a 1 006% or less chance of incorrectly accepting the H0 when |i = 2 SSL. 108 TUT ,.24S Highlight 7: Directions for the Chert Test Using Simple Random Sample Scheme (continued) STEP 6: Let C represent the number of specimens composited to form each of the N samples, where each of x-i, X2,.», XN is a composite sample consisting of C specimens selected so that each composite is representative of the EA as a whole. (If each of XL x2,..., XN is an individual random or pseudo-random sampling point, then C «1.) SSL If Max (xr x2,..., XN) < ..... , then no further data quality assessment is needed and the EA needs no further investigation. Otherwise proceed to Step 7. STEP 7: Calculate the sample estimate of the coefficient of variation, CV, for individual concentration measurements from across the EA. CV = NOTE: This calculation ignores measurement error, which results in conservatively large sample size requirements. STEP 8: Use the value of the sample CV calculated in Step 7 as the true CV of concentrations in Tables 25 through 30 to determine the minimum sample size, N', necessary to achieve a 10013% or less chance of incorrectly accepting H0 when u = 2 SSL. tf N > N", the EA needs no further investigation. If N < N*, further investigation of the EA is necessary. The further investigation may consist of selecting a supplemental sample and repeating this hypothesis testing procedure with the larger, combined sample. 4.1.14 Final Analysis. After either the Max test or the Chen test has been performed foi each EA of interest (0.5 acre or less) at an NPL site, the pattern of decisions for individual EAs (to "walk away" or to "investigate further") should be examined. If some EAs for which the decision wao to "walk away" are surrounded by EAs for which the decision was to "investigate further," it may be more efficient to identify an area including all these EAs for further study and develop a global investigation strategy. 4.1.15 Reporting. The decision process for surface soil screening should be thoroughly documented as part of the RI/FS process. This documentation should include a map of the site 109 TUT COS 1246 (showing the boundaries of the EAs and the sectors, or strata, within EAs that were used to select sampling points within the EAs); documentation of how composite samples were formed and the number of composite samples that were analyzed for each EA; the raw analytical data: the results of all hypothesis tests; and the results of all QA/QC analyses. 4.2 Sampling Subsurface Soils Subsurface soil sampling is conducted to estimate the mean concentrations of contaminants in each source at a site for comparison to inhalation and migration to ground water SSLs. Measurements of soil properties and estimates of the area and depth of contamination in each source are also needed to calculate SSLs for these pathways. Table 31 shows the steps in the DQO process necessary to develop a sampling strategy to meet these objectives. Each of these steps is described below. 4.2.1 State the Problem. Contaminants present in subsurface soils at the site may pose significant risk to human health and the environment through the inhalation of volatiles or b 'he migration of contaminants through soils to an underlying potable aquifer. The problem is to idc -fy the contaminants and source areas that do not pose significant risk to human health through t uier of these exposure pathways so that future investigations may be focused on areas and contaminants of true concern. Site-specific activities in this step include identifying the data collection planning team (including technical experts and key stakeholders) and specifying the available resources (i.e., the cost and time available for sampling). The list of technical experts and stakeholders should contain all key personnel who are involved with applying SSLs to the she. Other activities include developing the conceptual site model and identifying exposure scenarios, which are fully addressed in the Soil Screening Guidance: User's Guide (U.S. EPA, 1996). 4.2.2 Identify the Decision. The decision is to determine whether mean soil concentrations in each source area exceed inhalation or migration to ground water SSLs for specific contaminants. If so, the source area will be investigated further. If not, no further action will be taken under CERCLA. 4.2.3 Identify Inputs to the Decision. Site-specific inputs to the decision include the average contaminant concentrations within each source area and the inhalation and migration ground water SSLs. Calculation of the SSLs for the two pathways of concern also requires site-specific measurements of soil properties (i.e., bulk density, fraction organic carbon content. pH, and soil texture class) and estimates of the area! extent and depth of contamination. A list of feasible sampling and analytical methods should be assembled during this step EPA recommends the use of field methods where applicable and appropriate. Verify that Contract Laboratory Program (CLP) methods and field methods for analyzing the samples exist and that the analytical method detection limits or field method detection limits are appropriate for the site- specific or generic SSL. The Sampler's Guide to the Contract Laboratory Program (U.S. EPA, 1990) and the User's Guide to the Contract Laboratory Program (U.S. EPA, 1991d) contain further information on CLP methods. 110 TUT OO8 Table 31. Soil Screening DQOs for Subsurface Soils DQO Process Steps Soil Screening Inputs/Outputs State the Problem Identify scoping team Develop conceptual site model (CSM) •Define exposure scenarios Specify available resources Write brief summary of contamination problem Site manager and technical experts (e.g., toxtcologists, risk assessors, hydrogeoiogists, statisticians). CSM development (described in Step 1 of the User's Guide, U.S. ERA, 1996). Inhalation of volatiles and migration of contaminants from soil to potable ground water (and plant uptake for certain contaminants). Sampling and analysis budget, scheduling constraints, and available personnel. Summary of the subsurface soil contamination problem to be investigated at the site. Identify the Decision Identify decision Identify alternative actions Do mean soil concentrations for particular contaminants (e.g., contaminants of potential concern) exceed appropriate SSLs? Eliminate area from further action or study under CERCLA or Plan and conduct further investigation. Identify Inputs to the Decision Identify decision Define basis for screening Identify analytical methods Volatile inhalation and migration to ground water SSLs for specified contaminants Measurements of subsurface soil contaminant concentration Soil Screening Guidance Feasible analytical methods (both field and laboratory) consistent with program-level requirements. Specify the Study Boundaries Define geographic areas of field investigation Define population of interest Define scale of decision making Subdivide site into decision units Define temporal boundaries of study Identify (list) practical constraints The entire NPL site (which may include areas beyond facility boundaries), except for any areas with clear evidence that no contamination has occurred. Subsurface soils Sources (areas of contiguous soil contamination, defined by the area and depth of contamination or to the water table, whichever is more shallow). Individual sources delineated (area and depth) using existing information or field measurements (several nearby sources may be combined into a single source). Temporal constraints on scheduling field visits. Potential impediments to sample collection, such as access, health, and safety issues. Develop a Decision Rule Specify parameter of interest Specify screening level Specify "if..., then..." decision rule Mean soil contaminant concentration in a source (as represented by discrete contaminant concentrations averaged within soil borings). SSLs calculated using available parameters and site data (or generic SSLs if site data are unavailable). If the mean soil concentration exceeds the SSL, then investigate the source further. If the mean soil boring concentration is less than the SSL, then no further investigation is required under CERCLA. Ill TUT OO8 1248 Table 31. (continued) Specify Limits on Decision Errors Define QA/QC goals CLP precision and bias requirements 10% CLP analyses for field methods Optimize the Design Determine how to estimate mean For each source, the highest mean soil core concentration (i.c., depth- concentration in a source weighted average of discrete contaminant concentrations within a boring). Define subsurface sampling strategy by Number of soil borings per source area; number of sampling intervals with evaluating costs and site-specific depth, conditions Develop planning documents for the field Sampling and Analysis Plan (SAP) investigation Quality Assurance Project Plan (QAPjP) Field methods will be useful in defining the study boundaries (i.e., area and depth of contamination) during site reconnaissance and during the sampling effort. For example, soil gas survey is an ideal method for determining the extent of volatile contamination in the subsurface. EPA expects field methods will become more prevalent and useful because the design and capabilities of field portable instrumentation are rapidly evolving. Documents on standard operating procedures (SOPs) for field methods are available through NTIS and should be referenced in soil screening documentation if these methods are used. Soil parameters necessary for SSL calculation are soil texture, bulk density, and soil organic carbon. Some of these parameters can be measured in the field, others require laboratory measurement. Although laboratory measurements of these parameters cannot be obtained under the Superfund Contract Laboratory Program, they are readily available from soil testing laboratories across the country. Note that the size, shape, and orientation of sampling volume (i.e., "support") for heterogenous media have a significant effect on reported measurement values. For instance, particle size has a varying affect on the transport and fate of contaminants in the environment and on the potential receptors. Comparison of data from methods that are based on different supports can be difficult Defining the sampling support is important in the early stages of site characterization This may be accomplished through the DQO process with existing knowledge of the site, contamination, and identification of the exposure pathways that need to be characterized. Refer to Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies (U.S. EPA, 1992f) for more information about soil sampling support. Soil Texture. The soil texture class (e.g., loam, sand, silt loam) is necessary to estimate average soil moisture conditions and to estimate infiltration rates.-A soil's texture classification is determined from a particle size analysis and the U.S. Department of Agriculture (USDA) soil textural triangle shown at the top of Figure 9. This classification system is based on the USDA soil particle size classification at the bottom of Figure 9. The particle size analysis method in Gee and Bauder (1986) can provide this particle size distribution also. Other particle size analysis methods may be used as long as they provide the same particle size breakpoints for sand/silt (0.05 mm) and silt/clay (0.002 mm). Field methods are an alternative for determining soil textural class; an example from Brady (1990) is also presented in Figure 9. 112 TUT 1249 Figure 9: U.S. Department of Agriculture soil texture classification. 100 to Percent Sand Criteria Used with the Field Method for Determining Soil Texture Classes (Source: Brady, 1990) Criterion Sand Sandy loam Loam Silt loam Clay loam Clay V Individual grains Yes Yes visible to eye 2. Stability of dry Do not form Do not form clods 3. Stability of wet Unstable Slightly stable dods 4. Stability of Does not Does not form •ribbon'when form wetsoi rubbed between thumb and fingers Some Few No Easty Moderately . Hard and broken easily broken atable Moderately Stable Very stable atable Does not form Broken appearance Thin, wil break No Very hard and stable Very stable Very long, flexible 0.002 Particle Size, mm 0.05_____0.10 0.25 0.5 1.0 2.0 U.S. Department of Agriculture Clay Silt Very Fine |f=ine|Med.| Coarse J Sand Very Coarse Gravel Source: USDA. 113 TUT Dry Bulk Density. Dry soil bulk density (pb) is used to calculate total soil porosity and can be determined for any soil horizon by weighing a thin-walled tube soil sample (e.g., Shelby tube) of known volume and subtracting the tube weight to estimate field bulk density (ASTM D 2937). A moisture content determination (ASTM 2216) is then made on a subsample of the tube sample to adjust field bulk density to dry bulk density. The other methods (e.g., ASTM D 1556. D 2167, D 2922) are not generally applicable to subsurface soils. ASTM soil testing methods are readily available in the Annual Book of ASTM Standards, Volume 4.08, Soil and Rock; Building Stones, which is available from ASTM, 100 Barr Harbor Drive, West Conshohocken, PA, 19428. Organic Carbon and pH. Soil organic carbon is measured by burning off soil carbon in a controlled- temperature oven (Nelson and Sommers, 1982). This parameter is used to determine soil-water partition coefficients from the organic carbon soil-water partition coefficient, K^. Soil pH is used to select site-specific partition coefficients for metals and ionizing organic compounds (see Pan 5). This simple measurement is made with a pH meter in a soil/water slurry (McLean, 1982) and may be measured in the field using a portable pH meter. 4.2.4 Define the Study Boundaries. As discussed in Section 4.1.4, areas that are known to be highly contaminated (i.e., sources) are targeted for subsurface sampling. The information collected on source area and depth is used to calculate she-specific SSLs for the inhalation and migration to ground water pathways. Contamination is defined by the lower of the CLP practical quantitation limit for each contaminant or the SSL. For the purposes of this guidance, source areas are defined by area and depth as contiguous zones of contamination. However, discrete sources that are near each other may be combined and investigated as a single source if site conditions warrant. 4.2.5 Develop a Decision Rule. The decision rule for subsurface soils is: If the mean concentration of a contaminant within a source area exceeds the screening level, then investigate that area further. In this case "screening level" means the SSL. As explained in Section 4.1.5, statistics other than the mean (e.g., the maximum concentration) may be used as estimates of the mean in this comparison as long as they represent valid or conservative estimates of the mean. 4.2.6 Specify Limits on Decision Errors. EPA recognizes that data obtained from sampling and analysis can never be perfectly representative or accurate and that the costs of trying to achieve near-perfect results can outweigh the benefits. Consequently, EPA acknowledges that uncertainty in data must be tolerated to some degree. The DQO process attempts to control the degree to which uncertainty in data affects the outcomes of decisions that are based on data. The sampling intensity necessary to accurately determine the mean concentration of subsurface soil contamination within a source with a specified level of confidence (e.g., 95 percent) is impracticable for screening due to excessive costs and difficulties with implementation. Therefore, EPA has developed an alternative decision rule based on average concentrations within individual soil cores taken in a source: If the mean concentration within any soil core taken in a source exceeds the screening level, then investigate that source further. 114 TUT DOS 1251 For each core, the mean core concentration is defined as the depth-weighted average concentration within the zone of contamination (see Section 4.2.7). Since the soil cores are taken in the area(s) of ,.,»„, highest contamination within each source, the highest average core concentration among a set of core samples serves as a conservative estimate of the mean source concentration. Because this rule is not a statistical decision, it is not possible to statistically define limits on decision errors. Standard limits on the precision and bias of sampling and analytical operations conducted during the sampling program do apply. These are specified by the Superfund quality assurance program requirements (U.S. EPA, 1993d), which must be followed during the subsurface sampling effort If field methods are used, at least 10 percent of field samples should be split and sent to a CLP laboratory for confirmatory analysis (U.S. EPA, 1993d). Although the EPA does not require full CLP sample tracking and quality assurance/quality control (QA/QC) procedures for measurement of soil properties, routine EPA QA/QC procedures are recommended, including a Quality Assurance Project Plan (QAPjP), chain-of-custody forms, and duplicate analyses. 4.2.7 Optimize the Design. Within each source, the Soil Screening Guidance suggests taking two to three soil cores using split spoon or Shelby tube samplers. For each soil core, samples should begin at the ground surface and continue at approximately 2-foot intervals until no contamination is encountered or to the water table, whichever is shallower. Subsurface sampling depths and intervals can be adjusted at a site to accommodate site-specific information on surface and subsurface contaminant distributions and geological conditions (e.g.. large vadose zones in the West). The number and location of subsurface soil sampling (i.e., soil core) locations should be based on knowledge of likely surface soil contamination patterns and subsurface conditions. This usually means that core samples should be taken directly beneath areas of high surface soil contamination. Surface soils sampling efforts and field measurements (e.g., soil gas surveys) taken during she reconnaissance will provide information on source areas and high contaminant concentrations to help target subsurface sampling efforts. Information in the CSM also will provide information on areas likely to have the highest levels of contamination. Note that there may be sources buried in subsurface soils that are not discernible at the surface. Information on past practices at the site included in the CSM can help identify such areas. Surface geophysical methods also can aid in identifying such areas (e.g., magnetometry to detect buried drums). The intensity of the subsurface soil sampling needed to implement the soil screening process typically will not be sufficient to fully characterize the extent of subsurface contamination. In these cases, conservative assumptions should be used to develop hypotheses on likely contaminant distributions (e.g., the assumption that soil contamination extends to the water table). Along with knowledge of subsurface hydrogeology and stratigraphy, geostatistics can be a useful tool in developing subsurface contaminant distributions from limited data and can provide information to help guide additional sampling efforts. However, instructions on the use of geostatistics is beyond the scope of this guidance. , Samples for measuring soil parameters should be collected when taking samples for measuring contaminant concentrations. If possible, consider splitting single samples for contaminant and soil parameter measurements. Many soil testing laboratories have provisions in place for handling and testing contaminated samples. However, if testing contaminated samples is a problem, samples may be taken from clean areas of the site as long as they represent the same soil texture and series and are ~ • 115 TUT OO8 1252 taken from the same depth as the contaminant concentration samples. The SAP developed for subsurface soils should specify sampling and analytical procedures as well as the development of QA/QC procedures. To identify the appropriate analytical procedures, the screening levels must be known. If data are not available to calculate site-specific SSLs. then the generic SSLs in Appendix A should be used. Finally, soil investigation for the migration to ground water pathway should not be conducted independently of ground water investigations. Contaminated ground water may indicate the presence of a nearby source area, with contaminants leaching from soil into the aquifer. 4.2.8 Analyzing the Data. The mean soil contaminant concentration for each soil core should be compared to the SSL for the contaminant. The soil core average should be obtained by averaging analyses results for the discrete samples taken along the entire soil core within the zone of contaminav; n (compositing will prevent the evaluation of contaminant concentration trends with depth). If each subsurface soil core segment represents the same subsurface soil interval (e.g.. 2 feet), then the average concentration from the surface to the depth of contamination is the simple arithmetic average of the concentrations measured for core samples representative of each of the 2-foot segments fr-. *he surface to the depth of contamination or to the water table. However, if the intervals r Jl of the same length (e.g., -cm; are 2 feet while others are 1 foot or 6 inches), then the ~«i.:..^i:on of the average concentration in the total core must account for the different lengths of the intervals. If c, is the concentration measured in a core sample representative of a core interval of length 1,, and the n-th interval is considered to be the last interval in the source area (i.e., the n-th sample represents the depth of contamination), then the average concentration in the core from the surface to the depth of contamination should be calculated as the following depth-weighted average c = f Ic (61) i*l 11 II, If the leach test option is used, a sample representing the average contaminant concentration within the zone of contamination should be formed for each soil core by combining discrete samples into a composite sample for the test. The composites should include only samples taken within the zone of contamination (i.e., clean soil below the lower limit of contamination should not be mixed with contaminated soil). As with any Superfund sampling effort, all analytical data should be reviewed to ensure that Superfund quality assurance program requirements are met (U.S. EPA, 1993d). 4.2.9 Reporting. The decision process for subsurface soil screening should be thoroughly documented.. This documentation should contain as a minimum: a map of the site showing the contaminated soil sources and any areas assumed not to be contaminated, the soil core sampling points within each source, and the soil core sampling points -that were compared with the SSLs; the depth and area assumed for each source and their basis; the -average soil properties used to calculate 116 TUT OO8 1253 SSLs for each source; a description of how samples were taken and (if applicable) how composite samples were formed; the raw analytical data; the average soil core contaminant concentrations compared with the SSLs for each source: and theresults of all QA/QC analyses. 4.3 Basis for the Surface Soil Sampling Strategies: Technical Analyses Performed This section describes a series of technical analyses conducted to support the sampling strategy for surface soils outlined in the Soil Screening Guidance. Section 4.3.1 describes the sample design procedure presented in the December 1994 draft guidance (U.S. EAP, 1994H). The remaining sections describe the technical analyses conducted to develop the final SSL sampling strategy. Section 4.3.2 describes an alternative, nonparametric procedure that EPA considered but rejected for the soil screening strategy. Section 4.3.3 describes the simulations conducted to support the selection of the Max test and the Chen test in the final Soil Screening Guidance. These simulation results also can be used to determine sample sizes for site conditions not adequately addressed by the tables in Section 4.1. Quantitation limit and multiple comparison issues are discussed in Sections 4.3.4 and 4.3.5, respectively. Section 4.3.6 describes a limited investigation of compositing samples within individual EA sectors or strata. 4.3.1 1994 Draft Guidance Sampling Strategy. The DQO-based sampling strategy in the 1994 draft Soil Screening Guidance assumed a lognormal distribution for contaminant levels over an EA and derived sample size determinations from lognormal confidence interval procedures by C. E. Land (1971). This section summarizes the rationale for this approach and technical issues' raised- by peer review. por ^e ^994 draft §0jj Screening Guidance, EPA based the surface soil SSL methodology on the comparison of the arithmetic mean concentration over an EA with the SSL. As explained in Section 4.1, this approach reflects the type of exposure to soil under a future residential land use scenario. A person moving randomly across a residential lot would be expected to experience an average concentration of contaminants in soil. Generally speaking, there are few nonparametric approaches to statistical inference about a mean unless a symmetric distribution (e.g., normal) is assumed, in which case the mean and median are identical and inference about the median is the same as inference about the mean. However, environmental contaminant concentration distributions over a surface area tend to be skewed with a long right tail, so symmetry is not plausible, in this case the main options for inference about means are inherently parametric, i.e., they are based on an assumed family of probability distributions. In addition to being skewed with a long right tail, environmental contaminant concentration data must be positive because concentration measurements cannot be negative. Several standard two- parameter probability models are nonnegative and skewed to the right, including the gamma. lognormal, and Weibull distributions. The properties of these distributions are summarized in Chapter 12 of Gilbert (1987). The lognormal distribution is the distribution most commonly used for environmental contaminant data (see, e.g., Gilbert, 1987, page 164). The lognormal family can be easy to work with in some respects, due to the work of Land (1971, 1975) on estimating confidence intervals for lognormal parameters, which are also described in Gilbert (1987). TU"T 008 1254 The equation for estimating the Land upper confidence limit (UL) for a lognormal mean has the form TTT /- S» S*H ^ (62) UL = exp( y + -TJ- where y and sy are the average and standard deviation of the sample log concentrations. The lower confidence limit (LL) has a similar form. The factor H depends on sy and n and is tabulated in Gilbert (1987) and Land (1975). If the data truly follow a lognormal distribution, then the Land confidence limits are exact (i.e., the coverage probability of a 95 percent confidence interval is 0.95). The problem formulation used to develop SSL DQOs in the 1994 draft Soil Screening Guidance tested the null hypothesis HO. \i £ 2 SSL versus the alternative hypothesis HI: \i < 2 SSL, with a Type I error rate of 0.05 (at 2 SSL), and a Type II error rate of 0.20 at 0.5 SSL (\i represents the true EA mean). Tha*is the probability of incorrectly deciding not to investigate further when the true mean is 2 SSL was set net to exceed 0.05, and the probability of incorrectly deciding to investigate further when the true mean i& 0.5 SSL was not to exceed 0.20. This null hypothesis can be tested at the 5 percent level of significance by calculating Land's upper 95 percent confidence limit for a lognormal mean, if one assumes that the true EA concentrations are lognormally distributed. The null hypothesis is rejected if the upper confidence limit falls below 2 SSL. Simulation studies of the Land procedure were used to obtain sample size estimates that achieve these DQOs for different possible values of the standard deviation of log concentrations. Additional simulation studies were conducted to calculate sample sizes and to investigate the properties of the Land procedure in situations where specimens are composited. All of these simulation studies assumed a lognormal distribution of site concentrations. If the underlying site distribution is lognormal, then the composites, viewed as physical averages, are not lognormal (although they may be approximately lognormal). Hence, correction factors are necessary to apply the Land procedure with compositing, if the individual specimen concentrations are assumed lognormal. The correction factors were also developed through simulations. The correction factors are multiplied by the sample standard deviation, sy, before calculating the confidence limit and conducting the test. Procedures for estimating sample sizes and testing hypotheses about the site mean using the Land procedure, with and without compositing, are described in the 1994 draft Technical Background Document (U.S. EPA, 1994i). A peer review of the draft Technical Background Document identified several issues of concern. • The use of a procedure relying strongly on the assumption of a lognormal distribution • Quantitation limit issues • Issues associated with multiple hypothesis tests where multiple contaminants are present in site soils. 118 TUT 008 1255 The first issue is of concern because the small sample sizes appropriate for surface soil screening will not provide sufficient data to validate this assumption. To address this issue, EPA considered several alternative approaches and performed extensive analyses. These analyses are described in Sections 4.3.2 and 4.3.3. Section 4.3.3 describes extensive simulation studies involving a variety of distributions that were done to compare the Land, Chen. and Max tests and to develop the latter two as options for soil screening. 4.3.2 Test of Proportion Exceeding a Threshold. One of the difficulties noted for the Land test, described in Section 4.3.1, is its strong reliance on an assumption of lognormality (see Section 4.3.3). Even in cases where the assumption may hold, there will rarely be sufficient information to test h. A second criticism of applying the Land test (or another test based on estimating the mean) is that values must be substituted for values reported as less than a quantitation limit (<QL). (As noted in Section 4.3.4, how one does this substitution is of little relevance if the SSL is much larger than the QL. However, even if a moderate proportion of the data values fall below the QL and are censored, then the lognormal distribution may not be a good model for the observed concentrations.) A third criticism of using the Land test for screening is its requirement for large sample sizes when the contaminant variability across the EA is expected to be large (e.g., a large coefficient of variation). Because of these drawbacks to applying the Land procedure, EPA considered, alternative. nonparametnc procedures. One such alternative that was considered is the test described below. For a given contaminant, let P represent the proportion of all possible sampling units across the EA for which the concentration exceeds 2 SSL. In essence, P represents the proportion of the EA with true contaminant levels above 2 SSL. A nonparametric test involving P was developed as follows. Let P0 be a fixed proportion of interest chosen hi such a way that if that proportion (or more) of the EA has contamination levels above 2 SSL, then that EA should be investigated further. One way to obtain a rough equivalence between the test for a mean greater than 2 SSL and a test involving P is to choose 1 -Po to correspond to the percentile of the lognormal distribution at which the mean occurs. One can show that this is equivalent to choosing P0 = 1-* [O.5a] = 1-*[0.5V ln(l + CV 2)] . <63> where o = assumed standard deviation of the logarithms of the concentrations CV = assumed coefficient of variation of the contaminant concentrations <1> = distribution function of the standard normal distribution. Here, the fixed proportion P0 will be less than one-half. The hypotheses are framed as HO: P > PO (EA needs further investigation) versus 119 TUT OO8 1256 Hj: P<P 0 (EA does not need further investigation). The test is based on concentration data from a grid sample of N points in the EA (\\ -n compositing). Let p represent the proportion of these n points with observed concentrations L - ~r than or equal to 2 SSL. The test is carried out by choosing a critical value, pc, to meet the a. >.red Type I error rate, that is. a = Prob (p < pc | P = PO) = 0.05. (64) The sample size should be chosen to satisfy the Type II error rate at some specified alternative value PI, where PI < PO- For example, to have an 80 percent power at Pj. l-P«Prob(} pc | P = P,) « 0.80. (65) If the same type of rationale for choosing PO (corresponding to 2 SSL) is used to make PI correspond to 0.5 SSL, then one would choose P! = 1 - <& [ 0.5 c -i- 1.386/0]. (66) Sample sizes for this test were developed based on the preceding formulation and were found to be approximately the same as those required by the Land procedure, though they tended to be slightly higher than the Land sample sizes for small c, and slightly smaller for large c. The major advantage of this test, in contrast to the Land procedure, for example, is its generality, the only assumption required is that random sampling be used to select the sample points. Its principal disadvantages are: » • Compositing of samples cannot be included (since the calculation of p requires the count of the number of units with observed levels at or above 2 SSL). • The test does not deal directly with the mean contaminant level at the EA, which is the fundamental parameter for risk calculations. Because the test does not depend directly on the magnitude of the concentrations, it is possible that the test will give misleading results relative to a test based on a mean. This can occur, for example, when only a small portion of the EA has very high levels (i.e., a hot spot). In that case, the observed p will converge for increasing n to that proportion of the EA that is contaminated; it would do the same if the concentration levels in that same portion were just slightly above 2 SSL. A test based on a mean for large samples, however, is able to distinguish between these two situations; by its very nature, a test based on a proportion of measurements exceeding a single threshold level cannot. For these reasons, the test described here based on the proportion of observations exceeding 2 SSL was not selected for inclusion in the current guidance. 120 TUT 008 1257 4.3.3 Relative Performance of Land, Max, and Chen Tests. A simulation study was conducted to compare the Land, Chen, and Max tests and to determine sample sizes '"""" necessary to achieve DQOs. This section describes the design of the simulation study and summarizes its results. Detailed output from the simulations is presented in Appendix I. Treatment of Data Below the Quantitation Limit. Review of quantitation limits for . 110 chemicals showed that for more than 90 percent of the chemicals, the quantitation limit was less than 1 percent of the ingestion SSL. In such cases, the treatment of values below the QL is not expected to have much effect, as long as all data are used in the analysis, with concentrations assigned to results below the QL in some reasonable way. In the simulations, the QL was assumed to be SSL/100 and any simulated value below the QL was set equal to 0.5 QL. This is a conservative assumption based on the comparison of ingestion SSLs with QLs. Decision Rules. For the Land procedure, as discussed in Section 4.3.1, the null hypothesis H0: \i 2 2 SSL (where u represents the true mean concentration for the EA) can be tested at the 5 percent level by calculating Land's upper 95 percent confidence limit for a lognormal mean. The null hypothesis is rejected (i.e., surface soil contaminant concentrations are less than 2 SSL), if this upper confidence limit falls below 2 SSL. This application of the Land (1971) procedure, as described in the draft 1994 Guidance, will be referred to as the "SSL DQOs" and the "original Land procedure." For the Max test, one decides to walk away if the maximum concentration observed in composite samples taken from the EA does not exceed 2 SSL. As indicated in Section 4.1.6, it is viewed as providing a test of the original null hypothesis, HO: \L £ 2 SSL. The Max test does not inherently control either type of error rate (i.e., its critical region is always the region below 2 SSL, not where concentrations below a threshold that achieve a specified Type I error rate). However, control of error rates for the Max test can be achieved through the DQO process by choice of design (i.e., by choice of the number N of composite samples and choice of the number C of specimens per /*"~ w" composite). The Chen test requires that the null hypothesis have the form HO: \i £ V.Q, with the alternative hypothesis as Hj: H > M.Q (Chen, 1995). Hypotheses or DQOs of this form are referred to as "flipped hypotheses" or "flipped DQOs" because they represent the inverse of the actual hypothesis for SSL decisions. In the simulations, the Chen method was applied with HO = 0.5 SSL at significance levels (Type I error rates) of 0.4, 0.3, 0.2, 0.1, 0.05, 0.025, and 0.01. In this formulation, a Type I error occurs if one decides incorrectly to investigate further when the true site mean, p., is at or below 0.5 SSL. The two formulations of the hypotheses are equivalent in the sense that both allow achievement of soil screening DQOs. That is, working with either formulation, it is possible to control the probability of incorrectly deciding to walk away when the true site mean is 2 SSL and to also control the probability of incorrectly deciding to investigate further when the true site mean is 0.5 SSL. • In addition to the original Land procedure, the Chen test, and the Max test, the simulations also include the Land test of the flipped null hypothesis HO: u, ^ 0.5 SSL at the 10 percent significance level. This Land test of the flipped hypothesis was included to investigate how interchanging the null and alternative hypotheses affected sample sizes for the Land and Chen procedures. » Simulation Distributions. In the following description of the simulations, parameter acronyms used as labels in the tables of results are indicated by capital letters enclosed in parentheses. ,,-. , TUT 008 1258 Each distribution used for simulation is a mixture of a lower concentration distribution and a higher concentration distribution. The lower distribution represents the EA in its natural (unpolluted) state, and the higher distribution represents contaminated areas. Typically, all measurements of pollutants in uncontaminated areas are below the QL. Accordingly, the lower distribution is assumed to be completely below the QL. For the purposes of this analysis, it is unnecessary to specify any other aspect of the lower distribution, because any measurement below the QL is set equal to 0.5 QL. A parameter between 0 and 1, called the mixing proportion (MIX), specifies the probability allocated to the lower distribution. The remaining probability (1-MDQ is spread over higher values according to either a lognormal, gamma, or Weibull distribution. The parameters of the higher distribution are chosen so that the overall mixture has a given true EA mean (MU) and a given coefficient of variation (CV). Where s is the sample standard deviation, j is the sample mean, and C is the number of specimens per composite sample, CV is defined as. CV = or CV = The following parameter values were used in the simulations: EA mean (MU) = 0.5 SSL or 2 SSL EA coefficient of variation (CV) = 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, or 6 (i.e., 100 to 600 percent) Number of specimens per composite (C) = 1, 2, 3, 4, 5, 6, 8, 9, 12, or 16 Number of composites chemically analyzed (N) = 4, 5, 6, 7, 8, 9, 12, or 16. The true EA mean was set equal to 0.5 SSL or 2 SSL in order to estimate the two error rates of primary concern. Most CVs encountered in practice probably will lie between 1 and 2.5 (i.e.. variability between 100 and 250 percent). This expectation is based on data from the Hanford sue (see Hardin and Gilbert, 1993) and the Piazza .Road site (discussed in Section 4.3.6). EPA believes that the most practical choices for the number of specimens per composite will be four and six In some cases, compositing may not be appropriate (the case C - 1 corresponds to no compositing) EPA also believes that for soil screening, a practical number of samples chemically analyzed per EA lies below nine, and that screening decisions about soils in each EA should not be based on fewer than four chemical analyses. Fcr a given CV, there is a theoretical limit to how large the mixing proportion can be. The values of the mixing proportion used in the simulations are shown below as a function of CV. The case MIX = 0 corresponds to an EA characterized by a gamma, lognormal, or Weibull distribution A value of MIX near 1 indicates an EA where all concentrations are below the QL except those in a small portion of the EA. Neither of these extremes implies an extreme overall mean. If MIX = 0, the contaminating (higher) distribution can have a low mean, resulting in a low overall mean. If MIX is near 1 (i.e., a relatively small contamination area), a high overall mean can be obtained if the mean of the distribution of contaminant concentrations is high enough. 122 TUT 008 1259 cv 1.0 1.5 2.0 2.5 3.0 3.5 •4.0 5.0 6.0 Valu«s of MIX us«d in the simulations 0, 0.49 0, 0.50 0,0.50,0.75 0, 0.50, 0.85 0, 0.50, 0.85 0, 0.50, 0.90 0,0.50,0.90 0. 0.50, 0.95 0, 0.50, 0.95 Treatment Of Measurement Error. Measurement errors were assumed to be normally distributed with mean 0 (i.e., unbiased measurements) and standard deviation equal to 20 percent of the true value for each chemically analyzed sample. (Earlier simulations included measurement error standard deviations of 10 percent and 25 percent. The difference in results between these two cases was negligible.) Number Of Simulated Samples. Unique combinations of the simulation parameters considered (i.e., 2 values of the EA mean, 10 values for the number of specimens per composite, 8 values for the number of composite samples, 25 combinations of CV and MIX, and 3 contamination models—lognormal, gamma, Weibull), result in a total of 12,000 simulation conditions. One thousand simulated random samples were generated for each of the 12,000 cases obtained by varying the simulation parameters as described above. The average number of physical samples simulated from an EA for a hypothesis test (i.e., the product CN) was 56. The following 10 hypothesis tests were applied to each of the 12 million random samples: Chen test at significance levels of 0.4, 0.3, 0.2, 0.1, 0.05, 0.025, and 0.01 • Original Land test of the null hypothesis HO: H 2: 2 SSL at the 5 percent significance level Land test of the flipped null hypothesis H0: u £ 0.5 SSL at the 10 percent significance level • Maximum test. These simulations involved generation of approximately 650 million random numbers. Simulation Results. A complete listing of the simulation results, with 150 columns and 59 lines per page, requires 180 pages and is available from EPA on a 3.5-inch diskette. Representative results for gamma contamination data, with eight composite samples that each consist of six specimens, are shown in Table 32. The gamma contamination model is recommended for determining sample size requirements because it was consistently seen to be least favorable, in the sense that it required higher sample sizes to achieve DQOs than either of the lognormal or Weibull 123 TUT 003 models. Hence, sample sizes sufficient to protect against a gamma distribution of contaminant concentrations are also protective against a lognormal or Weibull distribution. Table 32. Comparison of Error Rates for Max Test, Chen Test (at .20 and .10 Significance Levels), and Original Land Test, Using 8 Composites of 6 Samples Each, for Gamma Contamination Data MU/SSL C=6 NsB CV=4 0.5 0.5 0.5 2.0 2.0. 2.0 MIX .00 .50 .90 .00 .50 .90 Max test .35 .40 .40 .06 .06 .04 0.20 Chen test .18 .22 .19 .10 .11 .16 0.10 Chen test .09 .11 .09 .18 .18 .29 Land test .99 .99 .98 .00 .00 .01 C=6 NsB CV=3 C=6 C=6 MU = MIX = C a N = CV = 0.;. .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 N-8 CV=2 0.5 .00 0.5 .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 Ns8 CV=1 0.5 .00 0.5 .49 2.0 .00 2.0 .49 .24 .25 .23 .04 .03 .03 .07 .06 .04 .02 .02 .01 .00 .00 .01 .01 .18 .19 .22 .03 .03 .06 .22 .19 .19 .00 .00 .00 .20 .20 .00 .00 .10 .10 .11 .06 .05 .12 .11 .09 .10 .00 .01 .01 .10 .12 .00 .00 .93 .94 .99 .00 .00 .00 .57 .68 .85 .01 .00 .00 .01 .12 .02 .00 True EA Mean - see subsection entitled "Simulation Distributions" in Section 4.3.3. Mixing Proportion - see subsection entitled "Simulation Distributions" in Section 4.3.3 Number of specimens in a composite. Number of composites analyzed. EA coefficient of variation (Jc }s * where s = sample standard deviation and x s mean sample concentration Table 32 shows that the original Land method is unable to control the error rates at 0.5 SSL for gamma distributions. This limitation of the Land method was seen consistently throughout the results for all nonlognonnal distributions tested. This limitation led to removal of the Land procedure from the Soil Screening Guidance. 124 • TUT 008 1261 Earlier simulation results for gamma and WeibuU distributions did not censor results below the QL and used pure unmixed distributions. In these cases, as the sample size N increased, with all other factors fixed, the Land error rates at 0.5 SSL increased toward 1. Normally, the expectation is that as the sample size increases, information increases, and error rates decrease. When using data from a Weibull or gamma distribution, the Land confidence interval endpoints converge to a value that does not equal the true site mean, m , and results in an increase in error rates. This phenomenon is easily demonstrated, as follows. Let X denote the concentration random variable, let Y = ln(X) denote its logarithm. Let m and oy denote the mean and standard deviation of logarithms of the soil concentrations. Then, as the sample size increases, the Land confidence interval endpoints (UL and LL) converge to f^ m, + -f) . (67) If X is lognormally distributed, this expression is the mean of X. If X has a Weibull or gamma distribution, this expression is not the mean of X. This inconsistency accounts for the increase in error rates with sample size. Table 32 also shows the fundamental difference between the Max test and the Chen test. For the Max test, the probability of error in deciding to walk away when the EA mean is 2.0 SSL is fairly stable, ranging from 0.01 to 0.06 across the different values of the CV. On the other hand, these error rates vary more across the CV values for the Chen test (e.g., from 0.00 to 0.29 for Chen test at the 0.10 significance level). This occurs because the Chen test is designed to control the other type of error rate (at 0.5 SSL). The Max test is presented in the 1995 Soil Screening Guidance (U.S. EPA, 1995c) because of its simplicity and the stability of its control over the error rate at 2 SSL. .,«,,. Table 33 shows error rate estimates for four to nine composite samples that each consist of four, six, or eight specimens for EAs with CVs of 2, 2.5, 3, or 3.5. and assuming a gamma distribution. Table. 33 should be adequate for most SSL planning purposes. However, more complete simulation results are reported in Appendix I. Planning for CVs at least as large as 2 is recommended because it is known that CVs greater than 2 occur in practice (e.g., for two of seven EAs in the Piazza Road simulations reported in Section 4.3.6). One conclusion that can be drawn from Table 33 is that composite sample sizes of four are often inadequate. Further support for this conclusion is reported in the Piazza Road simulations discussed in Section 4.3.6. Conclusions. The primary conclusions from the simulations are: • For distributions other than lognormal, the Land procedure is prone to decide to investigate further at 0.5 SSL, when the correct decision is to walk away. It is therefore unsuitable for surface soil screening. • Both the Max test and the Chen test perform acceptably under a variety of distributiona] assumptions and are potentially suitable for surface soil screening. 125 TUT 008 1262 Table 33. Error Rates of Max Test and Chen Test at .2 (C20) and .1 (C10) Significance Level for CV = 2, 2.5, 3, 3.5 N C = 4 4 4 r 6 7 7 8 8 9 9 C = 6 4 4 5 5 6 6 7 7 8 8 9 9 C s 8 4 4 5 5 6 6 7 7 8 8 9 9 MU/SSL 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 •: .5 .-. 0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 0.5 2.0 CV 8 Max .09 .13 .11 .10 .11 .06 .12 .04 .16 .02 .16 .01 .03 .14 .04 .09 .06 .04 .06 . .02 .06 .02 .06 .01 .02 .12 .03 .07 .02 .04 .03 .03 .04 .02 .04 .01 2.0 C20 .20 .08 .21 .05 .21 .03 .20 .03 .19 .02 .21 .01 .20 .03 .20 .02 .20 .01 .20 .00 ' .19 .00 .20 .00 .21 .02 .22 .01 .18 .00 .20 .00 .20 .00 .21 .00 C10 .11 .16 .10 .11 .12 .08 .10 .05 .09 .03 .11 ".02 .12 .08 .10 .05 .11 .02 .09 .01 .09 .01 .10 .01 .13 .05 .11 .02 .09 .01 .11 .00 .10 .00 .11 .00 CV s Max .14 .19 .15 .10 .21 .08 .25 .05 .25 .04 .28 .03 .08 .16 .11 .09 .14 .06 .12 .05 .15 .02 .18 .02 .06 .15 .05 .09 .08 .06 .09 .04 .11 .02 .11 .02 2.5 C20 .18 .17 .18 .09 .20 .08 .22 .04 .20 .03 .20 .03 .21 .08 .17 .04 .21 .03 .19 .02 .20 .01 .22 .01 .19 .04 .20 .02 .21 .01 .20 .01 .21 .01 .21 .00 C10 .09 .28 .09 .18 .10 .14 .11 .09 .09 .07 .09 .06 .12 .17 .09 .10 .10 .07 .10 .04 .10 .03 .11 .02 .10 .09 .11 .06 .11 .02 .11 ,01 .11 .01 .10 .01 CV = Max .19 .20 .26 .17 .28 .11 .31 ,08 .36 .05 .36 .04 .15 .17 .17 .13 .19 .09 .23 .06 .25 .03 .28 .03 .10 .17 .11 .09 .13 .07 .18 .04 .17 .04 .20 .01 3.0 C20 .18 .21 .20 .19 .21 .13 .20 .11 .20 .08 .18 .07 .20 .14 .20 .10 .20 .07 .22 .06 .19 .03 .20 .02 .21 .09 .20 .04 .19 .04 .21 .02 .21 .01 .19 .00 C10 .08 .33 .08 .30 .11 .23 .09 .18 .10 .14 .09 .13 .10 .24 .10 .18 .10 .14 .10 .10 .10 .05 .11 .04 .10 .17 .10 .10 .10 .07 .11 .04 .10 .03 .10 .01 CV = Max .24 .26 .26 .18 .31 .11 .36 .08 .42 .07 .44 .07 .16 .20 .22 .15 .25 .09 .29 .08 .30 .04 .34 .03 .14 .19 .17 .12 .20 .08 .22 .05 .26 .03 .30 .02 3.5 C20 .20 .29 .20 .23 .19 .18 .18 .14 .20 .13 .22 .12 .17 .19 .20 .13 .20 .10 .21 .09 .19 .06 .19 .05 .18 .14 .19 .08 .20 .07 .20 .05 .19 .03 .23 .02 C10 .10 .42 .09 .36 .09 .28 .10 .23 .09 .21 .12 .20 .08 .33 .10 .24 .10 .19 .10 .14 .10 .11 .09 .09 .08 .25 .09 .17 .10 .13 .11 .09 .10 .06 .12 .04 s True EA Mean - see subsection entitled "Simulation Distributions" in MIX= Mixing Proportion - see subsection entitled "Simulation Distributions* in Section 4.3.3 C = Number of specimens in a composite. N = Number of composites analyzed. CV = EA coefficient of variation (J~c }s where s = sample standard deviation and x = mean sample concentration 126 TUT 008 1263 4.3.4 Treatment of Observations Below the Limit of Quantitation. Test procedures that are based on estimating a mean contaminant level for an EA, such as the Land and Chen procedures, make use of each measured concentration value. For this reason, the use of all reported concentration measurements in such calculations should be considered regardless of their magnitude—that is, even if the measured levels fall below a quantitation level. One argument for this approach is that the, QL is itself an estimate. Another is that some value will have to be substituted for any censored data point (i.e., a point reported as <QL), and the actual measured value is at least as accurate as a substituted value. The peer review of the Draft Soil Screening Guidance raised the following issue: If such censored value's do occur in a data set, what values should be used? There is a substantial amount of literature on this subject and a variety of sophisticated approaches. In the context of SSLs, however, a simple approach is recommended. Consistent with general Superfund guidance, each observation reported as "<QL" shall be replaced with 0.5 QL for computation of the sample mean. The evidence suggests that the ingestion SSL generally will be 2 orders of magnitude or more greater than the QL for most contaminants. In these cases, the results of soil screening will be insensitive to alternative procedures that could be used to substitute values for observations reported as "<QL " When the SSL is not much greater than tlie QL (e.g., SSL < 50 QL), the outcome of the soil screening could be affected by the procedure used to substitute for "<QL" values. The most conservative approach would be to substitute the concentration represented by the QL itself for all observations reported as "<QL." In the context of the SSLs, however, the simple approach of using 0.5 QL is suggested. This will be sufficiently conservative given the conservative factors underlying the SSLs. 4.3.5 Multiple Hypothesis Testing Considerations. The Soil Screening Guidance addresses the following hypothesis testing problem for each EA: H0: mean concentration of a given chemical £ 2 SSL versus H]: mean concentration of a given chemical < 2 SSL. . The default value for the probability of a Type I error is a - 0.05, while the default value for the power of the test at 0.5 SSL is 1-6 = 0.80. The test is applied separately for each chemical, so that these probabilities apply for each individual chemical. Thus, there is an 80 percent probability of walking away from an EA (i.e., rejecting HO) when only one chemical is being tested and its true mean level is 0.5 SSL and a 5 percent probability of walking away if its true mean level is 2 SSL. However, the Soil Screening Guidance does not explicitly address the following issues: What is the composite probability of walking away from an EA if there are multiple contaminants? and If such probabilities are unacceptable, how should one compensate when testing for multiple contaminants within a single EA? 127 TUT 008 j.264 The answer to the first question cannot be determined, in general, since the concentrations of the various contaminants will often be dependent on one another (e.g.. this would be expected if they originated from the same source of contamination). The joint probability of walking away can be determined, however, if one makes the simplifying assumption that the contaminant concentrations for the different chemicals are independent (unconrelated). In that case, the probability of walking away is simply the product of the individual rejection probabilities For two chemicals (Chemical A and Chemical B, say), this is: Pr{walking away from EA} = Pr{reject HO for Chemical A) x Pr{reject HO for Chemical B}. While these joint probabilities must be regarded as approximate, they nevertheless serve to illustrate the effect on the error rates when dealing with multiple contaminants. Assume (for illustrative purposes only) that the probabilities for rejecting the null hypothesis (walking away from the EA) for each single chemical appear as follows: True concentration 0.2 SSL 0.5 SSL 0.7 SSL 1.0 SSL 1.5 SSL 2.0 SSL Probability of rejecting H0 0.95 0.80 (default 1-B) 0.60 0.50 0.20 0.05 (default a) Let C(A) denote the concentration of Chemical A divided by the SSL* and let P(A) denote the corresponding probability of rejecting HQ. Define C(B) and P(B) similarly for Chemical B. Assuming independence, the joint probabilities of rejecting the null hypothesis (walking away) are as shown in Table 34. Table 34. Probability of "Walking Away11 from an EA When Comparing Two Chemicals to SSLs Chemical A C(A) 0.2 0.5 0.7 1.0 1.5 2.0 P(A) 0.95 0.80 0.60 0.50 .0.20 0.05 Chemical B C(B) = 0.2 P(B) =.95 0.90 0.76 0.57 0.48 0.19 0.05 C(B) = 0.5 P(B) = .80 0.76 0.64 0.48 0.40 0.16 0.04 C(B) = 0.7 P(B) = .60 0.57 0.48 0.36 0.30 0.12 0.03 C(B) s 1.0 P(B) s .50 0.48 0.40 0.30 0.25 0.10 0.03 C(B) = 1.5 P(B) = .20 0.19 0.16 0.12 0.10 0.04 0.01 C(B) = 2.0 P(B) = .05 0.05 0.04 0.03 0.03 0.01 <0.01 128 TUT OOS 1265 These probabilities demonstrate that the test procedure will tend to be very conservative if multiple '"""" chemicals are involved—that is. all of the chemical concentrations must be quite low relative to their SSL in order to have a high probability of walking away from the EA. On the other hand, there will be a high probability that further investigation will be called for if the mean concentration for even a single chemical is twice the SSL. ' A potential problem occurs when there are several chemicals under consideration and when all or most of them have levels slightly below the SSL (e.g., near 0.5 SSL). For instance, if each of six independent chemicals had levels at 0.5 SSL, the probability of rejecting the null hypothesis would be 80 percent for each such chemical, but the probability of walking away from the EA would be only (0.80)6 = 0.26. If the same samples are being analyzed for multiple chemicals, then the original choice for the number of such samples ideally should have been based on the worst case (i.e., the chemical expected to have the largest variability). In this case, the probability of correctly rejecting the null hypothesis at 0.5 SSL for the chemicals with less variability will be higher. The overall probability of walking away will be greater than shown above if all or some of the chemicals have less variability than assumed as the basis for determining sample sizes. Here, the sample size will be large enough for the probability of rejecting the null hypothesis at 0.5 SSL to be greater than 0.80 for these chemicals. The probability values assumed above for deciding that no further investigation is necessary for individual chemicals, which are the basis for these conclusions, are equally applicable for the Land, Chen, and Max tests. They simply represent six hypothetical points of the power curves for these tests (from 0.2 SSL to 2.0 SSL). Therefore, the conclusions are equally applicable for each of the hypothesis 'testing procedures that have been considered in the current guidance for screening surface soils. If the surface soil concentrations are positively correlated, as expected when dealing with multiple chemicals, then it is likely that either all the chemicals of concern have relatively high concentrations or they all have relatively low concentrations. In this case, the probability of making the correct decision for an EA would be greater than that suggested by the above calculations that assume independence of the various chemicals. However, the potential problem of several chemicals having concentrations near 0.5 SSL is not precluded by assuming positive correlations. In fact, it suggests that if the EA average for one chemical is near 0.5 SSL, then the average for others is also likely to be near 0.5 SSL, which is exactly the situation where the probability of not walking away from the EA can become large because there is a high probability that HO will be rejected for at least one of these chemicals. An alternative would be to use multiple hypothesis testing procedures to control the overall error rate for the set of chemicals (i.e., the set of hypothesis tests) rather than the separate error rates for the individual chemicals. Guidance for performing multiple hypothesis tests is beyond the scope of the current document. Obtain the advice of a statistician familiar with multiple hypothesis testing procedures if the overall error rates for multiple chemicals is of concern for a particular site. The classical statistical guidance regarding this subject is Simultaneous Statistical Inference (Miller, 1991) 4.3.6 Investigation of Compositing Within EA Sectors, if one decides that an EA needs further investigation, then it is natural to inquire which portion(s) of the EA exceed the screening level. This is a different question than simply asking whether or not the EA average soil concentration exceeds the SSL. Conceivably, this question may require additional sampling, chemical ,x~v ^ 129 ~UT OO8 1266 analysis, and statistical analysis. A natural question is whether this additional effort can be avoided by forming composites within sectors (subareas) of the EA. The sector with the highest estimated concentration would then be a natural place to begin a detailed investigation. The simulations to investigate the performance of rules to decide whether further investigation is required, reported in Section 4.3.3, make specific assumptions about the sampling design. It is assumed that N composite samples are chemically analyzed, each consisting of C specimens selected to be statistically representative of the entire EA. The key point, in addition to random sampling, is that composites must be formed across sectors rather than within sectors. This assumption is necessary to achieve composite samples that are representative of the EA mean (i.e., have the EA mean as their expected value). If compositing is limited to sectors, such as quadrants, then each composite represents its sector, rather than the entire EA. The simulations reported in Section 4.3.3, and sample sizes based on them, do not apply to this type of compositing. This does not necessarily preclude compositing within sectors for both purposes, i.e., to test the hypothesis about the EA mean and also to indicate the most contaminated sector. However, little is known about the statistical properties of this approach when applying the Max test, which would depend on specifics of the actual spatial distribution of contaminants for a given EA. Because of the lack of extensive spatial data sets for contaminated soil, there is limited basis for determining what sample sizes would be adequate for achieving desired DQOs for various sites. However, one spatial data set was available and used to investigate the performance of compositing within sectors at one site. piazza Road Simulations. Data from the Piazza Road NPL site were used to investigate the properties of tests of the EA mean based on compositing within sectors, as compared to compositing between sectors. The investigation of a single site cannot be used to validate a given procedure, but it may indicate whether further investigation of the procedure is worthwhile. Seven honoverlapping 0 4-acre EAs were defined within the Piazza Road site. Each EA is an 8-by-12 grid composed of 14'xl4' squares. The data consist of a single dioxin measurement of a composite sample from each small square. These measurements are regarded as true values for the simulations reported in this section. Measurement error was incorporated in the same fashion as for the simulations reported in Section 4.3.3. Each of the seven EAs was subdivided into four 4-by-6 sectors, six 4-by-4 sectors, eight 4-by-3 sectors, twelve 2-by-4 sectors, and sixteen 2-by-3 sectors. Results are presented here for the cases of four, six, and eight sectors because composites of more than eight specimens are expected to be used rarely, if at all. Table 35 presents the "true" mean and CV for each EA, computed from all 96 measurements within the 0.4-acre EA. The CVs range from 1.0 to 2.2. Note-that two of the seven CVs equal or exceed 2 at this site. This supports EPA's belief that at many sites it is prudent, when planning sample size requirements for screening, to assume a CV of at least 2.5 and to consider the possibility of CVs as large as 3 or 3.5. As data on variability within EAs for different sites and contaminant conditions accrue over time, it will be possible to base the choice of procedures on a larger, more comprehensive database, rather than just a single site. Appendix J contains results of simulations from the seven Piazza Road EAs. Sampling with 130 TUT 008 1267 replacement from each sector was used, because this was felt to be more consistent with the planned compositing. To estimate the error rates at 0.5 SSL and 2 SSL for each EA, the SSL was defined so that the site mean first was regarded as 0.5 SSL and then was regarded as 2 SSL. Notation for Results from Piazza Road Simulations. The following notation is used in Appendix J. The design variable (DES) indicates whether compositing was within sector (DES=W) or across sectors (DES=X). As in Section 4.3.3, C denotes the number of specimens per composite, and N denotes the number of composite samples chemically analyzed. Results in Appendix J are for the Chen test at the 10 percent significance level and for the Max test. The true mean and CV are shown in the header for each EA. Table 35. Means and CVs for Dioxin Concentrations for 7 Piazza Road Exposure Areas _ Mean of EA CV of EA N 1 2 3 4 5 6 7 2.1 2.4 5.1 4.0 9.3 15.8 2.8 1.0 1.6 1.1 .1.2 2.0 2.2 1.4 96 96 96 96 96 96 96 Results and Conclusions from Piazza Road Simulations. Although the results from a single site cannot be assumed to apply to all sites, the following observations can be made based on the Piazza Road simulations reported in Appendix J. • The error rate at 0.5 SSL for the Chen test, using compositing across sectors (DES=X), is generally close to the nominal rate of 0.10. For compositing within sectors (DES=W), the error rate for Chen at 0.5 SSL is generally much lower than the nominal rate. • Except for plans involving only four analyses (N = 4), the error rate at 2 SSL is always below 0.05 for the Chen test. For the Max test, the error rate at 2 SSL fluctuated between 0 and 16 percent. The error rate at 2 SSL is smaller for the Chen test at the 10 percent significance level than for the Max test in virtually all cases The only two exceptions to this are for compositing within sector (DES=W) in EA No. 6. • This observation provides further support for the conclusion drawn from the simulations reported in Section 4.3.3: plans involving only four analyses can result in high error rates in determining the mean contaminant concentration of an EA with the Max test. In most cases the error rates of concern to EPA (at 2 SSL) are 0.10 or larger. 131 TUT 008 1268 In general, error rates estimated from Piazza Road simulations for compositing across sectors are at least as small as would be predicted on the basis of the simulation results reported in Section 4.3.3. The simulation results show that compositing within sectors using the Max test may be an option for she managers who want to know whether one sector of an EA is more contaminated than the other. However, use of the Max test when compositing within sectors may lead the site manager to draw conclusions about the mean contaminant concentration in that sector only, not across the entire EA. 132 TUT 008 1.269 Part 5: CHEMICAL-SPECIFIC PARAMETERS Chemical-specific parameters required for calculating soil screening levels include the organic carbon normalized soil-water partition coefficient for organic compounds (K<>£), the soil-water partition coefficient for inorganic constituents (Kj), water solubility (S), Henry's law constant (HLC, H'), air diffusivhy (D^). and water diffusivity (DirW). In addition, the octanol-water partition coefficient (Kow) is needed to calculate KOC values. This pan of the background document describes the collection and compilation of these parameters for the SSL chemicals. With the exception of values for air diffusivity (D^J, water diffusivity (D j,w), and certain Koc values, all of the values used in the development of SSLs can be found in the Superfund Chemical Data Matrix (SCDM). SCDM is a computer code that includes more than 25 datafiles containing specific chemical parameters used to calculate factor and benchmark values for the Hazard Ranking System (HRS). Because SCDM datafiles are regularly updated, the user should consult the most recent version of SCDM to ensure that the values are up to date. 5.1 Solubility, Henry's Law Constant, and Kow Chemical-specific values for solubility, Henry's law constant (HLC), and K<,w were obtained from SCDM. In the selection of the value for SCDM, measured or analytical values are favored over calculated values. However, in the event that a measured value is not available, calculated values are used. Table 36 presents the solubility, Henry's law constant, and K<,w values taken from SCDM and used to calculate SSLs. Henry's law constant values were available for all but two of the constituents of interest. Henry's law constants could not be obtained from the SCDM datafiles for either carbazole or mercury. As a consequence, this parameter was calculated according to the following equation: HLC = (VP)(M)/(S) (68) where HLC = Henry's law constant (atm-m3/mol) VP = vapor pressure (atm) M = molecular weight (g/mol) S - solubility (mg/L or g/m3). The SSL equations require the dimensionless form of Henry's law constant, or H1, which is calculated from HLC (atm-m3/mol) by multiplying by 41 (U.S. EPA, 1991b). The values taken from SCDM for HLC and the calculated dimensionless values for H' are both presented in Table 36. 5.2 Air (Dj,.) and Water (Dj,w) Diffusivities Few published difrusivrties were available for the subject chemicals for air (Du) and water (DIW) Water and air diffusivities were obtained from the CHEMDAT8 model chemical properties database (DATATWO.WK1). For chemicals not in CHEMDAT8, diffusivities were estimated using the 133 TUT O08 WATERS model correlations for air and water diffusivities. Both CHEMDAT8 and WATERS can be obtained from EPA's SCRAM bulletin board system, as described in Section 3.1.2. Table 37 presents the values used to calculate SSLs. Table 36. Chemical-Specific Properties Used in SSL Calculations CAS No. 83-32-9 67-64-1 309-00-2 120-12-7 ~-55-3 r .-43-2 205-99-2 207-08-9 65-85-0 50-32-8 111-44-4 117-81-7 75-27-4 75-25-2 71-36-3 85-68-7 86-74-8 75-15-0 56-23-5 57-74-9 106-47-8 108-90-7 124-48-1 67-66-3 95-57-8 218-01-9 72-54-8 72-55-9 50-29-3 53-70-3 84-74-2 95-50-1 106-46-7 91-94-1 75-34-3 Compound Acenaphthene Acetone Aldrin Anthracene Benz(a)anthracene Benzene Benzo(b)fluoranthene Benzo{/c)fluoranthene Benzoic acid Benzo(a)pyrene Bis(2-chk>roethyl)ether Bis(2-ethylhexyl)phthalate Bromodichloromethane Bromoform Butanol Butyl benzyl phthalate Carbazole Carbon disulftde Carbon tetrachloride Chlordane p-Chloroaniline Chlorobenzene Chlorodibromomethane Chloroform 2-Chlorophenol Chrysene ODD DDE DOT Dibenz(a,/))anthracene Dt-n-butyl phthalate 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 3,3-Dichlorobenzidine 1,1-Dichloroethane S (mg/L) 4.24E+00 1 .OOE+06 1.80E-01 4.34E-02 9.40E-03 1 .75E+03 1.50E-03 8.00E-04 3.50E+03 1.62E-03 1.72E+04 3.40E-01 6.74E+03 3.10E+03 7.40E+04 2.69E+00 7.48E-t-00 1.19E+03 7.93E+02 5.60E-02 5.30E+03 4.72E+02 2.60E+03 7.92E+03 2.20E+04 1 .60E-03 9.00E-02 1.20E-01 2.50E-02 .2.49E-03 1.12E+01 1.56E+02 7.38E+01 3.11E+00 5.06E-f03 HLC H (•tm-m3/mol) (dim«nsionl»ss) 1 .55E-04 3.88E-05 1.70E-04 6.50E-05 3.35E-06 5.55E-03 1.11E-04 8.29E-07 1.54E-06 1.13E-06 1 .80E-05 1.02E-07 1 .60E-03 5.35E-04 8.81 E-06 1.26E-06 1.53E-08» 3.03E-02 3.04E-02 4.86E-05 3.31 E-07 3.70E-03 7.83E-04 3.67E-03 3.91 E-04 9.46E-05 4. 00 E-06 2.10E-05 8.10E-06 1.47E-08 9.38E-10 1.90E-03 2.43E-03 4.00E-09 5.62E-03 6.36E-03 1.59E-03 6.97E-03 2.67E-03 1.37E-04 2.28E-01 4.55E-03 3.40E-05 6.31 E-05 4.63E-05 7.38E-04 4.18E-06 6.56E-02 2.19E-02 3.61 E-04 5.17E-05 6.26E-07 1.24E+00 1.25E+00 1 .99E-03 1 .36E-05 1.52E-01 3.21 E-02 1.50E-01 1 .60E-02 3.88E-03 1.64E-04 8. 61 E-04 3.32E-04 6.03E-07 3.85E-08 7.79E-02 9.96E-02 1.64E-07 2.30E-01 log KOW 3.92 -0.24 . 6.50 4.55 5.70 2.13 6.20 6.20 1.86 6.11 1.21 7.30 2.10 2.35 0.85 4.84 3.59 2.00 2.73 6:32 1.85 2.86 2.17 1.92 2.15 5.70 6.10 6.76 6.53 6.69 4.61 3.43 3.42 3.51 1.79 134 1271. Table 36 (continued) CAS No. 107-06-2 75-35-4 156-59-2 156-60-5 120-83-2 78-87-5 542-75-6 60-57-1 84-66-2 105-67-9 51-28-5 121-14-2 606-20-2 117-84-0 115-29-7 72-20-8 100-41-4 206-44-0 86-73-7 76-44-8 1024-57-3 118-74-1 87-68-3 319-84-6 319-85-7 58-89-9 77-47-4 67-72-1 193-39-5 78-59-1 7439-97-6 72-43-5 74-83-9 75-09-2 95-48-7 91-20-3 98-95-3 86-30-6 621-64-7 Compound 1 ,2-Dichloroethane 1 ,1-Dichloroethylene c/s-1 ,2-Dichloroethylene frans-1 ,2-Dichloroethylene 2,4-Dichlorophenol 1 ,2-Dichloropropane 1 ,3-Dichloropropene Dieldrin Diethylphthalate 2,4-Dimethylphenol 2,4-Dinitrophenol 2,4-Dinitrptoluene 2,6-Dinrtrotoluene Di-n-octyl phthalate Endosulfan Endrin Ethylbenzene Fluoranthene Fluorene Heptachlor Heptachlor epoxide Hexachlorobenzene Hexachloro-1 ,3-butadiene a-HCH (a-BHC) P-HCH 0-BHC) Y-HCH (Lindane) Hexachlorocyctopentadiene Hexachloroethane lndeno(1 ,2,3-ccQpyrene Isophorone Mercury Methoxychlor Methyl bromide Methylene chloride 2-Methylphenol Naphthalene Nitrobenzene A/-Nitrosodiphenylamine AANrtrosodi-n-propylamine S (mg/L) 8.52E+03 2.25E+03 3.50E+03 6.30E+03 4.50E+03 2.80E+03 2.80E+03 1.95E-01 1 .08E+03 7.87E+03 2.79E+03 2.70E+02 1 .82E+02 2.00E-02 5.IOE-01 2.50E-01 1 .69E+02 2.06E-01 1.98E+00 1 .80E-01 2.00E-01 6.20E+00 3.23E+00 2.00E+00 2.40E-01 6.80E+00 1 .80E+00 5.00E+01 2.20E-05 1 .20E+04 — ' 4.50E-02 1.52E+04 1 .30E+04 2.60E+04 3.10E+01 2.09E+03 3.51 E+01 9.89E+03 HLC H (atm-m'/mol) (dim«nsionl«ss) log K,,w 9.79E-04 2.61 E-02 4.08E-03 9.38E-03 3.16E-06 2.80E-03 1.77E-02 1.51E-05 4.50E-07 2.00E-06 4.43E-07 9.26E-08 7.47E-07 6.68E-05 1.12E-05 7.52E-06 7.88E-03 1 .61 E-05 6.36E-05 1.48E+00 9.50E-06 1 .32E-03 8.15E-03 1.06E-05 7.43E-07 1.40E-05 2.70E-02 3.89E-03 1.60E-06 6.64E-06 1.14E-02*> 1 .58E-05 6.24E-03 2.19E-03 1.20E-06 4.83E-04 2.40E-05 5.QOE-06 2.25E-06 4.01 E-02 1 .07E-I-00 1.67E-01 3.85E-01 1.30E-04 1.15E-01 7.26E-01 6.19E-04 1.85E-05 8.20E-05 1.82E-05 3.80E-06 3.06E-05 2.74E-03 4.59E-04 3.08E-04 3.23E-01 6.60E-04 2.61 E-03 6.07E+01 3.90E-04 5.41 E-02 3.34E-01 4.35E-04 3.05E-05 5.74E-04 1.1lE-fOO 1 .59E-01 6.56E-05 2.72E-04 4.67E-01 6.48E-04 2.56E-01 8.98E-02 4.92E-05 : 1.98E-02 9.84E-04 2.05E-04 9.23E-05 1.47 2.13 1.86 2.07 3.08 1.97 2.00 5.37 2.50 2.36 1.55 2.01 1.87 8.06 4.10 5.06 3-14 5.12 4.21 , 6.26 5.00 5.89 4.81 3.80 3.81 3.73 5.39 4.00 6.65 1.70 — ' 5.08 1.19 1.25 1.99 3.36 1.84 3.16 1.40 135 TUT 008 1272 Table 36 (continued) CAS No. Compound 87-86-5 Pentachlorophenol 108-95-2 Phenol 129-00-0 Pyrene 100-42-5 Styrene 79-34-5 1,1,2,2-Tetrachloroethane 1 27- 1 8-4 Tetrachloroethy lene 108-88-3 Toluene 8001 -35-2 Toxaphene 1 20-82-1 1 ,2,4-Trichlorobenzene 71 -55-6 1,1,1 -Trichloroethane 79-00-5 1,1,2-Trichloroethane 79-01-6 Trichloroethylene 95-95-4 2,4,5-Trichlorophenol 88-06-2 2,4,6-Trichlorophenol 108-05-4 Vinyl acetate 75-01-4 Vinyl chloride 108-38-3 m-Xylene 95-47-6 o-Xylene 106-42-3 p-Xylene CAS = Chemical Abstracts Service. S = Solubility in water (20-25 *C). HLC = Henry's law constant. H' = Dimensionless Henry's law constant KOW = Octanol/water partition coefficient. S (mg/L) 1 .95E+03 8.28E+04 1.35E-01 3.10E+02 2.97E+03 2.00E+Q2 5.26E+02 7.40E-01 3.00E+02 1 .33E+03 4.42E+03 1.10E+03 1 .20E+03 8.00E+02 2.006+04 2.76E+03 1.61 En-02 1.78E+02 1 .85E+02 (HLC Iatnvm3/mol] • HLC was calculated using the equation: HLC = vapor pressure ' HLC (atm-m^/mol) 2.44E-08 3.97E-07 1.10E-05 2.75E-03 3.45E-04 T.84E-02 6.64E-03 6.00E-06 1 142E-03 1.72E-02 9.13E-04 1.03E-02 4.33E-06 7.79E-06 5.11E-04 2.70E-02 7.34E-03 5.19E-03 7.66E-03 *41)(25'C). H (dim*nsionl«8s) 1.00E-06 1.63E-05 4.51 E-04 1.13E-01 1.41E-02 7.54E-01 2.72E-01 2.46E-04 5.82E-02 7.05E-01 3.74E-02 4.22E-01 1.78E-04 3.19E-04 2.10E-02 1.11E+00 3.01 E-01 2.13E-01 3;14E-01 ' molecular wt. / solubility. Vapor pressure log Kow 5.09 - 1.48 5.11 2.94 2.39 2.67 2.75 5.50 4.01 2.48 2.05 2.71 3.90 3.70 0.73 1.50 3.20 3.13 3.17 JS6.B3E-10 atm and molecular weight is 1 67.21 g/mol for carbazole. t> Value from WATERS model database. 136 TUT Table 37. Air Diffusivity (Di>a) and Water Diffusivity (Di>w) Values for SSL Chemicals (25°C)» CAS No. 83-32-9 67-64-1 309-00-2 120-12-7 56-55-3 71-43-2 205-99-2 207-08-9 65-85-0 50-32-8 111-44-4 117-81-7 75-27-4 75-25-2 71-36-3 85-68-7 86-74-8 75-15-0 56-23-5 57-74-9 106-47-8 108-90-7 124-48-1 67-66-3 95-57-8 218-01-9 72-54-8 72-55-9 50-29-3 53-70-3 84-74-2 95-50-1 106-46-7 91-94-1 75-34-3 107-06-2 75-35-4 156-59-2 156-60-5 120-83-2 Compound Acenaphthene Acetone Aldrin Anthracene Benz(a)anthracene Benzene Benzo(b)fluoranthene Benzo(ft)fluoranthene Benzoic acid Benzo(a)pyrene Bis(2-chloroethyl)ether Bis(2-ethylhexyl)ph1halate Bromodichloromethane Bromoform Butanol Butyl benzyl phthalate Carfoazole Carbon disulfide Carbon tetrachtoride Chlordane p-Chloroaniline Chlorobenzene Chlorodibromomethane Chloroform 2-Chlorophenol Chrysene ODD DDE DDT Dibenz(a,/?)anthracene Di-n-butyl phthalate 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 3,3-Dichlorobenzidine 1,1 -Dichloroethane 1 ,2-Dichloroethane 1 ,1 -Dichloroethylene c/s-1 ,2-Dichioroethylene f/ans-1 ,2-Dichloroethylene 2 ,4-Dichlorophenol Du (cm*/s) 4.21 E-02 1.24E-01 1.32E-02 3.24E-02 5.10E-02 8.80E-02 2.26E-02 2.26E-02 5.36E-02 4.30E-02 6.92E-02 3.51 E-02 2.98E-02 1.49E-02 8.00E-02 1.74E-02b 3.90E-02 b 1.04E-01 7.80E-02 1.18E-02 4.83E-02 7.30E-02 1.96E-02 1 .04E-01 5.01 E-02 2.48E-02 1.69E-02 b 1 .44E-02 1.37E-02 2.02E-02 b 4. 38 E-02 6.90E-02 6.90E-02 1.94E-02 7.42E-02 1.04E-01 9.00E-02 7.36E-02 7.07E-02 346E-02 D,.w <cm*/s) 7.69E-06 1.14E-05 4.86E-06 7.74E-06 9.00E-06 9.80E-06 5.56E-06 5.56E-06 7.97E-06 9.00E-06 7.53E-06 3.66E-06 1.06E-05 1.03E-05 9.30E-06 4.83E-06 b 7.03E-06 b 1.00E-05 8.80E-06 4.37E-06 1.01E-05 8.70E-06 1.05E-05 1.00E-05 9.46E-06 6.21 E-06 4.76E-06 b 5.87E-06 4.95E-06 5.18E-06 b 7.86E-06 7.90E-06 7.90E-06 6.74E-06 1.05E-05 9.90E-06 1:04E-05 1.13E-05 1.19E-05 8.77E-06 137 TUT 008 1274 Table 37 (continued) CAS No. 78-87-5 542-75-6 60-57-1 84-66-2 105-67-9 51-28-5 121-14-2 606-20-2 117-84-0 115-29-7 72-20-8 100-41-4 206-44-0 86-73-7 76-44-8 1024-57-3 118-74-1 87-68-3 319-84-6 319-85-7 58-89-9 77-47-4 67-72-1 193-39-5 78-59-1 7439-97-6 72-43-5 74-83-9 75-09-2 95-48-7 91-20-3 98-95-3 86-30-6 621-64-7 87-86-5 108-95-2 129-00-0 100-42-5 79-34-5 127-18-4 Compound 1 ,2-Dichloropropane 1 ,3-Dichloropropene DieWrin Diethylphthalate 2,4-Dimethylphenol 2,4-Dinttrophenol 2 ,4-Din rtrololuene 2,6-Dinrtrotoluene Di-noctyl phthalate Endosuttan Endrin Ethylbenzene Fluoranthene Fluorene Heptachlor Heptachlor epoxide Hexachlorobenzene Hexachloro-1 ,3-butadiene a-HCH (a-BHC) pMHCH (P-BHC) rHCH (Lindane) Hexachlorocyclopentadiene Hexachloroethane lndeno(1 ,2,3-cd)pyrene Isophorone Mercury Methoxychlor Methyl bromide Methylene chloride 2-Methylphenol Naphthalene Nitrobenzene N-Nitrosodiphenylamine N-Nitrosodi-/>propylamine Pentachlorophenol Phenol Pyrene Styrene 1 ,1 ,2,2-Tetrachloroethane Tetrachtoroethylene Dif.(cm*/s) 7.82E-02 6.26E-02 1.25E-02 2.56E-02 b 5.84E-02 2.73E-02 2.03E-01 3.27E-02 1.51E-02 1.15E-02 1.25E-02 7.50E-02 3.02E-02 3.63E-02 b 1.12E-02 1.32E-02 b 5.42E-02 5.61 E-02 1.42E-02 1.42E-02 1.42E-02 1.61E-02 2.50E-03 1.90E-02 6.23E-02 3.07E-02 b 1.56E-02 7.28E-02 1.01E-01 7.40E-02 . 5.90E-02 7.60E-02 3.12E-02 b 5.45E-02 b 5.60E-02 8.20E-02 2.72E-02 b 7.10E-02 7.10E-02 7.20E-02 D,.w (cm*/s) 8.73E-06 1.00E-05 4.74E-06 6.35E-06 b 8.69E-06 9.06E-06 7.06E-06 7.26E-06 3.58E-06 4.55E-06 4.74E-06 7.80E-06 6.35E-06 7.88E-06 b 5.69E-06 4.23E-06 b 5.91 E-06 6.16E-06 7.34E-06 7.34E-06 7.34E-06 7.21 E-06 6.80E-06 5.66E-06 6.76E-06 6.30E-06 b 4.46E-06 1.21E-05 1.17E-05 8.30E-06 7.50E-06 S.60E-06 6.35E-06 b 8.17E-06 b 6.10E-06 9.10E-06 7.24E-06b 8.00E-06 7.90E-06 8.20E-06 13S TUT 008 I27S Table 37 (Continued) CAS No. 108-88-3 8001-35-2 120-82-1 71-55-6 79-00-5 79-01-6 95-95-4 88-06-2 108-05-4 75-01-4 108-38-3 95-47-6 106-42-3 Compound Toluene Toxaphene 1 ,2,4-Trichiorobenzene 1,1,1 -Trichloroethane 1 ,1 ,2-Trichloroethane Trichloroethylene 2,4,5-Trichlorophenol 2,4,6-Trichtorophenol Vinyl acetate Vinyl chloride /n-Xylene o-Xylene p-Xylene Du (cm*/*) 8.70E-02 1.16E-02 3.00E-02 7.80E-02 7.80E-02 7.90E-02 2.91 E-02 3.18E-02 8.50E-02 1.06E-01 7.00E-02 8.70E-02 7.69E-02 DiiW (cm2/s) 8.60E-06 4.34E-06 8.23E-06 8.80E-06 8.80E-06 9.10E-06 7.03E-06 6.25E-06 9.20E-06 1.23E-06 7.80E-06 1.00E-05 8.44E-06 CAS = Chemical Abstracts Service. • Value from CHEMDATB model database unless indicated otherwise. b Estimated using correlations in WATERS rmdel. 5.3 Soil Organic Carbon/Water Partition Coefficients (Koe) Application of SSLs for the inhalation and migration to ground water pathways requires Koc values for each organic chemical of concern. K«c values are also needed for site-specific exposure modeling efforts. An initial review of the literature uncovered significant variability in this parameter, with reported measured values for a compound sometimes varying over several orders of magnitude. This variability can be attributed to several factors, including actual variability due to differences in soil or sediment properties, differences in experimental and analytical approaches used to measure the values, and experimental or measurement error. To resolve this difficulty, an extensive literature review was conducted to uncover all available measured values and to identify approaches and information that might be useful in developing valid K<,c values. The soil-water partitioning behavior of nonionizing and ionizing organic compounds differs because the partitioning of ionizing orgamcs can be significantly influenced by soil pH. For this reason, different approaches were required to estimate Koc values for nonionizing and ionizing organic compounds. 5.3.1 Koc for Nonionizing Organic Compounds. As noted earlier, there is significant variability in reported Koc values and an extensive literature search was conducted to collect all available measured Koc values for the nonionizing hydrophobic organic compounds of interest. . .- In the literature search, misquotation error was minimized by obtaining the original references whenever possible. Values from compilations and secondary references were used only when the original references could not be obtained. Redundancy of values was avoided, although in rare 139 TUT 008 1276 instances it was not possible to determine if compilations included such values, especially when data were reported as "selected" values. In certain references, soil-water partition coefficients (e.g., K0 or Kp) were reported along with the organic carbon content of the soil. In these cases, Koc was computed by dividing K<j by the fractional soil organic carbon content (foc, g/g). If the partition coefficient was normalized to soil organic matter (i.e., K,,m), it was converted to K^ as follows (Dragun, 1988): (69) where 1.724 = conversion factor from organic matter to organic carbon ({<„, = 1.724 foc ) K^ = partition coefficient normalized to organic matter (L/kg) fom = fraction organic matter (g/g). Once collected, K^ values were reviewed. It was not possible to systematically evaluate each source for accuracy or consistency or to analyze sources of variability between references because of wide variations in soil and sediment properties, experimental and analytical methods, and the manner in which these were reported in each reference. This, and the limited number of Koc values for many compounds, prevented any meaningful statistical analysis to eliminate outliers. Collected values were qualitatively reviewed, however, and some values were excluded. Values measured for low-carbon-content sorbents (i.e., foc £ 0.001) are generally beyond the range of the linear relationship between soil organic carbon and Ka and were rejected in most cases. Some references produced consistently high or low values and, as a result, were eliminated. Values were also eliminated if they fell outside the range of other measured values. The final values used are presented in Appendix K along with their reference sources. Summary statistics for the measured Koc values are presented in Table 38. The geometric mean of the KOC for each nonionizing organic compound is used as the the central tendency K^ value because it is a more suitable estimate of the central tendency of a distribution of environmental values with wide variability. The data contained in Table 38 are summarized in Table 39 for each of the nonionizing organic compounds for which measured KOC values were available. As shown, measured values are available for only a subset of the SSL compounds. As a consequence, an alternative methodology was applied to determine Koc values for the entire set of nonionizing hydrophobia organic compounds of interest. It has long been noted that a strong linear relationship exists between KOC and KOVk (octanol/water partition coefficient) (Lyman et al., 1982) and that this relationship can be used to predict KOC in the absence of measured data. One such relationship was reported by Di Toro (1985). This relationship was selected for use in calculating KOC values for most semivolatile nonionizing organic compounds (Group 1 in Table 39) because it considers particle interaction and was shown to be in conformity with observations for a large set of adsorption-desorption data (Di Toro, 1985). Di Toro's equation is as follows. log Koc = 0.00028 + (0.983 x iog K»w) (70) 140 008 12?7 For volatile organic compounds (VOCs), Equation 70 consistently overpredicted K,,c values when compared to measured data. For this reason, a separate regression equation was developed using log KOW and measured log KOC values for VOCs, chlorinated benzenes, and certain chlorinated pesticides: log K«e = 0.0784 + (0.7919 x log KoW) (71) Equation 71 was developed from a linear regression calculated at the 95 percent confidence level. The correlation coefficient (r) was 0.99 with an r? of 0.97. The compounds and data used to develop this equation are provided in Appendix K. Equation 71 was used to calculate K^ values for VOCs, chlorobenzenes, and certain chlorinated pesticides (i.e., Group 2 in Table 39). Log Koc values calculated using Equations 70 and 71 were rounded to two decimal places, and the resulting KOC values were rounded to two decimal places in scientific notation (i.e, as they appear in Table 39) prior to calculating SSLs. Table 38. Summary Statistics for Measured KOC Values: Nonionizing Organics* Kee (L/kg) Compound Acenaphthene Aldrin Anthracene Benz(a)anthracene Benzene Benzo(a)pyrene Bis(2-chloroethyl)ether Bis(2-e1hylhexyl)phthalate Bfomoform Butyl benzyl phthalate Carbon tetrachloride Chlordane Chlorobenzene Chloroform ODD DDE DOT Dibenz(a,h)anthracene 1 ,2-Dichlorobenzene (o) 1 ,4-Dichtorobenzene (p) 1 ,1 -Dichloroethane 1 ,2-Dichloroethane 1 ,1 -Dichloroethylene Geometric Moan 4,898 48,685 23,493 357,537 62 968,774 76 111,123 126 13,746 152 51,310 224 53 45,800 86,405 677,934 1,789,101 379 616 53 38 65 Average 5.028 48,686 24,362 459,882 66 1,166,733 76 114,337 126 14,055 158 51,798 260 57 45,800 86,405 792,158 2.029,435 390 687 54 44 65 Minimum 3,890 48.394 14,500 150,000 31 478,947 76 87,420 126 11.128 123 44,711 83 28 45,800 86,405 285.467 565,014 267 273 46 22 65 Sample Maximum Size 6,166 48,978 33,884 840,000 100 2,130,000 76 141,254 126 16.981 224 58,884 500 81 45,800 86.405 1,741.516 3,059.425 529 1.375 62 76 65 2 2 9 4 13 3 1 2 1 2 3 2 9 5 1 1 6 . 14 9 16 2 3 1 141 TUT 008 1278 Table 38 (continued) *oc (L/kg) Geometric Compound Mean frans-1 ,2-Dichloroethylene 1 ,2-Dichloropropane 1 ,3-Dichloropropene Dieldrin Die:hy!phthalate Di-n-butylphthalate Endosulfan Endrin Ethytbenzene Fluoranthene Fluorene Heptachlor Hexachlorobenzene a-HCH (a-BHC) P-HCH (p-BHC) Y-HCH (Lindane) Methoxychlor Methyl bromide Methyl chloride Methylene chloride Naphthalene Nitrobenzene Pentachlorobenzene Pyrene Styrene 1 ,1 ,2,2-Tetrachloroethane Tetrachtoroethylene Toluene Toxaphene 1 ,2,4-Trichlorobenzene 1 ,1 ,1 -Trichloroethane 1 ,1 ,2-Trichloroethane Trichtoroethylene o-Xylene m-Xylene p-Xylene 38 47 27 25,546 82 1,567 2,040 10,811 204 49,096 7,707 9,528 80,000 1,762 2,139 1,352 80,000 9 6 10 1,191 119 32,148 67,992 912 79 265 140 95,816 1,659 135 75 94 241 196 311 Average Minimum 38 47 27 25,604 84 1,580 2,040 .11.422 207 49,433 8,906 10,070 80,000 1,835 2,241 1 ,477 80,000 9 6 10 1,231 141 36,114 70,808 912 79 272 145 95,816 1,783 139 77 97 241 204 313 38 47 24 23,308 69 1,384 2,040 7,724 165 41 ,687 3,989 6,810 80,000 1,022 1,156 731 80,000 9 6 10 830 31 11,381 43,807 912 79 177 94 95,816 864 106 60 57 222 158 260 Sample Maximum Size 38 47 32 27,399 98 1,775 2,040 15,885 255 54,954 16,218 13,330 80,000 2,891 3,563 3,249 80,000 9 6 10 1.950 270 55.176 133,590 912 79 373 247 95,816 3,125 179 108 150 258 289 347 1 1 3 3 2 2 1 4 5 3 6 2 1 12 14 65 1 1 1 1 20 10 5 27 1 1 15 12 1 17 5 4 21 4 3 3 * See Appendix K lor spurces of measured values. 142 TUT 008 1279 Table 39. Comparison of Measured and Calculated Koc Values CAS No. 83-32-9 67-64-1 309-00-2 120-12-7 56-55-3 71-43-2 205-99-2 207-08-9 50-32-8 111-44-4 117-81-7 75-27-4 75-25-2 71-36-3 85-68-7 86-74-8 75-15-0 56-23-5 57-74-9 106-47-8 108-90-7 1 24-48-1 67-66-3 218-01-9 72-54-8 72-55-9 50-29-3 53-70-3 84-74-2 95-50-1 106-46-7 91-94-1 75-34-3 107-06-2 75-35-4 156-59-2 156-60-5 78-87-5 542-75-6 Chemical Log Compound Group • KOW Acenaphthene Acetone AWrin Anthracene Benz(a)anthracene Benzene Benzo(&)fluoranthene Benzo(/c)fluoranthene Benzo(a)pyrene Bis(2-chloroethyl)ether Bis(2-ethylhexyl)phthalate Bromodichloromethane Bromoform Butanol Butyl benzyl phthalate Carbazole Carbon disulfide Carbon tetrachtoride Chiordane p-Chloroaniline Chlorobenzene Chlorodtbromomethane Chloroform Chrysene ODD DDE DDT Dibenz(a, rt)anthracene Di-n-butyl phthalate 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 3,3-Dichlorobenzidine 1,1-Dichloroethane 1 ,2-Dichloroethane 1 ,1-Dichloroethylene c/s-1 ,2-Dichloroethylene fna/is-1 ,2-Dichtoroethylene 1 ,2-Dichloropropane 1 ,3-Dichloropropene 1 1 1 1 1 2 1 1 1 1 1 2 2 1 1 1 2 2 2 1 2 2 2 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3.92 -0.24 6.50 4.55 5.70 2.13 6.20 6.20 6.11 1.21 7.30 2.10 2.35 0.85 4.84 3.59 2.00 2.73 6.32 1.85 2.86 2.17 1.92 5.70 6.10 6.76 6.53 6.69 4.61 3.43 3.42 3.51 1.79 1.47 2.13 1.86 2.07 1.97 2.00 LogKoe (L/kg) 3.85 -0.24 6.39 4.47 5.60 1.77 6.09 6.09 6.01 1.19 7.18 1.74 1.94 0.84 4.76 3.53 1.66 2.24 5.08 1.82 2.34 1.80 1.60 5.60 6.00 6.65 6.42 6.58 4.53 2.79 2.79 2.86 1.60 1.24 1.77 1.55 1.72 1.64 1.66 Calculated KOC (L/kg) 7.08E+03 5.75E-01 2.45E+06 2.95E+04 3.98E+05 5.89E+01 1.23E+06 1.23E+06 1.02E+06 1.55E+01 1.51 £+07 5.50E+Q1 8.71 E-t-01 6.92E+00 5.75E+04 3.39E+03 4.57E+01 1.74E+02 1.20E+05 6.61 E+01 2.19E+Q2 6.31 E-t-01 3.98E+01 3.98E+05 1.00E+06 4.47E+06 2.63E+06 3.80E+06 3.39E+04 6.17E-4-02 6.17E+02 7.24E+02 3.16E+01 1.74E+01 5.89E+01 3.55E+01 5.25E+01 4.37E+01 4.57E+01 Measured Kee (L/kg) 4.90E+03 — 4.87E+04 2.35E+04 3.58E+05 6.17E+01 — — 9.69E+05 7.59E-f01 1.1lE-f05 — 1.26E+02 — 1.37E*04 — — . 1.52E+02 5.13E+04 — ' 2.24E+02 — 5.25E+01 — ' 4.58E+04 8.64E-t-04 6.78E+05 1.79E-f06 1.57E+03 3.79E+02 6.16E+02 — 5.34E+01 3.80E+01 6.50E+01 — 3.80E+01 4.70E+01 2.71 E+01 143 TUT COS 12SO Table 39 (continued) CAS No. 60-57-1 84-66-2 105-67-9 121-14-2 606-20-2 117-84-0 115-29-7 72-20-8 100-41-4 206-44-0 86-73-7 76-44-8 v;; 4-57-3 118-74-1 87-68-3 319-84-6 319-85-7 58-89-9 77-47-4 67-72-1 193r39-5 78-59-1 72-43-5 74-83-9 75-09-2 95-48-7 91-20-3 98-95-3 86-30-6 621-64-7 1336-36-3 108-95-2' 129-00-0 100-42-5 79-34-5 127-18-4 108-88-3 8001-35-2 120-82-1 Chemical Log Compound Group * KOW Dieldrin Diethylphthalate 2,4-Dimethytphenol 2,4-Dinitrotoluene 2,6-Dinitrotoluene Di-n-octyi phthalate Endosulfan Endrin Ethylbenzene Fluoranthene Fluorene Heptachbr Heptachlor epoxide Hexachlorobenzene Hexachloro-1 ,3-butadiene a-HCH (a-BHC) p-HCH (P-BHC) •y-HCH (Lindane) Hexachlorocyclopentadiene Hexachloroethane lndeno(1 ,2,3-cd)pyrene Isophorone Methoxychlor Methyl bromide Methylene chloride 2-Methylphenol Naphthalene Nitrobenzene /V-Nitrosodiphenylamine /V-Nitrosodi-n-propylamine PCBs Phenol Pyrene Styrene 1 ,1 ,2,2-Tetrachloroethane Tetrachloroethylene Toluene Toxaphene 1 ,2,4-Trichlorobenzene 2 1 1 1 1 1 2 2 2 1 1 1 1 2 1 2 2 2 1 2 1 1 1 2 2 1 1 1 1 1 1 1 1 1 2 2 2 1 2 5.37 2.50 2.36 2.01 1.87 8.06 4.10 5.06 3.14 5.12 4.21 6.26 5.00 5.89 4.81 3.80 3.81 3.73 5.39 4.00 6.65 1.70 5.08 1.19 1.25 1.99 3.36 1.84 3.16 1.40 5.58 1.48 5.11 2.94 2.39 2.67 2.75. 5.50 4.01 LoglC^ (L/kg) 4.33 2.46 2.32 1.98 1.84 7.92 3.33 4.09 2.56 5.03 4.14 6.15 4.92 4.74 4.73 3.09 3.10 3.03 5.30 3.25 6.54 1.67 4.99 1.02 1.07 1.96 3.30 1.81 3.11 1.38 5.49 1.46 5.02 2.89 1.97 2.19 2.26 5.41 3.25 Calculated K«c (L/kg) 2.14E+04 2.88E+02 2.09 E+02 9.55E+01 6.92E+01 8.32E+07 2.14E+03 1.23E+04 3.63E+02 1.07E+05 1.38E+04 1.41E+06 8.32E+04 5.50E+04 5.37E+04 1.23E+03 1.26E+03 1.07E+03 2.00E+05 1.78E+03 3.47E+06 4.68E+01 9.77E+04 1.05E4-01 1.17E+01 9.12E+01 2.00E+03 6.46E+01 1.29E+03 2.40E+01 3.09E+05 2.88E+01 1.05E+05 7.76E+02 9.33E-H01 1.55E+02 1.82E+02 2.57E-I.05 1 .78E+03 Measured Koc (L/kg) 2.55E+04 8.22E+01 — • — — — 2.04E+03 1.08E+04 2.04E+02 4.91 E+04 7.71 E+03 9.53E+03 — 8.00E+04 — 1.76E+03 2.14E+03 1.35E+03 — — — i- 8.00E-f04 9.00E-1-00 1.00E+01 — 1.19E+03 1.19E+02 — — — — 6.80E+04 9.12E+02 7.90E+01 2.65E+02 1.40E+02 9.58E+04 1.66E+03 144 TUT 1281 Table 39 (continued) CAS No. 71-55-6 79-00-5 79-01-6 108-05-4 75-01-4 108-38-3 95-47-6 106-42-3 Chemical Log Compound Group * KOW 1 , 1 , 1 -Trichloroethane 1 , 1 ,2-Trichloroethane Trichloroethylene Vinyl acetate Vinyl chloride m-Xylene o-Xylene p-Xylene 2 2 2 1 2 2 2 2 2.48 2.05 2.71 0.73 1.50 3.20 3.13 3.17 LogKoc (L/kg) 2.04 1.70 2.22 0.72 1.27 2.61 2.56 2.59 Calculated KOC (L/kg) 1.10E+Q2 5.01 E+01 1.66E+Q2 5.25E+00 1.86E+01 4.07E+02 3.63E+02 3.89E+02 Measured KOC (L/kg) 1.35E+02 7.50E+01 9.43E+01 — — 1.96E+Q2 2.41 E+02 3.11E+02 • Group 1 : log KM s 0.9B3 log K™ * 0.00028. Group 2: (VOCs, chlorobenzenes, and certain chlorinated pesticides) log K^ s 0.791 9 tog Note: Calculated values rounded as shown for subsequent SSL calculations. 0.0784. 5.3,2 KOC for ionizing Organic Compounds. Sorption models used to describe the behavior of nonionizing hydrophobic organic compounds in the natural environment are not appropriate for predicting the partitioning of ionizable organic compounds. Certain organic compounds such as amines, carboxylic acids, and phenols contain functional groups that ionize under subsurface pH conditions (Schellenberg et al., 1984). Because the ionized and the neutral species of such compounds have different sorption coefficients, sorption models based solely on the partitioning of the neutral species may not accurately predict soil sorption under different pH conditions. To address this problem, a technique was employed to predict Koc values for the 15 ionizing SSL organic compounds over the pH range of the subsurface environment. These compounds include: Organic Acids Organic Bases Benzole acid 2-Chlorophenol 2,4-Dichlorophenol 2,4-Dimethylphenol 2,4-Dinitrophenol 2-Methylphenol Pentachlorophenol • Phenol • 2,3,4,5-Tetrachlorophenol • 2,3,4,6-Tetrachlorophenol • 2,4,5-Trichlorophenol • 2,4,6-Trichlorophenol • p-Chloroaniline • W-Nitrosodiphenylamine • W-Nitrosodi-n-propylamine Estimation of KOC values for these chemicals involves two analyses. First, the extent to which the compound ionizes under subsurface conditions must be determined to estimate the relative proportion of neutral and ionized species under the conditions of concern. Second, die K<,-c values for the neutral and ionized forms (KOC,D and KoC,i) must be determined and weighted according to the extent of ionization at a particular pH to estimate a pH-specific Koc value. For organic acids, the ionized species is an anion (A-) with a lower tendency to sorb to subsurface materials than the neutral species. Therefore, KOCii for organic acids is likely to be less than KoC,n. In the case of organic bases, the ionized species is positively charged (HB+) so mat K^ is likely to be greater than KOC,,,. 145 TUT 008 128: It should be noted that this approach is based on the assumption that the sorption of ionizing organic compounds to soil is similar to hydrophobic organic sorption in that the dominant sorbent is soil organic carbon. Shimizu et al. (1993) demonstrated that, for several "natural solids," pentachlorophenol sorption correlates more strongly with cation exchange capacity and clay content than with organic carbon content. This suggests that this organic acid interacts more strongly with soil mineral constituents than organic carbon. The estimates of Koc developed here may overpredict contaminant mobility because they ignore potential sorption to soil components other than organic carbon. Extent Of lonization. The sorption potential of ionized and neutral species differs because most subsurface solids (i.e., soil and aquifer materials) have a negative net surface charge. Therefore, r-sitively charged chemicals have a greater tendency to sorb man neutral forms, and neutral species sorb more readily than negatively charged forms. Thus, predictions for the total sorption of any ionizable organic compound must consider the extent to which it ionizes over the range of subsurface pH conditions of interest. Consistent with the EPA/Office of Solid Waste (EPA/OSW) Hazardous Waste Identification Rule (HWIR) proposal (U.S. EPA, 1992a), the 7.5th, 50th, and 92.5th percentiles (i.e., pH values of 4.9, 6.8, and 8.0) for 24,921 field-measured ground water pH values in the U.S. EPA STORET database are defined as the pH conditions of interest for SSL development. The extent of ionization can be viewed as the fraction of neutral species present that, for organic acids, can be determined from the following pH-dependent relationship (Lee et al., 1990): + 10 where ^Mcid = fraction of neutral species present for organic acids (unitless) [HA] = equilibrium concentration of organic acid (mol/L) [A-] = equilibrium concentration of anion (mol/L) pKa = acid dissociation constant (unitless). Using Equation 68, one can show that, in ground water systems with pH values exceeding the pKa by 1.5 pH units, the ionizing species predominates, and, in ground water systems with pH values that are 1.5 pH units less than the pKa, the neutral species predominates. At pH values approximately equal to the pKa, a mixed system of both neutral and ionizing components occurs. The fraction of neutral species for organic bases is defined by: I^D 1 IE* I < J > S III I * • . 1 I . = nb"' [B*] + [HB+] where ~ traction of neutral species present for organic bases (unitless) [B°] = equilibrium concentration of neutral organic base (mol/L) [HB+] = equilibrium concentration of ionized species (mol/L). As with organic acids, pH conditions determine the relative concentrations of neutral and ionized species in the system. However, unlike organic acids, the neutral species predominates at pH values 146 TUT 008 1283 that exceed the pKa. and the ionized species predominates at pH values less than the pKa For the SSL organic bases, A'-nitrosodi-n-propylamine and A^-nitrosodiphenylamine have very low pKa values and the neutral species are expected to prevail under environmental pH conditions. The pKa for />-chloroaniline, however, is 4.0 and, at low subsurface pH conditions (i.e., pH = 4.9), roughly 10 percent of the compound will be present as the less mobile ionized species. Table 40 presents pKa values and fraction neutral species present over the ground water pH range for the SSL ionizing organic compounds. This table shows mat ionized species are significant for only some of the constituents under normal subsurface pH conditions. The pKa values for phenol. 2- methylphenol, and 2,4-dimethylphenol are 9.8 or greater. Hence, the neutral species of these compounds predominates under typical subsurface conditions (i.e., pH = 4.9 to 8), and these compounds will be treated as nonionizing organic compounds (see Section 5.3.1). The pKa value for 2,4-dinitrophenol is less than 4 and the ionized species of this compound predominates under subsurface conditions. However, the pKas for 2-chlorophenol, 2,4-dichlorophenol, 2,4,5-trichlorophenol, 2,4,6-trichlorophenol, 2,3,4,5-tetrachlorophenol, 2,3,4,6-tetrachlorophenol, pentachlorophenol, and benzole acid fall within the range of environmentally significant pH conditions. Mixed systems consisting of both the neutral and the ionized species will prevail under such conditions with both species contributing to total sorption. Table 40. Degree of lonization (Fraction of Neutral Species, 4>) as a Function of pH Compound Benzole acid p-Chloroahilineb 2-Chlorophenol 2,4-Dichlorophenol 2 ,4-Dimethylphenol 2,4-Dinrtrophenol 2-Methylphenol A/-Nitrosodiphenylamineb N-Nitrosodi-n-propylamineb Pentachlorophenol Phenol 2,3,4,5-Tetrachlorophenol 2,3,4,6-Tetrachlorophenol 2 ,4 ,5-Trichlorophenol 2,4,6-Trichlorophenol pKaa 4.18 4.0 8.40 7.90 10.10 3.30 9.80 <0 <1 4.80 10.0 6.35c 5.30 7.10 6.40 pH = 4.9 0.1600 0.8882 0.9997 0.9990 1.0000 0.0245 1.0000 1.0000 0.9999 0.4427 1.0000 0.9657 0.7153 0.9937 0.9693 pH = 6.8 0.0024 0.9984 0.9755 0.9264 0.9995 0.0003 0.9990 1 .0000 1 .0000 0.0099 0.9994 0.2619 0.0307 0.6661 0.2847 pH = 8.0 0.0002 0.9999 0.7153 0.4427 0.9921 0.00002 0.9844 1 .0000 1 .0000 0.0006 0..9901 0.0219 0.0020 0.1118 0.0245 • Kolligetal. (1993). b Denotes that the compound is an organic base. c Lee eta!. (1991). 147 TUT 008 12S34 Prediction of Soil-Water Partition Coefficients. Lee et ai. (1990) developed a relationship from thermodynamic equilibrium considerations to predict the total sorption of an ionizable organic compound from the partitioning of its ionized and neutral forms: (74) where ' = soil organic carbon/water partition coefficient (L/kg) KOC r, = partition coefficient for the neutral species (L/kg) 4>n - fraction of neutral species present for acids or bases KOC,J ~ partition coefficient for the ionized species (L/kg). This relationship defines the total sorption coefficient for any ionizing compound as the sum of the weighted individual sorption coefficients for the ionized and neutral species at a given pH. Lee et al. (1990) verified that this relationship adequately predicts laboratory-measured K«c values for pentachlorophenol . A literature review was conducted to compile the pKa and the laboratory-measured values of KoC,n and Koc,i shown in Table 41. Data collected during this review are presented in RTI (1994), along with the references reviewed. Sorption coefficients for both neutral and ionized species were reported for only four of the nine ionizable organic compounds of interest. Sorption coefficients reported for the remaining compounds were generally Koc>n, and estimates of K oc>; were necessary to predict the compound's total sorption. The methods for estimating K^ for organic acids and organic bases are discussed separately in the following subsections. Organic Acids. Sorption coefficients for both the neutral and ionized species have been reported for two chlorophenolic compounds. 2,4,6-trichlorophenol and pentachlorophenol. For 2,4,5- trichlorophenol and 2,3,4,5-tetrachlorophenol, soil-water partitioning coefficient (Kp) data in the literature were adequate to allow calculation of K^i from Kp and soil foc (Lee et al., 1991). From ihese measured values, the ratios of KoCi; to KOCiI1 are: 0.1 (2,4,6-trichlorophenol), 0.02 (pentachlorophenol), 0.015 (2,4,5-trichlorophenol), and 0.051 (2,3,4,5-tetrachlorophenol). A ratio of 0.015 (1.5 percent) was selected as a conservative value to estimate K.OCi; for the remaining phenolic compounds, benzoic acid, and vinyl acetate. Organic Bases. No measured sorption coefficients for either the neutral or the ionized species were found for the three organic bases of interest (#-nitrosodi-n-propylamine, N-nitrosodiphenylamine, and p-chloroaniline). Generally, the sorption of ionizable organic bases has not been as well investigated as that of the organic acids, and there has been no relationship developed between the sorption coefficients of the neutral and ionized species. EPA is currently initiating research on models for predicting the sorption of organic bases in the subsurface. As noted earlier, the neutral species of the organic base predominates at pH values exceeding the pKa. For N-nitrosodi-n-propylamine (pKa 1) and N-nhrosodiphenylamine (pKa 0), the neutral , species is present under environmentally significant conditions. The neutral species constitutes approximately 90 percent of the system for p-chloroaniline (Table 40). 148 TUT 008 Table 41. Soil Organic Carbon/Water Partition Coefficients and pKa Values for Ionizing Organic Compounds Compound Benzoic acid 2-Chlorophenol 2,4-Dichlorophenol 2,4-Dinrtrophenol Pentachlorophenol 2,3,4,5-Tetrachlorophenol 2,3,4,6-Tetrachlorophenol 2 ,4, 5-Trichlorophenol 2,4,6-Trichlorophenol Koc,n <«-feS) 32b 398b 159d 0.8a 19,953" 17,916* 6,190' 2,380' 1 ,070' K.e.. (L/Kfl) 0.5C 6.0C 2.4C 0.01° 398* 678 93C 36J 107k pKaa 4.18 8.40 7.90 3.30 4.80 6.35h 5.30 7.10 6.40 « Kolligetal. (1993). t> Meylanetal. (1992). c Estimate based on the ratio of Kocjfl^c^ for compounds for which data exist; KOC,I was estimated to be 0.01 5 « Calculated using data (Kp = 0.62, f^ = 0.0039) contained in Lee et al. (1 991 ); agrees well with Boyd (1982) reporting measured KOC =126 L/kg. * Lee etal. (1990). 1 Average of values reported for two aquifer materials from Schellenberg et al. (1984). o Calculated using data (Kp = 0.26, foc = 0.0039) contained in Lee et al. (1991 ). * Lee etal. (1991). 1 Schellenberg et al. (1984). I Calculated using data (Kp = 0.1 4, foc = 0.0039) contained in Lee et al. (1991 ). * Kukowski (1989). The neutral species has a lower tendency to sorb to subsurface materials than the positively charged ionized species. As a consequence, the determination of overall sorption potential based solely on the neutral species for jV-mtrosodi-w-propylamine, A^-nitrosodiphenylamine, and p-chloroaniline is conservative, and these three organic bases will be treated as nonionizing organic compounds (see Section 5.3.1). Soil-Water Partition Coefficients for Ionizing Organic Compounds. Partition coefficients for the neutral and ionized species (K^cf and K^, respectively) and pKa values for nine ionizable organic compounds are provided in Table 41. These parameters can be used in Equation 74 to compute Koc values for organic acids at any given pH. KQC values for each of the ionizable compounds of interest are presented in Table 42 for pMs of 4.9, 6.8, and 8.0. Appendix L contains pH-specific KOC values for ionizable organics over this entire range. 5.4 Soil-Water Distribution Coefficients (Kd) for Inorganic Constituents As with organic chemicals, development of SSLs for inorganic chemicals (i.e., toxic metals) requires a soil-water partition coefficient (Kd) for each constituent. However, the simple relationship between soil organic carbon content and sorption observed for organic chemicals does not apply to inorganic constituents. The soil-water distribution coefficient (Kd) for metals and other inorganic compounds is affected by numerous geochemical parameters and processes, including pH; sorption to clays, organic 149 TUT 008 1286 matter, iron oxides, and other soil constituents; oxidation/reduction conditions; major ion chemistry: and the chemical form of the metal. The number of significant influencing parameters, their variability in the field, and differences in experimental methods result in as much as seven orders of magnitude variability in measured metal Kd values reported in the literature (Table 43). This variability makes it much more difficult to derive generic Kj values for metals than for orgarucs. Table 42. Predicted Soil Organic Carbon/Water Partition Coefficients (K0c,L/kg) as a Function of pH: Ionizing Organics Compound Benzoic acid 2-Chlorophenol 2,4-Dichlorophenol 2,4-Dinitrophenol Pentachlorophenol 2,3,4,5-Tetrachlorophenol 2,3,4,6-Tetrachlorophenol 2,4,5-Trichlorophenol 2,4,6-Trichlorophenol pH s 4.9 5.5 398 159 0.03 9,055 17,304 4.454 2,365 1,040 pH = 6.8 0.6 388 147 0.01 592 4,742 280 1,597 381 pH s 8.0 0.5 286 72 0.01 410 458 105 298 131 Because of their great variability and a limited number of data points, no meaningful estimate of central tendency Kd Values for metals could be derived from available measured values. For this reason, an equilibrium geochemical speciation model (MINTEQ) was selected as the best approach for estimating Kj values for the variety of environmental conditions expected to be present at Superfund sites. This approach and model were also used by OSW to estimate generic Kj values for metals proposed for use in the HWIR proposal (U.S. EPA, 1992a). The HWIR MINTEQA2 analyses were conducted under a variety of geochemical conditions and metal concentrations representative of solid waste landfills across the Nation. The metal K<t values developed for this effort were reviewed for SSL application and were used as preliminary values to develop the September 1993 draft SSLs. Upon further review of the HWIR MINTEQ modeling effort, EPA decided it was necessary to conduct a separate MINTEQ modeling effort to develop metal Kd values for SSL application Reasons for this decision include the following: It was necessary to expand the modeling effort to include other metal contaminants likely to be encountered at Superfund sites (i.e., beryllium, copper, and zinc). HWIR work incorporated low, medium, and high concentrations of dissolved organic acids that are present in municipal solid waste (MSW) leachate. These organic acids are not expected to exist in high concentrations in pore waters underlying Superfund sites; therefore, their inclusion in tile Superfund contaminated soil scenario is not warranted. The HWIR modeling simulations for chromium (+3) were found to be in error. This error has been corrected in subsequent HWIR modeling work but corrected results were not available at the time of preliminary SSL development. 150 TUT 12Q: Table 43. Summary of Collected Kd Values Reported in Literature Metal Antimony Arsenic8 Arsenic (+3) Arsenic (4-5) Barium Beryllium Cadmium Chromium Chromium (+2) Chromium (+3) Chromium (+6) Mercury* Nickel Selenium Silver Thallium Vanadium Zinc AECL (1990)* Range 45-550 - - - - 250-3,000 2.7-17.000 1.7-2,517 - -' - - 60-4,700 150-1,800 2.7-33,000 - - 0.1-100,000 Baes and Sharp (1983) or Baas et al. (19M)*> Geometric Mean* 451 200« 3.30 6.78 60* 650' 6.4* 850' 2.2000 - 370 10' 150' 300' 46" 1,500' 1,000' 38* Range - - 1.0-8.3 1.9-18 - - 1.26-26.8 - 470-150,000 - 1.2-1,800 - - — 10-1,000 ' - - 0.1-8,000 No. Values - ~ 19 37 - - 28 -r 15 - 18 - ~ - 16 - - 146 Coughtrey at al. (1985)c Range - - - - - - 32-50 . - - . - -20 <9 50 - _ S20 Battelle (1989)0 Range 2.0-15.9 5.86-19.4 - . • - 530-16,000 70-8,000 14.9-567 - - 168-3,600 16.8-360 322-5,280 12.2-650 5.9-14.9 0.4-40.0 0.0-0.8 50-100.0 ~ • ' . • The Atomic Energy of Canada, Limited (AECL, 1990) presents the distribution of K<j values according to four major soil types—sand, sift, clay, and organic material. Their data were obtained from available literature. b Baes et al. (1984) present K<j values for approximately 220 agricultural soils in the pH range of 4.5 to 9. Their data were derived from available literature and represent a diverse mixture of soils, extracting solutions, and laboratory techniques. c Coughtrey et al. (1985) report best estimates and ranges of measured soil K^ values for a limited number of metals. a Battelle Memorial Institute (Battelle, 1989) reports a range in revalues as a function of pH (5 to 9) and sorbeni content (a combination of clay, aluminum and iron oxyhydroxides, and organic matter content), the sorbent content ranges were <10 percent, 10 to 30 percent, and >30 percent sorbent. Their data were based on available literature. * The valence of these metals is not reported in the documents. ' Estimated based on the correlation between K<j and soil-to-plant concentration factor (Bv). o Average value reported by Baes and Sharp (1983). h Represents the median of the logarithms of the observed values. For these reasons, a MINTEQ modeling effort was expanded to develop a series of metal-specific isotherms for several of the metals expected to be present in soils underlying Superfund sites. The model used was an updated version of MINTEQA2 obtained from Allison Geoscience Consultants. Inc. Model results are reported in the December 1994 draft Technical Background Document (U.S. EPA, 19941) and were used to calculate the SSLs presented in the December 1994 draft Soil Screening Guidance (U.S. EPA, 1994h). 151 TUT OOS 1.288 The MINTEQA2 model was further updated by Allison Geoscience Consultants. Inc.. in 1995 to include thermodynanuc data for silver, an improved estimate of water saturation in the vadose zone (i.e.. water saturation is assumed to be 77.7 percent saturated as opposed to 100 percent), and revised estimates of sorbent mass (i.e.. organic matter content, iron oxide content). This updated model, which is expected to be made public through EPA's Environmental Research Laboratory in Athens, Georgia, was used to revise the generic Kj values for the EPA/OSW HWIR modeling effort. The metal Kj values for SSL application were also revised. Model results are contained in this document. The following section describes the important assumptions and limitations of this modeling effort. 5.4.1 Modeling Scope and Approach. New MINTEQA2 modeling runs were conducted to develop sorption isotherms for barium, beryllium, cadmium, chromium (+3), copper, mercury (+2), nickel, silver, and zinc. The general approach and input values used for pH, iron oxide (FeOx) concentration, and background chemistry were unchanged from the HWIR modeling effort. The HWIR MINTEQA2 analyses were conducted under a variety of geochemical conditions and metal concentrations. Three types of parameters were identified as part of the chemical speciation modeling effort. (1) parameters that have a direct first-order impact on metal speciation and are characterized by a wide range in environmental variability; (2) parameters that have an indirect, generally less pronounced effect on metal speciation and are characterized by a relatively small or insignificant environmental variability; and (3) parameters that may have a direct first-order impact on metal speciation but neither the natural variability nor its significance is known. In the HWIR modeling effort, parameters of the first type ("master variables") were limited to those having a significant effect on model results, including pH, concentration of available amorphous iron oxide adsorption sites (i.e., FeOx content), concentration of solid organic matter adsorption sites (with a dependent concentration of dissolved natural organic matter), and concentration of leachate organic acids expected to be present in MSW leachate. High, medium, and low values were assigned to each of the master variables to account for their natural environmental variability. The SSL modeling effort used this same approach and inputs except that anthropogenic organic acids were not included in the model simulations. Furthermore, the SSL modeling effort incorporated a medium fraction of organic carbon (foc) that correlated to the HWIR high concentration. Parameters of the second type constitute the background pore-water chemistry, which consists of chemical constituents commonly occurring in ground water at concentrations great enough to affect metal speciation. These constituents were treated as constants in both the SSL and HWIR effort The third type of parameter was entirely omitted from consideration in both modeling efforts due to poorly understood geochemistry and the lack of reliable thermodynamic data. The most important of there parameters is the oxidation-reduction (redox) potential. To compensate, both modeling effort: incorporated an approach that was most protective of the environment with respect to the impact of redox potential on the partitioning of redox-sensitive metals (i.e., each metal was modeled in the oxidation state that most enhances metal mobility). For the HWIR modeling effort, metal concentrations were varied from the maximum contaminant level (MCL) to 1,000 times the MCL for each individual metal. This same approach was taken for SSL modeling, although for certain metals the concentration range was extended to determine the metal concentration at which the sorption isotherm departed from linearity. a Sorption isotherms for arsenic (+3), chromium (+6), selenium, and thallium are unchanged from the previous efforts and are based on laboratory-derived pH-dependent sorption relationships developed 152 TUT 008 1289 for HWIR. Using these relationships, the Kj distribution as a function of pH is presented for each of these four metals in Figure 10. Sorption isotherms for antimony and vanadium could not be estimated using MINTEQA2 because the thermodynamic databases do not contain the required reactions and associated equilibrium constants. Sufficient experimental research has not been conducted to develop pH-dependent relationships for these two metals. As a consequence, K<j values for antimony and vanadium were obtained from Baes et al. (1984) (Table 43). These Kj values are not pH-dependent. 5.4.2 Input Parameters. Table 44 lists high, medium, and low values for pH and iron oxide used for both the HWIR and SSL MINTEQ modeling efforts. Sources for these values are as follows (U.S. EPA, 1992a): • Values for pH were obtained from analysis of 24,921 field-measured pH values contained in the EPA STORET database. The pH values of 4.9, 6.8, and 8.0 correspond to the 7.5th, 50th, and 92.5th percentiles of the distribution. • Iron oxide contents were based on analysis of six aquifer samples collected over a wide geographic area, including Florida, New Jersey, Oregon, Texas, Utah, and Wisconsin. The lowest of the six analyses was taken to be the low value, the average of the six was used as the medium value, and the highest was taken as the high value. The development of the values presented in Table 44 is described in more detail in U.S. EPA (1992a). Thirteen chemical constituents commonly occurring in ground water were used to define the background pore-water chemistry for HWIR and SSL modeling efforts (Table 45). Because these constituents were treated as constants, a single total ion concentration, corresponding to the median total metal concentration from a probability distribution obtained from the STORET database, was assigned to each of the background pore-water constituents (U.S. EPA, 1992a). Although the HWIR and the SSL MINTEQ modeling efforts were consistent in the majority of the assumptions and input parameters used, the fraction of organic carbon (foc) used for the SSL modeling effort was slightly different than that used for the HWIR modeling effort. The foc used for the SSL effort was equal to 0.002 g/g, which better reflected average subsurface conditions at Superfund sites. This value is approximately equal to the high value of organic carbon used in the HWIR modeling effort. Table 44. Summary of Geochemical Parameters Used in SSL MINTEQ Modeling Effort Value Low Medium High PH 4.9 6.8 8.0 Iron oxide content (weight percent) 0.01 0.31 1.11 Source: U.S. EPA (1992a) 153 TUT 008 1290 Figure 10. Empirical pH-dependent adsorption relationship: araanic (+3), chromium (+6), salanium, thallium 154 TUT 008 i? 91 Table 45. Background Pore-Water Chemistry Assumed for SSL MINTED Modeling Effort* Parameter Aluminum Bromine Calcium Carbonate Chlorine Iron (43) Magnesium Manganese (+2) Nitrate Phosphate Potassium Sodium Sulfate Concentration (mg/L) 0.2 0.3 48 187 15 0.2 14 0.04 1 0.09 2.9*> 22 25 • Median values from STORE! database as reported in U.S. EPA (1992a). » Median values from STORE! database; personal communication from J. Allison, Allison Geosciences. 5.4.3. Assumptions and Limitations. The SSL MINTEQ modeling effort incorporates several basic simplifying assumptions. In addition, the applicability and accuracy of the model results are subject to limitations. Some, of the more significant assumptions and limitations are described below. The system is assumed to be at equilibrium. This assumption is inherent in geochemical aqueous speciation models because the fundamental equations of mass action and mass balance are equilibrium based. Therefore, any possible influence of adsorption (or desorption) rate limits is not considered. This assumption is conservative. Because the model is being used to simulate metal desorption from the solid substrate, if equilibrium conditions are not met, the desorption reaction will be incomplete and the metal concentration in pore water will be less than predicted by the model. Redoz potential is not considered. The redox potential of the system is not considered due to the difficulty in obtaining reliable field measurements of oxidation reduction potential (Eh), which are needed to determine a realistic frequency distribution of this parameter. Furthermore, the geochemistry of redox-sensitive species is poorly understood. Reactions involving redox species are often biologically mediated and the concentrations of redox species are not as likely to reflect thennodynamic equilibrium as other inorganic constituents. To provide a conservative estimate of metal mobility, all environmentally viable oxidation states are modeled separately for the redox-sensitive metals; the most conservative was selected for defining SSL metal Kj values. The redox-sensitive 155 TUT OO8 1292 constituents that make up the background chemistry are represented only by the oxidation state that most enhances metal mobility (U.S. EPA, 1992a). Potential sorbent surfaces are limited. Only metal adsorption to FeOx and solid organic matter is considered in the system. It is recognized that numerous other natural sorbents exist (e.g., clay and carbonate minerals); however, thermodynamic databases describing metal adsorption to these surfaces are not available and the potential for adsorption to such surfaces is not considered. This assumption is conservative and will underpredict sorption for soils with significant amounts of such sorption sites. The available thermodynamic database is limiting. As metal behavior increases in complexity, thermodynamic data become more rare. The lack of complete thermodynamic data requires simplification to the defined system. This simplification may be conservative or nonconservative in terms of metal mobility. • Metal competition is not considered. Model simulations were performed for systems comprised of only one metal (i.e., the potential for competition between multiple metals for available sorbent surface sites was not considered). Generally, the competition of multiple metals for available sorption sites results in higher dissolved metal concentrations than would sxist in the absence of competition. Consequently, this assumption is nonconservative but is significant only at metal concentrations much higher than the SSLs. Other assumptions and limitations associated with this modeling effort are discussed in RTI (1994). 5.4.4 Results and DiSCUSSion. MINTEQ model results indicate that metal mobility is most affected by changes in pH. Based on this observation and because iron oxide content is not routinely measured in site characterization efforts, pH-dependent Kjs for metals were developed for SSL application by fixing iron oxide at its medium value and fraction organic carbon at 0.002. For arsenic (+3), chromium (+6), selenium, and thallium, the empirical pH-dependent K<jS were used. Table 46 shows the SSL.Kd values at high, medium, and low subsurface pH conditions. Figure 11 plots MINTEQ-denved metal Kd values over this pH range. Figure 10 shows the same for the empirically derived meuil KdS. These results are discussed below by metal and compared with measured values. See RTI (1994) for more information. pH-dependent values are not available for antimony, cyanide, and vanadium. The estimated Kd values shown in Table 46 for antimony and vanadium are reported by Baes et al. (1984) and the Kd value for cyanide is obtained from SCDM. Arsenic. Kd values developed using the empirical equation for arsenic (+3) range from 25 to 31 L/kg for pH values of 4.9 to 8.0, respectively. These values correlate fairly well with the range of measured values reported by Battelle (1989)—5.86 to 19.4 L/kg. They are slightly above the range reported by Baes and Sharp (1983) for arsenic (+3) (1.0-8.3). The estimated Kd values for arsenic (+3) do not correlate well with the value of 200 L/kg presented by Baes et al. (1984). Oxidation state is not specified in Baes et al. (1984), and the difference between the empirical-derived Kd values presented here and the value presented by Baes et al. (1984) may reflect differences in oxidation states (arsenic (+3) is the most mobile species). 156 TUT OOS 1293 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 Note: Conditions depicted are medium iron oxide content (0.31 wt %) and organic matter of 0.2 wt %. Figure 11. Metal Kd as a function of pH. 157 TUT OOS 1294 Table 46. Estimated Inorganic Kd Values for SSL Application Metal Antimony8 Arsenic (+3)b Barium Beryllium Cadmium Chromium (+3) Chromium (+6)b Cyanide" Mercury (+2) Nickel Seleniumb Stiver Thallium^ Vanadium** Zinc pH = 4.9 2.5E+01 1.1E+01 2.3E+01 1.5E+01 1 .2E+03 3.1E+01 4.0E-02 1 .6E+01 1 .8E+01 1 .OE-01 4.4E+01 1 .6E+01 Estimated Kd (L/kg) pH = 6.8 4.5E+01 2.9E+01 4.1E+01 7.9E+02 7.5E+01 1.8E+06 1.9E+01 9.9E+00 5.2E+01 6.5E-I-01 5.0E+00 8.3E+00 7.1E+01 1.0E+03 6.2E+01 pH = 8.0 3.1E+01 5.2E+01 1.0E+05 4.3E+03 4.3E+06 1 .4E+01 2.0E+02 1 .9E+03 2.2E+00 1.1E+02 9.6E+01 5.3E+02 a Geometric mean measured value from Baes el at., 1984 (pH-dependent values not available). b Determined using an empirical pH-dependent relationship (Figure 10). c SCDM = Superfund Chemical Data Matrix (pH-dependent values not available). Barium. For ground water pH conditions, MINTEQ-estimated Kd values for barium range from 11 to 52 L/kg. This range correlates well with the value of 60 L/kg reported by Baes et al. (1984). Battelle (1989) reports a range in Kd values from 530 to 16,000 L/kg for a pH range of 5 to 9. The model-predicted Kd values for barium are several orders of magnitude less than the measured values, possibly due to the lower sorptive potential of iron oxide, used as the modeled sorbent, relative to clay, a sorbent present in the experimental systems reported by Battelle (1989). Beryllium. The Kd values estimated for beryllium range from 23 to 100,000 L/kg for the conditions studied. AECL (1990) reports medians of observed values for Kd ranging from 250 L/kg for sand to 3,000 L/kg for organic matter. Baes et al. (1984) report a value of 650 L/kg. Battelle (1989) reports a range of Kd values from 70 L/kg for sand to 8,000 L/kg for clay. MINTEQ results for medium ground water pH (i.e., a value of 6.8) yields a Kd value of 790 L/kg. Hence, there is reasonable agreement between the MINTEQ-predicted K<j values and values reported in the literature. Cadmium. For the three pH conditions, MINTEQ Kd values for cadmium range from 15 to 4,300 L/kg, with a value of 75 at a pH of 6.8. The range in experimentally determined Kd values for cadmium is as follows: 1.26 to 26.8 L/kg (Baes et al., 1983), 32 to 50 L/kg (Coughtrey et al., 1985), 14.9 to 567 L/kg (Battelle, 1989), and 2.7 to 17,000 L/kg (AECL, 1990). Thus the MINTEQ estimates are generally within the range of measured values. Chromium (+3). MINTEQ-estimated Kd values for chromium (+3) range from 1,200 to 4,300,000 L/kg. Battelle (1989) reports a range of Kj values of 168 to 3,600 L/kg, orders of 158 TUT 008 1295 magnitude lower than the MINTEQ values. This difference may reflect the measurements of mixed systems comprised of both chromium (+3) and (+6). The incorporation of chromium (+6) would tend to lower the Kd. Because the model-predicted values may overpredict sorption, the user should exercise care in the use of these values. Values for chromium (+6) should be used where speciation is mixed or uncertain. Chromium (+6). Chromium (+6) Kd values estimated using the empirical pH-dependent adsorption relationship range from 31 to 14 L/kg for pH values of 4.9 to 8.0. Battelle (1989) reports a range of 16.8 to 360 L/kg for chromium (+6) and Baes and Sharp (1983) report a range of 1.2 to 1,800. The predicted chromium (+6) Kd values thus generally agree with the lower end of the range of measured values and the average measured values (37) reported by Baes and Sharp (1983). These values represent conservative estimates of mobility the more toxic of the chromium species. Mercury (+2). MINTEQ-estimated Kd values for mercury (+2) range from 0.04 to 200 L/kg. These model-predicted estimates are less than the measured range of 322 to 5,280 L/kg reported by Battelle (1989). This difference may reflect the limited thermodynamic database with respect to mercury and/or that only the divalent oxidation state is considered in the simulation. Allison (1993) reviewed the model results in comparison to the measured values reported by Battelle (1989) and found reasonable agreement between the two sets of data, given the uncertainty associated with laboratory measurements and model precision. Nickel. MINTEQ-estimated Kd values for nickel range from 16 to 1,900 L/kg. These values agree well with measured values of approximately 20 L/kg (mean) and 12.2 to 650 L/kg, reported by. Coughtrey et al. (1985) and Battelle (1989), respectively. These values also agree well with the value of 150 L/kg reported by Baes et al. (1984). However, the predicted values are at the low end of the range reported by the AECL (1990)—60 to 4,700 L/kg. Selenium. Empirically derived Kd values for selenium range from 2.2 to 18 L/kg for pH values of 8.0 to 4.9. The range in experimentally determined Kd values for selenium is as follows: less than 9 L/kg (Coughtrey et al., 1985), 5.9 to 14.9 L/kg (Battelle, 1989), and 150 to 1,800 L/kg (AECL, 1990). Baes et al. (1984) reported a value of 300 L/kg. Although they are significantly below the values presented by the AECL (1990) and Baes et al. (1984), the MINTEQ-predicted Kd values correlate well with the values reported by Coughtrey et al. (1985) and Battelle (1989). Silver. The Kd values estimated for silver range from 0.10 to 110 L/kg for the conditions studied The range in experimentally determined Kd values for silver is as follows: 2.7 to 33,000 L/kg (AECL. 1990), 10 to 1,000 L/kg (Baes et al., 1984), 50 L/kg (Coughtrey et al., 1985), and 0.4 to 40 L/kg (Battelle, 1989). The model-predicted Kd values agree well with the values reported by Coughtrey et al. (1985) and Battelle (1989) but are at the lower end of the ranges reported by AECL (1990) and Baes etal. (1984) Thallium. Empirically derived Kd values for thallium range from 44 to 96 L/kg for pH values of 4.9 to 8.0. Generally, these values are about an order of magnitude greater than those reported by Battelle (1989)—0.0 to 0.8 L/kg - but are well below the value predicted by Baes et al. (1984). Zinc. MINTEQ-estimated Kd values for zinc range from 16 to 530 L/kg. These estimated Kd values are within the range of measured Kd values reported by the AECL (1990) (0.1 to 100,000 L/kg) and Baes et al. (1984) (0.1 to 8,000 L/kg). Coughtrey et al. (1985) reported a Kd value for zinc of greater than or equal to 20 L/kg. ' 159 TUT OOS 5.4.5 Analysis Of Peer-Review Comments. A peer review was conducted of the model assumptions and inputs used to estimate Kd values for SSL application. This review identified several issues of concern, including: The charge balance exceeds an acceptable margin of difference (5 percent) in most of the simulations. A variance in excess of 5 percent may indicate that the model problem is not correctly chemically poised and therefore the results may not be chemically meaningful. • The model should not allow sulfate to adsorb to the iron oxide. Sulfate is a weakly outer-sphere adsorbing species and, by including the adsorption reaction, sulfate is removed from the aqueous'phase at pH values less than 7 and is prevented from participating in precipitation reaction at these pH values. Modeled Kd values for barium and zinc could not be reproduced for all studied conditions. A technical analysis of these concerns indicated that, although these comments were based on true observations about the model results, these factors do not compromise the validity of the MINTEQ results in this application. This technical analysis is provided in Appendix M. 160 TUT 008 1297 Part 6: REFERENCES Abdul, A.S., T.L. Gibson, and D.N. Rai. 1987. Statistical correlations for predicting the partition coefficient for nonpolar organic contaminants between aquifer carbon and water. Hazardous Waste and Hazardous Materials 4(3):211-223. AECL (Atomic Energy of Canada Limited). 1990. A Critical Compilation and Review of Default Soil Solid/Liquid Partition Coefficients, K* for Use in Environmental Assessments, by D.H. Thibault, M.I. Sheppard, and P.A. Smith, Whiteshell Nuclear Research Establishment, Pinawa, Manitoba. Aller, L., T. Bennett, J.H. Lehr, R.J. Petty, and G. Hackett. 1987. DRASTIC: A Standardized System for Evaluating Ground Water Pollution Potential Using Hydrogeologic Settings. EPA-600/2- 87-035. Prepared for U.S. EPA Office.of Research and Development, Ada, OK. National Water Well Association, Dublin, OH. Allison, J. 1993. Review and Evaluation of the MINTEQA2 Kd Results for Mercury in the Finite Source Methodology. Prepared by Allison Geoscience Consultants, Inc., for the U.S. Environmental Protection Agency, Environmental Research Laboratory, Athens, GA. ASTM (American Society for Testing and Materials). 1994. Risk-Based Corrective Action Guidance. Philadelphia, PA. (Draft). Baes, C.F. (Ill), and R.D. Sharp. 1983. A proposal for estimation of soil leaching and leaching constants for use in assessment models. J. Environ. Qual. 12(l):17-28. Baes, C.F. (Ill), R.D. Sharp, A.L. Sjoreen, and R.W. Shor. 1984. A Review and Analysis of Parameters for Assessing Transport of Environmentally Released Radionuclides through Agriculture. Oak Ridge National Laboratory Report ORNL-5786. ORNL, Oak Ridge, TN. Battelle Memorial Institute. 1989. Chemical Data Bases for the Multimedia Environmental Pollutant Assessment System (MEPAS): Version 1. Prepared for the U.S. Department of Energy under Contract DE-AC06-76RKI 1830, by Pacific Northwest Laboratory, operated by Battelle Memorial Institute. Boyd, S.A. 1982. Adsorption of substituted phenols by soil. Soil Science, 134(S):337-343. Brady, N.C. 1990. The Nature and Properties of Soils. Macmillan Publishing Company, New York, NY. . Calabrese, E.J., H. Pastides, R. Barnes, et al. 1989. How much soil do young children ingest: an epidemiologic study. In: Petroleum Contaminated Soils, Vol. 2. E.J. Calabrese and P.T. Kostecki, eds. pp. 363-417. Chelsea, MI, Lewis Publishers. Carsel, R.F., and R.S. Panish. 1988. Developing joint probability distributions of soil water retention characteristics. Water Resources Research, 24(5):755-769. 161 TUT OO8 1298 Carsel, R.F., R.S. Fairish, R.L. Jones, J.L. Hansen, and R.L. Lamb. 1988. Characterizing the uncertainty of pesticide leaching in agricultural soils. J. ofContam. Hyd. 2:111-124. Chen, L. 1995. Testing the mean of skewed distributions. Journal of the American Statistical Association 90(430):767-772. Coughtry, P.J.. D. Jackson, and M.C. Thome. 1985. Radionuclide Distribution and Transporting Terrestrial and Aquatic Ecosystems: A Compendium of Data. A A. Balkem. Netherlands. Vol. 6. Cowherd, C., G. Muleski, P. Engelhart, and D. Gillette. 1985. Rapid Assessment of Exposure to Particulate Emissions from Surface Contamination. EPA/600/8-85/002. NTIS PB85- 192219. Office of Health and Environmental Assessment, U.S. Environmental Protection Agency, Washington, DC. Cnscenti, L.J., S.B. Christian, E.Z. Hosseinipour, and V.M. Gorokhovski. 1994. Evaluation of Unsaturated Zone/Vadose Zone Models for Super/and Sites. Draft Report. Environmental Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Athens, GA. Curtis, G. P., P. V. Roberts, and M. Reinhard. 1986. A natural gradient experiment on solute transport in a sand aquifer. Water Resources Research 22(13):2059-2067. Davis, S., P. Waller, R. Buschom, J. Ballou, and P. White. 1990. Quantitative estimates of soil ingestion in normal children between the ages of 2 and 7 years: population-based estimates using Al, Si, and Ti as soil tracer elements. Archives of Environmental Health 45:112-122. Di Toro, D.M. 1985. A particle interaction model of reversible organic chemical sorption. Chemosphere 14(10): 1503-1538. Dragun, J. 1988. The Soil Chemistry of Hazardous Materials. Hazardous Materials Control Research Institute, Silver Spring, MD. EQ (Environmental Quality Management). 1992. Limited Validation of the H\vang and Falco Model for Emissions of Soil-Incorporated Volatile Organic Compounds. Contract No. 68-D8-0111. Prepared for Office of Emergency and Remedial Response, U.S. Environmental Protection Agency. Washington, DC. EQ (Environmental Quality Management). 1993. Evaluation of the Dispersion Equations in the Risk Assessment Guidance for Superfund (RAGS): Volume 1—Human Health Evaluation Manual (Part B, Development of Preliminary Remediation Goals). Contract No. 68-02-D120. Prepared for Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. EQ (Environmental Quality Management). 1994. Revisions to VF and PEF Equations. Memorandum #5099-3. Prepared for Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. EQ (Environmental Quality Management) and Pechan (E. H. Pechan and Associates). 1994. A Comparison of Soil Volatilization Models in Support of Superfund Soil Screening Level Development. Contract No. 68-D3-0035. Prepared for Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. 162 TUT 008 1299 EQ (Environmental Quality' Management). 1995. Limited Validation of the Jury Infinite Source and Jury Reduced Solution Finite Source Models for Emissions of Soil-Incorporated Volatile Organic Compounds. Prepared for Office of Solid Waste and Emergency Response, U.S. Environmental Protection Agency, Washington. DC. Farmer, W. J., and J. Letey. 1974. Volatilization Losses of Pesticides from Soils. EPA-660/2-74/054. Office of Research and Development, U.S. Environmental Protection Agency. Washington. DC. Feenstra. S., D.M. MacKay, and J.A. Cherry. 1991. A method for assessing residual NAPL based on organic chemical concentrations in soil samples. Ground Water Monitoring Rev. 11(2): 128- 136. Gee, G.W., and J.W. Bauder. 1986. Particle size analysis. In: A. Klute (ed.). Methods of Soil Analysis. Part 1. Physical and Mineralogical Methods. 2nd Edition, 9(1):383-411, American Society' of Agronomy, Madison, WI. Gelhar, L.W., and C.J. Axness. 1981. Stochastic Analysis of Macro-Dispersion in Three- Dimensionally Heterogeneous Aquifers. Report No. H-8. Hydraulic Research Program. New Mexico Institute of Mining and Technology, Socorro, NM. Gilbert, Richard O. 1987. Statistical Methods for Environmental Pollution Monitoring. New York: Van Nostrand Reinhold. Hardin, J.W., and R.O. Gilbert. 1993. Comparing Statistical Tests for Detecting Soil Contamination Greater than Background. Pacific Northwest Laboratory (PNL-8989), Richard, WA. Heath, R.C. 1984. Ground-Water Regions of the United States. USGS Water Supply Paper 2242. U.S. Geological Survey, Reston, VA. Howard, P. H. 1989. Handbook of Environmental Fate and Exposure Data for Organic Chemicals. Vol. I. Large Production and Priority Pollutants. Lewis Publishers, Chelsea. MI. Hwang, S. T., and J. W. Falco. 1986. Estimation of Multimedia Exposure Related to Hazardous Waste Facilities. Y. Cohen, Ed. Plenum Publishing. Johnson, N.L 1978. Modified t tests and confidence intervals for asymmetrical populations. Journal of the American Statistical Association 73(363):536-544. Johnson, P.C. and R.A. Ettinger. 1991. Heuristic model for predicting the intrusion rate of contaminant vapors into buildings. Environmental Science and Technology 25(8): 1445- 1452. Jury, W.A., W.F. Spencer, and W.J. Farmer. 1983. Behavior assessment model for trace organics in soil: I. Model description. J. Environ. Qual. 12(4):558-564. Jury, W.A., W.J. Fanner, and W.F. Spencer. 1984. Behavior assessment model for trace organics in soil: II. Chemical classification and parameter sensitivity. J. Environ. Qual. 13(4):567-572. Jury, W.A., D. Russo, G. Streile, and H.E. Abd. 1990. Evaluation of volatilization by organic chemicals residing below the soil surface. Water Resources Research 26(1): 13-20. Karickhoff, S.W. 1984. Organic pollutant sorption in aquatic systems. J. Hydraul. Eng. 110(6):707- 735. . . 163 TUT OO8 1300 Kollig, H.P. 1993. Environmental Fate Constants for Organic Chemicals under Consideration for EPA's Hazardous Waste Identification Projects. Office of Research and Development U.S. Environmental Protection Agency, Washington, DC. Kukowski, H. 1989. Adsorption and Desorption of lonizable Organic Chemicals by Soils. Ph.D. Dissertation. University of Kiel. Land, C.E. 1971. Confidence intervals for linear functions of the normal mean and variance. Annals of Mathematical Statistics 42:1187-1205. Land, C.E. 1975. Tables of confidence limits for linear functions of the normal mean and variance. Selected Tables in Mathematical Statistics, 3:385-419. American Mathematical Society, Providence, RI. . - Lee, L. S., P. S. C. Rao, P. Nkedi-Kizza, and J. J. Delfino. 1990. Influence of solvent and sorbent characteristics on distribution of pentachlorophenol in octanol-water and soil-water systems. Environ. Sci. Tech. 24:654-661. Lee, L. S., P. S C. Rao, and M. L. Brusseau 1991. Nonequilibrium sorption and transport of neutral and ionized compounds. Environ. Sci. Tech. 25:722-729. Lyman, W. J., W. F. Reehl, and D. H. Rosenblatt. 1982. Handbook of Chemical Property Estimation Methods. McGraw-Hill, New York. Marshall, J. 1971. Drag measurements in roughness arrays of varying density and distribution Agricultural Meteorology, 8:269-292. McCarty. P.L., M. Reinhard, and 6.E. Rittmann. 1981. Trace organics in ground water. Environ Sci. Technol. 15:40-51. McLean, E.O. 1982. Soil pH and lime requirement. In: A.L. Page (ed.), Methods of Soil Analysis Part 2. Chemical and Microbiological Properties. 2nd Edition, 9(2): 199-224, American Society of Agronomy. Madison, WI. Meylan, W., P.H. Howard, and R.S. Boethling. 1992. Molecular topology/fragment contribution method for predicting soil sorption coefficients. Environ. Sci. Technol. 26(8): 1560-1567 Miller, R.G. 1991. Simultaneous Statistical Inference. 2nd edition. Springer-Verlag, New York. Millington, R. J., and J. M. Quirk. 1961. Permeability of porous soils. Trans. Faraday Soc. 57:1200- 1207. Nelson, D.W., and L.E. Sommers. 1982. Total carbon, organic carbon, and organic matter. In A.L Page (ed.), Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties 2nd Edition, 9(2):539-579, American Society of Agronomy, Madison, WI. Newell, C.J., L.P. Hopkins, and P.B. Bedient. 1989. Hydrogeologic Database for Ground Water Modeling. API Publication No. 4476. American Petroleum Institute, Washington, DC. Newell. .'.J., L.P. Hopkins, and P.B. Bedient. 1990. A hydrogeologic database for ground water- niodeling. Ground Water, 28(5):703-714. 164 TUT 008 j.301 Nofziger, D.L., J. Chen, and C.T. Haan 1994. Evaluation of UnsaturatedWadose Zone Models for Superfund Sites. EPA/600/R-93/184. Office of Research and Development, U.S. Environmental Protection Agency. Robert S. Kerr Environmental Research Laboratory Ada, OK • _ . Piwoni. M. D., and P. Banerjee. 1989. Sorption of volatile organic solvents from aqueous solution onto subsurface solids. J. Contam. Hydro!. 4(2): 163-179. Radian. 1989. Short-term Fate and Persistence of Motor Fuels in Soils. Report to the American Petroleum Institute, Washington, DC. RTI (Research Triangle Institute). 1994*. Chemical Properties for Soil Screening Levels. Draft Report. Prepared for Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, DC. Salhotra, A.M., P. Mineart, S. Sharp-Hansen, and T. Allison. 1990. Multimedia Exposure Assessment Model (MULTIMED) for Evaluating the Land Disposal of Wastes—Model Theory. Environmental Research Laboratory, U.S. Environmental Protection Agency, Athens. GA. Schellenberg K., C. Leuenberger, and R.P. Schwarzenbach. 1984. Sorption of chlorinated phenols by natural sediments and aquifer materials. Environ. Sci. Technol. 18(9):652-657. Schwarzenbach, R. P., and J. C. Westall. 1981. Transport of non-polar organic compounds from surface water to ground water. Environ. Sci. Tech. 15(11): 1360-1367. Shan, C., and D.B. Stephens. 1995. An analytical solution for vertical transport of volatile chemicals in'the vadose zone. Journal of Contaminant Hydrology 18:259-277. Sharp-Hansen, S., C. Travers, P. Hummel, and T. Allison. 1990. A Subtitle D Landfill Application Manual for the Multimedia Exposure Assessment Model (MULTIMED). EPA Contract No 68-03-3513. Environmental Research Laboratory, Office of Research and Development. U.S. Environmental Protection Agency, Athens, GA. Shimizu, Y., N. Takei, S. Yamakazi, and Y. Terashima. 1993. Sorption of Organic Pollutants onto Natural Solids: lonizable Organics in a Saturated System and Volatile Organics in an Unsaturated System. Selected Papers on Environmental Hydrogeology, 29th International Geologic Congress, Kyoto, Japan, Volume 4, August 24-September 3, 1993. U.S. EPA (Environmental Protection Agency). 1980. Land Disposal ofHexachloroberaene Wastes .Controlling Vapor Movement in Soil. EPA-600/2-80-119. Office of Research and Development, Cincinnati, OH. NTIS PB80-216575. U.S. EPA (Environmental Protection Agency). 1988. Superfund Exposure Assessment Manual OSWER Directive 9285.5-1. EPA/540/1-88/001. Office of Emergency and Remedial Response, Washington, DC. NTIS PB89-135859 U.S. EPA (Environmental Protection Agency). 1989a. Methods for Evaluating the Attainment of Soil Cleanup Standards. Volume 1: Soils and Solid Media. EPA 230/02-89-042. Statistical Policy Branch, Office of Policy, Planning and Evaluation, Washington, DC. U.S. EPA (Environmental Protection Agency). 1989b. Risk Assessment Guidance for Superfund. Volume 1: Human Health Evaluation Manual, Part A, Interim Final EPA/540/1-89/002, Publication 9285.7-01A. Office of Emergency and Remedial Response, Washington, DC. NTIS PB90-155581/CCE. 165 TUT COS 13O2 U.S. EPA (Environmental Protection Agency). 1990. Guidance on Remedial Actions for Superfund Sites with PCB Contamination. Office of Solid Waste and Emergency Response. Washington. DC. NTIS PB91-1921206CDH. U.S. EPA (Environmental Protection Agency). 1990. Sampler's Guide to the Contract Laboratory Program. Office of Emergency and Remedial Response. Washington. DC. NTIS-PB91- 921330CDH. U.S. EPA (Environmental Protection Agency). 1991a. Human Health Evaluation Manual Supplemental Guidance: Standard Default Exposure Factors. Publication 9285.6-03. Office of Emergency and Remedial Response, Washington, DC. NTIS PB91-921314. U.S. EPA (Environmental Protection Agency). 1991b. Risk Assessment Guidance for Superfund, Volume I: Human Health Evaluation Manual (Part B, Development of Risk-Based Preliminary Remediation Goals). Interim. Publication 9285.7-01B. Office of Emergency and Remedial Response, Washington, DC. NTIS PB92-963333. U.S EPA (Environmental Protection Agency). 1991c. Role of the Baseline Risk Assessment in Superfund Remedy Selection Decisions. Publication 935S.O-30. Office of Emergency and Remedial Response, Washington, DC. NTIS PB91-921359/CCE. U.S. EPA (Environmental Protection Agency). 1991d. User's Guide to the Contract Laboratory Program. Office of Emergency and Remedial Response, Washington, DC. NTIS-PB91"- 921278CDH. U.S. EPA (Environmental Protection Agency). 1992a. Background Document for Finite Source Methodology for Wastes Containing Metals. HWEP-S0040. Office of Solid Waste, Washington, DC. U.S. EPA (Environmental Protection Agency). 1992b. Dermal Exposure Assessment: Principles and Applications. Interim Report. EPA/600/8-91/OHB. Office of Research and Development, Cincinnati, OH. U.S. EPA (Environmental Protection Agency). 1992c. Estimating the Potential for Occurrence of DNAPL at Superfund Sites. Pub. 9355.4-07FS. Office of Solid Waste and Emergency Response, Washington, DC. R.S. Kerr Environmental Research Laboratory, Ada, OK NTIS - PB92-963358. U.S. EPA (Environmental Protection Agency). 1992d. Technical Support Document for Land Application of Sewage Sludge. Volumes 1 and 2. EPA 822/R-93-001A,B. Office of Water. Washington, DC. U.S. EPA (Environmental Protection Agency). 1992e. Handbook of RCRA Ground-Water Monitoring Constituents: Chemical and Physical Properties. EPA/530-R-92-022: Office of Solid Waste, Washington, DC. NTIS PB92-233287. U.S. EPA. (Environmental Protection Agency). 1992f. Preparation of Soil Sampling Protocols: Sampling Techniques and Strategies. Office of Research and Development, Washington, DC EPA/600/R-92/128. U.S. EPA (Environmental Protection Agency). 1993a. Background Document for EPA's Composite Model for Leachate Migration with Transformation Products, EPACMTP. Office of Solid Waste, Washington, DC. July. . 166 TUT OO8 1303 U.S. EPA (Environmental Protection Agency). 1993b. Data Quality Objectives for Superfiind: Interim Final Guidance. EPA 540-R-93-071. Publication 9255.9-01. Office of Emergency and Remedial Response, Washington, DC. NTIS PB94-963203. U.S. EPA (Environmental Protection Agency). 1993c. Provisional Guidance for the Quantitative Risk Assessment ofPolycyclic Aromatic Hydrocarbons. Office of Health Effects Assessment. Washington, DC. U.S. EPA (Environmental Protection Agency). 1993d. Quality Assurance for Superfund Environmental Data Collection Activities. Quick Reference Fact Sheet. Office of Emergency and Remedial Response, Washington, DC. NITS PB93-963273. U.S. EPA (Environmental Protection Agency). 1993e. Science Advisory Board Review of the Office of Solid Waste and Emergency Response draft Risk Assessment Guidance for Superfiind (RAGS), Human Health Evaluation Manual (VOIEM). EPA-SAB-EHC-93-007. Science Advisory' Board, Washington, DC. U.S. EPA (Environmental Protection Agency). 1993f. The Urban Soil Lead Abatement Demonstration Project. Vol 1: Integrated Report Review Draft. EPA-600/AP93001/A. NTIS PB93-222-651. U.S. EPA (Environmental Protection Agency). 1994. Revised Interim Soil Lead Guidance for CERCLA Sites and RCRA Corrective Action Facilities. Office of Solid Waste and Emergency Response, Washington, DC. Directive 9355.4-12. U.S. EPA (Environmental Protection Agency). 1994a. Frame-work for Assessing Ground Water Modeling Applications. EPA-500-B-94-004. Resource Management and Information Staff, Office of Solid Waste and Emergency Response, Washington, DC. U.S. EPA (Environmental Protection Agency). 1994b. Ground Water Modeling Compendium, Second Edition. EPA-500-B-94-003. Resource Management and Information Staff, Office of Solid Waste and Emergency Response, Washington, DC. U.S. EPA (Environmental Protection Agency). 1994c. Guidance for the Data Quality Objectives Process. EPA QA/G-4. Quality Assurance Management Staff, Office of Research and Development, Washington, DC. . . . U.S. EPA (Environmental Protection Agency). 1994d. Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry. EPA/600/8-90/066F. Office of Research and Development, Washington, DC. U.S. EPA (Environmental Protection Agency). 1994e. Methods for Evaluating the Attainment of Cleanup Standards—Volume 3: Reference-Based Standards for Soils and Solid Media. EPA 230-R-94-004. Environmental Statistics and Information Division, Office of Policy, Planning^ and Evaluation, Washington, DC. U.S. EPA (Environmental Protection Agency). 1994f. Risk Assessment Issue Paper for: Derivation of a Provisional RfD for 1,1-Dichloroethane (75-34-3). Environmental Criteria Assessment Office, Cincinnati, OH. U.S. EPA (Environmental Protection Agency). 1994g. Risk Assessment Issue Paper for: Provisional Oral RfD for Naphthalene (91-20-3). Environmental Criteria Assessment Office, Cincinnati, OH 167 TUT 008 j.304 U.S. EPA (Environmental Protection Agency). 1994h. Soil Screening Level Guidance. Quick Reference Fact Sheet. Draft. Office of Emergency and Remedial Response Washington. DC. NTIS PB93-963508. U.S. EPA (Environmental Protection Agency). 1994i. Technical Background Document for Soil Screening Guidance. Draft. EPA/540/R-94/106. Office of Emergency and Remedial Response, Washington, DC. NTIS PB95-963532. U.S. EPA (Environmental Protection Agency). 1995a. Drinking Water Regulations and Health Advisories. Office of Water, Washington, DC. U.S. EPA (Environmental Protection Agency). 1995b. Integrated Risk Information System (IRIS). Cincinnati, OH. December. U.S. EPA (Environmental Protection Agency). 1995d. Health Effects Assessment Summary Tables: FY-1995 Annual. EPA540/R-95-036. Office of Research and Development. Office of Emergency and Remedial Response, Washington, DC. NTIS PB95-921199. U.S. EPA (Environmental Protection Agency). 1996. Soil Screening Guidance: User's Guide. EPA/540/R-96/018. Office of Emergency and Remedial Response, Washington, DC. NTIS PB96.9634501 Van der Heijde, P.M. 1994. Identification and Compilation of Unsaturated/Vadose Zone Models. EPA/600/R-94/028. R.S. Ken Environmental Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Ada, OK. Van Wijnen, J.H., P. Clausing, and B. Brunekreef. 1990. Estimated soil ingestion by children. Environmental Research, 51:147-162. Wester, R.C., H.I. Maibach, D.A.W. Bucks, L. Sedik, J. Melendres, C. Liao, and S. DiZio. 1990. Percutaneous absorption of [14C]DDT and [MC]Bcnzo[o]pyrene from soil. Fund. App. Toxicol. 15:510-516. Wester, R.C., H.I. Maibach, and L. Sedik. 1993. Percutaneous absorption of pentachlorophenol from soil. Fundamentals of Applied Toxicology, 20. Will, M.E. and G.W. Suter II. 1994. Toxicological Benchmarks for Screening Potential Contaminants of Concern for Effects on Terrestrial Plants: 1994 Revision. ES/ER/TM-85/R1. Prepared for the U.S. Department of Energy by the Environmental Sciences Division of the Oak Ridge National Laboratory. 168 1305 APPENDIX A Generic SSLs TUT O08 1306 APPENDIX A Generic SSLs Table A-l provides generic SSLs for 110 chemicals. Generic SSLs are derived using default values in the standardized equations presented in Part 2 of this document. The default values (listed in Table A-2) are conservative and are likely to be protective for the majority of she conditions across the nation. However, the generic SSLs are not necessarily protective of all known human exposure pathways, reasonable land uses, or ecological threats. Thus, before applying generic SSLs at a site, it is extremely important to compare the conceptual site model (see the User's Guide) with the assumptions behind the SSLs to ensure that the site conditions and exposure pathways match those used to develop generic SSLs (see Parts 1 and 2 and Table A-2). If this comparison indicates mat the site is more complex than the SSL scenario, or that there are significant exposure pathways not accounted for by the SSLs, then generic SSLs are not sufficient for a full evaluation of the site. A more detailed site-specific approach will be necessary to evaluate the additional pathways or site conditions. Generic SSLs are presented separately for mrjor pathways of concern in bom surface and subsurface soils. The first column to the right of the chemical name presents levels based on direct ingestion of soil and the second column presents levels based on inhalation. As discussed in the User's Guide, the fugitive dust pathway may be of concern for certain metals but does not appear to be of concern for organic compounds. Therefore, SSLs for the fugitive dust pathway are only presented for inorganic compounds Except for mercury, no SSLs for the inhalation of volatiles pathway are provided for inorganic compounds because these chemicals are not volatile. The user should note that several of the generic SSLs for the inhalation of volatiles pathway are determined by the soil saturation concentration (€,„), which is used to address and screen the potential presence of nonaqueous phase liquids (NAPLs). As explained in Section 2.4.4, for compounds that are liquid at ambient soil temperature, concentrations above Cllt indicate a potential for free-phase liquid contamination to be present and the need for additional investigation. The third column presents generic SSL values for the migration to ground water pathway developed using a default DAF (dilution-attenuation factor) of 20 to account for natural processes that reduce contaminant concentrations in the subsurface (see Section 2.5.6). SSLs in Table A-l are rounded to two significant figures except for values less than 10, which are rounded to one significant figure Note that the 20 DAF values in Table A-l are not exactly 20 times the 1 DAF values because each SSL is calculated independently in both the 20 DAF and 1 DAF columns, with the final value presented according to the aforementioned rounding conventions. The fourth column contains the generic SSLs for the migration to ground water pathway developed assuming no dilution or attenuation between the source and the receptor well (i.e., a DAF of 1). These values can be used at sites where little or no dilution or attenuation of soil leachate concentrations is expected at a site (e.g., sites with shallow water tables, fractured media, karst topography, or source size greater than 30 acres). Generally, if an SSL is not exceeded for a pathway of concern, the user may eliminate the pathway or areas of the site from further investigation. If more than one exposure pathway is of concern, the lowest SSL should be used. A-l TUT O08 1307 Table A-1. Generic SSLs* Organics CAS No. Migration to ground water Compound Ingestion (n>g/kg) Inhalation volatile (mg/kg) 20 DAF (ma/kg) 1 DAF (mg/kg) ' 83-32-9 67-64-1 309-00-2 120-12-7 56-55-3 71-43-2 205-99-2 207-08-9 65-85-0 50-32 -t 111-44-4 117-81-7 75-27-4 75-25-2 71-36-3 85-68-7 86-74-8 75-15-0 56-23-5 57-74-9 106 -'7-8 108-90-7 124-48-1 67-66-3 95-57-8 218-01-9 72-54-8 72-55-9 50-29-3 53-70-3 84-74-2 95-50-1 106-46-7 91-94-1 75-34-3 107-06-2 75-35-4 156-59-2 156-60-5 120-83-2 Acenaphthene Acetone Aldrin Anthracene Benz(a)anthracene Benzene Benzo(d)fluoranthene Benzof K)f luoranthene Benzoic acid Benzo(a)pyrene Bis(2-chloroethyl)ether Bis(2-ethylhexyl)phthalate Bromodichloromethane Bromoform Butanol Butyl benzyl phthalate Carbazole Carbon disutftde Carbon tetrachloride Chlordane p-Chloroaniline Chlorobenzene Chlorodibromomethane Chloroform 2-Chlorophenol Chrysene ODD DDE DOT Dibenz(a,h)anthracene Di-n-butyl phthalate 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 3,3-Dichlorobenzidine 1 ,1 -Dichloroethane 1 ,2-Dichloroethane 1 ,1-Dichloroethylene c/s-1 ,2-Dichloroethylene frans-1 ,2-Dichloroethylene 2,4-Dichlorophenol 4.700 D 7,800 b 0.04 • 23,000 b 0.9 • 22 • 0.9 • 9 • 3.1E+05 b 0.09 ••' 0.6* 46 • 10 • 81 • 7.800 b 16,000 b 32 • 7,800 b 5 • 0.5 • 310 b 1 ,600 b 8 • 100 • 390 b 88 • 3 • 2 • 2 • 0.09 •* 7,800 b 7,000 b 27 • 1 • 7,800 b 7 • 1 • 780 b 1,600 b 230 b _ c 1.0E+05 d 3 • _,... C — c 0.8 • — c __ c _ c — e 0.2 *f 31,000 d 3.000 d 53 • 10.000 d 930 d _ c 720 d 0.3 • 20 • _ c 130 b 1.300 d 0.3 • 53.000 d __ c _ c — c — g _ c 2,300 d 560 d — g — c 1,300 b 0.4 • 0.07 • • 1 .200 d 3,100 d __ c 570 D 16 b 0.5 • 12,000 b 2 • 0.03 5 • 49 • 400 W 8 0.0004 «-f 3,600 0.6 0.8 17 b 930 d 0.6 • 32 b 0.07 10 0.7 b 1 0.4 0.6 4 b.i 160 * 16 • 54 • 32 * 2 • 2.300 d 17 2 •0.007 •* 23 b 0.02 0.06 0.4 0.7 1 b>l 29 D 0.8 b 0.02 • 590 b 0.08 •* 0.002 f 0.2 ••« 2 • 20 W 0.4 2E-05 •* 180 0.03 0.04 0.9 b 810 b 0.03 e-f 2 b 0.003 f 0.5 .-.. 0.03 b.' 0.07 0.02 0.03 0.2 b-f'' 8 • 0.8 • 3 e 2 '• 0.08 ••' 270 b .0.9 0.1 < 0.0003 ••' 1 b 0.001 f 0.003 f 0.02 0.03 0.05 »>•'•' A-2 TIJT 008 1308 Table A-1 (continued) Organics CAS No. 78-87-5 542-75-6 60-57-1 84-66-2 105-67-9 51-28-5 121-14-2 606-20-2 117-84-0 115-29-7 72-20-8 100-41-4 206-44-0 86-73-7 76-44-8 1024-57-3 118-74-1 87-68-3 319-84-6- 319-85-7 58-89-9 77-47-4 67-72-1 193-39-5 78-59-1 7439-97-6 72-43-5 74-83-9 75-09-2 95-48-7 91-20-3 98-95-3 86-30-6 621-64-7 1336-36-3 87-86-5 108-95-2 129-00-0 100-42-5 79-34-5 Compound 1 ,2-Dichloropropane 1 ,3-Dichloropropene Dieldrin Diethylphthalate 2,4-Dimethylphenol 2,4-Dinitrophenol 2,4-Dinrtrotoluene 2,6-Dinrtrotoluene Di-ft-octyl phthalate Endosutfan Endrin Ethylbenzene Fluoranthene Fluorene Heptachlor Heptachlor epoxide Hexachlorobenzene Hexachloro-1 ,3-butadiene a-HCH (a-BHC) P-HCH 0-BHC) rHCH (Lindane) Hexachlorocyclopentadiene Hexachloroethane lndeno(1 ,2,3-cd)pyrene Isophorone Mercury Methoxychlor Methyl bromide Methylene chloride 2-Methylphenol Naphthalene Nitrobenzene /V-Nttrosodiphenylamine /^Nitrosodi-n-propyiamine PCBs Pentachlorophenol Phenol Pyrene Styrene 1 ,1 ,2,2-Tetrachloroetharie Ingastion (mg/kg) 9 • 4 • 0.04 • 63,000 b 1,600 b 160 b 0.9 • 0,9 • 1,600 b 470 b 23 b 7,800 b 3,100 b 3.100 b 0.1 • 0.07 • 0.4 • 8 • ' . 0.1 • 0.4 • 0.5 » 550 b 46 • 0.9 • 670 • 23 b-' 390 b 110 b 85 • 3,900 b 3,100 b 39 b 130 • 0.09 »-f 1 h 3 »J 47,000 b 2,300 b 16,000 b 3 • Inhalation volatile (mg/kg) 15 *> 0.1 • 1 • 2,000 d _ c __ c _ c _ c 10,000 d _ c _ c 400 d _ c _ c 0.1 • 5 • 1 • 8 • 0.8 • — g _ c 10 b 55 • __ c 4,600 d 10 b-' __ c 10 b 13 • _ c _ c 92 b _ c _ c _ h __ c __ c __ c 1,500 d 0.6 • Migration to 20 DAF (mg/kg) 0.03 0.004 • 0.004 • 470 b 9 b 0.3 b<fii 0.0008 •* 0.0007 •* 10,000 d 18 b 1 13 4,300 b 560 b 23 0.7 2 2 0.0005 ••' 0.003 * 0.009 400 0.5 e 14 • 0.5 • 2 j 160 0.2 b 0.02 • 15 b 84 b 0.1 W 1 • 5E-05 •* _ h 0.03 '•' 100 b 4,200 b 4 0.003 •* ground water 1 DAF (mg/kg) 0.001 ' 0.0002 • 0.0002 ••' 23 b 0.4 b 0.01 b-f-' 4E-05 «-f 3E-05 «-f 10,000 d 0.9 b 0.05 0.7 210 b 28 b 1 0.03 0.1 ' 0.1 ' 3E-05 ••' 0.0001 ••' 0.0005 f 20 0.02 «-1 0.7 • 0.03 ••' 0.1 ' 8 0.01 b-' 0.001 ••' 0.8 b 4 b 0.007 W 0.06 ^ 2E-06 ••* _ h 0.001 (-' 5 b 210 b 0.2 0.0002 0} A-3 008 1309 Table A-1 (continued) O>- ?nics Migration to ground water CAS No. 127-18-4 Te1 Ingastion Compound (mg/kg) trachloroethylene 12* inhalation volatiles (mg/kg) 11 • 20 DAF (mg/kg) 0.06 1 DAF (mg/kg) 0.003 ' 127-18-4 108-88-3 8001-35-2 120-82-1 71-55-6 79-00-5 79-01-6 95-95-4 88-06-2 108-05-4 75-01-4 108-38-3 95-47-6 106-42-3 Tetrachloroethylene Toluene Toxaphene 1 ,2,4-Trichlorobenzene 1,1,1 -Trichloroethane 1 ,1 ,2-Trichloroethane Trichloroethylene 2 ,4 , 5-Trichlorophenoh 2,4,6-Trichlorophenol Vinyl acetate Vinyl chloride m-Xylene o-Xylene p-Xylene 12 •• 16,000 b 0.6 • 780 b _ e 11 • 58 • 7,800 b 58 • . 78.000 b 0.3 • 1.6E+05 b 1.6E+05 b 1.6E+05 b 11 • 650 d 89 • 3,200 d 1,200 d 1 • 5 • __ c 200 • 1,000 b 0.03 • 420 d 410 d 460 d 0.06 12 31 5 2 0.02 0.06 270 bJ 0.2 ••" 170 b 0.01 f 210 190 200 0.003 ' 0.6 2 0.3 f 0.1 0.0009 f 0.003 f 14 b-' 0.008 ••'•' 8 b 0.0007 f 10 9 10 A-4 TUT OO8 Table A-1 (continued) Inorganics CAS No. 7440-36-0 7440-38-2 7440-39-3 7440-41-7 7440-43-9 7440-47-3 16065-83-1 18540-29-9 57-12-5 7439-92-1 7440-02-0 7782-49-2 7440-22-4 7440-28-0 7440-62-2 7440-66-6 Migration to ground water Compound Antimony Arsenic Barium Beryllium Cadmium Chromium (total) Chromium (III) Chromium (VI) Cyanide (amenable) Lead Nickel Selenium Silver Thallium Vanadium Zinc Ingestion (mg/kg) 31 b 0.4* 5,500 b 0.1 • TO b»m / o 390 b 78,000 b 390 b 1,600b 40? k 1 ,600 b 390 b 390 b __ c 550 b 23,000 b Inhalation fugitive particulate (mg/kg) __ c 750* 6.9E+05 b 1,300- 1 ,800 • 270* — 'c 270 • _ c __ k 1 3,000 • _ c __ c _ c _ c _ c 20 DAF (mg/kg) 5 29 ' 1.6001 63' 8 ' 38' _ a 38' 40 _ k 130' 5 i 34 w 0.7 ' 6,000 b 12,000 b-' 1 DAF (mg/kg) 0.3 1 ' 82 ' 3 ' 0.4' 2 ' _ B 2 1 2 _ k 7* 0.3' o b>' 0.04' 300 b 620 b<i OAF = Dilution and attenuation factor. a Screening levels based on human health criteria only. b Calculated values correspond to a noncancer hazard quotient of 1. c No toxiclty criteria available for that route of exposure. d Soil saturation concentration (Cnt). * Calculated values correspond to a cancer risk level of 1 in 1,000,000. . ' Level is at or below Contract Laboratory Program required quantitation limit for Regular Analytical Services (RAS). 0 Chemical-specific properties are such that this pathway is not of concern at any soil contaminant concentration. h A preliminary remediation goal of 1 mg/kg has been set for PCBs based on Guidance on Remedial Actions for Superfund Sites with PCB Contamination (U.S. ERA, 1990) and on ERA efforts to manage PCB contamination. | SSL for pH of 6.8. ' Ingestion SSL adjusted by a factor of 0.5 to account for dermal exposure. k A screening level of 400 mg/kg has been set for lead based on Revised Interim Soil Lead Guidance for CERCLA Sites and RCRA Corrective Action Facilities (U.S. EPA, 1994). I SSL is based on RfD for mercuric chloride (CAS No. 007487-94-7). m SSL is based on dietary RfD. • ' A-5 TUT COS 1311 Table A-2. Generic SSLs: Default Parameters and Assumptions SSL pathway Parameter Migration to Inhalation ground water Default Source Characteristics Continuous vegetative cover Roughness height Source area (A) Source length (L) Source depth • O 50 percent 0.5 cm for open terrain; used to derive Uti7 0.5 acres (2,024 m2); used to derive L for MTG 45 m (assumes square source) Extends to water table (i.e., no attenuation in unsaturated zone) Soil Characteristics Soil texture Dry soil bulk density (pb) Soil porosity (n) Vol. soil water content (6W) Vol. soil air content (6a) Soil organic carbon (foc) Soil pH Mode soil aggregate size Threshold windspeed @ 7 m (L), 7) O O Loam; defines soil characteristics/ parameters I.5 kg/L 0.43 0.15 (INH); 0.30 (MTG) 0.28 (INH); 0.13 (MTG) 0.006 (0.6%, INH); 0.002 (0.2 %, MTG) 6.8; used to determine pH-specHic K<j (metals) and KOC (ionizable organics) 0.5 mm; used to derive U, 7 II.32m/s Meteorological Data Mean annual windspeed (Um) Air dispersion factor (Q/C) Volatilization Q/C Fugitive paniculate Q/C 4.69 nVs (Minneapolis, MN) 90th percentile conterminous U.S. 68.81; Los Angeles, CA; 0.5-acre source 90.80; Minneapolis, MN; 0.5-acre source Hydrogeologic Characteristics Hydrogeologic setting Dilution/attenuation factor (DAF) Generic (national); surficial aquifer 20 • Indicates input parameters directly used in SSL equations. O Indicates parameters/assumptions used to develop SSL input parameters. INH = Inhalation pathway. MTG = Migration to ground water pathway. A-6 TUT 1312 Analysis of Effects of Source Size on Generic SSLs A large number of comm enters on the December 1994 Soil Screening Guidance suggested that most contaminated soil sources were 0.5 acre or less. Before changing this default assumption from 30 acres to 0.5 acre, the Office of Emergency and Remedial Response (OERR) conducted an analysis of the effects of changing the area of a contaminated soil source on generic SSLs calculated for the inhalation and migration to ground water exposure pathways. This analysis includes: An analysis of the sensitivity of SSLs to a change in source area from 30 acres to 0.5 acre • Mass-limit modeling results showing the depth of contamination for a 30-acre source that corresponds to a 0.5-acre SSL. All equations, assumptions, and model input parameters used in mis analysis are consistent with those described in Part 2 of this document unless otherwise indicated. Chemical properties used in the analysis are described in Part 5 of this document. In summary, the results of this analysis indicate that: • The SSLs are not particularly sensitive to varying the source area from 30 acres to 0.5 acre. This reduction in source area lowers SSLs for the inhalation pathway by about a factor of 2 and lowers SSLs for the migration to ground water pathway by a factor of 2.9 under typical hydrogeologic conditions. • • Half-acre SSLs calculated for 43 volatile and semivolatile contaminants using the infinite source models correspond to mass-limit SSLs for a 30-acre source uniformly contaminated to a depth of about 1 to 21 meters (depending on contaminant and pathway); the average depth is 8 meters for. the inhalation pathway (21 contaminants) and 11 meters for the migration to ground water pathway (43 contaminants). Sensitivity Analysis. For the inhalation pathway, source area affects the Q/C value (a measure of dispersion), which directly affects the final SSL and is not chemical-specific. Higher Q/C values result in higher SSLs. As shown in Table 3 (Section 2.4.3), the effect of area on the Q/C value is not sensitive to meteorological conditions, with the ratio of a 0.5-acre Q/C to a 30-acre Q/C ranging from 1.93 to 1.96 over the 29 conditions analyzed. Decreasing the source area from 30 acres to 0.5 acre will therefore increase inhalation SSLs by about a factor of 2. For the migration to ground water pathway, source area affects the DAF, which also directly affects the final SSLs and is not chemical-specific. The sensitivity analysis for the dilution factor is more complicated than for Q/C because increasing source area (expressed as the length of source parallel to ground water flow) not only increases infiltration to the aquifer, which decreases the dilution factor, but also increases the mixing zone depth, which tends to increase the dilution factor. The first effect generally overrides the second (i.e., longer sources have lower dilution factors) except for very thick aquifers (see Section 2.5.7). The sensitivity analysis described in Section 2.5.7 shows that the dilution model is most sensitive to the aquifer's Darcy velocity (i.e., hydraulic conductivity x hydraulic gradient). For a less conservative Darcy velocity (90th percentile), decreasing the source area from 30 acres to 0.5 acre increased the dilution factor by a factor of 3.1 (see Table 9, Section 2.5.7). For the conditions analyzed, decreasing the source area from 30 acres to 0.5 acre affected dilution factor from no increase to a factor of 4.3 increase. No increase in dilution factor for a '0.5-acre source was observed for the less conservative A-7 TUT OOS 131.: (higher) aquifer thickness (46 m). In this case the decrease in mixing zone depth balances the decrease in infiltration rate for the smaller source. Mass-Limit Analysis. The infinite source assumption is one of the more conservative assumptions inherent in the SSL models, especially for small sources. This assumption should provide adequate protection for sources with larger areas than those used to calculate SSLs. To test this hypothesis the SSL mass-limit models (Section 2.6) were used to calculate, for 43 volatile and semivolatile chemicals, the depth at which a mass-limit SSL for a 30-acre source is equal to a 0.5-acre infinite-source SSL. The mass-limit models are simple mass-balance models that calculate SSLs based on the conservative assumption that the entire mass of contamination in a source either volatilizes (inhalation model) or leaches (migration to ground water model) over the exposure period of interest. These models were developed to correct the mass-balance violation in the infinite source models for highly volatile or soluble contaminants. Table A-3 presents the results of mis analysis. These results demonstrate that 0.5-acre infinite source SSLs are protective of uniformly contaminated 30-acre source areas of significant depth. For the 21 chemicals analyzed for the inhalation pathway, these source depths range up to 21 meters, with an average depth of 8 meters and a standard deviation of 5.7. For the migration to ground water pathway, source depths for 43 contaminants range to 21 meters, with an average of 11 meters and a standard deviation of 5.4. References IJ.f liPA (Environmental Protection Agency). 1990. Guidance on Remedial Actions for Superfund Sites with PCB Contamination. Office of Solid Waste and Emergency Response, Washington. DC. NT1S PB91-921206CDH. U.S. EPA (Environmental Protection Agency). 1994.'Revised Interim Soil Lead Guidance for CERCLA Sites and RCRA Corrective Action Facilities. Office of Solid Waste and Emergency Response, Washington, DC. Directive 9355.4-12. A-8 TUT 008 1314 Table A-3.Source Depth where 30-acre* Mass-Limit SSLs = 0.5-acreb Infinite-Source SSLs0 Chemical Acetone Benzene Benzole acid Bis(2-chloroethyl)ether Bromodichioromethane Bromoform Butanol Carbon disutfide Carbon tetrachloride Chlorobenzene Chlorodibromomethane Chloroform 2-Chlorophenol 1 ,2-Dichlorobenzene 1 ,4-Dichlorobenzene 1,1-Dichloroethane 1 ,2-Dichloroethane 1 ,1-Dichloroethylene c/s-i,2-Dichloroethylene frans-1 ,2-Dichloroethylene 2,4-Dichlorophenol 1 ,2-Dichloropropane 1 ,3-Dichloropropene 2,4-Dimethylphenol 2,4-Dinrtrophenol 2,4-Dinrtrotoluene 2,6-Dinttrotohjene Ethylbenzene Methyl bromide Methylene chloride 2-Methylphenol Nitrobenzene 1 ,1 ,2,2-Tetrachloroethane Tetrachtoroethylene Toluene 1 ,1 ,1 -Trichloroethane 1 ,1 ,2-Trichloroethane Trichloroethylene Vinyl acetate Source Inhalation NA 8.1 NA 0.7 NA 0.9 NA 19 11 3.5 NA 8.3 NA NA NA 9.1 5.6 15 NA NA NA 6.2 12 NA NA NA NA NA 12 8.9 NA 0.5 1.6 8.7 NA NA 3.4 6.8 4.6 depth (m) Migration to ground water' 21 12 21 18 13 11 20 11 6 6 13 14 4 3 3 15 18 10 15 12 8 14 12 7 21 11 12 4 17 18 11 13 11 7 7 9 14 7 20 A-9 TUT 008 13 Table A-3. (continued) Source depth (m) Chemical ________________inhalation Migration to ground water0 Vinyl chloride 21 13 m-Xylene NA 4 o-Xylene NA 4 p-Xylene ___________________NA_____________ 4________ NA = Risk-based SSL not available. » CVC = 35.15;DAF=10. b Q/C « 68.81 ;DAF = 20. c Migration to ground water mass-limit analysis based on 70-yr exposure duration and 0.18 nVyr infiltration rate. A-10. TUT OO! 1316 APPENDIX B Route-to-Route Extrapolation of Inhalation Benchmarks TUT O08 131: APPENDIX B Route-to-Route Extrapolation of Inhalation Benchmarks Introduction For a number of thj contaminants commonly found at Superfund sites, inhalation benchmarks for toxicity are not available from IRIS or HEAST. As pointed out by commenters to the December 1994 Soil Screening Guidance, ingestion SSLs tend to be higher than inhalation SSLs for most volatile chemicals with both inhalation and ingestion benchmarks. This suggests that ingestion SSLs may not be adequately protective for inhalation exposure to chemicals that lack inhalation benchmarks. To address this concern, the Office of Emergency and Remedial Response (OERR) evaluated potential approaches for deriving inhalation benchmarks using route-to-route extrapolation from oral benchmarks (e.g., inhalation reference concentrations [RfCs] from oral reference doses [RfDs]). OERR evaluated Agency initiatives concerning -vjte-to-route extrapolation, including: the potential reactivity of airborne toxicants (e.g., portal-of-entry effects), the pharmacokinetic behavior of toxicants for different routes of exposure (e.g., absorption by the gut versus absorption by the lung), and the significance of physicochemical properties in determining dose (e.g., volatility, speciation). During this process, OERR consulted with staff in the EPA Office of Research and Development (ORD) to identify appropriate techniques and key technical aspects in performing route-to-route extrapolation. The following sections describe OERR's analysis of route-to-route extrapolation and the conclusions reached regarding the use of extrapolated inhalation benchmarks to support inhalation SSLs. B.1 Extrapolation of Inhalation Benchmarks The first step taken in considering route-to-route extrapolation of inhalation benchmarks was to compare existing inhalation benchmarks to inhalation benchmarks extrapolated from oral studies. This comparison was important to determine whether a simple route-to-route extrapolation could provide a defensible inhalation benchmark for chemicals lacking appropriate inhalation studies. OERR identified nine chemicals found in IRIS (Integrated Risk Information System) that have verified RfDs and RfCs for noncancer effects, including three chemicals found in the SSL guidance (ethylbenzene, styrene, and toluene). Reference concentrations for inhalation exposure were extrapolated from oral reference doses for adults using the following formula: extrapolated RfC (mg/m3) = RID (mg/kg-d) x ™*B ^ (B~l* 20 m / d • It is important to note that dosimetric adjustments were not made to account for respiratory tract deposition efficiency and distribution; physical, biological, and chemical factors; and other aspects of exposure (e.g., discontinuous exposure) that affect uptake and clearance. Consequently, this simple extrapolation method relies on the implicit assumption that the route of administration is irrelevant to the dose delivered to a target organ, an assumption not supported by the principles of dosimetry or phaimacokinetics. B-l TUT OO8 1318 The limited data on noncarcinogens suggest that more volatile constituents tend to have extrapolated RfCs closer to the RfCs developed by EPA (i.e., extrapolated RfC within a factor of 3 of the RfC in IRIS). The less volatile chemicals (e.g., dichlorvos) tend to be below the RfCs developed by EPA workgroups by 1 to 3 orders of magnitude. Although this data set is insufficient to discern trends in extrapolated versus IRIS RfCs, two points are reasonably clear: (1) for some volatile chemicals, route-to-route extrapolation results in inhalation benchmarks reasonably close to the RfC. and (2) as volatility decreases and/or chemical speciation becomes important (e.g., hydrogen sulfidej with respect to environmental chemistry and toxicology, the uncertainty in extrapolated inhalation benchmarks is likely to increase. For carcinogens, OERR identified 41 chemicals in IRIS for which oral cancer slope factors (CSFotat) and inhalation unit risk factors (URFs) are available, including 23 chemicals covered under the SSL guidance. Unit risk factors for inhalation exposure were extrapolated from oral carcinogenic slope factors for adults using the following formula: (mg/kg-d)-1 ' . (B-2) 3 "3 7Q kg x 20m/d x 10" Using the extrapolated URF, risk-specific air concentrations were calculated as a lifetime average exposure concentration as shown in equation 6-3: . '. . ,3 target risk 10'6 (B-3) extrapolated air concentration M-g/m = ———————— - Not surprisingly, the risk-based (i.e., 10- 6) air concentrations in IRIS are the same as the air concentrations extrapolated from the CSForai for 30 of the 41 carcinogenic chemicals evaluated (at one significant figure). Historically, oral and inhalation slope factors have been based on oral studies for chemicals for which pharmacokmetic or portal-of-entry effects were considered insignificant. As a result, route of exposure extrapolations were often included in the development of the carcinogenic slope factors. However, the divergence of extrapolated air concentrations. with risk-based (i.e., 10- 6) air concentrations in IRIS reflects newer methods in use at EPA that address portal-of-entry effects, dosimetry, and pharmacokmetic behavior. For example, 1,2-dibromomethane has an extrapolated 10-6 a-' concentration that is 2 orders of magnitude below the value in IRIS. This difference is probab: attributable to differences in: (1) the endpoint for inhalation exposure (nasal cavity carcinoma) versus oral exposure (squamous cell carcinoma), and/or (2) portal-of-entry effects directly related to deposition physiology and absorption of 1,2-dibromomethane B.2 Comparison of Extrapolated Inhalation SSLs with Generic SSLs Having performed a simple extrapolation of inhalation benchmarks, the next step was to compare the inhalation SSLs (SSL.^) based on extrapolated data to the soil saturation concentrations* (C,at) and generic SSLs for soil ingestion (SSLjng) and ground water ingestion (SSLgw). Table B-l presents the 50 organic chemicals in the SSL guidance that lack inhalation benchmarks. The table presents oral benchmarks found in IRIS (columns 2 and 3) and extrapolated inhalation benchmarks as * The derivation of C,,t and its significance is discussed in Section 2.4.4 of this Technical Background Document. B-2 TUT 008 1319 described in Equations B-l and B-2 (columns 4 and 5). In addition, the table presents volatilization- based SSLs and SSLs based on paniculate emissions derived from the extrapolated toxicity values. For each column of extrapolated inhalation SSLs in this table, values are truncated at 1,000.000 mg/kg because the soil concentration cannot be greater than 100 percent (i.e., 1,000,000 ppm). B. 2.1 Comparison of Extrapolated SSLs Based on Volatilization The extrapolated SSLjnh for volatilization (SSLja^v) was calculated with Equation 4 in Section 2.4 using a chemical -specific volatilization factor (VF). In Table B-l, the SSLjnh-v values based on extrapolated inhalation benchmarks (column 6) are compared with the soil saturation concentration (Cut* column 7) and generic migration to ground water SSLs assuming a dilution attenuation factor (DAF)of20(SSLgw). ' - As. described in Section 2.4.4, Cut represents the concentration at which soil pore air is saturated with a chemical and maximum volatile emissions are reached. A comparison of the Ct(t with the extrapolated SSL^.v values indicates that, for 36 of the SO contaminants, SSLj^v exceeds the soil saturation concentration, often by several orders of magnitude. Because maximum volatile emissions occur at CMt, these 36 contaminants are not likely to pose significant risks through the inhalation pathway, and therefore the lack of inhalation benchmarks is not likely to underestimate risk through the volatilization pathway. For the remaining 14 contaminants with extrapolated SSL^v values below Cltt, all are above the generic SSLg* values. This analysis suggests that SSLs based on the migration^o-groundwater pathway are likely to be protective of the inhalation pathway as well. However, for sites where groundwater is not of concern, the SSLs based on ingestion may not necessarily be protective of the inhalation pathway. The analysis indicates that the extrapolated inhalation SSLs are below SSLs based on direct ingestion for the following chemicals: acetone, brpmodichloromethane, chlorodibromomethane, CIS- 1, 2-dichloroethylene, and rran5-l, 2-dichloroethylene. This analysis supports the possibility that the SSLs based on direct ingestion for the listed chemicals may not be adequately protective of inhalation exposures. However, a more rigorous evaluation of the route-to-route extrapolation methods used to derive the toxicity criteria for this analysis is warranted (refer to section B.3). B.2.2 Comparison of Extrapolated SSLs Based on Paniculate Emissions The extrapolated paniculate inhalation SSLs (SSL^p) were calculated with Equation 4 in Section 24 using the paniculate emission factor (PEF) of 1.32 x 109 m3/kg. Table B-l compares the SSL,,^ values based on extrapolated benchmarks (column 10) and generic SSLs based on direct ingestion (SSLjng, Column 9). This comparison indicates that the extrapolated SSL^.p values that are based on the PEF are well above the SSLs for soil ingestion. Thus,. ingestion SSLs are likely to be protective of inhalation risks from fugitive dusts from surface soils. B.3 Conclusions and Recommendations Based on the results presented in this appendix, OERR reached several conclusions regarding route- to-route extrapolation of inhalation benchmarks for the development of generic inhalation SSLs. First, it is reasonable to assume that, for some contaminants, the lack of inhalation benchmarks may underestimate risks due to inhalation exposure. Of the 17 volatile organics for which both the ingestion and inhalation SSLs are based on IRIS benchmarks, all had inhalation SSLs that were below the ingestion SSLs. Nevertheless, generic SSLs for ground water ingestion (DAF of 20) are lower, B-3 . TUT 008 1320 often significantly lower, than both extrapolated and HUS-based inhalation SSLs with the exception of vinyl chloride, which is gaseous at ambient temperatures. Thus, at sites where ground water is of concern, migration to ground water SSLs generally will be protective from the standpoint of inhalation risk. However, if the ground water is not of concern at a site (e.g.. if ground water below the site is not potable).'the use of SSLs for soil ingestion may not be adequately protective of the inhalation pathway. Second, the extrapolated SSLinh values are not intended to be used as generic SSLs for site investigations; the extrapolated inhalation SSLs are useful in determining the potential for inhalation risks but should not be misused as SSLs. Route-to-route extrapolation methods must account for the relationship between physicochemical properties and absorption and distribution of toxicants, the significance of portal-of-entry effects, and the potential differences in metabolic pathways associated with the intensity and duration of inhalation exposure. However, methods required to generate sufficiently rigorous inhalation benchmarks have recently been developed by the ORD. A final guidance document was made available by ORD in November of 1995 mat addresses many of the issues critical to the development of inhalation benchmarks described above. The document, entitled Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994), describes the application of inhalation dosimetry to derive inhalation reference concentrations and represents the current state-of-the-science at EPA with respect to inhalation benchmark development. The fundamentals of inhalation dosimetry are presented with respect to toxicokinetics and the physicochemical properties of chemical contaminants. Thus, at sites where the migration to ground water pathway is not of concern and a she manager determines that the inhalation pathway may be significant for contaminants lacking inhalation benchmarks, route-to-route extrapolation may be performed using EPA-approved methods on a case-by-case basis. Chemical-specific route-to-route extrapolations should be accompanied by a complete discussion of the data, underlying assumptions, and uncertainties identified in the extrapolation process. Extrapolation methods should be consistent with the EPA guidance presented in Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry. If a route-to-route extrapolation is found not to be appropriate based on the ORD guidance, the information on extrapolated SSLs may be included as part of the uncertainty analysis of the baseline risk assessment for the site. Reference U.S. EPA (Environmental Protection Agency). 1994 .thods for Derivation of Inhalation Reference Concentrations and Application of In: .Jtion Dosimetry. EPA/600/8-90/066F Office of Research and Development, Washington, DC B-4 Table B-1. Comparison of Extrapolated Inhalation SSLs (SSLlnh) with Soil Concentrations (C.,t), and Migration to Ground Water (SSL^) Compound Acenaphthene Acetone Anthracene Benzfc) anthracene Benzofty nuoranthene Benzoftl fkioranthene BenzofaJ pyrene Benzole acid Bl»(2-ethylhexyl)phthalate Bromodlchloromefhane Butanol Butyl benzyl pMhaJate Carbazoto p -CMoroanWrw CMorodtbromomettiane 2-CMorophanol Chryseno ODD DDE Dlbanzfoty anthracene Dl-n -butyl phthalate a.ff-DfcMorobenzkine. da -1,2-Dtehterorthylen* mm •1i2-DleMwo«lhyl«M 2,4-DfcMorophenol Dlethylphthalate 2,4-Dhnethylphenol 2,4-DlnHrophenol 2.4-Dlnttrotokiem 2,6-Dlnltrototuene Dl-n -octyl phthalate Endosulfan Endrln Fluoranthene IRIS oral benchmarks RfD CSF (mg/ko-d) (moftfl-d)' 6E-02 IE-01 3E-01 73 E-Ot 73 E-01 73 E-02 7.3 E+OO 4E+00 14 E-02 6 2 E-02 IE-01 2E-01 20 E-02 4E-03 84 E-02 5E-03 7.3 E-03 24 E-01 34 E-01 73 E+OO IE-01 4.5 E-01 1E-02 2E-02 3E-03 8E-01 2E-02 2E-03 68E-01 68E-01 2E02 6E 03 3E-04 4E02 Extrapolated Inhalation benchmarks RfC URF (ma/m«) (maftnT 2.1E-01 35E-01 1.1E+00 2. IE-04 2. IE-04 2 IE-05 2. IE-03 1.4E+01 40E-08 1.8E-05 35E-01 7.0E-01 57E-06 1.4E-02 2.4E-05 18E-02 2 IE-06 69E-05 97E-05 2.1E-03 3.5E-01 1.3E-04 35E-02 7.0E-02 HE-02 28E+00 7 OE-02 7.0E-03 19E-04 19E-04 70E-02 21E-02 1 IE-03 1 4E-01 VF-basedSSL«(mg/kg) Extrapolated volatlllzaHon Generic SSI^. SSU^. C« IDAF20) 48.000 181 570 4,600 103,747 IS 860.000 8 12.000 110 22 2 54 11 5. 4,600 6 49 ' 28 10 8 > 1.000,000 363 400 130,000 30,804 3,600 1.14 2,981 0.8 14.000 10.477 17 > 1,000.000 928 930 1.10T. 153 0.8 4. 10C 2,832 0.7 1.S 1,280 0.4 550 53,482 4 3.200 4 180 820 540 18 620 3,218 54 120 57 2 > 1,000,000 2,279 2.300 24 14 0 110 1,205 0.4 170 3,087 0.7 2.500 4.419 1 > 1,000,000 1.974 470 19.000 10.656 9 1.000 279 0.3 4 182 0.0008 4 94 0.0007 > 1.000.000 9,984 10.000 18,000 7 18 2.500 18 1 450.000 132 4300 Generic SSL. 4.700 7,800 23.000 0.9 0.9 9 0.09 310.000 48 10 7.800 18,000 32 310 8 390 88 3 2 0.09 7.800 1 780 1,800 240 63.000 1.600 160 09 09 1.600 470 23 3.100 PEF-basedSSLa (mgfcfl) . Extrapolated parttculate SSLt*. > 1.000.000 > 1.000.000 > 1.000.000 15.000 15.000 150.000 1.500 > 1,000.000 800,000 180.000 > 1,000,000 > 1.000.000 560,000 > 1,000.000 130,000 > 1.000,000 > 1.000,000 47.000 33,000 1,500 > 1.000.000 25.000 > 1.000,000 > 1,000,000 > 1,000.000 > 1.000,000 > 1.000.000 >1. 000.000 17.000 17.000 > 1.000,000 > 1.000.000 > 1,000.000 >1.000.000 w CO TabIeB 1.C.mpaH,ono,E^p....e-Inh.I.«onSSL,,SSL»,wHhS.H ————— *.« —————— 00 C__ Fluorene rHexachlorocyctoh»xane lndeno(1.2.3-cd;pyrene Isophorone MethoxycWor 2-Methylphenol Naphthalene N -Nltrosodphenylamlne ; N-Nltrosodl-n-propytamlne Pentachlorophenol Phenol Pyrene 2,4.5-Trtehlorophenol m-Xytene o-Xytene IRIS oral benchmarks RID CSF fmg/kg-d) 4E-02 5E-03 5E-02 4E-02 6E-01 3E-02 IE-01 2E+00 2E+00 2E+00 Extrapolated Inhalation benchmarks RfC URF (mg/kg-d)' (ma/in') 1 .4E-01 13E+00 73 E-01 95 E-04 4.9 E-03 7.0 E+00 1 2 E-01 18E-02 1.8E-01 14E-01 21 E+00 1 IE-01 35E-01 7.0E+00 70E+00 7.0E+00 VF-basedSSL>(mg/kg) Extrapolated volatilization Generic SSL^ c .. (DAF 20) (mg/nvr OOEH— 75,000 3 7E-04 ° ' 2.1 E-04 2.7E-07 14E-08 20E-03 34E-05 •3.1 660 720 74.000 37.000 D *W\ O.ZUO 1.000 0.1 83 400.000 420.000 250,000 45.000 45.000 ft nnn —— ^m ———— - — . 44 0.5 4,570 26 16,827 473 Of 9 275 2.413 7,121 22.588 85 11.618 418 413 461 >;ftn 3UV Generic ssu. 3.100 PEF-basedSSLs (mgftfl) Extrapolated partlculate SSL^ > 1,000.000 0.009 05 8,600 14 . 0.9 15,000 0.5 670 4tu\ ion itxi jyu 15 84 > 1.000,000 > 1,000.000 3.900 >1. 000.000 3.100 > 1,000.000 1 130 > 1.000.000 0.00005 0.09 1.600 0.03 3 94,000 100 4.200 270 210 190 200 47,000 2.300 7.800 160,000 160.000 160.000 > 1.000.000 > 1.000.000 > 1,000,000 > 1,000.000 >1. 000,000 > 1.000.000 p-Xvtene___________I £Cfw_________: . ..—_________ NR a SSlw,, I* gwater than 1.000,000 ppm (= no volatile Inhalation risk at any soil concentration) Bold Indicates where extrapolated SSL» value* are leu than SSL value* bated on direct Ingestlon l_» ww APPENDIX C Limited Validation of the Jury Infinite Source and Jury Finite Source Models (EQ, 1995) TUT OOS 1324 LIMITED VALIDATION OF THE JURY INFINITE SOURCE AND JURY REDUCED SOLUTION FINITE SOURCE MODELS FOR EMISSIONS OF SOIL-INCORPORATED VOLATILE ORGANIC COMPOUNDS by Environmental Quality Management, Inc. Cedar Terrace Office Park, Suite 250 3325 Chapel Hill Boulevard Durham, North Carolina 27707 Contract No. 68-D30035 Work Assignment No. 1-55 Subcontract No. 95.5 PN 5099-4 Janine Dinan, Work Assignment Manager U.S. ENVIRONMENTAL PROTECTION AGENCY OFFICE OF SOLID WASTE AND EMERGENCY RESPONSE WASHINGTON, D.C. 20460 July 1995 C-l TUT OOS DISCLAIMER This project has been performed under contract to E.H. Pechan & Associates, Inc. It was funded with Federal funds from the U.S. Environmental Protection Agency under Contract No. 68-030035. The content of this publication does not necessarily reflect the views or policies of the U.S. Environmental Protection Agency nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. C-2 "ft ...... oos CONTENTS Figures iv Tables vi Acknowledgment • ' vii 1. Introduction 1 Project objectives 3 Technical approach 3 2. Review of the Jury Volatilization Models 4 Finite source model derivation 6 Infinite source model derivation 10 Summary of model assumptions and limitations 11 3. Model Validation 14 Validation of the Jury Infinite Source Model 14 VaJidation of the Jury Reduced Solution Finite Source Model 41 4. Parametric Analysis of the Jury Volatilization Models 46 Affects of soil parameters 48 Affects of nonsoil parameters 49 5. Conclusions . 51 References . 54 Appendices A. Validation Data for the Jury Infinite Source Model A-1 B. Validation Data for the Jury Reduced Solution Finite Source Mode! B-1 c-3 TUT 008 1327 FIGURES Number 1 Predicted and measured emission flux of dieldrin versus time (C0 = 5 ppmw) 19 2 Comparison of log-transformed modeled and measured emission flux of dieldrin (C0 = 5 ppmw) 20 3 Predicted and measured emission flux of dieldrin versus time (C0 « 10 ppmw) 21 4 Comparison of log-transformed modeled and measured emission flux of dieldrin (Q, = 10 ppmw) 22 5 Predicted and measured emission flux of lindane versus time (C0 = 5 ppmw) 24 6 Predicted and measured emission flux of lindane versus time (C0 = 10 ppmw) 25 7 Comparison of log-transformed modeled and measured emission flux of lindane (C0 = 5 ppmw) 26 8 Comparison of log-transformed modeled and measured emission flux of lindane (C0 =10 ppmw) 27 9 Predicted and measured emission flux of benzene (Q, =110 ppmw) 33 10 Predicted and measured emission flux of toluene (C0 « 880 ppmw) 34 11 Predicted and measured emission flux of ethylbenzene (C0 = 310 ppmw) 35 12 Comparison of log-transformed modeled and measured emission fluxof benzene (Cb = 110 ppmw) . 37 C-4 TUT 008 Number FIGURES (continued) 13 Comparison of log-transformed modeled and measured emission flux of toluene (C0 = 880 ppmw) 38 14 Comparison of log-transformed modeled and measured emission flux of ethylbenzene (C0 =310 ppmw) 39 15 Predicted and measured emission flux of triallate versus time 43 16 Comparison of log-transformed modeled and measured emission flux of triallate 44 C-5 TUT OOS 1329 TABLES Number Volatilization Model Input Values for LJndane and Dieldrin 17 2 Summary of the Bench-Scale Validation of the Jury Infinite Source Model 28 3 Volatilization Mode! Input Variables for Benzene, Toluene, and Ethylbenzene 31 4 Summary of Statistical Analysts of Pilot-Scale Validation 36 5 Volatilization Model Input Values for Triallate 42 C-6 TUT COS 133O ACKNOWLEDGMENT This report was prepared for the U.S. Environmental Protection Agency by Environmental Quality Management, Inc. of Durham, North Carolina under contract to E.H. Pechan & Associates, Inc. Ms. Annette Najjar with E.H. Pechan & Associates, Inc. served as the project technical monitor and Craig Mann with Environmental Quality Management, Inc. managed the project and was author of the report. Janine Dinan of the U.S. Environmental Protection Agency's Toxics Integration Branch provided overall project direction and served as the Work Assignment Manager. C-7 . TUT 008 1331 C-8 TUT 008 1331 A SECTION 1 INTRODUCTION In December 1995, the U.S. Environmental Protection Agency (EPA) Office of Solid Waste and Emergency Response published the Draft Technical Background Document (TED) for Soil Screening Guidance (U.S. EPA, 1994). This document provides the technical background behind the development of the Soil Screening Guidance for Superfund, and defines the Soil Screening Framework. The framework consists of a suite of methodologies for developing Soil Screening Levels (SSLs) for 107 chemicals commonly found at Superfund sites. An SSL is defined as "a chemical concentration in soil below which there is no concern under the Comprehensive Environmental Response Compensation and Liability Act (CERCLA) for ingestion, inhalation, and migration to ground water exposure pathways....* (U.S. EPA, 1994). The SSL inhalation pathway considers exposure to vapor-phase contaminants emitted from soils. Inhalation pathway SSLs are calculated using air pathway fate and transport models. Currently, the models and assumptions used to calculate SSLs for inhalation of voiatiles are updates of risk assessment methods presented in the Risk Assessment Guidance for Superfund (RAGS) Part B (U.S. EPA, 1991). The RAGS Part B methodology employs a reverse calculation of the concentration in soil of a given contaminant that would result in an acceptable risk-level in ambient air at the point of maximum long-term air concentration. Integral to the calculation of the inhalation pathway SSLs for voiatiles, is the soil- to-air volatilization factor (VF) which defines the relationship between the concentration of contaminants in soil and the volatilized contaminants in air. The VF (rtf/kg) is calculated as the inverse of the ambient air concentration at the center of a ground- level, nonbouyant area source of volatile emissions from soil. The equation for C-9 TUT OO8 1332 calculating the VF consists of two parts: 1) a volatilization model, and 2) an air dispersion model. The volatilization model mathematically predicts volatilization of contaminants fully incorporated in soils as a diffusion-controlled process. The basic assumption in the mathematical treatment of the movement of volatile contaminants in soils under a concentration gradient is the applicability of the diffusion laws. The changes in contaminant concentration within the soil as well as the loss of contaminant at the soil surface by volatilization can then be predicted by solving the diffusion equation for different boundary conditions. As noted in the TBD, Environmental Quality Management, Inc. (EQ) under a subcontract to E. H. Pechan conducted a preliminary evaluation of several soil volatilization models for the U.S. EPA Office of Emergency and Remedial Response (OERR) that might be suitable for addressing both infinite and finite sources of emissions (EQ, 1994). The results of this study indicated that simplified analytical solutions are presented in Jury et al. (1984 and 1990) for both infinite and finite emission sources. These analytical solutions are mathematically consistent and use a common theoretical approximation of the effective diffusion coefficient in soil. Under a subcontract with E. H. Pechan for OERR, EQ performed a limited validation of the Jury Infinite Source emission model (Jury et al., 1984, Equation 8) and the Jury Reduced Solution finite source emission model (Jury et al., 1990, Equation B1), hereinafter known as the Jury volatilization models. This document reports on several studies in which volatilization of contaminants from soils was directly measured and data were obtained necessary to calculate emissions of contaminants using the Jury Infinite Source model and the Jury Reduced •• Solution finite source model. These data are then compared and analyzed by statistical methods to determine the relative accuracy of each model. c-io TUT oos 1.333 1.1 PROJECT OBJECTIVES The primary objective of this project was to assess the relative accuracy of the Jury volatilization models using experimental emission flux data from previous studies as a reference data base. 1.2 TECHNICAL APPROACH The following series of tasks comprised the technical approach for achieving the project objectives: 1. Review the theoretical basis and development of the Jury volatilization models to verify the applicable model boundary conditions and variables, and to document model assumptions and limitations. 2. Perform a literature search and survey (not to exceed nine contacts) for the purpose of determining the availability of acceptable emission flux data from experimental and field-scale measurement studies of volatile organic compound (VOC) emissions from soils. Acceptable data must have undergone proper quality assurance/quality control (QA/QC) procedures. 3. Determine if the emission flux measurement studies referred to in Task No. 2 also provided sufficient site data as input variables to the volatilization models. Again, acceptable variable input data must have undergone proper QA/QC procedures. 4. Review, collate, and normalize emission flux measurement data and volatilization model variable data, and compute chemical-specific emission rates for comparison to respective measured emission rates. 5. Perform statistical analysis of the results of Task No. 4 to establish the extent of correlation between measured and modeled values and perform parametric analysis, of key model variables. C-ll TUT OOS 1334 S££T 800 mi zio SECTION 2 REVIEW OF THE JURY VOLATILIZATION MODELS The Jury Reduced Solution finite source volatilization model calculates the instantaneous emission flux from soil at time, t, as: s = C0 e-" (DEln f)m [l - exp (-I2/4 D£ t)] (1) where J, = Instantaneous emission flux,//g/cnf-day C0 = Initial soil concentration (total volume), A/g/cm3-soil jj = Degradation rate constant, 1/day t as Time, days DE = Effective diffusion coefficient, cm2/day L = Depth from the soil surface to the bottom of contamination, cm and, D e I/»NW3 n* If * «10/3 n«*\/^2ll/- * f ^ O ^ m V\ t2) ^f * Ks L/g KH + V Ut where DE = Effective diffusion coefficient, cm2 /day a = Soil volumetric air content, cm3/cm3 DB* = Gaseous diffusion coefficient in air, cm2/day C-13 TUT 008 133^ ^ = Henry's law constant, unitless 0 = Soil volumetric water content, cm3 /cm3 D|w = Liquid diffusion coefficient in pure water, cm2 /day 4> - Total soil porosity, unitless pb = Soil dry bulk density, g/cm3 k = Soil organic carbon fraction - Organic^carbon partition coefficient, cm3/g. The model assumes no boundary layer at the soil-air interface, no water flux through the soil, and an isotropic soil column contaminated uniformly to some depth L The initial and boundary conditions for which Equation 1 is solved are: c * Ce at f=0, 0 ^ x £ L c = 0 at f=0, x > L c = 0 at r>-0, x - 0 where c and C0 are, respectively, the soil concentration and initial soil concentration (g/cm3 - total volume), x is the distance measured normal to the soil surface (cm), and t is the time (days). The average flux over time (J^*) is computed by integrating the time-dependent flux over the exposure interval. The Jury Infinite Source volatilization model calculates the instantaneous emission flux from soil at time, t, as: TUr 1337 (3) where J, , = Instantaneous emission flux, j/g/cm2 -day C0 = Initial soil concentration (total volume), fjg/cm3 -soil t = Time, days Dfe = Effective diffusion coefficient, cm2 /day (Equation 2). The model assumes no boundary layer at the soil-air interface, no water flux through the soil, and an isotropic soil column contaminated untformly to an infinite depth. The boundary conditions for which Equation 3 is solved are: c • Ce at t £ 0, x - oo The average flux over time (J/*0) is calculated as: C0 (4 Df In 2.1 FINITE SOURCE MODEL DERIVATION The Jury Reduced Solution finite source model is derived from the methods presented by Mayer et ai. (1974), and Carslaw and Jaeger (1959). Mayer et al. (1974) considered a system where pesticide is uniformly mixed with a layer of soil and volatilization occurs at the soil surface. If diffusion is the only mechanism supplying pesticide to the surface of an isotropic soil column, and if the diffusion coefficient, Dg, is • assumed to be constant, the general diffusion equation is: C-li TUT OOS 1338 where c = Soil concentration, g/crri3 - total volume x = Distance measured normal to soil surface, cm DE = Effective diffusion coefficient in soil, cnf/d t = Time, days. If the pesticide is rapidly removed by volatilization from the soil surface and is maintained at a zero concentration, the initial and boundary conditions which also allow for diffusion across the lower boundary at x = L are identical to those of Equation 1. Recognizing the analogy between the heat transfer equation (Fourier's Law) and the transfer of matter under a concentration gradient (Pick's Law), Mayer et al. (1974) employed the heat transfer equation of Carsiaw and Jaeger (1959, page 62, Equation 14) to solve the diffusion equation given these initial and boundary conditions as: c = (Ce/2){2 erf [x!2(De r)1/2] - erf [(x-i)/2(D£fl10] - erf[( The flux is obtained by differentiating Equation 6 with respect to x, determining 0c/9x at x = 0, and multiplying by D^. The result is: Jt - D£ [8c/axL.o = \DE Cj(n D£t)m] [l -exp (-I2/4 De f)]. (7) Note that Equation 7 is equivalent to the Jury Reduced Solution given in Equation 1 with the exception of the first-order degradation expression (e*1). Jury et al. (1983 and 1990) expanded upon the work of Carsiaw and Jaeger » • (1959) and Mayer et a!. (1974) by developing an analytical solution for Equation 5 which C-16 TUT OO8 •1339 includes water flux through the soil column and a soil-air boundary layer. In addition, the Jury et al. solution also includes a theoretical approximation of the effective diffusion coefficient (Equation 2) which was not included in Mayer et al. (1974). Given these conditions, the flux equation from Jury et al. (1983) is given as: J, = - D£ (dCr/dx) + Vg CT <8) where C, = Soil total concentration x = Depth normal to soil surface VE = Effective solute convection velocity. The minus sign is used because the x direction is positive downward. Given the initial and boundary conditions: c = C0 at t=0, 0 *z x < L c = 0 at t=0, x > L c = 0 at t>0, x = 0 J, = -hQ; att>0, x = 0 where h = Transport coefficient across the soil-air boundary layer of thickness d (h = D9*/d) CG = Vapor-phase concentration The Jury et al. (1983) analytical solution for the volatilization flux is: C-17 TUT 008 134O JS(U) = +C0 V£ erfc 2(D£ f)1/2 exp - erfc 2(D£ r)m (9) exp erfc L * (2HE 2(DE f)1/2 - erfc (2H£ + Vj 2 (D£ I)' ,1/2 where HE is the transport coefficient across the boundary layer divided by the gas- phase partition coefficient, HE * h/fcb U *W^ + e/K< + a)- Jury et al. (1990) explains that compounds with large values of KH are insensitive to the thickness of the soil-air boundary layer (i.e., as HE •* »). Therefore, for the case where HE - oo and in the absence of water flux (VE = 0) Equation 9 is reduced to Equation 1 where the approximation is used to expand the error function for targe values of x (Carslaw and Jaeger, 1959). The Jury Reduced Solution given in Equation 1 is therefore a reduced form of the analytical solution given in Equation 9 for the conditions of zero water flux and no soil-air boundary layer. As such, the Jury Reduced Solution (discounting degradation) is equivalent to the Mayer et al. (1974) solution for diffusion across both the upper and lower boundaries (Equation 7). C-18 TUT oos 22 INFINITE SOURCE MODEL DERIVATION The Jury Infinite Source volatilization model (Equation 3) is derived from Mayer et al. (1974) Equations 3 and 4. Mayer et al. (1974) employed the heat transfer equation of Carslaw and Jaeger (1959, page 97, Equation 8) to solve the diffusion equation given the boundary conditions: c * Ce at r«0, 0 £ x £ L c - 0 at t>Q, x « 0 dc/dx * 0 at x « L The Mayer et al. (1974) solution for the volatilization flux is: - Dt c> 'of r)"2 1*25: (-1)" exp (~n*L*/D£ t) Ml) Therefore, Equation 11 is the analytical solution for a finite emission source, but accounts only for diffusion across the upper boundary. The summation expression in Equation 11 decreases with increasing L and decreasing DE and t. If this term is small enough to be negligible, Equation 11 reduces to: J, « Df CJ(n Df f)1/2 (12) C-19 TUT OO8 134i Use of Equation 12 will result in less than 1 percent error if t < L2/18.4 Dt (Mayer et al., 1974). Jury et al. (1984 and 1990) gave the solution for the semi-infinite case in Equation 3 where C = Qo at t ^ 0, x = oo as: >. * C. (Deln Equation 3 is equivalent to the semi-infinite solution of Mayer et al. (1974) as given in Equation 12 and provides a bounding estimate of the maximum volatilization flux but does not account for source depletion. As with Equation 12, use of Equation 3 on a finite system will result in less than 1 percent error if t < L2/18.4 D^. For the purposes of calculating SSLs based on volatilization from soils, let t be set equal to the exposure interval. If t > L2/18.4 0^, Equation 1 should be used to calculate the volatilization factor. As an alternative, an estimate of the average emission flux over the exposure interval, < J, > , can be obtained from a simple mass balance: <Jt> = CeLlt O3) where Cc = Initial soil concentration (total volume), j/g/crrf-soil L = Depth from soil surface to the bottom of contamination, cm t = Exposure interval, days. 2.3 SUMMARY OF MODEL ASSUMPTIONS AND LIMITATIONS The Jury Reduced Solution finite source volatilization model is analogous to the mathematical solution for heat flow in a solid such that the region 0 < x < L is initially at constant temperature, the region x > L is at zero, and the surface x = 0 is maintained at zero for t > 0 (Carslaw and Jaeger, 1959). As such, the model's C-20 • . ' ' TUT 008 1343 applicability to diffusion processes is limited to the initial and boundary conditions upon which the model is derived. The following represents the major model assumptions for these conditions: 1. Contamination is uniformly incorporated from the soil surface to depth L 2. The soil column is isotropic to an infinite depth (i.e., uniform bulk density, soil moisture content, porosity and organic carbon fraction). 3. Liquid water flux is zero through the soil column (i.e., no leaching or evaporation). 4. No soil-air boundary layer exists. 5. The soil equilibrium liquid-vapor partitioning (Henry's law) is instantaneous. 6. The soil equilibrium adsorption isotherm is instantaneous, linear, and reversible. 7. Initial soil concentration is in dissolved form (i.e., no residual-phase contamination). 8. Diffusion occurs simultaneously across the upper boundary at x = 0 and the lower boundary at x = L The model is therefore limited to surface contamination extending to a known depth and cannot account for subsurface contamination covered by a layer of clean soil. Also, the model does not consider mass flow of contaminants due to water movement in the soil nor the volatilization rate of nonaqueous-phase liquids (residuals). Finally, the model does not account for the resistance of a soil-air boundary layer for contaminants with low Henry's law constants. The Jury Infinite Source volatilization model is analogous to the mathematical solution for heat flow hi a semi-infinite solid. The major model assumptions are the same as those of the Jury Reduced Solution finite source model except that/the contamination is assumed to be uniformly incorporated from the soil surface to an infinite depth, and that diffusion occurs only across the upper boundary. C-21 TUT OO8 1344 In general, both models describe the vapor-phase diffusion of the contaminants to the soil surface to replace that lost by volatilization to the atmosphere. Each model predicts an exponential decay curve over time once equilibrium is achieved. In actuality, there is a high initial flux rate from the soil as surface concentrations are depleted. The lower flux rate characteristics of the latter portion of the decay curve are thus determined by the rate at which contaminants diffuse upward. This type of desorption curve has been well documented in the literature. It is important to note that both models do not account for the high initial rate of volatilization before equilibrium is attained and will tend to underpredict emissions during this period. Finally, each model is most applicable to single chemical compounds fully incorporated into isotropic soils. Effective solubilities and activity coefficients in multicomponent systems are not addressed in the determination of the effective diffusion coefficient nor is the effect of nonlinear soil adsorption and desorption isotherms. However, because of the complexities involved with theoretical solutions to these effects, their contribution to model accuracy is difficult to predict, especially in multicomponent systems. C-22 TUT 008 134g SECTIONS MODEL VALIDATION To achieve the project objective, EQ executed a literature search and a survey of professional environmental investigation/research firms as well as regulatory agencies to obtain experimental and field data suitable for comparing modeled emissions with actual emissions. The literature search uncovered several papers and bench-scale . experimental studies concerned with the volatilization and vapor density of pesticides and chlorinated organics incorporated in soils (Farmer et al., 1972, 1974, and 1980; Spencer and Cltath, 1969 and 1970; Spencer, 1970; and Jury et al., 1980). 3.1 VALIDATION OF THE JURY INFINITE SOURCE MODEL From the literature search, one bench-scale study was found that approximated the boundary conditions of the Jury Infinite Source model and met the data requirements for this project, Farmer et al., (1972). The Farmer et al. (1972) study reports the experimental emissions of lindane (1,2,3,4,5,6-hexachlorocydohexane, gamma isomer) and dieldrin (1,2,3,4,10,10-hexachloro-6,7-epoxy-1,4,4a,5,6,7,8,8a- octahydro-1,4-endo, exo-5, 8-dimethanonapthalene) incorporated in Gila silt loam. The objective of the survey of professional firms and regulatory agencies was to find pilot-scale or field-scale studies of volatilization of organic compounds using the U.S. ERA emission isolation flux chamber. The candidate flux chamber studies must also have provided adequate data for input to the volatilization models. . Flux chamber studies were chosen to provide pilot-scale or field-scale measurement data needed for model validation. Flux chambers have been widely used to measure flux rates of VOCs and inorganic gaseous pollutants from a wide variety of sources. The flux chamber was originally developed by soil scientists to measure C-23 TUT 008 1346 biogenic emissions of inorganic gases and their use dates back at least two decades (Hill et al., 1978). In the early 1980's, EPA became interested in this technique for estimating emission rates from hazardous wastes and funded a series of projects to develop and evaluate the flux chamber method. The initial work involved the development of a design and approach for measuring flux rates from land surfaces. A test cell was constructed and parametric tests performed to assess chamber design and operation (Kienbusch and Ranum, 1986 and Kienbusch et al., 1986). A series of field tests were performed to evaluate the method under field conditions (Radian Corporation, 1984 and Balfour, et al., 1984). A user's guide was subsequently prepared summarizing guidance on the design, construction, and operation of the EPA recommended flux chamber (Keinbusch, 1985). The emission isolation flux chamber is presently considered the preferred in-depth direct measurement technique for emissions of VOCs from land surfaces (EPA, 1990). EQ contacted several environmental consulting firms as well as State and local agencies. In addition, the EPA data base of emission flux measurement data was reviewed (EPA, 1991 a). Although several flux measurement studies were found, only one applicable study was identified with adequate QA/QC documentation and the necessary input data for the Jury Infinite Source model (Radian Corporation, 1989). From Farmer et al. (1972) the influence of pesticide vapor pressure on volatilization was measured by comparing the volatilization from Gila silt loam of dieidrin with that of lindane. Volatilization of dieidrin and lindane was measured in a closed air- flow system by collecting the volatilized insecticides in ethylene glycol traps. Ten grams of soil were treated with either 5 or 10//g/g of C-14 tagged insecticide in hexane. The hexane was evaporated by placing the soils in a fume hood overnight Sufficient water was then added to bring the initial soil water content to 10 percent. For the volatilization studies, the treated soil was placed in an aluminum pan 5 mm deep, 29 mm wide, and 95 mm long. This produced a bulk density of 0.75 g/cm3. The aluminum pan was then introduced into a 250 mL bottle which served as the volatilization chamber. A relative humidity of 100 percent was maintained in the incoming air stream to prevent water evaporation from the soil surface. Air flow was C-24 °OS 1347 maintained at 8 mL/s equivalent to approximately 0.018 miles per hour. The temperature was maintained at 30° C. The soil was a Gila silt loam, which contained 0.58 percent organic carbon. The volatilized insecticides were trapped in 25 ml of ethylene glycol. Insecticides were extracted into hexane and anhydrous sodium sutfate was added to the hexane extract to remove water. Aliquots of the dried hexane were analyzed for lindane and dieldrin using liquid scintillation. The extraction efficiencies for lindane and dieldrin were 100 and 95 percent, respectively. The concentrations of volatilized compounds were checked using gas-liquid chromatography. All experiments were run in duplicate. To ensure that the initial soil concentrations of lindane and dieldrin were in dissolved form, the saturation concentration (mg/kg) of both compounds under experimental conditions was calculated using the procedures given in U.S. EPA (1994): — (^ KOC P* * 9 «• KH a) (14) Pt> where S is the pure component solubility in water. C^ for lindane and dieldrin were calculated to be 34 mg/kg and 12 mg/kg, respectively. Therefore, the initial soil concentrations of 10 and 5 mg/kg were below saturation for both compounds. Table 1 gives the values of each variable employed to calculate the emissions of lindane and dieldrin using the Jury Infinite Source volatilization model (Equation 3). The potential for loss of contaminant at the lower boundary at each time-step was checked to see if t > L2/18.4 Dfe. If this condition was true at any time-step, the boundary conditions of the infinite source model were violated. In such a case, emissions were also calculated using the finite source model of Mayer et al. (1974) as presented in Equation 11. The difference between the predictions of both models were compared at each time-step and a percent error was calculated for the infinite source model. The C-25 TUT 008 1348 TABLE 1. VOLATILIZATION MODEL INPUT VALUES FOR UNDANE AND DIELDRIN Variable Initial soil concentration Soil depth Soil dry bulk density Soil particle density Gravimetric soil moisture content Water-filled soil porosity Total soil porosity Air-filled soil porosity Soil organic carbon Organic carbon partition coefficient Diffusivity in air (Undane) DifUsivity in air (Dieidrin) Diffusivity in water (Undane) Diffusivity in water (Dieidrin) Henry's law constant (Undane) Henry's law constant (Dieidrin) Degradation rate constant (Undane and Dieidrin) Symbol C0 L Pb P. w e <t> a te K* V D8' or or KH K« // Units mg/kg cm g/cm3 g/cm3 percent cm5 /cm* cm3 /cm3 cm /crrr fraction crriVg cnf/d cnf/d cnf/d cnf/d unitless unitless 1/day Value 5 and 10 0.5 0.75 2.65 10 0.075 0.717 0.642 0.0058 1380 1521 1080 0.480 « 0.410 1.40 E-04 2.74 E-06 0 Reference/equation Farmer et al. (1972) Farmer et al. (1972) Farmer et al. (1972) U.S. EPA (1988) Farmer et al. (1972) *Pb 1-fo»/P.) 0 - e Farmer et al. (1972) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994a) U.S. EPA (19943) U.S. EPA (1994) U.S. EPA (1994) Default to eliminate effects of degradation C-26 TUT 1 .....Cv instantaneous emission flux values predicted by Equation 3 and Equation 11 (where applicable) were plotted against the measured flux values for dieldrin and lindane at both 5 and 10 ppmw. Figure 1 shows the comparison of the predicted and measured values of dieldrin at an initial soil concentration of 5 ppmw. For dieldrin, the boundary conditions of the infinite source model were not violated until the last time-step. A best curve was fit to both the measured and predicted values. As expected, both curves indicate an exponential decrease in emissions with time. The ratio of the modeled emission flux to the measured emission flux was determined as a measure of the relative difference between the modeled and measured values. The natural log of this ratio was then analyzed by using a standard paired Student's t-test This analysis is equivalent to assuming a tognormal distribution for the emission flux and analyzing the log-transformed data for differences between modeled and measured values. The data were also analyzed by using standard linear regression techniques (Figure 2). Again, the data were assumed to follow a lognormal distribution. A simple linear regression model was fit to the log-transformed data and the Pearson correlation coefficient was determined. The Pearson correlation coefficient is a measure of the strength of the linear association between the two variables. From a limited population of four observations, the correlation coefficient was v. calculated to be 0.994 with a mean ratio of modeled-to-measured values of 0.42. The actual significance (p-vaiue) of the paired Student's t-test was p = 0.0001. The lower and upper confidence limits were calculated to be 0.38 and 0.48, respectively. On average, this indicates that at the 95 percent confidence limit, the modeled emission flux is between 0.38 and 0.48 times the measured emission flux. Figure 3 shows the modeled and measured flux values of dieldrin at an initial soil concentration of 10 ppmw, while Figure 4 shows the relationship of the log-transformed data and the upper and lower confidence limits. At 10 ppmw, the correlation coefficient was 0.974 with a mean ratio of 0.45, p-value of 0.0001, and a 95 percent confidence interval of 0.37 to 0.54. C-27 TUT 1350 H •-••4 cn O s> oe 0.25 0.2 0 DIELDRIN (Initial Soil Cone.« 8 mgflcol • Measured flux o Infinite source model predicted flux 50 100 150 200 TIME FROM SAMPLING (t), hrs 250 Figure 1. Predicted and measured emission flux of dieldrin versus time (C0 = 5 ppmw). -1- -2- 5c -3- r> K> -4 T—i—i—i—i—i—i—i—|—r—i—i—i—i—i—i—i—i—|—i—i—i—i—i—i—i—i—r 4 5 Ln Time 000 Jury Model Predicted Flux ODD Measured Flux Figure 2. Comparison of log-transformed modeled and measured emission flux of dieMrin (C0 = 5 ppmw). o 00 Ui U) O 0.45 0.4 0.35 ^0.25 G! 0.2 0.15 DIELDRIN (Initial Soil Cone. • 10 mg/kg) 50 o Infinite source model predicted flux • Measured flux 100 150 200 TIME FROM SAMPLING (t), hrs Figure 3. Predicted and measured emission flux of dleldrln versus time (C0 = 10 ppmw). o- -1- 3c -2- n U) -3- I——I——t——I——|——I——I——I——I——I——I——I——T——I——j——1——I——I——I——I——I——I——I——T 4 5 Inline 000 Jury Model Predicted Flux ODD Measured Flux Figure 4. Comparison of log-transformed modeled and measured emission flux of dieldrin (C0 = 10 ppmw). As can be seen from Figures 1 and 3, the model underpredicts the emissions during the initial stages of the experiment. This is to be expected in that during this phase, contaminant is evaporating from the soil surface. The apparent discrepancy -between measured and predicted values decreases with time as equilibrium is achieved .and diffusion becomes the rate-limiting factor. • For lindane, the boundary conditions of the infinite source model were violated after the first time-step (i.e., t > L2/18.4 D^ at 24 hours). Therefore, the Mayer et al. (1974) finite source model was used to derive a percent error at each succeeding time- step. At an initial soil concentration of 5 ppmw, the infinite source model predicted 114 percent total mass loss of the finite source model over the entire time span of the experiment. At a concentration of 10 ppmw, the infinite source model predicted 107 percent total mass loss of the finite source model. Rgures 5 and 6 show the comparison of modeled to measured values of lindane at initial soil concentrations of 5 and 10 ppmw, respectively. Likewise, Figures 7 and 8 show the comparisons of the log-transformed data. At an initial soil concentration of 5 ppmw, the correlation coefficient between modeled and measured values was 0.997 with a mean modeled-to-measured ratio of 0.81, a p-value of 0.3281, and a 95 percent confidence interval of 0.46 to 1.44. At an initial soil concentration of 10 ppmw, the correlation coefficient was calculated to be 0.998, the mean ratio 0.73, the p-value 0.1774, and the confidence interval 0.41 to 1.28. The p-values for dieldrin are considerably lower than those of lindane. This is due to the very narrow confidence interval around the modeled values. In the case of dieldrin, Equation 3 did not predict a loss of contaminant at the lower boundary until the last time-step (i.e., t > L2/18.4 D^ at 12 days). This results in a nearly perfect straight line when the log-transformed data are plotted. For dieldrin, therefore, Equations 3 and 11 predict identical values until the last time-step. Table 2 summarizes statistical analysis for the bench-scale comparative validation of the Jury Infinite Source volatilization model. In general, the data support good agreement between modeled and measured values and show relatively narrow confidence intervals and high correlation coefficients. C-32 TUT 008 1355 cH CO O 0 LINDANE {Initial Soil Cone. - 6 mg/kg) o Infinite source model predicted flux A Finite source model predicted flux 60 80 100 120 TIME FROM SAMPLING (t), hre Figure 8. Predicted and measured emission flux of flndana veraua time (C0 '»• 8 ppmw). LINDANE (Initial Cone.•10 ppmw) c. o CO I-*- i>i i'f; O o Infinite source model predicted flux Finite source model predicted flux 60 80 100 120 TIME FROM SAMPUNO (t), hrs Figure e. Predicted and measured emission flux o! llndane versus time (C0 * 10 ppmw). H 03 0- -1- g C _ -3- o UJ -4- | I I I | I I I | I I I [ I I I | I I I | I I I | I I I | I I I [ I l I | I I I | I I I | I I I | I I I | I I I | I I I | I I I | I I I | I I I | I I I | I I I | I 3.1 32 33 3.4 3.S 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.6 4.9 5.0 5.1 In Time 000 Jury Model Predicted Flux ODD Measured Flux Figure 7. Comparison of log-transformed modeled and measured emission flux of Itndane (CB = 5 ppmw). 1- I -1- -2- n -3-jTTTyrTTjrn|iii j i . . | . . . j . . • | i i i | . . . j . . . j i i . j i i i j i . i j i i i | i i rj n i j , , , j i i^p-TTyrTryi 3.1 32 3.3 3.4 3.5 3.8 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.S 4.6 4.7 4.8 4.9 5.0 5.1 LnThro 000 Jury Model Predicted Phix ODD Measured Flux Figure 8. Comparison of log-transformed modeled and measured emission flux of llndane (C0 = 10 ppmw). TABLE 2. SUMMARY OF THE BENCH-SCALE VALIDATION OF THE JURY INFINITE SOURCE MODEL Chemical LJndane (5 ppmw) Lindane (10 ppmw) Dieldrin (5 ppmw) Dieidrin (10 ppmw) N 4 4 7 7 Correlation coefficient 0.997 0.998 0.994 0.974 Mean ratio: Modeled-to- rneasured 0.81 0.73 0.42 0.45 p-value 0.3281 0.1774 0.0001 0.0001 95% confidence interval (0.46, 1.44) (0.41, 1.28) (0.38, 0.48) (0.37, 0.54) Appendix A contains the spreadsheet calculations for the bench-scale validation of the Jury Infinite Source volatilization model. From Radian Corporation (1989), a pilot-scale study was designed to determine how different treatment practices affect the rate of loss of benzene, toluene, xylenes, and ethylbenzene (BTEX) from soils. The experiment called for construction of four piles of loamy sand soil, each with a volume of approximately 4 cubic yards (7900 pounds), a surface area of 8 square meters, and a depth of 0.91 meters. Each test cell was lined with an impermeable membrane and the soil in each cell was sifted to remove particles larger than three-eighth inch in diameter. The contaminated soil for each pile was prepared in batches using 55-gallon drums. In the "high level" study, each soil batch was brought to 5 percent moisture content and 6 liters of gasoline added. Additional water was then added to bring the soil to 10 percent moisture by weight. The drums were capped and sat undisturbed overnight The drums were then opened the next day and shoveled into the test cell platform. Twenty-two soil batches were prepared for each soil pile. Each batch consisted of 360 pounds of soil and 6.0 liters of fuel. Therefore, each soil pile contained 7900 pounds of soil and 132 liters of gasoline. Each soil pile was then subjected to one of the following management practices: 0 A control pile that was not moved or treated C-37 TUT OOS 1360 0 An "aerated" or "mechanically mixed" pile 0 A soil pile simulating soil venting or vacuum extraction 0 A soil pile heated to 38° C. Losses due to volatilization during the mixing and transfer process and during a • 28 hour holding time in the test bed before initial sampling reduced the residual BTEX in soil. For the purpose of this validation study, however, these losses caused initial soil rorcerrtrations of benzene, toluene, and ethylbenzene to be below or within a factor of two of their respective single component saturation concentrations. Because the mixed pile, vented pile, and heated pile were subject to mechanical disturbances or thermal treatment, only the control pile data were used in this study. In general, the test schedule called for collection of soil samples and air emission toss measurements during the first, sixth, and seventh weeks. Soil samples were collected randomly within specified grid areas by composite core collection to the maximum depth of the pile. Emission losses were measured similarly using an emission isolation flux chamber as specified in Kienbusch (1985). Only data for which soil samples and flux chamber measurements were taken on the same day were used for this study. Analysis of BTEX in soil samples was accomplished by employing the EPA 5030 extraction method and the EPA 8020 analytical method. The BTEX method was modified to reduce the sample hold time to one day in an .effort to improve the accuracy of the method. Five soil samples were submitted in duplicate. The relative percent differences (RPD) ranged from 8.0 to 48.9 percent. The average RPD for the five samples was 26.8 percent In addition, EPA QC sample analysis indicated average percent recoveries ranging from 89 percent for m-xylene to 119 percent for toluene. The pooled coefficient of variation (CV) for all the BTEX analysis was 10.5 percent. Spiked sample recoveries (eight samples) ranged from 75 percent for m-xylene to 168 percent for toluene. The average spike recoveries ranged from 108 percent for benzene to 146 percent for toluene. Finally, both system blanks and reagent blanks indicated no contamination was found in the analytical system. C-38 1361 It should be noted that the standard method used for BTEX analysis was observed to have contributed to the variabilities in soil concentrations. The EPA acceptance criteria based on 95 percent confidence intervals from laboratory studies are roughly 30 to 160 percent for the BTEX compounds during analysis of water samples. The necessary extraction step for soil samples would increase this already . large variability. Analysis of vapor-phase organic compounds via the emission isolation flux chamber was accomplished using a gas chromatograph (GC). Gas samples were collected from the flux chamber in 100 ml, gas-tight syringes and analyzed by the GC hi laboratory facilities adjacent to the test site. During the study, a multicomponent standard was analyzed daily to assess the precision and daily replication of the analytical system. The results of the analysis Indicated a good degree of reproducibility with coefficients of variation ranging from 5.1 to 16.3 percent From these data, instantaneous emission fluxes were calculated for benzene, toluene, and ethylbenzene corresponding to each time period at which flux chamber measurements were made. Table 3 gives the values of each variable employed to calculate emissions of each compound using the Jury infinite Source model and the Mayer et al. (1974) finite source model. Appendix A contains the spreadsheet data for benzene, toluene, and ethylbenzene at initial soil concentrations of 110 ppm, 860 ppm, and 310 ppm, respectively. It should be noted that the fraction of soil organic carbon ($*) was not available from Radian (1989). For this reason, the default value for tc of 0.006 from U.S. EPA (1994) was used for ail calculations. Figures 9,10, and 11 show the comparison of modeled and measured emission fluxes of benzene, toluene, and ethylbenzene, respectively. The Radian Corporation study noted that the second measured value in each figure represented a data outlier, possibly due to the formation of a soil fissure, reducing the soil path resistance and increasing the emission flux. Table 4 presents the results of the statistical analysis of the comparison of modeled and measured values. For both benzene and ethylbenzene, measured values C-39 TUT OOB TABLE 3. VOLATILIZATION MODEL INPUT VARIABLES FOR BENZENE, TOLUENE, AND ETHYLBENZENE Variable Initial soil concentration - benzene • toluene • ethylbenzene ?:•:••! depth &-« dry bulk density Soil particle density Gravimetric soil :. me nure content Water-filled soil porosity Total soil porosity Air-filled soil porosity Soil organic carbon Organic carbon partition coefficient - benzene - toluene - ethylbenzene Diffusivity in air - benzene • toluene - ethylbenzene Diffusivity in water - benzene - toluene - ethylbenzene Symbol C0 L Pt P, w e 0 a U K* V or Units mg/kg cm g/cm3 g/cm3 Percent cnr/cnr. cnT/crrr crrr/crrr Fraction cm3/g crrf/s crrf/s Value 110 880 310 91 1.5 2.65 10 0.150 0.434 0.284 0.006 57 131 221 0.0870 0.0870 0.0750 9.80 E-06 8.60 E-06 8.64 E-06 Reference/equation Radian (1989) Radian (1989) Radian (1989) U.S. EPA (1988) Radian (1989) v*,, 1-fcb/P.) 0 -e U.S. EPA (1994) default value U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994a) U.S. EPA (1994a) U.S. EPA (19943) (continued) C-40 TUT 003 TABLE 3 (continued) Variable Henry's law constant • benzene - toluene - ethylbenzene Degradation rate constant Symbol K, V Units Unitless 1/day Value 0.22 0.26 0.32 0 Reference/equation U.S. EPA (1994) U.S. EPA (1994) U.S. EPA (1994) Default to eliminate effects of degradation C-41 TUT 003 1364 o-- 'Jl o 1400 BENZENE (Initial Soil Cone. = 110 mg/kg) o Infinite source model predicted flux A FlnHe source model predicted flux 200 300 400 600 TIME FROM SAMPLING (t), hrs 600 700 Figure 9. Predicted and measured emission flux of benzene (C0 = 110 ppmw). o 00 &-o 8000 7000 6000 1 6000 4000 3000 2000 1000 TOLUENE (Initial Soil Cone. • 880 mg/lcg) o Infinite source model predicted flux A Finite source model predicted flux 200 400 800 800 TIME FROM SAMPLING (t), hrs 1000 1200 Figure 10. Predicted and measured emission flux of toluene (C, = 880 ppmw). Nj 2500 2000 I dL 3I 1500 1000 w2 ETHYLBENZENE (Initial Soil Cone. * 310 ppmw) o Infinite source model predicted flux; A Finite source model predicted flux 300 400 TIME FROM SAMPLING (t), hre Figure 11. Predicted and measured emission flux of ethyfbenzene (C0 = 310 ppmw). TABLE 4. SUMMARY OF STATISTICAL ANALYSIS OF PILOT-SCALE VALIDATION Chemical Benzene (110 ppm) Toluene (880 ppm) Ethylbenzene (310 ppm) N 5 7 5 Correlation coefficient 0.982 0.988 0.999 Mean ratio: Modeled-to measured 2.5 6.3 7.8 p-value 0.0149 0.0002 0.0008 95% confidence interval (1.4,4.5) (3.9, 10.4) (4.9, 12.4) were below the detection limits after the fifth observation; measured values for toluene were below the detection limit after the seventh observation. Figures 12,13, and 14 show the comparison of the log-transformed data for the modeled and measured emission fluxes of benzene, toluene, and ethylbenzene, respectively. As can be seen from Table 4, correlation coefficients ranged from 0.982 for benzene to 0.999 for ethylbenzene, while p-values and 95 percent confidence intervals indicate a significant statistical difference between modeled and measured values. The boundary conditions of the infinite source model were violated after the first time-step for benzene, and after the third time-step for both toluene and ethylbenzene. The infinite source model predicted 134 percent, 117 percent, and 103 percent of the total mass loss of the finite source model for benzene, toluene, and ethylbenzene, respectively. In general, the predicted values were higher than the measured values throughout the time-span of the experiment for all three compounds. It is also interesting to note that during the initial stage of the experiment the predicted values were considerably higher than measured values even when contaminant loss at the soil surface due to evaporation was expected. Although the relative differences between predicted and measured values are not excessive fi.e., the highest modeled-to- measured mean ratio is within a factor of approximately 10), they are considerably higher than those of the bench-scale studies. Any one or a combination of the following could account for the larger discrepancies between measured and predicted values in the pilot-scale study: C-45 TUT OO8 1368 8 - § C o 7- 6- 5 ™ 4 - 3 - 0 I ' ' 4 I I I I I I I LnTlim I I I I I I I 0- 000 Jwy Model Predlded fhix ODD Meawired Fhix Figure 12. Comparison of log-transformed modeled and measured emission flux of benzene (C0 = 110 ppmw). 9- ••o- 7 - n Jx -O -i—i—i—i—i—i—i—i—i—i—i—i—i—r—i—i—i—|—i—i—i—i—i—i—i—i—i—|—i—i—i—i—i—i—i—i—i—j- 4 5 8 7 In Time 000 Jury Model Predicted Flux D D D Measured Flux Figure 13. Comparison of log-transformed modeled and measured emission flux of toluene (C0 = 880 ppmw). 5 ojt 00 o 5 e 8 3 4 LnTlme 000 Jury Model Predicted Rux D G Q Measured Flux Figure 14. Comparison of log-transformed modeled and measured emission flux of ethylbenzene (C0 = 310 ppmw). 1. Although the initial soil concentrations of the three compounds were below or within B factor of two of their respective single component saturation concentrations, they may have been greater than the component concentrations for which a residual-phase of gasoline existed. If this were the case, measured emissions may have been in part due to the presence of nonaqueous-phase liquids (NAPL) which would have violated the model's assumptions of equilibrium partitioning. 2. Soil mixing processes and transfer to the test bed may have resulted in heterogenous incorporation of the contaminants. If surface con- centrations were reduced due to incomplete mixing, measured emissions would have been reduced during the initial stages of the experiment. 3. Sampling and/or analytical variability may have resulted in under reporting of emission fluxes and/or over reporting of initial soil concentrations. 4. Contaminants sorbed to the test bed liner may have acted to reduce emissions. 5. Variability in the relative humidity of the air above the test bed may have induced surface water evaporation ft between flux chamber samples. Water evaporation would have moved contaminants to the surface by convection and depleted soil concentrations in between sampling events. 6. The model is not as accurate for compounds with relatively high Henry's law constants. From these observations, it appears more likely that the larger discrepancies between modeled and measured emissions in the pilot-scale study are due.to experimental conditions. Sufficient uncertainty exists as to whether all model boundary conditions were maintained during the experiment For this reason, the results of the pilot-scale validation should be considered less reliable than those of the bench-scale validation. This conclusion suggests that controlled studies should be considered for validation of model predictions for compounds with relatively high Henry's law constants. C-49 /UT °°8 137? 3.2 VAUDATION OF THE JURY REDUCED SOLUTION FINITE SOURCE MODEL From the literature search, one bench-scale study was found that replicated the boundary conditions of the Jury Reduced Solution model (Equation 1). Jury et al. (1980) reports the emissions of the herbicide triallate [S-(2,3,3-trichloroallyl) diisopropytthiocarbamate] incorporated in San Joaquin sandy loam. This study replicated the model boundary conditions in that a dean layer of soil underlayed the contaminated soil allowing diffusion across the tower boundary as well as the upper boundary. Volatilization of triallate was measured in a dosed volatilization chamber (Spencer et al., 1979). The air chamber above the soil was 2 mm deep and 3 cm wide, matching the width of the evaporating surface. An average air flow rate of 1 liter per minute was maintained across the surface equivalent to a windspeed of 1 km/h. Triallate was applied by atomizing the material in hexane onto the air-dry autodaved soil. The soil was mixed and allowed to equilibrate in a vented fume hood. The soil was then transferred to the chamber and wetted from the bottom. To prevent water evaporation at the soil surface, the chamber was maintained at 100 percent relative humidity and a temperature of 25° C. The volatilized triallate was trapped daily on polyurethane plugs and extracted and analyzed as described in Grover et al. (1978). The volatilization of triallate at an initial soil concentration of 10 ppmw was measured over a 29 day period in the absence of water evaporation. Calculation of the saturation concentration (C^) confirmed that the initial concentration of 10 ppmw was in dissolved form. Table 5 gives the values of each variable employed to calculate emissions of triallate using the Jury Reduced Solution volatilization model. Figure 15 shows the comparison of the predicted and measured values for triallate at an initial soil concentration of 10 ppmw. The data plots indicate very good agreement between modeled and measured values. Figure 16 shows the comparison of the log-transformed data and confidence intervals. From the population of 32 observations, the correlation coefficient was calculated to be 0.998 with a mean C-50 TUT 008 TABLE 5. VOLATILIZATION MODEL INPUT VALUES FOR TRIALLATE Variable Initial soil concentration Sol! depth Soil dry bulk density Soil particle density Gravimetric soil moisture content Water-filled soil porosity Total soil porosity Air-filled soil porosity Soil organic carbon Organic carbon partition coefficient Diffusivity in air Diffush/ity in water Henry's law constant Degradation rate constant Symbol 0, L Pb P* w e <f> a to K* D0' Pw KH V Units mg/kg cm g/cm3 g/cm3 percent cm /cm cm /crrr cm3 /cm 3 fraction cnf/g crrf/d crrf/d unitless 1/day Value 10 10 1.34 ., 2.65 21 0.279 0.494 0.215 0.0072 3600 3888 0.432 1.04E-03 0 Reference/equation Jury et al. (1980) Jury et al. (1980) Jury et al. (1980) U.S. ERA (1988) Calculated from Jury et al. (1980) Jury et al. (1980) Jury et al. (1980) Jury et al. (1980) Calculated from Jury et al. (1980) Jury et al. (1990) Jury et al. (1980) Jury et al. (1980) Jury et al. (1980) Default to eliminate effects of degradation C-51 TUT 008 TRIALLATE (Initial Cone.•10ppmw) n •tn K) C H 00 xl Ul o Jury model predicted flux 2QO 300 400 600 TIME FROM SAMPLING (t), hr* Figure 15. Predicted and measured emission flux of trlaflate versus time. 1- o- c -i 5 -2- -3- I I I I I I I I I I [ I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | 1 2 3 4 5 6 7 LnTlme 0 0 Q Jury Model Predicted Flux DOB Measured Flux xi Figure 16. Comparison of log-transformed modeled and measured emission flux of triallale. modeled-to-measured ratio of 1.11. The p-value was calculated at 0.0001, and the confidence interval was 1.07 to 1.16. The degree of agreement between modeled and measured emission flux values for triallate may be due to soil adsorption studies conducted to experimentally derive the organic carbon partition coefficient specific to the San Joaquin sandy loam used in the experiment. With experimentally derived values of K^, more accurate phase partitioning was possible resulting in an experimental-specific value of the effective diffusion coefficient (Equation 2). Appendix B contains the spreadsheet calculations for the bench-scale validation of the Jury Reduced Solution finite source volatilization model. C-54 TUT SECTION 4 PARAMETRIC ANALYSIS OF THE JURY VOLATILIZATION MODELS This section presents the results of parametric analysis of the key variables of the Jury volatilization models (Equations 1 and 3). The Jury volatilization models are applicable for the case of no boundary layer resistance at the soil-air interface and no water flux through the soil column. Because the models are equivalent to the Mayer et al. (1974) solutions to the general diffusion equation (Equation 5), the parametric observations of Mayer et al. (1974) and Farmer, et al. (1980) are also directly applicable. Jury et al. (1983) established the relationship between vapor and solute diffusion and adsorption by defining total phase concentration partitioning as it relates to the , effective diffusion coefficient. The effective diffusion coefficient is a theoretical expression of the combination of soil parameters and chemical properties which govern the rate at which soil contaminants move to the surface to replace those lost by evaporation. As such, the effective diffusion coefficient is the rate-limiting factor governing the general diffusion equation in soils given the initial and boundary conditions for which the models are applicable. The remainder of this section discusses the key soil and nonsoil parameters used in the expression of the effective diffusion coefficient and the general diffusion equation. 4.1 AFFECTS OF SOIL PARAMETERS In this section, the experimental results of Farmer, et al. (1980) are discussed as they relate to the effect of soil water content, soil bulk density, air-filled soil porosity, and temperature on diffusion in soil. C-55 TUT 008 Soil Moisture Content Farmer, et at. (1980) indicates that the effect of soil moisture content on the volatilization flux of contaminants through soils is exponential. Increasing soil water content decreases the pore spaces available for vapor. Diffusion and will decrease volatilization flux. In contrast, increasing soil water content has also been shown to increase the volatility of pesticides in soil under certain conditions (Gray, et al.,. 1965; and Spencer and Cliath, 1969 and 1970). In essence, the soil water content affects the contaminant adsorption capacity by competing for soil adsorption sites. Under these conditions, an increase in soil moisture above a certain point will tend to desorb contaminants, increasing the flux dependent on the relative water and contaminant adsorption isotherms. Bulk Density Soil compaction or bulk density also determines the porosity of soil and thus affects the diffusion through the soil. Experimental results from Farmer et al. (1980) indicate that soil bulk density also has an exponential effect on volatilization flux through the soil. From previous considerations of the effect of soil water content, a higher bulk density will have similar effects to that of an increased soil moisture content. Soil Air-Filled Porosity The effects of soil water content and soil bulk density on volatilization can be contributed to their effect on the air-filled porosity, which in turn is the major soil factor controlling volatilization. The effect of air-filled porosity is manifested in the expression of the effective diffusion coefficient. The effective diffusion coefficient, however, does not depend only on the amount of air-filled pore space. The presence of liquid film on the solid surfaces not only reduces porosity, but also modifies the pore geometry increasing tortuosity and the length of the gas passage. The Jury et al. (1983) expression of the effective diffusion coefficient uses the model of Millington and Quirk (1961) to account for the porosity and the tortuosity of soil as a porous medium. C-56 TUT OOS .|,37'9 Soil Temperature The effect of soil temperature on the volatilization flux is multifunctional. The diffusion in air, DB*, is theoretically related to temperature, T, and the collision integral, Q, in the following manner (Lyman, et al., 1990): g (proportional to) - . (15) The exponential coefficient for temperature varies from 1.5 to 2 over a wide range of temperatures. Barr and Watts (1972) found that 1.75 gave the best values for gaseous diffusion. Farmer, et al. (1980) estimates the .effective diffusion coefficient at temperature T2 as : 0,-D, (7,/Tj06 <16) where Dj = Diffusion coefficient at T2 D, = Diffusion coefficient at T, T = Absolute temperature. A temperature increase will effect the vapor pressure function of the Henry's Law constant, which causes ah increase in the vapor concentration gradient across the soil layer. In actual fact, temperature gradients will exist across the soil due primarily to seasonal variations. Vapor diffusion is influenced by such gradients; however, these effects of fluctuating soil temperatures will tend to cancel one another over time. C-57 TUT COS 138O 4.2 AFFECTS OF NONSOIL PARAMETERS The nonsoil variables in the Jury volatilization models include the initial soil concentration, C0, the Henry's law constant (K,), the.soil/water partition coefficient, (^>) and the depth of contaminant incorporation (L). • > . Initial Soil Concentration The effect of change in the initial soil concentration is linear; i.e., an increase in Q, of 100 percent causes an increase in the emission rate of 100 percent Probably the greatest degree of uncertainty in the value of Q, is likely to be either insufficient soil sampling to adequately characterize site soil concentrations, or the variability in percent recovery of contaminants as it applies to existing sampling and analysis methods for organic compounds in soils. Typically, present extraction and analysis method recovery variability increases the likelihood of underprediction of the emission rate (i.e., more contaminant is present in the soil than is reported by sampling and analysis methods). Henry's Law Constant and Soil/Water Partition Coefficient Jury et al. (1984) showed that a given chemical can be grouped into three main categories depending on the ratio ^>/KH. These categories are defined as a function of which phase dominates diffusion. A Category I chemical is dominated by the vapor- phase, a Category III chemical by the liquid-phase, and Category II chemicals by vapor- phase diffusion at low soil water content and liquid-dominated at high water content. Desorption from the solid-phase to the liquid-phase is a function of the soil/water partition coefficient, while volatilization from the liquid to the vapor-phase is a function of. the Henry's law constant. Therefore, the interstitial vapor density, and thus emission flux, is directly proportional to \<* and inversely proportional to KD. Because the Jury volatilization models do not account for a soil-air boundary layer, the effects of K^ and H^j are exponential for all three categories of chemicals. C-58 . • TUT 008 1381 Depth of Contaminant Incorporation The Jury Reduced Solution finite source model accounts for diffusion across both the upper and lower boundaries. Therefore for chemicals with high effective diffusion coefficients, the residual soil concentration wiU decrease rapidly, in this regard, the emission flux curve will become asymptotic more rapidly than for the semi- infinite case (Equation 3). The exponential term [1 - exp (~L2/4 Dyt)] in Equation 1 accounts for diffusion across the lower boundary such that the term.decreases rapidly with time for small values of L and large values of Dfe. C-59 TUT COS This page left blank on purpose. C-60 TUT OOS 1383 SECTIONS CONCLUSIONS * * From the results of this study, it can be concluded that for the compounds included in the experimental data, both models showed good agreement with measured data given the conditions of each test. Each model demonstrated superior agreement with bench-scale measured values and to a lesser extent the infinite source model with pilot-scale data. The results indicate high correlation coefficients across all experimental data with mean modeled-to-rr cssured ratios as low as 0.37 and as high as 7.8. From a review of test conditions, it was concluded that the bench-scale studies better approximated the initial and boundary conditions of the infinite source model. This is evident in the lower modeled-to-measured mean ratios and narrow 95 percent confidence intervals. Although the pilot-scale study data showed reasonable agreement with predicted values, questions remain as to whether the test conditions were in agreement with model assumptions and accurately replicated all model boundary conditions. Overall, each model provided reasonably accurate predictions. Clearly, this validation study is limited by the range of conditions simulated, the assumptions under which the models operate, and the initial and boundary conditions of each model. Important limitations include: 1. The duration of the experiments examined range from 7 to 36 days. Model performance for longer periods could not be validated. . 2. Both models assume no mass flow of contaminants due to water movement in the soil. Mass flow due to capillary action or redistribution of contaminates due to rain events may be significant if applicable to site- specific condition. C-61 TUT 008 1384 3. . The models are valid only if the effective diffusion coefficient in soil is constant. This assumes isotropic soils and completely homogeneous incorporation of contaminants. In reality, soils are usually heterogeneous, with properties that change with depth (e.g., fraction of organic carbon, water content, porosity, etc.). The user will need to carefully consider the characterization of soil properties before assigning model input parameters. 4. The equilibrium partitioning relationships used in the models are no longer valid for pure-phase chemicals or when high dissolved concentrations are present. Therefore, the models should not be used when these conditions exist 5. The models do not consider the effects of a soil-air boundary layer on the volatilization rate. For chemicals with Henry's law constants less than approximately 2.5 x 10~5, volatilization is highly dependent on the thickness of the boundary layer .(Jury et al., 1984). A boundary layer will restrict volatilization if the maximum flux through the boundary layer is small compared to the rate at which the contaminant moves to the surface. In this case, the volatilization rate is inversely proportional to the boundary layer thickness. 6. In the case of the infinite source model, validation for chemicals with relatively high Henry's law constants requires that the depth of contamination be sufficient to prevent loss at the lower boundary over the duration of the experiment, i.e., L > (18.4 D^ t)"2. Although this study indicates that, the Jury Infinite Source model exhibited a relatively small maximum error (i.e., 134% of the Mayer et al. finite source .model total mass loss for benzene), any future validation studies should maintain a sufficient depth of incorporation to prevent violation of the model boundary conditions. 7. No experimental data could be found in the literature for validation of the Jury Reduced solution finite source model for compounds with high Henry's law constants. Emission rates predicted by the Jury Infinite Source volatilization model and the Jury Reduced Solution finite source volatilization model indicate good correlation to measured emission rates under controlled conditions, but predicted values for field conditions would be subject to error because the boundary conditions and environmental conditions are not as well defined as they are in the laboratory. Nonetheless, results of this study indicate that both models should make reasonable C-62 TUT COS 1385 estimates .of loss through volatilization at the soil surface given the boundary conditions of each model. C-63 TUT OO8 1386 This page left blank on pnrpose. C-64 . ' TUT 008 1387 REFERENCES Balfour, W. D., B. M. Eklund, and S. J. WDIiamson. Measurement of Volatile Organic Emissions from Subsurface Contaminants. In Proceedings of the National Conference on Management of Uncontrolled Hazardous Waste Sites. September. 1984, pp. 77-81. Hazardous Materials Control Research Institute, Silver-Springs, Maryland. Barr, R. F. and H. F. Watts. 1972. Diffusion of Some Organic and Inorganic Compounds in Air. J. Chem. Eng. Data 17:45-46. Carslaw, H. S., and J. C. Jaeger. 1959. Conduction of Heat in Solids. 2* Edition Oxford University Press, Oxford. Environmental Quality Management, Inc. 1994. A Comparison of Soil Volatilization Models in Support of Superfuhd Soil Screening Level Development. U.S. Environmental Protection Agency, Contract No. 68-030035, Work Assignment No. 0-25. Farmer, W. J., K. Igue, W. F. Spencer, and J. P. Martin. ;1972. Volatility of Organochlorine Insecticides from Soil. I and II Effects. Soil Sci. Soc. Amer. Proc. 36:443-450. Farmer, W. J., and J. Letey. 1974. Volatilization Losses of Pesticides From Soils. Office of Research and Development. EPA-660/2-74/Q54. Farmer, W. J., M. S. Yang, J. Letey, and W. F. Spencer. 1980. Land Disposal of Hexachlorobenzene Wastes. Office of Research and Development EPA-600/2-80/119. Gray, R. A., and A. J. Weierch. 1965. Factors Affecting the Vapor Loss of EPTC from Soil. Weeds 13:141-147. Grover, R., W. F. Spencer, W. J. Farmer, and T. D. Shoup. 1978. Triallate Vapor Pressure and Volatilization from Glass Surfaces. Weed Sci., 26:505-508. Hill, F. B., V. P. Aneja* and R. M. Felder. 1978. A Technique for Measurement of Biogenic Sulfur Emission Fluxes. J. Env. Sci. Health AIB (3), pp. 199-225. C-65 TUT 008 138S Howard, P. H., R. S. Boethling, W. F. Jarvis, W. M. Meylan, and E. O. Michaelenko. 1991. Handbook of Environmental Degradation Rates. Lewis Publishers, Cheisea, Michigan. Jury, W. A., R. Qrover, W. F. Spencer, and W. J. Farmer. 1980. Modeling Vapor Losses of Soil-Incorporated Triallate. Soil Science Society Am. J., 44:445-450. Jury, W. A., W. F. Spencer, and W. J. Farmer. 1983. Behavior Assessment Model of Trace Organics in Soil; I. Model Description. J. Environ. Qua!., Vol. 12, No. 4:558:564. Jury, W. A., W. J. Farmer, and W. F. Spencer. 1984. Behavior Assessment Model for Trace Organics in Soil: II. Chemical Classification and Parameter Sensitivity. J. Environ. Qua!., Vol. 13, No. 4:567-572. Jury, W. A., D. Russo, G. Streile, and H. El Abd. 1990. Evaluation of Volatilization by Organic Chemicals Residing Below the Soil Surface. Water Resources Res., Vol. 26, No. 1:13-20. Kienbusch, M. Measurement of Gaseous Emission Rates from Land Surfaces Using an Emission Isolation Flux Chamber - User's Guide. Report to EPA-EMSL, Las Vegas under EPA Contract No. 68-02-3889, Work Assignment No. 18, December 1985. Kienbusch, M. and D. Ranum. 1986. Validation of Flux Chamber Emission Measurements On a Soil Surface - Draft Report to EPA-EMSL, Las Vegas, Nevada. Kienbusch, M., W. D. Baifour, and S. Williamson. The Development of an Operations Protocol for Emission Isolation Flux Chamber Measurements on Soil Surfaces. Presented at the 79th Annual Meeting of the Air Pollution Control Association (Paper 86-20.1), Minneapolis, Minnesota, June 22-27,1986. Lyman, W. J., W. F. Reehl, and D. H. Rosenblatt. 1990. Handbook of Chemical Property Estimation Methods. American Chemical Society, Washington, D.C. Millington, R. J., and J. M. Quirk. 1961. Permeability of Porous Solids. Trans. Faraday Soc. 57:1200-1207. Radian Corporation. Soil Gas Sampling Techniques of Chemicals for Exposure Assessment - Data Volume. Report to EPA-EMSL, Las Vegas under EPA Contract No. 68-02-3513, Work Assignment No. 32, March 1984. Radian Corporation. Short-term Fate and Persistence of Motor Fuels in Soils. Report to the American Petroleum Institute, Washington, D.C. July 1989. C-66 TLJT 008 Spencer, W. F., M. Cliath, and W. J. Farmer. 1969. Vapor Density of Soil Applied HEOD as Related to Soil Water Content, Temperature, and HEOD Concentration. So/7 Sci. Soc. Amer, Proc. 33:509-511. Spencer, W. F., and M. Cliath. 1970. Vapor Density and Apparent Vapor Pressure of Undane (r-BHC). J. Agr. Food Chem. 18:529-530. Spencer, W. F., T. D. Shoup, M. M. Cliath, W. J. Farmer, and R. Haque. 1979. Vapor Pressures and Relative Volatility of Ethyl and Methyl Parathion. J. Agric. Food Chem., 27:273-278. Spencer, W. F. 1970. Distribution of Pesticides Between Soil, Water and Air. In Pesticides in the Soil: Ecology, Degradation and Movement. A symposium, February 25-27,1970. Michigan State University. U.S. Environmental Protection Agency. 1988. Superfund Exposure Assessment Manual. Office of Emergency and Remedial Response. EPA-540/1 -88-001. U.S. Environmental Protection Agency. 1990. Procedures for Conducting Air Pathway Analyses for Superfund Activities, Interim Final Documents: Volume 2 - Estimation of Baseline Air Emissions at Superfund Sites. Office of Air Quality Planning and Standards. EPA-450/1-89-002a. U.S. Environmental Protection Agency. 1991. Risk Assessment Guidance for Superfund, Volume I, Human Health Evaluation Manual (Part B). Office of Emergency and Remedial Response. Publication No. 9285.7-01 B. U.S. Environmental Protection Agency. 1991 a. Database of Emission Rate Measurement Projects - Technical Note. Office of Air Quality Planning and Standards. EPA-450/1-91-003. U.S. Environmental Protection Agency. 1994. Technical Background Document for Soil Screening Guidance - Review Draft. Office of Solid Waste and Emergency Response. EPA-540/R-94/102. U.S. Environmental Protection Agency. 1994& CHEMDAT8 Data Base of Compound Chemical and Physical Properties. Office of Air Quality Planning and Standards Technology Transfer Network, CHIEF Bulletin Board. Research Triangle Park, North Carolina. C-67 TUT OOS 139O This page left blank on purpose. C-68 TUT OOS 1391 APPENDIX A VALIDATION DATA FOR THE JURY INFINITE SOURCE MODEL C-69 TUT DOS 1392 This page left blank on purpose. C-70 . - TUT OO8 1393 DIELDRIN 5 PPM Chemteat DMdrin DteWrin OteWrtn Dtekfrtn OteWfwi OMdnn OteWrin Ssnipta PoM 1 2 3 4 S 6 7 InJHaJ IflRWI •oR cone., c. (mo/kg) 5 5 5 5 5 5 S InnUn Ml cone., c. (M) 5.00E46 S.OOE-06 5.00E-08 S.OOE-08 5.00E46 S.OOE-08 S.OOE-08 EmMlnp, •rea (em1) 27.55 27.55 27.55 27.85 27.55 27.55 27.55 Sol Depth (L) (em) 05 0.5 0.5 O.S 0.5 0.5 O.S Sol Typ» GtaSltUwm GRaSlllmm OtaSKLoKn OMSlLoMn OtaSMLotm OtoSMLom OtaSMLoMn Sol bulk ftmftmtt,. oonsny. i% (Oton1) 0.75 0.75 0.75 0.75 0.75 0.75 0.75 Sol . Perth* ttmnmWti UVIIWI/t P. (p/om1) 185 2.65 2.65 2.65 2.65 2.65 2.65 OfWnTwtnc 8oN moisture. w (wl. fraction) 0.10 0.10 0.10 0.10 0.10 0.10 0.10 Wttw-flHed Ml porosity, V (unlttett) 0.07SO 0.0750 0.0750 L 0.0750 0.0750 0.0750 0.0750 SohiMly, S (man.) 0.1970 0.1870 0.1870 0.1870 0.1870 0.1870 0.1870 .Sal of^tnlo moon, ** (nidlon) 0.0068 0.0058 0.0058 0.0058 0.0058 0.0058 0.0058 Saturation cone., c* (mgftg) 12 12 12 12 «2 12 12 c^-c., (YM/NO) No No No No No No No Manured •mission flux (no/cm'-day) 200 115 75 65 60 55 40 Onjmte CMDon part, coftff., K« Jcm'/ft) 10900 10900 10900 10900 10900 10900 10900 o •ll c o £ <i 1 Of 2 DIELDRIN 5 PPM C hmtf mtftm\ iWfTnCn DteMrin DtoMnn OtefcWn DtoMrin OtoMrin DWdrtn DMdrhi SoW water part, coeff., Ko (cm'/B) 63.22 . 63.22 63.22 63.22 63.22 63.22 63.22 OtfhMMy In air. o; (cm'/>) 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 OtftoMly In wstef, DIW (cm'/s) 4.74E-06 4.74E-06 4.74E-06 4.74E-08 4.74E-06 4.74E-06 4.74E-06 EffecUv* oHUislon coofflctefit, Dt (cm1^) 1.32E-06 1.32E-OB 1.32E-06 1.32E-OB 1.32E-08 1.32E-08 1.32E-06 Henn/s tew fitin*tmnl COnStofn, •^ (unlfless) 0.00011 0.00011 0.00011 0.00011 0.00011 0.00011 0.00011 TOM •OR porosity. • (unMess) 0.7170 0.7170 0.7170 0.7170 0.7170 0.7170 0.7170 AMBted Ml porosity, a (unKless) 0.6420 0.6420 0.6420 0.6420 0.6420 0.6420 0.6420 Measured €fiwsslon nux (jig/ctn'-day) 0.2000 0.1150 0.0750 0.0650 0.0600 0.0550 0.0400 Tim*. t CumuMrve (hours) 24 72 120 144 168 216 266 l>LJ7l4.4Df (Yes/No) No No No No No No No LifliJfj. NltRMV •outcc inoctel •mission flux (Mtem'-day) 0.0714 0.0412 0.0319 0.0292 0.0270 0.0236 0.0206 -H C cs O 03 •-G Oi 2 of 2 DIELDRIN 10 PPM Chemical DWdrin DteMrin DteWrin OteMrin Oteldrin DteWrtn Dtekkln Ssnipto Point 1 2 3 4 5 6 7 mRM sol COOC.j c. (mpyVfl) 10 10 10 10 10 10 10 InMal sol oonc.( c. W9Y 1.00E-05 1.00E-05 1.00E-05 1.00E-05 1.00E-05 1.00E-05 1.00E-05 EniMNiQ area (em 1) 27.58 27.55 27.55 27.55 27.55 27.55 27.55 Sol Depth (L) (cm) 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Sol Type Ola SM Loam Ola Si Loam GtaSKLoam Ola SIR Loam Ola SK Loam Ola SM Loam Ola Sit Loam Sol bulk QBnvJiy, p» (gW) 0.75 0.75 0.75 0.75 0.75 0.75 0.75 Sol »A«41fl.|A parncw density, Pi (flfcm4) . 2.65 2.65 2.65 2.65 2.65 2.65 2.65 Of 1 Vn 1 Wll lU SOl molStUS, W (M, fraction) 0.10 0.10 0.10 0.10 0.10 0.10 0.10 Water-rated Ml porosity, • (unfttoss) 0.0750 0.0750 0.0750 0.0750 0.0750 0.0750 0.0750 SokjbMy, S (i"8A) 0.1870 0.1870 0.1870 0.1870 0.1870 0.1870 0.1870 Sol ofgmio cifuon( loc (racoon) 0.0068 0.0058 0.0058 0.0058 0.0058 0.0068 0.0058 ^Saturation conc.t c- (mgyhg) 12 12 12 12 12 12 12 C^C.., (Ytt/No) No No No No No No No Measured WTNSSKM1 flux (np/cm'-day) 400 260 140 110 105 90 85 Organic carbon part, cod?., K- (cm'/B) » 10900 10900 10900 10900 10900 10900 10900 0) •-0 1 of 2 DIELDRIN 10 PPM CnfefinCM DteWrtn DtoMrtn DWdrtn DteWrtn Dtetdnn Dtetdnn DteWrtn SoW wtltr pwt. co%9T.t KD (on'W 63.22 63.22 63.22 63.22 63.22 63.22 63.22 wirfUWrnj In air, b.1 (cm1/*) 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 ONftttMy tf (cm'te) 4.74E-08 4.74E-OB 4.74E-OB 4.74E-06 4.74E-06 4.74E-06 4.74E-06 EffMdv* dHfUtlon linalilrl-nt OCWITICWn, Dl (cm'/t) 1.32E-08 1.32E-08 1.32E-06 1.32E-08 1.32E-OB 1.32E-08 1.32E-08 Hwwyt taw _ mfl mtm^A OOnVUHH| KM (undteM) 0.00011 0.00011 0.00011 0.00011 0.00011 0.00011 0.00011 Total Ml pototHy, • (unMess) 0.7170 0.7170 0.7170 0.7170 0.7170 0.7170 0.7170 Air-Wed •ON fWWiWIlyi • (unNten) 0.6420 0.6420 0.6420 0.6420 0.6420 0.6420 0.6420 RMMUTM • mlaalnn vnrasion flux (H^cm'-day) 0.4000 0.2600 0.1400 0.1100 0.1050 0.0900 0.0650 Tlm«, 1 CumuMM (hours) 24 72 120 144 168 216 288 t>L'/14.4D, (YM/No) No No No No No No •No hJkaMM NNRNW •OWMmoaX •mtMtondux (ufltem'-dty) 0.1428 00825 0.0639 0.0583 0.0540 0.0476 0.0412 O -ic -I C-J *»-, of 2 LINDANE 5 PPM f*li •••llnal vnemicei Llndane Llndane Llndane Undane Sample Point 1 2 3 • 4 InttM toll cone., c. (mg/kg) 5 5 S 5 Initial soil cone., c. (0fe) 5.00E-06 S.OOE-06 5.00E-08 5.00E-06 EiiMUiig area (cm1) 27.55 27.55 27.58 27.55 Sol Depth (I) (cm) 0.5 0.5 0.5 0.5 Sod Type Ola SI Loam Gta Si loam Ola SI Loam Ola SI Loam Sol bulk density, Pk (kg/I) . 0.75 0.75 0.75 0.75 Sol particle density. P. (kgfl.) 2.65 2.65 2.65 2.65 OravlmeMc SOR moWure, w (wt. fraction) 0.10 0.10 0.10 0.10 Water-fMed sofl poroshy. 0 (unRtess) 0.0750 0.0750 0.0750 0.0750 SdubWy, S (mgrt.) 4.2000 4.2000 4.2000 4.2000 Sol OTQMIlC CttlXNl, foe (fraction) 0.0058 0.0058 0.0058 0.0058 Saturation cone., CM (mg/kg) 34 34 34 34 C.>C^ (Yes/No) No No No No Measured •mission flux (ngfcm'-day) 500 160 60 40 Organic carbon part, coeff.. K» (cm'/g) 1360 1380 1380 1380 Oil I/I H- 1 of 2 LINDANE 5 PPM Chemical Ltndsfw Undane Undarw UndflM SoW water part, coeff., Ko (cm'/g) 8.00 8.00 8.00 8.00 DnfUSWny In air. D,* (cmj/s) 0.0176 0.01.78 0.0176 0.0178 Dufusivwy Inwilor, v (cm'/a) 5.57E-08 S.57E-08 S.57E-06 S.57E-06 EffacHva diffusion _ _ -••_ t~ — • vwnicivfii! DC (cm'/a) 1.60E-07 1.60E-07 1.80E-07 1.60E-07 Mercy's law a-.. -t j,,j COfnunll, KM (unttlesa) 0.00014 0.00014 0.00014 0.00014 Total •on. poresRy, * (unMess) 0.7170 0.7170 0.7170 0.7170 Ak-fHted aol porosity, « (unMess) 0.6420 0.6420 0.6420 0.6420 MMMured wvnsslon flux (ug/cm2-day) 0.5000 0.1600 0.0600 0.0400 Tkna, 1 Gutnutatlva (hours) 24 72 120 168 •• l>LI/14.40f (Yeaffto) No Yea Yes Yea I«4WMA IIHimV •oUPMfnooil 0fiHMion mix Oigfem'Hlay) OJ841 0.1525 0.1161 0.0998 Plnia aouroa model emission flux (MQYcm*-day) 0.2641 0.1510 0.1066 0.0797 tntbiM* IIIIHnlV •KNJfM IfKXwl •fTOf (percent) 0.0000 0.9604 8.7691 252965 O c _! CO -0 2 of 2 LINDANE 10 PPM Chtmlcflt Llndane Llndww Llndane Llndane Swnpte Point 1 2 3 4 InKW ad cone., c. (mg/kg) 10 10 10 10 mMal aol cone., c. (M) 1.00E-05 1.00E-05 t.ODE-05 1.00E-05 cfnUHng area (cm 2) 27.55 27.55 27.55 27.55 Sol Depth (I) (em) 0.5 0.5 0.5 0.5 Sod Type GfaMLoam OtaSMLoam OtaSMLoam OHaSMLoem SoR bulk (tensity, pk (kgrt.) 0.75 0.75 0.75 0.75 Sol _ -,M,t- pmuCW -a ——— a^'. cwnsny, PS (HB«.) 2.65 2.65 2.65 2.65 Gravimetric SOB molsturOi w (wt. fraclloii) 0.10 0.10 0.10 0.10 >ftf-j— r nt*(t Wm€T*fM6O aoR porosity, 0 (unWws) 0.0750 0.0750 0.0750 0.0750 SohiWMy. S (mart.) 4.2000 4.2000 4.2000 4.2000 Sol oipinlc CWwOftt foe (fraction) 0.0058 0.0058 0.0058 0.0058 Saturation cone., c- (mg/kg) 34 34 34 34 C.>C«, (YeVNo) No No No No M688Urod -l,i,i--i,,n Mnmon flux (ngfem -day) 1160 320 140 90 Organic cwbon part* coefr., K. (cm/g) 1380 1360 1380 1360 O •Ij-J 1 of 2 LINDANE 10 PPM , ChtffnCw Llndane Undano Undsrw UndsM SoW Vntef pwt. cooff., Ko (cm'/8) a.oo 8.00 8.00 8.00 DffiusMty In*. D,4 (cmVs) 00178 0.0176 0.0178 0.0176 OffiMMty InMter. V (cm'/«) 5.57E-06 5.S7E-06 5.57E-06 S.57E-06 En«c«w diffusion coefficient, Dt (crn'M) 1.80E-07 1.60E-07 1.80E-07 1.80E-07 H«nn/s taw nnnalaait vonsunii, 1C. (unWest) 0.00014 0.00014 0.00014 0.00014 ToM sol porotdy, • (unMess) 0.7170 0.7170 0.7170 0.7170 Air-filed •on porosity. *• (unMess) 0.6420 0.6420 0.6420 0.6420 Measured _ -_.i^ m t,M oimsion flux (jioycm'-day) 1.1600 0.3200 0.1400 0.0900 Time, 1 Cumulative (houra) 24 72 120 168 ' t>lz/14.4D, (Yes/No) No Yet YM Ye* NMtt •OUTM flKXNt •fnlcslon flux Oi9fcm2-day) O.S282 0.3049 0.2362 0.1996 Ftate MUTMtlKMnl •nwssipn IKix (|iQrcm ^toy) O.S282 0.3020 0.2171 0.1593 . tnflnto •OUTC9 fnodn WTOf (perceot) 0.0000 09604 8.7691 25.2965 r>•li oo O O 2 of 2 BENZENE 110 PPMW Chcfnlcsl Benzene Benzene Benzene Benzene Benzene Benzene Sample PoM 3 4 5 6 7 8 MIW art cone., C. (mo/kg) 110 110 110 110 110 110 Initial art cone.. C. WW .10E-04 .10E-04 .10E-04 .10E-04 .10E-04 .10E-04 Flux 4j_~^_L__ cnamDar aurfaca area (cm 1) 1300 1300 1300 1300 1300 1300 Sol Depth W (cm) 91 91 91 91 91 91 Sol Typ« - Loamy Sand Loamy Sand Loamy Sand Loamy Sand Loamy Sand Loamy Sand Sol bulk jlaauilti density. * (tart-) 1.5 1.5 1.5 1.5 1.5 1.5 Sol _^^j|_|_ density. P* (KpA) 2\65 2.65 2.65 2.65 2.65 2.65 GrsvfrneWc Ml iiwlsluia, w (wt. fiaulloii) 0.10 0.10 0.10 0.10 0.10 0.10 Water-fflted tol pornlty, e (unMesa) 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 SoMMRy, s (mat) 1780 1780 1780 1780 1780 1780 Sol organic Carbon, foe (traction) 0.008 0.008 0.008 0.006 0.008 0.006 c ————— conc.f C«, (mgVkg) 882 862 862 .862 862 862 c.*c«, (Ye«/No) No No No No No No Measured •mission Dux OigAn'-fflkt) 2780 9000 910 400 290 0 Organic carbon partcoeff., K» (cm'/a) 57 57 57 57 57 57 o o 03 1 of 2 BENZENE 110 PPHW ^li • ulln •! vnemiCW Benzene. Benzene Benzene Benzene Benzene Benzene SoW water pert, coeff.i Ko (em'fg) 0.34 0.34 0.34 0.34 0.34 0,34 OHfushrity In sir, o,' (cm'/s) - 0.0870 0.0870 0.0870 0.0870 0.0870 0.0870 DHtUsMy In WMfif* or (cm'/s) 9.80E-06 9.80E-08 9.80E-08 9.80E-08 9.80E-06 9.80E-Q6 cfrecthfB dlffiislon , „ . fflnlaaii vwvincwnii o. (cmaM) 2.14E-03 2.14E-03 Z14E-03 2.14E-O3 Z14E-03 2.14E-03 H (atm-m'/mol) 0.00543 0.00543 0.00543 0.00543 0.00543 0.00543 Henrys tow constant, K» (unWess) 0.22283 0.22263 0.22263 0.22263 0.22263 0.22263 Toisl soR porosity, * . (unMess) 0.4340 0.4340 0.4340 0.4340 0.4340 0.4340 Mr-IMed sod porosity, • (unMess) 0.2840 0.2840 0.2840 0.2840 0.2840 0.2840 Measured —— i —— • —— vinnwon flux Oigfcm'-day) 397 1296 131 58 42 0 Time, t CumuMN* (hours) 26.40 76.25 119.73 508.83 |_ «»S5 863.17 , — -7 —— . *>\.*n4.40t (Yes/No) No Yet Yes Yes Yes Yes IHJIWMA miRHW sourca model •mission flux (ugtem'-dty) 1207 710 567 275 L 235 211 FMte •ouros mixJel •mission flux Oigfem'-day) 1207 710 567 209 135 92 InflnRe sourosmoosl •nor {percent) 0.0000 0.0002 0.0253 31.5053 73.9743 129.3941 n 00o H f~H 03 -IS: ft 2 Of 2 TOLUENE 880 PPMW ^liaMilnal ^nennCal Toluene Toluene Toluene Toluene Toluene Toluene Toluene Sample Point 3 4 5 8 7 8 9 InRM to* cone., c. (mgAg) 880 880 880 880 880 880 880 MIM toll cone., c. (9/9) 8.80E-04 8.80E-04 8.80E-04 8.80E-04 8.80E-04 8.80E-04 8.80E-04 FhM CnBmDBT aurfaea area . (cm1) 1300 1300 1300 1300 1300 1300 1300 So* Depth (D (cm) 91 91 91 91 91 91 91 Sol Type Loamy Sand Loamy Sand Loamy Sand Loamy Sand Loamy Sand Loamy Sand Loamy Sand Sol bulk j^-_— ij-- QnlSnj, p» (kgA) 1.5 1.5 1.5 1.5 1.5 1.5 1.5 Sol pffnCW JM_— ij,- oonsny, pi (koA) 2.65 2.65 2.65 2.65 2.65 2.65 2.65 Gravimetric sod moisture, w (wt. fraction) 0.10 0.10 0.10 0.10 0.10 0.10 0.10 Water-fined SOl porosity, e (unMen) 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 0.1500 SohMRy, S (mpA) 558 558 558 558 558 558 558 Sol jMw^MMijk onjanc carbon, toe (fraction) 0.008 0.008 0.008 0.008 0.008 0.006 0.008 Samraaon cone., c« (mo/kg) .522 522 522 522 522 522 522 c.>c«, (Yet/No) YM Ye* L Y« YM Ym Yn YM Measured mmmimmtan emission flux (ugftn'-min) 14800 17300 4910 1340 830 340 280 Organic carbon ptn. COCff.f K. (cm'/fl) 131* , 131 131 131 131 131 131 o oo H 1 of 2 TOLUENE 880 PPHW Chemical Toluene Tohiene Tohwne Toluene Toluene Toluene ToHwnQ SoW water part, coefi., Ko (cm'/o.) 0.79 0.79 0.79 0.79 0.79 0.79 0.79 DtffusMty In air. D.' (cm'/») 0.0670 0.0670 0.0870 0.0670 0.0670 0.0670 0.0670 OtffilsMty In witef, V (cm'/t) 8.80E-06 8.80E-06 8.80E-06 8.60E-06 8.60E-08 8.60E-06 6.80E-08 Effective dMtoton , --tilalB^i ooornpvni! D, (cm'fe) 1.30E-03 1.30E-03 1.30E-03 1.30E-03 1.30E-03 1.30E-03 1.30E-03 H (•bn-m^/lnol) 0.00637 0.00637 0.00637 0.00637 0.00637 0.00637 0.00637 Henrys law constant, KM (unMess) 0.26117 0.26117 0.28117 0.26117 0.26117 0.28117 0.26117 Total •o* poresMy, 4 (unNteSB) 0.4340 0.4340 0.4340 0.4340 0.4340 0.4340 0.4340 Alr-flHed aoR potoaMy, a. (unMess) 0.2640 0.2840 0.2840 0.2640 0.2840 0.2640 0.2840 Measured •flnSWOn flux (ugfcm*-day) 2131 2491 707 193 120 49 37 Time, I Cumulative (hours) 26.40 76.25 119.73 506.83 898.55 863.17 1007.17 f t>L'/14.40. (Yes/No) No No No Yea Yea Yea Yea iniiijij- RHINMI Muroa modal •mission flux (ug/cm'-day) 7524 4427 3533 1717 1463 1316 1218 Finite aoume model •mission flux (ug/cm'-day) 7524 4427 3533 1813 1231 978 800 •- ••-••- mime source modal •nor (percent) 0.0000 . 0.0000 0.0001 6.4806 18.8541 34.5481 52.2596 o oo 03 4* .•**'. •«•• cn 2 Of 2 O t-4n W 03 It i II ft i § § «M O C-83 TUT 008 1406 ETHYLBENZENE 310 PPMW Clia>nln»l nCfinCSI Ethylbenzena Ethylbenzene Ethylbenzene Ethylbenzene Elhylbenzene SoW ...-I- r WKet p0n. 006rf.| Ko (cm'/9) 1.33 1.33 1.33 1.33 1.33 i •»•» DfflMNty In St. o; (cm'/s) 0.0750 0.0750 0.0750 0.0750 0.0750 nnron OMusMty In wvteff V (cm'/s) 7.80E-06 7.80E-06 7.80E-06 7.60E-06 7.60E-06 7HOFJW Effective diffusion ___jgi_i_-j COofilCIVfW, oc (cm'/s) 8.64E-04 8.64E-04 8.64E-04 8.64E-04 8.64E-04 HH4FM Henn/s taw nnn al^nl UUIWUnn, K« (unltless) 0.32021 0.32021 0.32021 0.32021 0.32021 032021 Total sol poresXy, • (unMess) 0.4340 0.4340 0.4340 0.4340 0.4340 04340 Alr-nited sod porosity, • • (unRtess) 0.2840 0.2840 0.2840 0.2840 0.2840 09840 Messured Mntotion flux (HQ/cm'-day) 380 245 156 36 26 0 Time. 1 CumuMlve (hours) 26.40 76.25 119.73 506.83 696.55 88317 t>L'/14.4Dl (Yes/No) No No No Yes Yes YM , •,, Hull a HNNMW •OUTCf fflOuP •mission flux (uflfcnrMtay) 2162 1272 1015 493 420 378 FWe •OUTM fHOdol •mission flux (uoybm'-day) 2162 1272 1015 486 402 343 InfsmNs 1 source model error (percent) 0.0000 0.0000 0.0000 1.0596 4.6357 in mm O oe •u c —; CO O xj 2 of 2 APPENDIX B VALIDATION DATA FOR THE JURY REDUCED SOLUTION FINITE SOURCE MODEL C-85 TUT 008 1408 This page left blank on purpose. C-86 TUT COS 1.409 TRIALLATE 10 PPM Chcfnic&l TriaRate TrlaHale TrtaHate TrlaHale TrlaHste TrlaDate TriaDate TriaMe Triable TrtaRale TrtaNate TriaRate TrtaRate TriflKflfa TfMtete TrtaRate TrtaRale TriaRate TriaRate TrtaRate TriaRate Tvtaaataataii 1 itOTHH Yot^H^A TnMUNB TfWWe TfWWo TrWtate TrtaRata TriaRata TriaRate y-l—H—f — Truman TriaRate Sample Point 1 2 3 4 5 6 7 9 9 10 11 12 13 14 IS 16 17 18 19 20 21 22 23 24 25 29 27 28 29 30 31 32 Initial aod cone., c. (mo/Kg) 10 10- 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 to 10 10 10 10 InNW aoR cone., c. (M) 1.00E-05 1.00E-05 1.00E-05 1.006-05 1.006-05 1.00E-05 IOOE-05 t.OOE-05 1.00E-05 1.006-05 1.00E-05 1.00E-05 1.00E-05 1.006-05 1.00E-05 1.006-05 1.00E-05 1. ODE-05 1.006-05 1. 006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 1.006-05 Emitting area (cm 1) 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 '30 30 •30 30 30 SoR Depth (I) (cm) 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 . 10 10 10 10 10 10 10 SoR Type San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquti Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam San Joaquin Sandy Loam SoR bulk (tensity, p» (W-) 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 1.34 SoR — — -ttmtm particle density. P* <*»A) 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.6$ 2.6S 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 2.65 Gravimetric aoR .moisture, w (wt. -fraction) 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 0.21 1Af_|_r Maia aoR . porosity, e (unttteaa) 0.2787 0.2787 0.2787 0.2787 - 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 0.2787 07787 0.2787 05787 0.2787 0.2787 0.2787 0.2787 . 0.2787 0.2787 SolubRNy. S 0"*). 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 . 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 SoR organic carbon. foe (fraction) 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 • 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 0.0072 .0.0072 0.0072 0.0072 Saturation cone., C* (mgftg) 105 105 105 105 105 105 105 105 105 105 105 105 105 105 .105 105 105 105 105 105 . 105 105 105 106 105 105 105 105 105 105 105 105 c.»c«, (Yea/No) No No No No No No No No No No No No No No No No •No No No • No No No No No No No No No No No No No Measured emission flux Gig/cm'-day) 1.700 0.975 0.750 0.490 0.330 0.280 0.210 0.180 0.155 0.145 0.135 0.125 0.123 0.115 0.107 0.105 0.103 0.102 0.095 0.094 0.093 0.094 0.085 0.083 0.062 0.083 0.080 0.080 0.072 '0.071 0.070 0.070 Organic naili mi caroon part.coef K« (cmty) % , ' 360 360 360 360( 360 360( 3601 3601 360 360C 3601 360C 360C 360C 360C 3800 3600 3600 3600 3600 3600 3600 3600 3800 3600 3600 3600 3600 3600 3600 3600; 3600 I n oo H 1 Of 2 TRIALLATE 10 PPM O oo 00 O 03 CnfefTncn TrWWt TrWWt TrWMt TrWMt TrWMt TrWMt TrWWt TrWMt TrWWt TrWMt TrWMt TnwMfft TiWM» TrWMt TIMMN TrWMt TrWMt TfWMt TrWMt TrWMt TrWWt TnMNM TnMMtl TrWMt TrWMt TrWMt TrWMt TrWMt TrWMt TrWMt TrWMt TrWMt SoW wiwf ptft. cooff., KO (cm'/o) 25.02 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 . 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 • 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 25.92 DHftoMy kt*, D,' (cmVt) 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 0.0450 OrmitMty Inwtttr, or (cm'fc) S.OOE-08 S.OOE-08 5.00E-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 8.00E-08 S.OOE-08 S.OOE-O8 S.OOE-06 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 8.00E46 S.OOE-08 5.00E-06 5.00E-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 S.OOE-08 5.00E-08 S.OOE-08 EfrtcHvt dnfuwon coefficient, Of (Cffl'/S) 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E4B 4.14E-08 4.14E-OS 4.14E-08 4.14E-08 4.14E-08 4.14E-08 4.14E-06 4.f4E-OB 4.14E-08 Htnr/t tow «M^M*tWaJ OOnSwfH, Ki, (unlHess) 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 0.00104 ToW M* porosity, • (unlHess) 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 0.4943 AMRted Mi porosity. t (unNtess) 0.2156 0.2156 0.2156 0.2156 0.2156 0.2158 0.2156 0.2156 0.2156 0.2156 0.2156 0.2156 0.2156 0.2156 0.2156 0.2156 0.2158 0.2158 0.2156 0.2156 0.2156 0.2158 0.2156 0.2156 0.2156 0.2156 0.2198 0.2158 0.2156 0.2156 0.2156 0.2156 M6MUTM Qfniswon flux Gigtem'-dty) 1.700 0.975 0.750 0.490 0.330 0.280 0.210 0.160 0.155 0.145 0.135 0.125 0.123 0.115 0.107 0.105 0.103 0.102 0.095 0.094 0.093 0.094 0.065 0.083 0.082 0.083 0.060 0.080 0.072 0.071 0.070 0.070 Tkm, 1 CumuMhw (hou«) 3 6 12 24 48 72 96 120 144 168 192 . 216 240 264 288 312 338 380 364 406 432 458 480 804 828 552 876 600 624 648 672 696 Jury (Into •OUTM tnodsV • ••nlaalnn Ibw •iTwSWonnux OigVcm'-day) 1.278 0.904 0.639 0.452 0.320 0261 0.226 0.202 0.164 0.171 0.160 0.151 0.143 0.136 0.180 . 0.125 0.121 0.117 0.113 0.1 1Q 0.107 0.104 0.101 0.099 0.098 0.094 0.092 0.090 0.069 0.067 0.085 0.064 2 of 2 APPENDIX D Revisions to VF and PEF Equations (EQ, 19945) oos ENVIRONMENTAL QUALITY MANAGEMENT, INC. MEMORANDUM TO: Ms. Janine Dinan ' DATE: July 11,1994 SUBJECT: Revisions to VF and PEF Equations . FROM: Craig Mann FILE: 5099-3 cc: Subsequent to the evaluation of the dispersion equations in the RAGS - Part B performed by Environmental Quafity Management, Inc. (EQ, 1993), questions have arisen as to the accuracy of the modeling protocol used to derive the dispersion coefficient (Q/C) used in the volatilization factor (VF) and the paniculate emission factor (PEF) presently employed to calculate the air pathway Soil Screening Levels (SSLs). EQ, 1993 used the Industrial Source Complex model (ISC2-ST) to derive a normalized concentration (kg/nf per g/nf-s) for a series of square and rectangular area sources of differing size. This modeling protocol employed a source subdivision scheme similar to that recommended in the ISC2-ST Model User's Manual (ERA, 1992) whereby the source was subdivided into smaller sources closest to the center of the area. The center of the area was found to represent the point of maximum annual average concentration for all source shapes analyzed. Consecutive model runs were performed whereby source subdivision was increased between runs. Final source subdivision was reached when the model results converged within a factor of three percent or less. From these data, a simple linear regression was used to evaluate the nature of the relationship between the normalized concentration and the size of the area Preliminary plots of the data indicated that the relationship was exponential. Therefore, the relationship was linearized by taking the natural logarithms (In) of each variable. The resulting linear regression for a square area of 0.5 acres resulted in a normalized concentration (C/Q) of 0.0098 kg/nf per g/nf-s; the inverse of the normalized concentration resulted in a dispersion coefficient (Q/C) of 101.8 g/nf-s per kg/rrf. On May 5,1994 a teleconference was held between representatives of the Toxics Integration Branch of the Office of Emergency and Remedial Response (OERR) and the Source Receptor Analysis Branch of the Office of Air Quality Planning Standards (OAQPS) to discuss the relative merits of the available area source algorithms as applied to near- field and on-site receptors exposed to ground-level nonbuoyant emissions. The conclusions drawn from this teleconference were that a new algorithm recently developed by OAQPS would yield more accurate results for the exposure scenario in question. D-i TUT The new algorithm is incorporated into the ISC2 model platform in both short-term mode (AREA-ST) and long-term mode (AREA-LT). Both models employ a double numerical integration over the area source in the upwind and crosswind directions as follows: dyldx (1) where Q* = Area source emission rate (g/nf-s) K = Units scaling coefficient V «Vertical term . D = Decay term. The integral in the lateral (i.e., crosswind or y) direction is solved analytically as: (2) where erfc is the complementary error function. The integral in the longitudinal (Le.. upwind or x) direction is solved by using a weighted average of successive estimates of the integral using a trapezoidal approximation. The model uses three separate criteria to determine convergence of the upwind integral. The result of these numerical methods is an estimate of the full integral that is essentially equivalent to, but much more efficient than, the method of estimating the integral as a series of line sources, such as the method used by the Point, Area, Une (PAL 2.0) model. Wind tunnel tests have also shown that the new algorithm performs well with on-s'rte and near-field receptors. D-2 TUT 008 Because the new algorithm provides better concentration estimates and does not require source subdivision, a revised dispersion analysis was performed for both volatile and paniculate matter contaminants using the new algorithm. The first part of the analysis involved a determination of the relationship between concentration and source size. In addition, this part of the analysis included a determination of the point of maximum annual average concentration for a square area source. This assessment employed the AREA-ST model as acquired from tne OAOPS Technology Transfer Network, Support Center for Regulatory Air Models (SCRAM) Bulletin Board. Meteorological data used for this analysis were 1989 hourly data for the Los Angeles National Weather Service (NWS) surface station, upper air data were from the Oakland NWS station for the same year. Rural dispersion coefficients were employed and all regulatory default options used. Modeling assumed flat terrain with no flagpole receptors; source rotation angle was set equal to zero. Five source sizes were included in the assessment: 0.5,5,30,200, and 600 acres. A coarse cartesian receptor-grid was employed within and extending beyond the source perimeter; a discrete receptor was also placed at the center of each source (x,y = 0,0). Emissions from each source were set equal to 1.0 g/nf-s; concentrations were calculated in units of kg/m3. Figure 1 shows the relationship between source size (acres) and annual average concentration (kg/nf) for the five source sizes modeled. In each case, the point of maximum concentration was located at the center of the source. As an example, Attachment A is the model run sheets for the 0.5 acre source. As can be seen from Figure 1, the relationship between concentration and source size is exponential. Results also show that the maximum concentration representing the 600 acre source is 2.9 times higher than that of the 0.5 acre source. Having established that when using the AREA-ST model the point of maximum concentration for a square area source is the center receptor, the second part of the analysis was to determine which of the 29 meteorological sites from EQ, 1993 best represents the average exposure and the high end exposure to volatile and paniculate matter emissions. It was determined that the average exposure case should be represented by the 50th percentile site concentration, while the high end exposure is best represented by the 90th percentile site concentration. Each of the 29 sites from EQ, 1993 were subsequently modeled at an emission rate of 1.0 g/nf-s with a single discrete receptor at the center of the square area source. Source sizes modeled were 0.5 acres and 30 acres. Hourly meteorological data for each site were from EQ, 1993. From tine set of 29 normalized annual average concentrations, D-3 TUT 008 i415 c 0- 0.10 CO O H a ^ s 8 Oo 0.010.1 0.01453 I I I I 111 0.04238 0.03680 0.02849 '0.02167 1 10 100 SOURCE SIZE (Acres) Figure 1. Normalized annual average concentration versus source size. 1000 the 50th percentile site was determined to be Salt Lake City, Utah; Los Angeles, California (89th percentile site) was determined to be the closest approximation of the 90th percentile site. Table 1 shows the resulting dispersion coefficients for the two source sizes and the percentile ranking of each site. TABLE 1. VOLATILE DISPERSION SITE RANKINGS , . City Huntington Fresno Phoenix Los Angeles Winnemucca Boise Hartford Little Rock Portland Salem Charleston Denver Atlanta Raleigh-Durham Salt Lake City Houston Lincoln Harrisburg Bismarck Seattle Cleveland Albuquerque Miami San Francisco Philadelphia Minneapolis LasVegas Chicago Casper NWS surface station number 13860 93193 23183 24174 24128 24131 14740 13963 14764 24232 13880 23062 13874 13722 24127 12960 14939 14751 24011 24233 14820 23050 12839 23234 13739 14922 23169 94846 24089 0.5 Acre (Q/C) (g/m2-sper ko/m3) 52.77 62.00 64.06 68.82 69.25 69.40 71.33 73.37 74.24 73.42 74.91 75.59 . 77.16 77.46 78.06 79.24 81.63 81.90 83.40 62.71 83.19 84.18 85.40 89.53 60.09 90.74 95.51 97.75 100.00 30 Acre (Q/C) (g/m2-sper ko/m3) 27.08 31.85 32.63 .' 36.10 35.49 35.69 36.64 37.68 37.86 37.88 38.42 38.80 39.68 39.87 40.14 40.70 41.56 42.34 42.72 42.81 43.03 43.31 43.57 46.06 46.38 46.84 49.48 50.45 51.68 Site ranking pereentHe (%) 100 96 93 89 86 62 79 75 71 68 64 61 57 54 60 46 43 39 36 32 29 25 21 18 14 11 7 4 0 D-5 TUT OO8 1417 In order to determine the average and high end sites for paniculate matter exposures resulting from wind erosion, a normalized concentration could not be used because meteorological conditions other than simple dispersion (i.e., wind velocity and frequency) influence emissions and therefore actual concentrations. For this reason, actual concentrations were calculated for each site using the existing PEF equation as follows: - • C - 3600 where C (C/Q) U..7 F(x) Annual average PM,0 concentration, kg/rrf Normalized annual average concentration (kg/rrf per Fraction of continuous vegetative cover Mean annual windspeed, m/s Equivalent threshold value of windspeed at 7 m, m/s Windspeed distribution function from Cowherd, 1985. The value of (C/Q) for each site was the normalized concentration previously estimated for volatile emissions (i-e., the inverse of each dispersion coefficient in Table 1). The value of V was set equal to 0.5. The mean annual windspeed (UJ for each site was taken from Weather of U.S. Cities, Second Edition, Volume 2 by J. A. Ruffner and F. E. Bair, Gale Research Co., Detroit, Michigan. The value of F(x) was estimated for each site from Figure 4-3 or calculated from Appendix B of Cowherd 1985, as appropriate. The value of U,.7 was calculated as follows: D-6 TUT 008 1418 where U,.7 «= Equivalent threshold value of windspeed at 7 m, m/s z* * Surface roughness height, cm fc - 0.5 cm for open terrain) U, • . • Threshold friction velocity, m/s (U, - 0.625 m/s). Table 2 gives the results of this analysis and shows the relative PM,0 concentrations for each site by source size and the percentile rankings. As can be seen from Table 2, the 50th percentile site was Salt Lake City, Utah, while the 89th percentile site was Minneapolis, Minnesota Table 3 summarizes the results of the dispersion coefficient analysis for both the VF and PEF equations. In addition, Table 3 also gives the default values of the PEF variables for both average and high end exposures. D-7 TUT OO8 1419 TABLE 2. PEF CALCULATIONS AND SITE RANKINGS o oo Cisptf CtevHand Lincoln ChfcMO Attorta SMttl* Bob* UsVtQM SMFHMJNO UHto Rock NWS numbtf 24009 14820 14939 14*12 24011 13738 12839 13074 24233 24131 23188 23050 23002 I4W T4764 14740 23234 13003 24120 muri tfrindMMd •5T 12.9 10.1 10.4 H.I 10.3 10.4 9.6 8.2 9.1 9.1 0.9 0.1 9.0 OJu T 8.0 LocAngtlM FfMno 14751 24174 24232 23163 7.1 7.4 6.8 1.4 vAnt^tM (mO) S.77 4.83 4.65 4.«* 4.60 4.65 4.11 407 4.07 3.98 4.07 4.02 M 0.5 0.5 0.5 .0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 "OS 0.8 041 OJ 2.02 o Ss mCuOII InCVOVI Mturtee* 0.625 0.625 0.625 9J28 0.625 0.625 0.625 0.625 0.625 0.625 0.625 0.625 0.625 0.625 0.625 0.629 0.625 0.625 0.625 0.625 0.629 0.625 0.625 0.629 0.625 : 0.629 0.625 O.tOS •17m 11.32 11J2 11.32 1.74 2.08 11JJ 11.32 11.32 11.32 1.32 1.32 1.32 1.32 1.32 1.32 1-32 141 132 .32 1.32 142 42 42 42 42 2.16 M4 2.18 2.16 2.34 2.44 2.47 2.47 252 2.47 2.55 2.68 2.60 2.64 2JM 3.03 331 3.4 13? 3.56 0.57 NA NA NA NA NA NA NA 2.32E-01 1.82E-01 1J4E-81 1.70E-01 1.826-01 9.93E-O2 6.62E-02 6.16E-02 8.1SE-02 4.95E-02 6.18E-02 5.53E-02 4.41E-02 3.91E-O2 3.91E-02 1.46E-P2 3J1E-O2 1.456-02 1336-02 .036-02 .606-03 .(06-03 1.B7E-O3 (.74E-03 3.196-04 2356-04 cow IKiaSSS 0.50 0.50 0.50 0.80 0.50 0.50 OJO OJO 0.80 OJO OJO 0.50 PM10 HUK "SUSS 3.77E-07 9.01E-08 6.306-08 8J26-88 5.736-08 6.306-08 2.716-OB 1.646-08 1.436-08 1.436-08 1.07E-OB 1.436-08 OJAcf* 100.00 83.10 •1.63 98.74 63.40 97.75 90.09 85.40 77.16 82.71 80.40 8SJ1 84.18 75.59 • T8J8 0.8 Ant •nnwl oonc. 3.77 1.06 0.77 0.89 0.84 0.30 0.19 0.19 0.17 0.15 0.15 0.15 0.12 8.1J - ESZ 0.11 an •9.24 0.000 LOOS 1:031 1027 0.019 SOAcra taftnS) 51.68 43.03 41.56 41.94 42.72 50.45 48.38 43.57 39.88 42.81 35.80 49.48 43.31 38.00 4t,14 37.88 38.42 38.84 48.08 37.88 35.49 40.70 39.87 42.34 aso 1.736-11 'nili"il ' .' 30Acn MMNMl cone. (ug/tn3) 7^29 2.09 1.52 1.48 1.34 1.25 0.58 0.38 0.38 0.33 0.30 0.29 0.29 0.24 9.23 0.21 0.21 0.18 0.17 0.081 0.052 0X137 0.030 0.029 0.017 100 96 82 -2 _75 7' 6' _57 54 4! 21 J« J4 11 3X83 531E-04 TABLE 3. VF AND PEF VALUES OF (Q/C) FOR AVERAGE AND HIGH END EXPOSURES Site size 0.5 Acres 30 Acres Average annual cone., PM10 (uo/m3) 0.12 0.23 High End annual cone., PM10 (ug/m3) 0.76 1.48 PEF Average (Q/C), (g/m2-sper ko/m3) 78.06 40.14 PEF High End. (Q/C), (g/rn2-sper ko/m3) 90.74 46.84 VF Average (Q/C), (g/m2-sper taftn3) 78.06 40.14 VF High End (0/0), (g/m2-sper Ko/m3) 68.82 35.10 Average Site for PM10= Salt Lake City Average Site for Volatile* «= Salt Lake City . High End Site for PM10 - Minneapolis High End Site for Volatile* * Los Angeles Average Site for PM10: Mean annual windspeed (Um)« 3.93 m/s; F(x)« 0.044, at x « 2.55. High End Site for PM10: Um = 4.69 m/s; F(x) * 0.194, at x * 2.14. Where: Vegetative cover (V) = 0.5. Surface roughness height (Zo) * 0.5 cm. Threshold friction velocity (Ut)« 0.625 m/s at surface. Threshold windspeed at 7 meters (Ut-7) = ut/0.4 x In(700/Zo) = 11.32 m/s. D-9 TUT 008 j.42i This page left blank on purpose. D-10 TUT 008 1422 - ATTACHMENT A AREA-ST MODEL RUN SHEETS FOR A 0.5 ACRE SQUARE AREA SOURCE D-ll TUT OO8 CO STARTING CO TITLEQNE AREA SOURCES- CO MODELOPT OFAULT CONC CO AVERT IHE PERIOD CO POLLUT1D PM10 CO IUNORNOT RUN CO ERRORFIL AREA1.ERR CO FINISHED •• 1/2 acre run RURAL SO STARTING •* SRCID SRCTYP •* • ..... ...... SO LOCATION At/2 AREA •* SRCID OS •* ' ..... .... SO SKPARAM A1/2 1.0 XS YS JS -22.5 -22.5 .0000 MS XINIT YINIT 0.0 45. 45. SO BHSUNIT .100000E-02 (GRA*S/(SEC-N**2)> SO SICGMUP AREA1 A1/2 SO FINISHED KILOGRAMS/CUBIC-NETER RE STARTING RE OISCCMT 'E OISCCART * DISCCART *E OISCCART RE OISCCART RE OISCCART RE OISCCART RE OISCCART RE OISCCART RE OISCCART RE DISCCART IE OISCCART RE OISCCART RE DISCCART RE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART IE DISCCART 0. 25. -25. 25. 25. •25. •25. 50.. -50. 50. 50. •50. -50. 75. -75. 75. 75. •75. •75. 100. •100. 100. 100. •100. •100. 0. 0. 0. 25. -25. •25. 25. 0. 0. SO. •so. -50. 50. 0. 0. 75. •75. -75. 75. 0. 0. 100. -100. -100. 100. RE FINISHED RE STARTING ME INPUTFIL RE ANEKHGHT ME SURFDATA NE UAIRDATA NE WINDCATS NE FINISHED C:\CRA1G\23174-89.ASC 10.0 METERS 23174 1989 LOS ANGELES 25250 1989 OAKLAND 1.54 3.09 5.14 8.23 10.80 OU STARTING OU RECTABLE ALLAVE FIRST OU FINISHED «•«•••••••«»«»••«•*•««••••«•••«•«»« •*• SETUP Finishes Successfully •** D-12 TUT 008 1424 «*• AREAST • VERSION TESTA.»•• ••* AREA SOURCES-" 1/2 acre run TEST OF ST AREA SOURCE ALGORITHM •»* •*• MODELtKG OPTIONS USED: CONC RURAL FLAT DFAULT *** MCOEL SETUP OPTIONS SUMMARY ••Model Is Setup For Calculation of Average concentration Values. "Model Uses RURAL Dispersion. "•Model Uses Regulatory DEFAULT Options: 1. Final Pluae Rise. 2. Stack-tip Downwash. 3. Buoyancy-induced Dispersion. 4. Use Cauat Processing Routine. 5. Not Use Missing Data Processing Routine. e. Default Wind Profile Exponents. 7. Default Vertical Potential Jseperaturt Gradients. 8. nipper found" Values for Suparsquat Buildings. 9. No Exponential Decay for RURAL Node ••Nodal Ossuasi Receptors on FLAT Terrain. . ••Model Assuaei No FLAGPOLE Receptor Heights. ••Model Calculates PERIOD Averages Only • ••This Run Includes: 1 Source(s); 1. Source GroujXs); «nd 25 Receptor<s) ••The Model ASSUME A Pollutant Type of: PM10 ••Model Set To Continue RUNoing After the Setup Testing. * ••Output Options Selected: Model Outputs Tables of PERIOD Averages by Receptor Model Outputs Tables of Highest Short Tena Values by Receptor (RECTABLE Keyword) ••MOTE: The Following Flags Nay Appear Following CONC Values: e for CaUi Hours • for Missing Hours b for Both CalB and Missing Hours •"Misc. Inputs: Anea. Hgt. (•} > 10.00 ; Decay Coef. * .0000 ; Rot. Angle • .0 Emission Units * (6RAMS/(SEC*M**2» ; Emission Rate Unit Factor « .10000E-02 Output Units « KILOGRANS/CUBIC-METER ••Input Runstrean. File: area1.dat ; ••Output Print File: areal.out ••Detailed Error/Message File: AREA1.ERR' TUT OOS 1425 ••" AJtEAST • VERSION TESTA "»» »M AREA SOURCES-•• 1/2 «cre run *•* TEST OF ST AREA SOURCE ALGORITHM ••* - •** * . •••.MODELING OPTIONS USED: COMC RURAL FLAT OFAULT ••• AREA SOURCE DATA •*• NUMBER EMISSION RATE COORD <SU CORNER) BASE RELEASE X-OIM Y-BIM ORIENT. EMISSION RATE SOURCE PART. {USER UNITS X Y ELEV. HEIGHT OF AREA OF AREA OF AREA SCALAR VARY ID CATS. /METER**2) (METERS) (METERS) (METERS) (METERS) (METERS) (METERS) (DEC.) BY A1/2 0 .100006+01 -22.5 -22.5 .0 .00 45.00 45.00 .00 D-14 TUT 008 1426 ••• AREAST • VERSION TESTA ••* •** MEA SOURCES-•• 1/2 acre TEST OF ST AREA SOURCE ALGORITHM ••• *•• MODELING OPTIONS USED: CONC RURAL FLAT DFAULT SOURCE IDs DEFINING SOURCE GROUPS SOURCE IDs AREA1 A1/2 D-15 TUT 008 1427 *•* AREAST - VERSION TESTA •« •** AREA SOURCES--• 1/2 «cr« rui TEST Of ST AREA SOURCE ALGORITHM •** •*• MODELING OPTIONS USED: CONC RURAL FLAT OFAUIT . •*« DISCRETE CARTESIAN RECEPTORS •" (X-COORD. Y-COORD, ZELEV. ZFLAG) (METERS) < .0, .0, .0. .0); ( 25.0. .0, .0, .0); ( -25.0. .0, .0. .0); ( 25,0. 25.0. .0. .0); ( 25.0, -25.0, .0. .0); ( -S.O, -25.0, .0, .0>; -25.0, 25.0, .0. .0); < 50.0. .0, .0, .0); -«« « n .0 .0>: < 50.0, 50.0, „ .0, .0); « u* * 'f I i:s: ™:s: ;: . -as --s ft § ' -™- - ?S:S: :S: ( -100.0. 100.0, .0, .0); D-16 TUT COS 1428 *** MEAST • VERSION TESTA ••* •** AREA SOURCES-•• 1/2 «crt run TEST OF ST AREA SOURCE ALGORITHM •** *** MODELING OPTIONS.USED: CONC RURAL FLAT DFAULT **• METEOROLOGICAL DAYS SELECTED FOR PROCESSING (1*TES; 0»NO> 1 ' 1 111111 1 1 I 1 1 •^ • . 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 .1 111 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1-11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 4 1 1 ^. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1.1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 • « « 1 « 1 1 r i 4 « «| 1 1 < 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 NOTE: METEOROLOGICAL DATA ACTUALLY PROCESSED WILL ALSO DEPEND 0» HUT IS IMCUBED IN TIE OATA FILE. «*• UPPER SOUND OF FIRST TNROUCH FIFTH WIND (METERS/SIC) CATEGORIES 1.S4. 3.09, 5.U. 8.23. 10.80, •*• mm PROFILE EXPONENTS *** STABILITY CATEGORY A B C D. E F MIND .700006-01 .700006-01 .100006*00 .150006*00 .350006*00 .550006*00 .700006-01 .700006-01 .100006400 .150006400 .350006400 .550006*00 SPEED CATEGORY . 3 i70000E-01 .700006*01 .100006400 .150006400 . .SSOOOE400 .S5000&MM _jk_ .700006-01 .700006-01 .100006400 .150006400 .350006*00 .550006400 .TOgOOE-OI .7DOOOE-01 .10000E+00 .1MOOE+00 .SSCiOO&HX) «*. •7DOOOE-01 .70000E-01 .10000E«00 .150006*00 .35000E*00 .S5000E+00 VERTICAL POTENTIAL TEMPERATURE GRADIENTS (DEGREES KELVIN PER METER) STABILITY CATEGORY A B C D E F WIND .000006400 .000006*00 .000006*00 .000006*00 .200006*01 .350006*01 .000006400 .000006*00 .000006400 .000006400 .200006*01 .350006*01 SPEED CATEGORY 3_ .000006*00 .000006*00 .000006*00 .000006400 .200006*01 .350006-01 .000006400 .000006400 .000006400 .000006400 .200006-01 .350006-01 .000006*00 .000006*00 .0000064QO .000006400 .200006*01 .350006*01 6 .000006400 .000006400 .000006400 .200006*01 .350006*01 D-17 TUT DOS 1429 **• MEAST • VERSION TESTA *•• *•* AREA SOURCES--• 1/2 «er« TEST OF ST AREA SOURCE ALGORITHM •*• ••• MODELING OPTIONS USED: CONC RURAL FLAT OFAULT •"..THE FIRST 24 HOURS OF METEOROLOGICAL DATA FILE: C:\CRAIG\23174-89.ASC SURFACE STATION NO.: 23174 NAME: LOS YEAR: 1989 YEAR MONTH DAY HOUR FLOW VECTOR SPEED (N/S) FORMAT: (4I2,2F9.4.F6.1,12.2F7.1> UPPER AIR STATION KO.: 23230 HAKE: OAKLAND YEAR: 1989 STAB CUSS NIXING HEIGHT (N) 89 89 89 89 89 89 89 89- 89 89 89 89 89 89 89 89 89 89 89 89 89 89 89 89 1 2 3 4 5 67 8 9 10 11 12 13U 15 16 17 18 19 20 21 22 23 251.0 228.0 194.0 143.0 173.0 272.0 265.0 233.0 257.0 261.0 44.0 56.0 83.0 59.0 82.0 74.0 81.0 87.0 154.0 167.0 280.0 252.0 220.0 260.0 3.09 3.09 2.57 4.63 2.06 3.09 2.06 2.06 2.06 .00 2.M 3.60 4.12 4.12 4.12 3.60 3.60 3.09 4.12 2.06 2.57 2.06 3.09 1.54 282.6 282.0 282.0 282.0 282.0 280.4 280.4 282.0 283.7 285.9 288.2 289.3 289.3 290.4 287.6 287.6 285.9 284.3 286.5 285.4 285.4 284.3 283.2 283.7 4 4 44 5 6 6 5 4 3 3 3 3 3 3 4 5 65 6 6 6 6 7 533.0 604.1 639.6 675.2 710.7 746J 04.9 278,2 421.6 564.9 708.3 851.6 995.0 995.0 995.0 992.3 975.8 959.2 942.7 926.2 909.6 893.1 876.5 533.0 568.6 604.1 09.6 151.0 151.0 151.0 285.4 387.0 508.6 630.2 751.8 873.4 995.0 995.0 995.0 979.1 880.6 782.2 683.8 585.3 486.9 388.4 290.0 NOTES: STABILITY CLASS 1«A. 24, 3*C, 4*>, 5«£ AK> 6-F. FLOW VECTOR IS DIRECTION TOWARD WHICH WIND IS BLOWING. D-18 TUT **• AREAST • VERSION TESTA •*• TEST OF ST AREA SOURCE ALGORITHM AREA SOURCES-- 1/2 acre run MODELING OPTIONS USED:" CONC RURAL FLAT DFAULT THE PERIOD ( 8760 HRS) AVERAGE CONCENTRATION INCLUDING SOURCE(S): A1/2 VALUES FOR SOURCE GROUP: AREA1 "*• DISCRETE CARTESIAN RECEPTOR POINTS • CONC OF PN10 IN KILOGRANS/CUBIC-NETER X-COORC (H> Y-COORD (M) CONC X-CCORD (N) Y-COORD (N) CONC .00 •25.00 25.00 •25.00 •50.00 50.00 •50.00 -75.00 75.00 •75.00 •100.00 100.00 -100.00 .00 .00 •25.00 25.00 .00 -50.00 50.00 .00 •75.00 75.00 .00 -100.00 100.00 .01453 .0059* .00104 , .00223 .00158 .00018 .00037 .00078 .00008 .00016 .00047 .00005 ^ .00009 ~" 25.00 25.00 •25.00 •50.00 50.00 •50.00 75.00 . 75.00 -75.00 100.00 100.00 -100.00 •00 25.00 . -25.00 .00 50.00 •50.00 .00 75.00 •75.00 .00 100.00 •100.00 .00679 .00414 .00220 .00175 .00060 .00034 .00076 .00024 . .00015 .00041 .00013 .00009 • . D-19 TUT 008 1431 •*• AKEAST • VERSION TESTA •*• •*• AREA SOURCES---'1/2 «er« run •* TEST Of ST AREA SOURCE ALCORITHH *** •» •••MODELING OPTIONS USED}''cONC RURAL FLAT DFAULT *•• THE SUNNARY OF MAXIMUM PERIOD < 6760 HRS) RESULTS •** •* CONC OF PMIP IN KILOCRANS/CUBIC-METER •* NETWORK OROUP 10 MEA1 * 1ST 2ND 310 ATM STH 6TH HIGHEST HIGHEST HIGHEST HIGHEST HIGHEST HIGHEST VALUE VALUE VALUE VALUE VALUE VALUE AVERAGE CONC IS IS IS IS IS IS .01453 .00679 .00594 .00414 .00223 .00220 AT AT *T v AT AT AT RECEPTOR (XR, YR. 2ELEV, ZFLAG) OF .00. 25.00, •25.00. 25.00, •25.00, •25.00, .00, .00, .00, 25.00. 25.00, •25.00, •OQ. 00 ,00, .00. .00, .00. .00) .00) .00) .00) .00) .00) TYPE GRID- ID OC DC DC OC DC DC RECEPTOR TYPES: GC * GRIOCART 6P « GRIDPOtR " DC « DISCCART OP * DISCPOLR n • BOUNDARY D-20 TUT oos •143? •** MEAST - VERSION TESTA *•* •*• AREA SOURCES-•- 1/2 acre run TEST OF ST AREA SOURCE ALGORITHM **• •*• MODELING OPTIONS USED: CONC RURAL FLAT DFAULT *** Message Senary For ISC2 Model Execution *•• ......... Sumary of Total Messages "••••-• » Total of 0 Fatal Error Message'(s) A Total of 0 Warning Message(s) A Total of 653 Informational Messaged) A Total Of 653 Cat* Hours Identified FATAL ERROR MESSAGES UMtNING MESSAGES *** NONE *** ISCST2 Finishes Successfully D-21 TUT 008 1433 APPENDIX E Determination of Ground Water Dilution Attenuation Factors TUT COS 1434 DETERMINATION OF GROUNDWATER DILUTION ATTENUATION FACTORS FOR FIXED WASTE SITE AREAS USING EPACMTP BACKGROUND DOCUMENT EPA OFFICE OF SOLID WASTE May 11,1994 R04-M.489 E-l TUT 008 1435 PREFACE The work documented in this report was conducted by HydroGeoLogic, Inc. for the EPA Office of Solid Waste. The work was performed partially under Contract No. 68-WO-0029 and partially under Contract No. 68-W3-0008, subcontracted through ICF Inc. This documentation was prepared under Contract No. 68-W4-0017. Technical direction on behalf of the Office of Solid Waste was provided by Dr. Z.A. Saleem. E-2 TUT 008 1436 ABSTRACT , # The EPA Composite Model for Leachate Migration with Transformation Products (EPACMTP) was applied to generate Dilution Attenuation Factors (DAF) for me groundwater pathway in support of the development of Soil Screening Level Guidance. The model 'was applied on a nationwide basis, using Monte Carlo simulation, to determine DAFs as a function of the area of me contaminated site at various probability levels. The analysis was conducted in two stages: First, the number of Monte Carlo iterations required to achieve converged results was determined. Convergence was^ defined as a change of less man 5% in the 85th percentile DAF value. A number of 15,000 Monte Carlo iterations was determined to yield convergence; subsequent analyses were performed using mis number of iterations. Second, Monte Carlo analyses were performed to determine DAF values as a function of the conomiinated area. The effects of different placements of the receptor well war evaluated. E-3 TUT OOS 1437 TABLE OF CONTENTS Page PREFACE . . . . . . . . . . . . . . . . .,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii 1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1,1 2.0 GROUNDWATER MODEL ................................. .2-1 2.1 Description of EPACMTP Model ......................... 2-1 2.2 Fate and Transport Simulation Modules ...................... 2-3 2.2.1 Unsaturated zone flow and transport module .............. 2-3 2.2.2 Saftmf** zone flow and transport module ................ 2-3 2.2.3 Model capabilities and limitations ..................... 2-5 2.3 Monte Carlo Module ................................ .2-8 2.3.1 CapatrilitfeE and Limitations of Monte Carlo Module ........ 2-11 3.0 MODELING PROCEDURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-1 3.1 Modeling Approach ......... . . . . . . . . . . . . . . . . . . . . . . . .3-1 3.1.1 Determination of Monte Carlo Repetition Number and Sensitivity Analysis ................................... .3-1 3.1.2 Analysis of DAF Values tor Different Source Areas ......... 3-4 3.1.2.1 Model Options and Input Parameters . ............. 3-4 4.0 RESULTS. ............................................ .4-1 4.1 Convergence of Monte Carlo Simulation ................ . . ... 4-1 4.2 Parameter Sensitivity Analysis .......................... .4-1 4.3 DAF Values as a Function of Source Area ................... .4-6 R-l E-4 . TUT 008 1438 LIST OF FIGURES Page Figure 1 Conceptual view of the ""ywryted zone-saturated zone system simulated by EPACMTP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-2 Figure 2 Conceptual Monte Carlo framework for deriving probability distribution of model output from probability distributions of input parameters. .... 2-10 Figure 3 Flow chart of EPACMTP for Monte Carlo simulation. ........... 2-12 Figure 4 Plan view and cross-section view showing location of receptor well. .... 3-8 Figure 5 Variation of DAF with size of source area for the base case scenario (x«=25 ft, y ""uniform in phone, z«natkmwide distribution). ........ .4-7 Figure 6 Variation of DAF with size of source area for the default nationwide scenario (Scenario 2: x-nationwide distribution, y~imifonn in plume, z^nationwide distribution). ............... .- ............. 4-8 Figure 7 Variation of DAF with size of source area for' Scenario 3 (x«0, y -uniform within half-width of source area, z«nationwide distribution). ..................................... .4-9 Figure 8 Variation of DAF -with size of source area for Scenario 4 (x«=25 ft, yarunifonn within half-width of source area, z«nationwide distribution). ................... ^ ................. 4-10 'Figure 9 Variation of DAF with size of source area for Scenario 5 (x-100 ft, y«uniform within half-width of source area, z«nationwide distribution). ..................................... 4-11 Figure 10 Variation of DAF with size of source area for Scenario 6 (x«=25 ft, y«.width of source area + 25 ft, z-25 ft). .................. 4-12 E-5 TUT 008 LIST OF TABLES * * . Page Table 1 Parameter input values for model sensitivity analysis. . . . . . . . . . . . . . 3-3 Table 2 Summary of EPACMTP modeling options. . ^ . . . . . . . . . . . . . . . . . . 3-5 Table 3 Summary of EPACMTP input-parameters. 'i .................... .-3-6 Table 4 Receptor well location scenarios. ....:..................... 3-9 Table 5 Distribution of aquifer particle diameter. .................... 3-11 Table 6 Variation of DAF with number of Monte Carlo repetitions. .......... 4-2 Table 7 Sensitivity of model parameters. .......................... .4-3 Table8 DAF values for waste site area of 150,000 ft 2. .................. 4-13 Table Al DAF values as a function of source area for base case scenario (x=25 ft, yuniform in plume, z-nationwide distribution). ............... A-l Table A2 DAF values as a function of source area for Scenario 2 (x^nationwide distribution* y«uniform in phone, z*nationwide distribution). ...... A-2 Table A3 DAF values as a function of source area for Scenario 3 (x«0 ft, yKuniform within naif-width of source area, z«nationwide distribution). ..................................... A-3 Table A4 DAF values as a function of source area for Scenario 4 (x-=25 ft, y«sunifonn wimin half-width of source area, z-nationwide distribution). ..................................... A-4 Table A5 DAF values as a function of source area for Scenario 5 (x«100 ft, within half-width of source, z«*nationwide distribution). . . . A-5 Table A6 DAF values as a function of source area for Scenario 6 (x«25 ft, y= source width + 25 ft, z«25 ft). . . . . . . . . . . . . . . . . . . . . . . . A-6 E-6 TUT ooe 1440 1JO INTRODUCTION The Agency is developing estimates for threshold values of chemical concentrations in soils at contaminated sites that represent a level of concentration above which there is sufficient concern to warrant further site-specific study. These concentration levels are called Soil Screening Levels (SSLS). The primary propose of the SSLs is to accelerate deeigum making eftnrymJng contaminated soils. Generally, if contaminant concentrations in soil fall below the screening level and die site meets specific residential use conditions, no further, study or action is warranted for that area under CERCLA (EPA, 1993b). The Soil Screening Levels have been developed using residential land use human exposure assumptions and considering multiple pathways of exposure to the contaminants, including migration nf contaminants through soil to an underlying potable aquifer. Contaminant migration through the unsamrated zone to the water table fc^«rally reduces the soil kachate concentration by attenuation processes such as adsorption and degradation. Groundwater transport in the saturated zone further reduces concentrations through attenuation atv^ dilution. The contaminant concentration arriving at a receptor point in the saturated zone, e.g., a domestic drinking water well, is therefore generally lower than the original contaminant concentration in the soil leachate. The reduction in concentration can be expressed succinctly in a Dilution-Attenuation Factor (DAF) defined as the ration of original soil leachate concentration to the receptor point concentration. The lowest possible value of DAF if therefore one; a value of DAF«1 means that there is no dilution or attenuation at all; the concentration at ffr 8 receptor point is die same as mat in the soil leachate. High values of DAF on the other hand correspond to a high degree of dilution *pd attenuation. For any specific site, the DAF depends on the interaction of a multitude of site-specific factors and physical and bio-chemical processes. The DAF also depends on the nature of the contaminant itself; i.e., whether or not the chemical degrades or sorbs. As a result, it is impossible to predict DAF values without the aid of a suitable computer fate and transport E-7 ' - TUT 008 1441 Simulation model that simulates the migration of a contaminant through the subsurface, and accounts for the relevant mechanisms and processes that affect the receptor concentration. The Agency has ,. developed the EPA Composite Model for T garhatr Migration with Transformation Products (EPACMTP; EPA, 1993a, 1994) to assess the groundwater quality impacts due to migration of wastes from surface waste sites." This model simulates the fate and f rnmamtnamc after their release from me land disposal mrit into the aoU, downwards to the water table and subsequently through the saturated zone. The fate and transport model has been coupled to a Monte Carlo driver to permit determination of DAFs on a generic, nationwide basis. The EPACMTP model has been applied to determine DAFs for the subsurface pathway for fixed waste site areas, as pan of the development of Soil Screening Levels. This report describes the application of EPACMTP for this purpose. E-8 TUT 008 2.0 GROUNDWATER MODEL •• * 2.1 Description of EPACMTP Model The EPA Composite Model for Leachatr Migration with Transformation froducts (EPACMTP, EPA, 1993a, 1994) is a computer model for simulating me subsurface fate and transport of t contaminants mat are released at or near me soil surface. A schematic view of the conceptual subsurface system as sforotetpd by EPACMTP, is shown in Figure 1. The g^p^n*™*11** are initially released over a «^*Mig»i^r source area representing the waste site. The modeled subsurface system consists of an unsaturated zone underneath the source area, and an underlying water table aquifer. Contaminants move vertically downward through the unsaturated zone to the water table. The contaminant is assumed to be dissolved in the aqwosiy phase; it migrates through the soft under the influence of downward infiltration. The rate of infiltration may reflect the combined effect of precipitation and releases from the source area. Once the contaminant enters the saturated zone, a three-dimensional plume develops under the combined influence of advection with the ambient groundwater flow and dispersive mixing. . , The EPACMTP accounts for the following processes affecting contaminant fate and transport: advection, dispersion, equilibrium sorption, first-order decay reactions, and recharge dilution in the saturated zone. For contaminants mat transform into one or more daughter products, the model can account for the fate and transport of those transformation products also. The EPACMTP model consists of three main modules: • An unsaturatrd zone flow and transport module • A saturated zone flow and transpon module • A Monte Carlo driver module, which generates model input, parameter values from specified probability distributions The assumptions of the unsaturated zone and saturated zone flow and transpon modules are described in Section 2.2. The Monte Carlo modeling procedure is described in Section 2.3. E-9 TUT 008 1443 r Receptor wen. Leakage from Contaminated Area Unsaturated Zone Saturated ~ Zone Ambient Groundwater Row Figure 1 Conceptual view of the unsaturated zone-saturated zone system simulated by EPACMTP. E-10 TUT O08 1444 2.2 Fate and Transport Simulation Modules .. » /*w»v . _ Details on the mathematical formulation and solution techniques of the »Tim*"rntt>d zone flow and transport module are provided in the EPACMTP background document (EPA, 1993a). For completeness, the major features and assumptions are summarized below: • The source area is a rectangular area. - • Contaminants are distributed uniformly over the source area. • The soil is a uniform, isotropic porous medium. • Flow and transport in me unsaturated zone are one-dimensional, downward. • Flow is steady state, and driven by a prescribed rate of infiltration. • Flow is isothermal and governed by Darcy's Law. • The leachate concentration entering the soil is either constant (with a finite or infinite duration), or decreasing with time following a "^" first-order decay process. • The chemical is dilute and present in solution or son solid phase only. • Sorption of chemicals onto the soD solid phase is described by a (Freundtich) Chemical and biological transformation process can be repr by an effective, first-order decay coefficient 2.2.2 Saturated zone flow and transport module The imgflturated rone mnrtiil* eftmpMtag fly e«nt*tnmant enmeentratinn arriving jf tfa» <vj^ff tah|«» as a function of time. Multiplying mis concentration by the rate of infiltration through the i zone yields the contaminant mass flux entering the ff^'^Tfd y»iffr This tn*8* flux is specified as the source boundary condition for the saturated zone flow and transport module. E-ll TUT 008 1445 Groundwater flow in the saturated zone is simulated using a (quasi-) three-dimensional steady state solution for predicting hydraulic head and Darcy velocities hi a constant thickness groundwater system subject to infiltration and recharge along the top of the aquifer and a regional hydraulic .gradient defined by upstream and downstream head boundary conditions. In addition to modeling fully thny-d«nenginnai groundwater flow «nd contaminant fate transport, EPACMTP offers the option to perform quasi-3D modeling. When mis option is •elected, the model ignores either the flow component in the horizontal transverse (-y) direction, wor the vertical (-z) direction. The appropriate 2D approximation is selecttd automatically in the code, based on the relative gignffieanee of phone movement in the horizontal transverse versus vertical directions. Details of mis procedure are provided in the saturated zone background document (ERA, 1993a). The switching criterion mat is implemented in the code will select the 2D area! solution for situations with a relatively thfo saturated zone m which the contaminant plume would occupy the entire saturated thickness; conversely, the solution in which advection in the horizontal transverse direction is ignored is used in situations with a large saturated thickness, in which the effect of vertical phone movement is more important. The saturated zone transpon module describes the advective-dispersive transpon of dissolved in a three-dimensional, constant »friffrn» >ss aquifer. The initial boundary is zero, and the lower aquifer boundary is taken to be impermeable. No-flux conditions are set for the upstream aquifer boundary. Contaminants enter the sanitated zone through a patch source of either constant concentration or constant mass flux on the upper aquifer boundary, representing the area directly underneath the waste site at me soil surface. The source may be of a finite or infinite duration. Recharge of contaminant-free infiltration water occurs along the upper aquifer boundary outside the patch source. Transpon T'lMfratitgm* considered are advection, longitudinal, vertical and transverse hydrodynamic dispersion, linear or nonlinear equilibrium adsorption, first-order decay and daughter product formation. As in the unsaturated zone, the saturated zone transport module CUP fjjtnniate multi-species transpon involving ^V"**** decay reactions. The saturated zone transpon module of EPACMTP can perform either a fully three- dimensional transpon simulation, or provide a quasi-3D approximation. The latter ignores E-12 TUT advection in either the horizontal transverse (-y) direction, on the vertical (-z) direction, consistent with the quasi-3D flow solution. In the course of a Monte Carlo simulation, the appropriate 2D approximations are selected automatically for each individual Monte Carlo iteration, thus yielding an overall quasi-3D simulation. The saturated zone and transport module is based on the following assumptions: • The aquifer is uniform and initially contaminant-free. • The flow field is at steady state; seasonal fluctuations in groundwater flow are neglected. The MflnriTrd f*"rfrrw?g of the mprifrr is represented by the head distribution along the top boundary of modeled saturated tone system. Flow is isothermal and governed by Darcy's Law. The chemical is dilute *i*d present in flip solution or aouifer solid phase only. Adsorption onto the solid phase is described by a linear or nonlinear equilibrium isotherm. Chemical and/or biochemical transformation of the contaminant can be described as a first-order process. 2.2.3 Model capability ^nd Hniltttions EPACMTP is based on a number of simplifying assumptions which make the code easier to use and ensure its computational efficiency. These assumptions, however, may cause application of the model to be inappropriate in certain situations. The main assumptions embedded in the fate and transport model are summarized in the previous sections and are discussed in more dc**'! here. The user should verify th«f the assumptions are reasonable for a given application. E-13 TUT DOS 1447 Uniform Porous Soil and Aquifer Medium. EPACMTP assumes that the soil and aquifer behave as uniform porous media and that flow and transport are described by Darcy's law and the advection-dispersion equation, respectively. The model does not account for the presence of cracks, macro-pores, and fractures. Where these features are present, EPACMTP may underpredict the rate of contaminant movement. Single Phase Flow and Transport. The model assumes that the water phase is the only mobile phase and disregards interphase transf er processes other man reversible adsorption onto the solid phase. For example, the model does not account for volatilization in the unsaturated zone, which will tend to give conservative predictions for volatile chemicals. The model also does not account for the presence of a second liquid phase (e.g., oil). When a mobile oil phase is present, the movement of hydrophobia chemicals may be underpredicted by me model, since significant migration may occur in the oil phase rather than in the water phase. Equilibrium Adsorption. The model assumes that adsorption of contaminants onto the soil or aquifer solid phase occurs instantaneously, or at least rapidly relative to the rate of contaminant movement. In addition, the adsorption process is taken to be entirely reversible. Geochemistry. The EPACMTP model does not account for complex geochemical processes, such as ion exchange, precipitation and complexation, which may affect the migration of chemicals in the subsurface environment. EPACMTP can only approximate such processes as an effective equilibrium retardation process. The effect of geochemical interactions may be especially important in the fate and transport analyses of metals. Enhancement of the model for a wide variety of geochemical conditions is currently underway. First-Order Decay. It is assumed that the rate of contaminant toss due to decay reactions is proportional to the dissolved contaminant concentration. The model is based on one overall decay constant and does not explicitly account for multiple degradation processes, such as oxidation, hydrolysis, and biodegradation. When multiple decay processes do occur, the user E-14 TUT 008 1443 •must determine the overall, effective decay rate. In order to increase flexibility of the model, the user" may instruct the model to detennine the overall decay coefficient from cbemicaTspecifk hydrolysis constants plus soil and aquifer temperature and pH. Prescribed Decay Reaction Stoichiometry. For scenarios -Involving chained decay reactions, EPACMTP assumes that the reaction Stoichiometry is always prescribed, and the speciaucn factors are specified by the user as constants (see EPACMTP Background Document, EPA, 1993a). In reality, these coefficients may change as ranctions of aquifer cx>nditiom(t pH, etc.) and/or concentration levels of other rldnical Uniform SoU. EPACMTP assumes mat the unsamrated zone profile is homogeneous. The model does not account for the presence of cracks and/or macropores in the soil, nor does it account for lateral soil variability. The latter condition may significantly affect the average transpon behavior when the waste source covers a large area. Steady-State Flow in the Unsaturuted-Zone. Flow in the unsansated zone is always treated as steady state, with the flow rate determined by the long tetm, average infiltration rate through a disposal unit, or by the average depth of ponding in a surface inuwondment. Considering me time scale of most practical problems, •«q'«»n£ steady-state flow conditions in the unsamrated zone is reasonable. Groundwater Mounting. The ««*»r«««*i zone module of EPACMTP is Hftigno! to simulate flow and transport m an unconfined aquifer.. Ground water mounding beneath the source is lepieseuied only by increased head values on top of the aquifer. The sanitated thickness of the aquifer remains constant in the model, and therefore the model Heats the aquifer as a confined system. This approach is reasonable as long as the mound height is small relative to the saturated «Mdfn»y« of the aquifer and. the thickness of the unsamrated zone. For composite modeling, the effect of mounding is partly accounted for in the unsamrated zone module, since the soil is allowed to become saturated. The aquifer porous material is assumed to be uniform. E-15 - • 008 1449 although the model does account for anisotropy in the hydraulic conductivity. The lower aquifer boundary is assumed to be impermeable. Flow in the Saturated Zone. Flow in the saturated- zone is taken to be at steady state. The concept is that of regional flow in the horizontal longitudinal"drrection, with vertical disturbance .due to recharge and infiltration from the overiying unsamxated zone and waste site (source area). EPACMTP tny-^T"rc for variable recharge rates underneath and outside the source area. It is, however t pgyiitpeH that the mtnmted «me lag • constant thielcness, which may in the predicted groundwater flow and contaminant transport in cases where the infiltration rate from the waste disposal facility is high. . Transport in the Saturated Zone. Contaminant transport in the saturated zone is by advection *n$ dispersion. The aquifer is fl^p*!"^ to be initially contaminant free and contaminants enter the aquifer only from the it"<anTratffl zone immediately underneath the waste site, which is modeled as a rectangular horizontal plane source. EPACMTP can simulate both steady state and transient transport in the saturated zone. In the former case, the contaminant mass flux entering at the water table must be constant with time. In the latter case, the flux at the water table can be constant or vary as a function of time. The transport module accounts for equilibrium adsorption aTV* decay reactions, bom of which are modeled in the same tranter as in the unsaturated zone. The adsorption and decay coefficients are assumed to be uniform throughout saturated zone. 23 Monte Carlo Module EPACMTP was designed to perform simulations on .a nationwide basis, and to account for variations of model input parameters reflecting variations in site and hydrogeological conditions. The fate and transport model is therefore linked to a Monte Carlo driver which generates model input parameter values from the probability distribution of each parameter. The Monte Carlo modeling procedure is described in more detail in mis section. E-16 TUT 008 1.450 The Monte Carlo method requires that for each input parameter, except constant parameters a probability distribution is provided. The method involves the repeated generation of pseudo- random values of the uncertain input variable(s) (drawn from the known distribution and within the range of any imposed bounds) and the application of the model using these values to generate a series of model responses (receptor well concentration). These responses are men statistically analyzed to yield the cumulative probability distribution of die model output. Thus, me various steps involve"! in the application of the Monte Carlo ihimlation technique are: (1) Selection of representative cumulative probability distribution functions for die relevant input variables. (2) Generation of a pseudo-random number from the distributions selected in (1). These values represent a possible set of values (a realization) for the input variables. (3) Application of the fate *nd transport «j""ii«tV*n nKvfyiijy to compute the output(s), i.e., downstream well concentration. (4). Repeated application of steps (2) and (3) for a specified number of iterations. • (5) Presentation of the series of output (random) values generated in step (3). (6) Analysis of the Monte Carlo output to derive regulatory DAF values. The Monte Carlo module designed for implementation with the EPACMTP composite model performs steps 2-5 above. This process is shown conceptually in Figure 2. Step 6 is performed as a post-processing step. This last step supply involves converting the normalized receptor well concentrations to DAF values, and ranking then for high to low values. Each Monte Carlo iteration yields one DAF value for the constituent of concern (phis one DAF value for each of the transformation products, if the constituent is a degrader). Since each Monte Carlo iteration has equal probability, ordering the DAF values from high to low, directly yields men- cumulative probability distribution (CDF). If appropriate, CDF curves representing different regional distributions may be combined into a single CDF curve, which is a weighted average of the regional curves. E-17 TUT COS 1451 Input Para**i»r$ O O E L' On I Value Input Distributions Vfelue Output Distribution Figure 2 Conceptual Monte Carlo framework for deriving probability distribution of model output from probability distributions of input parameters. E-18 TUT 008 A simplified flow chart that illustrated the linking of the Monte Carlo module to the simulation modules of the EPACMTP composite model is presented in Figure 3. The modeling input data is read first, and subsequently the desired random numbers are generated. The generated random and/or derived parameter values are men assigned to the model variables. Following mis, the enntammant transport fate *i»d transport simulation is performed. The result is given in terms of the predicted contaminant concentration(s) in a down-stream receptor well. The generation of random parameter values and fate and transport simulation is repeated as many times as desired to determine the probability distribution of down-stream well concentrations. 23.1 Capabilities and Limitations of Monte Carlo Module The Monte Carlo module in EPACMTP is implemented as a flexible module f»»* can accommodate a wide variety of input distributions. These include: constant, normal, log- normal, exponential, uniform, Iog10 uniform,, Johnson SB, empirical, or derived. In addition, specific upper and/or lower'bounds can be provided for each parameter. The empirical distribution is used when the data does not fit any of the other probability distributions. When the empirical distribution is used, me probability distribution is specified in tabular form as a list of parameter values versus cumulative probability, from zero to one. It is important to realize that the Monte Carlo method accounts for parameter variability and uncertainty; h does, however, not provide a way to account or compensate for process uncertainty. If the actual flow and transport processes mat may occur at different sites, are different from those sJrmilntftd in the fate and transport module, the result of a Monte Carlo analysis may not accurately reflect the actual variation in groundwater concentrations. EPACMTP does not directly account for potential statistical dependencies, i.e., correlations between parameters. The probability distributions of individual parameters are considered to be statistically independent. At the same time, EPACMTP does incorporate a number of safeguards against generating impossible combinations of model parameters. Lower and upper bminris nn th» rffTn"rtm prrvrm wir<l*li«rie»11y |«w * E-19 008 Read input data and desired modeing options Yes Post Processing Perform Sftnutation Print Results Figure 3 Flow chart of EPACMTP for Mome Carlo simulation. E-20 TUT OO8 1454 In the case of model parameters that have a direct physical dependence on other parameters, these parameters can be specified as derived parameters. For instance, the ambient groundwater flow rate is determined by the regional hydraulic gradient and the aquifer hydraulic conductivity. In the Monte Carlo analyses, the ambient groundwater flow rate is therefore calculated as the product of conductivity and gradient, rather than generated independently. A detailed discussion of the derived parameters used in the model is provided in the EPACMTP User's Guide (EPA, 1994). E-21 TUT Oos 1455 This page left blank on purpose. E-22 .• . T U T 0 0 8 3.0 MODELING PROCEDURE This section documents the modeling procedure followed in determining the groundwater pathway DAF values for the Soil Screening Levels. Section 3.1 describes the overall approach for the modeling analysis; section 3.2 describes the model options used and summarizes tbe input parameter values. . 3.1 Modeling Approach The overall modeling approach consisted of two stages. First, a sensitivity analysis was performed to determine tbe optimal number of Monte Carlo repetitions required to achieve a stable and converged result, and to determine which site-related parameters have the greatest impact on the DAFs. Secondly, Monte Cano analyses were performed to determined DAF values as a function of the size of tbe source area, for various scenarios of receptor well placement. 3.1.1 Determination of Monte Carlo Repetition Knmber and Sensitivity Analysis The criterion for determining the optimal number of Monte Carlo repetitions was set to a change in DAF value of no more than 5 percent when die number of repetitions is varied. A Monte Carlo simulation comprising 20,000 repetitions was first made. Tbe results from <frfo simulation were analyzed by calculating tbe 83th perceotile DAF value obtained by sampling model output sequences of different length, from 2,000 to the full 20,000 repetitious. The modeling scenario considered in <frfc analysis was the same as mat in the base case scenario diEnisfted in the next section, with the size of the source area set to 10,000 m2. The sensitivity analysis on site-related model parameters was performed by fixing one parameter at a time, while remaining model parameters were varied according to their default, nationwide probability distributions as discussed in the EPACMTP User's Guide (EPA, 1993b). E-23 TUT 008 1457 For each parameter, the low, medium, and high values were selected, corresponding to the 15th, 50th, and'S5th percentile, respectively, of that parameter's probability distribution. As -a result, the sensitivity analysis reflects, in part, the width of each parameter's probability distribution. Parameters with a narrow range of variation will tend to be among the less sensitive parameters, and vice versa for parameters that have a wide range of variation. By conducting the sensitivity analysis as a series of Monte Carlo simulations, any parameter interactions on die model output are automatically acc^un?*^ for. Each of the Monte Carlo simulations yields a probability distribution of predicted receptor well concentrations. Evaluating die distributions obtained with different fixed values of the same parameter provides a measure of the overall sensitivity and impact of that parameter. In each case the model was run for 2000 Monte Carlo iterations. Steady-state conditions (continuous source) were simulated in all cases. In a complete Monte Carlo analysis, over 20 different model parameters are involved. These parameters may be divided into two broad categories. The first includes parameters mat are independent of contaminant-specific chemical properties, e.g., depth to water table, aquifer thickness, receptor well flisfrnce, etc. The second category encompasses those parameters mat are related to contaminant-specific sorption and biochemical transformation characteristics. This category includes the organic carbon partition coefficient, but also parameters such as aquifer pH, temperature and fraction organic carbon. The sensitivity of the model to the first category of parameter* hag eratninftd toy e^ngvterfrig » tinturiggraditig, «muMifwi£ enantamtnant Under these conditions, any parameters in the second category will have zero sensitivity. In addition, all unsaturated zone parameters can be left out of the analysis, since the predicted steady state contaminant concentration at the water table will always be toe same as mat entering the vnsyniprte^ zone. The only exception to mis is the soil type parameter. In the nationwide Monte Carlo modeling approach, different soil types are distinguished. Each of the three different soil types (sandy loam, silt loam or siity clay loam) has a different distribution of infiltration rate, with the sandy loam soil type having the highest infiltration rates, silty clay loam having the lowest, and silty loam having intermediate rates. The effect of the soil type parameter is thus intermixed with that of infiltration rate. Table 1 lists the input 'low', •medium' and *high' values for all the parameters examined. E-24 TUT OO8 145s Table 1 •Parameter input values for model sensitivity analysis. Parameter Source Parameters Source Area (m2) Infiltration Rate (m/yr) Recharge Rate (m/yr) Saturated Zone Parameters Saturated Thickness (m) Hydraulic conductivity (m/yr) Regional gradient Ambient groundwater velocity (m/yr) Porosity Longitudinal Dispersivity (m) Transverse Dispersivity (m) Vertical Dispersivity (m) Low 4.8X10* 6.0XHT* 6.0x10-* 15.55 1.9X10? 4.3 xl(T3 53.2 0.374 4.2 0.53 0.026 Median . 2.8 x 10s 6.4XHT3 8.0 X10-3 60.8 1.5X10* 1.8 X10-2 404.0 0.415 12.7 1.59 0.079 High l.lxlO6 1.7 x 10-' 1.5x10-' 159.3 5.5x10* 5.0 xia2 2883.0 0.455 98.5 12.31 0.62 E-25 TUT OO8 1459 3.1.2 Analysis of DAF Values for Different Source Areas Following completion of the sensitivity analysis discussed above, an analysis was performed of the variation of DAF values with size of the contaminated area. The sensitivity analysis, results of which are presented in Section 4.1, showed that the size of the contaminated source area is one of the most sensitive parameters in the model. For the purpose of deriving DAF values for the groundwater pathway in determining soil screening levels, it would therefore be appropriate to correlate the DAF value to the size of the contaminated area. ' / ' The EPACMTP modeling analysis was designed to determine the .size of the contaminated area that would result in DAF values of 10 and 100 at the upper 85th, 90th, and 95th percentile of probability, respectively. Since it is not possible to directly determine the source area that results hi a specific DAF value, the model was executed for a range of different source areas, using a different but fixed source area value in each Monte Carlo simulation. The 85th, 90th, and 95th percentile DAF values were then plotted against source area, in order to determine the value of source area corresponding to a specific DAF value. 3.1.2.1 Model Options and Input Parameters Table 2 gmn^arfrj* the EPACMTP model options used in performing the simulations. Model input parameters used are gmmwTfwi m Table 3. The selected options and input parameter distributions and values are consistent with those used in the default nationwide modeling, and are discussed individually in the EPACMTP User's Guide (EPA, 1994). Exceptions to this default modeling scenario are discussed below. Source Area In the default, nationwide modeling scenario, the waste she area, or source area, is treated as a Monte Carlo variable, with a distribution of values equal to that of the type of waste unit, e.g. landfills, considered. In the present modeling analyses, the source area was set to a different but constant value hi each simulation run. E-26 TUT 008 1460 Table 2 Summary of EPACMTP modeling options. OPTION Simulation Type Number of Repetitions Nationwide Aggregation Source Type Unsat. Zone Present Sat. Zone Model Contaminant Degradation Contaminant Sorption Value Selected Monte Carlo . 15,000 Yes Continuous Yes Quasi-3D No No E-27 TUT 008 1461 Table 3 Summary of EPACMTP mput parameters. Parameter Value or Distribution Type Source-Specific Area Infiltration Rate Recharge Rate Leacfaate Concentration Chemical-Specific Hydrolysis Rate Constants Organic Carbon Partition Coen. Unmunted ZOTI* Specific Depth to Water Table Dispenivity Soil Hydraulic Properties Soil Chemical Properties Constant Soil-type dependent Soil-type dependent • 1.0 • 0.0 • 0.0 Empirical •Soil-depth dependent Soil-type dependent Soil-type dependent Varied in each ran defult defmlt default mant ooes not default default Sat. Zone Thickness Hydraulic Conductivity Hydraulic Gradient Seepage Velocity Panicle Diameter Porosity Bulk Density Longitudinal Dispenivity Transverse Dispenivity Vertical Dispenivity Receptor Well x-coordmate Receptor Well y-coordinate Receptor Well z-coordinate Derived from Pan. Diam. Derived from Conductivity and Gradient Empirical Derived from Pan. Diam Derived from Porosity Derived from Long. Dispenivity Derived from Long. Dispenivity • 25 feet Within plume Empirical defmlt default default defmlt defmlt defmlt defmlt defmlt Set to fixed value defmlt defmlt Note: •Default' represents defmlt nationwide Monte Carlo scenario as presented in EPACMTP User's Guide (EPA. 1994). E-28 TUT OOQ ±462 Receptor Well Location In the default nationwide modeling scenario, the position of the nearest downgndient. receptor well in the saturated zone is treated as a Monte Carlo variable. The position of the well is defined by its x-, y-, and z-coordinates. The x-coordinate represents the distance along the ambient groundwater flow direction from die downgradient edge of the contaminated area. The y-coordinate represents the horizontal transverse distance of the well from the plume centerline. The x-, and y-coordinate in turn can be defined in terms of an overall downgradient distance, and an angle off-center (EPA, 1994). The z-coordinate represent the depth of the well intake "point below the water table. This is illustrated schematically in Figure 4, which shows the receptor well location in bom plan view and cross-sectional view. In the default nationwide modeling scenario, the x-, and z-coordinates of the well are determined from Agency surveys on the distance of residential wells from municipal landfills, and data on the depth of residential drinking water weds, respectively. The y-coordinate value is determined so that the well location rails within the approximate area! extent of the contaminant plume (see Figure 4). For the present modeling analysis, a number of different receptor well placement scenarios were considered. These scenarios are summarized is Table 4. The base use scenario (scenario 1) mvolved setting the x-distanc£ of the receptor well to 25 feet from the edge of the source area. Nationwide default options were used for the receptor well y. and z-coordinates. The y-coordinate of the well was assigned a uniform probability distribution within the boundary of the phme. The depth of the well intake pomt (z-coordinate) was assmped to vary within upper and lower bounds of IS and 300 feet below the water table, reflecting a national sample distribution of depths of residential drinking water wells (EPA, 1994). In addition to this base case scenario, a number of other well placement scenarios were investigated also. These are numbered in Table 4 as scenarios 2 through 6. Scenario 2 E-29 TL>T 008 1463 GROUNDWATER FLOW CONTAMINANTED AREA RECEPTOR WELL LAND SURFACE UNSATURATED ZONE WATER TABLE GROUNDWATER FLOW Figure 4 Plan view and cross-section view showing legation of receptor well. E-30 oos 1464 Table 4 Receptor well location scenarios. Scenario Xwell 1 (Base 2 3 4 5 6 Xwell Ywell Zweil - Case) 25 ft from edge of source area Nationwide Distribution 0 ft firom edge of source area 25 ft firom_cdge of source area 100 ft firom edge of source area 25 ft firom edge of source •area Downgradicnt distance of receptor i HnnTftftfal trancvWRft idisfflnp^ In Mil Ywell Monte Carlo within plume * Monte Cario within plume Monte Carlo within half- width of source area Monte Carlo vffhn1 half- width of source area M0nt» Curio «Tthili half- width of source area Width of source area + £5 ft Zwell Nationwide Distribution Nationwide Distribution Nationwide Distribution Nationwide Distribution Nationwide Distribution 25 ft below water table Bvell firom edge of source area, nhnne centerline. «= Depth of well intake point below water table. E-31 TUT 008 1465 corresponds to the default, nationwide Monte Carlo modeling scenario in which the x, y, and z locations of the well are all variable. In scenarios 3, 4 and 5, die distance between the receptor well and the source area is varied from zero to 100 feet. In these scenarios, the y- coordinate of the well was constrained to the central portion of the plume. In scenario number 6, the x-, y-, and z-coordinates of the receptor well were-all set to constant values. These additional scenarios were included in the analysis in order to assess the sensitivity of the model results to the location of the receptor well. • . Aouifer Particle Size Distribution In the default Monte Carlo modeling scenario, the aquifer hydrtulk conductivity, porosity, and bulk density are dctermi***^ from the mean particle diameter. The particle ^ipTvfgr distribution used is based on data compiled by Shea (1974). In the present modeling analyses for fixed waste site areas, the same approach and data were' used, but the distribution was shifted somewhat to assign more weight to the smallest particle diameter interval. The result is that lower values of the hydraulic conductivity values generated, and also of the ambient groundwater seepage velocities, received more emphasis. Lower ambient groundwater velocities reduce the degree of dilution of the incoming contaminant plume and therefore result in lower, i.e. more conservative, DAF values. Table 5 «««marh*« the distribution of particle size diameters used in both the default nationwide modeling scenario and in the present analyses. . E-32 TUT OO8 1466 Table 5 Distribution of aquifer particle diameter. Nationwide Particle Diameter (cm) 3.9 10-4 7.8 104 1.6 ID"3 3.1 1(T3 6.3 1(T3 1.25 10-2 2.5 10-2 5.0 1(T2 1.0 Ifr1 2.0 10-1 4.0 10-' 8.0 10-1 Default Cumulative Probability 0.000 0.038 0.104 0.171 0.262 0.371 0.560 0.792 0.904 0.944 0.946 1.000 Present A Particle Diameter (on) 4.0JO-4 8.0 104 1.6 10-3 3.1 10-3 6.3 Ifr3 1.25 10-2 2.5 10-2 5.0 10-2 1.0 10-' 2.0 10-1 4.0 10-' 7.510-' Cumulative Probability 0.100 0.150 0.200 0.270 0.330 0.440 0.590 0.790 0.880 0.910 0.940 1.000 E-33 TUT 008 1467 This page left blank on purpose. E-34 . TUT OO8 1.468 4.0 RESULTS .. \ • • ** This section presents the results of the modeling analyses performed. The analysis of the convergence of the Monte Carlo simulation is presented first, followed by the parameter sensitivity analysis, and thirdly the analysis of DAF values as a function of source area for various well placement scenarios. 4.1 Convergence of Monte Carlo Simulation Table 6 agiminarfcey the results of this convergence analysis. It shows the variation of the 85th percentile DAF value with the number of Monte Cario repetitions, from 2,000 to 20,000. The variations in DAF values are shown both as absolute and relative differences. The table shows that for this example, the DAF generally increases with the number of Monte Carlo repetitions. It should be kept in mind that the results from different repetition numbers as presented in the table, are not independent of one another. For instance, the first 2000 repetitions are also incorporated in the 5000 repetition results, which in turn is in the 10,000 repetition result, etc. The .rightmost column of Table 6 shows the percentage difference in DAF value between different repetition numbers. At repetition numbers of 14,000 or less, the percentage difference varies in a somewhat irregular manner. However, for repetition numbers of 15,000 or greater, the DAF remained relatively constant, with incremental changes of DAF fgmafriing at 1% or less. Based upon these results, a repetition number of 15,000 was selected for use in the subsequent runs with fixed source area. 4.2 Parameter Sensitivity Analysis Results of the parameter sensitivity analysis are summarized in Table 7. The parameters are ranked in this table in order of relative sensitivity. Relative sensitivity is defined for this purpose as the absolute difference between the "high" and "low" DAF at the 85th percentile level, divided by the 85th percentile DAF for the "median" case. E-35 TUT 008 1469 Table 6 'Variation of DAF with number of Monte Carlo repetitions. No. of Repetitions 85-th PercentUe DAF 2,000 5,000 10,000 11,000 12,000 13,000 14,000 15,000 16,000 17,000 18,000 19,000 20,000 347.8 336.9 354.2 359.2 387.4 369.3 369.1 387.3 387.4 388.0 387.3 390.2 392.8 Difference Relative Difference (%) -10.9 4-17.3 4-5.0 4-28.2 -18.1 -0.2 4-18.2 4-0.1 4-0.6 -0.7 4-2.9 4-2.6 -3.1 4-5.1 4-1.4 4-7.9 -4.7 -0.05 4-4.9 4-0.03 4-0.15 -0.18 4-0.75 4-0.67 E-36 TUT OO8 1470 Table? Sensitivity of model parameters. 85% DAF Value Parameter Infiltration Rate Saturated Thickness G.W. Velocity ' Source Area Hydr. Conductivity Vertical Well Position G.W. Gradient Long. Dispersivity Vert. Dispersivity Porosity Receptor Well Distance Transv. Dispersivity Receptor Well Angle Ambient Recharge Low 4805.4 25.3 7.6, 357.1 19.8 49.1 32.4 182.6 179.6 41.3 163.9 156.7 127.3 108.3 Median 418.8 198.5 97.7 85.2 180.4 206.1 168.3 104.2 114.9 49.9 117.9 156.3 130.8 100.0 High ir.6 2096.9 816.3 35.6 660.1 491.4 383.0 78.8 66.6 79.7 84.5 173.5 113.6 114.4 • Relative ^ Sensitivity 11.4 10.4 8.3 3.8 3.5 2.1 2.1 1.0 1.0 0.8 0.7 0.1 0.1 0.06 Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 * Relative Sensitivity = |High-Low]/Median E-37 TUT 008 .1.471 The table shows that the most sensitive parameters included the rate of infiltration, which is a function of soil type, the saturated thickness of the aquifer, the size of source-area, the groundwater seepage velocity, and the vertical position of the receptor well below the water table. The least sensitive parameters included porosity, downstream distance of the receptor well in both the x- and y-directions, the horizontal transverse dispersivity, and the area! recharge rate. To interpret these results, it should be kept in mind mat the rankings reflect in part the range of variation of each parameter in the data set used for the sensitivity analysis. The infiltration rate was a highly sensitive parameter since, for a given leachate concentration, h directly affects the mass flux of contaminant entering the subsurface. The size of the source are would be expected to be equally sensitive, were it not for the fact mat in the sensitivity analysis, the source area had a much narrower range of variation than the infiltration rate. The "high" and "low" values of the source area, which were taken from a nationwide distribution of landfill waste units, varied by a factor of 23, while the ratio of "high* to "low* infiltration rate was almost 300. ; In the simulations performed for the sensitivity analysis, no constraint was imposed on the vertical position of the well. The well was modeled as having a uniform distribution with the well intake point located anywhere between the water table and the base of the aquifer. The aquifer saturated thickness and vertical position of the well were both among the sensitive parameters, with similar effects on DAF values. Increasing either the saturated thickness, or the fractional depth of the receptor well below the water table, increases the likelihood mat the receptor well will be located underneath the contaminant plume ygd sample uncontaminated groundwater, leading to a high DAF value. The dilution-attenuation Actors were also sensitive to the groundwater velocity, and the parameters that determine the groundwater velocity, i.e., hydraulic conductivity and ambient gradient. Table 7 shows mat a higher groundwater velocity results in an increase of the dilution-attenuation factor. Since a conservative contaminant was simulated under steady-state conditions, variations in travel time do not affect the DAF. The increase of DAF with increasing flow velocity reflects the greater mixing and dilution of the contaminant as it enters the saturated zone in systems with high groundwater flow rate. Porosity E-38 TUT 008 also directly affects the groundwater velocity, but was not among the sensitive parameters. This is a reflection of the narrow range of variation "dgn*ti to mis parameter. The off-center angle which determines the y position of the well relative to the plume center line would be expected to have a similar effect as the well depoVbut is seen to have a much smaller sensitivity. This was a result of constraining the y-locatibn of the receptor well to be always inside the approximate areal extent of the contaminant plume. The effect is that the relative sensitivity of the off-center angle was much less than mat of the vertical coordinate of the well. The low relative sensitivity of recharge rate reflects the Act that this parameter has an only indirect effect on phnne concentrations. Overall, the Monte Carlo results were not very sensitive to dispersivity and downstream distance of the receptor well. The probable explanation for these parameters is that variations of the parameters produce opposing effects which tended to cancel one. another. Low dispersivity values will produce a compact phnne which mcreases the probability mat a randomly located receptor well will lie outside (underneath) the phnne. Higher dispersivities will increase the chance that the well will intercept the phnne. At the same tone, however, mass balance considerations dictate *fa»* in th« case average concentrations inside the phnne will be lower *h*« in the low dispersivity case. Similar •^•nfnrc applies to the effect of receptor well distance. If the well is located near the source, concentrations in the plume will be relatively high, but so is the chance that the well does not intercept the phnne at all. At greater dimmm from the source, the likelihood that the well is located inside the phnne is greater, but the phnne will also be more diluted. In the course of a full Monte Carlo simulation these opposing effects would tend to average out. The much lower sensitivity of transverse dispersivity, otr, compared to aL and o» can be contributed to the imposed constraint that me weU must always be withm me areal extent of the plume. The results of the sensitivity analysis show that the she characteristic which tends itself best for a classification system for correlating sites to DAF values is the size of the contaminated (or E-39 TUT OOS 1473 source) area. In the subsequent analyses, the DAF values were therefore determined as a function of the source area size. These results are presented in the following section. 4.3 DAF Values as a Function of Source Area This section presents the DAF value as a function of source area for various well location scenarios. The results for each of the scenarios examined are presented in tabular and graphical form. Figure 5 shows the variation of the 85th, 90th. and 95th percentile DAF with source area --for the base case scenario. The source area is expressed in square feet. The figure displays w DAF against source area in a log-log graph. The graph shows an approximately linear relationship except that at very large values of the source area, the DAF starts to level off. Eventually the DAF approaches a value of 1.0. As expected, the curve for the 95th percentile DAF always shows the lowest DAF values, white the 85th percentile shows me highest DAFs. The DAF versus source area relationship for the other well placement scenarios are shown in Figures 6 through 10. The numerical results for each scenario are summarized in Tables Al through A6 in the appendix. Inspection and comparison of the results for each scenario indicate mat the relationship follows the same general shape in each case, but the TnTignfr" 1* of DAF values at a given source area can be quite different for different well placement scenarios. In order to allow a direct comparison between the various scenarios analyzed, the DAF values obtained for a source area of 150,000 ft2 (3.4 acres) are shown in Table 8 as a function of the receptor well location scenario. Inspection of the DAF values shows that the default nationwide scenario for locating the receptor well results in the highest DAF values, as compared to the base case scenario and the other scenarios, in which the receptor well location was fixed at a relatively close distance from the waste source. In the default nationwide modeling scenario, the well location is assigned from nationwide data on both the distance from the waste source and depth of the well intake point below the water table. In the default nationwide modeling scenario, the receptor well is allowed to be located up to 1 mile from the waste source. In the base case (Scenario 1) the well is E-40 TUT 008 1474 T o 8 f,OE+07| 1.0E 1.0E 1.0E 1.0E 1.0E 1.0E+01 1.0E+00 1.0E+03 407 MK- 85TH-W- 90 TH iTH Figures Variation otDAF with size of! (x=25 10 1 .OE+04 j..___4........I......J 1.0E+03r Q 1.0E+02; 1.0E+01: 1.0E+00- 1.0E+03 JV;~ 1.0E+04 Bel : w - rt strlb. .] 2w»H-Na!fc>nwfctodt»trib. | ""SL 1.0E4-OS AREA OF LANDFILL .oe+oe 1.0E+07 85 TH 90 TH -®- 95 TH Figure 6 Variation of DAF with size of source area for the default nationwide scenario (Scenario 2: x=nationwide distribution, y=uniform in phime, z-nationwide distribution). m H I—-f CO I.OE+OSa 1.0E407 1.0E+06- LOE+OSs 1.0E403 1.0E+05 AREA OF LANDFILL 65 TH -*- 90 TH -**- 05 THJ 1.0E+07 Figure 7 Variation of DAP with size of source area for Scenario 3 <x«0. y-uniform within half-width of source area, z-nationwide distribution). 8 * £ 8I wI E-44 TUT OOS 1478 c H oo 03 «.'~5 -3 wj, . Wf 1.0E+05: 1.0E+04 1.0E+03 1.0E+02 1.0E-I-01 1.0E4-00 1.0E+03 1.0E-f04 1.0E+05 AREA OF LANDFILL (sq ft) 1.0E+06 85 TH -l~ 90 TH -*- 95 TH 1.0E+07 Figure 9 Variation of DAF with size of source area for Scenario 5 (x=100 ft, y=uniform within half-width of source area, z=nationwide distribution). 8 8 e» N «»+! I •5! >» c ^ & *o &I -^ 1 •sII t>I ^'»I a E-46 TUT 008 1480 Table 8 DAF values for waste she area of 150,000 ft2. Model Scenario 1 (base case) 2 3. 4 5 6 85 237.5 300.1 158.8 132.1 98.8 94.7 DAF Pelcentile 90 26.4 114.7 .17.9 16.6 15.1 25.3 95 2.8 26.8 1.7 1.8 2.0 4.4 E-47 TUT COS 1481 allowed to be located anywhere within the areal extent of the contaminant plume for a fixed x- distancc'of 25 feet. This allows the well to be located near the fringes of the contaminant plume where concentrations are relatively low and DAF values are correspondingly high. In contrast,- in Scenarios 3, 4, and 5, the well location was constrained to be wuhinthe half-width of the waste source. In other words, the well was always placed in the central portion of the contaminant phone where concentrations are highest. As a result, these scenarios show lower DAF values then the base case scenario. The results far Scenarios 3,4, and 5, which differ only in the x-distaace of the receptor well, show that placement of the well w either 25 or 100 feet away from the waste source results in 85% and 90% DAF values mat are actually tower, i.e. more conservative, thanplacememof the weU directly at the edge oft he waste source. This is a counter-intuitive result, but may be explained from the interaction between distance from the waste source and vertical extent of the contaminant Close to the waste source, the contaminant concentrations within the phone are highest, but the plume may not have penetrated very deeply into the saturated zone (Figure 2). Because the vertical position of the well was taken as a random variable, with a maximum value of up to 300 feet, the probability that a receptor well samples pristine groundwater underneath the contammant phone is higher at close **<****** from the waste area. Convener/, as the distance from the source increases, the plume becomes more dilute but also extends deeper below the water table. The final result is that the overall DAF may actually decrease with distance from the source. The table also shows mat at the 95% level, me lowest DAF is obtained in the case where me well is located at the edge of the waste source. This reflects mat the highest concentration values will be obtained only very close to the waste source. . The results for the last scenario, in which the x, y, and z locations of the receptor well were all fixed, show that fixing the well depth at 25 feet ensures mat the wen is placed shallow enough that it will be located inside the phone in nearly all cases, resulting in low DAF values at the 85th and 90th percentile values. On the other hand, the well in mis case is never placed immediately at the plume centerline, so mat the highest concentrations sampled in this scenario .are always lower than in the other scenarios. This is reflected in the higher DAF value at the 95th percentile level. - • E-48 TUT 008 .1482 One of tbe key objectives of the present analyses was to detennine tbe appropriate groundwater DAF value for a waste area of given size. For die base case scenario, tbe 90th percentile DAF value is on the order of 100 or higher for a waste area size of 1 acre (43,560 ft7) and less. For waste areas of 10 acres and greater, the 90th percentile DAF is 10 or less. £-49 OO8 This page left blank on purpose. E-50 TUT OOS 1484 REFERENCES Shea, J.H., 1974. Deficiencies of elastic panicles of certain sizes. Journal of Sedimentary Petrology, 44:905-1003. U.S. EPA, 1993a. EPA's Composite Model for I.radmc Migration with Transformation Products: EPACMTP. , Volume I: Background Document. Office of Solid Waste, July 1993. * * U.S. EPA, 1993b. Draft Soil Screening Level Guidance Quick Reference Fact Sheet. Office of Solid Waste and Emergency Response, September 1993. U.S. EPA, 1994. Modeling Approach for Simulating Three-Dimensiontl Migration of Land Disposal I**"***** with Transformation Products and Consideration of Water-Table Mounding. Volume II: User's Guide for EPA's Composite Model for Leachatc . Migration with Transformation Products (EPACMTP). Office of Solid Waste, May 1994. E-51 TUT oos 14SS E-52 TUT OO8 1486 APPENDIX A E-53 TUT OO8 1487 E-54 TUT OOS 1488 Table Al DAF values as a function of source area for base case scenario (x< y=unifonn in plume, z-nationwide distribution). =25 ft; Area <«nr 1000 2000 5000 10000 30000 50000 70000 80000 150000 4EUUUUU 500000 1000000 2000000 3000000 5000000 OAF 85 TH 1.09E+06 1.86E+05 2.91 E+04 9.31 E+ 03 1647.18 669.57 569.80 477.33 237.47 174.86 64.52 »F"^«Wfc 3227 17.83 12.04 8.91 90 TH 3.76E+04 9.63E+03 2.00E+03 68027 15521 8425 5928 5036 26.36 20.19 O12 5.61 3.68 2A4 ^••^F^ 2,33 95 TH 609.01 167.69 53.02 22,57 7.82 5.41 4.34 3.97 2.77 2.37 1.61 122 1.16 1.11 1.06 E-55 TUT OOS 1489 Table A2 DAF values as a function of source area for Scenario 2 (x«nationwide distribution, y«unifonnin plume, z«nttionwide distribution). Area (sq.tl) 5000 8000 10000 45000 50000 100000 150000 220000 500000 1000000 5000000 6000000 DAF 85 TH 6222.78 3977.72 3215.43 817.66 745.16 424.81 300.12 218.67 110.35 63.45 21.03 10.06 90 TH 2425.42 1573.32 1286.01 315.06 26827 160.62 114.71 62.30 40.10 23.75 . 7.85 7.01 95 TH 565.61 371.06 298.78 73.48 6720 38.11 26.82 20.00 1022 622 2JSS 229 E-56 TUT 008 1490 Table A3 DAF values as a function of source area for Scenario 3 (x-0 ft, y "uniform within half-width of source area, z«*nationwide distribution). Area (•q.tt.) 1000 2000 5000 10000 30000 . 50000 70000 80000 150000 200000 500000 1000000 2000000 SQOQOOO OAF 85 1.42E+07 9.19E+05 534E+04 1.16E+04 1.43E+03 668.45 417.19 850.39 158.76 114.63 4055 91 4*9 cl.lo 1138 866 90 2.09E+05 2.83E+04 2.74E+03 644.33 120.42 60,02 87.97 33.16 17.87 12,96 534 8 JO W4MPW £38 1 08 95 946.07 211.15 44.23 15^9 4.46 3.10 233 2.34 1.74 136 1^3 1.15 1.08 1.O6 E-57 TUT OOS 1491 Table A4 DAF values as a function of source area for Scenario 4 (x«2S ft v«,~-. Within half-width of source area, x-iationwLtelSbUon) ymvutm. Area (*q:1t) 1000 2000 5000 10000 4htf%tfk4Ut 90000 50000 70000 80000 150000 200000 500000 1000000 2000000 3000000 DAF 85 5.93E+05 1.09E+05 1.64E+04 4.89fc>03 82631 49020 323.42 272.85 132.05 97.94 37^9 20.06 1155 8.49 90 2.07E+04 4A2E+03 1.03E+03 352.49 93.96 49.76 34.79 29.62 1655 1229 530 330 2.40 2.00 95 348.31 118.11 29.86 13.14 • 4.73 328. 2.69 2.47 1.82 1.61 129 1.17 1.10 1.07 E-58 rUT OO8 1492 Table AS DAF values as a function of source area for Scenario 5 (x« 100 ft, y«*uniform within half-width of source, z«*nationwide distribution). Area (sq.ft.) 1000 2000 5000 10000 9OQQO ^nrinnf 50000 70000 60000 150000 200000 500000 1000000 2000000 29LAAJLJLIU DAF 85 424E+04 1.52E+04 424E+03 1.61E+03 49727 293.3* 207.77 164.57 98.81 74.63 32M 18.66 11.14 fi.33 90 3.43E+03 1.33E-I-03 43725 20429 6621 40.72 29.69 26.86 15.05 11.55 5.83 3.71 2£3 2-Qfi 95 181J8 74.79 2723 13.09 5.10 3.71 2*6 2.73 2.03 1JB2 1 40 126 1.16 1.13 E-59 TUT OOS 1493 Table A6 DAF Values as a function of source area for Scenario 6 (x=25 ft, y«source width-f 25 ft, z=25 ft). AREA (SOFT) 1200 1500 5000 7500 * 4lM^\r 23000 26000 29000 100000 170000 250000 800000 1800000 DAF 85 TH 44247.70 30759.77 478927 269633 637.76 544.66 462.63 139.66 76.69 50.40 18.10 1026 90 TH 10478.98 7215.01 1273.40 725.69 • •••PMI^^ 155.16 135.91 121.43 3535 2124 15.04 6.04 3.87 95 TH 1004.72 744.05 140.61 8231 21.82 18.84 16.52 5.56 3.94 . 3.19 1.81 1.48 E-60 TUT 008 1494 APPENDIX F Dilution Factor Modeling Results TUT O08 1495 -0 0- III I •••i •i*I HHIH If !M 9 IHHHHHHr !linil ! 11IS 'Iff!! > f 3 R c o o g ft s I a - 9 2 C M C M C M C M O t C M C M t S - p l C M a t B 8 ! i 8 B 8 t l ! S K 8 B ! J B B 5 B 5 J ! 2 B 8 B a t i q ^ 8 8 8 i J 9 f t 4 f t 6 8 9 9 & f e 8 ! 1 8 8 9 8 f e S 1 8 l t 8 S | l : t 9 f t f t f t S I 8a8aa^889289888^29 sas 9aa9 aaas 92H233Si32 si iis328*82i 122 133218 • M 8 8 - ft g ft * « t^ -itt ** ^* § 35 3 o . « •* S - S S £-•:£-•(*«» ft£ -»B**-*<*:£-*-* 18 CM CM - 8 5 - - - * g M HI -. ^ ft •» •» - «S I I! § s 8 g s Iff!| j ~ i ft £ M s •D K t| •_. *> u 5 8 is C o o Oi§ 38c C2 t U g Dilution Factor Model Results: DNAPL Sites Source aba (i 05 10 30 100 600 45 201 349 636 1559 ________9.1 •MM " — •• •• ii aMB nBfne mtmvm Unemaster Switch Corp. Maryland Sand, Gravel & Stone McNnCo. MaW Banks Mottoto WQ FeWn Mw MviutKturtnQ NCR Corp. MBsboro Norwood PCBS Nyanza Chemicals OCpnrwr Company OMCttyofYortcUndlUl Old SObiteJiiykni LJbvJQtt Old Sprtnpfitld LwflDl Otbonw LrfwiJiUl Ofe A* Natl. Guard Ottatf * Goes/KJnoston Drums Pease Air Force Base PelanorVPurttan. Inc. PteUoFarm Plneae* Salvage Yard PSC Resources Re-Solve, toe. RecttcervABed Steel nhkiohaWt Tire RfB Saoo Tannery Waste Pits Sawders Supply Co. Savage Mun. Water Supply Sttaakn Chamtaal Corp. Somersworth San. Landm South Municipal Water Supply Southern MD Wood Treating Stamina MBs, toe. Ski CWortne/ryboiirs Comer LF SvasaurgUandni SuBverts Ledge Ouenif County UndBH tS QylMBiter'i TansJtor Elactrartcs TfcbetsRoad US Detanee Qenarnl Supply US Dover AFB US Naval Mr Development US Newport Nav. Educ ATm. Ctr. CT MD ME PA NH PA DE MA MA ME PA CT VT PA MA NH NH Rl Rl ME MA MA PA VA ME VA NH MA NH MM Pin MD Rl DE PA MA DE NH VT NH VA DE PA Rl MA Infiltration by Uu*4 lleinann nyv* nBBpVQfi Region 7 10 9 9 9 8 7 8 7 9 9 9 9 9 9 9 9 6 6 8 10 9 . 9 9 9 10 9 10 8 9 10 9 9 .9 8 10 6 9 9 (rrvyr) 022 0.15 022 0.15 022 020 024 022 022 022 0.15 020 022 020 022 022 022 022 022 022 022 022 0.15 0.15 022 024 022 022 022 022 024 022 024 0.15 022 024 022 022 022 0.15 024 0.15 022 022 Average OW Velocity (m/yr) Csapagi Oarcy 1.113 2 680 5 131 21.027 223 389 39 214 779 134 30 1.113 312 46 11 56 834 333 45 634 73 1.346 66 26 235 26 139 80 2 2506 39 2,160 11fc 196 490 103 28 37 4 66 3 445 369 1 312 2 46 7,359 . 78 136 14 75 273 47 11 389 109 16 4 19 187 117 16 282 26 471 19 10 82 8 49 32 1 663 14 756 39 69 171 36 10 13 • 1 19 1 156 Mixlnc 05 10 5 21 10 30 5 21 6 29 5 22 5 21 5 22 5 22 5 . 24 5. 22 5 21 5 22 6 25 5 21 5 22 5 24 7 28 5 23 5 22 5 22 5 24 5 21 5 22 5 21 5 23 6 25 5 22 6 25 5 22 5 23 12 30 5 21 6 24 6 21 5 22 6 22 5 22 5 22 6 25 5 23 10 30 5 23 11 30 5 22 1 VM9MA j4a%8»VA»j * ! ZOO* Ovplti ate (acres) •0 100 37 . 68 46 78 37 68 46 76 36 70 37 87 36 68 37 66 41 74 36 68 37 68 36 70 42 74 37 68 36 68 41 73 45 76 40 72 37 68 36 68 41 73 37 68 38 70 37 68 40 72 42 75 38 68 42 75 36 70 » «M 7l 46 76 37 87 41 74 37 67 39 70 36 68 37 68 38 70 42 75 40 72 46 76 38 71 .'46 76 37 68 600 166 174 166 174 170 165 KB. 167 174 168 166 170 174 166 166 173 174 173 167 167 173 166 171 165 173 174 168 174 170 171 174 165 174 165 171 KB 167 171 174 173 174 172 174 167 Dilution factor 05 189 2 152 3 24 3JB6 36 68 8 38 184 27 7 206 54 10 4 11 82 66 8 142 20 334 11 6 42 6 25 17 2 475 8 635 21 33 84 19 7 11 2 . 16 2 77 Srto eta (acre*) 10 30 100 600 61 47 1 1 65 38 1 1 10 6 1473 866 16 10 28 17 4 3 16 10 64 49 12 7 3 2 69 82 24 14 4 . 3 2 " 1 S .3 40 23 25 15 4 3 61 36 9 5 144 63 5 3 3 2 18 11 3 2 11 7 8 5 1 1 204 118 4 2 _230 133 8 6 14 9 36 21 • 5 3 2 5 3 1 1 7. 4 1 1 33 20 26 11 1 1 21 9 1 1 4 2 530 217 6 3 10 5 2 1 6 3 27 12 4 2 2 1 29 12 8 4 2 1 1 1 2 2 13 6 » 4 2 1 •— 20 9 3 2 46 19 2 2 2 1 6 3 2 1 4 2 3 2 1 1 65 27 2 1 73 31 4 2 5 3 12 6 3 2 2 1 2 2 1 1 3 2 1 1 11 5 F-2 TUT 008 1497 Dilution Factor Model Results: DNAPL Sites Source stn(icfM) Source tangtti (m) as. 10 so too aoo 45 201 349 636 1,559 •_____9.1 /•"•-v. A^te U«*MA •A^AA •Mi NWIW TOM Wwtam Sand & Gnvti WwttnghouM Etovttor iMnVwop LwJnU **•- ~~**1I««» f^MIM^I 1 BM^MII Rl PA ME MO MUtttenby Oktfl P^filnr> rryB* nvsicin R^ten <m/yr) 9 022 6 0.15 9 O22 8 0.15 AVMQCOW * wodty (nvyvj SMpage Dwey 48 17 882 197 16 6 S57 195 Mixing zone depth SHt *»(«*••} &5 10 30 100 800 5 24 40. 73 173 8 21 37 88 188 6 27 44 7ft 174 •5 21 37 88 186 Dilution factor OS 10 141 5 140 SI»«ln(aefM) 10 30 100 4 3 2 81 35 20 2 2 1 80 35 20 800 1 9 1 ' 9 F-3 TUT O08 1498 Dllntlon Factors (DFs) tor 208 Sites In the fiydrogeotoglc Database (HGDB) - National Average I Source Ungth(m) Seure* An* f MIM) 0.5 I 10 45 1 201 30 349 100 636 600 1,559 Hytfrogootoglc SaWn0 1.11 1.11 1.3 1.6 1.6 1.7 1.7 1.8 1.9 1.9 2.12 2.12 2.12 2.12 2.12 2.12 2.12 ' 2.13 2.13 2.13 2.3 2.3 2.4 2.4 2.4 2.4 2.4 2.4 2.4 2.4 Infiltration 0.30 0.30 0.03 0.08 0.08 0.14 0.14 0.03 0.08 0.08 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 AVMWQO 1C 63 946 63 946 5,676 157,660 192.370 63,072 125.829 2,759,400 126 946 1.369 1,577 1.577. 23.652 31.536 95 . 158 2.838 318 5,992 315 315 631 1,892 4,100 11,038 16,714 107.222 Hya. Qrao> (mAn) 3.00E-02 1.00E-02 8.00E-02 9.306-02 2.006-03 1.006-04 1.006-02 5.00E-03 1.00E-03 3.006-02 2.006-03 2.006-03 3.006-03 1.006-03 5.006-03 3.006-03 1.006-03 3.006-04 1.006-03 2.006-03 5.706-03 1.006-03 1.006-03 2.006-03 1.006-02 1.006-03 1.006-03 2.006-03 4.006-03 Darcyv 2 9 S 68 .11 16 1.924 315 126 •2.762 0.3 2 4 2 8 71 32 0.03 0.2 '6 2 6 0.3 1 6 ' 2 4 22 67 536 Aq.Tnlck. 30 305 23 15 21 3 6 2 8 23 5 3 91 914 24 6 24 9 130 30 46 183 15 3 9 37 3 . 13 6 7 Calcutatatf Mhlng ZOM Depth (<Q 0.5 11 6 5 5 5 5 5 5 5 5 6 5 5 5 5 5 5 14 12 5 10 6 18 8 6 10 6 5 5 5 Source ATM (Km) 10 41 28 22 21 23 23 21 21 21 21 26 23 22 25 22 21 21 30 50 22 40 28 ' 37 24 26 38 24 23 22 21 30 62 48 39 37 39 39 37 37 37 37 42 39 39 43 38 37 37 46 83 38 64 49 32 40 44 61 40 40 38 37 100 96 87 70 68 71 70 67 67 68 67 72 70 71 77 69 68 68 76 138 70 104 89 83 70 78 99 70 72 69 68 600 195 211 172 166 173 168 165 165 168 165 170 188 174 190 170 188 166 174 276 171 210 213 160 168 174 201 168 174 168 166 Mutton Factor (OF) Some* Area (acres) 0.5 3 5 . 23 124 16 9 1.459 421 169 114.973 2 6 19 8 ' 35 298 133 1 3 26 3 8 . 1 1 8 3 2 13 35 265 10 2 5 23 89 17 3 419 95 39 114.973 1 ' 2 19 9 35 88 . 133 1- 3 26 3 5 1 1 2 3 1 8 11 91 30 2 5 14 51 10 2 242 55 ' 23 71.160 1 2 19 9 23 50 88 1 2 21 2 5 1 1 2 2 1 5 7 S3 100 " . 1 5 8 29 6 2 133 31 13 39.049 1 1 19 9 13 28 49 1 2 12 2 5 1 1 1 2 1 3 4 • 30 . moi 5 4 12 3 1 55 13 6 15.931 1 1 11 9 6 12 20 1 2 5 1 4 2 2 13 r ;•-•( o ill •43 ••0 DIhltlon Factors (DFs) for 208 Sites In the Hydrogeologic Database (HGDB) - National Average [ Source Length (Hi) source wee (eciMj 0.5 45 10 201 30 349 100 638 800 1,559 Hydregaetegte Salting 2.4 2.4 2.5 2.5 2.5 2.5 2.5 2.5 2.6 2.8 2.8 2.9 2.9 2.9 2.9 2.9 3.7 ; 3.7 4.1 4.2 4.2 4.4 4.4. 5.2 5.3 5.8 5.8 8.11 8.11 6.11 HtflltMnlOII (nW) 032 0.22 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.03 0.03 0.03 0.03 0.03 0.14 0.14 0.03 0.03 0.03 0.14 0.14 0.03 0.03 0.03 0.03 0.22 0.22 0.22 (nVW 190,793 3,311.280 948 1,281 4,418 6.938 23,337 58.134 1.577 13,878 50.773 126 1.281 3,469 22.075 220,752 220,752 298,438 • 32 22 284 946 9,776 242,827 2417.898 831 33,113 1,577 4.415 4.415 HyAOrad (mini) 1.006-03 6.006-03 2.00E-03 3.006-03 7.006-04 3.006-03 4.006-03 2.006-03 1.006-03 2.806-02 5006-03 2.006-03 1.006-04 2.006-02 1.006-03 1.006-03 2.006-03 2.006-04 1.006-01 2.806-02 3.206-03 8.006-03 1.306-02 2.006-03 2.006-03 3.006-03 2.006-06 1.006-02 5.006-03 1.006-02 Darcyv 191 16,556 2 4 3 21 93 112 2 389 254 0.3 0.1 09 22 . 221 442 59 3 1 1 8 127 486 4,636 2 0.07 16 22 44 Aq. Thick, (m) .8 18 8 305 38 23 37 10 12. 34 9 11 18 IS 9,1 15 9 9 21 - 11 3 3 • 3 17 12 24 34 24 15 21 r CefeuletMf (ailing Zone Depth (d) Source Area (acre*) 0.5 5 5 10 8 9 5 5 5 11 5 5 8 12 5 5 5 5 5 5 6 6 5 5 5 5 5 18 5 5 6 10 21 21 29 37 37 24 22 22 33 21 22 31 38 21 22 21 21 22 23 27 24. 23 21 21 21 24 51 24 23 22 30 37 37 45 84 60 42 38 38 49 37 37 47 58 37 37 37 37 38 40 45 40 40 37 37 37 41 70 41 40 39 100 68 67 76 114 98 75 89 69 79 68 68 78 88 68 68 67 w 69 72 77 70 70 68 67 67 75 101 75 72 70 600 167 165 173 288 202 179 ; 170 188 177 168 167 176 183 188 167 165 165 168 174 178 188 168 188 185 165 179 199 179 175 171 • DttufIon Factor (OF) Source Area (acres) 0.5 96 8,118 2 3 3 9 34 41 2 137 90 3 2 291 94 922 338 47 . 15 4 3 5 83 2.025 19.317 10 2 10 13 24 10 35 6,978 1 3 3 0 34 19 1 137 39 2 1 208 94 680 . 145 20 14 2 2 2 IS 1,598 11.072 10 1 10 9 23 30 20 4.019 1 3 2 5 33 12 1 123 23 1 ,. 1 120 94 381 84 12 9 2 . 1 1 9 919 8.377 6 1 6 5 14 100 12 2.208 1 3 2 3 19 7 1 66 13 1 1 66 94 209 46 7 5 1 1 1 5 505 3.500 4 1 4 3 8 6> 1 9C 1 ' 3 1 2 8 3 1 21 6 ' 1-!. I 28 52 80 20 3 3 1 1 1 3 207 1,421 2I 2 2 4 c o 03 cn Dilution Factors (DFs) for 208 Sites In the Hydrogeologlc Database (HGDB) - National Average | Source Length (m) Source Area (aeree) 0.5 45 10 —201 30 . 349 100 638 800 1,559 1 ———— 1 — -1 — ••* ———————— . Soltln| 6.11 6.12 6.12 6.12 6.14 6.14 6.14 6.14 6.14 6.2 6.2 6.2 6.2 6.3 6.3 6.4 6.S 6.5 6.S 6.S 6.S 6.S 6.8 7.11 7.11 7.12 7.12 7.12 7.13 7.13 IbtArt Wf*Jl 042 0.03 o.«e 0.03 0.22 0.22 0.22 0.22 0.22 0.14 0.14 0.14 0.14 0.03 0.03 0.06 0.14. 0.14 0.14 0.14 0.14 0.14 0.14 0.22 0.22 0.22 0.22 0.22 0.14 0.14 Avenge K (nvw 81.994 946 3,154 315 1,577 1.892 5,670 14.191 33,113 126 3 1.325 2.208 1.892 31.538 0.776 83 189 315 315 31,536 34.690 2£08 95 2,523 4,100 12,614 116.052 3.154 5.519. I —— f —— n Hyd\Ofwl. 3.006-03 8.00E-03 6.006-03 1.706-02 4.00E-02 2.00E-03 1.006-03 7.006-04 1.00E-02 4.00E-03 1.006-02 S.OOE-03 3.306-02 4.306-02 1.40E-01 1.206-02 4.006-02 2.306-02 S.OOE-03 2.506-02 5.006-02 8.006-03 2.506-02 6.006-03 2.006-02 3.006-03 4.906-02 4.006-03 1.306-02 1.006-02 246 8 19 5 63 4 6 10 331 1 0.03 7 73 81 4.418 117 3 4 2 8 1,577 278 55 0.6 SO 12 618 464 41 55 Aq. Thick. 9 6 3 9 8 6 6 18 23 . 8 S 21 30 6 3 30 20 61 21 19 6 5 2 4 3 32 6 . 76 17 5 Calculated Mixing Zone Dtpth (d) 0.5 5 5 5 5 5 7 6 6 5 11 9 6 5 5 5 5 7 6 8 6 S 5 . 5 9 S 6 S 5 5 5 Source Area (acres) 10 21 22 22 22 22 26 26 25 21 29 26 25 22 21 21 21 30 27 33 25 21 . 21 22 25 22 25 21 21 22 22 30 37 38 37 38 38 43 42 43 37 45- 42 43 38 37 37 37 49 47 53 42 37 37 38 41 38 43 37 . 37 38 38 100 68 69 68 70 69 73 73 77 68 75 72 77 69 68 67 68 " 85 87 76 67 88 68 71 69 77 68 68 69 69 166 169 16t 170 169 171 171 180 168 173 170 182 168 165 165 168 185 199 186 180 185 166 168 168 168 183 166 166 170 168 Source Arm (acres) 0.5 123 34 51 24 33 3' 5 •7 164 • 2 1 7 57 341 11.774 165 4 5 . 3 8 1,197 203 14 1 17 8 305 230 33 44 10 49 10 12 11 12 2 2 S 164 1 1 8 57 98 2.637 165 3 s. 2 6 343 46 4 1 5 8 92 230 25 12 30 29 r 6 8 7 7 1 1 3 101 1 1 4 47 57 1.519 135 2 S 2 4 . 198 27 3 1 3 6 54 230 15 7 100 16 4 5 4 5 1 1 2 56 1 1 3 26 32 834 75 2 4 1 3 109 IS 2 1 2 4 30 230 9 4 7 2 2 2 2 1 1 2 23 1 1 2 11 14 341 31 1 2 1 2 45 7 1 1 1 2 13 106 4 2 Z 03 Dilution Factors (DFs) for 208 Sites (n the Hydrogeologlc Database (HGDB) • National Average 1 SourctLangHifrn) 8ovrc4 AIM (acne) 0.8 45 10 I 30 I 100 201 1 349 J 630 600 1,559 Hydrog*ologle Settfng 7.13 7.14 7.14 7.14 • 7.14 7.14 7.14 7.15 7.15 7.16 7.18 7.17 7.17 7.17 T.17 7.17 7.17 7.17 7.17 7.17 7.17 7.17. 7.16 7.3 7.3 7.4 7.4 7.4 7:5 7J tnffltratton t^tt il t"Vyi 0.14 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 933. 022 0.22 0.30 0.30 AwfB06 K IfflM 15,453 6,307 6.938 11,038 14.507 17,660 23,652 7.253 24,314 221 3.154 19 32 32 63 126 318 319 946 3,164 3,469 21.760 1.892 •48. 25344 • 189 2.681 3,784 63 11,038 Hyvf* OHM* (mfrn) 6.00E-03 4.906-02 4.00E-03 2.SOE-01 1.206-02 2.006-03 3.30E-02 6.006-04 6.806-03 4.00E-03 3.00E-03 8.006-03 9.006-03 3.006-02 2£06-02 1.506-01 1.006-03 7.00E-03 5.006-02 1.00E-02 1.706-02 4.00E-03 8.00E-03 5.00E-03 9.00E-04 1.206-02 9.006-03 4.006-02 7.006-03 8.006-04 Darcyv (mfV) 93 309 28 2.759 174 35 781 4 165 1 9 0.2 0.3 1 1 19 0.3 2 47 32 59 67 9 8 23 2 24 151 0.4. 6 Aq. Thick, (m) 6 5 8 5 18 43 18 37 11 8 9 5 3 11 3 30 12 23 14 5 85 15 1 8 4 61 2 2 518 23 Cateutotod MMno ZOM Offrth (d) Sown Aim (acrm) 0.5 5 5 5 5 5 5 5 8 5 9 5 10 ' 8 10 7 5 15 7 8 5 8 8 8 8 8 9 8 8 35 7 10 22 21 23 21 22 23 21 33 22 29 24 27 24 31 24 23 33 31 22 22 22 22 22 28 22 38 23 22 143 30 30 37 37 40 37 38 40 37 88 38 48 41 42 40 48 40 39 49 61 38 38 38 37 36 41 39 63 39 37 230 50 100 68 68 72 67 68 72 68 93 68 78 73 73 70 78 70 72 79 66 69 69 69 68 68 72 70 108 70 68 363 65 600 167 166 172 165 168 177 168 200 168 173 173 170 168 176 168 175 177 188 169 169 169 167 166 170 168 221 167 166 618 187 DHutton Factor (OP) Source AH» (acres) 0.5 72 109 . 12 921 62 14 273 3 59 2 . 9 1 1 ' 2 ' 2 16 2 4 36 24 47 68 2 8 14 3 7 25 2 4 10 27 27 5 207 S3 14 234 3 30 1 4 1 1 1 1 16 1 3. 24 6 47 48 1 2 . 4 3 2 6 2 3 30 16 16 3 120 31 14 135 2 18 1 3 1 1 1 1 13 1 2 14 4 47 28 1 2 3 3 2 4 2 2 100 9 9 2 66 17* 9 75 2 10 1 2 1 1 1 1 7 1 2 . 8 3 37 16 1 1 2 2 1 3 2 2 6 2 1 t a ' 1 • 1 1 1 2 16 7 2 1 1 Dilution Factors (DFs) for 208 Sites In the Hydrogeologic Database (HGDB) • National Average | Source Length (m) Source AIM (aerta) 0.5 45 10 201 30 349 100 638 600 1.559 Hydrogeologic Selling 7.6 7.6 7.6 7.7 7.7 7.8 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9- 7.9 7.9 7.9 7.9 7.9 7.9 8:6 9.12 ______ Infiltration (nVy) 0.14 0.14 0.14 0.14 0.14 0.14 • 0.22 0.22 0.22 0.22 0.22 0.22 0.22 032 0.22 0.22 0.22 0.22 . 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 0.22 *zr 63 126 •6,623 158 8.830 631 1.892 2,208 3.879 8,678 6.307 7,253 7.884 9,776 13.245 13,878 14,822 15,768 18,922 23,967 29,959 34,374 37.843' 44.150 99.654 110.378 ,. 662.256 64.018.080 2.523 22 Hytl* CnaHl* (mAn) 7.00E-03 1.00E-02 2:OOE-02 3.00E-03 5.00E-04 5.00E-03 3.00E-02 9.00E-04 4.00E-03 1.00E-03 1.00E-03 6.00E-04 3.00E-02 7.00E-04 O.OOE-03 2.00E-03 1.00E-03 1.00E-03 S.OOE-03 2.00E-03 4.00E-03 6.00E-03 3.00E-03 2.00E-03 7.00E-04 4.00E-03 3.00E-03 9.00E-04 1.10E-02 Darcyv 0.4 1 132 0.5 4 3 57 2 16 6 6 4 237 7 79 28 15 16 95 48 120 208 114 88 70 442 1.987 57,616 28 0.09 Aq. Thick. 4 15 21 5 46 8 32 23 8 6 61 40 3 15 12 122 61 24 8 23 19 26 9 19 7 21 8 76 6 14 Calcutated Miring Zone Depth TO Source Area (acres) 0.5 9 9 5 9 6 7 5 9 5 6 6 7 5 6 5 5 8 5 5 5 5 5 5 .5 5 5 5 5 5 18 10 25 33 21 26 27 27 22 35 24 26 28 30 21 26 22 23 24 24 22 22 22 21 22 22 22 21 21 21 23 35 30 41 51 37 42 47 44 38 55 41 42 48 51 37 45 38 •40 42 41 38 38 38 37 38 38 38 37 37 37 39 51 100 71 82 68 72 84 75 70 89 73 73 88 89 68 78 6» 72 76 75 69 70 68 68 68 69 69 68 67 67 71 81 600 169 180 167 170 195 ' 173 170 188 172 171 201 199 186 180 169 177 184 179 168 171 168 167 168 168 168 168 165 165 170 179 Dilution Factor (DF) Source Area (acres) 0.5 1 3 102 1 S 4 30 3 10 8 5 4 75 5 41 18 9 10 48 25 61 103 SB 45 36 218 976 28.244 18 1 10 1 2 102 1 5 2 30 2 4 2 5 ' 4 18 3 23 16 . 9 10 18 25 S3 103 25 39 . 12 218 260 28.244 5 1 30 1 1 59 1 5 1 25 2 3 1 5 3 11 2 14 16 9 6 11 16 31 73 15 23 7 126 162 28.244 3 1 100 1 1 33 1 3 1 14 1 2 1 4 2 6 2 8 16 8 4. 6 9 17 40 9 13 5 70 89 28.244 2 1 1 1 14 1 2 1 6 1 1 1 2 2 3 1 4 11 4 2 3 4 8 17 4 6 2 29 37 13.045 2 1 en Dilution Factors (DFs) for 208 Sites In the Hydrogcologic Database (HGDB) - National Average I 8ourcaLangth(m) Source ATM (aeras) • • 0.5 45 10 201 30 349 100 636 600 1,559 Hydreflaotegte Setting 9.12 9.13 9.14 9.14 9.14 9.14 9.15 9.15 9.15 9.15 9.15 9.15 9.15 9.15 9.15 9.15 9.7 9.9 9.9 9.9 ' 10.2 10.2 10.2 10.2 10.2 10.2 10.2 10.2 10.2 10.2 tttM,m»tmm tiitnumtvii (mfr) 0.22 0.30 .0.22 0.22 0.22 0.22 0.30 0.30 0.30 0.30 0.30 0.30 • 0.30 0.30 0.30 0.30 0.06 0.22 0.22 0.22 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 AvWQft K (rrVy) 158 318 126 126 631 4,100 631 2.206 8,046 11,036 19.237 19,237 27.782 27,782 33.113 60.684 126 284 318 8.830 . 25- 32 126 126 158 284 318 316 631 2.208 Hyd. Orad. (mfrn) 1.20E-02 6.006-03 5.006-02 2.006-02 1.506-01 1.006-02 1.006-02 2.006-02 3.006*03 7JJ06-02 6.006-03 1.306-02 2.006-03 2.006-03 4.006-04 3.006-03 3.006-02 1.006-02 8.106-01 4.006-03 9.806-03 1.706-02 3.006-03 £806-02 6.006-04 1.006-02 4.00E-03 1.006-02 8.006-03 1.006-05 Darcyv (nVy) 2 2 6 3 95 41 6 44 IS 828 154 250 56 56 13 163 4 - 3 161 35 0.2 0.5 0.4 3 0.1 3 1 3 3 0.02 AQ. Thick, (m) 3 5 5 8 2 6 11 . 4 . 12 3 12 11 ' 24 24 30 30 107 9 6 18 8 7 4 12 3 8 6 11 1 8 Calculated Mixing Zone Depth (d) Source Area (acres) 0.5 7 8 6 8 5 5 7 5 6 5 5 5 5 5 6 8 6 8 8 8 9 11 9 8 8 6 10 . 6 6 12 10 24 26 25 28 22 22 28 22 25 21 22 22 22 22 26 22 25 29 22 22 26 28 25 31 24 28 27 30 22 29 30 40 42 41 44 38 39 45 39 42 37 38 37 39 39 44 38 44 46 37 39 42 44 41 .48 40 44 43 47 36 48 100 70 72 72 75 68 TO 77 70 75 68 69 68 71 71 79. 66 79 76 68 71 72 74 71 79 70 75 73 78 68 78 600 168 170 170 173 167 ; 1W 176 168 176 166 168 167 172 172 188 167 192 174 167 172 170 172 169 177 168 173 171 176 166 173 Dilution Factor (OF) Source Area (acres) 0.5 2 2 4 3 22 22 4 13 7 188 55 69 . 21 21 7 65 •7 3 61 19 1 1 1 3 1 3 2 3 1 1 10 1. 1 . 2 1 6 7 . 2 4 4 42 32 . « 21 21 7 65 7 & 24 16 1 1 1 2 1 1 1 2 1 1 30 1 1 1 1 4 4 2 3 3 25 19 26 14 14 6 53 7 1 14 10 •_ 100 1 1 1 1- 2 3 1 2 2 14 11 15 8 8. 3 30 7 1 8 6 - 6 : ; .. • i ii 7 4 4 2 1! 4 1 . 4 3 1 1 1 1 1 1 1 1 1 1 oo CO H- cn Dilution Factors (DFs) for 208 Sites In the Hydrogeologlc Database (HGDB) - National Aterage I Source length (m) 0.5 45 •owe* AIM (act**) 10 201 90 I 100 349 1 636 600 1.559 Hydregeotegte Sotting 10.2 10.2 10.2 10.2 10.2 10.2 10.5 10.5 10.5 11.3 11.3 11.3 11.4 11.4 11.4 11.4 11.4 11.4 11.4 11.4 ' 11.4 11.4 11.4 11.4 11.4 12.4 13.4 13.4 Infiltration (nVW 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30 . 0.06 O.OS Average K ftflM 2.206 3,469 4.415 4,415 19.552 807.066 315 631 4,415 631 7,569 12.614 32 264 315 316 946 1,261 1,281 1,577 2,523 3.154 8,186 13.676 176.602 309,053 5,361 7.684 Hyo* Grao* 1.00E-02 2.00E-03 5.00E-03 1.40E-02 '3.00E-04 2.00E-03 2.00E-03 1.00E-03 2.00E-03 1.00E-02 6.00E-03 5.00E-03 5.00E43 3.00E-03 9.00E-02 1.00E-03 2.00E-04 2.00E-03 1.706-02 2.30E-02 2.00E-03 1.50E-01 3.30E-03 2.00E-03 1.90E-02 S.OOE-04 1.00E-03 Dareyv (HW) 22 7 22 62 6 1,214 0.6 0.6 9 6 45 63 0.2 0.9 16 0.3 0.2 3 21 36 8 473 27 28 .3,355 155 5 158 Aq. Thick. H 8 3 55 9 21 15 3 0 20 6 46 6 15 30 2 24 2 11 3 5 2 6 6 61 4 43 6 3 Calculated Mining Zone Depth (d) Source Arm (acres) 0.5 5 6 5 5 7 5 8 5 6 7 5 5 20 17 5 25 6 9 8 5 6 5 5 5 5 5 5 . 5 10 24 24 24 22 30 21 24 22 27 26 23 22 37 49 23 46 23 31 23 23 23 21 23 23 21 22 24 21 30 41 40 42 39 49 37 40 37 46 43 39 38 52 67 38 61 39 47 39 39 39 37 40 41 37 38 40 37 100 73 70 75 70 64 67 70 68 81 73 71 70 83 9* 69 92 69 78 70 70 69 68 72 74 67 69 72 68 600 172 168 183 170 186 165 168 165 183 171 174 189 180 195 167 189 167 178 168 169 167 166 171 180 165 168 171 166 Mutton Factor (OF) Source Area (acres) 0.5 10 3 10 28 4 424 1 1. 5 4 18 22 1 2 3 2 1 3 6 13 2 168 11 12 1.045 56 9 141 10 4 1 10 10 3 303 1 1 4 2 18 6 • • 2 4 1 48 4 12 235 56 3 32 30 3 ' 1 10 6 2 175 1 1 3 1 18 4 1 1 . 1 1 1 1 2 3 1 26 3 12 138 56 2 19 100 2 1 7 4 - 2 96 1 1 2 1 12 2 1 1 1 1 1 1 1 2 1 16 2 10 75 35 2 11 600 1 1 4 2 1 40 1 1 1 1 5 2 . 7 1 5 31 i* 1 5 c .•*"•.v CO en rjj lor MUUB SI16S Region Setting Reference Number Western Mountain Ranges Mountain Slopes Facing East 1.1 Mountain Ranks Facing East 1.3 Mountain Ranks Facing West 1.4 Wide A*ivial Valleys Facing East 1.6 Wide Alluvial Valleys Facing West 1.7 Alluvial Mountain Valteys Facing West 1.8 . Alluvial Mountain Valleys Facing East 1.9 Coastal Beaches . 1.11' VUuwa/Basins Mountain Slopes 2.1 Altemafing Sedimentary Rocks 2.3 River Alluvium With Overbank Deposits 2.4 River Alluvium Without Overbank Deposits 2-5 Coastal Lowlands 2.6 Alluvial Fans . 2.9 Alluvial Basins with Internal Drainage 2.13 PlayaLakes 2.11 Continental Deposits 2.12 Columbia Lava Plateau Lava Flows: Hydraulically Connected 3.3 Alluvial Fans 3.5 fliver Alluvium 3.7 Colorado Plateau and Wyoming Basin Resistant Ridges 4.1 Coneolidated Sedimentary Rocks -42. AMuviua and Dune Sand 4.3 River Afluvium 4.4 High Plains River Alluvium with Overbank Deposits 52 River Alluvium without Overbank Deposits 5.3 PlayaLakes 5.7 Ogalalla 5.8 Non-Glaciated Cental Region Triassic Basins 6<2 Mountain Slopes 6.3 Mountain Flanks 6.4 Alternating Beds of Sandstone, Limestone, or Shale Under Thin Soil 6.5 Alternating Beds of Sandstone. Limestone, or Shate Under Deep RegoHth 6.6 Alluvial Mountain Valleys 6,8 Braided River Deposits 6.9 River Alluvium with Overbank Deposits 6.14 River Alluvium without Overbank Deposits 6.11 F-ll TUT 008 1506 Hydrogeologic Settings for HGDB Sites Region Setting Reference Number Unconsolidated and Semi-Consolidated Auifers 6.12 Solution Limestone 6.13 Glaciated Central Region Till Over Solution Limestone 7.1 Outwash Over Solution Limestone 7.2 . Till Over Bedded Sedimentary Rock 7.3 Thin Till Over Bedded Sedimentary Rock 7.4' Outwash Over Bedded Sedimentary Rock 7.5 Till Over Sandstone 7.6 Till Over Shale 7.7 Glaciated Lake Deposits 7.8 Outwash 7.9 Till Over Outwash . 7.18 Moraine 7.11 Buried Valley • 7.12 River Alluvium with Overbank Deposits 7.13 River Alluvium without Overbank Deposits 7.14 Beaches, Beach Ridges, and Sand Dunes 7.15 Swamp/Marsh 7.16 Till 7.17 Piedmont Blue Ridge Region Thick Regolith 8.1 River Alluvium 8.6 Northeast and Superior Uplands Glacial Till. Over Crystalline Bedrock 9.1 Glacial Lakes/Glacial Marine Deposits 82 Bedrock Uplands 9.4 Swamp/Marsh 9.5 Mountain Flanks 9.7 Glacial Till Over Outwash 9.9 Outwash 9.15 Alluvial Mountain Valleys 9.11 River Alluvium with Overbank Deposits 9.12 River Alluvium without Overbank Deposits 9.13 . Till . 9.14 ASanSc and Gulf Coast Confined Regional Aquifers 10.1 . Unconsolidated and Semi-Consolidated ;' Shallow Surfacial Aquifers 10-2 . River Alluvium with Overbank Deposits 10.3 River Alluvium without Overbank Deposits 10.4 Swamp 105 Southeast Coastal Plain • Solution Limestone and Shallow Surfacial ' ' Aquifers 11-1 F-12 . ' TUT COS 1507 Hydrogeoiogic Settings for HGDB Sites Region Setting Reference Number Swamp 11.2 Beaches and Bars 11.3 Coastal Deposits 11.4 HawBi Volcanic Uplands 12.1 Coastal Beaches 12.4 . Atata ' Coastal Lowland Deposits 13.2 Glacial and Glado-Iacustrine Deposits of the Interior Uplands 13.4 F-13 TUT OO8 1508 APPENDIX G Background Discussion for Soil-Plant-Human Exposure Pathway TUT 008 APPENDIX G Background Discussion for Soil-Plant-Human Exposure Pathway Introduction The U.S. Environmental1 Protection Agency (EPA) has identified the consumption of garden fruits and vegetables as a likely exposure pathway to contaminants in residential soils. To address this pathway within the guidance, the Office of Emergency and Remedial Response (OERR) evaluated methods to calculate soil screening levels (SSLs) tor the soil-plant-human exposure pathway. In particular, OERR evaluated algorithms and approaches proposed by other EPA offices or identified in the open literature. Key sources of information included me Technical Support Document for Land Application of Sewage Sludge (U.S. EPA, 1992), Estimating Exposure to Dioxin-like Compounds (U.S. EPA, 1994), Plant Contamination (Trapp and McFarlane, 1995), Current Studies on Human Exposure to Chemicals with Emphasis on the Plant Route (Paterson and Mackay, 1991), Uptake of Organic Contaminants by Plants (McFarlane, 1991), and Air-to-Leaf Transfer of Organic Vapors to Plants (Bacci and Calamah, 1991). Although empirical data on plant uptake from soil (either through root or leaf transfer) are limited, a comprehensive collection of available empirical data on plant uptake is presented in the Technical Support Document for the Land Application of Sewage Sludge (U.S. EPA, 1992), hereafter referred to as the "Sludge Rule." The Sludge Rule presents uptake-response slopes, or bioconcentration factors, for a number of heavy metals found in sewage sludge, including six metals addressed in the Soil Screening Guidance (i.e., arsenic, cadmium, mercury, nickel, selenium, and zinc). These empirical bioconcentration factors were used in the development of the generic plant SSLs presented in mis appendix. The Sludge Rule does not present uptake-response slopes for organic chemicals because of a lack of empirical data. Therefore, generic plant SSLs for organic contaminants are not presented in this appendix. Currently, EPA is evaluating mathematical constructs to estimate plant uptake of organic chemicals for several initiatives (e.g., Hazardous Waste Identification Rule, Office of Solid Waste; Indirect Exposure to Combustion Emissions, Office of Research and Development). In addition, new mathematical models are becoming available mat use a fugachy-based approach to estimate plant uptake of organic compounds (e.g., PLANTX, Trapp and McFarlane, 1995). Once these methods are reviewed and finalized, OERR may be able to address the soil-plant-human exposure pathway for organic contaminants. The methods and data used to calculate the generic plant SSLs for arsenic, cadmium, mercury, nickel, selenium, and zinc are presented below. For comparative purposes, data on the potential phytotoxicity of metals have also been included. In addition, the site-specific factors mat influence the bioavailability and uptake of metals by plants are discussed. The potentially significant effect of these site-specific factors on plant uptake underscores the need for she-specific assessments where the soil-plant-human pathway may be of concern. G.1 SSL Calculations from Empirical Data For uptake of chemicals into edible plants, EPA recommends a simple equation to determine SSLs for the soil-plant-human exposure pathway. The equation is appropriate for both belowground and G-l TUT 008 1.510 aboveground vegetation, provided that the appropriate bioconcentration factor (Br) is used (see Section G.4). The screening level equation for the soil-plant-human pathway is given by: SSL equation for the Soil-Plant-Human Pathway Screening Level (nig /kg) (G-1) Parameter/Definition (units) Default CpUnl/acceptable plant concentration (mg/kg DW) see Section G.2 . Br/plant-soi] bioconcentration factor (mg contaminant/kg chemical- and plant-specific plant tissue DW)(mg contaminant/kg soil)-' (see Section G.2) It is important to note that the plant concentration is in dry weight (DW) instead of fresh weight (FW). Consequently, the consumption rates for plants must also be given in dry weight. For convenience, Table G-1 presents conversion factors with which to convert fresh weight to dry weight for a variety of garden fruits and vegetables. For example, because the conversion factor for lettuce is 0.052, 10 kg of lettuce fresh weight is equivalent to 0.52 kg of lettuce dry weight. Several inputs to Equation G-1 are either derived from other equations or identified from empirical studies in the literature. Specifically, the derivation and data sources for Cp|Ut and Br are discussed below. G.2 Acceptable Concentration in Plant Tissue ( The .acceptable contaminant concentration in plant tissues (Cpunt) in mg/kg DW for fruits and vegetables is backcalculated using the following equation: . Acceptable Plant Concentration for Fruits and Vegetables ' I x B W (G'2) F x C R Parameter/Definition (units) . Default I/acceptable daily intake of contaminant (mg/kg-d) see Section G.3 BW/body weight (kg) . 70 F/fraction of fruits and vegetables consumed that are 0.4 (see Section G.4) contaminated (unitless) CR/consumption rate for fruits and vegetables 0.0197 (aboveground) (kg-plant DW-d) 0.0024 (belowground) (see Section G.4) G-2 008 1511 Table G-1. Fresh-to-Dry Conversion Factors for Fruits and Aboveground Vegetables Vegetables Asparagus Snap beans Cucumber Eggplant Sweet pepper Squash Tomato Broccoli Brussels sprouts Cabbage Cauliflower Celery Escarde Green onions Lettuce Spinach green Average for vegetables 0.070 0.111 0.039 0.073 0.074 0.082 0.059. 0.101 0.151 0.076 0.083 0.063 0.134 0.124 0.052 0.073 0.085 Fruits Apple Bushberry. Cherry Grape Peach Pear Strawberry Plum/prune - • Average for fruits* 0.159 0.151 0.170 0.181 0.131 0.173 0.101 0.540 • 0.15 • PlunVprime was omitted from the average as an outlier. Source: Baesetal. (1984). G.3 Acceptable Daily Intake (I) of Contaminants For carcinogens, the acceptable daily intake (I) in mg/kg-day is calculated at the target risk level, using default assumptions for exposure duration, exposure frequency, and averaging time. At the target risk level,, the acceptable daily intake of carcinogens may be.calculated as follows: Acceptable daily intake for carcinogens (G-3) » TR x AT x 36Sd/yr ED x EF » CSF^., Parameter/Definition (units) Default TR/target risk level (unitless) 1(H AT/averaging time (years) 70 ED/exposure duration (years) 30 EF/exposure frequency (d/yr) 350 cancer slope factor (mg/kg-d)-1 chemical-Specific (see Part 2, Table 1) G-3 TUT i512 For noncarcinogens, the acceptable daily intake (I) in mg/kg-day is calculated at a hazard quotient of 1 using the following equation: Acceptable daily intake (I) for noncarcinogens HQ xRfD x AT x 36Sd/yr ' (G-4) Parameter/Definition (units) HQ/target hazard quotient (unhless) AT/averaging time (years) ED/exposure duration (years) EF/exposurc frequency (d/yr) RfD/oral reference dose (mg/kg-d) Default 1 30 30 . 350 chemical-specific (see Pan 2, Table 1) G.4 Contaminated Fraction (F) and Consumption Rate (CR) Default values for the fraction of vegetables assumed to be contaminated (F) are recommended in the Exposure Factors Handbook (U.S. EPA, 1990). For home gardeners, a high-end dietary fraction of 0.40 is assumed for the ingestion of contaminated fruits and vegetables grown onsite'. The default values for total fruit and vegetable consumption rates (CR) cited in the Exposure Factors Handbook are 0.140 and 0.2 kg/d fresh weight, respectively. Assuming mat the homegrown fraction is roughly 0.25 to 0.40, EPA estimated fresh weight consumption rates of: (1) 0.088 kg/d of aboveground unprotected fruits, (2) 0.076 kg/d of aboveground unprotected vegetables, and (3) 0.028 kg/d of unprotected belowground vegetables (U.S. EPA, 1994). The consumption rates for fruits and vegetables are converted to dry weight based on the average frcsh-to-dry conversion of 0.15 for fruits and 0.085 for vegetables presented in Table G-l. For unprotected belowground vegetables, the consumption rate (CR) is calculated by multiplying the fresh weight consumption rate (0.028 kg FW/d) by the average conversion factor of 0.085 resulting in a CR of 0.0024 kg DW/d. Using this same method, dry weight consumption rates of 0.0132 and 0.0065 kg DW/d were calculated for unprotected aboveground fruits and vegetables, respectively. Consequently, the overall consumption rate (CR) for aboveground, unprotected fruits and vegetables is 0.0197 kg DW/d. The distinction between protected and unprotected produce reflects evidence that, for protected plants such as cantaloupe and citrus, mere is very little translocatton of contaminants to the edible parts of the plant. EPA recognizes mat, while these assumptions for contaminated fraction and consumption rates are reasonable for general assessment purposes, there is likely to be wide variability on the types of produce grown at home, the percentage that is •unprotected, and other exposure-related characteristics (U.S. EPA, 1994). G.5 Soil-to-Plant Bioconcentration Factors (Br) For metals, soil-to-plant bioconcentration factors (Br) for both aboveground and belowground plants must be identified from empirical studies because the relationship between soil concentration and plant concentration has not been described adequately to' provide a mathematical construct for G4 TUT COS 1513 modeling. Table G-2 provides empirical plant uptake values for six metals identified in the Technical Support Document for Land Application of Sewage Sludge (U.S. EPA, 1992). Because of the variability in site-specific assessments, bioconccntration factors mat are appropriate for the type of produce considered in a particular risk assessment should be selected- For general screening purposes, the geometric mean Br values for leafy vegetables and root vegetables are typically selected to represent aboveground and belowground plants, respectively. These values may be used to calculate SSLs for six metals for the soil-plant-human exposure pathway. G.6 Example Calculation of Soil-Plant-Human SSL: Cadmium V, . To demonstrate how the methods described in this appendix may be used to calculate an SSL tor the soil-plant-human pathway, a sample calculation is provide below for cadmium.' Cadmium is considered a noncarcinogen via oral exposure and, therefore, the acceptable daily intake (I) is calculated using Equation G-4. Using the RID for cadmium ingested in food of 1.0 x 10*3 mg/kg (the RfD is 5.0 x 10-4 in water), Equation G-4 may be solved for acceptable daily intake (I) of cadmium from a dietary source: HQ xRfD x AT x365d/yr Jcl) x EF 1 x 1.0xlO'3mg/kg-d * 30yr x 365d/yr SOyrs x 350d/yr 1= 1.0xlO-3mg/kg-d x The acceptable daily intake (I) is used in Equation G-2 to estimate the acceptable contaminant concentration in plant tissue (Cpum). However, Equation G-2 is designed to solve for the acceptable plant concentration (Cpunt) in either aboveground fruits and vegetables or belowground vegetables. Consequently, Equations G-l and G-2 must be combined to calculate the screening level for the ingestion of both aboveground and belowground produce. These equations are combined by summing the product of the category-specific produce intake and bioconcentxation factors. Since the default contaminated fraction applies to both categories of produce, Equations G-l and G-2, are combined to solve for the soil screening level: (G-5) Screening Uvd (mg/kg) G-5 TUT 008 1514 Tab!* G-2. Summary Tablt of Empirical Bioconcantration Factors for M*tals . • (in mg contaminant par kg plant DW / mg contaminant par kg soil) Bioconcantration factors (Br) Study observations pH Rang* Min. Max G*om*tric M*an Br Ars*nfc grains and cereals potatoes leafy vegetables legumes root vegetables garden imits sweet corn 1 6 7 7 7 5 3 7.5 5.5 -7.5 5.5 - 7.5 NB-7.5 NR-7.5 NR-7.5 MR 0.026 0:002 0.002 . 0.002 0.002 0.002 0.002 0.026 0.24 0.068 0.004 0.28 0.006 0.002 0.026 0.004 0.036 0.002 0.008 0.002 0.002 Cadmium grains and cereals potatoes leafy vegetables legumes root vegetables garden fruits sweet com 14 14 71 14 25 19 12 4.4 - 8.0 4.7- 8.0 4.6-8.4 5.1 - 7.7 4.6- 8.0 4.6 - 7.1 5.1-7.1 0.002 0.002 0.002 0.002 0.002 0.002 0.02 0.346 0.076 14.12 0.054 1.188 1.272 0.666 0.36 0.008 0.364 0.004 0.064 0.09 0.118 Mercury grains and cereals potatoes leafy vegetables legumes root vegetables garden tarts sweet com 1 1 9 3 6 7 default 5.3 - 7.1 5.3 - 7.1 5.3 - 7.1 5.3 - 7.1 5.3 - 7.1 5.3 - 7.1 ND 0.0854 0.002 0.002 0.002 0.002 0.002 0.002 0.0854 0.002 0.092 0.002 0.086 0.086 0.002 0.0854 0.002 0.008 0.002 0.014 0.01 0.002 Nickal grains and cereals potatoes leafy vegetables legumes root vegetables garden fruits sweet com 10 14 56 ' 11 25 14 4 6.2 - 8.0 6.4 - 8.0 5.3 - 8.0 5.9 - 7.7 5.9 - 8.0 5.9 - 7 .3 5.9 - 7.1 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.11 0.06 30 1.004 0.232 0.19 0.002 0.01 0.01 0.032 0.062 0.008 0.006 0.002 S*l*ol4Mn grains and cereals potatoes leafy vegetables legumes root vegetables garden fruits sweet com 4 2 7 4 8 8 default 5.5-7.0 5.5 - 6.8 5.5 - 7.8 5.5 • 6.8 5.5 - 7.6 5.5 - 6.8 ND 0.002 0.018 0.002 0.024 0.004 OO08 0.002 0.11 0.096 0.076 0.11 0.096 0.078 0.002 0.002 0.042 0.016 0.024 0.022 0.02 0.002 008 j.515 Table G-2. (continued) Bioconcentration factors (Br) Study observations pH Rang* Min Max . Geometric Mean Br Zinc grains and cereals •potatoes leafy vegetables legumes root vegetables *" garden fruits • sweet com 13 14 47 10 20 21 8^ 5.3 - 8.0 4.7-8,0 4.6 - 8.0 5.1-7.7 4.6-8.0 4.6-7.3 5.1 - 6.5 O.D16 . „ 0.01 0.012 0.002 0.002 0.002 0.002 0.368 0.122 4.488 0.11 0.412 0.394 0.19 0.1 0.024 0.25 0.036 0.044 0.046 0.02 NR = Not reported ND«Nodata The input parameters in Equation G-5 correspond to input parameters in Equations G-1 and G-2, with a contaminated fraction (F) of 0.4, and consumption rates (CR.g and CRbg) and bioconcentration factors (BrM and Brbg) specific to either aboveground or belowground produce. Solving Equation G-5 for cadmium using the default parameters in Equation G-2 for F, CR,g, and results in: . . . . . I x B W Screening Level Screening Level 0.4 xL(CR., x BrM) + (CRb, x Brbg) ______ 1.0xlO'3mg/kg-d x 70kg_______ 0.4 x L(0.0197 x 0.364) + (0.0024 x 0.064) kg soil /d Screening Level » 24 mg /kg soil As described above, the geometric mean Br values for leafy vegetables and root vegetables were selected to represent the bioconcentration factors (Br) for aboveground fruits and vegetables (Br,j) and belowground vegetables (Br^), respectively (see Table G-2). SSLs for the plant pathway that are calculated using the bioconcentration factors for leafy and root vegetables are considered to be generic SSLs by OERR During site-specific assessments, OERR recommends mat a weighted average bioconcentration factor be used to reflect the type of produce grown and eaten locally. ' • G.7 Generic SSLs for Selected Metals Table G-3 presents the generic SSLs for the soil-plant-human exposure pathway along with the SSLs for direct soil ingestion. In addition, this table presents plant toxicity values identified in the Toxicological Benchmarks for Screening Potential Contaminants of Concern for Effects on Terrestrial Plants: 1994 Revision (Will and Suter, 1994). The phytotoxicity values are either. (1) the G-7 • . TUT OOS estimated 90th percentile of lowest observed effects concentrations (LOECs) from a data set consisting of 10 or more values, or (2) the lowest LOEC from a data set with less than 10 values. The lexicological endpoints for the phytotoxicity were limited to growth and yield parameters because they are the most common endpoints reported in phytotoxicity studies and are ecologically significant in terms of plant populations. Table G-3. Comparison of Generic SSLs for Plant Pathway with the SSL* for Soil Ingestion and LOEC Values for Phytotoxicity (all values in ing/kg) : , . •Arsenic Cadmium Mercury Nickel Selenium Zinc Generic plant SSL Soil ingestion SSL Migration to ground water SSL* Phytotoxicity LOEC 0.4 0.4 29(1) 10 24 78 8(0.4) 3 270 23 2(0.1) 0.3 5400 1600 130(7) 30 2400 390 5(0.3) 1 10000 23000 12000(620) 50 • Values based on DAF of 20 (OAF of 1). The comparison of the generic SSLs for the plant pathway with SSLs for soil ingestion and migration to ground water suggests mat this pathway may be of concern at sites .contaminated with arsenic or cadmium. For mercury, nickel, and selenium, the generic plant SSLs are well above the SSLs based on soil ingestion and migration to ground water. Thus, although SSLs based on these other pathways are likely to be protective of the soil-plant-human pathway, other data suggest that phytotoxicity is likely to be the factor limiting exposure through plant uptake for these metals. Phvtofoxicitv • The data in Table G-3 suggest that, for cadmium, mercury, nickel, and selenium, toxicity to plants will be observed at levels well below those estimated to elicit adverse effects in humans. The phytotoxicity of arsenic, nickel, and zinc have been well documented. However, despite the low phytotoxicity value for selenium, some authors have demonstrated that selenium can accumulate in certain plants at high levels (Bitton et al., 1980). Moreover, many phytotoxicity values are based on a reduction in yield that may result in higher levels in the surviving produce. Thus, with the exception of zinc, phytotoxicity should not be used to rule out this exposure pathway unless empirical data are available that are relevant to the site conditions (e.g., similar pH, organic matter) and the type of crops likely to be grown. Soil Characteristics - Because the majority of the plant uptake data for metals were generated in sludge application studies, the empirical bioconcentration factors listed in Table G-2 may not be appropriate for use at all sites. For example, the adsorption "power" of sludge in the presence of phosphates, manganese, hydrous oxides of iron, and Ca+2 may reduce the amount of metal that is bioavailable to plants. In addition, soil pH strongly influences the ability of plants to absorb metals from soil. Several studies document that, as pH decreases, the bioavailability of many metals increases. In fact, agricultural practices maintain a soil pH of 5.5 or greater to protect against aluminum and manganese phytotoxicity. However, 40 percent of the data evaluated for the Sludge Rule were from studies in which the pH was less than 6, and, as a result, bioconcentration factors may be artificially skewed. Chemical Characteristics - Another factor mat heavily influences plant uptake of metals is the chemical form of the metal. Researchers have observed that plant uptake rates of metal salts in sludge tend to be higher than plant uptake rates in studies on elemental metals. Metal salts do not G-8 TUT 008 adsorb to sludge the same way as "metals in nonsalt forms" and, consequently, they are more bioavailable to plants. Type of produce - The bioconcentration potential of metals varies with plant type. As shown in Table G-2, the range of bioconcentration factors covers an order of magnitude for most metals across the seven categories of produce. Certain types of plants are resistant to some metals while these same metals may be highly toxic to other plant species. Depending on the type of crops grown, the generic soil-plant-human SSLs may not reflect the most appropriate measures of bioconcentration. • Dietary habits - The dietary habits of the home gardener may result in an increase or decrease in exposure. The default values for consumption rate (CR) and contaminated fraction (F) represent reasonably conservative estimates for these exposure parameters. However, individual consumers may ingest significantly different quantities of produce and, depending on their fruit/vegetable preferences, may rely on crops that are efficient accumulators of metals. G.8 SSL Calculations for Organics Lacking Empirical Data The lack of plant bioconcentration data on organics presented in the Technical Support Document for Land Application of Sewage Sludge (U.S. EPA, 1992) has been discussed in several other sources. For example; the status of empirical data on plant uptake and accumulation of organics was recently evaluated for a database on uptake/accumulati"" translocation, adhesion, and biotransformation of chemicals in plants (Nellessen and Fletcher, 1993). This database, referred to as UTAB, is one of the most comprehensive data sources available on chemical processes in plants and contains over 42,000 records taken from more than 2,100 published papers. The authors found that, with the exception of pesticides, uptake-response data for organic.chemicals are available for roughly 25 percent of the chemicals monitored by EPA. Given the comprehensive nature of the UTAB database, modeling may be the only alternative to evaluating the soil-plant-human pathway in the near future for many organic chemicals. Recently, several authors have developed models to predict the uptake and accumulation of organic chemicals in plants (e.g., Matthies and Behrendt, 1994; McKone, 1994; Trapp et al., 1994). One of the most promising models for use as a risk assessment tool is PLANTX, a peer-reviewed partitioning model that describes the dynamic uptake from soil, or solution, and the metabolism and accumulation of xenobiotic chemicals in roots, stems, leaves, and fruits (Trapp et al., 1994). Unlike a number of other models used to estimate plant uptake, PLANTX is not based on regression equations that correlate log Kew with plant bioconcentration; it is a mechanistic model mat accounts for major plant processes and requires only a few well-known input data. Moreover, it was designed as a risk assessment tool and has been validated for the herbicide bromicil and several nitrobenzenes. A follow-on model (PLANTE) has recently been made available that also incorporates plant uptake during transpiration (i.e., accumulation directly from the air). The results on bromocil, nitrobenzene, etc., as well as ongoing validation studies suggest mat the PLANT models may be a scientifically defensible alternative to the uptake-response slopes generated by log K«* regressions.. G.9 Conclusions and Recommendations The comparison of generic plant SSLs with generic SSLs for soil Digestion and migration to ground water indicate that the soil-plant-human exposure pathway may be of concern for two of the six metals evaluated (arsenic and cadmium). For mercury, nickel, and selenium, SSLs based on the other pathways are likely to be adequately protective of the soil-plant-human exposure pathway." In addition, data presented on the phytotoxicity of these metals and zinc suggest that toxic effects in plants are likely to be observed below levels mat would be harmful to humans. Although mis pathway « • . G-9 ooa may not be of concern from a human health standpoint, these data suggest that metals could be of particular concern for ecological receptors. Currently, EPA is developing methods to evaluate the uptake of organics into plants. In addition to the efforts of the Office of Solid Waste and the Office of Research and Development mentioned in the Introduction, OERR has jointly funded research on plant uptake of organics with the State of California. These studies support ongoing revisions to the indirect, multimedia exposure model. CalTOX. Until these efforts are reviewed and finalized, OERR will continue to address the potential for plant uptake of organics on a case-by-case basis. References Bacci, Eros, and David Calamari. 1991. Air-to-leaf transfer of organic vapors to plants. In: Municipal Waste Incineration Risk Assessment. C.C. Travis editor, Plenum Press, New York. Baes, C.F., R.D. Sharp, A.L. Sjoreen, and R_W. Shor. 1984. Review and Analysis of Parameters and Assessing Transport of Environmentally Released Radionuclides Through Agriculture. Oak . Ridge National Laboratory, Oak Ridge, TN. Bitton, G., B.L. Damron, G.T. Edds, and J.M. Davidson. 1980. Sludge - Health Risks of Land Application. Ann Arbor Science Publishers, Inc., Ann Arbor, MI. Matthies, M., and H. Behrendt. 1994. Dynamics of leaching, uptake, and translocation: The Simulation Model Network Atmosphere-plant-soil (SNAPS). In: Plant Contamination: Modeling and Simulation of Organic Chemical Processes. Stefan Trapp and J. Craig Me Farlane eds, Lewis Publishers, Michigan. McKone, T. 1994. Uncertainty and variability in human exposures to soil contaminants through homegrown food: a Monte carlo assessment. Risk Analysis 14 (4): 449-463. McFarlane, Craig. 1991. Uptake of organic contaminants by plants. In: Municipal Waste Incineration Risk Assessment. C.C. Travis editor, Plenum Press, New York. Nellessen, J.E., and J.S. Fletcher. 1993. Assessment of published literature pertaining to the uptake/accumulation, translocation, adhesion and biotransformation of organic chemicals by vascular plants. Environmental Toxicology and Chemistry 12: 2045-2052. Paterson, Sally and Donald Mackay. 1991. Current studies on human exposure to chemicals with emphasis on the plant route. In: Municipal. Waste Incineration Risk Assessment. C.C. Travis editor, Plenum Press, New York. . Trapp, S., and J.C. McFarlane. 1995. Plant Contamination. Modeling and Simulation of Organic Chemical Processes. Lewis Publishers, Boca Raton, FL. Trapp, S., and J.C. McFarlane, and M. Matthies. 1994. Model for uptake of xenobiotic into plants: Validation with bromocil experiments. Environmental Toxicology and Chemistry 13 (3): 413-422. Travis, C.C., and A.D. Arms. 1988. Bioconcentration of organics in beef, milk and vegetation. Environmental Science and Technology 22(3):271-274. G-10 TUT COS 1519 U.S. EPA (Environmental Protection Agency). 1990. Exposure Factors Handbook. Office of Health •and Environmental Assessment, Exposure Assessment Group, Washington, DC. March. U.S. EPA (Environmental Protection Agency). 1992. Technical Support Document for Land Application of Sewage Sludge, Volume 1 and II. EPA 822/R-93-001a. Office of Water. Washington, DC. U.S. EPA (Environmental Protection Agency). 1994. Estimating Exposure to Dioxin-like Compounds. Volumes I-III: Site-specific Assessment Procedures EPA/600/6-88/005C. Office of Research and Development, Washington, DC. June. Will, M.E. and G.W. Suter II. 1994. Toxicological Benchmarks for Screening Potential Contaminants of Concern for Effects on Terrestrial Plants: 1994 Revision. ES/ER/IM-85/R1. Prepared for the U.S. Department of Energy by the Environmental Sciences Division of the Oak Ridge National Laboratory. G-ll 008 1520 APPENDIX H Evaluation of the Effect on the Draft SSLs of the Johnson and Ettinper Model (EQ, 1994a) TUT 1521 ENVIRONMENTAL QUALITY MANAGEMENT, INC. MEMORANDUM TO: Janine Dinan SUBJECT: Evaluation of the Effect on the Draft SSLs of the Johnson and Ettinger (1991) Model for the Intrusion of Contaminant Vapors Into Buildings , FILE: 5099-3 DATE: October 7,1994 FROM: Craig S. Mann ce: Under U.S. Environmental Protection Agency .(ERA) Contract No. 68-D3-0035, Task order No. 0-25, Environmental Quality Management, Inc. (EQ) was directed to evaluate the sffect on the draft soil screening levels (SSLs) of employing the Johnson and Ettinger (1991) model for estimating the intrusion rate of contaminant vapors from soil into buildings. This memorandum summarizes the evaluation. Model Review: Johnson and Ettinger (1991) is a closed-form analytical solution for both convective and diffusive transport of vapor-phase contaminants fully incorporated in soil into enclosed structures. The nondimensionalized mass balance is written as: r L (1) where * » Nondimensional variables e: * Volume fraction of phase i, unlttess Q «= Concentration of contaminant in phase i, g/cnf t * Time, s Lp « Convection path length, cm H-i TUT OOS 1522 LO = Diffusion path length, cm P = Pressure in vapor-phase, g/cm-s2 v = Del operator, 1/cm Q, = Contaminant concentration in vapor phase, g/cm3 D*" « Effective diffusion coefficient, cnf/s fj - Vaporviscosity, g/cm-s K, = Soil permeability to vapor flow, cm2 A Pr = Reference indoor-outdoor pressure differential, g/cm-s2 R. * Formation rate of contaminant in phase i, g/cnf-s and, Q* -Q/Q P* = P/AP, t* = t ( K A R.* -R. where Q, Lp and LD are .characteristic concentration, convection pathway length, and diffusion pathway length, chosen to give the dependent concentration variable and derivatives of Q* and P* magnitudes of order unify. The mass balance solution includes the following assumptions: 1. The soil column is isqtropic within any horizontal plane. 2. The effective diffusion coefficient is constant within any horizontal plane. 3. Concentration at the soil-air interface is zero (i.e., boundary layer resistance is zero). H-2 TUT OO8 1523 4. No loss of contaminant occurs across the lower boundary (i.e., no leaching). 5. Source degradation and transformation are not considered. 6. Convective vapor flow near the building foundation is uniform. 7. Contaminant vapors enter the building primarily through openings in the . walls and foundation at or below grade. 8. Convective velocities decrease with increasing contaminant source-building distance. 9. All contaminant vapors directly below a basement will enter the basement, unless the floor and walls are perfect vapor barriers. 10. the building contains no oi<er contaminant sources or sinks, and the air volume is well mixed. Therefore, .£ (2) where Q^^, C^^, and E represent the volumetric flow rate or ventilation rate of the building (crrr/s), contaminant concentration within the building (g/crn3), and rate of contaminant entry (g/s), respectively. Also, a » C_^_JC__ (3) where C^, is the vapor-phase contaminant concentration within the soil at the source, and a represents the attenuation coefficient C^. is written as: H-3 TUT OO3 1524 where H = Henry's law constant, unrtiess C* = Soil bulk concentration, g/g pb « Soil dry bulk density, g/cnf 6W = Soil water-filled porosity, unrtiess K, « Soil-water partion coefficient, cnf/g 6. « Soil air-filled porosity, unitless. The authors derive a solution for er for both steady-state conditions (i.e., depth of contamination, z = ») and for quasi-steady-state conditions (0 < z < L). For steady- state conditions a is written as: x exp exp exp -1 where CT" « Effective diffusion coefficient, crrf/s AB «= Area of basement, cm 2 * Lp = Source-building separation, cm H-4 TUT 008 1525 Volumetric flow rate of soil gas into the building, cm3/s Building foundation thickness, cm Effective diffusion coefficient through crack, cnf/s (DP* 011 « Cf) Area of crack, cm2 . = Building ventilation rate, crrf/s. For quasi-steady-state conditions the long-term average attenuation coefficient <a> is: . where pb = Soil dry bulk density, g/cnf CR = Average contaminant level in soil, g/g \ & He = Thickness of depth over which contaminant is distributed, cm Aa = Area of basement, cm2 Building ventilation rate, cnf/s Vapor-phase soil concentration at source, g/cm3 r «= Exposure averaging period, s LT° = Source-building separation at t*0, cm and, H-5 TUT 008 1526 . exp - (7) The time required to deplete a finite source (r0) of depth A He is given as: 2* If the exposure period (r) is greater than rDl the average emission rate into the building <E> is given as a simple mass balance: t (10) and the average building concentration (C^M^) is: c^M -<£>/a^. (ID Evaluation in order to evaluate the effects of using the model on the SSLs for volatile contaminants, a case example was constructed which best estimates a reasonable high end exposure point concentration for residential land use. Where possible, values of model variables were taken directly from Johnson and Ettinger (1991). • The case example assumes that a residential dwelling with a basement is constructed within the area of homogeneous residual contamination such that the contaminant source lies directly below the basement floor at t * 0. Therefore, the H-6 OOS 1577 " diffusion and convection path lengths were set equal to the thickness of the basement slab (15 cm). Soil permeability to vapor flow from the basement floor to the bottom of contamination was set equal to 1.0 x 1Qe cm2 (1 darcy) which is representative of silty to fine sand. Soil column-building pressure differential was set equal to 1 pascal (10 g/cnv s2) as a reasonable long-term average value (Johnson and Ettinger, 1991). Values for all other soil properties were set equal to those of the Generic SSLs in the July 1994 Technical Background Document for Draft Soil Screening Level Framework (TBD). Building variables, i.e., basement area, ventilation rate, etc., were taken from Johnson and Ettinger (1891). In the analysis, the values for C,^^, (kg/rn3) were calculated for the 42 chemicals in the TBD for which human health benchmarks.are available. Please note that the values of C^e. and QuKfeg were calculated for an initial soil concentration of 1 mg/kg instead of 1 x 10*6 g/g. This was done to facilitate reverse calculation of the SSL in units of mg/kg. Therefore, these values are artificially high by a factor of 1 x 10*. The inverse of the value of 0*^ (nf/kg) was used as the indoor volatilization factor (vT^) and substituted into Equations 2-4 or 2-5 of the < &D as appropriate to calculate the resulting carcinogenic and noncartinogenic inhalation SSLs. SSLs were calculated for both steady- state conditions (infinite source depth) and quasi-steady-slate conditions (finite source depth). In each case were the exposure period exceeded the time required for source depletion {finite source depth), the volatilization factor was normalized to an average contaminant level in soil (Q) of 1 mg/kg. For quasi-steady-state conditions, the depth to '-- the bottom of contamination was set equal to 2 meters below the basement floor. * • The value of the indoor SSL for each contaminant was compared to the respective SSL calculated for outdoor exposures of the same duration using the Generic SSL calculations found in the TBD. The outdoor SSLs were computed for a 30 acre square area source of emissions. Table 1 summarizes the results of this comparison. The attachment to this memorandum gives the detailed computations for this evaluation. As can be seen from Table 1, results on a chemical-specific basis indicate a rate of change as high as three orders of magnitude between the outdoor SSL and the infinite source indoor SSLs in the case of highly volatile contaminants. For very persistent contaminants, the relative difference was considerably less, and in some cases there was no difference in SSL concentrations. . This variability is due to: 1) the variability in the human health benchmarks used to calculate the risk-based SSLs, and 2) the apparent diffusion coefficient of each compound. The apparent diffusion coefficient can be expressed as the effective diffusion coefficient through soil divided by the liquid-phase partition coefficient (Jury et al., 1983). The apparent diffusion coefficient pA) is given here so as not to be confused with the effective diffusion coefficient (D*") from Johnson and Ettinger (1991): H-7 TUT 008 1528 TABLE 1. SUMMARY OF INDOOR AND OUTDOOR INHALATION SSLs FOR VOLATILE CONTAMINANTS Indoor SSL, Infinite source Chemies! (mo/kg) AWrin Benzene Bi*<2-chtoroethyt)ether Bfomofbfm Carbon disulfide Carbon letiacnTbrtde CMordane Chlorebenzene Chloroform DOT 1,2-Dichlorobenzone . 1,4-Dichlorobenrene 1,1-Dichtefoetharte 1.2-Dichloroethane 1.1-Diehtoroethyt*ne 1.2-Dichtoropropan* 1,3-DichlofeprDpene Dieldrin Bhylbenzone Heptachlor Heptachlor epoxide ' Hexachlon>.1.34utadione HexBchlorobanzene HCH-alphe(elph»-BHC) HCH-oeta(bst»-8HC) HexecMorocyclopsntadiene Hexachloroethane Methyl bromide M*hytmchk*id« Nttrobtnane Gtyiww • ai n ri ~r • ii •*• • i i«Z»2*TvtJ tUJllOf IMUUU M Tetraehtoroetriytene Toluene Toaphene 12,4-Trichlorobsflzene 1.1.1-Thchteroethane 11^ Til ililn • srfliaiiiai TriCRtoreethytane 2.4.6-Triehlorcphenol Vinyl acetate Vinyl chloride 0.4 0.002 0.02 0.6 0.03 0.0007 61 0.7 0.001 s» 26 102 4 0.002 0.0001 0.06 0.0007 3 21 0.04 1 0.03 0.3 0.5 7* 0.06 0.6 0.0V 0.04 9 165 0.007 0.05 6 2 6 5 0.009 0.01 64 5 0.00002 Indoor Outdoor ssu ' s s u finite • infinite source source (mo/kg) (mo/kfl) 0.4 0.02 • 0.05 0.9 0.7 0.01 53 2 0.007 5' 85 944 • 299 35 0.007 0.003 0.3 0.004 4 69 0.04 1 0.05 0.6 0.6 T 0.07 0.6 0.3 0.3 25 472 . 0.02 0.3 26 2 9 69 0.02 0.09 94 14 0.002 0.5 0.5 • 0.3 43' 11 A M 54 87 02 6* 297' 235* 939 0.3 0.04 10 0.1 2 257' 0.3 1 1 1 0.9 7* ' 2 45 3 7 100 1439* 0.4 11 821* 2* 214 . 980* 1 3 190 351 0.01 • « SSL based on CM- H-8 TUT OOS 1529 DA - [(el*3 D. H «• er D>f]f(P. *, * BW «• e.H) 02) where DA = Apparent diffusion coefficient, cnf/s 6. = Air-filled soil porosity, unitiess D. « Diffusivity in air, cnf/s H «= Henry's law constant, unitiess 6W = Water-filled soil porosity, unitiess DU » Diffusivity in water, cnf/s 6, «= Total soil porosity, unitiess pb • Soil dry bulk density, g/cnf K, • Soil-water partition coefficient, cnf/g. With all nonchemical-spedfic variables held constant, Figure 1 shows the exponential relationship between the apparent diffusion coefficient and the building concentration for quasi-steady-state conditions (finite source). For nonchemical-specffic variables, a sensitivity analysis was performed for soil permeability to vapor flow OO, soil-building pressure differential (AP), depth of contamination (&H), source-building separation att « 0 (I*0), crack-to-totalarea ratio (7), and building ventilation rate Table 2 shows the results of the sensitivity analysis for the quasi-steady-state condition (finite source). As can be seen from Table 2, the effect of the building ventilation rate is linear if the value of Q^ is not included in limiting the value of the SSL Depth of contamination (AHJ has the greatest effect for contaminants with higher apparent diffusion coefficients (e.g., benzene, chloroform, vinyl chloride, etc.), in that as Ati increases, the time required for source depletion (r0) also increases. Therefore, with greater initial contaminant mass in the soil, these compounds are emitted for a longer period of time thus reducing the SSL For the more persistent contaminants, an increase in K, or AP produces the greatest results. This is to be expected as values of r0 for these contaminants exceed the exposure duration. Table 2 also indicates that an order of magnitude change in values of LT° and n produce same order of magnitude results. It must be remembered, however, that in the case of U°, the mode! assumes isotropic soil H-9 TUT 008 1530 o CO csw t-* 1.006*01 1 1.00E-01 1.00C-03 t 1.00E4S • * 1.00E-07 ———— v —— 1.00E-09 A i . . * : 4 • 1.00E-07 ,f ^ • 4 • i 1.00E-05 >*- ^ \ ' ' I A —— P 4 1 1.00E-03 W 1.00E «nt MfftMfon CiMfndml (aitft) Flgurt 1. BuNdlng concentration vertut apparent dlffutton coefflclent TABLE 2. MODEL SENSITIVITY TO NONCHEMICAL SPECIFIC VARIABLES Rrto af V*ttUi4o-Tmt CenMan SSL Ctwniea! S.04E-04 6.18&04 7.78E-04 •57&04 8.64&04 1J4E4O 1JSEXO 1J4&OJ 1JM&03 UK-03 2.12M3 2.44&03 '•SSL H-ll TUT OOS 1532 conditions from the point of bulding entry to the bottom of contamination. As V increases, a decreases until diffusion not convection limits the rate of contaminant vapor transport The effect of changes in the value of/? decrease as values of K, decrease such that for very permeable soils and convection-dominated vapor transport, the effect of crack size is relatively insignificant Conclusions . ' ' • Use of the Johnson and Ettinger (1991) model to calculate SSLs based on indoor chronic exposures can have significant impacts on the values of the SSLs for contaminants with high apparent diffusion coefficients. When comparing the infinite source indoor model to the infinite source outdoor model for these contaminants, values of the SSL differ by orders of magnitude for case example conditions. Under these conditions, diffusion is the limiting transport mechanisms for all but one contaminant for both steady-state and quasi-steady-state conditions. To effect case example conditions, the following must be true: 1. The contaminant source must be relatively dose or directly beneath the structure. 2. The soil between the structure and the source must be very permeable (K, «> 10* cnf). ." 3. The structure must be underpressurized. 4. The air within the structure must be well mixed (Le., tttie or no soil-air boundary layer resistance). 5.- The combination of diffusion coefficient through the cracks, area of the cracks, and bunding underpressurization must offer no more resistance than the soil column beneath the structure. From this evaluation, the four most important factors affecting the average long- term building concentration and thus the SSL are budding ventilation rate, source-building separation, soil permeability to vapor flow,, and source depth. H the source of contamination is relatively deep and dose to the building, and if the soB between the source and the building is very permeable, building concentrations of contaminants with relatively high apparent diffusion coefficients will increase dramatically. It should be noted, however, that son permeability, K,, te the most variable parameter at any given site, and may vary by three orders of magnitude across a typical residential lot (Johnson and Ettinger, 1991). For this reason, the overall effective diffusion coefficient should be determined by integration across each soS type. Overall H-12 TUT 008 diffusion/convection vapor transport will therefore be limited by the soil stratum offering the greatest resistance to vapor flow. References • Johnson, Paul C., and Robert A. Ettinger. 1991. Heuristic model for predicting the intrusion rate of contaminant vapors into buildings. Environ. Sci. Technoi., 25(8): 1445- 1452. Jury, W. A., W. J. Farmer, and W. F. Spencer. 1983. Behavior Assessment Model for Trace Organics in Soil, I, Model description. J. Environ. Qua!., 12:558*564. Attachment H-13 TUT COS 152 This page left blank on purpose. H-14 TUT 008 1535 ATTACHMENT DETAILED MODEL EVALUATION H-15 TUT 008 1536 This page left blank on purpose. H-16 TUT 008 1537 ) . CCMfWmONOFmOCXKMOOlJfOOm MHM^ M M 801 toW *4M Ml CbmM Ottffo M l%U^!ft^^l) Af8tfl9^^Ml dhodl^^At ______^mljfOTf^hpniilVMn^MmlvMI OltaMty lit*. o. Mk) Mwpor i^idrf ncvD n. MM** DftMton Oaflwdton fev* u r, bwcon*t*< Pt H K« «?<**»*«<*« Oitouttnun* CMM DOT * 1.34fcMorapn|M» OKMrtn —I O 03 Tehttm 1.2.471 1.1. TilcHunmhytlo* 30940-2 19 18 1.9 19 79-19* 19 98-2M 1.9 97-749 1.9 10840.7 19 97484 19 90-294 1.9 9940-1 1.9 10848.7 1.9 79444 19 10749-2 19 79494 1.9 79474 19 942-794 19 0047-1 19 100414 1.9 79444 1.9 1024474 U 97494 U 11*741 U 319444 1.9 31949-7 19 •77474 1:9 9742-1 19 ' 74434 19 79494 1.9 99494 18 10042-8 19 79448 IS 127-194 1.8 109494 1.9 9001494 1.9 12042-1 1.9. VlJttJI *1 • 1 f t*3W 1,9 79404 19 79414 1.9 8848-2 1.9 109404 1.9 794)14 1.9 ai ai ai 01 ai ai ai ai ai ai 01 01 ai 01 ai 01 ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai ai or 0.15 aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw aw 0.434 0434 0.284 0.284 O434 0434 0.434 0.434 O434 0.434 0.434 0434 0434 0.434 0.434 0.434 0.434 0434 0.434 O434 0.434 0434 0.434 0434 O434 O434 O434 0434 0434 0434 0434 0434 0434 O434 0.434 0434 0.434 0.434 0.434 0.434 0434 O434 0.434 0.294 0294 0.294 0.284 0.284 0.284 0.284 0284 0.284 0.284 0.284 0.284 0.284 0284 0.284 0.284 0284 0.284 0.284 0.284 0.284 0.284 0.284 0.284 6.294 0284 0.284 0.284 0.294 0.284 0284 0.284 0.284 0.284 0.284 0.284 0.284 0.284 0.294 0.130 tt 190 a 190 awfl 0.190 0,190 0.190 a 190 0.190 aiso 0.150 aiso 0.190 0.190 0.150 ai90 0.190 0.130 0.190 0.190 a 190 aiso 0.130 0130 a 130 0.199 a 130 0.190 a 190 owo O.WO a wo 0.190 0.190 0.190 0130 0.190 0.190 0.190 0150 0.190 1.32640 8.70602 092642 1*8602 104641 7.80642 1.18602 7.30602 1.04601 1.37602 4.86648 • 80608 7.93E40 103649 1.00605 8.90642 7.42642 1.04641 9.00642 7.82642 0286-02 1.25642 7.90642 1.12642 122642 9816-02 9426-02 1.78642 1.78642 1.81602 149640 7.28602 1.01601 7.80602 7.10602 7.W602 7.20602 8.70642 118642 300642 7.80642 7.80642 7.90642 3.14642 8.90642 1.08641 4.37648 070648 1.00649 4.99E48 7.90609 7.90648 1.04645 9.80648 1.01609 9.73648 1.00649 4.74649 7.80648 989648 4.88648 8.18608 9.91649 &9T648 997648 7.21608 121649 1.17605 9.90649 9.00649 7.90649 020640 8,80648 4.34E48 9.23649 9.80648 8.80648 9.10608 038648 9.20648 1.23645 1.05640 099640 993649 119640 0.31640 023649 9.43644 9.83649 8.31603 109640 9.91640 9.91640 5.036X0 9.31649 7.19643 029649 9.00E03 9.99604 999643 095604 9.79644 4.48640 4.33649 141640 141649 1.29643 1.99644 982603 8.07603 8.07603 987603 9.97649 979603 095649 9.27644 2.40640 9.23640 8.23640 8.31643 2.91643 8.70640 9.47640, 100648 100640 100648 1.00601 100648 1.00648 1.00640 1.00648 100648 100648 1.00648 1.00648 1.00648 100648 1.00640 100640 1.00649 100648 100648 1.00648 100608 100648 1.00648 1.00648 100648 1.00649 1.00640 1.00648 100648 100649 1.00640 1.00648 100649 1.00648 1.00649 100640 1.00648 1.00648 1.00648 100648 1.00640 1.00640 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 to w 10 to 13 19 19 19 W 19 19 W 19 W W 19 W W W. W W W W W W W w w w w w 19 W 19 W 19 19 19 19 19 W W IS IS IS 15 IS 15 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 W 19 19 19 19 19 19 19 19 19 19 19 19 19 15. 15 15 15 190644 1.80644 190644 1.90644 1.80644 •1.90644 190644 1.80644 1.90644 1.80644 1.90644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.80644 1.90644 1.90644 1.90644 1.90644 1.90644 1.80644 1.90644 1.90644 1.90644 1.90644 1.80644 1.80644 1.80644 1.80644 1806-04 180644 1.80644 1.80644 180644 053 008 0.10 0.4V 0* 0.00 0.59 010 007 051 0.10 0.10 009 0.07 0.08 009 Oil O.S6 009 0.82 0.57 0.12 0.13 039 039 0.43 2.7» 010 0.07 009 010 010 010 0.08 080 023 009 009 009 022 008 007 420643 2.20641 880644 4.846*04 9.706*01 7.806*01 LM«02 020641 1.206*00 2.70643 180641 1.80641 2.20643 8.80642 1.20641 2.40601 920642 1.006*00 1.20601 1.20641 1.10644 3.20641 2.40642 3.40644 • 90641 220642 2.80644 140605 7.10641 130641 580641 9.70642 • 40644 140601 150642 7.10641 2.50641 140644 1.10641 7.60641 410602 430641 170644 230642 3906*00 1.846*02 5f36*04 2.046*02 9.806*01 2.376*09 3,786*02 3.186*02 9.206*01 3806*01 8906*01 4.706*01 2.606*01 1.096*04 2.216*02 8816*03 7.246*03 8.996*03 3756*04 1.746*03 2.286*03 9.996*03 1836*03 • 006*00 1.606*01 1.316*02 8.126*02 .7.906*01 3.006*02 1.316*02 9016*02 1.546*03 • 906*01 7.806*01 • 406*01 2836*02 5006*00 1106*01 ii I i JI I'M 88888888888 88888888888 $$$$$$$$£$1 g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g l g g g 22222222222222222222222222222222 — '" ' i til til lit 111 li.! IJJ iii UJ iif ii-1 111 lif li> tif IAi tu ti) fl) llf tlJ >JJ <1J W :8^8^8^888^8888-3S8^.38^8^868s8B! o o i ' - - " V r < < N « 2222222 S 5 S 8 f B S S S 5 B o 5 S S 5 5 5 8 6 8 B 8 2 2 f 2 5 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 S o S S B S o S B U J i l l U i l I i U J i L i i i t i J f - f C i S R S S S -M A 9 W n M 0) ^i 5 5 5 5 5 5 5 5 5 5 5 S S 5 5 5 5 5 5 S 5 5 5 5 5 5 5 o e e e > c > e > o c > c i c > e > e i c > e e > c i c > c 9 o o o e e > c > c > e c > iii§iiiii^ ul W W lil KU U U-l U III O o in v <n n *f mm -.n •>•> w <•>« a •>« IIIIIIIII 9? 9 9 9 U lit UJ fi ; 5 ill i n t n i n i n i n i n i n i n i n •8 8888 ni fit IM |ii '" R R R R R R R . « 8 8 a « R B a R a R R . S R R R 8 J J « a R R B B R R 8 a 8 8 R R S a F 55555555555555555555551555.5555555555555515 g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g g ill I I »-i H-18 TUT 008 1539 f HtfJi I filil-i 8 8 8 5 8 $ ?8 g S K g g 8 P SI £ g 9 W i r t W r t W r t w w i r f i < « « « < i w i l « t » r i « i f < i W « 0 « < w « r i ^ W f » r f « t f w « ( W f « t < « « « i « < r t w W c i r t « t f 5 p S S p o S e S S S o S S S S q g S o S B S S S S S S o S S S E S o S S o B S o o S ^ipi^iliHiggoiig^^giggpiSggiiHg^Hi « » - « > « « > ^ i r i ^ ^ ^ « i f » « n « o « » « « « » t f « « . i o « i « » « j « i c i r t r i » i i » » « - r « f i ( » ; o i « f > i o > e i ^ w II *«» •> W iu fi r< r« K •»• rt w «j r» 2ii2 .•»' «< «* J S J W W <4 tf w « £ •> w «* « £ S^ i w » w •' 855 gill w w w n 5 5 w 5 5 'I 8 f £ 8 8 S E 9 9 9 m U | W U f | H W W W W B B R B B R R z l I 8 8 8 8 8 9 ' ui tij ui u ui ut 1 6 w B 6 fe B'i s?ss ii w w * w * I ;i ii 8 ii i I i:! Jit 35*:S5.J? sssiati. H-19 TUT OOS .154O Put* Outdoor OuMoor SSL 8*. m«toar SSL 89L. a (mpftj AUrin Kl HOI «*»(•»»• »C) HCHM*M»8HC) —H j-4. rn H- U« 1.1.1-TiteNonNttew I.U-TrtcMoKMttwiw TifcNorottfiyten* mSSBiS—SSSSLm NA 4.496*08 NA NA 7.81641 2.516*00 1.906*02 NA 9.29643 W NA NA HA 2.236*00 HA 1946*00 4.216*09 9.996*01 1.096*04 HA HA 1.086*03 HA 1146*02 1.426*09 NA 3916*02 NA 4.89641 9.33641 2.74641 4.296*01 1.086*01 2.28641 9.426*01 8.866*01 2.04641 7.416*01 1.756*09 8.946*09 9.396*02 2.81641 438642 9.836*00 1.29641 1.976*00 3.336*03 3.14641 1-306*00 1.316*00 9.88641 8.91641 1.476*01 1236*00 4.486*01 1946*00 8976m? 9.956*01 1.096*04 3.81641 1.076*01 1.086*03 4.496*00 1.426*03 7.81641 1916*00 1.906*02 3.916*02 9.29643 7.84E42 1.786*09 1.186*04 3.216*09 2.876*03 7.926*02 2.19641 4096*02 7.986*03 3.41643 1.296*02 7.306*01 9.186*03 8,316*03 3.006*03 1886*03 1.986*09 1.87641 1.736*02 173641 180641 1946*00 8.82649 1406*00 9.42641 1.936*00 4.086*01 1.496*04 1.746*04 1.926*09 2.576*02 3.076*03 1326*02 9.586*02 8.70641 3.076*01 1.176*09 4.406*03 1286*01 8816*02 8.586*09 2.786*09 1.396*03 1.046*03 8.746*01 9.556*02 3.716*09 4.856*00 1976*02 1356*02 1386*03 1816*09 1046*09 1.086*03 4.326*02 1.226*01 1976*02 1.126*01 1.176*01 1.076*02 1.946*00 2.986*01 7.476*00 8.846*01 4.936*02 3.836*09 3.736*09 1.706*09 1.446*09 1.776*09 4.726*02 9.216*02 1116*00 1876*02 2.486*03 7.936*02 1.356*09 2.246*04 3.016*03 1736*09 2.286*03 0.4 0.002 0.02 08. 0.09 0.0007 91 0.7 0.001 9* 28 102 4 0002 0.0001 008 •0.0007 3 21 0.04 1 0.09 03 0.9 7 • 0.08 08 0.01 004 t 189 0.007 0.09 • 2 8 9 0.009 0.01 84 . 9 0.00002 0.4 002 0.09 0.9 0.7 • 0.01 93 2 0007 ; 9« 89 239* 39 0.007 0.003 ' 0.3 0.004 4 89 0.04 1 0.09 0.8 0.8 7 * 0.07 08 0.3 0.3 29 472 002 0.3 2 9 89 0.02 0.09 94 14 0.002 05 05 03 43 11 02 54 87 0.2 9 297 239 939 0.3 0.04 10 0.1 2 297 03 1 1 I 0.9 7 ' 2 45 3 7m 1439 ' 0.4 II 921 2 214 960 1 3 190 351 0.01 APPENDIX I SSL Simulation Results TUT OOS 1542 APPENDIX I SSL Simulation Results Section 4.3.3 contains a complete description of the simulation setup and parameters. The following notation is used in the tables of this appendix. C = the nMBber of specimens per composite N * the number of composite samples chemically analyzed MU = the assumed true site mean (•* 0.5 SSL or 2 SSL) CV « the aifiiuned true value of the site coefficient of variation, (i.e. the true site standard deviation divided by the true she mean MU) MIX = the proportion of the site which is uncontaminated The remaining variables give the estimated probability of deciding to investigate further (PDIF) for a given method and simulation distribution. The variable names indicate the method of testing (Mx = Max test, C * Chen test, L = Land test) and the type of probability distribution used to generate values for the contaminated part of the EA (L - './gnormal, G *= gamma, W » Weibull). MxL,MxG,MxW C40L, C40G, C40W C30L, C30G, C30W C20L.C20G.C20W C10L, C10G, CWW C05L, C05G, C05W LfL, LfG, LfW LoL, LoG, LoW PDIF for Max rule applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal .40 significance level applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal JO significance level applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal .20 significance level applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal .10 significance level applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal .05 significance level applied to lognormal, gamma or Weibull data PDIF for Chen test at the nominal .01 significance level applied to lognormal, gamma or Weibull data PDIF for Land test of the flipped null hypothesis at the nominal .10 significance level applied to lognormal, gamma or Weibull data PDIF for Land test of the original null hypothesis at the nominal .05 significance level applied to lognormal, gamma or Weibull data. 1-1 TUT COS 1543 Appendix I. SSI Slmidtlon ftesultti EttliMted ProbabllltUi of invtttlgctfng furthw. rf -».' o \_.! CD J-fc cn NUNIX O.S .00 0.5 .50 2.0 .00 2.0 .50 NUNIX 0.5 .00 0.5 .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 NUNIX O.S .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 NUNIX 0.5 .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 NUNIX | 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 NxL C40L .124 .314 .192 .371 .750 .974 .848 .863 NxL C40L .140 .296 .213 .320 .343 .387 .700 .919 .753 .813 .699 .698 NxL C40L .157 .266 .212 .296 .459 .445 .642 .857 .692 .752 .481 .481 Nxl C40L .180 .280 .206 .277 •3oo «3o4 .632 .820 .631 .683 .457 .457 Nxl C40L .187 .267 .210 .262 .312 .306 .616 .791 .607 .659 .346 .346 C30L C20L .246 .161 .295 .182 .960 .926 .816 .745 C30L C20L .227 .150 .240 .167 .298 .193 .890 .838 .766 .688 .689 .657 C30L C20L .203 .138 .227 .165 .357 .208 .816 .764 .699 .620 .481' .480 C301 C201 .234 .163 .225 .158 .289 .207 .778 .724 .637 .558 .456 .438 C30L C20L .221 .152 .217 .154 .275 .199 .746 .674 .601 .532 .346 .346 C10L COSL .087 .042 .089 .039 .858 .759 .569 .361 ClOL COSL .073 .036 .080 .036 .067 .021 .733 .593 .524 .339 .456 .176 ClOL COSL .072 .026 .085 .038 .060 .014 .648 .512 .468 .312 .438 .082 ClOL COSL .091 .034 .074 .029 .090 .023 .593 .429 .434 .277 .336 .106 ClOL COSL .085 .035 .082 .029 .070 .021 .555 .397 .403 .270 .310 .131 COIL .016 .016 .426 .132 COIL .007 .010 .004 .260 .112 .028 COIL .002 .007 .001 .192 .089 .009 COIL .002 .002 .001 .143 .068 .005 COIL .002 .002 .002 .103 .058 .002 Lfl .099 .297 .873 .703 Lfl .088 .214 .241 .765 .680 .433 Lfl .088 .186 .103 .672 .617 .421 LfL .119 .170 .135 .631 .574 .314 Ifl .104 .151 .071 .580 .521 .307 lol .835 .937 .972 .947 lol .887 .932 .673 .964 .941 .704 lol| .900 .927 .490 .962 .930 .481 Lol| .912 .941 .501 .965 .939 .459 lol| .905 .897 .336 .962 .945 .346 NxG C40G .195 .350 .194 .347 .757 .874 .792 .813 NxG C40G .255 .338 .273 .343 .366 .407 .676 .740 .694 .712 .675 .673 NxG C40G .256 .317 .267 .324 .449 .437 .620 .668 .613 .630 .474 .471 NxG C40G .263 .291 .277 .297 .316 .309 .566 .589 .531 .555 .451 .450 NxG C40G .231 .252 .268 .273 .301 .299 .489 .509 .512 .516 .363 .363 C«l H«4 CV-1.5 C30G C200 C10G .276 .195 .088 .270 .170 .072 .838 .775 .650 .774 .699 .540 C«1 N*4 CV»2.0 C30G C20G C10C .277 .198 .106 .273 .191 .084 .302 .170 .061 .698 .645 .493 .675 .625 .466 .658 .615 .430 C«1 N*4 CV-2.5 C30G C20G C10G .257 .197 .089 .260 .187 .093 .350 .215 .063 .623 .562 .423 .592 .537 .404 .474 .474 .425 C«1 N>4 CV-3.0 C30G C20G C10G .238 .178 .093 .244 .179 .075 .265 .198 .075 .552 .502 .377 .501 .447 .329 .444 .422 .356 C«1 N*4 CV»3.5 C30G C20G C10G .203 .149 .078 .242 .186 .098 .266 .222 .101 .466 .416 .311 .492 .445 .338 .362 .356 .321 COSG C010 .043 .011 .028 .013 .494 .188 .334 .106 COSG COlG .046 .009 .031 .007 .023 .008 .341 .072 .291 .072 .161 .030 COSG COlG .034 .004 .033 .003 .015 .006 .266 .044 .219 .040 .096 .008 COSG COlG .030 .004 .028 .002 .015 .002 .225 .030 .179 .024 .123 .002 COSG COlG .019 .003 .022 .002 .018 .000 .179 .017 .199 .014 .111 .000 Lfo LOO .172 .975 .257 .931 .742 .991 .659 .925 LfG LoG .182 .965 .201 .905 .205 .680 .624 .988 .572 .926 .406 .681 LfG LoO .183 .944 .194 .868 .117 .486 .526 .972 .496 .885 .406 .474 LfG LoG .164 .859 .142 .764 .105 .450 .455 .921 .388 .860 .348 .465 LfG LoG | .124 .784 .145 .722 .094 .341 .361 .862 .376 .819 .311 .364 NxU C40U C30U C20U ClOU COSU COIU LfU LeU .176 .360 .286 .209 .104 .042 .009 .151 .945 .194 .373 .294 .197 .086 .039 .017 .265 .929 .760 .903 .870 .808 .707 .SSI .340 .768 .988 .825 .839 .799 .729 .989 .370 .117 .66$ .952 • HxU C40U C30U C20U ClOU COSU COIU IfU ' LoU .219 .315 .261 .169 .084 .033 .006 .143 .961 .241 .330 .261 .167 .075 .032 .009 .214 .902 .377 .414 .305 .175 .055 .016 .003 .215 .676 .695 .809 .770 .716 .596 .433 .138 .663 .995 .713 .747 .698 .620 .481 .295 .073 .585 .938 .669 .664 .642 .596 .442 .175 .027 .425 .694 NxU C40U C30U C20U ClOU COSU COIU LfU LoU .225 .306 .236 .168 .078 .028 .DOS .145 .948 .264 .310 .263 .189 .091 .032 .006 .191 .896 .441 .432 .350 .213 .064 .012 .005 .120 .499 .624 .718 .671 .60S .469 .331 .073 .558 .986 .635 .672 .627 .567 .412 .233 .036 .530 .918 .473 .473 .470 .464 .419 .091 .011 .404 .474 NxU C40U C30U C20U C10U COSU COIU LfU LoU .226 .275 .226 .163 .087 .029 .005 .142 .950 .270 .302 .253 .181 .087 .024. .003 .163 .872 .359 .353 .306 .235 .098 .028 .003 .133 .480 .581 .657 .615 .558 .430 .280 .048 .514 .986 .591 .616 .588 .530 .395 .242 .038 .489 .90S .471 .468 .455 .435 .353 .138 .005 .348 .491 NxU C40U C30U C20U ClOU COSU COIU LfU LoU .226 .259 .214 .163 .080 .034 .002 .137 .9*5 .231 .266 .214 .170 .079 .026 .004 .141 .860 .299 .293 .266 .216 .096 .023 .001 .099 .348 .558 .631 .587 .509 .398 .247 .037 .477 .983 .524 .549 .506 .448 .331 .190 .023 .410 .891 .328 .328 .327 .324 .291 .120 .004 .289 .331 Appendix I. SSL Simulation Result*: Estimated Probabilities of Investigating Further. C H til £> ul Ml MIX ] O.S .00 O.S .50 O.S .90 2.0.00 2.0 .50 2.0 .90 MINI* O.I .00 O.I .10 2.0 .00 2.0 .90 Ml MIX 0. .00 0. .50 0. .75 2. .00 2. .50 2.0 .75 Ml MIX O.S .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 Ml MIX O.S .00 0.5 .50 O.S .85 2.0 .00 2.0 .50 2.0 .85 MxL C40L C30L C20L C10L C05L COIL LfL lot .173 .241 .203 .141 .041 .024 .003 .091 .920 .198 .239 .190 .142 .Oft .027 .001 .135 .916 .273 .268 .235 .184 .Oft .012 .000 .089 .329 .594 .750 .708 .632 .5$ .387 .104 .534 .961 .590 .655 .596 .501 .3B .239 .039 .513 .942 .354 .354 .351 .340 .30) .123 .001 .296 .355 MxL C40L C301 C20L ClOt C05L COIL LfL lot «181 .3fO .252 .180 .00 .039 .005 .099" .634 .633 .347 .285 .206 .100 .OSS .614 .$20 ,9fct .860 .906 .979 .971 .939 .89$ .498 .9*5 .944 .931 .924 .901 .844 .722 .565 .242 .799 .992 Mxt C40L C30L C20L C10L C05L COIL LfL lot .220 .301 .237 .168 .081 .032 .004 .099 .728 .260 .304 .253 .176 .085 .034 .005 .223 .965 .459 .385 .293 .197 .083 .038 .007 .188 .844 .816 .960 .946 .917 .849 .760 .481 .855 .947 .891 .891 .842 .777 .633 .497 .189 .732 .989 .819 .796 .769 .697 .506 .353 .068 .461 .824 Nxl C40L C30L C20L C10L C05L COIL LfL Lot .267 .292 .241 .163 .095 .047 .004 .105 .796 .312 .333 .266 .190 .092 .035 .004 .219 .956 .591 .409 .300 .194 .082 .031 .003 .154 .631 .804 .932 .90S .868 .791 .681 .340 .801 .947 .827 .834 .789 .724 .591 .435 .151 .710 .973 .628 .628 .628 .621 .456 .221 .014 .226 .628 MxL C40L C30L C20L C10L C05L COIL If I LoL .249 .282 .230 .159 .083 .034 .001 .091 .808 .305 .296 .235 .164 .083 .031 .008 .173 .950 .495 .377 .290 .188 .072 .024 .003 .118 .647 .797 .899 .865 .833 .745 .595 .261 .760 .956 .778 .780 .740 .670 .544 .379 .114 .644 .980 .637 .623 .607 .556 .403 .215 .023 .230 .638 MxG .255 .232 .280 .443 .453 .330 KxQ .210 .276 .873 .926 MxG .348 .366 .486 .822 .837 .804 .MxG .356 .371 .597 .759 .765 .596 MxG .380 .393 .486 .705 .690 /• C40G C30G .265 .234 .235 .191 .269 .242 .450 .426 .459 .424 .330 .325 -•- C«1 1 C40G C30G .353 .270 .371 .298 .928 .902 .899 .868 C40G C30G .348 .280 .369 .281 .395 .296 .832 .802 .818 .786 .777 .747 C40G C30G .324 .254 .329 .252 .434 .311 .745 .705 .734 .700 .596 .596 — -- C«1 1 C40G C30G .327 .275 .337 .274 .397 .329 .669 .621 .646 .602 .585 .564 1 ™ % C V% • v C20G CIOO .177 .093 .146 .078 .196 .111 .387 .294 .380 .283 .314 .277 1=6 CV*1.5 C20G CIOG .173 .090 .204 .104 .861 .709 .830 .730 1*6 CV*2.0 CMC CIOG .189 .100 .183 .088 .199 .082 .752 .630 .718 .609 .676 .518 ••A PU«9 C C20G CIOG .172 .088 .179 ,073 .200 .085 .649 .521 .636 .520 .586 .431 l«6 CV-3.0 C20G CIOG .211 .101 .208 .100 .228 .099 .560 .444 .537 .429 .524 .403 C050 C010 .029 .002 .027 .001 .032 .003 .162 .009 .150 .009 .132 .004 COSG COlG .047 .006 .049 .007 .177 .356 .973 .235 COSG C01G .039 .005 .034 .002 .030 .003 .497 .167 .453 .155 .360 .058 COSQ COlG .044 .005 .034 .002 .027 .000 .362 .086 .367 .076 .223 .019 COSG COlG .036 .003 .040 .001 .033 .005 .307 '060 .287 .049 .209 .029 LfO LoG .141 .727 .106 .443 .115 .$75 .333 .753 .315 .751 .270 i$40 KG LoG .172 .953 .300 .971 .843 .992 .781 .980 LfG loG .182 .957 .209 .931 .179 .833 .721 .993 .657 .966 .478 .812 LfG LoG .162 .921 .152 .892 .154 .637 .585 .980 .552 .951 .226 .596 LfG LoG .155 .854 .161 .844 .132 .647 .487 .947 .443 .892 .239 .627 MXWC40U .194 .213 .235 .254 .240 .234 .511 .579 .505 .517 .330 .326 MXHC40U .218 .357 .320 .418 .881 .956 .920 .891 MxUC40U .317 .346 .387 .372 .504 .393 .831 .887 .846 .834 .786 .768 MXUC40H .286 .297 .373 .340 .559 .434 .781 .815 .795 .756 .650 .649 NxU C40U .336 .322 .337 .296 .460 .371 .720 .752 .704 .676 .583 .560 C30U C20U .178 .131 .215 .147 .207 .155 .526 .466 .486 .427 .320 .301 C30U C20U .279 .184 .30) .200 .940 .900 .869 .821 C30UC20U .276 .202 .289 .217 .298 .187 .849 .805 .787 .713 .738 .665 C30UC20U .243 .162 .263 .183 .318 .203 .772 .715 .708 .641 .646 .629 C30U C20U .254 .191 .239 .183 .306 .211 .707 .652 .634 .557 .536 .500 C10U .068 .074 .082 .342 .318 .254 C10U .061 .107 .141 .709 C10U .101 .114 .068 .689 .607 .473 C10U .090 .089 .073 .599 .512 .470 C10U .100 .094 .081 .517 .442 .406 COSU COIU .025 .001 .023 .000 .020 .000 .213 .018 .166 .017 .145 .000 COSU COIU .027 .005 .041 .007 .733 .417 .543 .213 COSU COIU ,040 .004 .039 .003 .023 .003 .539 .229 .450 .147 .313 .058 COSU COIU .040 .005 .035 .004 .019 .001 .474 .150 .370 .107 .230 .020 COSU COIU .039 .004 .035 .007 .027 .001 .391 .112 .31' .068 023 LfU LoU .105 .910 .125 .822 .092 .324 .445 .979 .380 .878 .254 .353 LfU LoU .128 .916 .304 .969 .070 .992 * fOw •TwW LfU LoU .157 1947 .243 .958 .172 .809 .758 .990 .675 .980 .423 .804 LfU LoU .138 .942 .178 .919 .156 .618 .664 .990 .576 .963 .232 .651 LfU LoU .149 .941 .144 .877 .112 .624 .595 .985 .477 .952 .224 .604 Appendix I. SSI Simulation Results* Estimated Probabilities of InvtttlMtlng Further. C 00 !— NUNIX 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 NUNIX 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 NUNIX 0.5 .00 0.5 .50 2.0 .00 2.0 .50 NUNIX 0.5 .00 0.5 .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 NUNIX 0.5 .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 Mxl C401 C30L C20L ClOL C051 C01L Ifl lol NxO C40G C300 C200 ClOO C05G COlO .240 .251 .203 .140 .079 .028 .002 .082 .833 .361 .309 .271 .201 .102 .034 .002 .289 .271 .212 .154 .080 .040 .003 .137 .931 .369 .307 .240 .164 .087 .037 .003 .407 .326 .258 .185 .093 .032 .001 .091 .441 .417 .340 .290 .202 .079 .037 .002 .771 .870 .832 .777 .680 .544 .217 .706 .958 .616 .586 .529 .470 .373 .234 .034 .756 .744 .687 .629 .513 .354 .099 .598 .980 .647 .602 .564 .508 .410 .259 .038 .442 .440 .440 .433 .354 .123 .014 .123 .442 .449 .445 .442 .428 .350 .140 .014 CM 4 M»A f*U*»i fl Nxl C40L C30L C20L ClOL COSL COIL IfL lol NxG C40G C30G C20G C10G C05G COIG .246 .255 .200 .131 .062 .023 .004 .084 .839 .342 .275 .229 .181 .092 .028 .001 .299 .271 .213 .153 .086 .037 .002 .134 .922 .315 .269 .220 .160 .084 .020 .000 ,396 .324 .269 .196 .082 .022 .000 .065 .453 .400 .340 .296 .21 1 .099 .041 .001 .723 .847 .797 .724 .600 .470 .179 .626 .956 .559 .523 .494 .450 .334 .214 '.023 .745 .735 .683 .609 .482 .342 .093 .597 .968 .576 .532 .495 .429 .334 .200 .023 .477 .472 .464 .439 .337 .140 .007 .145 .479 .471 .449 .437 .415 .335 .164 .013 .. — ..'...... ————————————————— - — . ——————— c-1 M«9 CV-1.5 ————— • Nxl C40L C30L C20L ClOL COSL COIL IfL Lol NxG C40G C30G C20G C10G C05G COIG .286 .335 .274 .205 .112 .057 .010 .112 .452 .363 .390 .293 .202 .089 .043 .006 .365 .380 .298 .203 .108 .050 .014 .380 .984 .420 .411 .301 .213 .103 .046 .009 .948 .999 .999 .995 .989 .965 .891 .987 .950 .955 .973 .963 .936 .887 .815 .577 .983 .956 .940 .904 .820 .719 .449 .856 .995 .978 .954 .930 .900 .817 .697 .430 Mxl C40L C301 C20L ClOl COSL COIL LfL lol | NxG C40G C30G C20G C106 C05G COIG .312 .314 .247 .173 .101 .049 .005 .110 .592 .472 .369 .297 .199 .096 .034 .005 .425 .366 .289 .205 .092 .045 .007 .287 .962 .496 .384 .285 .199 .102 .045 .010 .629 .418 .307 .215 .107 .053 .008 .198 .840 .642 .417 .324 .208 .102 .051 .005 .913 .987 .983 .974 .955 .910 .714 .953 .948 .913 .897 .864 .828 .742 .619 .341 .951 .923 .892 .842 .756 .630 .338 .816 .989 .930 .908 .884 .844 .710 .576 .271 .918 .873 .826 .752 .615 .464 .138 .456 .920 .925 .872 .842 .761 .642 .469 .170 ..................................'. — ... ———————— ... c«f M*9 CV'2.5 ----•—•«• NxL C40L C301 C20L C10C COSL COIL LfL Lol NxG C40G C30G C20G C10G C05G COIG .364 .323 .256 .179 .092 .044 .006 .097 .667 .477 .346 .283 .196 .090 .046 .004 .416 .315 .250 .172 .096 .042 .006 .216 .957 .525 .377 .297 .209 .101 .045 .007 .706 .384 .280 .186 .083 .027 .002 .118 .744 .737 .407 .315 .197 .093 .036 .004 .910 .980 .970 .954 .905 .846 .593 .912 .952 .867 .821 .782 .725 .623 .486 .191 .931 .900 .868 .823 .731 .607 .322 .790 .990 .888 .823 .788 .729 .628 .489 .179 .743 .741 .731 .681 .458 .367 .073 .364 .743 .764 .762 .750 .686 .432 .355 .064 LfO .130 .103 .086 .371 .389 .141 IfG .099 .087 .078 .323 .299 .168 IfG .216 .336 .933 .841 LfO .206 .250 .187 .808 .743 .472 IfG .166 .184 .118 .656 .629 .351 LoG .785 .784 .460 .873 .851 .451 LoG .725 .691 4*97 .821 .792 .489 loG .937 .980 .999 .994 loG .934 .940 .841 .997 .987 .927 LoG .864 .862 .754 .983 .961 .764 MxU .296 .340 .403 .714 .672 .472 MxU .289 .325 .350 .675 .623 .446 MxU .336 .416 .957 .973 MxU .397 .495 .656 .933 .938 .920 MxU .435 .473 .719 .886 .902 .782 C40U .265 .287 .357 .727 .636 .468 C40U .268 .275 .301 .679 .574 .428 C40U .375 .392 .985 .955 C40U .317 .385 .396 .950 .888 •866 C40U .350 .341 .406 .892 .834 .773 C30U .216 .237 .311 .684 .596 .459 C30U .216 .226 .254 .634 .531 .411 C30U .287 .290 .974 .936 C30U .239 .301 .306 .930 .860 .838 C30U .287 .270 .307 .861 .798 .760 C20U .160 .176 .223 .614 .535 .448 C20U .153 .173 .196 .568 .468 .390 C20U .188 .201 .965 .907 C20U .159 .220 .209 .905 .790 .764 C20U .199 .188 .185 .811 .738 .708 CIOU .085 .099 .085 .489 .433 .375 CIOU .088 .080 .078 .457 .348 .326 CIOU .092 .095 .927 .828 CIOU .087 .112 .102 .826 .696 .609 CIOU .103 .107 .081 .712 .615 .459 COSU .037 .040 .033 .344 .270 .143 COSU .034 .035 .030 .317 .236 .160 COSU .036 .046 .855 .713 COSU .048 .058 .037 .727 .558 .465 COSU .049 .049 .025 .601 .489 .370 COIU .003 .002 .000 .080 .058 .009 COIU .002 .001 .003 .057 .032 .013 COIU .008 .008 .648 .403 COIU .009 .010 .003 .409 .267 .170 COIU .003 .004 .001 .316 .203 .090 IfU LoU .117 .927 .126 .854 .084 .457 .562 .987 .463 .936 .144 .476 LfU loW .121 .883 .112 .818 .058 .438 .519 .982 .380 .906 .175 .473 IfU LoU .168 .880 .335 .972 .950 .994 .863 .991 LfU LoU .151 .907 .255 .940 .193 .840 .872 .989 .720 .986 .470 .923 LfU loU .159 .906 .179 .871 .100 .746. .788 .992 .649 .976 .366 .782 Appendix t. SSL Simulation Results: fit limited Probabilities of Investigating further. c • CO NUNIX 0.3 .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 Ml NIX 0.5 .00 0.5 .50 0.5 .90 2.0. .00 2.0 .50 2.0 .90 NUNIX 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0..90 NUNIX 0.5 .00 0.5 .50 2.0 .00 2.0 .50 Ml NIX 0.5 .00 0.5 .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 Nxl .354 .443 .615 .888 .900 .763 Nxl .353 .427 .572 .880 .885 .624 Nxl .385 .431 .520 .858 .864 .636 Nxl .350 .450 .985 •Wo Nxl .390 .496 .738 .967 .988 .963 C40L C30L .292 .218 .313 .239 .391 .310 .953 .930 .878 .848 .726 .690 C401 C30L .271 .222 .319 .259 .350 .276 .927 .902 .841 .795 .620 .614 C40L C301 .301 .230 .307 .252 .326 .267 .895 .862 .808 .774 .609 .582 C40L C30L .368 .276 .397 .311 1.00 1.00 .977 .965 C40L C30L .330 .273 .367 .281 .396 .290 .997 .995 .967 .949 .907 .875 C20L ClOl .153 .074 .166 .091 .203 .103 .895 .840 .803 .693 .618 .467 C20L ClOl .158 .085 .186 .103 .201 .099 .876 .800 .739 .606 .577 .390 C20L ClOl .168 .097 .181 .097 .197 .099 .809 .723 .719 .610 .523 .395 C20L ClOl .206 .102 .209 .122 1.00 .999 .950 .904 cm CIOL .186 .084 .189 .093 ;181 .100 .993 .977 .918 .852 .821 .702 C05L COIL .030 .006 .041 .003 .040 .003 .739 .438 .570 .232 .306 .069 COSl COIL .046 .002 .038 .005 .036 .002 .699 .385 .503 .198 .195 .030 COSL COIL .042 .005 .033 .003 .042 .007 .619 .292 .475 .173 .223 .040 COSl COIL .054 .014 .057 .009 .993 .962 .822 .598 COSL COIL .041 .004 .046 .008 .057 .008 .959 .865 .763 .500 .580 .301 Ifl lol .093 .729 .191 .948 .101 .686 .834 .937 .767 .983 .287 .764 Ifl lol .096 .742 .179 .911 .053 .595 .819 .951 .688 .983 .189 .624 Ifl Lol .108 .755 .152 .876 .057 .546 .740 .952 .690 .982 .205 .636 Ifl lol .109 .280 .441 .989 .998 .955 .905 .998 Ifl Lol .094 .439 .313 .964 .171 .856 .978 .946 .867 1.00 .565 .905 ........... c«i w cV-3.0 -• —— • — .......... NXG C40G C30G C20G C10G COSO COlG IfG LoG .494 .338 .276 .205 .115 .049 .008 .144 .817 .507 .337 .265 .173 .082 .029 -.002 .121 .801 .604 .363 .284 ..200 .085 .030 .002 .080 .660 .851 .785 .731 .676 .548 .399 .099 .537 .955 .833 .744 .707 .644 .519 .368 .111 .485 .927 .753 .708 .682 .622 .479 .315 .089 .303 .764 ——— . — c«f n*9 CV«3.5 .-.» —— > —————— • NxG C40Q C30G C20G C10G COSG COlG LfG LoG .474 .323 .271 .198 .092 .046 .006 .112 .728 .471 .329 .265 .194 .094 .053 .002 .102 .701 .537 .353 .288 .199 .097 .040 .003 .058 .560 .766 .690 .652 .604 .483 .333 .083 .434 .896 .779 .699 .659 .612 .493 .335 .087 .409 .873 .610 .592 .581 .560 .401 .209 .056 .203 .611 NxG C40G C3DG C20G C1' G COSO COlG IfO LoG .475 .315 .255 .182 .099 .043 .003 .079 .673 .444 .323 .270 .215 .116 .040 .004 .089 .628 .483 .324 .267 .191 .095 .037 .003 .044 .527 .730 .640 .596 .535 .429 .280 .056 .348 .845 .709 .608 .568 .520 .399 .257 .046 .308 .822 .594 .544 .523 .484 .385 .201 .031 .171 .601 NxG C40G C306 C20G ClOO COSG COlG IfG LoG .480 .386 .296 .207 .097 .045 .004 .241 .922 .494 .398 .312 .208 .094 .048 .014 .398 .974 .989 .993 .984 .971 .939 .906 .751 .967 .998 .995 .981 .963 .948 .897 .815 .571 .902 .996 .. —— ... c.t N»12 CV«2.0 —————— • ———— - NxG C40G C30G C20G C10G COSG COlG IfG LoG .576 .383 .300 .199 .101 .043 .006 .223 .949 .593 .351 .264 .178 .085 .046 .009 .231 .943 .747 .414 .320 .201 .090 .041 .011 .186 .822 .977 .952 .927 .899 .821 .746 .480 ;890 .999 .971 .933 .910 .878 .778 .682 .399 .786 .991 .9?" 928 .900 .840 .724 .598 .305 .575 .929 NxUC40U .426 .313 .464 .330 .609 .379 .863 .856 .855 .80S .736 .692 NxU C40U .418 .293 .427 '.293 .553 .383 .836 .796 .819 .721 .620 .594 NXUC40U .408 .267 .431 .299 .471 .321 .836 .774 .797 .733 .588 .544 NXUC40U .438 .353 .507 .396 •y»r *YTD .992 .971 NXUC40U .524. .377 .590 .393 .741 .402 .977 .978 .981 .947 .954 .900 C30UC20U .256 .172 .264 .194 .284 .187 .823 .777 .761 .700 .660 .612 C30UC20U .249 .190 .241 .175 .322 .214 .748 .693 .683 .618 .573 .537 C30U C20U .211 .154 .239 .174 .275 .181 .729 .665 .674 .603 .523 .478 C30UC20U .276 .189 .313 .223 .990 .986 .948 .926 C30U C20U .30$ .214 .321 .225 .302 .198 .96$ .946 .920 .880 .866 .818 C10U .079 .105 .083 .684 .597 .476 C10U .106 .080 .108 .572 .506 .424 C10U .076 .088 .077 .558 .488 .373 C10U .096 .107 .972 .877 C10U .124 .119 .085 .89$ .793 .706 COSU COIU .038 .003 .047 .004 .032 .005 .560 .226 .446 .166 .306 .077 COSU COIU .051 .004 .040 .003 .043 .003 .435 .171 .369 .097 .213 .041 COSU COIU .038 .002 .034 .005 .030 .001 .410 .143 .330 .089 .201 .035 COSU COIU .056 .008 .052 .013 .948 .808 .800 .560 COSU COIU .059 .008 .057 .012 .047 .007 .840 .595 .684 402 .5? 11 LfU LoU .133 .850 .156 .854 .085 .671 .751 .988 .610 .958 .299 .744 IfU LeW .127 .835 .119 .790 .059 .584 .639 .982 .507 .954. .204 .624 IfU LoU .118 .798 .098 .747 .038 .510 .607 .977 .477 .924 .179 .594 IfU LoU .185 .821 .386 .975 .980 .990 .876 .994 LfU LoU .204 .919 .285 .940 .181 .809 .937 .998 .811 .995 .542 .907 Appendix I. Itl Simulation Results: EttlMted Probabilities of Investigating rurtMr. C —! cn .£,. Ml NIX 0.1 .00 0,9 .50 0.9 .85 2.0 .00 2.0 .90 a.tf .as Ml NIX 0.5 .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 Ml NIX 0.9 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Ml NIX 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Ml NIX 0.5 .00 0.5 .50 2.0 .00 2.0 .50 NM. C40L C30L C20L ClOL COSl COU LfL tol .4lS .296 .22) .19> .084 .038 .005 .089 .520 .4*2 .305 .234 .171 .083 .041 .006 .227 .942 .8)3 .413 .312 .194 .104 .036 .003 .092 .596 .968 .991 .984 .981 .959 .921 .720 .96) .956 .919 .940 .914 .8/7 .799 .686 .3*9 .844 .995 .8)9 .844 .812 .7ft .544 .406 .If 6 .250 .842 NxL C40L C30L C20L C101 COSL COU Ift LoL .429 .311 .2)9 .162 .079 .042 .005 .092 .612 .510 .322 .253 .187 .109 .043 .005 .217 .930 .728 .381 .290 .205 .117 .056 .012 .088 .593 .93) .972 .959 .932 .900 .84) .605 .909 .932 .955 .917 .883 .843 .748 .634 .318 .797 .991 .844 .769 .718 .631 .511 .364 .124 .252 .760 HxL C40L C30L C20L ClOL COSL COU Ifl LoL .4)7 .307 .237 .174 .095 .044 .005 .111 .660 .508 .305 .238 .174 .094 .040 .003 .173 .908 .680 .377 .299 .196 .085 .0)7 .001 .033 .409 .940 .957 -.934 .912 .857 .777 .522 .863 .952 .945 .900 .863 .809 .715 .596 .284 .765 .989 .72) .710 .684 .628 .420 .295 .08) .147 .70) NxL C40L C)OL C20L ClOl COSL COIL LfL LoL .470 .290 .226 .169 .090 .04) .005 .10) .699 .495 .)1) .263 .17) .098 .04) .002 .152 .904 '.638 .351 .289 .19) .094 .051 .007 .052 .425 .942 .950 .926 .899 .824 .740 .471 .8)8 .955 .9)0 .849 .819 .769 .67) .561 .227 .736 .985 .71) .672 .640 .575 .425 .286 .074 .151 .666 NxL C40L OOL C20L ClOL COSL COIL LfL Id .438 .354 .279 .202 .102 .053 .007 .092 .161 .535 .387 .303 .205 .110 .062 .014 .482 .986 .996 1.00 1.00 1.00 1.00 1.00 .997 1.00 .957 1.00 .990 .982 .973 .950 .910 .761 .945 1.00 NxG C40G .627 .375 ..623 .384 .802 .411 .939 .879 .951 .882 .851 .835 HxG C406 .600 .365 .610 .365 .707 .358 .925 .821 .909 .826 .852 .786 . — ..... | NxO C40G .608 .360 .589 .325 .666 .388 .870 .760 .866 .749 .709 .685 ——— ... | NxG C40G .559 .317 .563 .319 .60S .365 .830 .705 .817 .689 .710 .652 ..... —— ( HxG C400 .565 .393 .586 .411 .994 .995 .999 .988 C«l W»1Z CV*2.5 C30G C20G C100 .293 .201 .092 .287 .195 .087 .308 .216 .097 .852 .803 .703 .850 .795 .691 .807 .715 .568 >1 N*12 CV*3.0 C30G C20G C10G .283 .197 .099 .300 .211 .114 .263 .198 .107 .784 .724 .608 .786 .719 .599 .745 .680 .535 >1 N«12 CV-3.5 C30G C20G C10G .294 .214 .108 .259 .182 .082 .304 .204 .100 .712 .648 .529 .690 .627 .506 .656 .598 .439 >1 N*12 CV*4.0 C30G C20G C10G .261 .197 .107 .260 .182 .108 .304 .204 .095 .667 .605 .490 .640 .586 .481 .621 .569 .437 •1 N*16 CV«1.5 C30G C20G C10G .307 .216 .100 .291 .200 .111 .992 .989 .975 .982 .959 .925 COSO C010 .040 .006 .042 .007 .046 .009 .583 .289 .577 .257 .455 .120 COSO C010 .043 .004 .059 .005 .052 .004 .492 .183 .458 .167 .378 .130 COSO C01G .042 .005 .046 .004 .050 .004 .394 .134 .359 .108 .283 .092 COSO C010 .049 .005 .043 .004 .043 .004 .347 .093 .317 .087 .257 .051 COSO C01G .058 .012 .052 .004 .956 .872 .875 .714 Lffl .176 .159 .091 .742 .6*4 .264 LfG .119 .148 .089 .587 .520 .267 LfO .109 .088 .055 .445 .398 .158 LfO .081 .080 .040 .374 .325 .139 LfG .293 .459 .989 .932 LoO .902 .878 .558 .991 .976 .830 LOG .806 .778 .540 .964 .943 .784 LOG .7)5 .708 .424 .929 .889 .681 LoO .640 .6)2 .442 .861 .8)0 .654 LoG .900 .976 .999 .998 NxW .5)0 .614 .814 .965 .954 .856 NxU .547 .565 .71) .9)) .912 .820 NxU .505 .541 .6)2 .918 .886 .708 NxU .525 .559 .585 .88) .880 .706 NxU .542 .595 .993 .998 C40W C30W .340 .273 .377 .285 .387 .290 .948 .919 .894 .860 .829 .798 C40W C30U .337 .258 .355 .269 .404 .315 .901 .873 .852 .814 .755 .716 C40U C30U .316 .260 .301 .247 .386 .316 .870 .838 .783 .737 .672 .654 C40W C30U .328 .273 .340 .274 .343 .282 .803 .760 .774 .735 .647 .611 C40U C30U .371 .287 .400 .292 .997 .996 .987 .976 C20UC10U .190 .088 .197 .093 .195 .086 .877 .816 .824 .736 .730 .549 C20U C10U .190 .090 .192 .093 .225 .118 .832 .754 .752 .634 .657 .529 C20U C10U .194 .110 .179 .093 .235 .121 .792 .677 .668 .572 1605 .445 C20U C10U .201 .099 .197 .097 .196 .093 .710 .618 .678 .555 .568 .443 C20W C10W .196 .111 .204 .103 .990 .983 .962 .928 COSW C01W .038 .005 .045 .005 .037 .003 .709 .406 .624 .307 .438 .134 COSU C01U .047 .006 .048 .006 .053 .004 .634 .310 .510 .227 .380 .119 COSU C01U .046 .004 .039 .006 .047 .005 .563 .238 .438 .176 .288 .091 COSU C01U .047 .004 .045 .006 .049 .002 .497 .181 .415 .136 .252 .061 COSU C01U .054 .010 .062 .007 .973 .924 .885 .734 LfU lc .164 .8< .187 .9' .086 .52 .873 .* .731 .9f .255 .0; LfU lc .145 .8! .149 .8< .090 .5* .808 .9". .637 .93 .250 .7* .LfU lc .139 .8A .110 .71 .049 .43 .747 .99 .561 .96 .166 .67 LfU lo .116 .84 .107 .78 .046 .41 .668 .98 .508 .94 .148 .6A LfU lc .223 .80 .447 .97 .988 .99 .922 .99 Apptndlx I. SSt SlfluUtton Results: Estimated Probabilities of Investigating Further. H MUMIX 0 .00 0 .50 0 .75 2 .00 2 .50 2 .75 Mil MIX 0.) .00 0.) .50 O.S .85 2.* .80 2.0 .50 2.0 .85 MUMIX O.S .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 MUNIX O.S .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Mxl C40L .502 .331 .594 .379 .804 .401 .990 .997 .997 .984 .992 .950 Mxl C401 .541 .327 .429 .344 .192 .401 •To4 «9w .991 .941 .924 .899 Mxl C40L .539 .322 .418 .330 .808 .391 .974 .984 .982 .948 .924.. 850 Nxl C40L .SSt .287 .442 .331 .787 .387 .978 .984 .973 .932 .803 .772 C30L .258 .291 .314 .997 .975 .927 CSOL .245 .294 .319 .991 .948 .847 CSOL .240 .248 .304 .981 .929 .800 csoi .224 .273 .302 .978 .913 .728 C20L ClOt .173 .09 .202 .11 .229 .09 • 9W •TT, .957 .921 .884 .80! C201 ClOt .188 .111 .209 .111 .204 .10 .985 .971 .923 .87 .772 .431 C20LC10I .173 .OH .181 .081 .220 .123 .944 .941 .899 .83 .734 .41! cm ci« .154 .081 .200 .102 .211 .09) .941 .924 .872 .792 .453 .SM . COSl COIL .041 .007 .058 .011 .051 .007 .987 .93* .850 .45* 1 .497 .404 COSL COIL 1 .047 .009 .053 .009 .043 .003 .962 .84* .799 .557 ' .489 .199 COSL COIL t .034 .005 r .039 .006 1 .060 .007 t .907 .747 .732 .45.7 ! .478 .195 COSL COIL 1 .038 .004 ' .056 .007 ' .044 .004 .880 .491 .477 .397 > .380 .128 LfL Id .103 .291 .371 .964 .197 .805 .992 .997 .925 .999 .599 .946 LfL lei .112 .420 .300 .147 .085 .176 .971 .95J .88) .995 .285 .719 Lfl Id .094 .504 .224 .924 .079 .543 .949 .948 .840 .995 .285 .720 Ifl Lot .084 .520 .209 .904 .033 .415 .938 .940 .830 .995 .190 .504 NxO C400 C300 C200 C100 C050 COIO LfO LoG .682 .401 .319 .219 .113 .057 .009 .279 iM7 .736 .394 .296 .213 .092 .041 .006 .276 «WS .809 .389 .308 .206 .093 .046 .004 .173 .F83 .994 .981 .976 .957 .919 .851 .623 .956 .998 .992 .971 .952 .929 .866 .778 .528 .862 .996 .987 .950 .920 .873 .789 .472 .413 .563 .928 MxG C406 C30G C20G C10G COSQ COlG IfG loO .700 .375 .298 .204 .104 .057 .009 .195 .872 .725 .347 .276 .196 .107 .060 .014 .155 .134 .898 .420 .307 .212 .099 .050 .009 .079 .404 ' .*/9 .922 .905 .844 .785 .488 .425 .813 .9*2 .972 .918 .899 .860 .784 .671 .384 .734 .974 .935 .904 .868 .794 .710 .520 .244 .337 .755 ————— O1 n.16 CV-3.0 —————— —. ——— • MxG C406 C30B C200 ClOt COSQ COlO If 6 LoO .705 .347 .247 .184 .094 .045 .007 .119 .774 .714 .375 .289 .210 .113 .052 .008 .T28 .729 .821 .394 .309 .203 .103 .060 .009 .076 .540 .973 .895 .868 .814 .710 .579 .294 .654 .979 .962 .874 .848 .790 .682 .550 .257 .594 .952 .905 .830 .773 .721 .574 .449 .193 .277 .705 ... —— .. c-1 N-16 CV-3.5 —————— —. ..." NxO C40G C30G C20G ClOG COSO C018 IfO loO .676 .354 .286 .209 .108 .046 .007 .094 .483 .447 .352 .278 .192 .099 .048 .004 .079 .434 .760 .390 .303 .197 .097 .048 .004 .034 .419 .935 .804 .769 .710 .599 .473 .221 .499 .914 .915 .787 .756 .695 .582 .455 .180 .432 .876 .799 .744 .714 .439 .485 .357 .125 .168 .493 NXWC40W .628 .378 .691 .400 .849 .404 .997 .995 .995 .960 .987 .941 NxWC40W .636 .335 .493 .392 .908 .398 .984 .971 .982 .932 .928 .894 NKWC40W .421 .333 .480 .358 .812 .405 •Woo »T*J .967 .902 .909 .841 MxW C40W .421 .331 .475 .359 .750 .381 .952 .892 .955 .874 .813 .758 C30WC20W .304 .209 .308 .205 .300 .218 .991 .980 .947 .928 .917 .876 C30WC20W .252 .179 .305 .194 .306 .206 .955 .940 .912 .873 .861 .787 C30WC20W .262 .179 .281 .200 .311 .210 .925 .890 .868 .822 .805 .739 C30WC20W .268 .202 .291 .201 .288 .212 .869 .820 .849 .801 .725 .665 ClOW .108 .100 .103 .961 .871 .781 ClOW .094 .096 .111 .897 .807 .675 ClOW .082 .107 .092 .824 .733 .611 C10W .103 .113 .114 .750 .705 .510 COSW COlW .050 .008 .047 .009 .038 .005 .925 .753 .790 .566 .690 .394 COSW COlW .044 .006 .046 .004 .049 .002 .837 .581 .717 .423 .478 .201 COSW COlW .042 .004 .054 .007 .045 .006 .739 .467 .627 .338 .451 .177 COSW COlW .045 .005 .053'-.004 .052 .012 .652 .337 .565 .256 .360 .111 LfW loW .217 .870 .290 .950 .187 .794 .982 1.00 .875 .997 .607 .924 LfW LoW .170 .868 .193 .872 .090 .600 .935 .998 .799".96B' .297 .731 LfW loW .144 .845 .160 .822 .061 .539 .877 .996 .732 .974 .282 .716 LfW LOW .130 .820 .123 .746 .045 .399 .791 .996 .657 .968 .167 .524 Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating further. i00 —i C.H Ml NIX 0.5 .00 0.5 .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Ml NIX 0.5 .00 0.5 .50 2.0 .00 2.0 .50 Ml NIX 0.3 .00 0.5 .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 Ml NIX 0. .00 0. .50 0. .85 2. .00 2. .50 2.0 .85 Ml NIX 0.5 .00 0.5 .50 0.5 .85 2.0 .00 2.0 .50 2.0 .85 Nxl C40L .543 .306 .611 .302 .721 .351 .973 .980 .976 .898 .810 .745 HxL MOL .035 .336 .026 .374 .835 1.00 .887 .994 NxL MOL .070 .331 .091 .354 .055 .385 •788 «TTT .855 .976 .884 .939 Nxl MOL .093 .295 .102 .319 .129 .401 .764 .994 .832 .966 .893 .888 NxL MOL .107 .321 .119 .321 .195 .364 .762 .987 .791 .947 .819 .826 C30L .249 .250 .280 •9oo .869 .714 C30L .261 .280 1.00 .989 C30L .255 .292 .293 .997 .964 .9lf C30L .240 .241 .303 .992 .959 .833 C30L .249 .241 .279 .979 .920 .782 cm CIOL .184 .112 .168 .087 .185 .095 .944 .900 .823 .738 .642 .493 C20L CIOL .176 .088 .211 .113 1.00 .997 .978 .943 cm cioi .188 .100 .206 .120 .197 .117 .993 .983 .947 .890 .864 .742 cm CIOL .172 .096 .172 .097 .179 .076 .982 .963 .935 .846 .751 .525 cm CIOL .176 .090 .165 .076 .187 .087 .971 .943 .878 .784 .710 .549 COSL C01L .056 .003 .042 .005 .045 .004 .841 .598 .634 .352 .346 .106 COSL COIL .045 .018 .073 .033 .991 .890 .845 .631 COSL €011 .059 .029 .066 .024 .074 .036 .964 .762 .791 .512 .560 .283 COSL C01I .047 .010 .050 .021 .050 .038 .904 .653 .735 .422 .327 .129 COSL COIL .041 .013 .039 .014 .036 .010 .872 .586 .664 .340 .355 .121 LfL .108 .144 .025 .905 .787 .158 Lfl .084 .156 .997 .989 Lfl .106 .161 .339 .983 .960 .900 LfL .101 .144 .394 .964 .928 .668 Lfl .098 .134 .328 .948 .890 • OoO LoL .576 .858 .383 .957 .994 .533 LoL .687 .852 .960 .990 LoL .753 .889 .939 .966 .985 .987 LoL .769 .918 .914 .957 .991 .922 lol .844 .918 .927 .964 .990 .928- NxG C40G C30G C20G C10G C050 COlO IfO LoO NxW C40W C30W C20W C10W COSW COIW LfW LoW .652 .327 .262 .190 .095 .038 .003 .058 .576 .629 .311 .264 .173 .086 .035 .002 .108 .663 .336 .273 .199 .100 .044 .006 .064 .555 .647 .314 .245 .179 .086 .037 .005 .094 .705 .700 .346 .279 .189 .093 .037 .003 .018 .359 .707 .370 .296 .221 .095 .046 .011 .029 .356 .897 .753 .709 .649 .535 .399 .127 .372 .866 .954 .857 .820 .766 .682 .574 .275 .722 .987 .896 .742 .695 .635 .513 .393 .144 .342 .836 .932 .825 .785 .718 .606 .491 .195 .560 .951 .784 .700 .663 .611 .456 .304 .099 .138 .504 .793 .708 .672 .611 .466 .329 .099 .155 NxG C40G C30G C20G C10G C05G C01G LfG loG NxW MOW C30W C20W C10W COSW C01W LfW LoW .029 .403 .307 .221 .122 .082 .043 .126 .814 .030 .377 .291 .204 .095 .059 .031 .107 .797 .019 .390 .295 .204 .125 .073 .039 .177 .868 .013 .398 .301 .202 .117 .076 .042 .178 .850 .857 .994 .988 .984 .966 .924 .674 .977 .983 .840 .996 .994 .993 .980 .946 .739 .983 .886 .982 .976 .958 .909 .823 .572 .974 .994 .891 .989 .979 .955 .897 .795 .540 .965 .9 .... —— .. c«4 N-4 CV«2.0 — - —— ..—. ——— • —— - —— ...................... —— ........... NxG C40G C30G C20G C10G COSG C01G LfG LoG NxW C40W C30W C20W C10W COSW COIW LfW LoW .089 .383 .292 .183 .103 .058 .020 .136 .931 .072 .381 .288 .195 .104 .054 .015 .126 .871 .089 .385 .303 .205 .110 .064 .029 .187 .930 .067 .364 .277 .187 .097 .052 .026 .059 .394 .300 .205 .107 .074 .042 .358 .932 .056 .379 .268 .183 .104 .065 .030 .312 .947 .820 .964 .952 .931 .860 .737 .436 .905 .994 .821 .983 .979 .966 .923 .858 .568 .939 .982 .870 .967 .953 .923 .837 .683 .360 .928 .996 .839 .974 .963 .927 .832 .704 .408 .885 .954 .924 .861 .723 .543 .267 .914 .991 .889 .933 .904 .846 .695 .518 .243 .87 . ————— c«4 M«4 CV«2.5 —. ——— -.'« — - —— ->.- — ...... ..„"..--.... — ......;..........- .NxG C40G C30G C20G C10G COSG COlG LfG LoG NxW MOW C30W C20W C10W COSW COIW LfW LoW .133 .354 .283 .194 .085 .043 .018 .162 .963 .117 .332 .267 .181 .091 .045 .015 .112 .898 .136 .362 .260 .180 .094 .050 .015 .201 .967 .128 .342 .255 .170 .078 .038 .016 .153 .956 .121 .396 .283 .174 .080 .058 .046 .415 .907 .126 .378 .274 .160 .081 .046 .036 .371 .814 .945 .910 .868 .771 .611 .296 .836 .991 .797 .969 .952 .927 .868 .762 .427 .889 .98 .808 .915 .883 .832 .720 .573 .268 .842 .988 .810 .925 .902 .855 .754 .610 .279 .857 .993 .886 .886 ,845 .757 .538 .355 .143 .677 .917 .880 .882 .836 .761 .554 .339 .140 .683 .913 .. ——— .. c«4 N«4 CV-3.0 -.— - — - ———————— ............."... —— .................. — ... NxG C40G C30G C20G C10G COSG COlG LfG LoG NxW C40W C30W C20W C10W COSW COIW LfW LoW .192 .371 .277 .181 .077 .033 .009 .194 .975 .149 .328 .258 .178 .090 .041 .014 .122 . .188 .351 .273 .180 .082 .036 .011 .197 :977 .177 .349 .270 .175 .082 .639 .009 .176 .95 .211 .404 .317 .212 .096 .052 .017 .351 .918 .212 .368 .298 .209 .084 .037 .016 .327 .9 .762 .875 .840 .783 .656 .462 .171 .761 .995 .749 .933 .909 .864 .777 .628 .306 .8 .801 .887 .853 .793 .668 .491 .163 .810 .997 .762 .897 ,863 .817 .697 .511 .211 .820 .824 .833 .796 .731 .568 .342 .111 .690 .929 .828 .842 .797 .738 .552 .332 .109 .696 Appendix I. SSL Simulation Results: estimated Probabilities of Investigating Further. u Ml NIX O.S .00 O.S .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Ml NIX O.S .00 O.S .50 0.5 .90 2.0 .00 2.0 .50 2.0 .90 Ml NIX O.S .00 O.S .50 2.0 .00 2.0 .50 Ml NIX O.S .00 O.S .50 0.5 .75 2.0 .00 2.0 .50 2.0 .75 Ml NIX 0.5 .00 O.S .50 0:5 .85 2.0 .00 2.0 .50 2.0 .85 NxL C40L C30L C201 C10L C05L COIL .116 .279 .220 .154 .082 .038 .008 .140 .308 .239 .166 .083 .042 .009 .260 .371 .289 .188 .083 .038 .013 .749 .979 .974 .952 .902 .829 .517 .753 .928 .901 .857 .756 .629 .306 .784 .780 .744 .675 .510 .278 .074 Nxl C40L C30L C20L C10L COSl COIL .125 .263 .210 .148 .079 .038 .007 .162 .303 .240 .173 .085 .039 .008 .297 .371 .293 .196 .070 .024 .007 .711 .966 .953 .929 .856 .768 .441 .737 .907 .871 .825 .728 .559 .238 .721 .718 .684 .626 .442 .244 .056 Nxl C40L C30L C20L C10L COSl COIL .048 .336 .262 .185 .092 .039 .013 .047 .401 .309 .209 .120 .060 .028 .928 1.00 1.00 1.00 1.00 .999 .993 .964 .999 .998 .995 .981 .947 .803 Nxl C40L C30L C20L ClOi COSL COIL .091 .368 .288 .193 .105 .058 .020 .089 .378 .293 .205 .108 .049 .011 .075 .390 .298 .209 .114 .066 .026 .906 1.00 1.00 1.00 .998 .996 .966 .937 .994 .992 .982 .965 .916 .715 .952 .967 .951 ,924 .848 .753 .429 » NxL C40L C30L C20L C10L COSL COIL .134 .347 .274 .187 .109 .062 .009 .186 .383 .293 .210 .097 .048 .010 .175 .363 .264 .163 .087 .050 .023 .889 .997 .997 .997 .990 .983 .891 .911 .989 .983 .963 .936 .887 .642 .980 .941 .913 .854 .710 .534 .204 Lfl .091 .134 .315 .909 .853 .574 Lfl .090 .146 .276 .865 .814 .544 Lfl .087 .205 1.00 .997 LfL .111 .174 .457 .999 .995 .941 ifl .119 .176 .406 .993 .977 .837 Lol .834 .934 .806 .970 .988 .826 Lol .849 .945 .823 .959 .990 .814 Lol .315 .685 .965 .992 Lol .481 .805 .944 .963 .992 .998 lot .554 .844 .970 .968 .997 .984 . —————— c«4 ft.* CV«3.5 •• —————— ......... ————— ......... — ... —— .................. NxG C400 C30G C20G C10G COSO C010 LfG LoQ NxW C4W C30W C20W ClOU COSW COlW LfW loW .242 .376 .295 .196 .090 .033 .008 .219 .975 .171 .303 .246 .174 .082 .038 .008 .126 .932 .236 .362 .287 .201 .097 .042 .009 .239 .978 .183 .327 .255 .168 .083 .038 .010 .165 .980 .291 .403 .304. …