United States
Environmental Protection
Agency
Office of Solid Waste and
Emergency Response
Washington, DC 20460
EPA/540/R95/128
May 1996
&EPA
Superfund
Soil Screening Guidance:
Technical Background
Document
Second Edition
-------
Publication 9355.4-17A
May 1996
Soil Screening Guidance:
Technical Background Document
Second Edition
Office of Emergency and Remedial Response
U.S. Environmental Protection Agency
Washington, DC 20460
-------
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 rulemaking. 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.
-------
TABLE OF CONTENTS
Section Page
Disclaimer ii
List of Tables vi
List of Figures viii
List of Highlights viii
Preface 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 Organization of the Document 6
Part 2: Development of Pathway-Specific Soil Screening Levels
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 Ingestion 18
2.3 Dermal Absorption 20
2.4 Inhalation of Volatiles and Fugitive Dusts 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 Particulate 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 Dilution/Attenuation Factor Development 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-Limit Model 58
2.6.3 Inhalation Mass-Limit Model 60
2.7 Plant Uptake 61
2.8 Intrusion of Volatiles into Basements: Johnson and Ettinger Model 62
ill
-------
TABLE OF CONTENTS (continued)
Section Page
Part 3: Models for Detailed Assessment
3.1 Inhalation of Volatiles: 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 Unsaturated Zone Models 68
Part 4: Measuring Contaminant Concentrations in Soil
4.1 Sampling Surface Soils 82
4.1.1 State the Problem 82
4.1.2 Identify the Decision 82
4.1.3 Identify Inputs to the Decision 84
4.1.4 Define the Study Boundaries 84
4.1.5 Develop a Decision Rule 85
4.1.6 Specify Limits on Decision Errors for the Max Test 86
4.1.7 Optimize the Design for the Max Test 87
4.1.8 Using the DQA Process: Analyzing Max Test Data 96
4.1.9 Specify Limits on Decision Errors for Chen Test 99
4.1.10 Optimize the Design Using the Chen Test 100
4.1.11 Using the DQA Process: Analyzing Chen Test Data 107
4.1.12 Special Considerations for Multiple Contaminants 107
4.1.13 Quality Assurance/Quality Control Requirements 107
4.1.14 Final Analysis 109
4.1.15 Reporting 109
4.2 Sampling Subsurface Soils 110
4.2.1 State the Problem 110
4.2.2 Identify the Decision 110
4.2.3 Identify Inputs to the Decision 110
4.2.4 Define the Study Boundaries 114
4.2.5 Develop a Decision Rule 114
4.2.6 Specify Limits on Decision Errors 114
4.2.7 Optimize the Design 115
4.2.8 Analyzing the Data 116
4.2.9 Reporting 116
4.3 Basis for the Surface Soil Sampling Strategies: Technical Analyses Performed 117
4.3.1 1994 Draft Guidance Sampling Strategy 117
4.3.2 Test of Proportion Exceeding a Threshold 119
4.3.3 Relative Performance of Land, Max, and Chen Tests 121
4.3.4 Treatment of Observations Below the Limit of Quantitation 127
4.3.5 Multiple Hypothesis Testing Considerations 127
4.3.6 Investigation of Compositing Within EA Sectors 129
IV
-------
TABLE OF CONTENTS (continued)
Section Page
Part 5: Chemical-Specific Parameters
5.1 Solubility, Henry's Law Constant, and Kow 133
5.2 Air (Di>a) and Water (Di)W) Diffusivities 133
5.3 Soil Organic Carbon/Water Partition Coefficients (Koc) 139
5.3.1 Koc for Nonionizing Organic Compounds 139
5.3.2 Koc for Ionizing Organic Compounds 145
5.4 Soil-Water Distribution Coefficients (K(j) for Inorganic Constituents 149
5.4.1 Modeling Scope and Approach 152
5.4.2 Input Parameters 153
5.4.3 Assumptions and 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
B Route-to-Route Extrapolation of Inhalation Benchmarks 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-1
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 J-l
K Soil Organic Carbon (Koc) /Water (Kow) Partition Coefficients K-l
L Koc Values for Ionizing Organics as a Function of pH L-l
M Response to Peer-Review Comments on MINTEQA2 Model Results M-l
-------
LIST OF TABLES
Table 1. Regulatory and Human Health Benchmarks Used for SSL Development 10
Table 2. SSL Chemicals with Noncarcinogenic 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 Csat 29
Table 4. Physical State of Organic SSL Chemicals 30
Table 5. Variation of DAF with Size of Source Area for SSL EPACMTP Modeling Effort 48
Table 6. Recharge Estimates for DNAPL Site Hydrogeologic Regions 50
Table 7. SSL Dilution 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 RITZ Model 67
Table 11. Input Parameters Required for VIP 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,
Given Expected 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
-------
LIST OF TABLES (continued)
Table 33. Error Rates of Max Test and Chen Test at .2 (C20) and .1 (CIO)
Significance Level for CV = 2, 2.5, 3, 3.5, C = # of Specimens per Composite,
N = # of Composite Samples 124
Table 34. 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 Exposvire Areas 129
Table 36. Chemical-Specific Properties Used in SSL Calculations 132
Table 37. Air Diffusivity (D; a) and Water Diffusivity (D; w) Values for SSL Chemicals (25°C) 135
Table 38. Summary Statistics for Measured Koc Values: Nonionizing Organics 139
Table 39. Comparison of Measured and Calculated Koc Values 141
Table 40. Degree of lonization (Fraction of Neutral Species, F) as a Function of pH 145
Table 41. Soil Organic Carbon/Water Partition Coefficients and pKa Values for Ionizing
Organic Compounds 147
Table 42. Predicted Soil Organic Carbon/Water Partition Coefficients (KoC,L/kg) as a
Function of pH: Ionizing Organics 148
Table 43. Summary of Collected IQ Values Reported in Literature 149
Table 44. Summary of Geochemical Parameters Used in SSL MINTEQ Modeling Effort 151
Table 45. Background Pore-Water Chemistry Assumed for SSL MINTEQ Modeling Effort 152
Table 46. Estimated Inorganic IQ Values for SSL Application 154
vn
-------
LIST OF FIGURES
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Conceptual Risk Management Spectrum for Contaminated Soil
Exposure Pathways Addressed by SSLs
Migration to ground water pathway — EPACMTP modeling effort
The Data Quality Objectives process
Design performance goal diagram
Systematic (square grid points) sample with systematic compositing scheme
(6 composite samples consisting of 4 specimens)
Systematic (square grid points) sample with random compositing scheme
(6 composite samples consisting of 4 specimens)
Stratified random sample with random compositing scheme
(6 composite samples consisting of 4 specimens)
U.S. Department of Agriculture soil texture classification
Empirical pH-dependent adsorption relationship: arsenic (+3),
chromium (+6), selenium, thallium
Metal K(j as a function of pH
2
4
46
80
87
90
91
92
112
155
156
LIST OF HIGHLIGHTS
Highlight 1-
Highlight 2-
Highlight3:
Highlight 4:
Highlight 5:
Highlight 6:
Highlight 7:
Key Attributes of the Soil Screening Guidance
Simplifying Assumptions for the Migration to Ground Water Pathway
Procedure for Compositing of Specimens from a Grid Sample
Using a Systematic Scheme (Figure 6)
Procedure for Compositing of Specimens from a Grid Sample
Using a Random Scheme (Figure 7)
Procedure for Compositing of Specimens from a Stratified Random Sample
Using a Random Scheme (Figure 8)
Directions for Data Quality Assessment for the Max Test
Directions for the Chen Test Usine Sirrrole Random Sanrole Scheme
3
34
90
91
92
96
106
Vlll
-------
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 soil 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 detailed models that may be used to develop
site-specific SSLs. Part 4 addresses sampling schemes for measuring soil contaminant levels during the soil
screening process. Part 5 provides technical background on the determination of chemical-specific properties for
calculating SSLs.
IX
-------
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 modeling effort for the inhalation pathway
and provided background information on that effort. Dr. Zubair Saleem 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«Allen & Hamilton.
-------
Part 1: INTRODUCTION
This document provides the technical background for the Soil Screening Guidance. 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 on the National
Priorities List (NPL) with anticipated future residential land use scenarios.1 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. "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 more 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 the 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 Superfimd program defines "soil" as having a particle size under 2 millimeters, while the RCRA program
allows for particles under 9 millimeters in size.
-------
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 site-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, site 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. EPA created the Soil Screening Guidance to be
consistent with and to enhance current Superfund processes.
The process presented in this guidance to develop and apply simple, site-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 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
-------
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 that are specific to the user's site.
Applying site-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 site-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 site-specific cleanup
levels unless a site-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 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.
1.3 Scope of Soil Screening Guidance
The Soil Screening Guidance incorporates readily obtainable site data into simple, standardized
equations to derive site-specific screening levels for selected contaminants and exposure pathways.
Key attributes of the Soil Screening Guidance are given in Highlight 1.
-------
Highlight 1: Key Attributes of the Soil Screening 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 site-specific basis using mass-limit models.
• Parameters are identified for which site-specific information is needed to develop site-specific
SSLs.
• Default values are provided to calculate generic SSLs where site-specific information is not
available.
• SSLs are generally based on a 10-6 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 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 Part 2 for further discussion).
Direct Ingestion
of Ground
Water and Soil
Inhalation
Blowing^
Dust and
Volatization
Also Addressed:
• Plant Uptake
• Dermal Absorption
Figure 2. Exposure Pathways Addressed
by SSLs.
-------
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 with 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, 1991a). The default intake and duration assumptions presented in U.S. EPA
(1991a) 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, 1995b) 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-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) (1O6 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 that setting a 10-6 risk level for individual chemicals and pathways will
generally lead to cumulative risks within the risk range (1O4 to 10-6) for the combinations of
5
-------
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 it to be unlikely that such data will be available from the limited subsurface sampling
that is done to apply SSLs. However, EPA also recognizes that 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, this 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 that no biological
or chemical degradation occurs in the soil or in the aquifer. Again, models capable of addressing
heterogeneities or degradation processes require collection of site-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
Part 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 site-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 site data are available or can be obtained.
-------
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 unsaturated 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.
-------
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 Organization of the Document 6
-------
Figure 1. Conceptual Risk Management Spectrum for Contaminated Soi 2
Figure 2. Exposure Pathways Addressed
by SSLs 4
-------
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 2.2)
• 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 assuming 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
volatiles 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.
-------
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, 1995b) 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 1O6 risk level for carcinogens and a hazard quotient of 1 for noncarcinogens. For
carcinogens, EPA believes that setting a 1O6 risk level for individual chemicals and pathways
generally will lead to cumulative risks within the 104 to 106 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.
-------
Table 1. Regulatory and Human Health Benchmarks Used for SSL Development
Number Chemical Name
83-32-9 Acenaphthene
67-64-1 Acetone (2-Propanone)
309-00-2 Aldrin
120-12-7 Anthracene
7440-36-0 Antimony
7440-38-2 Arsenic
7440-39-3 Barium
56-55-3 Benz(a (anthracene
71-43-2 Benzene
205-99-2 Benzo(fa (fluoranthene
207-08-9 Benzo(fc (fluoranthene
65-85-0 Benzoic acid
50-32-8 Benzo(a (pyrene
7440-41-7 Beryllium
1 1 1 -44-4 Bis(2-chloroethyl)ether
1 1 7-81-7 Bis(2-ethylhexyl)phthalate
75-27-4 Bromodichloromethane
75-25-2 Bromoform (tribromomethane)
71-36-3 Butanol
85-68-7 Butyl benzyl phthalate
7440-43-9 Cadmium
86-74-8 Carbazole
75-15-0 Carbon disulfide
56-23-5 Carbon tetrachloride
57-74-9 Chlordane
106-47-8 p -Chloroaniline
108-90-7 Chlorobenzene
124-48-1 Chlorodibromomethane
67-66-3 Chloroform
95-57-8 2-Chlorophenol
Maximum
Contaminant Level
Goal
(mg/L)
MCLG ,
(PMCLG) Ref-
6.0E-03 3
2.0E+00 3
4.0E-03 3
5.0E-03 3
1.0E-01 3
6.0E-02 3
Maximum
Contaminant Level
(mg/L)
MCL (PMCL) Ref. "
6.0E-03 3
5.0E-02 3
2.0E+00 3
5.0E-03 3
2.0E-04 3
4.0E-03 3
6.0E-03 3
1.0E-01* 3
1.0E-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 Based
Limits
(mg/L)
HBL " Basis
2E+00 RfD
4E+00 RfD
5E-06 SFD
1E+01 RfD
1 E-04 SFD
1 E-04 SF0
1 E-03 SFD
1E+02 RfD
8E-05 SF0
4E+00 RfD
7E+00 RfD
4E-03 SF0
4E+00 RfD
1E-01 RfD
2E-01 RfD
Cancer Slope Factor
(mg/kg-d)-1
<££• SF° ™-a
D
B2 1.7E+01 1
D
A 1.5E+00 1
B2 7.3E-01 4
A 2.9E-02 1
B2 7.3E-01 4
B2 7.3E-02 4
B2 7.3E+00 1
B2 4.3E+00 1
B2 1.1E+00 1
B2 1.4E-02 1
B2 6.2E-02 1
B2 7.9E-03 1
D
C
B2 2.0E-02 2
B2 1.3E-01 1
B2 1.3E+00 1
D
C 8.4E-02 1
B2 6.1 E-03 1
Unit Risk Factor
(ug/m3)-1
aaTs* URF «-••
D
B2 4.9E-03 1
D
A 4.3E-03 1
B2
A 8.3E-06 1
B2
B2
B2
B2 2.4E-03 1
B2 3.3E-04 1
B2
B2
B2 1.1E-06 1
D
C
B1 1.8E-03 1
B2 1.5E-05 1
B2 3.7E-04 1
D
C
B2 2.3E-05 1
Reference Dose
(mg/kg-d)
RfD Ref. a
6.0E-02 1
1.0E-01 1
3.0E-05 1
3.0E-01 1
4.0E-04 1
3.0E-04 1
7.0E-02 1
4.0E+00 1
5.0E-03 1
2.0E-02 1
2.0E-02 1
2.0E-02 1
1.0E-01 1
2.0E-01 1
1.0E-03** 1
1.0E-01 1
7.0E-04 1
6.0E-05 1
4.0E-03 1
2.0E-02 1
2.0E-02 1
1.0E-02 1
5.0E-03 1
Reference
Concentration
(mg/m3)
RfC Ref. a
5.0E-04 2
7.0E-01 1
2.0E-02 2
"Proposed MCL = 0.08 mg/L, Drinking Water Regulations and Health Advisories , U.S. EPA (1995).
** Cadmium RfD is based on dietary exposure.
-------
Table 1 (continued)
Number Chemical Name
7440-47-3 Chromium
16065-83-1 Chromium (III)
18540-29-9 Chromium (VI)
218-01-9 Chrysene
57-12-5 Cyanide (amenable)
72-54-8 ODD
72-55-9 DDE
50-29-3 DDT
53-70-3 Dibenz(a,h (anthracene
84-74-2 Di-fl -butyl phthalate
95-50-1 1,2-Dichlorobenzene
106-46-7 1,4-Dichlorobenzene
91-94-1 3,3-Dichlorobenzidine
75-34-3 1,1-Dichloroethane
107-06-2 1,2-Dichloroethane
75-35-4 1,1-Dichloroethylene
156-59-2 cis -1 ,2-Dichloroethylene
156-60-5 trans -1 ,2-Dichloroethylene
120-83-2 2,4-Dichlorophenol
78-87-5 1,2-Dichloropropane
542-75-6 1,3-Dichloropropene
60-57-1 Dieldrin
84-66-2 Diethylphthalate
105-67-9 2,4-Dimethylphenol
51-28-5 2,4-Dinitrophenol
121-14-2 2,4-Dinitrotoluene**
606-20-2 2,6-Dinitrotoluene**
1 1 7-84-0 Di-fl -octyl phthalate
115-29-7 Endosulfan
72-20-8 Endrin
Maximum
Contaminant Level
Goal
(mg/L)
MCLG .
(PMCLG) Ret'
1.0E-01 3
(2.0E-01) 3
6.0E-01 3
7.5E-02 3
7.0E-03 3
7.0E-02 3
1.0E-01 3
2.0E-03 3
Maximum
Contaminant Level
(mg/L)
MCL(PMCL) Ref.a
1.0E-01 3
1.0E-01 3*
(2.0E-01) 3
6.0E-01 3
7.5E-02 3
5.0E-03 3
7.0E-03 3
7.0E-02 3
1.0E-01 3
5.0E-03 3
2.0E-03 3
Water Health Based
Limits
(mg/L)
HBL " Basis
4E+01 RfD
1E-02 SF0
4E-04 SF0
3E-04 SF0
3E-04 SFD
1E-05 SF0
4E+00 RfD
2E-04 SF0
4E+00 RfD
1E-01 RfD
5E-04 SFD
5E-06 SF0
3E+01 RfD
7E-01 RfD
4E-02 RfD
1 E-04 SF0
1 E-04 SFD
7E-01 RfD
2E-01 RfD
Cancer Slope Factor
(mg/kg-d)-1
££• SF- «••
A
A
B2 7.3E-03 4
D
B2 2.4E-01 1
B2 3.4E-01 1
B2 3.4E-01 1
B2 7.3E+00 4
D
D
B2 2.4E-02 2
B2 4.5E-01 1
C
B2 9.1E-02 1
C 6.0E-01 1
D
B2 6.8E-02 2
B2 1.8E-01 2
B2 1.6E+01 1
D
B2 6.8E-01 1
B2 6.8E-01 1
D
Unit Risk Factor
(ug/m3)-1
aaTs' URF ™-'
A 1.2E-02 1
A 1.2E-02 1
D
B2
B2
B2 9.7E-05 1
B2
D
D
B2
B2
C
B2 2.6E-05 1
C 5.0E-05 1
D
B2
B2 3.7E-05 2
B2 4.6E-03 1
D
D
Reference Dose
(mg/kg-d)
RfD Ref . a
5.0E-03 1
1.0E+00 1
5.0E-03 1
2.0E-02 1
5.0E-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 1
5.0E-05 1
8.0E-01 1
2.0E-02 1
2.0E-03 1
2.0E-03 1
1.0E-03 2
2.0E-02 2
6.0E-03 2
3.0E-04 1
Reference
Concentration
(mg/m3)
RfC Ref. a
2.0E-01 2
8.0E-01 1
5.0E-01 2
4.0E-03 1
2.0E-02 1
* MCL for total chromium is based on Cr (VI) toxicity.
* Cancer Slope Factor is for 2,4-, 2,6-Dinitrotoluene mixture.
-------
Table 1 (continued)
Number Chemical Name
100-41-4 Ethylbenzene
206-44-0 Fluoranthene
86-73-7 Fluorene
76-44-8 Heptachlor
1024-57-3 Heptachlor epoxide
118-74-1 Hexachlorobenzene
87-68-3 Hexachloro-1,3-butadiene
319-84-6 a-HCH (a-BHC)
319-85-7 p-HCH (p-BHC)
58-89-9 y-HCH (Lindane)
77-47-4 Hexachlorocyclopentadiene
67-72-1 Hexachloroethane
193-39-5 lndeno(1,2,3-cc/)pyrene
78-59-1 Isophorone
7439-97-6 Mercury
72-43-5 Methoxychlor
74-83-9 Methyl bromide
75-09-2 Methylene chloride
95-48-7 2-Methylphenol (o -cresol)
91-20-3 Naphthalene
7440-02-0 Nickel
98-95-3 Nitrobenzene
86-30-6 N -Nitrosodiphenylamine
621-64-7 N -Nitrosodi-n -propylamine
87-86-5 Pentachlorophenol
108-95-2 Phenol
129-00-0 Pyrene
7782-49-2 Selenium
7440-22-4 Silver
100-42-5 Styrene
79-34-5 1,1,2,2-Tetrachloroethane
Maximum
Contaminant Level
Goal
(mg/L)
MCLG _ f ,
(PMCLG) Ret'
7.0E-01 3
1.0E-03 3
2.0E-04 3
5.0E-02 3
2.0E-03 3
4.0E-02 3
5.0E-02 3
1.0E-01 3
Maximum
Contaminant Level
(mg/L)
MCL(PMCL) Ref.a
7.0E-01 3
4.0E-04 3
2.0E-04 3
1.0E-03 3
2.0E-04 3
5.0E-02 3
2.0E-03 3
4.0E-02 3
5.0E-03 3
1.0E-03 3
5.0E-02 3
1.0E-01 3
Water Health Based
Limits
(mg/L)
HBL " Basis
1E+00 RfD
1E+00 RfD
1 E-03 SFD
1E-05 SFD
5E-05 SF0
6E-03 SF0
1 E-04 SFD
9E-02 SFD
5E-02 RfD
2E+00 RfD
1E+00 RfD
1E-01 HA*
2E-02 RfD
2E-02 SFD
1E-05 SF0
2E+01 RfD
1E+00 RfD
2E-01 RfD
4E-04 SFD
Cancer Slope Factor
(mg/kg-d)-1
££• SF- «••
D
D
D
B2 4.5E+00 1
B2 9.1E+00 1
B2 1.6E+00 1
C 7.8E-02 1
B2 6.3E+00 1
C 1.8E+00 1
B2 1.3E+00 2
D
C 1.4E-02 1
B2 7.3E-01 4
C 9.5E-04 1
D
D
D
B2 7.5E-03 1
C
D
A
D
B2 4.9E-03 1
B2 7.0E+00 1
B2 1.2E-01 1
D
D
D
D
C 2.0E-01 1
Unit Risk Factor
(ug/m3)-1
S URF «••
D
D
B2 1.3E-03 1
B2 2.6E-03 1
B2 4.6E-04 1
C 2.2E-05 1
B2 1.8E-03 1
C 5.3E-04 1
C
D
C 4.0E-06 1
B2
C
D
D
D
B2 4.7E-07 1
C
D
A 2.4E-04 1
D
B2
B2
B2
D
D
D
D
C 5.8E-05 1
Reference Dose
(mg/kg-d)
RfD Ref . a
1.0E-01 1
4.0E-02 1
4.0E-02 1
5.0E-04 1
1.3E-05 1
8.0E-04 1
2.0E-04 2
3.0E-04 1
7.0E-03 1
1.0E-03 1
2.0E-01 1
3.0E-04 2
5.0E-03 1
1.4E-03 1
6.0E-02 1
5.0E-02 1
4.0E-02 6
2.0E-02 1
5.0E-04 1
3.0E-02 1
6.0E-01 1
3.0E-02 1
5.0E-03 1
5.0E-03 1
2.0E-01 1
Reference
Concentration
(mg/m3)
RfC Ref. a
1.0E+00 1
7.0E-05 2
3.0E-04 2
5.0E-03 1
3.0E+00 2
2.0E-03 2
1.0E+00 1
* Health advisory for nickel (MCL is currently remanded); EPA Office of Science and Technology, 7/10/95.
-------
Table 1 (continued)
Number Chemical Name
127-18-4 Tetrachloroethylene
7440-28-0 Thallium
108-88-3 Toluene
8001-35-2 Toxaphene
120-82-1 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
7440-62-2 Vanadium
108-05-4 Vinyl acetate
75-01-4 Vinyl chloride (chloroethene)
108-38-3 m -Xylene
95-47-6 o -Xylene
106-42-3 p -Xylene
7440-66-6 Zinc
Maximum
Contaminant Level
Goal
(mg/L)
MCLG ,
(PMCLG) KeT'
5.0E-04 3
1.0E+00 3
7.0E-02 3
2.0E-01 3
3.0E-03 3
zero 3
1.0E+01 3*
1.0E+01 3*
1.0E+01 3*
Maximum
Contaminant Level
(mg/L)
MCL (PMCL) Ref. a
5.0E-03 3
2.0E-03 3
1.0E+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 Based
Limits
(mg/L)
HBL " Basis
4E+00 RfD
8E-03 SF0
3E-01 RfD
4E+01 RfD
1E+01 RfD
Cancer Slope Factor
(mg/kg-d)-1
<££• SF- Ref-a
5.2E-02 5
D
B2 1.1E+00 1
D
D
C 5.7E-02 1
1.1E-02 5
B2 1.1 E-02 1
A 1.9E+00 2
D
D
D
D
Unit Risk Factor
(ug/m3)-1
aaTs' URF «*••
5.8E-07 5
D
B2 3.2E-04 1
D
D
C 1.6E-05 1
1.7E-06 5
B2 3. 1 E-06 1
A 8.4E-05 2
D
D
D
D
Reference Dose
(mg/kg-d)
RfD Ref. a
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
2.0E+00 1 **
3.0E-01 1
Reference
Concentration
(mg/m3)
RfC Ref. a
4.0E-01 1
2.0E-01 2
1.0E+00 5
2.0E-01 1
* MCL for total xylenes [1330-20-7] is 10 mg/L.
** RfD for total xylenes is 2 mg/kg-day.
' References: 1 = IRIS, U.S. EPA (1995b)
2 = HEAST, U.S. EPA(1995d)
3= U.S. EPA(1995a)
4 = OHEA, U.S. EPA(1993c)
5 = Interim toxicity criteria provided by Superfund
Health Risk Techincal Support Center,
Environmental Criteria Assessment Office
(ECAO), Cincinnati, OH (1994)
6 = ECAO, U.S. EPA (1994g)
7 = ECAO, U.S. EPA(1994f)
1 Health Based Limits calculated for 30-year exposure duration, 1 &B risk or hazard quotient = 1.
c Categorization of overall weight of evidence for human carcinogenicity:
Group A: human carcinogen
Group B: probable human carcinogen
B1: limited evidence from epidemiologic studies
B2: "sufficient" evidence from animal studies and "inadequate" evidence or
"no data" from epidemiologic studies
Group C: possible human carcinogen
Group D: not classifiable as to health carcinogenicity
Group E: evidence of noncarcinogenicity for humans
-------
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 (Csat)
above which these chemicals may occur as nonaqueous phase liquids (NAPLs) in soil (see Section
2.4.4). Therefore, the potential for additive effects must be carefully evaluated at every 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 Csat. 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 toxicologist 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 (e.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 site-related 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
-------
Table 2. SSL Chemicals with Noncarcinogenic Effects on Specific Target
Organ/System
Target Organ/System
Effect
Kidney
Acetone
1,1-Dichloroethane
Cadmium
Chlorobenzene
Di-n-octyl phthalate
Endosulfan
Ethylbenzene
Fluoranthene
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 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 liver-to-brain weight ratios
Histopathology
Increased weight; increased SCOT 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
-------
Table 2: (continued)
Target Organ/System
Effect
Circulatory System
Antimony
Barium
frans-1,2-Dichloroethene
c/s-1,2-Dichloroethylene
2,4-Dimethylphenol
Fluoranthene
Fluorene
Nitrobenzene
Styrene
Zinc
Reproductive System
Barium
Carbon disulfide
2-Chlorophenol
Methoxychlor
Phenol
Respiratory System
1,2-Dichloropropane
Hexachlorocyclopentadiene
Methyl bromide
Vinyl acetate
Gastrointestinal System
Hexachlorocyclopentadiene
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
Hematologic 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
-------
However, with the exception of vinyl chloride (which is gaseous at ambient temperatures), migration
to ground water SSLs are significantly lower than 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 site, 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 RfDorai). 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 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., 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 its results.
As part of this analysis, inhalation benchmarks were derived using simple route-to-route
extrapolation for 50 contaminants lacking inhalation benchmarks. A review of SSLs calculated from
these extrapolated benchmarks indicated that for 36 of the 50 contaminants, inhalation SSLs exceed
the soil saturation concentration (Csat), often by several orders of magnitude. Because maximum
volatile emissions occur at Csat (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 Csat have inhalation SSLs above generic SSLs for the migration to
ground water pathway (dilution attenuation factor [DAF] of 20). This suggests that migration to
ground water SSLs will be adequately protective of volatile inhalation risks at sites where ground
water is of concern.
At sites where ground water is not of concern (e.g., where ground water beneath or adjacent to the
site 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,
chlorodibromomethane, cw-l,2-dichloroethylene, and fra«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 site. However, the analysis in
Appendix B suggests that this 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 (|ig/m3)
17
-------
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 that 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 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 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 site.
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 B (U.S. EPA, 1991b). 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, Part B 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"
-------
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)
\-6
l/RfD0 x 10 kg/mg x EF x ED x IR
Parameter/Definition (units)
THQ/target hazard quotient (unitless)
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
6a
chemical-specific
350
6
200
a 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 50 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 (rng /kg) =
TR x AT x 365 d/yr
(2)
SFo x ID'6 kg/mg xEFxIFsoil/,
19
-------
Parameter/Definition (units)
Default
TR/target cancer risk (unitless)
AT/averaging time (yr)
SF0 /oral slope factor (mg/kg-d)-1
EF/exposure frequency (d/yr)
IFsoN/adj /age-adjusted soil ingestion factor (mg-yr/kg-d)
10-6
70
chemical-specific
350
114
Equation for Age-Adjusted Soil Ingestion Factor, IFSOji/adj
TF
11 soil/adj
(mg-yr/kg-d)
IR-soil/agel-6 X
agel-6
BW
- soil /age? -31
EL>
age7.31
agel.6
BW
age7.31
(3)
Parameter/Definition (units)
"%oii/adj /age-adjusted soil ingestion factor (mg-yr/kg-d)
IRsoii/age1.6/ingestion rate of soil age 1-6 (mg/d)
EDggei-e /exposure duration during ages 1-6 (yr)
IRsoii/age7-3i /ingestion rate of soil age 7-31 (mg/d)
EDage7-3i /exposure duration during ages 7-31 (yr)
BWggei-e /average body weight from ages 1-6 (kg)
BWage7_31 /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 site-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 half 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
-------
further investigation is needed. As adequate dermal 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. The
models and assumptions used to calculate SSLs for inhalation of volatiles are 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 |ig/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 (Csat), and the dispersion portion of the VF and particulate 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
-------
Screening Level Equation for Inhalation of Carcinogenic Contaminants in Residential
Soil
Volatile Screening Level
(mg/kg)
TR x AT x 365 d/yr
URF x l,000|lg/mg x EF x ED x
VF
(4)
Particulate Screening Level
(mg/kg)
TR x AT x 365 d/yr
URF x l,000|lg/mg x EF x ED x
PEF
Parameter/Definition (units)
TR/target cancer risk (unitless)
AT/averaging time (yr)
URF/inhalation unit risk factor (u,g/m3)-1
EF/exposure frequency (d/yr)
ED/exposure duration (yr)
VF/soil-to-air volatilization factor (m3/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. EPA, 1991b).
Screening Level Equation for Inhalation of Noncarcinogenic Contaminants in
Residential Soil
Volatile Screening Level
(mg/kg)
THQ x AT x 365 d/yr
EF x ED x (_L x —
V RfC VF
(5)
Particulate Screening Level
(mg/kg)
THQ x AT x 365 d/yr
EF x ED x ( 1 x _L)
V RfC PEF '
22
-------
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)
VF/soil-to-air volatilization factor (m3/kg)
PEF/particulate emission factor (m3/kg) (Equation 10)
Default
1
30
350
30
chemical-specific
chemical-specific
1.32x109
Source: RAGS HHEM, Part B (U.S. EPA, 1991b).
To calculate inhalation SSLs, the volatilization factor and particulate 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 Part 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 site measured values for these parameters.
Derivation of Volatilization Factor
,1/2
(6)
(3 14 x D x T)
VF(m3/kg) = Q/C x — — x I(r4(m2/cm2)
(2 xpb x DA)
where
PbKd + 6,
+ 6 H'
23
-------
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/m3)
T/exposure interval (s)
pb/dry soil bulk density (g/cm3)
9a /air-filled soil porosity (Lajr/LSOj|)
n/total soil porosity (Lpore/LSOj|)
9w/water-filled soil porosity (Lwater/LSoii)
ps/soil particle density (g/cm3)
Dj /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 foc
Koc/soil organic carbon-water partition coefficient (cm3/g)
foc/organic carbon content of soil (g/g)
68.81
9.5x108
1.5
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 (1991 b)
n-9w
1 - (Pb/Ps)
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 VF 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 B is based on the 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 (Farmer 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 VF 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 (De;).
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
-------
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:
where
ea =
n =
ea = n -
air-filled soil porosity (Lair/Lsoil)
total soil porosity (Lpore/Lsoil)
water-filled soil porosity (Lwater/Lsoji)
(7)
and
where
n = 1 - (Pb/ps)
(8)
pb = dry soil bulk density (g/cm3)
ps = 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 0W (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 ps (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
-------
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 site 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
algorithm performs well with onsite and near-field receptors. In addition, subdivision of the source is
not required for 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 particulate 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 that best represents a site's size and
meteorologic condition.
To develop a reasonably conservative default Q/C for calculating generic SSLs, a default site (Los
Angeles, CA) was chosen that best approximated the 90th percentile of the 29 normalized
concentrations (kg/m3 per g/m2-s). The inverse of this concentration results in a default VF Q/C
value of 68.81 g/m2-s per kg/m3 for a 0.5-acre site.
26
-------
Table 3. QIC Values by Source Area, City, and Climatic Zone
QIC (g/m2-s per kg/m3)
Zone 1
Seattle
Salem
Zone II
Fresno
Los Angeles
San Francisco
Zone III
Las Vegas
Phoenix
Albuquerque
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
Huntington
Harrisburg
Zone VIM
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
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
-------
2.4.4 Soil Saturation Limit. The soil saturation concentration (Csat) 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
e
(9)
Parameter/Definition (units)
CSat/soil saturation concentration (mg/kg)
S/solubility in water (mg/L-water)
pb/dry soil bulk 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)
9w/water-filled soil porosity (Lwater/LSoii)
H'/dimensionless Henry's law constant
H/Henry's law constant (atm-m3/mol)
Gg/air-filled soil porosity (Lajr/LSOj|)
n/total soil porosity (Lpore/LSOj|)
ps/soil particle density (kg/L)
Default
-
chemical-specific
1.5
KOC x foc (organics)
chemical-specific
0.006 (0.6%)
0.15
H x41, where 41 is a
conversion factor
chemical-specific
0.28
0.43
2.65
Source
see Part 5
U.S. EPA, 1991b
see Part 5
Carsel etal., 1988
EQ, 1994
U.S. EPA, 1991b
see Part 5
n-9w
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 B,
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 Csat 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 Csat). Thus, the
VF-based inhalation SSLs are applicable only if the soil concentration is at or below Csat. When
calculating volatile inhalation SSLs, Csat 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 Csat 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 is added to the soil. Table 3-A shows
that for compounds with generic volatile inhalation SSLs greater than Csat, the risks at Csat are
significantly below the screening risk of 1 x 1O6 and an HQ of 1. Since Csat corresponds to maximum
28
-------
volatile emissions, the inhalation route is not likely to be of concern for those chemicals with SSLs
exceeding Csat concentrations.
Table 3-A. Risk Levels Calculated at Csat for Contaminants that have
SSLjnh Values Greater than Csat
Chemical name
DDT
1,2-Dichlorobenzene
1,4-Dichlorobenzene
Ethylbenzene
(3-HCH (P-BHC)
Styrene
Toluene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
URF RfC
(u^g/m3)-1 (mg/m3)
9.7E-05
2.0E-01
8.0E-01
1.0E+00
5.3E-04
1.0E+00
4.0E-01
2.0E-01
1.0E+00
VF
(m3/kg)
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
Csat
(mg/kg)
4.0E+02
6.0E+02
2.8E+02
4.0E+02
2.0E+00
1.5E+03
6.5E+02
3.2E+03
1.2E+03
Carcinogenic
Risk
5.2E-07
_
_
3.4E-07
—
_
—
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 Csat for liquid compounds, the SSL is set at Csat. 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. EPA believes that further investigation is warranted when free nonaqueous phase liquids may
be present in soils at a site.
Table 4. Physical State of Organic SSL Chemicals
Compounds liquid at soil temperatures
CAS No. Chemical
67-64-1 Acetone
71-43-2 Benzene
117-81-7 Bis(2-ethylhexyl)phthalate
111-44-4 Bis(2-chloroethyl)ether
75-27-4 Bromodichloromethane
75-25-2 Bromoform
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
(°C)
-94.8
5.5
-55
-51.9
-57
8
-89.8
-35
-115
-23
-45.2
-20
-63.6
Compounds solid at soil 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(6)fluoranthene
207-08-9 Benzo(/c)fluoranthene
65-85-0 Benzole acid
86-74-8 Carbazole
57-74-9 Chlordane
106-47-8 p-Chloroaniline
218-01-9 Chrysene
72-54-8 ODD
Melting
Point
(°C)
93.4
104
215
84
176.5
168
217
122.4
246.2
106
72.5
258.2
109.5
29
-------
Table 4. (continued)
Compounds liquid at soil temperatures
Melting
CAS No.
Chemical
Point
Compounds solid at soil temperatures
Melting
CAS No. Chemical Point
95-57-8 2-Chlorophenol 9.8
84-74-2 Di-n-butyl phthalate -35
95-50-1 1,2-Dichlorobenzene -16.7
75-34-3 1,1-Dichloroethane -96.9
107-06-2 1,2-Dichloroethane -35.5
75-35-4 1,1-Dichloroethylene -122.5
156-59-2 c/s-1,2-Dichloroethylene -80
156-60-5 f/-ans-1,2-Dichloroethylene -49.8
78-87-5 1,2-Dichloropropane -70
542-75-6 1,3-Dichloropropene NA
84-66-2 Diethylphthalate -40.5
117-84-0 Di-n-octyl phthalate -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 Methylene 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
72-55-9 DDE 89
50-29-3 DDT 108.5
53-70-3 Dibenzo(a,/i)anthracene 269.5
106-46-7 1,4-Dichlorobenzene 52.7
91-94-1 3,3-Dichlorobenzidine 132.5
120-83-2 2,4-Dichlorophenol 45
60-57-1 Dieldrin 175.5
105-67-9 2,4-Dimethylphenol 24.5
51-28-5 2,4-Dinitrophenol 115-116
121-14-2 2,4-Dinitrotoluene 71
606-20-2 2,6-Dinitrotoluene 66
72-20-8 Endrin 200
206-44-0 Fluoranthene 107.8
86-73-7 Fluorene 114.8
75-44-8 Heptachlor 95.5
1024-57-3 Heptachlor epoxide 160
118-74-1 Hexachlorobenzene 231.8
319-84-6 a-HCH(a-BHC) 160
319-85-7 I3-HCH (I3-BHC) 315
58-89-9 y-HCH (Lindane) 112.5
67-72-1 Hexachloroethane 187
193-39-5 lndeno(1,2,3-cd)pyrene 161.5
72-43-5 Methoxychlor 87
95-48-7 2-Methylphenol 29.8
621-64-7 A/-Nitrosodi-n-propylamine NA
86-30-6 A/-Nitrosodiphenylamine 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 Endosullfan 106
NA = Not available.
30
-------
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 Csat 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., DDT), Table 3-A indicates that 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 Csat for solid compounds,
soil screening decisions should be based on the appropriate SSLs for other pathways of concern at the
site (e.g., migration to ground water, ingestion).
2.4.5 Particulate Emission Factor. The particulate 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 part of the PEF equation is based on the "unlimited reservoir" model
from Cowherd et al. (1985) developed to estimate particulate 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 |im 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 500 |im yields an uncorrected threshold friction velocity of 0.5 m/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
part 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 particulate emissions. Given that 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
-------
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 particulate matter. For this reason, a separate default Q/C
value was derived for particulate matter [nominally 10 u,m and less (PMio)] 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
concentration.
The results produced a revised default PEF Q/C value of 90.80 g/m2-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 m3/kg, which corresponds to a receptor point concentration of approximately 0.76 |ig/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 Particulate Emission Factor
PEF(m3/kg) = Q/C
3,600s/h
0.036 x (1-V) x (Um/Ut) x F(x)
(10)
Parameter/Definition (units)
PEF/particulate emission factor (m3/kg)
Q/C/inverse of mean cone, at center of square source
(g/m2-s per kg/m3)
V/fraction of vegetative cover (unitless)
Um/mean annual windspeed (m/s)
lit/equivalent threshold value of windspeed at 7 m (m/s)
F(x)/function dependent on Um/Ut derived using
Cowherd et al. (1985) (unitless)
Default
1.32x109
90.80
0.5 (50%)
4.69
11.32
0.194
Source
--
Table 3 (for 0.5-acre source 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
-------
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 site-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
-------
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 soil/water 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 unconsolidated and unconfined (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 site 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 <">
d s w
where
Kd = Freundlich soil/water partition coefficient (L/kg)
Cs = concentration sorbed on soil (mg/kg)
Cw = solution concentration (mg/L)
n = Freundlich exponent (dimensionless).
34
-------
Assuming that adsorption is linear with respect to concentration (n=l)* and rearranging to
backcalculate a sorbed concentration (Cs):
C
For SSL calculation, Cw is the target soil leachate concentration.
(12)
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
Mt = Ms + Mw + Ma
Mt = total contaminant mass in sample (mg)
Ms = contaminant mass sorbed on soil materials (mg)
Mw = contaminant mass in soil water (mg)
Ma = contaminant mass in soil air (mg).
(13)
Furthermore,
and
where
Pb =
Ow =
Ca =
ea =
Mt = Ct pb Vsp ,
Ms = Cs pb Vsp ,
Mw = Cw 0W Vsp ,
Ma = Ca 0a Vsp
dry soil bulk density (kg/L)
sample volume (L)
water-filled porosity (Lwater/Lsoil)
concentration on soil pore air (mg/Lso;i)
air-filled soil porosity (Lajr/Lsoji).
(14)
(15)
(16)
(17)
For contaminated soils (with concentrations below Csat), Ca may be determined from Cw and the
dimensionless Henry's law constant (FT) using the following relationship:
a - Cw FT
(18)
The linear assumption will tend to overestimate sorption and underestimate desorption for most organics at higher
concentrations (i.e., above 10-5 M for organics) (Piwoni and Banerjee, 1989).
35
-------
thus
Substituting into Equation 13:
a = CwH'0aV
CsPb
or
sp
CH'9
w a
Pb
(19)
(20)
c = c. - c.
e,,, + e H'
pb
(21)
Substituting into Equation 12 and rearranging:
Soil-Water Partition Equation for Migration to Ground Water Pathway: Inorganic
Contaminants
Pb
(22)
Parameter/Definition (units)
Default
Source
Cj/screening level in soil (mg/kg)
Cw/target soil leachate concentration
(mg/L)
Kd/soil-water partition coefficient (L/kg)
9w/water-filled soil porosity (Lwater/l-soii)
Gg/air-filled soil porosity (Lair/Lsoi|)
n/total soil porosity (Lpore/LSOj|)
pb/dry soil bulk density (kg/L)
ps/soil particle density (kg/L)
H'/dimensionless Henry's law constant
H/Henry's law constant (atm-m3/mol)
(nonzero MCLG, MCL,
or HBL) x 20 DAF
chemical-specific
0.3 (30%)
0.13
0.43
1.5
2.65
H x41, 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-9w
1 -Pb/Ps
U.S. EPA, 1991 b
U.S. EPA, 1991 b
U.S. EPA, 1991 b
see Part 5
36
-------
Equation 22 is used to calculate SSLs (total soil concentrations, Ct) 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, 0a 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 K
-------
Parameter/Definition (units)
Default
Source
Ct/screening level in soil mg/kg)
Cw/target leachate concentration (mg/L)
il organic carbon-water partition
coefficient (L/kg)
foc/organic carbon content of soil (kg/kg)
9w/water-filled soil porosity (Lwater/l-soii)
Gg/air-filled soil porosity (Lair/Lsoi|)
n/total soil porosity (Lpore/LSOii)
pb/dry soil bulk density (kg/L)
ps/soil particle density (kg/L)
H'/dimensionless Henry's law constant
H/Henry's law constant (atm-m3/mol)
(nonzero MCLG, MCL,
or HBL) x 20 DAF
chemical-specific
0.002 (0.2%)
0.3 (30%)
0.13
0.43
1.5
2.65
H x41, 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-9w
1 - Pb/Ps
U.S. EPA, 1991b
U.S. EPA, 1991b
U.S. EPA, 1991b
see Part 5
Part 5 of this document provides Koc 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 of 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 halogenated 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 this 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 foc* 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 foc = 0.0003 by considering Koc 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):
where
Kj
fio + foc
- (Koc
soil inorganic partition coefficient
fraction of inorganic material
fio)
(25)
38
-------
Although this equation is considered conceptually valid, K;0 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, r2 = 0.85) that can be used to estimate Kd values for hydrophobic organic
chemicals from Kowfor low-carbon soils:
log Kd = 1.01 log Kow - 0.36 (26)
where
Kow = octanol/water partition coefficient.
The authors indicate that this equation should provide a K^ estimate that is within a factor of 2 or 3
of the actual value for nonpolar sorbates with log Kow < 3.7. This Kd 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 (Kocp ) as follows:
l-On) (27)
where
Koc n, Koc j = sorption coefficients for the neutral and ionized species (L/kg)
On = (1 + 10pH-pKa)-l
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, 1996) provides guidance on conducting site-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
-------
2.5.3 Inorganics (Metals)—Partition Theory. Equation 22 is used to estimate SSLs 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 Kd values
are largely controlled by a single parameter (soil organic carbon), Kd values for metals are
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^ values for
individual metals reported in the literature (over 5 orders of magnitude). Thus, it is much more
difficult to derive generic Kd values for metals than for organics.
The Kd values used to generate SSLs for Ag, Ba, Be, Cd, Cr+3, 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, Kd values are selected as a function of site-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
-------
4. Adsorption is reversible. The methodology assumes that desorption processes operate
in the same way as adsorption processes, since most of the Koc 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 than 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 ground 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 unsaturated 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
-------
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 site area in question, SSLs
should not be used for this area (i.e., further investigation is required). The Csat 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 + Ugw 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
-------
Option 2 (EPA Ground Water Forum):
dilution factor = (Qp + QA)/Qp (31)
where
Qp = percolation flow rate (m3/yr)
QA = aquifer flow rate (m3/yr)
For percolation flow rate:
Qp = IA (32)
where
A = facility area (m2) = WL.
For aquifer flow rate:
QA = WdKi (33)
where
W = width of source perpendicular to flow (m)
d = mixing zone depth (m).
Thus
dilution factor = (IA + WdKi)/IWL
= 1 + (Kid/IL) (34)
Option 3 (Summers Model):
Cw = (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
-------
1/dilution factor = QP/(QP + QA)
or
dilution factor = (Qp + CLO/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)
(Note 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/IL) (37)
Parameter/Definition (units)
K/aquifer hydraulic conductivity (m/yr)
i/hydraulic gradient (m/m)
d/mixing zone depth (m)
I/infiltration rate (m/yr)
L/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 = (2avL)o.s + da {1 - exp[(-LI)/(Vsneda)]} (38)
44
-------
where
Oy = vertical dispersivity (m/m)
Vs = horizontal seepage velocity (m/yr)
ne = effective aquifer porosity (Lpore/Laquifer)
da = aquifer depth (m).
The first term, (2avL)°-5, 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 the source parallel to flow
L. Vertical dispersivity can be estimated by the following relationship (Gelhar and Axness, 1981):
av = 0.056 aL (39)
where
(XL = longitudinal dispersivity = 0.1 xr
xr = horizontal distance to receptor (m).
Because the potential receptor is assumed to have a well at the edge of the facility, xr = L and
av = 0.0056 L (40)
Thus
dav = (0.0112L2)0.5 (41)
The second term, da {1 - exp[(-LI) / (Vsneda)]}, estimates the depth of mixing due to the downward
velocity of infiltrating water, div. In this equation, the following substitution may be made:
Vs = Ki/ne (42)
so
dIv = da{l-exp[(-LI)/(Kida)]} (43)
Thus, mixing zone depth is calculated as follows:
d = dav + dlv (44)
Estimation of Mixing Zone Depth
d = (0.0112 L2)o.s + da {1 - exp[(-LI)/(Kida)]} (45)
45
-------
Parameter/Definition (units)
d/mixing zone depth (m)
L/source length parallel to ground water flow (m)
I/infiltration rate (m/yr)
K/aquifer hydraulic conductivity (m/yr)
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 (da). 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. The 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 (da, 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
-------
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, this
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
-------
• 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, site data from over 1,300 municipal
landfill sites in OSW's 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.
Parameters:
• X (distance from source to well) = 0 ft
• Y (transverse well location) = Monte Carlo within
1/2 width of source
• Z (well intake point below water table) = Monte
Carlo, range 15 -» 300 ft
• Rainfall = Monte Carlo
• Soil type = Monte Carlo
• Depth to aquifer = Monte Carlo
• Assumes infinite source term
Figure 3. Migration to ground water pathway—EPACMTP modeling
effort.
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
-------
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
2.09E+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
-------
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:
da = 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 site-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 1O5 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/ne (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 nonconservative 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
-------
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 not 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 needed 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
-------
Table 6. Recharge Estimates for DNAPL Site Hydrogeologic Regions
Hydrogeologic setting
Nonglaciated Central (Region 6)
Alluvial Mountain Valleys
Alter. SS/LS/Sh., Thin Soil
Alter. SS/LS/Sh., Deep Regolith
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
Glaciated Central (Region 7)
Glacial Till Over Bedded Rock
Glacial Till Over Outwash
Glacial Till Over Sol. Limestone
Glacial Till Over Sandstone
Glacial Till 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 Till Over Bedded Rock
Beaches, B. Ridges, Dunes*
Swamp/Marsh
Recharge
Min. Max.
0.10 0.18
0.10 0.18
0.10 0.18
0.25 0.38
0.18 0.25
0.18 0.25
0.10 0.18
0.10 0.18
0.10 0.18
0.00 0.05
0.00 0.05
Overall Average:
0.10 0.18
0.10 0.18
0.10 0.18
0.10 0.18
0.10 0.18
0.18 0.25
0.25 0.38
0.25 0.38
0.18 0.25
0.18 0.25
0.10 0.18
0.25 0.38
0.10 0.18
0.18 0.25
0.25 0.38
0.10 0.18
Overall Average:
(m/yr)
Avg.
0.14
0.14
0.14
0.32
0.22
0.22
0.14
0.14
0.14
0.03
0.03
0.15
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
Recharge (m/yr)
Hydrogeologic setting
Piedmont/Blue Ridge (Region 8)
Alluvial Mountain Valleys
Regolith
River Alluvium
Mountain Crests
Swamp/Marsh
Min.
0.18
0.10
0.18
0.00
0.10
Max.
0.25
0.18
0.25
0.05
0.18
Overall Average:
Northeast & Superior Uplands (Region 9)
Alluvial Mountain Valleys
Till 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./Semiconsol. Surficial Aquifer*
Alluvium w/ Overbank Deposits
Alluvium w/o Overbank Deposits*
Swamp*
0.00
0.25
0.18
0.25
0.25
0.05
0.38
0.25
0.38
0.38
Overall Average:
Avg.
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
0.03
0.32
0.22
0.32
0.32
0.24
Source: Aller et al. (1987); hydro-geologic regions from Heath (1984).
* 0.25m to 0.38m (9.8 in to 15 in) used as recharge range for 25+ m setting values from Aller etal. (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
1 0
15
138
2
4
13
60
144
10
829
1
2
6
30
134
11
617
1
2
8
35
144
30
10
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
-------
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, the 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 site-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 that the parameter ranges selected do not
result in impossible combinations of these parameters. Because the effects of the soil parameters on
the SSLs 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. In 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
chloroform. As expected, the effect of foc increases with increasing Koc. The greatest effect was seen
for benzo(a)pyrene whose SSL showed a 50 percent increase at an foc of 0.03. An foc of 0.005 will
increase the benzo(a)pyrene SSL by 150 percent.
54
-------
Table 8. Sensitivity Analysis for SSL Partition Equation
Parameter assignments
All default parameter values
Less conservative parameter value
Organic carbon
Bulk density
Soil moisture
More conservative parameter value
Organic carbon
Bulk density
Soil moisture
Chloroform
SSL Percent
(mg/kg) change
0.59 —
0.67 14
0.69 18
0.74 26
0.51 -14
0.51 -13
0.27 -54
Trichloroethylene
SSL
(mg/kg)
0.057
0.074
0.065
0.062
0.040
0.051
0.046
Percent
change
—
29
14
9
-29
-10
-19
Naphthalene
SSL Percent
(mg/kg) change
84 —
124 48
85 1
86 2
44 -48
83 -1
80 -4
Benzo(a)pyrene
SSL Percent
(mg/kg) change
8 —
12 50
8 0
8 0
4 -50
8 0
8 0
Conservatism
Chemical-specific parameters
KOC
H'
cw
Input parameters
Fraction org. carbon
(g/g)
Bulk density (kg/L)
Average soil moisture
(L/L)
a n = 0.53; qa = 0.23.
b n = 0.34; qa = 0.04.
Chloroform
3.98E+01
1.50E-01
2. DC
Less
0.003
1.25a
0.43
Nominal
0.002
1.50
0.30
More
0.001
1.75b
0.02
Trichloroethylene Naphthalene
1.66E+02
4.22E-01
0.1C
2.00E+03
1.98E-02
20d
Benzo(a)pyrene
1.02E+06
4.63E-05
0.004C
<= MCL x 20 DAF.
d HBL(HQ=1)x20DAF.
-------
Dilution Factor. Site-specific parameters for the dilution factor model include aquifer hydraulic
conductivity (K), hydraulic gradient (i), infiltration rate (I), aquifer thickness (d), and source length
parallel to ground water flow (L). Because they are somewhat 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 that
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 (Csat) 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
-------
Table 9. Sensitivity Analysis for SSL Dilution Factor Model
Dilution Factor
Source area
Parameter
assignments
30-acre 0.
All central parameters 5.2
Less conservative
Darcy velocity
Aquifer thickness
Infiltration rate
More conservative
Darcy velocity
Aquifer thickness
Infiltration rate
85
15
39
1.2
2.1
3.2
Input parameters
Mixing
Ratio of 0.5-
5-acre acre/30-acre 30-acre
15
263
15
118
2.1
9.1
8.7
Darcy velocity (DV, m/yr)
Aquifer thickness (da
Infiltration rate (m/yr)
,m)
2.9 12
3.1 12
1.0 40
3.0 12
1.8 12
4.3 3.0
2.7 12
Conservatism
Less Nominal More
442 22 0.8
46 12 3
0.02 0.18 0.35
depth (m)
0.5 acre
5.1
4.8
5.1
4.8
12
3.0
5.5
Parameter sources
Percentile
10th
25th
50th
75th
90th
Average:
Geomean:
DVa
0.8
4
22
121
442
800
22
(m/yr) dab (m)
3.0
5.5
11
23
46
28
12
300 DNAPL & HGDB sites.
208 HGDB sites.
57
-------
• 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, more 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 (AIHC) 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
MI = Cw x I x As x ED (47)
where
MI = mass of contaminant leached (g)
Cw = leachate contaminant concentration (mg/L or g/m3)
I = infiltration rate (m/yr)
As = source area (m2)
ED = exposure duration (yr).
58
-------
The total mass of contaminants present in a source can be expressed as
MT = Q xpb x As x ds (48)
where
MT = total mass of contaminant present (g)
Q = total soil contaminant concentration (mg/kg or g/Mg, dry basis)
pb = dry soil bulk density (kg/L or Mg/m3)
As = source area (m2)
d<, = source depth (m).
To avoid a mass balance violation, the mass of contaminant leached cannot exceed the total mass of
contaminants present (i.e., MI 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
MI = MT
or
Cw x I x As x ED = Q x pb x As x ds
Rearranging to solve for the total soil concentration (Ct) corresponding to this situation (i.e.,
maximum possible leaching)
Mass-Limit Model for Migration to Ground Water Pathway
Q = (Cw x I x ED)/(pb x ds)
(49)
Parameter/Definition (units)
Default
Cj/screening level in soil (mg/kg)
Cw/target soil leachate concentration (mg/L)
I/infiltration rate (m/yr)
ED/exposure duration (yr)
pb/dry soil bulk density (kg/L)
ds/average source depth (m)
(nonzero MCLG, MCL, or HBL) x 20 DAF
site-specific
70
1.5
site-specific
59
-------
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 Ct, the soil leachate concentration averaged over the ED time period cannot exceed Cw.
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 (CT°/Jsave) x 10-10 m2kg/cm2mg (50)
where
VF = volatilization factor (m3/kg)
Q/C = inverse concentration factor for air dispersion (g/m2-s per kg/m3)
CT° = total soil contaminant concentration at t=0 (mg/kg or g/Mg, dry basis)
Jsave = 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 pb x As x ds (51)
where
Mt = total mass of contaminant within the source (g)
CT° = 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)
dg = 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
Jsave = Mt/[(As x 104 Cm2/m2) x (T x 3.15E7 s/yr)] (52)
where
T = exposure period (yr).
Substituting Equation 51 for Mt in Equation 52 yields
jsave = (cxo x pb x ds) / (1Q4 cm2/m2 x T x 3.15E7 s/yr) (53)
Rearranging Equation 53 yields
60
-------
CTo/jsave = (1Q4 Cm2/m2 x T x 3.15E7 s/yr)/(pb x ds)
(54)
Substituting Equation 54 into Equation 50 yields
Mass-Limit Model for Inhalation of Volatiles
VF = (Q/C) x [(T x 3.15E7 s/yr)/(pb x ds x 1Q6 g/Mg)]
(55)
Parameter/Definition (units)
Default
VF/volatilization factor (m3/kg)
Q/C/inverse of mean cone, at center of source (g/m2-s per kg/m3)
T/exposure interval (yr)
pb/dry soil bulk density (kg/L)
ds/average source depth (m)
Table 3
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
-------
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
-------
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 (VF;ncjoor) to calculate carcinogenic
or noncarcinogenic 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 standardization 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 site-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
-------
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 fractionation 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 site conditions (e.g., a thick unsaturated zone),
the additional cost and time required to apply these models may be offset by the potential 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
(within 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
-------
program is available from the Center for Subsurface Modeling Support (CSMOS) at EPA's
Environmental Research Laboratory in Ada, Oklahoma (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 derive 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
Csat)
• No boundary layer thickness at ground level (no stagnant air layer)
• No water evaporation or leaching
• No chemical reactions, biodegradation, or photolysis
• ds » (4DAt)1/2 (ramifications of this are discussed below).
Under these assumptions, the Jury et al. (1990) simplified finite source model is
Js = C0(DA/7it)i/2[l-exp(-ds2/4DAt)] (56)
where
Js = contaminant flux at ground surface (g/cm2-s)
C0 = uniform contaminant concentration at t=0 (g/cm3)
DA = apparent diffusivity (cm2/s)
TI = 3.14
t = time (s)
dg = depth of uniform soil contamination at t=0 (cm),
and
DA = [(9ai°/3 0; H' + 0wio/3 Dw)/n2]/(pb Kd + 0W + 0a H') (57)
where
0a = air-filled soil porosity (Lajr/Lsoji) = n - 0W
n = total soil porosity (Lpore/Lsoil) = 1 - (pb/ps)
0W = water-filled soil porosity (Lwater/Lsoil) = wpb/pw
pb = soil dry bulk density (g/cm3)
ps = soil particle density (g/cm3)
w = average soil moisture content (g/g)
65
-------
pw = water density (g/cm3)
D; = diffusivity in air (cmVs)
H' = dimensionless Henry's law constant = 41 x HLC
HLC = Henry's law constant (atm-m3/mol)
Dw = diffusivity 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/Jsave) x 10-4 m2/cm2 (58)
where
VF = volatilization factor (m3/kg)
Q/C = inverse concentration factor for air dispersion (g/m2-s per kg/m3)
C0 = uniform contaminant concentration at t=0 (g/cm3)
pb = soil dry bulk density (g/cm3)
Jsave = 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 ISCST3 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
-------
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 EST).
The user registers in the first call and then has full access to the BBS.
The ISCST3 model will output an air concentration (in |ig/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 corner of the square (e.g., SO LOCATION
sourcename AREA -^length -1/2width height=0). For the source parameter input line, the
contaminant's area emission rate (in units of g/m2-s) must be entered. The area 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 = Jsave x 104 cm2/m2). Alternatively, an area emission rate of 1 g/m2-s 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 = (Jsave x 104 Cm2/m2)/(Cair x 10-9 kg/|ig) (59)
where
Q/C = inverse concentration factor for air dispersion (g/m2-s per kg/m3)
Jsave = average rate of contaminant flux (g/cm2-s)
Cair = ISC output maximum contaminant air concentration (|ig/m3).
Note: If an area emission rate of 1 g/m2-s is assumed, then (Jsave x 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 that can address such site 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
-------
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 (CRP) as follows:
DAF = CW/CRP = 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 summarized 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:
• RITZ (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
-------
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 RITZ model was obtained primarily from Nofziger et al. (1994). RITZ is
a steady-state analytical model used to simulate the transport and fate of 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
Oil 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
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
-------
the source to be infinite. VIP differs from RITZ 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 RITZ 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
Site
characteristics
Plow zone depth
Pollutant
properties
Concentration in
Oxygen properties
Oil-air partition
Oil
properties
Density of oil
Bulk density
Saturated hydraulic
conductivity
Clapp-Hornberger
constant
Treatment zone
depth
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
sludge
Oil-water partition
coefficient3
Air-water partition
coefficient3
Soil-water partition
coefficient3
Degradation constant
in oil3
Degradation constant
in water3
Dispersion coefficient
Adsorption-desorption
rate constant (water/oil)
Adsorption-desorption
rate constant
(water/soil)
Adsorption-desorption
rate constant (water/air)
coefficient3
Water-air partition
coefficient3
Oxygen half-saturation
constant in air phase 3
Oxygen half-saturation
constant in oil phase 3
Oxygen half-saturation
constant in water
phase3
Oxygen half-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
Degradation
rate constant
of oil
a Parameters required for 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 chemical properties defined for each layer.
70
-------
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 evapotranspiration Amount applied
Bulk density — Depth of application
Saturated water content — Date of application
Field capacity — Koc
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 on 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
-------
Table 13. Input Parameters Required for HYDRUS
Soil properties
Depth of soil layers
Saturated water
content
Saturated hydraulic
conductivity
Bulk density
Site characteristics
Uniform orstepwise
rainfall intensity
Contaminant
concentrations in soil
—
—
Pollutant properties
Molecular diffusion
coefficient
Dispersivity
Decay coefficient
(dissolved)
Decay coefficient
Root uptake
parameters
Power function in stress-
response function
Pressure head where
transpiration is reduced by
50%
Root density as a function of
depth
—
Retention
parameters
Residual water
content
(adsorbed)
Freundlich isotherm
coefficients
Table 14. Input Parameters Required for SUMMERS
Parameters required
Target concentration in ground water
Volumetric infiltration rate into aquifer
Downward porewater velocity
Ground water seepage velocity
Void fraction
Horizontal area of pond or spill
Thickness of aquifer
Width of pond/spill perpendicular to flow
Initial (background) concentration
Equilibrium partition coefficient
Darcy velocity in aquifer
Volumetric ground water flow rate
MULTIMED. Information on the MULTIMED model was obtained from Criscenti 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 using 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 dispersivity 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
-------
Table 15. Input Parameters Required for MULTIMED
Unsaturated zone parameters
Saturated hydraulic
conductivity
Porosity
Air entry pressure head
Depth of unsaturated 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 linear 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
K,
oc
Henry's law constant
Aqueous solubility
Recharge rate
Contaminant concentrations in
recharge
Depth to ground water
Free air diffusion coefficient Dimensions of "polygons"
73
-------
SESOIL. Information on the SESOIL model was obtained from Criscenti et al. (1994). SESOIL 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
Chemical data
Application data
Mean air temperature3
Mean cloud cover
fraction3
Mean relative humidity3
Short wave albedo
fraction3
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
Silt, 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
Site 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
Wash load area
Average slope and
slope length
Erodibility factor
Practice factor
Manning coefficient
a SESOIL uses these parameters to calculate evapotranspiration if an evapotranspiration value is not specified.
74
-------
PRZM-2. Information on PRZM-2 was obtained from Criscenti et al. (1994). PRZM-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, VADOFT, 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.
VADOFT can be run independently of PRZM and output from the PRZM model can be used to set
the boundary conditions for VADOFT. 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 VADOFT 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 that 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
-------
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
erodibility
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
Foliar extraction
coefficient
Diffusion coefficient in
air
Initial concentration
levels
Number of
applications
(50 maximum)
Incorporation depth
Enthalpy of
vaporization
Parent/daughter
transform rates
Number of chemicals
(3 maximum)
Plant uptake factor
Kd and KQC
Aqueous, sorbed, vapor
decay rates
Application dates
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
Albedo
Avgerage monthly
bottom boundary
temperature
Reflectivity of soil surface
Initial horizon
temperature
Height of windspeed
measurement
Sand and clay content
Biodegradation and irrigation parameters (not presented)
76
-------
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
-------
Table 20. Characteristics of Unsaturated Zone Models Evaluated
Model
RITZ
VIP
CMLS
HYDRUS
Mill TIMFD
SUMMERS
PRZM-2
SESOIL
VLEACH
Type
"ro
o
5,
ro
c
<
.
.
•
emianalytical
OT
umerical
-z.
•
•
•
•
•
Fate and Transport Processes Considered
nite source
LL
.
•
•
•
•
.
•
•
•
•
t
C~
03
(0
Q.
O
Q.
(0
•
«
c
o
'«
-e
ro
Q_
onequilibrium
-z.
•
•
^
ro
o
,
O
T3
>,
I
.
•
.
•
odegradation
CO
.
J/J
'5
T3
OJ
.
•
«
^
ro
ro
^
ro
O
"c
o
^
•
c
ro
^
0
ro
0
later balance
^
.
•
•
•
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 following
text highlights some of the differences between the models, outlines their advantages 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 particles).
Sorption, degradation, volatilization, and first-order decay processes are considered in the subsurface
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 additional
78
-------
input parameters are required to simulate transient partitioning between the air, soil, water, and oil
phases. Like RITZ, VIP's focus on land treatment of oily waste limits its application to SSLs.
CMLS. CMLS differs from RITZ and VIP in that 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
-------
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 model 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.
80
-------
Part 4: MEASURING CONTAMINANT CONCENTRATIONS IN SOIL
The 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 4.2 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
-------
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.
4.1 Sampling Surface Soils
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
A sampling strategy for surface soils is presented in this section,
organized by the steps of the DQO process. The first five steps
of this 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 on 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
-------
Table 21. Sampling Soil Screening DQOs for Surface 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., 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 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 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" (|i) individual contaminant concentration in each EA. (since the
determination of the "true mean" would require the collection and analysis of
many samples, the "Max 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
-------
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 site 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 site 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 site that are within the legal boundaries of the property but
84
-------
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 site-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 site 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 site 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 site contamination
data. It may be identical to the SSL, or it 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
-------
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 that 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 making 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 H0, 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, site 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 times the SSL (2 SSL). By specifying the upper edge
of the gray region as twice the SSL, it is possible that exposure areas with mean values slightly higher
than the SSL may be screened from further study. However, EPA believes that the exposure scenario
86
-------
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. The 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 that 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
-------
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'strue 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 I: 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 of 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
-------
Probability of
Deciding that
the Mean
Exceeds the
Screening Level
1.0
0.9
0.8
0.7
0.6
Q 5
Q 4
0.3
0.2
0.1
0
1.0
Tolerable
Type II
Decision
Error
Rates
f
•^-
1
V
Tolerable Type 1
Decision Error
Rates
Gray Region
(Relatively Large
Decision Error Rates
are Considered
Tolerable.)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1 1 1
0.5 x SSL SSL 2x 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
-------
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).
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 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 specimen(s) 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
-------
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 site-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
-------
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 Chapters 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. 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.
•i
•3 *4
•5 *6
•i
•3 *4
•5 *6
•i
•3 *4
•i
•3 *4
Figure 6. Systematic (square grid points) sample with systematic compositing scheme
(6 composite samples consisting of 4 specimens).
92
-------
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 the N
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
• 1 *4
• 1 *5
•6
•2 *4
•6 *5
• 1 *4
•3
•4 *6
•3 *5
•2 •!
Figure 7. Systematic (square grid points) sample with random compositing scheme
(6 composite samples consisting of 4 specimens).
93
-------
Highlight 5: Procedure for Compositing of Specimens from a Stratified Random
Sample Using a Random Scheme (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, 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.
• 1
•2 *4
•5
•6
•6 *3
•2
•4 •!
•5
• 1
•4 *6
•3
5
•2
•2
•4
•3
•6
• 1
Figure 8. Stratified random sample with random compositing scheme
(6 composite samples consisting of 4 specimens).
94
-------
Table 23. Probability of Decision Error at 0.5 SSL and 2 SSL Using Max Test
Sample
Sizeb
4
5
6
7
8
9
4
5
6
7
8
9
CV=1.0a
Eo.5C
F d
C2.0
CV=1.5
Eo.5
E2.0
CV=2.0
Eo.5
E2.0
CV=2.5
Eo.5
E2.0
CV=3.0
Eo.5
E2.0
CV=3.5
E0.5
E2.0
CV=4.0
E0.5
E2.0
C = 4 specimens per composite6
<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
0.11
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
0.36
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
0.09
0.08
C = 6 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
0.06
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
0.26
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
a The CV is the coefficient of variation for individual, uncomposited measurements across the entire EA, including measurement error.
b Sample size (N) = number of composite samples.
c E0.5 = Probability of requiring further investigation when the EA mean is 0.5 SSL.
d E2.o = Probability of not requiring further investigation when the EA mean is 2.0 SSL.
e 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: All 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 quantitation limit (i.e., SSL/100), and half the EA has concentrations that follow a gamma distribution (a conservative
distributional assumption).
95
-------
The error rates shown in Table 23 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 (C = number of sectors or
strata).
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 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., lognormal), rather than gamma, then 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 xls x2, ..., XN represent concentration
measurements for N composite samples that each consist of C specimens selected so that each
96
-------
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 (xls x2, ..., XN) > 2 SSL, then investigate the EA further;
If Max (xls 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 against 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 1//C, 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///~C . 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
-------
Highlight 6: Directions for Data Quality Assessment for the Max Test
Let x-i, X2, ..., XN represent contaminant concentration measurements for N composite samples that
each consist of C specimens selected so that each composite is representative of the EA as a whole.
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 =
STEP 3: Calculate the sample standard deviation
s =
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^, ..., * )<—7^,then no further data quality assessment is needed and the EA
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 II 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
-------
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: (i > 2 SSL, versus the alternative hypothesis,
HI: (i < 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, H0: (i < 0.5 SSL, versus the alternative hypothesis,
Hj: (i > 0.5 SSL, at the 20 percent significance level using a sample size chosen to
achieve a Type II 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
by increasing 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, HQ: (i < SSL, instead of HQ: ^ < 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 II 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
-------
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
-------
Table 24. Sampling Soil Screening DQOs for Surface Soils under Chen
Test
DQO 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 not investigate further ("walk away from") when
an EAtrue 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 I: 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 Plan (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 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),
101
-------
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 that 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
-------
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 that 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 Chen 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 that 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 than 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
-------
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*5
2
3
4
5
6
Coefficient of variation (CV)a
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
aThe 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 composite*5
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)a
2.5
>9
>9
9
7
6
5
3.0
>9
>9
>9
>9
8
8
3.5
>9
>9
>9
>9
>9
9
aThe 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).
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*5
1
2
3
4
5
6
Coefficient of variation (CV)a
1.0
5
4
4
4
4
4
1.5
9
4
4
4
4
4
2.0
>9
8
5
4
4
4
2.5
>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
aThe 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 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*5
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)a
2.5
>9
>9
7
6
5
3.0
>9
>9
>9
8
7
3.5
>9
>9
>9
>9
9
aThe 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).
105
-------
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 composite13
1
2
3
4
5
6
Coefficient of variation (CV)a
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
aThe 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 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)a
specimens
percompositeb 1-0 1-6 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
aThe 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).
106
-------
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 that 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 site 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 that 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 that 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
-------
Highlight 7: Directions for the Chen Test Using Simple Random Sample Scheme
Letx-i, 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 it were a simple random sample).
The following describes the steps for a one-sample test for H0: u < 0.5 SSL at the 100a% significance
level that is designed to achieve a 100B% chance of incorrectly accepting H0 when u = 2 SSL.
STEP 1: Calculate the sample mean x =
I X,
I I
I I
| STEP 2: Calculate the sample standard deviation |
I _ I
s=
STEP 3: Calculate the sample skewness
£(*.-*)'
b-NT
= N
(N-l) (N-2) s'
STEP 4: Calculate the Chen test statistic, t.2, as follows:
b
a = -
x-0.5 SSL
s /
t2 = t+a (l + 2t2) + 4a2 (t+2t3)
STEPS: Compare t2 to 2^, the 100(1 -a) percentile of the standard normal probability distribution.
If \2 > Za,tne null hypothesis is rejected, and the EA needs further investigation.
If t.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 100B% or less chance of incorrectly
accepting the H0 when u = 2 SSL.
108
-------
Highlight 7: Directions for the Chen 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-|, 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 x-i, X2,..., XN is an
individual random or pseudo-random sampling point, then C = 1 .)
SSL
If Max (x x^..., x )<— y^, 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.
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
100B% or less chance of incorrectly accepting HQ when u = 2 SSL.
If 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 for
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 was
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
-------
(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 by the
migration of contaminants through soils to an underlying potable aquifer. The problem is to identify
the contaminants and source areas that do not pose significant risk to human health through either
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 site. 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 areal 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
-------
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., toxicologists, risk assessors,
hydrogeologists, statisticians).
CSM development (described in Step 1 of the User's Guide, U.S. EPA, 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
-------
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.e., 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
-------
Figure 9: U.S. Department of Agriculture soil texture classification.
100
90
Percent Sand
<4
Criteria Used with the Field Method for Determining Soil Texture Classes (Source: Brady, 1990)
Criterion
1.
2.
3.
4.
Individual grains
visible to eye
Stability of dry
clods
Stability of wet
clods
Stability of
"ribbon" when
wet soil rubbed
between thumb
and fingers
Sand Sandy loam
Yes Yes
Do not form Do not form
Unstable Slightly stable
Does not Does not form
form
Loam
Some
Easily
broken
Moderately
stable
Does not form
Silt loam
Few
Moderately
easily broken
Stable
Broken appearance
Clay loam
No
Hard and
stable
Very stable
Thin, will break
Clay
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
Fine
Med.
Coarse
Very Coarse
Sand
Gravel
Source: USDA.
113
-------
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, Koc. Soil pH is used to
select site-specific partition coefficients for metals and ionizing organic compounds (see Part 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 site-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
-------
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 site 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
-------
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
contamination (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 from the surface to the depth of contamination or to the water table. However, if the
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 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),
I
_ 1=1
c =
(61)
I 1,
1=1
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
-------
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.
For the 1994 draft Soil 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).
117
-------
The equation for estimating the Land upper confidence limit (UL) for a lognormal mean has the
form
(62)
UL = exp( v + —^-
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 HQ: (i> 2 SSL versus the alternative hypothesis Hj: (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 (\a represents the true EA mean).
That is, the probability of incorrectly deciding not to investigate further when the true mean is 2 SSL
was set not to exceed 0.05, and the probability of incorrectly deciding to investigate further when the
true mean is 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
-------
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 it.
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 ( P0 (EA needs further investigation)
versus
119
-------
HI: P < P0 (EA does not need further investigation).
The test is based on concentration data from a grid sample of N points in the EA (without
compositing). Let p represent the proportion of these n points with observed concentrations greater
than or equal to 2 SSL. The test is carried out by choosing a critical value, pc, to meet the desired
Type I error rate, that is,
ot = Prob(p
-------
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 SSL (where (i 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, H0: (i > 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
composite).
The Chen test requires that the null hypothesis have the form HQ: (i < (IQ, with the alternative
hypothesis as Hj: (i > (i0 (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 (i0 = 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, \i, 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 H0: (i < 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.
121
-------
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-MIX) 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:
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 site
(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.
For 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
-------
cv
1.0
1.5
2.0
2.5
3.0
3.5
4.0
5.0
6.0
Values of MIX
used 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 H0: (i > 2 SSL at the 5 percent significance
level
Land test of the flipped null hypothesis H0: (i < 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
-------
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 MIX
C=6
C=6
C=6
C=6
MU =
MIX =
C =
N =
CV =
N=8
0.5
0.5
0.5
2.0
2.0
2.0
N=8
0.5
0.5
0.5
2.0
2.0
2.0
N=8
0.5
0.5
0.5
2.0
2.0
2.0
N=8
0.5
0.5
2.0
2.0
True
CV=4
.00
.50
.90
.00
.50
.90
CV=3
.00
.50
.85
.00
.50
.85
CV=2
.00
.50
.75
.00
.50
.75
CV=1
.00
.49
.00
.49
Max test
.35
.40
.40
.06
.06
.04
.24
.25
.23
.04
.03
.03
.07
.06
.04
.02
.02
.01
.00
.00
.01
.01
EA Mean - see subsection entitled
0.20 Chen test 0.10 Chen test
.18
.22
.19
.10
.11
.16
.18
.19
.22
.03
.03
.06
.22
.19
.19
.00
.00
.00
.20
.20
.00
.00
.09
.11
.09
.18
.18
.29
.10
.10
.11
.06
.05
.12
.11
.09
.10
.00
.01
.01
.10
.12
.00
.00
Land test
.99
.99
.98
.00
.00
.01
.93
.94
.99
.00
.00
.00
.57
.68
.85
.01
.00
.00
.01
.12
.02
.00
"Simulation Distributions" in Section 4.3.3.
Mixing Proportion - see subsection entitled "Simulation Distributions" in
Number of specimens in a composite.
Number of composites analyzed.
EA coefficient of variation (J~c)s
where s
= sample standard
X
deviation and x
= mean sample concentration
Section 4.3.3
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 nonlognormal distributions tested. This limitation led to removal of the Land procedure from
the Soil Screening Guidance.
124
-------
Earlier simulation results for gamma and Weibull 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, (ix , 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 \iy and ay 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
2
UL = LL = exp (ny + -y ) - (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
distributional assumptions and are potentially suitable for surface soil screening.
125
-------
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
5
5
6
6
7
7
8
8
9
9
C = 6
4
4
5
5
6
6
7
7
8
8
9
9
C = 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
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
0.5
2.0
CV =
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 =
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
MU = True EA Mean - see subsection entitled "Simulation Distributions" in Section 4.3.3.
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
-------
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 2 SSL
versus
HI: 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-B = 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 H0) 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
-------
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 (uncorrelated). 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 H0 for Chemical A} x Prjreject H0 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 Away" 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)
P(B)
0
0
0
0
0
0
= 0.2
= .95
90
76
57
48
19
05
C(B)
P(B)
0
0
0
0
0
0
= 0.5
= .80
76
64
48
40
16
04
C(B) = 0.7
P(B) = .60
0.57
0.48
0.36
0.30
0.12
0.03
C(B) = 1.0
P(B) = .50
0.48
0.40
0.30
0.25
0.10
0.03
C(B)
P(B)
0
0
0
0
0
0
= 1.5
= .20
19
16
12
10
04
01
C(B) =
P(B) =
0.05
0.04
0.03
0.03
0.01
<0.01
2.0
.05
128
-------
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 H0 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
129
-------
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 nonoverlapping 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
-------
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 (DBS) 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
EA
1
2
3
4
5
6
7
Mean of EA
2.1
2.4
5.1
4.0
9.3
15.8
2.8
CV of EA
1.0
1.6
1.1
1.2
2.0
2.2
1.4
N
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
-------
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 site 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
-------
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 (Kd), water solubility (S), Henry's law constant (HLC, H'), air
diffusivity (Dia), and water diffusivity (Diw). In addition, the octanol-water partition coefficient
(Kow) is needed to calculate Koc values. This part 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; a), water diffusivity (D; 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 KQW 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 Kow 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 H', 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,a) and Water (Dj
-------
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
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
Compound
Acenaphthene
Acetone
Aldrin
Anthracene
Benz(a)anthracene
Benzene
Benzo(Jb)fluoranthene
Benzo(/<)fluoranthene
Benzole acid
Benzo(a)pyrene
Bis(2-chloroethyl)ether
Bis(2-ethylhexyl)phthalate
Bromodichloromethane
Bromoform
Butanol
Butyl benzyl phthalate
Carbazole
Carbon disulfide
Carbon tetrachloride
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
S
(mg/L)
4.24E+00
1.00E+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+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+03
HLC H
(atm-m3/mol) (dimensionless) '°9 Kow
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
8.81E-06
1.26E-06
1.53E-08a
3.03E-02
3.04E-02
4.86E-05
3.31E-07
3.70E-03
7.83E-04
3.67E-03
3.91E-04
9.46E-05
4.00E-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
6.31E-05
4.63E-05
7.38E-04
4.18E-06
6.56E-02
2.19E-02
3.61E-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
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
-------
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-Dinitrotoluene
2,6-Dinitrotoluene
Di-n-octyl phthalate
Endosulfan
Endrin
Ethylbenzene
Fluoranthene
Fluorene
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro-1 ,3-butadiene
a-HCH (a-BHC)
p-HCH (p-BHC)
Y-HCH(Lindane)
Hexachlorocyclopentadiene
Hexachloroethane
lndeno(1 ,2,3-ccf)pyrene
Isophorone
Mercury
Methoxychlor
Methyl bromide
Methylene chloride
2-Methylphenol
Naphthalene
Nitrobenzene
/V-Nitrosodiphenylamine
/V-Nitrosodi-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
5.10E-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
3.51E+01
9.89E+03
HLC H
(atm-m3/mol) (dimensionless) '°9 Kow
9.79E-04
2.61E-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
1.61E-05
6.36E-05
1.09E-03
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-02b
1.58E-05
6.24E-03
2.19E-03
1.20E-06
4.83E-04
2.40E-05
5.00E-06
2.25E-06
4.01E-02
1.07E+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
2.61E-03
4.47E-02
3.90E-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
1.47
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
-------
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
127-18-4 Tetrachloroethylene
108-88-3 Toluene
8001-35-2 Toxaphene
120-82-1 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+02
5.26E+02
7.40E-01
3.00E+02
1.33E+03
4.42E+03
1.10E+03
1.20E+03
8.00E+02
2.00E+04
2.76E+03
1.61E+02
1.78E+02
1.85E+02
(HLC [atm-mVmol]
a HLC was calculated using the equation: HLC = vapor pressure
atm and molecular weight is 167.21 g/mol for carbazole.
b Value from WATERS model database.
HLC
(atm-m3/mol)
2.44E-08
3.97E-07
1.10E-05
2.75E-03
3.45E-04
1.84E-02
6.64E-03
6.00E-06
1.42E-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'
(dimensionless)
1.00E-06
1.63E-05
4.51E-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
3.01E-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
is6.83E-10
136
-------
Table 37. Air Diffusivity (Diia) and Water Diffusivity (DiiW) Values
for SSL Chemicals (25°C)a
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(Jb)fluoranthene
Benzo(/<)fluoranthene
Benzole acid
Benzo(a)pyrene
Bis(2-chloroethyl)ether
Bis(2-ethylhexyl)phthalate
Bromodichloromethane
Bromoform
Butanol
Butyl benzyl phthalate
Carbazole
Carbon disulfide
Carbon tetrachloride
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-Dichloroethylene
frans-1 ,2-Dichloroethylene
2,4-Dichlorophenol
Dia(cm2/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-02b
1.44E-02
1.37E-02
2.02E-02 b
4.38E-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
DiiW(cm2/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
-------
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
Dieldrin
Diethylphthalate
2,4-Dimethylphenol
2,4-Dinitrophenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Di-n-octyl phthalate
Endosulfan
Endrin
Ethylbenzene
Fluoranthene
Fluorene
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro-1 ,3-butadiene
a-HCH (a-BHC)
P-HCH (p-BHC)
y-HCH(Lindane)
Hexachlorocyclopentadiene
Hexachloroethane
lndeno(1 ,2,3-ccf)pyrene
Isophorone
Mercury
Methoxychlor
Methyl bromide
Methylene chloride
2-Methylphenol
Naphthalene
Nitrobenzene
/V-Nitrosodiphenylamine
/V-Nitrosodi-n-propylamine
Pentachlorophenol
Phenol
Pyrene
Styrene
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
Dja(cm2/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-02b
5.42E-02
5.61E-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-02b
5.45E-02 b
5.60E-02
8.20E-02
2.72E-02 b
7.10E-02
7.10E-02
7.20E-02
Djw(cm2/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
8.60E-06
6.35E-06 b
8.17E-06 b
6.10E-06
9.10E-06
7.24E-06 b
8.00E-06
7.90E-06
8.20E-06
138
-------
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-Trichlorobenzene
1,1,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,5-Trichlorophenol
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
m-Xylene
o-Xylene
p-Xylene
Dja(cm2/s)
8.70E-02
1.16E-02
3.00E-02
7.80E-02
7.80E-02
7.90E-02
2.91E-02
3.18E-02
8.50E-02
1.06E-01
7.00E-02
8.70E-02
7.69E-02
Djw(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.
a Value from CHEMDAT8 model database unless indicated otherwise.
b Estimated using correlations in WATERS model.
5.3 Soil Organic Carbon/Water Partition Coefficients (Koc)
Application of SSLs for the inhalation and migration to ground water pathways requires Koc values
for each organic chemical of concern. Koc 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 Koc values.
The soil-water partitioning behavior of nonionizing and ionizing organic compounds differs because
the partitioning of ionizing organics 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
-------
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., Kd or Kp) were reported along with the
organic carbon content of the soil. In these cases, Koc was computed by dividing IQ by the fractional
soil organic carbon content (foc, g/g). If the partition coefficient was normalized to soil organic
matter (i.e., Kom), it was converted to Koc as follows (Dragun, 1988):
Koc=1.724Kom (69)
where
1.724 = conversion factor from organic matter to organic carbon (fom = 1.724 foc)
Kom = partition coefficient normalized to organic matter (L/kg)
fom = fraction organic matter (g/g).
Once collected, KQ,, 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 IQ 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 Koc 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 KQ,, 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 hydrophobic organic compounds of interest.
It has long been noted that a strong linear relationship exists between KOC and Kow (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 log Kow) (70)
140
-------
For volatile organic compounds (VOCs), Equation 70 consistently overpredicted KQC 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 Koc = 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 r2 of 0.97. The compounds and data used to develop
this equation are provided in Appendix K. Equation 71 was used to calculate KQC 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
Organicsa
Koc (L/kg)
Compound
Acenaphthene
Aldrin
Anthracene
Benz(a)anthracene
Benzene
Benzo(a)pyrene
Bis(2-chloroethyl)ether
Bis(2-ethylhexyl)phthalate
Bromoform
Butyl benzyl phthalate
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
ODD
DDE
DDT
Dibenz(a,h)anthracene
1,2-Dichlorobenzene (o)
1,4-Dichlorobenzene (p)
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
Geometric
Mean
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
-------
Table 38 (continued)
Koc (L/kg)
Geometric
Compound Mean
frans-1 ,2-Dichloroethylene
1,2-Dichloropropane
1,3-Dichloropropene
Dieldrin
Diethylphthalate
Di-n-butylphthalate
Endosulfan
Endrin
Ethylbenzene
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
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
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
a See Appendix K for sources of measured values.
142
-------
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
124-48-1
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 a Kow
Acenaphthene
Acetone
Aldrin
Anthracene
Benz(a)anthracene
Benzene
Benzo(Jb)fluoranthene
Benzo(/<)fluoranthene
Benzo(a)pyrene
Bis(2-chloroethyl) ether
Bis(2-ethylhexyl)phthalate
Bromodichloromethane
Bromoform
Butanol
Butyl benzyl phthalate
Carbazole
Carbon disulfide
Carbon tetrachloride
Chlordane
p-Chloroaniline
Chlorobenzene
Chlorodibromomethane
Chloroform
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-Dichloroethylene
frans-1 ,2-Dichloroethylene
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
Log Koc
(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
1.50
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
1.51E+07
5.50E+01
8.71E+01
6.92E+00
5.75E+04
3.39E+03
4.57E+01
1.74E+02
1.20E+05
6.61E+01
2.19E+02
6.31E+01
3.98E+01
3.98E+05
1.00E+06
4.47E+06
2.63E+06
3.80E+06
3.39E+04
6.17E+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
KOC
(L/kg)
4.90E+03
—
4.87E+04
2.35E+04
3.58E+05
6.17E+01
—
—
9.69E+05
7.59E+01
1.11E+05
—
1.26E+02
—
1.37E+04
—
—
1.52E+02
5.13E+04
—
2.24E+02
—
5.25E+01
—
4.58E+04
8.64E+04
6.78E+05
1.79E+06
1.57E+03
3.79E+02
6.16E+02
—
5.34E+01
3.80E+01
6.50E+01
—
3.80E+01
4.70E+01
2.71E+01
143
-------
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
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
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 a Kow
Dieldrin
Diethylphthalate
2,4-Dimethylphenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Di-n-octyl phthalate
Endosulfan
Endrin
Ethylbenzene
Fluoranthene
Fluorene
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro-1 ,3-butadiene
a-HCH (a-BHC)
p-HCH (p-BHC)
y-HCH(Lindane)
Hexachlorocyclopentadiene
Hexachloroethane
lndeno(1 ,2,3-ccf)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
Log Koc
(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
KOC
(L/kg)
2.14E+04
2.88E+02
2.09E+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
1.05E+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+01
1.55E+02
1.82E+02
2.57E+05
1.78E+03
Measured
KOC
(L/kg)
2.55E+04
8.22E+01
—
—
—
—
2.04E+03
1.08E+04
2.04E+02
4.91E+04
7.71E+03
9.53E+03
—
8.00E+04
—
1.76E+03
2.14E+03
1.35E+03
—
—
—
—
8.00E+04
9.00E+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
-------
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
Compound
1,1,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
Vinyl acetate
Vinyl chloride
m-Xylene
o-Xylene
p-Xylene
Chemical Log
Group a KOW
2
2
2
1
2
2
2
2
2
2
2
0.
1.
3.
3.
3.
.48
.05
.71
.73
.50
.20
.13
.17
Log Koc
(L/kg)
2
1
2
0
1
2
2
2
.04
.70
.22
.72
.27
.61
.56
.59
Calculated
KOC
(L/kg)
1.10E+02
5.01E+01
1.66E+02
5.25E+00
1.86E+01
4.07E+02
3.63E+02
3.89E+02
Measured
KOC
(L/kg)
1.
7.
9.
1.
2
3.
.35E+02
.50E+01
.43E+01
—
—
.96E+02
.41E+02
.11E+02
Group 1: log K^ = 0.983 log Kow + 0.00028.
Group 2: (VOCs, chlorobenzenes, and certain chlorinated pesticides) log Koc = 0.7919 log KQW + 0.0784.
Note: Calculated values rounded as shown for subsequent SSL calculations.
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
A/-Nitrosodiphenylamine
A/-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, the Koc values for the neutral
and ionized forms (Kocn and KQC ;) 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, Koci for organic acids is likely to be less than
In the case of organic bases, the
ionized species is positively charged (HB+) so that Koc ; is likely to be greater than K
oc n.
145
-------
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,
positively charged chemicals have a greater tendency to sorb than 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):
o = 2^3 = d + iQpwy1
n'acid [HA] + [A'] l ' (72)
where
^acid = 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:
n £ase
l Q pKa-pH V
[Ef] + [HB + ] V ' (73)
where
^base = fraction 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
-------
that exceed the pKa, and the ionized species predominates at pH values less than the pKa. For the
SSL organic bases, 7V-nitrosodi-«-propylamine and TV-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 that 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 benzoic 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, O) as a
Function of pH
Compound
Benzoic acid
p-Chloroanilineb
2-Chlorophenol
2,4-Dichlorophenol
2,4-Dimethylphenol
2,4-Dinitrophenol
2-Methylphenol
/V-Nitrosodiphenylamineb
A/-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
aKolligetal. (1993).
b Denotes that the compound is an organic base.
cLeeetal. (1991).
147
-------
Prediction of Soil-Water Partition Coefficients. Lee et al. (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:
Koc = KOC;n(Dn + Koc,1(l-(Dn) (74)
where
Koc = soil organic carbon/water partition coefficient (L/kg)
Koc,n = partition coefficient for the neutral species (L/kg)
On = fraction of neutral species present for acids or bases
Koc,i = 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 KQC values for
pentachlorophenol.
A literature review was conducted to compile the pKa and the laboratory-measured values of Koc n
and Koci 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 Koc; 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 Koci from Kp and soil foc (Lee et al., 1991). From
these measured values, the ratios of K^ to Kocn 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 Koc; 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 (7V-nitrosodi-«-propylamine,
TV-nitrosodiphenylamine, and />-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 7V-nitrosodi-«-propylamine (pKa < 1) and 7V-nitrosodiphenylamine (pKa < 0), the neutral
species is present under environmentally significant conditions. The neutral species constitutes
approximately 90 percent of the system for/>-chloroaniline (Table 40).
148
-------
Table 41. Soil Organic Carbon/Water Partition Coefficients and pKa
Values for Ionizing Organic Compounds
Compound
Benzole 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
Koc,n (L/kg)
32 b
398b
159d
0.8a
19,953e
17,916f
6,190'
2,380'
1,070'
KOCii(L/kg)
0.5C
6.0C
2.4C
0.01C
398e
67g
93C
36j
107k
pKaa
4.18
8.40
7.90
3.30
4.80
6.35h
5.30
7.10
6.40
a Kollig etal. (1993).
b Meylan et al. (1992).
c Estimate based on the ratio of Koc j/Kocn for compounds for which data exist; Koc j was estimated to be 0.015 x
'xjc.n •
d Calculated using data (Kp = 0.62, foc = 0.0039) contained in Lee et al. (1991); agrees well with Boyd (1982)
reporting measured KOC = 126 L/kg.
e Leeetal. (1990).
f Average of values reported fortwo aquifer materials from Schellenberg et al. (1984).
g Calculated using data (Kp = 0.26, foc = 0.0039) contained in Lee etal. (1991).
h Leeetal. (1991).
i Schellenberg et al. (1984).
J Calculated using data (Kp = 0.14, foc = 0.0039) contained in Lee etal. (1991).
k 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 7V-nitrosodi-«-propylamine, jV-nitrosodiphenylamine, and />-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^ n and Koc;, 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 pHs 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 (IQ) 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
-------
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 Kd values for metals than for organics.
Table 42. Predicted Soil Organic Carbon/Water Partition Coefficients
(Koc>l-/kg) as a Function of pH: Ionizing Organics
Compound
Benzole 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 = 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 = 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 Kd 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 IQ 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 IQ 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 IQ 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 the 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
-------
Table 43. Summary of Collected Kd Values Reported in Literature
Metal
Antimony
Arsenic6
Arsenic (+3)
Arsenic (+5)
Barium
Beryllium
Cadmium
Chromium
Chromium (+2)
Chromium (+3)
Chromium (+6)
Mercury6
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
Baes et al. (1984)"
Geometric
Mean6
45f
200f
3.39
6.79
6Qf
650f
6.4h
850f
2,2009
-
379
10f
150f
30Qf
46^
1,500f
1,000f
38h
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
-
15
-
18
-
—
—
16
—
—
146
Coughtrey et
al. (1985)c
Range
-
—
-
-
—
-
32-50
-
-
-
-
-
-20
<9
50
—
—
>20
Battelle
(1989)d
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
-
a The Atomic Energy of Canada, Limited (AECL, 1990) presents the distribution of Kd values according to four major
soil types—sand, silt, clay, and organic material. Their data were obtained from available literature.
b Baes et al. (1984) present Kd 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 Kj values for a limited number of metals.
d Battelle Memorial Institute (Battelle, 1989) reports a range in Kd values as a function of pH (5 to 9) and sorbent
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.
6 The valence of these metals is not reported in the documents.
f Estimated based on the correlation between Kj and soil-to-plant concentration factor (Bv).
9 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, 1994i) and were used to calculate the SSLs presented in the December 1994 draft Soil Screening
Guidance (U.S. EPA, 1994h).
151
-------
The MINTEQA2 model was further updated by Allison Geoscience Consultants, Inc., in 1995 to
include thermodynamic 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 IQ values for the EPA/OSW HWIR
modeling effort. The metal IQ 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 these parameters is the oxidation-reduction (redox) potential. To compensate, both modeling
efforts 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.
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
-------
for HWIR. Using these relationships, the Kd 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, Kd values for antimony and vanadium were obtained from Baes
et al. (1984) (Table 43). These Kd 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
0.01
0.31
1.11
percent)
Source: U.S. EPA(1992a)
153
-------
Figure 10. Empirical pH-dependent adsorption relationship: arsenic (+3), chromium
154
-------
Table 45. Background Pore-Water Chemistry Assumed for SSL MINTEQ
Modeling Efforts
Parameter
Aluminum
Bromine
Calcium
Carbonate
Chlorine
Iron (+3)
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
2.9b
22
25
a Median values from STORE! database as reported in U.S. EPA (1992a).
b 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.
Redox 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
thermodynamic 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 IQ values. The redox-sensitive
155
-------
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 exist 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 Kds 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 Kds were used.
Table 46 shows the SSL Kd values at high, medium, and low subsurface pH conditions. Figure 11 plots
MINTEQ-derived metal Kd values over this pH range. Figure 10 shows the same for the empirically
derived metal Kjs. 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 IQ 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 IQ 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
-------
G.
5 _
O) ^
§> 6
o
0
3.5 4
«*<
*
x
^
x
g
X
^
x
x
^
^
0 4
^
f
*f
,••
t
**
*f
f
/
^
(
/
ff
f
S
*•
**
0
••
-
^
—
*
y
i*
^
^
^
^
f
^
f
"
0
f
«
^
S
y
^
•
.
•
^
x
x
x
••
^
x
^
(
5 5.0 5.5 6.0 6
PH
x
X
^
^
x
«
^
^
eS
,**
(
^
z=
/
^
x
X
/
^
x
••
x
(
*t
x
^
^
y
X
x
^
x
^
^
^
Chromium (+3]
(
'
f
s
f
t
1
x
^
Beryll
um
Cadmium
^
^
\l
c
se
5 7.0 7.5 8.0 8
5
0
O
•j _
I
x
§>
.•I .
-2
.3 .
7inr
Barium
S
Iv
er
Mercury
i i
^
^
x
^
/
^
x
/
,«•
^
^
/
^
/
3.5 4.0 4
••
x
^
*•
*r
*
t
1
x
/
/
*f
x
/
••
/
,x
/
^
/
^
^
—
••
s-
— •
/
/
X
^
/
^a
/
x
/
X
—
'
/
x
x
5 5.0 5.5 6.0 6
PH
x
i~
7
2
X
X
X
x
tf
^
x
•;
*
*•
&
x
5 7.0 7
x
^
^
^
^
^
'
**
x
h
Zinc H
Mercury
Silver
- Barium
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
-------
Table 46. Estimated Inorganic Kd Values for SSL Application
Metal
Antimony3
Arsenic (+3)&
Barium
Beryllium
Cadmium
Chromium (+3)
Chromium (+6)b
Cyanide0
Mercury (+2)
Nickel
Seleniumb
Silver
Thalliumb
Vanadium3
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
1.0E-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+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 et al., 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 IQ 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 K^ 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 IQ values estimated for beryllium range from 23 to 100,000 L/kg for the
conditions studied. AECL (1990) reports medians of observed values for IQ 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 Kd values and values reported in the literature.
Cadmium. For the three pH conditions, MINTEQ IQ 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 IQ values for chromium (+3) range from 1,200 to
4,300,000 L/kg. Battelle (1989) reports a range of Kd values of 168 to 3,600 L/kg, orders of
158
-------
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) IQ 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 IQ 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 Kj 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
Baesetal. (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 IQ values for zinc range from 16 to 530 L/kg. These estimated Kd values
are within the range of measured IQ 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 IQ value for zinc of
greater than or equal to 20 L/kg.
159
-------
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 Kj 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
-------
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, Kj, 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 Kj 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(1): 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(5):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. Parrish. 1988. Developing joint probability distributions of soil water retention
characteristics. Water Resources Research, 24(5):755-769.
161
-------
Carsel, R.F., R.S. Parrish, R.L. Jones, J.L. Hansen, and R.L. Lamb. 1988. Characterizing the
uncertainty of pesticide leaching in agricultural soils. J. of Contain. 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. Thorne. 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
Paniculate 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.
Criscenti, L.J., S.B. Christian, E.Z. Hosseinipour, and V.M. Gorokhovski. 1994. Evaluation of
Unsaturated Zone/Vadose Zone Models for Superfund 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 Hwang 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 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, 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
-------
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 I. 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. Farmer, 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
-------
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 B.E. Rittmann. 1981. Trace organics in ground water. Environ. Sci.
Technol. 15:40-51.
McLean, E.G. 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, C.J., L.P. Hopkins, and P.B. Bedient. 1990. A hydrogeologic database for ground water-
modeling. Ground Water, 28(5):703-714.
164
-------
Nofziger, D.L., J. Chen, and C.T. Haan. 1994. Evaluation of Unsaturated/Vadose 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. Hydrol. 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 of Hexachlorobenzene 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
-------
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 9355.0-30. 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/600/8-91/01 IB. 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
-------
U.S. EPA (Environmental Protection Agency). 1993b. Data Quality Objectives for Superfund:
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 of Polycyclic 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. NTIS 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 Superfund
(RAGS;, Human Health Evaluation Manual (RREM). 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 I: 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. Framework 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 I,I-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
-------
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. Kerr 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 [14C]Benzo[a]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
-------
APPENDIX A
Generic SSLs
-------
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 site 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 that 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 major pathways of concern in both 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 (Csat), 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 Csat 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
-------
Table A-1. Generic SSLs a
Organics
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
Migration to ground water
Compound
Acenaphthene
Acetone
Aldrin
Anthracene
Benz(a)anthracene
Benzene
Benzo(Jb)fluoranthene
Benzo(/<)fluoranthene
Benzole acid
Benzo(a)pyrene
Bis(2-chloroethyl)ether
Bis(2-ethylhexyl)phthalate
Bromodichloromethane
Bromoform
Butanol
Butyl benzyl phthalate
Carbazole
Carbon disulfide
Carbon tetrachloride
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-Dichloroethylene
frans-1 ,2-Dichloroethylene
2,4-Dichlorophenol
Ingestion
(mg/kg)
4,700 b
7,800 b
0.04 e
23,000 b
0.9 e
22 e
0.9 e
9 e
3.1E+05 b
0.09 e>f
0.6 e
46 e
10 e
81 e
7,800 b
16,000 b
32 e
7,800 b
5 e
0.5 e
310 b
1,600 b
8 e
100 e
390 b
88 e
3 e
2 e
2 e
0.09 e>f
7,800 b
7,000 b
27 e
1 e
7,800 b
7 e
1 e
780 b
1,600 b
230 b
Inhalation
volatiles
(mg/kg)
c
1.0E+05 d
3 e
— C
— C
0.8 e
c
c
c
c
0.2 e>f
31,000 d
3,000 d
53 e
10,000 d
930 d
c
720 d
0.3 e
20 e
c
130 b
1,300 d
0.3 e
53,000 d
— C
c
c
— g
c
2,300 d
560 d
... g
c
1,300 b
0.4 e
0.07 e
1,200 d
3,100 d
— C
20 DAF
(mg/kg)
570 b
16 b
0.5 e
12,000 b
2 e
0.03
5 e
49 e
400 b>'
8
0.0004 e>f
3,600
0.6
0.8
17 b
930 d
0.6 e
32 b
0.07
10
0.7 b
1
0.4
0.6
4 b,i
160 e
16 e
54 e
32 e
2 e
2,300 d
17
2
0.007 e>f
23 b
0.02
0.06
0.4
0.7
1 b,i
1 DAF
(mg/kg)
29 b
0.8 b
0.02 e
590 b
0.08 e>f
0.002 f
0.2 e.f
2 e
20 b>'
0.4
2E-05 e>f
180
0.03
0.04
0.9 b
810 b
0.03 e>f
2 b
0.003 f
0.5
0.03 b.f
0.07
0.02
0.03
0.2 b.f.'
8 e
0.8 e
3 e
2 e
0.08 e>f
270 b
0.9
0.1 f
0.0003 e>f
1 b
0.001 f
0.003 f
0.02
0.03
0.05 b>f>'
A-2
-------
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
Migration to ground water
Compound
1,2-Dichloropropane
1,3-Dichloropropene
Dieldrin
Diethylphthalate
2,4-Dimethylphenol
2,4-Dinitrophenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Di-n-octyl phthalate
Endosulfan
Endrin
Ethylbenzene
Fluoranthene
Fluorene
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Hexachloro-1 ,3-butadiene
a-HCH (a-BHC)
P-HCH (P-BHC)
Y-HCH(Lindane)
Hexachlorocyclopentadiene
Hexachloroethane
lndeno(1 ,2,3-ccf)pyrene
Isophorone
Mercury
Methoxychlor
Methyl bromide
Methylene chloride
2-Methylphenol
Naphthalene
Nitrobenzene
/V-Nitrosodiphenylamine
/V-Nitrosodi-n-propylamine
PCBs
Pentachlorophenol
Phenol
Pyrene
Styrene
1 ,1 ,2,2-Tetrachloroethane
Ingestion
(mg/kg)
9 e
4 e
0.04 e
63,000 b
1,600 b
160 b
0.9 e
0.9 e
1,600 b
470 b
23 b
7,800 b
3,100 b
3,100 b
0.1 e
0.07 e
0.4 e
8 e
0.1 e
0.4 e
0.5 e
550 b
46 e
0.9 e
670 e
23 b>'
390 b
110 b
85 e
3,900 b
3,100 b
39 b
130 e
0.09 e,f
1 h
3 ej
47,000 b
2,300 b
16,000 b
3 e
Inhalation
volatiles
(mg/kg)
15 b
0.1 e
1 e
2,000 d
c
c
c
c
10,000 d
— C
c
400 d
c
c
4 e
5 e
1 e
8 e
0.8 e
... g
c
10 b
55 e
— C
4,600 d
10 b>'
c
10 b
13 e
— C
— C
92 b
c
c
... h
c
— C
— C
1,500 d
0.6 e
20 DAF
(mg/kg)
0.03
0.004 e
0.004 e
470 b
9 b
0.3 W'
0.0008 e,f
0.0007 e,f
10,000 d
18 b
1
13
4,300 b
560 b
23
0.7
2
2
0.0005 e,f
0.003 e
0.009
400
0.5 e
14 e
0.5 e
2 '
160
0.2 b
0.02 e
15 b
84 b
0.1 b>f
1 e
5E-05 e,f
... h
0.03 f>i
100 b
4,200 b
4
0.003 e,f
1 DAF
(mg/kg)
0.001 f
0.0002 e
0.0002 e,f
23 b
0.4 b
0.01 W'
4E-05 e,f
3E-05 e,f
10,000 d
0.9 b
0.05
0.7
210 b
28 b
1
0.03
0.1 f
0.1 f
3E-05 e,f
0.0001 e,f
0.0005 f
20
0.02 e,f
0.7 e
0.03 e,f
0.1 '
8
0.01 b>f
0.001 e,f
0.8 b
4 b
0.007 b>f
0.06 e,f
2E-06 e,f
... h
0.001 f>i
5 b
210 b
0.2
0.0002 e,f
A-3
-------
Table A-1 (continued)
Organics
CAS No.
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
Migration to ground water
Compound
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,5-Trichlorophenol
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
m-Xylene
o-Xylene
p-Xylene
Ingestion
(mg/kg)
12 e
16,000 b
0.6 e
780 b
c
11 e
58 e
7,800 b
58 e
78,000 b
0.3 e
1.6E+05 b
1.6E+05 b
1.6E+05 b
Inhalation
volatiles
(mg/kg)
11 e
650 d
89 e
3,200 d
1,200 d
1 e
5 e
c
200 e
1,000 b
0.03 e
420 d
410 d
460 d
20 DAF
(mg/kg)
0.06
12
31
5
2
0.02
0.06
270 b>'
0.2 e-f-'
170 b
0.01 f
210
190
200
1 DAF
(mg/kg)
0.003 f
0.6
2
0.3 f
0.1
0.0009 f
0.003 f
14 b>'
0.008 e>f>'
8 b
0.0007 f
10
9
10
A-4
-------
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
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 e
5,500 b
0.1 e
78 b'm
390 b
78,000 b
390 b
1,600 b
400 k
1,600 b
390 b
390 b
... c
550 b
23,000 b
Inhalation
fugitive
particulate
(mg/kg)
c
750 e
6.9E+05 b
1,300 e
1,800 e
270 e
... c
270 e
c
... k
13,000 e
c
c
— c
— c
c
Migration to
20 DAF
(mg/kg)
5
29 '
1,600 '
63 '
8 j
38 '
...9
38 '
40
... k
130 '
5 j
34 b'
ground water
1 DAF
(mg/kg)
0.3
1 j
82 '
3 j
0.4 '
2 j
...9
2 j
2
... k
7 j
0.3 '
2 b.i
0.04 '
300 b
620 b>'
DAF = Dilution and attenuation factor.
a Screening levels based on human health criteria only.
Calculated values correspond to a noncancer hazard quotient of 1.
c No toxicity criteria available for that route of exposure.
Soil saturation concentration (Csat).
e 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).
Chemical-specific properties are such that this pathway is not of concern at any soil contaminant concentration.
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. EPA, 1990) and on EPA efforts to manage PCS contamination.
' SSL for pH of 6.8.
J Ingestion SSL adjusted by a factor of 0.5 to account for dermal exposure.
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).
1 SSL is based on RfD for mercuric chloride (CAS No. 007487-94-7).
m SSL is based on dietary RfD.
h
A-5
-------
Table A-2. Generic SSLs: Default Parameters and Assumptions
Parameter
SSL pathway
Inhalation
Migration to
ground water
Default
Source Characteristics
Continuous vegetative cover
Roughness height
Source area (A)
Source length (L)
Source depth
O
•
O
O
50 percent
0.5 cm for open terrain; used to derive Ut 7
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 O
Dry soil bulk density (pb) •
Soil porosity (n) •
Vol. soil water content (0W) *
Vol. soil air content (0a) •
Soil organic carbon (foc) •
Soil pH O
Mode soil aggregate size O
Threshold windspeed @ 7 m (Ut 7) •
O Loam; defines soil characteristics/
parameters
• 1.5kg/L
O 0.43
• 0.15(INH);0.30(MTG)
• 0.28 (INH); 0.13 (MTG)
• 0.006 (0.6%, INH); 0.002 (0.2 %, MTG)
O 6.8; used to determine pH-specific Kd
(metals) and KOC (ionizable organics)
0.5 mm; used to derive U, 7
11.32m/s
Meteorological Data
Mean annual windspeed (Um)
Air dispersion factor (Q/C)
Volatilization Q/C
Fugitive particulate Q/C
4.69 m/s (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)
O 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
-------
Analysis of Effects of Source Size on Generic SSLs
A large number of commenters 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 this 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
-------
(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 this 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
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 PB91-921206CDH.
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
-------
Table A-3.Source Depth where 30-acrea Mass-Limit SSLs = 0.5-acreb
Infinite-Source SSLsc
Chemical
Acetone
Benzene
Benzole acid
Bis(2-chloroethyl)ether
Bromodichloromethane
Bromoform
Butanol
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chlorodibromomethane
Chloroform
2-Chlorophenol
1,2-Dichlorobenzene
1,4-Dichlorobenzene
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
c/s-1 ,2-Dichloroethylene
frans-1 ,2-Dichloroethylene
2,4-Dichlorophenol
1,2-Dichloropropane
1,3-Dichloropropene
2,4-Dimethylphenol
2,4-Dinitrophenol
2,4-Dinitrotoluene
2,6-Dinitrotoluene
Ethylbenzene
Methyl bromide
Methylene chloride
2-Methylphenol
Nitrobenzene
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
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 water0
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
-------
Table A-3. (continued)
Source depth (m)
Chemical Inhalation Migration to ground water
Vinyl chloride 21 13
m-Xylene NA 4
o-Xylene NA 4
p-Xylene NA 4
NA = Risk-based SSL not available.
a Q/C = 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 m/yr infiltration rate.
A-10
-------
APPENDIX B
Route-to-Route Extrapolation of Inhalation
Benchmarks
-------
APPENDIX B
Route-to-Route Extrapolation of Inhalation Benchmarks
Introduction
For a number of the 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 route-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) = RfD (mg / kg - d;
70kg (B-l)
20m3/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 pharmacokinetics.
B-l
-------
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 sulfide)
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 (CSForai)
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:
3 l CSF^mg/kg-d)-1 3 3 (B-2)
URF (^ig/m3)"1 = ?()k x 20m3/d x 10"3 mg/^ig .
Using the extrapolated URF, risk-specific air concentrations were calculated as a lifetime average
exposure concentration as shown in equation B-3:
3 target risk 10"6 (B-3)
extrapolated air concentration |lg / m =
URF (|lg/m3)
Not surprisingly, the risk-based (i.e., 1O6) 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 pharmacokinetic 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 pharmacokinetic behavior. For example, 1,2-dibromomethane has an extrapolated
lO-6 air concentration that is 2 orders of magnitude below the value in IRIS. This difference is
probably 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* (Csat)
and generic SSLs for soil ingestion (SSL;ng) 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 Csat and its significance is discussed in Section 2.4.4 of this Technical Background Document.
B-2
-------
described in Equations B-l and B-2 (columns 4 and 5). In addition, the table presents volatilization-
based SSLs and SSLs based on particulate 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 SSL^ for volatilization (SSLj^.y) was calculated with Equation 4 in Section 2.4
using a chemical-specific volatilization factor (VF). In Table B-l, the SSLinh.v values based on
extrapolated inhalation benchmarks (column 6) are compared with the soil saturation concentration
(Csat, column 7) and generic migration to ground water SSLs assuming a dilution attenuation factor
(DAF)of20(SSLgw).
As described in Section 2.4.4, Csat represents the concentration at which soil pore air is saturated with
a chemical and maximum volatile emissions are reached. A comparison of the Csat with the
extrapolated SSLj^.y values indicates that, for 36 of the 50 contaminants, SSLj^.y exceeds the soil
saturation concentration, often by several orders of magnitude. Because maximum volatile emissions
occur at Csat, 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 SSLj^.y values below Csat, all are above the
generic SSLgw values. This analysis suggests that SSLs based on the migration-to-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, bromodichloromethane, chlorodibromomethane, cis-
1,2-dichloroethylene, and fra«5-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 Particulate Emissions
The extrapolated particulate inhalation SSLs (SSL^.p) were calculated with Equation 4 in Section 2.4
using the particulate emission factor (PEF) of 1.32 x 109 m3/kg. Table B-l compares the SSL^.p
values based on extrapolated benchmarks (column 10) and generic SSLs based on direct ingestion
(SSLing, Column 9). This comparison indicates that the extrapolated SSLin]l_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
-------
often significantly lower, than both extrapolated and IRIS-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 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. 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 that 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 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 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. Methods for Derivation of Inhalation
Reference Concentrations and Application of Inhalation Dosimetry. EPA/600/8-90/066F.
Office of Research and Development, Washington, DC.
B-4
-------
(O
w.
1_
0)
•4-i
i
•o
c
i
O
o
•4-i
c
o
'v^
5
O)
c
re
O,
in
c
o
'vj
2
•4-i
0)
o
c
o
o
'o
(O
(O
w.
in
(O
c
o
"re
_c
•o
0)
o
Q.
2
•4-i
X
LJJ
"S
C
o
(A
'C
re
a.
E
o
O
m
re
U)
•D "™
re c
•9 —
LL
II 1
Q.
I
I
I/I
i
•D
I/I
LL
C
O
re
"re i/>
c Is
•D J
•" O
2 c
o
•D C
a) o
re ~i
II
2 S
£ JS
X 0
LJJ >
LL
3
o
JS
^~
LL
CO
O
s
0£
,
_f
"5
u
re
Q.
_f
o1
Ol
LL
Q
S
O
=•
.n
_f
tn
I
D1
I
Ol
5.
Ol
"Bi
£
*•"*
^-^
"9
51
"Si
£
Compound
§o
Q
Q
§o
Q
Q
TT T~
0 0
o o
1 — 00
° «°
in '"
h.
i «-
o
§1.
OO ^
9 9
LJJ LJJ
*- in
CM CO
CM T-
9 9
LJJ LJJ
CO T-
Acenaphthene
Acetone
0
8 o
8 §
o m
TT
O
8 co
CO" °
CM
O
O
O CM
CM
00 CM
O
8 o
o ^
CO
CO
9
LJJ
CM
8
LJJ
^~
O
LJJ
CO
1^
^
O
LJJ
CO
Anthracene
Benz(a) anthracene
o
o
0
in"
CO
d
in
^
in
9
LJJ
CM
O
LJJ
CO
l~~
(D
Benzo(b) fluoranthen
o
8
o
in
CO
»
CO
8
CO
^~
in
p
LJJ
CM
CM
p
LJJ
CO
l~~
(D
Benzo(k) fluoranthen
8
in
CO
o
d
CO
o
CO
CM
CO
p
LJJ
CM
o
0
LJJ
CO
Benzo(a) pyrene
o
o
0
o"
o
0
TT~
o"
CO
1
8
o
o
o
o"
o
o
•ft-
o
LJJ
"*"
o
o
LJJ
Benzole acid
0
o
o
8
00
s?
o
o
CO
co"
s
CO
s
o
o
o
o"
CO
8
LJJ
0
^f
LJJ
~^-
•^
(D
"TO
"m
Bis(2-ethylhexyl)phth
8
o
o"
CO
o
CO
o
,_
CO
o>
of
?
^
g
LJJ
CO
••-
LJJ
C\l
CO
,
1 HI!
-------
(O
w.
I_
s
I
•o
c
I
O
2
c
o
'^
5
O)
c
re
(A
C
o
'^
2
•4-i
0)
o
c
o
o
'o
(O
(O
(O,
(A
(O
(O
c
o
re
JZ
c
o
Q.
2
•4-i
X
UJ
II-
o
c
o
W)
'C
re
a.
E
o
O
m
re
00
w _
•o 5>
« "o>
II
UJ
Q.
"3
^i
E.
U)
W
W
I
re
•9
LL
.0
]i
_c ^
.E re
«l
re
•5
Ul
U)
E
u
c
15
0
^
—
9
rapolated
ulate SSLlnl
•55 o
Ul t
Q
U
1 5?
O W
J
W o
U LL
11
;_»
u. "c
o: 5
3 c1
•=•
«!
E
^
?
w 51
0 0)
g
?
S £
a 01
_E
Compound
o
o
o
o"
o
o
"T
o
0
co"
I
s
o
8
in
o
LLJ
"^
"
CM
9
LLJ
^
Fluorene
o
8
00
in
0
CO
8
o
?
CO
^.
cp
LLJ
CO
0
O
LLJ
CO
"-
CO
c
ra
X
CO
JZ
y-Hexachlorocyclol
o
ll
A
§ 1
•3- m
•>- o
o
in r-
8 CM
CD r-
f r-
9 9
LLJ LLJ
CM CM
9?
LLJ LLJ
co in
r^ oi
CO
G)
>••
Indenof 1 ,2,3-cd) \
Isophorone
,000,000
,000,000
A A
8|
8 ¥.
s|
O O
So
o
1^ CO
CM T-
9 9
LLJ LLJ
00 00
CO CM
9 9
LLJ LLJ
in in
Methoxychlor
2-Methylphenol
,000,000
,000,000
A A
co" *~
S -
12 12
CO CM
O O
o o
CM O
00" T-~
CD
cp
LLJ
•^
O
LU
"^
"
CO
cp
LLJ
CO
"*
CM
9
LLJ
^
CO
ra
Naphthalene
N -Nitrosodipheny
II
CO
CO
O CO
d
in
8 9
O O
d
CO T-
CM" K
T CO
o oo
co in
9 9
LLJ LLJ
O t
CM CO
O ^
? 9
LLJ LLJ
0 CM
r- -
CO
,g
Q. —
N-Nitrosodi-n -pro
Pentachloropheno
,000,000
,000,000
A A
§O
O
- m.
O o
O CM
00
00 „
in g
CM"
CM
400,000
420,000
?9
LLJ LLJ
T- -^
CM ^~
T- CM
9 9
LLJ LLJ
CD CO
g g
§ S.
-C >,
CL CL
,000,000
,000,000
A A
o
0 0
o o
°° o"
1^ CD
0 0
CM CM
00
to 2
II
0 y,-
9?
LLJ LLJ
in o
cd r-^
T- 0
9 ?
LLJ LLJ
•<- CM
O
C
2,4,5-Trichlorophe
m -Xylene
,000,000
,000,000
A A
O O
O O
O O
o" o"
CD CD
0 0
CO O
T- CM
CO T-
8 8
o o
in" T-~
8 8
+ +
LLJ LLJ
q q
i^ i^
0 0
o o
LLJ LLJ
CM CM
Q) Q)
C C
CO CO
x" x
o 1
c
O
to
a)
O)
,-v .E
"S" ^
._ ^
"*-* "O
o o
11
° S
co -Q
>- «
C Q)
(0 =
2 >
•^ w
c W
= =
c *%
.b (/)
CO —
= tf)
g I
Ji, >
E 1
Q 1
Q. (/)
0 W
o" is
o .2
°- a
c. S
1 1
*- Q)
CO S
ra «
NR = SSLlnh_v isgre
Bold indicates wl
CQ
-------
APPENDIX C
Limited Validation of the Jury Infinite
Source and Jury
Finite Source Models (EQ, 1995)
-------
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
-------
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-D30035. 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.
-------
CONTENTS
Figures iv
Tables vi
Acknowledgment vii
1 . Introduction C-1
Project objectives C-2
Technical approach C-2
2. Review of the Jury Volatilization Models C-3
Finite source model derivation C-4
Infinite source model derivation C-6
Summary of model assumptions and limitations C-7
3 . Model Validation C-9
Validation of the Jury Infinite Source Model C-9
Validation of the Jury Reduced Solution Finite Source Model C-31
4. Parametric Analysis of the Jury Volatilization Models C-35
Affects of soil parameters C-35
Affects of nonsoil parameters C-36
5. Conclusions C-38
References C-40
APPENDICES
A. Validation Data for the Jury Infinite Source Model C-43
B. Validation Data for the Jury Reduced Solution Finite Source Model C-58
-------
FIGURES
Number Page
1 Predicted and measured emission flux of dieldrin versus time
(C0 = 5 ppmw) C-12
2 Comparison of log-transformed modeled and measured emission
flux of dieldrin (C0 = 5 ppmw) C-14
3 Predicted and measured emission flux of dieldrin versus time
(C0 = 10 ppmw) C-15
4 Comparison of log-transformed modeled and measured emission
flux of dieldrin (C0 = 10 ppmw) C-16
5 Predicted and measured emission flux of lindane versus time
(C0 = 5 ppmw) C-17
6 Predicted and measured emission flux of lindane versus time
(C0 = 10 ppmw) C-18
7 Comparison of log-transformed modeled and measured emission
flux of lindane (C0 = 5 ppmw) C-19
8 Comparison of log-transformed modeled and measured emission
flux of lindane (C0 = 10 ppmw) C-20
9 Predicted and measured emission flux of benzene (C0 =110 ppmw) C-24
10 Predicted and measured emission flux of toluene (C0 = 880 ppmw) C-25
11 Predicted and measured emission flux of ethylbenzene
(C0 = 310 ppmw) C-26
12 Comparison of log-transformed modeled and measured emission
flux of benzene (C0 =110 ppmw) C-28
13 Comparison of log-transformed modeled and measured emission
flux of toluene (C0 = 880 ppmw) C-29
14 Comparison of log-transformed modeled and measured emission
flux of ethylbenzene (C0 = 310 ppmw) C-30
15 Predicted and measured emission flux of triallate versus time C-33
16 Comparison of log-transformed modeled and measured emission
flux of triallate C-34
C-v
-------
TABLES
Number Page
1 Volatilization Model Input Values for Lindane and Dieldrin C-11
2 Summary of the Bench-Scale Validation of the Jury Infinite
Source Model C-21
3 Volatilization Model Input Variables for Benzene, Toluene,
and Ethylbenzene C-23
4 Summary of Statistical Analysis of Pilot-Scale Validation C-27
5 Volatilization Model Input Values for Triallate C-32
C-vi
-------
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.
-------
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 (TBD) 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 volatiles 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 volatiles, 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 (m3/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 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 Bl), 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-l
-------
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-2
-------
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:
Js = C0 e-m (DE/p t)1/2 [1 - exp (-L2/4 DE t)] (1)
where Js = Instantaneous emission flux, ^ug/cm2 -day
C0 = Initial soil concentration (total volume), ^g/cm3-soil
[i = Degradation rate constant, 1 /day
t = Time, days
DE = Effective diffusion coefficient, cm2 /day
L = Depth from the soil surface to the bottom of contamination, cm
and,
DE = [(a1073 D* KH + Q110/3 Df)/f2]/(pb foc Koc + Q + aKH) (2)
where DE = Effective diffusion coefficient, cm2 /day
a = Soil volumetric air content, cmVcm3
Dga = Gaseous diffusion coefficient in air, cm2/day
KH = Henry's law constant, unitless
0 = Soil volumetric water content, cm3/cm3
Df = Liquid diffusion coefficient in pure water, cm2/day
(f) = Total soil porosity, unitless
pb = Soil dry bulk density, g/cm3
foc = Soil organic carbon fraction
Koc = 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 = C0 att = 0, 0 < x < L
c = 0 at t = 0, x >- L
C-3
-------
c = 0 at t >- 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 (Jsavg) 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:
Js = C0 (DE/7Tt)1/2 (3)
where Js = Instantaneous emission flux, [ig/cm2-day
C0 = Initial soil concentration (total volume), ^g/cm3-soil
t = Time, days
DE = 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 uniformly to an infinite depth. The boundary
conditions for which Equation 3 is solved are:
c = C0 at t > 0, x = oo
c = Oatt>0, x=0
The average flux over time (J!!V8)is calculated as:
C8 = C0(4DE/;rt)1/2 (4)
2.1 FINITE SOURCE MODEL DERIVATION
The Jury Reduced Solution finite source model is derived from the methods presented by
Mayer et al. (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, DE, is assumed to be constant, the general diffusion equation is:
0 - if t - ° (5)
dx DE (A
where c = Soil concentration, g/cm3 - total volume
x = Distance measured normal to soil surface, cm
DE = Effective diffusion coefficient in soil, cm2/d
t = Time, days.
C-4
-------
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 Carslaw and Jaeger (1959, page 62, Equation 14) to solve the diffusion
equation given these initial and boundary conditions as:
C = C0/2){2erf[x/2(DEt)1/2]- erf [(x - L)/2(DEt)1/2 ] - erf[(x + L)/2(DEt)1/2]} (6)
The flux is obtained by differentiating Equation 6 with respect to x, determining dc/ ck at
x = O. and multiplying by DE. The result is:
Js = DE [*/&]_ = [DE C0/ (;rDEt)1/2] [1-exp (-L2/4 DEt)] (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'^).
Jury et al. (1983 and 1990) expanded upon the work of Carslaw and Jaeger (1959) and
Mayer et al. (1974) by developing an analytical solution for Equation 5 which 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:
Js = - DE ( L
c = O at t>0, x = 0
Js = - hCG at t>0, x = 0
where h = Transport coefficient across the soil-air boundary layer of
thickness d (h = Dga/d)
CG = Vapor-phase concentration (CG = KH C:),
C-5
-------
The Jury et al. (1983) analytical solution for the volatilization flux is:
Js(t,L)= + -C0VE
erfc
vEt
2(DEtr ,
1 crfcL + VEt
- LI 1C ,„
J ^ 2(DEt)1/2
PE + vE)0
(9)
exp
(2HE + VE)t
2(DEt)1/2 ,
2(DEt)1
where HE Is the transport coefficient across the boundary layer divided by the gasphase partition
coefficient, HE =h/(pb foc Koc/KH + 0/KH + 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.,asHE —» °°). Therefore, for the case where
HE —» 0° and in the absence of water flux (VE = 0) Equation 9 is reduced to Equation 1 where the
approximation
erfc [x] =
1
(10)
is used to expand the error function for large 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).
2.2 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 (19SS, page 97, Equation 8) to solve the diffusion equation given the boundary conditions:
c = C0 at t = 0, 0 < x < L
c = 0 at t >- 0, x = 0
del dx = Oatx = L
The Mayer et al. (1974) solution for the volatilization flux is:
= DE C0/(7TDEt)1/2l + 2
2[l
exp(
-n2L2/DEt)J
(11)
Therefore, Equation 11 is the analytical solution for a finite emission source, but accounts only for
diffusion across the upper boundary.
C-6
-------
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:
Js = DEC0/(7TDEt)1/2 (12)
Use of Equation 12 will result in less than 1 percent error if t < L2/18.4 DE (Mayer et al., 1974) .
Jury et al. (1984 and 1990) gave the solution for the semi-infinite case in Equation 3 where
C = C0 at t > 0, x = oo as:
1/2
Js = C0(DE/7T t)"
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 DE. 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 DE, Equation 1
should be used to calculate the volatilization factor. As an alternative, an estimate of the average
emission flux over the exposure interval, , can be obtained from a simple mass balance:
= C0L/t (13)
where C0 = Initial soil concentration (total volume), |lg/cm3-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 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.
C-7
-------
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 in 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.
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-8
-------
SECTION 3
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 Cliath, 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-hexachlorocyclohexane, gamma isomer) and dieldrin
(1,2,3,4,10,10-hexachloro-6,7-epoxy-l,4,4a,5,6,7,8,8a-octahydro-l,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. EPA
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 biogenic emissions of inorganic gases and their use dates
back at least two decades (Flill 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, 199la).
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 dieldrin with that of lindane.
Volatilization of dieldrin and lindane was measured in a closed airflow system by collecting the
volatilized insecticides in ethylene glycol traps. Ten grams of soil were treated with either 5 or 10
|ig/g of C-14 tagged insecticide in hexane. The hexane was evaporated by placing the soils in a
C-9
-------
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 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 sulfate 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):
Csat = — (focKocpb +0 + KHa) (14)
Pb
where S is the pure component solubility in water. Csat 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 DE. 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 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 lognormal distribution for the emission flux and analyzing the logtransformed data
for differences between modeled and measured values.
C-10
-------
TABLE 1.
VOLATILIZATION MODEL INPUT VALUES FOR LINDANE AND DIELDRIN
Variable
Initial soil
concentration
Soil depth
Soil dry bulk
density
Soil particle
density
Gravimetric soil
moisture content
Water-filled soil
porosoty
Total soil porosity
Air-filled soil
porosity
Soil organic carbon
Organic carbon
partition coefficient
Diffusivity in air
(Lindane)
Diffusivity in air
(Dieldin)
Diffusivity in water
(Lindane)
Diffusivity in water
(Dieldrin)
Henry's law
constant (Lindane)
Henry's law
constant (Dieldrin)
Degradation rate
constant (Lindane
and Dieldrin)
Symbol
C0
L
Pb
Ps
w
0
a
fnn
Koc
D;
D;
Df
Df
KH
KH
!-*
Units
mg/kg
cm
g/cm3
g/cm3
percent
cm3/cm3
cm3/cm3
cm3/cm3
fraction
cm3/g
cm2/d
cm2/d
cm2/d
cm2/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
2.75 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)
wps
i-(A,/P.)
(f) - 0
Farmer et al. (1972)
U.S. EPA (1994)
U.S. EPA (1994)
U.S. EPA (1994)
U.S. EPA(1994a)
U.S. EPA(1994a)
U.S. EPA (1994)
U.S. EPA (1994)
Default to eliminate
effects of degradation
C-ll
-------
0.25
0.2
.o
oi
X
O
55
(0
5
UJ
0.15
0.05
DIELDRIN
(Initial Soil Cone. = 5 mg/kg)
m Measured flux
o Infinite source model
predicted flux
0 50 100 150 200 250 300
TIME FROM SAMPLING (t), hrs
Figure 1
Predicted And Measured Emission Flux Of Dieldrin Versus Time (C0 = 5 ppmw)
C-12
-------
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 calculated to
be 0.994 with a mean ratio of modeled-to-measured values of 0.42. The actual significance
(p-value) 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.
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 DE at 24 hours). Therefore, the Mayer et al. (1974) finite source
model was used to derive a percent error at each succeeding timestep. 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.
Figures 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 DE
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 timestep.
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-13
-------
— 1 -
— 9 -
-3-
-4-
n i i i I i i i
t i i i i i i i i i i i i i i i i i
Ln Time
000 Jury Model Predicted Flux
Measured Flux
Figure 2
Predicted And Measured Emission Flux Of Dieldrin Versus Time (C0 = 10 ppmw)
C-14
-------
0.45
0.4
0.35
J? 0.3
,0
^i
^0.25
X
u! 0.2
z
g
V)
J2 0.15
UJ
0.1
0.05
xa.
50
DIELDRIN
(Initial Soil Cone. = 10 mg/kg)
o Infinite source model
predicted flux
• Measured flux
100 150 200
TIME FROM SAMPLING (t), hrs
250
300
Figure 3
Predicted And Measured Emission Flux Of Dieldrin Versus Time (C0 = 10 ppmw)
C-15
-------
o -
x
c
-3 -
-4 -
n i i i i i i i r
i i i i i i
i T i i i i i
Ln Time
000 Jury Model Predicted Flux EHEHD Measured Flux
Figure 4
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Dieldrin (C0 = 10 ppmw)
C-16
-------
0.6
LINDANE
(Initial Soil Cone. = 5 mg/kg)
0.5
0.4
111
• Measured flux
o Infinite source model
predicted flux
A Finite source model
predicted flux
20
40
60 80 100 120
TIME FROM SAMPLING (t), hrs
140
160
180
Figure 5
Predicted And Measured Emission Flux Of Lindane Versus Time (Co = 5 ppmw)
C-17
-------
1.2
eg
D)
dL
X
0.6
V)
go.
Ul
0.2
LINDANE
(Initial Cone. = 10 ppmw)
o Infinite source model
predicted flux
A Finite source model
predicted flux
20
40
60 80 100 120
TIME FROM SAMPLING (t), hrs
140
160
180
Figure 6
Predicted And Measured Emission Flux Of Lindane Versus Time (C0 = 10 ppmw)
C-18
-------
o -
-1 -
x
-2 -
-3 -
-4 -
i i i i i i i i i i i i i i i i i i i I i ii | i ii | 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 TTJTTT
3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5.0 5.1 5.2
Ln Time
000 Jury Model Predicted Flux
Measured Flux
Figure 7
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Lindane (C0 = 5 ppmw)
C-19
-------
1 -
o-
-H
-2 -
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 [ II I | I I I | I I I | I II | I II | I I I | I I I j I II |
3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 5.0 5.1 5.2
Ln Time
000 Jury Model Predicted Flux
Measured Flux
Figure 8
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Lindane (Cc
= 10 ppmw)
C-20
-------
TABLE 2.
SUMMARY OF THE BENCH-SCALE VALIDATION OF
THE JURY INFINITE SOURCE MODEL
Chemical
Lindane (5 ppmw)
Lindane (10 ppmw)
Dieldrin (5 ppmw)
Dieldrin (10 ppmw)
N
4
4
7
7
Correlation
coefficient
0.997
0.998
0.994
0.974
Mean ratio:
Modeled-to-
measured
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:
• A control pile that was not moved or treated
• An "aerated" or "mechanically mixed" pile
• A soil pile simulating soil venting or vacuum extraction
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 concentrations 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 loss
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.
C-21
-------
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.
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 in 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, 880 ppm, and 310 ppm, respectively.
It should be noted that the fraction of soil organic carbon (foc) was not available from
Radian (1989). For this reason, the default value for foc of 0.006 from U.S. EPA (1994) was used
for all 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 were below the detection
limits after the fifth observation; measured values for toluene were below the detection limit after
the seventh observation.
C-22
-------
TABLE 3.
VOLATILIZATION MODEL INPUT VARIABLES FOR BENZENE,
TOLUENE, AND ETHYLBENZENE
Variable
Initial soil concentration
- benzene
- toluene
- ethylbenzene
Soil Depth
Soil dry bulk density
Soil particle density
Gravimetric soil moisture
content
Water-filled soil porosoty
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
Henry's law constant
- benzene
- toluene
- ethylbenzene
Degradation rate constant
Symbol
C0
L
ph
ps
w
0
a
foe
Koc
°:
D™
KH
!-*
Units
mg/kg
cm
g/cm3
g/cm3
percent
cm3/cm3
cm3/cm3
cm3/cm3
Fraction
cm3/g
cm2/s
cm2/s
Unitless
1/day
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
0.22
0.26
0.32
0
Reference/Equation
Radian (1989)
Radian (1989)
Radian (1989)
U.S. EPA (1988)
Radian (1989)
wph
l-(Pb/P.)
0 - 0
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(1994a)
U.S. EPA (1994)
U.S. EPA (1994)
U.S. EPA (1994)
Default to eliminate
effects of degradation
C-23
-------
1400
BENZENE
(Initial Soil Cone. = 110 mg/kg)
o Infinite source model
predicted flux
Finite source model
predicted flux
100
200 300 400
TIME FROM SAMPLING (t), hrs
500
600
700
Figure 9
Predicted And Measured Emission Flux Of Benzene (C0 = 110 ppmw)
C-24
-------
TOLUENE
(Initial Soil Cone. = 880 mg/kg)
ra
J
_o
O)
8000
7000
6000
5000
*- 4000
D
u.
Z
2 3000
(/)
co
LU
2000 •
1000
o Infinite source model
predicted flux
A Finite source model
predicted flux
200
400 600 800
TIME FROM SAMPLING (t), hrs
Figure 10
Predicted And Measured Emission Flux Of Toluene (Cc
1000
= 880 ppmw)
1200
C-25
-------
2500
ETHYLBENZENE
(Initial Soil Cone. = 310 ppmw)
2000
n
1500
1000
CO
CO
5
til
• Measured flux
o Infinite source model
predicted flux
A Finite source model
predicted flux
100
200 300 400
TIME FROM SAMPLING (t), hrs
500
600
700
Figure 11
Predicted And Measured Emission Flux Of Ethylbenzene (C0 = 310 ppmw)
C-26
-------
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)
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 timestep
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 (i.e., the highest
modeled-to-omeasured 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:
1. Although the initial soil concentrations of the three compounds were below or
within a 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 concentrations 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.
C-27
-------
X
\L
c
7 -
6 -
5 -
4 -
3 -
0
Ln Time
0—0—B Jury Model Predicted Flux O~~B~D Measured Flux
Figure 12
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Benzene (C0 = 110 ppmw)
C-28
-------
9 -
8 -
7 -
"• 6
5
4 -
\ r
Ln Time
000 Jury Model Predicted Rux B—B—B Measured Flux
Figure 13
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Toluene (C0 = 880 ppmw)
C-29
-------
7 -
x
2
\L
3 -
n—i—i—i—i—i—i—r
Ln Time
O-~G—9 Jury Model Predicted Flux B—B~O Measured Flux
Figure 14
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Ethylbenzene (C0 = 310 ppmw)
C-30
-------
5. Variability in the relative humidity of the air above the test bed may have induced
surface water evaporation in 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.
3.2 VALIDATION 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) diisopropyithiocarbamate] incorporated
in San Joaquin sandy loam. This study replicated the model boundary conditions in that a clean
layer of soil underlayed the contaminated soil allowing diffusion across the lower boundary as well
as the upper boundary.
Volatilization of triallate was measured in a closed 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 autoclaved 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 (Csat) 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 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.
C-31
-------
TABLE 5.
VOLATILIZATION MODEL INPUT VALUES FOR TRIALLATE
Variable
Initial soil concentration
Soil depth
Soil dry bulk density
Soil particle density
Gravimetric soil moisture content
Water-filled soil porosoty
Total soil porosity
Air-filled soil porosity
Soil organic carbon
Organic carbon partition
coefficient
Diffusivity in air
Diffusivity in water
Henry's law constant
Degradation rate constant
Symbol
cn
L
ph
ps
w
0
0
a
foe
KOC
°:
Df
KH
!-*
Units
mg/kg
cm
g/cm3
g/cm3
percent
cm3/cm3
cm3/cm3
cm3/cm3
fraction
cm3/g
cm2/d
cm2/d
unitless
1/day
Value
10
10
1.34
2.65
21
0.279
0.494
0.215
0.0072
3600
3888
0.432
1.04 E-03
0
Reference/Equation
Jury et at. (1980)
Jury et at. (1980)
Jury et at. (1980)
U.S. EPA (1988)
Calculated from Jury et al.
(1980)
Jury et at. (1980)
Jury et at. (1980)
Jury et at. (1980)
Calculated from Jury et al.
(1980)
Jury et at. (1980)
Jury et at. (1980)
Jury et at. (1980)
Jury et at. (1980)
Default to eliminate
effects of degradation
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 Koc, 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-32
-------
0.01
TRIALLATE
(Initial Cone. = 10ppmw)
o Jury model predicted flux
100
200 300 400
TIME FROM SAMPLING (t), hrs
500
600
700
Figure 15
Predicted And Measured Emission Flux Of Triallate Versus Time
C-33
-------
1 -
0-
-1
-2 -
-3 -
I i i—i—i i i i—rr~|—r~rn i i i i i i |
1 2 3
i i i i i i—i i i i i—r~r~i—r~i—r
Ln Time
I I I I I I I I ! I I I—PT
000 Jury Model Predicted Flux
Measured Flux
Figure 16
Comparison Of Log-Transformed Modeled And Measured Emission Flux Of Triallate
C-34
-------
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.
Soil Moisture Content
Farmer, et al. (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
C-35
-------
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.
Soil Temperature
The effect of soil temperature on the volatilization flux is multifunctional. The diffusion in
air, D*, is theoretically related to temperature, T, and the collision integral, Q, in the following
manner (Lyman, et al., 1990):
rpO.5
D" (proportional to) (15)
8 Q(T)
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:
D2 = Dj (Tj/Tj)05 (16)
where D2 = Diffusion coefficient at T2
Dj = Diffusion coefficient at T:
T = Absolute temperature.
A temperature increase will effect the vapor pressure function of the Henry's Law constant,
which causes an 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.
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 (KH), the soil/water partition coefficient, (KD) 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 C0 of 100
percent causes an increase in the emission rate of 100 percent. Probably the greatest degree of
uncertainty in the value of C0 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).
C-36
-------
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 KD/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 KH and inversely proportional to KD. Because the
Jury volatilization models do not account for a soil-air boundary layer, the effects of KH and KD are
exponential for all three categories of chemicals.
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 will 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 DEt)] 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 DE.
C-37
-------
SECTION 5
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-measured 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.
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 DE t)1/2. 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
C-38
-------
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 estimates of loss through volatilization at the soil surface given the boundary conditions
of each model.
C-39
-------
REFERENCES
Balfour, W. D., B. M. Eklund, and S. J. Williamson. 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, zd Edition Oxford
University Press, Oxford.
Environmental Quality Management, Inc. 1994. A Comparison of Soil Volatilization Models in
Support of Superfund Soil Screening Level Development. U.S. Environmental Protection Agency,
Contract No. 68-D30035, 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-660/2-74/054.
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.
Howard, P. H., R. S. Boethling, W. F. Jarvis, W. M. Meylan, and E. O. Michaelenko. 1991.
Handbook of Environmental Degradation Rates. Lewis Publishers, Chelsea, Michigan.
Jury, W. A., R. Grover, 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. Qual, 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. Qual., 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.
C-40
-------
Kienbusch, M. and D. Ranum. 1986. Validation of Flux Chamber Emission Measurements on
Soil Surface - Draft Report to EPA-EMSL, Las Vegas, Nevada.
Kienbusch, M., W. D. Balfour, 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.
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. Soil Sci. Soc. Amer.
Proc. 33:509-511.
Spencer, W. F., and M. Cliath. 1970. Vapor Density and Apparent Vapor Pressure of Lindane
(y-BHC). J. Agr. Food Chem. 75:529-530.
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/l-89-002a.
U.S. Environmental Protection Agency. 1991. Risk Assessment Guidance for Superfund, Volume
1, Human Health Evaluation Manual (Part B). Office of Emergency and Remedial Response.
Publication No. 9285.7-01B.
U.S. Environmental Protection Agency. 199 la. 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.
C-41
-------
U.S. Environmental Protection Agency. 1994a. CHEMDAT8 DataBase 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-42
-------
APPENDIX A
VALIDATION DATA FOR THE JURY INFINITE SOURCE MODEL
C-43
-------
DIELDRIN 5 PPM
(1 of 2)
Chemical
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Sample
Point
1
2
3
4
5
6
7
Initial soil
cone.,
C0
(mg/kg)
5
5
5
5
5
5
5
Initial soil
cone.,
C0
(g/g)
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
Emitting
area
(cm2)
27.55
27.55
27.55
27.55
27.55
27.55
27.55
Soil
Depth
(L)
(cm)
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Soil Type
GilaSlitLoam
Gila Slit Loam
Gila Slit Loam
Gila Slit Loam
Gila Slit Loam
Gila Slit Loam
Gila Slit Loam
Soil bulk
density,
Pb
Kg/L
0.75
0.75
0.75
0.75
0.75
0.75
0.75
Soil
particle
density,
Ps
Kg/L
2.65
2.65
2.65
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
0
(unitless)
0.0750
0.0750
0.0750
0.0750
0.0750
0.0750
0.0750
Solubility
S
(mg/L)
0.1870
0.1870
0.1870
0.1870
0.1870
0.1870
0.1870
Soil
organic
carbon,
foe
(fraction)
0.0058
0.0058
0.0058
0.0058
0.0058
0.0058
0.0058
Saturation
cone.,
csal
(mg/kg)
12
12
12
12
12
12
12
c0>csal
(Yes/No)
No
No
No
No
No
No
No
Measured
emission
flux
(|ig/m2-
min)
200
115
75
65
60
55
40
Organic
carbon part.
coeff.,
KOC
(cm3/g)
10900
10900
10900
10900
10900
10900
10900
C-44
-------
DIELDRIN 5 PPM
(2 of 2)
Chemical
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Soil/water part.
coeff.,
Kn
(cmj/g)
63.22
63.22
63.22
63.22
63.22
63.22
63.22
Diffusivity in
air,
D/
(crrr/s)
0.0125
0.0125
0.0125
0.0125
0.0125
0.0125
0.0125
Diffusivity in
water,
Dw
(cm'/s)
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
Effective
diffusion
coefficient,
DF
(cm'/s)
1.32E-08
1.32E-08
1.32E-08
1.32E-08
1.32E-08
1.32E-08
1.32E-08
Henry's law
constant,
KH
(unitless)
0.00011
0.00011
0.00011
0.00011
0.00011
0.00011
0.00011
Total soil
porosity,
0
(unitless)
0.7170
0.7170
0.7170
0.7170
0.7170
0.7170
0.7170
Air-filled soil
porosity,
a
(unitless)
0.6420
0.6420
0.6420
0.6420
0.6420
0.6420
0.6420
Measured emission
flux emission flux
(|ig/cm'-day)
0.2000
0.1150
0.0750
0.0650
0.0600
0.0550
0.0400
Time, t
Cumulative
(hours)
24
72
120
144
168
216
288
t>L2/14.4 DF
(Yes/No)
No
No
No
No
No
No
No
Infinite source
model emission
flux
(|ig/cm'-day)
0.0714
0.0412
0.0319
0.0292
0.0270
0.0238
0.0206
C-45
-------
DIELDRIN 10 PPM
(1 of 2)
Chemical
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Sample
Point
1
2
3
4
5
6
7
Initial soil
cone.,
Cn
(mg/kg)
10
10
10
10
10
10
10
Initial soil
cone.,
C0
(g/g)
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
Emitting
area
(cm')
27.55
27.55
27.55
27.55
27.55
27.55
27.55
Soil
Depth
(L)
(cm)
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Soil Type
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Soil bulk
density,
Pb
(g/cmj)
0.75
0.75
0.75
0.75
0.75
0.75
0.75
Soil
particle
density,
P,
(g/cmj)
2.65
2.65
2.65
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
e
(unitless)
0.0750
0.0750
0.0750
0.0750
0.0750
0.0750
0.0750
Solubility,
S
(mg/L)
01870
0.1870
0.1870
0.1870
0.1870
01870
0.1870
Soil
organic
carbon,
for
(fraction)
0.0058
0.0058
0.0058
0.0058
0.0058
0.0058
0.0058
Saturation
cone.,
c ,
(mg/kg)
12
12
12
12
12
12
12
cn>c,al
(Yes/No)
No
No
No
No
No
No
No
Measured
emission
flux
(ng/cm'
-day)
400
260
140
110
105
90
85
Organic
carbon
part.
coeff.,
Km
(cmj/g)
10900
10900
10900
10900
10900
10900
10900
C-46
-------
DIELDRIN 10
(2 of 2)
PPM
Chemical
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Dieldrin
Soil/ water
part, coeff.,
Kn
(cmj/g)
63.22
63.22
63.22
63.22
63.22
63.22
63.22
Diffusivity in
air,
Daa
(crrWs)
0.0125
0.0125
0.0125
0.0125
0.0125
0.0125
0.0125
Diffusivity in
water,
D;W
(cnr7s)
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
4.74E-06
Effective
diffusion
coefficient,
DF
(crrWs)
1 .32E-08
1 .32E-08
1 .32E-08
1 .32E-08
1 .32E-08
1 .32E-08
1 .32E-08
Henry's law
constant,
KH
(unitless)
0.0001 1
0.0001 1
0.0001 1
0.0001 1
0.0001 1
0.0001 1
0.0001 1
Total soil
porosity,
*
(unitless)
0.7170
0.7170
0.7170
0.7170
0.7170
0.7170
0.7170
Air-filled soil
porosity,
a
(unitless)
0.6420
0.6420
0.6420
0.6420
0.6420
0.6420
0.6420
Measured
emission flux
(ng/cnr'-day)
0.4000
0.2600
0.1400
0.1100
0.1050
0.0900
0.0850
Time,
t
Cumulative
(hrs)
24
72
120
144
168
216
288
t>L2/14.4DF
(Yes/No)
No
No
No
No
No
No
No
Infinite source
model emission
flux
(ng/cnr'-day)
0.1428
0.0825
0.0639
0.0583
0.0540
0.0476
0.0412
C-47
-------
LINDANE 5 PPM
(1 of 2)
Chemical
Lindane
Lindane
Lindane
Lindane
Sample
Point
1
2
3
4
Initial soil
cone
Cn
(mg/kg)
5
5
5
5
Initial soil
cone.
C0
(g/g)
5.00E-06
5.00E-06
5.00E-06
5.00E-06
Emitting
area
(cm')
27.55
27.55
27.55
27.55
Soil
Depth
(L)
(cm)
0.5
0.5
0.5
0.5
Soil Type
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Soil bulk
density,
Pb
(g/cmj)
0.75
0.75
0.75
0.75
Soil
particle
density,
P,
(g/cmj)
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
0
(unitless)
0.0750
0.0750
0.0750
0.0750
Solubility,
S
(mg/L)
4.2000
4.2000
4.2000
4.2000
Soil organic
carbon,
for
(fraction)
0.0058
0.0058
0.0058
0.0058
Saturation
cone.,
c ,
(mg/kg)
34
34
34
34
cn>c,al
(Yes/No)
No
No
No
No
Measured
emission flux
(ng/cm'-day)
500
160
60
40
Organic
carbon
part.
coeff.,
K«.
(cmj/g)
1380
1380
1380
1380
C-48
-------
LINDANE 5 PPM
(2 of 2)
Chemical
Linda ne
Linda ne
Linda ne
Linda ne
Soil/water
part, coeff.,
Kn
(cmj/g)
8.00
8.00
8.00
8.00
Diffusivity in
air,
D0a
(crrr/s)
0.0176
0.0176
0.0176
0.0176
Diffusivity in
water,
pw
(cm'/s)
5.57E-06
5.57E-06
5.57E-06
5.57E-06
Effective
diffusion
coefficient,
DF
(cm'/s)
1.80E-07
1.80E-07
1.80E-07
1.80E-07
Henry's law
constant,
KH
(unitless)
0.00014
0.00014
0.00014
0.00014
Total soil
porosity,
*
(unitless)
0.7170
0.7170
0.7170
0.7170
Air-filled soil
porosity,
a
(unitless)
0.6420
0.6420
0.6420
0.6420
Measured
emission flux
(u,g/cm' day)
0.5000
0.1600
0.0600
0.0400
Time,
t
Cumulative
(hours)
24
72
120
168
t>L2/14.4 DF
(Yes/No)
No
Yes
Yes
Yes
Infinite source
model emission
flux
(u,g/cm' day)
0.2641
0.1525
0.1181
0.0998
Finite source
model
emmision flux
(u,g/cm' day)
0.2641
0.1510
0.1086
0.0797
Infinite source
model error
(percent)
0.0000
0.9604
8.7891
25.2965
C-49
-------
LINDANE 10 PPM
(1 OF 2)
Chemical
Linda ne
Linda ne
Linda ne
Linda ne
Sample
Point
1
2
3
4
Initial soil
cone.,
cr
(mg/kg)
10
10
10
10
Initial soil
cone.,
cn
(g/g)
1.00E-05
1.00E-05
1.00E-05
1.00E-05
Emitting
area
(cm')
27.55
27.55
27.55
27.55
Soil Depth
(L)
(cm)
0.5
0.5
0.5
0.5
Soil Type
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Gila Silt Loam
Soil bulk
density,
Ph
(g/cmj)
0.75
0.75
0.75
0.75
Soil
particle
density,
P,
(g/cmj)
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
e
(unitless)
0.0750
0.0750
0.0750
0.0750
Solubility,
S
(mg/L)
4.2000
4.2000
4.2000
4.2000
Soil
organic
carbon,
L
(fraction)
0.0058
0.0058
0.0058
0.0058
Saturation
cone.,
c ,
(mg/kg)
34
34
34
34
cn>c,al
(Yes/No)
No
No
No
No
Measured
emission
flux
(ng/cm'
-day)
1160
320
140
90
Organic
carbon
part.
coeff.,
K
(crrr/g)
1380
1380
1380
1380
C-50
-------
LINDANE 10
(2 of 2)
PPM
Chemical
Linda ne
Linda ne
Linda ne
Linda ne
Soil/water
part, coeff.,
Kn
(cmj/g)
8.00
8.00
8.00
8.00
Diffusivity in
air,
Dra
(crrr/s)
0.0176
0.0176
0.0176
0.0176
Diffusivity in
water,
Dw
(cm'/s)
5.57E-.06
5.57E-.06
5.57E-.06
5.57E-.06
Effective
diffusion
coefficient,
DF
(cm'/s)
1.80E-07
1.80E-07
1.80E-07
1.80E-07
Henry's law
constant,
KH
(unitless)
0.00014
0.00014
0.00014
0.00014
Total soil
porosity,
0
(unitless)
0.7170
0.7170
0.7170
0.7170
Air-filled soil
porosity,
a
(unitless)
0.6420
0.6420
0.6420
0.6420
Measured
emission flux
(u,q/cm'
-day)
1.1600
0.3200
0.1400
0.0900
Time, t
Cumulative
(hours)
24
72
120
168
t>L2/14.4 DF
(Yes/No)
No
Yes
Yes
Yes
Infinite
source model
emission flux
(u,q/cm'
-day)
0.5282
0.3049
0.2362
0.1996
Finite source
model
emmision
(u.q/cm'
-day)
0.5282
0.3020
0.2171
0.1593
Infinite
source model
error
(percent)
0.0000
0.9604
8.7891
25.296
C-51
-------
BENZENE 110 PPMW
(1 of 2)
Chemical
Benzene
Benzene
Benzene
Benzene
Benzene
Benzene
Sample
Point
3
4
5
6
7
8
Initial soil
cone.,
C-
(mg/kg)
110
110
110
110
110
110
Initial soil
cone.,
Cn
(g/g)
1.10E-04
1.10E-04
1.10E-04
1.10E-04
1.10E-04
1.10E-04
Flux
chamber
surface
area
(cm')
1300
1300
1300
1300
1300
1300
Soil Depth
(L)
(cm)
91
91
91
91
91
91
Soil Type
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Soil bulk
density,
P,
(Kg/L)
1.5
1.5
1.5
1.5
1.5
1.5
Soil
particle
density,
P,
(Kg/L)
2.65
2.65
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
e
(unitless)
0.1500
0.1500
0.1500
0.1500
0.1500
0.1500
Solubility,
S
(mg/L)
1780
1780
1780
1780
1780
1780
Soil organic
carbon,
L
(fraction)
0.006
0.006
0.006
0.006
0.006
0.006
Saturation
cone.,
CM
(mg/kg)
862
862
862
862
862
862
cn>c,al
(Yes/No)
No
No
No
No
No
No
Measured
emission
flux
(u,g/cm'
-day)
2760
9000
910
400
290
0
Organic
carbon
part.
coeff.,
K™.
(cmj/g)
57
57
57
57
57
57
C-52
-------
BENZENE 110 PPMW
(2 of 2)
Chemical
Benzene
Benzene
Benzene
Benzene
Benzene
Benzene
Soil/water
part.
coeff.,
Kn
(cmj/g)
0.34
0.34
0.34
0.34
0.34
0.34
Diffusivity
in air,
Dra
(crrr/s)
0.0870
0.0870
0.0870
0.0870
0.0870
0.0870
Diffusivity
in water,
nw
(crrr/s)
9.80E-06
9.80E-06
9.80E-06
9.80E-06
9.80E-06
9.80E-06
Effective
diffusion
coefficient,
DF
(cm'/s)
2.14E-03
2.14E-03
2.14E-03
2.14E-03
2.14E-03
2.14E-03
H
(atm-mj/mol)
0.00543
0.00543
0.00543
0.00543
0.00543
0.00543
Henry's law
constant,
KH
(unitless)
0.22263
0.22263
0.22263
0.22263
0.22263
0.22263
Total soil
porosity,
0
(unitless)
0.4340
0.4340
0.4340
0.4340
0.4340
0.4340
Air-filled soil
porosity,
a
(unitless)
0.2840
0.2840
0.2840
0.2840
0.2840
0.2840
Measured
emission flux
(u,q/cm'
-Jay)
397
1296
131
58
42
0
Time,
t
Cumulative
(hours)
26.40
76.25
119.73
506.83
698.55
863.17
t>L2/14.4 DF
(Yes/No)
No
Yes
Yes
Yes
Yes
Yes
Infinite
source model
emission flux
(u,q/cm'
-day)
1207
710
567
275
235
211
Finite source
model
emmision
flux
(u,q/cm'
-clay)
1207
710
567
209
135
92
Infinite
source model
error
(percent)
0.0000
0.0002
0.0253
31.5053
73.9743
128.3941
C-53
-------
TOLUENE 880 PPMW
(1 OF 2)
Chemical
Toluene
Toluene
Toluene
Toluene
Toluene
Toluene
Toluene
Sample
Point
3
4
5
6
7
8
9
Initial soil
cone.
Cn
(mg/kg)
880
880
880
880
880
880
880
Initial soil
cone.
Cn
(g/g)
8.80E-04
8.80E-04
8.80E-04
8.80E-04
8.80E-04
8.80E-04
8.80E-04
Flux
chamber
surface
area
(cm2)
1300
1300
1300
1300
1300
1300
1300
Soil
Depth
(L)
(cm)
91
91
91
91
91
91
91
Soil Type
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Soil bulk
density,
Ph
(kg/L)
1.5
1.5
1.5
1.5
1.5
1.5
1.5
Soil
particle
density,
P,
(kg/L)
2.65
2.65
2.65
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
e
(unitless)
0.1500
0.1500
0.1500
0.1500
0.1500
0.1500
0.1500
Solubility
S
(mg/L)
558
558
558
558
558
558
558
Soil
organic
carbon,
fnr
(fraction)
0.006
0.006
0.006
0.006
0.006
0.006
0.006
Saturation
cone.,
C,al
(mg/kg)
522
522
522
522
522
522
522
Comsat
(Yes/No)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Measured
emission
flux
(|ig/m'
-min)
14800
17300
4910
1340
830
340
260
Organic
carbon part.
coeff.,
K«-
(cm3/g)
131
131
131
131
131
131
131
C-54
-------
TOLUENE 880 PPMW
(2 OF 2)
Chemical
Toluene
Toluene
Toluene
Toluene
Toluene
Toluene
Toluene
Soil/water
part, coeff.,
Kn
(cmj/g)
0.79
0.79
0.79
0.79
0.79
0.79
0.79
Diffusivity
in air,
D/
(crm/s)
0.0870
0.0870
0.0870
0.0870
0.0870
0.0870
0.0870
Diffusivity
in water,
nw
(crm/s)
8.60E-06
8.60E-06
8.60E-06
8.60E-06
8.60E-06
8.60E-06
8.60E-06
Effective
diffusion
coefficient,
DF
(cm'/s)
1.30E-03
1.30E-03
1.30E-03
1.30E-03
1.30E-03
1.30E-03
1.30E-03
H
(atm-
mj/mol)
0.00637
0.00637
0.00637
0 00637
0.00637
0.00637
0.00637
Henry's
law
constant,
KH
(unitless)
0.26117
0.26117
0.26117
0.26117
0.26117
0.26117
0.26117
Total soil
porosity,
0
(unitless)
0.4340
0.4340
0.4340
0.4340
0.4340
0.4340
0.4340
Air-filled
soil
porosity,
a
(unitless)
0.2840
0.2840
0.2840
0.2840
0.2840
0.2840
0.2840
Measured
emission
flux
(u,q/cm'
clay)
2131
2491
707
193
120
49
37
Time, t
Cumulative
(hours)
26.40
76.25
119.73
506.83
698.55
863. 17
1007.17
t>L2/14.4 DF
(Yes/No)
No
No
No
Yes
Yes
Yes
Yes
Infinite source
model emission
flux
(\iglcmf day)
7524
4427
3533
1717
1463
1316
1218
Finite source
model
emmision flux
(\iglcmf day)
7524
4427
3533
1613
1231
978
800
Infinite source
model error
(percent)
0.0000
0.0000
0.0001
6.4806
18.8541
34.5481
52.2568
C-55
-------
ETHYLBENZENE 310 PPMW
(1 of 2)
Chemical
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethyibenzene
Sample
Point
3
4
5
6
7
8
Initial soil
cone
Cn
(mg/kg)
310
310
310
310
310
310
Initial soil
cone.
C0
(g/g)
3.1 OE-04
3.1 OE-04
3.1 OE-04
3.1 OE-04
3.1 OE-04
3.1 OE-04
Flux
chamber
surface
area
(cm')
1300
1300
1300
1300
1300
1300
Soil Depth
(L)
(cm)
91
91
91
91
91
91
Soil Type
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Loamy Sand
Soil bulk
density,
Pb
(g/cmj)
1.5
1.5
1.5
1.5
1.5
1.5
Soil
particle
density,
Ps
(g/cmj)
2.65
2.65
2.65
2.65
2.65
2.65
Gravimetric
soil moisture,
w
(wt. fraction)
0.10
0.10
0.10
0.10
0.10
0.10
Water-
filled soil
porosity,
e
(unitless)
0.1500
0.1500
0.1500
0.1500
0.1500
0.1500
Solubility
S
(mg/L)
173
173
173
173
173
173
Soil
organic
carbon,
for
(fraction)
0.006
0.006
0.006
0.006
0.006
0.006
Satura-
tion cone.,
c,.,
(mg/kg)
257
257
257
257
257
257
cn>c,al
(Yes/No)
Yes
Yes
Yes
Yes
Yes
Yes
Measured
emission
flux
(|ig/cm'
-min)
2640
1700
1080
250
180
0
Organic
carbon
part.
coeff.,
Km
(cmj/g)
221
221
221
221
221
221
C-56
-------
ETHYLBENZENE 310
(2 of 2)
PPMW
Chemical
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethylbenzene
Ethyibenzene
Soil/ water
part, coeff.,
Kn
(cmj/g)
1.33
1.33
1.33
1.33
1.33
1.33
Diffusivity
in air,
Dna
(cm'/s)
0.0750
0.0750
0.0750
0.0750
0.0750
0.0750
Diffusivity
in water,
pw
(cm'/s)
7.80E-06
7.80E-06
7.80E-06
7.80E-06
7.80E-06
7.80E-06
Effective
diffusion
coefficient,
DF
(cm'/s)
8.64E-04
8.64E-04
8.64E-04
8.64E-04
8.64E-04
8.64E-04
Henry's
law
constant,
KH
(unitless)
0.32021
0.32021
0.32021
0.32021
0.32021
0.32021
Total soil
porosity,
*
(unitless)
0.4340
0.4340
0.4340
0.4340
0.4340
0.4340
Air-filled
soil
porosity,
a
(unitless)
0.2840
0.2840
0.2840
0.2840
0.2840
0.2840
Measured
emission flux
(|iq/cm'
day)
380
245
156
36
26
0
Time,
t
Cumulative
(hours)
26.40
76.25
119.73
506.83
698.55
863.17
t>L2/14.4 DF
(Yes/No)
No
No
No
Yes
Yes
Yes
Infinite source
model emission
flux
(|iq/cm'
day)
2162
1272
1015
493
420
373
Finite source
model
emmision flux
(|iq/cm'
day)
2162
1272
1015
488
402
343
Infinite source
model error
(percent)
0.0000
0.0000
0.0000
1.0596
4.6357
10.0850
C-57
-------
APPENDIX B
VALIDATION DATA FOR THE JURY REDUCED SOLUTION FINITE
SOURCE MODEL
C-58
-------
TRIALLATE 10 PPM
(1 OF 2)
Chemical
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Sample
Point
1
2
3
4
5
6
7
8
9
10
n
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Initial
soil cone.
Cn
(mg/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
10
10
10
10
10
Initial soil
cone
C0
(g/g)
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
1.00E-05
Emitting
area
(cm')
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
Soil
Depth
(L)
(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
Soil Type
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
San Joaquin Sandy Loam
San Joaquin Sandy Loam
San Joaquin Sandy Loam
San Joaquin Sandy Loam
San Joaquin Sandy Loam
Soil bulk
density,
Pb
(Kg/L)
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
Soil
particle
density,
Ps
(Kg/L)
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
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
Gravi-
metric soil
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
Water-
filled soil
porosity,
e
(unitless)
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
0.2787
0.2787
0.2787
0.2787
0.2787
0.2787
0.2787
0.2787
0.2787
Solubility
S
(mg/L)
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
Soil
organic
carbon,
for
(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
0.00.72
0.00.72
0.00.72
0.0072
0.0072
0.0072
Saturation
cone.,
c ,
(mg/kg)
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
105
Cn>CQI
(Yes/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
(u.g/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
0.082
0.083
0.080
0.080
0.072
0071
0070
0.070
Organic
carbon
part.
coeff.,
K«.
(cmj/g)
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
3600
C-59
-------
TRIALLATE 10
(2 OF 2)
PPM
Chemical
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triallate
Triailate
Trialiate
Triallate
Triallate
Soil/ water
part, coeff.,
Kn
(cmj/g)
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
25.92
Diffusivity in
air,
Dra
(crm/s)
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
Diffusivity in
water,
Dw
(cm'/s)
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
5.00E-06
Effective
diffusion
coefficient,
DF
(cm'/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
4.14E-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
Henry's law
constant,
KH
(unitless)
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
Total soil
porosity,
0
(unitless)
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
Air-filled soil
porosity,
(unitless)
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
0.2156
Measured
emission flux
(|ig/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
0.082
0.083
0.080
0.080
0.072
0.071
0.070
0070
Time, t
Cumulative
(hours)
3
6
12
24
48
72
96
120
144
168
192
216
240
264
288
312
336
360
384
408
432
456
480
504
528
552
576
600
624
648
672
696
Jury finite
source model
emission flux
(|ig/cm' day)
1.278
0.904
0.639
0.452
0.320
0.261
0.226
0.202
0.184
0.171
0.160
0.151
0.143
0.136
0.130
0.125
0.121
0.117
0.113
0.110
0.107
0.104
0.101
0.099
0.096
0.094
0.092
0.090
0.085
0.087
0.085
0.084
C-60
-------
APPENDIX D
Revisions to VF and PEF Equations (EQ, 1994b)
-------
ENVIRONMENTAL QUALITY MANAGEMENT, INC.
MEMORANDUM
TO: Ms. Janine Dinan
SUBJECT: Revisions to VF and PEF Equations
FILE: 5099-3
DATE:
FROM:
cc:
July 11, 1994
Craig Mann
Subsequent to the evaluation of the dispersion equations in the RAGS - Part B performed
by Environmental Quality 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/m3 per g/m2-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 (EPA, 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/m3 per g/m2-s; the inverse
of the normalized concentration resulted in a dispersion coefficient (Q/C) of 101.8 g/m2-s per
kg/m3.
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 nearfield 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.
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:
X =
QAK
2;ru
VD
f -^- \
J x ^ ^ J J
exp
dy
dx
(1)
where
QA
K
Area source emission rate (g/m2-s)
= Units scaling coefficient
D-l
-------
V = Vertical term
D = Decay term.
The integral in the lateral (i.e., crosswind or y) direction is solved analytically as:
(Y"
cfy = erfc — | (2)
\°y
where erfc is the complementary error function.
The integral in the longitudinal (i.e., 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, Line (PAL 2.0) model. Wind tunnel tests have also shown that the new
algorithm performs well with on-site and near-field receptors.
Because the new algorithm provides better concentration estimates and does not require
source subdivision, a revised dispersion analysis was performed for both volatile and particulate
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 the OAQPS 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/m2-s; concentrations were calculated in units of kg/m3.
Figure 1 shows the relationship between source size (acres) and annual average
concentration (kg/m3) 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 particulate 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.
D-2
-------
0.10
CO
E:
O
CL
LU
O
Z
O
O
0.04238
0.03680
0.02849
0.02167
0.01453
0.01 "1 1 1 1—I I I I I I 1 1 1—I I I I I I 1 1 1—I I I I I I 1 1 1—Mil!
0.1 1 10 100 1000
SOURCE SIZE (Acres)
Figure 1
Normalized Annual Average Concentration Versus Source Size
D-3
-------
Each of the 29 sites from EQ, 1993 were subsequently modeled at an emission rate of 1.0
g/m2-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
the set of SS normalized annual average concentrations, 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.
In order to determine the average and high end sites for particulate 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:
_ ro.036(l-V)x(U./U..7)'xF(x)-
V ; [ 3600 s/h
where C = Annual average PM10 concentration, kg/m3
(C/Q) = Normalized annual average concentration (kg/m3 per g/m2 -s)
V = Fraction of continuous vegetative cover
Um = Mean annual windspeed, m/s
Ut_7 = Equivalent threshold value of windspeed at 7 m, m/s
F(x) = 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 (Um) 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 Ut 7 was calculated as follows:
"--XT)
where Ut7 = Equivalent threshold value of windspeed at 7m, m/s
Zo = Surface roughness height, cm (zo = 0.5 cm for open terrain)
Ut = Threshold friction velocity, m/s (Ut = 0.625 m/s).
Table 2 gives the results of this analysis and shows the relative PM10 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.
D4
-------
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
Las Vegas
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
(QIC)
(g/m2-s per
kg/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
82.71
83.19
84.18
85.40
89.53
90.09
90.74
95.51
97.75
100.00
30 Acre
(QIC)
(g/m2-s per
kg/m3)
27.08
31.85
32.63
35.10
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
Percentile
(%)
100
96
93
89
86
82
79
75
71
68
64
61
57
54
50
46
43
39
36
32
29
25
21
18
14
11
7
4
0
D-5
-------
TABLE 2. PEF.CALCULATIONS AND SITE RANKINGS
City
Casper
Cleveland
Lincoln
Nbmeapois
Bismarck
Chicago
Philadelphia
Miami
Altanta
Seattle
Boise
Las Vegas
Albuquerque
Denver
Salt Lake City
Portland
Charleston
Hartford
San Francisco
Little Rock
Winnemucca
Houston
Raleigh-
Durham
Harrisburg
LosAngeles
Salem
Huntington
Fresno
Phoenix
NWS
surface
station
number
24089
14820
14939
14922
24011
94846
13739
12835
13874
24233
24131
23165
23050
23062
24127
14762
13880
14764
23234
13963
24128
12960
13722
14751
24174
2423 2
13860
93193
23183
Mean
annual
wind-
speed
(mph)
12.9
10.8
10.4
10.5
10.3
10.4
9.6
9.2
9.1
9.1
8.9
9.1
9.0
8.8
8.8
8.7
8.7
8.6
8.7
8.0
7.9
7.8
7.7
7.7
7.4
7.0
6.5
6.4
6.3
Mean
annual
wind-
speed
(mis)
5.77
4.83
4.65
4.69
4.60
4.65
4.29
4.11
4.07
4.07
3.98
4.07
4.02
3.93
3.93
3.89
3.89
3.84
3.89
3.58
3.53
3.49
3.44
3.44
3.31
3.13
2.91
2.86
2.82
Roughness
height, Zo t
(cm)
0.5
0.5
0.5.
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Threshold
friction
velocity at
surface
(mis)
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
0.625
Threshold
friction
velocity at
7m
(mis)
11.32
11.32
11.32
11.32
11. 3Z
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
11.32
X
1.74
2.08
2.16
2.14
2.18
2.16
2.34
2.44
2.47
2.47
2.52
2.47
2.49
2.55
2.55
2.58
2.58
2.61
2.58
2.80
2.84
2.88
2.91
2.91
3.03
3.21
3.45
3.51
3.56
F(x),
x<=2
0.57
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
F(x),
x>2
NA
2.32E-01
1.82E-01
1.94E-01
1.70E-01
1.82E-01
9.93E-02
6.82E-02
6.16E-02
6.16E-02
4.95E-02
6.16E-02
5.53E-02
4.41E-02
4.41E-02
3.91 E-02
3.91 E-02
3.45E-02
3.91 E-02
1.45E-02
1.23E-02
1.03E-02
8.60E-03
8.60E-03
4.74E-03
1.87E-03
4.45E-04
3.19E-04
2.25E-04
Vegetative
cover
(fraction)
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
PM10
emission
flux
(g/m2-s)
3.77E-07
9.01 E-08
6.30E-08
6.92E-08
5.73E-08
6.30E-08
2.71 E-08
1.64E-08
1.43E-08
1.43E-08
1.07E-08
1.43E-08
1.24E-08
9.25E-09
9.25E-09
7.93E-09
7.93E-09
6.76E-09
7.93E-09
2.29E-09
1.86E-09
1.51E-09
1.21E-09
1.21E-09
5.92E-10
1.98E-10
3.76E-11
2.58E-11
1.73E-11
0.5 Acre
(QIC)
(g/m2-s
per
kg/m3)
100.00
83.19
81.63
90.74
83.40
97.75
90.09
85.40
77.16
82.71
69.40
95.51
84.18
75.59
78.06
74.24
74.91
71.33
89.53
73.37
69.25
79.24
77.461
81.90
68.82
73.42
52.77
62.00
64.06
0.5 Acre
annual
average
cone.
(ug/m3)
3.77
1.08
077
0.76
0.69
0.64
0.30
0.19
0.19
0.17
0.15
0.15
0.15
0.12
0.12
0.11
0.11
0.095
0.089
0.031
0.027
0.019
0.016
0.015
8.60E-03
2.69E-03
7.13E-04
4.16E-04
2.71 E-04
30 Acre
(QIC)
(g/m2-s
per
kg/ m3)
51.68
43.03
41.56
46.84
42.72
50.45
46.38
43.57
39.68
42.81
35.69
49.48
43.31
38.80
40.14
37.86
38.42
36.64
46.06
37.68
35.49
40.70
39.87
42.34
35.10
37.88
27.08
31.85
32.63
30 Acre
annual
average
cone
(ug/ m3)
7.29
2.09
1.52
1.48
1.34
1.25
0.58
0.38
0.36
0.33
0.30
0.29
0.29
0.24
0.23
0.21
0.21
0.18
0.17
0.061
0.052
0.037
0.030
0.029
0.017
5.22E-03
1.39E-03
8.09E-04
5.31 E-04
Site
ranking
percentile
(%)
100
96
93
89
86
82
79
75
71
68
64
61
57
54
50
46
43
39
36
32
29
25
21
18
14
n
7
4
0
F(x) <= 2 from Cowherd (1985), Figure 4-3.
F(X) > 2 from Cowherd (1985), Appendix B.
NA = Not Applicable.
D-6
-------
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.
TABLE 3.
VF AND PEF VALUES OF (QIC) FOR AVERAGE
AND HIGH END EXPOSURES
Site size
0.5 Acres
30 Acres
Average
annual
cone.,
PM10
(ug/m3)
0.12
0.23
High End
annual
cone.,
PM10
(ug/m3)
0.76
1.48
PEF
Average
(QIC),
(g/m2-s per
kg/m3)
78.06
40.14
PEF
High End
(QIC),
(g/m2-s per
kg/m3)
90.74
46.84
VF
Average
(QIC),
(g/m2-s per
kg/m3)
78.06
40.14
VF
High End
(QIC),
(g/m2-s per
kg/m3)
68.82
35.10
Average Site for PM10= Salt Lake City
Average Site for Volatiles = Salt Lake City
High End Site for PM10 = Minneapolis
High End Site for Volatiles = Los Angeles
Average Site for PM10: Mean annual windspeed (UJ = 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 (Z0) = 0.5 cm.
Threshold friction velocity (Ut) = 0.625 m/s at surface.
Threshold windspeed at 7 meters (Ut-7) = U/0.4 x In(700/Z0) = 11.32 m/s.
D-7
-------
ATTACHMENT A
AREA-ST MODEL RUN SHEETS FOR A 0.5 ACRE SQUARE AREA SOURCE
CO STARTING
CD TITLEONE
CD MODELOPT
CD AVERTIME
CD POLLUTID
CD RUNORNOT
CD ERRORFIL
CD FINISHED
SO STARTING
SRCID
SO LOCATION Al/2
SRCID
AREA SOURCES-
DFAULT CONC
PERIOD
PM10
RUN
AREAl.ERR
SRCTYP
AREA
CS
-- 1/2 acre run
RURAL
XS YS
-22.5 -22.5
HS XINIT
ZS
.0000
YINIT
SO SRCPARAH Al/2 1.0
SO EHISUNIT .100000E-02
SO SRCGROUP AREA1 Al/2
SO FINISHED
0.0
45.
RE
ME
ME
ME
ME
ME
FINISHED
STARTING
INPUTFIL
ANEHHGHT
SURFDATA
UAIRDATA
(GRAMS/(SEC-M**2) ]
45.
KILOGRAMS/CUBIC-METER
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
STARTING
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DSSCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
DISCCART
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.
o.
0.
o.
25.
-25.
-25.
25.
0.
0.
50.
-50.
-50.
50.
0.
0.
75.
-75.
-75.
75.
0.
0.
100.
-100
-100
100.
C:\CRAIG\23174-89.ASC
10.0 METERS
23174 1989 LOS ANGELES
23230 1989 OAKLAND
D-8
-------
ME WINDCATS 1.54 3.09 5.14 8.23 10.80
ME FINISHED
OU STARTING
OU RECTABLE ALLAVE FIRST
OU FINISHED
*** SETUP Finishes Successfully ***
***************************************************
D-9
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES---1/2 acre run***
TEST OF ST AREA SOURCE ALGORITHM *** ***
*** MODELING OPTIONS USED: CONG RURAL FLAT DFAULT
**Model Is Setup For Calculation of Average CONCentration Values.
**Model Uses RURAL Dispersion.
**Model Uses Regulatory DEFAULT Options:
1. Final Plume Rise.
2. Stack- tip Downwash.
3. Buoyancy- induced Dispersion.
4 . Use Calms Processing Routine .
5. Not Use Missing Data Processing Routine.
6. Default Wind Profile Exponents.
7. Default Vertical Potential Temperature Gradients.
8. "Upper Bound" Values for Supersquat Buildings.
9. No Exponential Decay for RURAL Mode
**Model Assumes Receptors on FLAT Terrain.
**Model Assumes No FLAGPOLE Receptor Heights.
**Model Calculates PERIOD Averages Only
**This Run Includes: 1 Source (s); 1 Source Group (s); and 25 Receptore(s)
**The Model Assumes A Pollutant Type of: PM10
**Model Set To Continue RUNning After the Setup Testing. ff
**0utput Options Selected:
Model Outputs Tables of PERIOD Averages by Receptor
Model Outputs Tables of Highest Short Term Values by Receptor (RECTABLE Keyword)
**NOTE: The Following Flags May Appear Following CONG Values:
c for Calm Hours
m for Hissing Hours
b for Both Calm and Missing Hours
**Misc. Inputs:
Anem. Hgt. (m) = 10.00 ; Decay Coef . =.0000 ; Rot. Angle = .0
Emission Units = (GRAMS/ (SEC-M**2) ); Emission Rate Unit Factor = .1COOOE-02
Output Units = KILOGRAMS/CUBIC-METER
**input Runstream File: areal.dat, **0utput Print File: areal.out
**Detailed Error /Message File: AREA1.ERR
D-10
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run
TEST OF ST AREA SOURCE ALGORITHM ***
***
MODELING OPTIONS USED: CONG RURAL FLAT
DFAULT
***
***
*** AREA SOURCE DATA ***
SOURCE
ID
A1/2
NUMBER
PART.
CATS.
0
EMISSION RATE
(USER UNITS
/METERS"2)
.10000E+01
*** AREAST - VERSION
TEST OF
COORD
X
(METERS)
-22.5
TESTA
(SW CORNER)
Y
(METERS)
-22.5
*** ***
BASE
ELEV.
(METERS)
.0
AREA SOURCES -
RELEASE
HEIGHT
(METERS)
.00
-- 1/2
X-DIM OF
AREA
(METERS)
45.00
acre run
Y-DIM OF
AREA
(METERS)
45.00
ORIENT.
OF AREA
PEG.)
.00
ST AREA SOURCE ALGORITHM ***
EMISSION
RATE SCALAR
VARY BY
***
***
MODELING OPTIONS USED: CONG RURAL FLAT DFAULT
GROUP ID
AREA1 Al/2
*** SOURCE IDs DEFINING SOURCE GROUPS **"
SOURCE IDs
D-ll
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run
TEST OF ST AREA SOURCE ALGORITHM *** ***
*** MODELING OPTIONS USED: CONG RURAL FLAT DFAULT
*** 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); ( -25.0, -25.0, .0, .0);
-25.0 25.0, .0, .0); ( 50.0, .0, .0, .0);
-50.0, .0, .0, .0); ( 50.0, 50.0, .0, .0);
50.0 -50.0, .0, .0); ( -50.0, -50.0, .0, .0);
-50.0 50.0, .0, .0); ( 75.0, .0, .0, .0);
-75.0, .0, .0, .0); ( 75.0, 75.0, .0, .0);
75.0, -75.0, .0, .0); ( -75.0, -75.0, .0, .0);
-75.0, 75.0, .0, .0); ( 100.0, .0, .0, .0);
-100.0, .0, .0, .0); ( 100.0, 100.0, .0, .0);
100.0, 100.0, .0, .0); ( -100.0, -100.0, .0, .0);
-100.0, 100.0, .0, .0);
D-12
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run
TEST OF ST AREA SOURCE ALGORITHM *** ***
*** MODELING OPTIONS USED CONG RURAL FLAT DFAULT
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
*** METEOROLOGICAL DAYS SELECTED FOR PROCESSING ***
(1=YES; 0=NO)
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
1111111111
NOTE: METEOROLOGICAL DATA ACTUALLY PROCESSED WILL ALSO DEPEND ON WHAT IS INCLUDED IN THE DATA FILE
*** UPPER BOUND OF FIRST THROUGH FIFTH WIND SPEED CATEGORIES ***
(HETERS/SEC)
1.54, 3.09, 5.14, 8.23, 10.80,
*** WIND PROFILE EXPONENTS ***
STABILITY
CATEGORY
A
B
C
D
E
F
.70000E-01
.70000E-01
.10000E+00
.15000E+00
.35000E+00
.55000E+00
WIND
2
70000E-01
70000E-01
10000E+00
15000E+00
35000E+00
55000E+00
SPEED CATEGORY
3 4
70000E-01 .70000E-01
70000E-01 .70000E-01
10000E+00 .10000E+00
15000E+00 .15000E+00
35000E+00 .35000E+00
55000E+00 .55000E+00
.70000E-01
.70000E-01
.10000E+00
.15000E+00
.35000E+00
.55000E+00
.70000E-01
.70000E-01
.10000E+00
.15000E+00
.35000E+00
.55000E+00
STABILITY
CATEGORY
A
B
C
D
E
F
*** VERTICAL POTENTIAL TEMPERATURE GRADIENTS ***
(DEGREES KELVIN PER METER)
WIND SPEED CATEGORY
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
.OOOOOE+00
.OOOOOt+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.20000E-01
.35000E-01
D-13
-------
*** AREAST - VERSION TESTA ***
TEST OF ST AREA SOURCE ALGORITHM
*** AREA SOURCES--- 1/2 acre run
***
***
***
*** MODELING OPTIONS USED: CONG RURAL FLAT DFAULT
*** THE FIRST 24 HOURS OF METEOROLOGICAL DATA ***
FILE: C:\CRAIG\23174-89.ASC
SURFACE STATION NO : 23174
NAME: LOS
YEAR: 1989
FORMAT: (412,2F9.4,F6.1,12,2F7.1)
UPPER AIR STATION NO.: 23230
NAME: OAKLAND
YEAR: 1989
YEAR
MONTH DAY
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
HOUR
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
FLOW
VECTOR
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
SPEED
(M/S)
3.09
3.09
2.57
4.63
2.06
3.09
2.06
2.06
2.06
.00
2.06
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
TEMP
(K)
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
STAB
MIXING
CLASS RURAL
4
4
4
4
5
6
6
5
4
3
3
3
3
3
3
4
5
6
5
6
6
6
6
7
533.0
568.6
604.1
639.6
675.2
710.7
746.3
134.9
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
HEIGHT
URBAN
533.0
568.6
604.1
639.6
151.0
151.0
151.0
265.4
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
(M)
*** NOTES: STABILITY CLASS 1=A, 2=B, 3=C, 4=D, 5=E AND 6=F.
FLOW VECTOR IS DIRECTION TOWARD WHICH WIND IS BLOWING.
D-14
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run
TEST OF ST AREA SOURCE ALGORITHM ***
***
***
*** MODELING OPTIONS USED: CONG RURAL FLAT
DFAULT
*** THE PERIOD ( 8760 HRS) AVERAGE CONCENTRATION VALUES FOR SOURCE GROUP: AREA1 ***
INCLUDING SOURCE(S): Al/2,
*** DISCRETE CARTESIAN RECEPTOR POINTS ***
** CONG OF PM10
IN KILOGRAMS/CUBIC-METER
X-COORD (M)
Y-CCORD (M)
CONE
X-COORD (M) )
Y-COORD (M
CONG
.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
.00594
.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-15
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run
TEST OF ST AREA SOURCE ALGORITHM ***
*** MODELING OPTIONS USED: CONG RURAL FLAT
DFAULT
*** THE SUMMARY OF MAXIMUM PERIOD ( 8760 HRS) RESULTS ***
** CONG OF PM10IN KILOGRAMS/CUBIC-METER **
GROUP ID
AVERAGE CONG
NETWORK
RECEPTOR (XR, YR, ZELEV, ZFLAG) OF TYPE GRID-ID
AREA1 1ST HIGHEST VALUE IS .01453 AT
2ND HIGHEST VALUE IS .00679 AT
3RD HIGHEST VALUE IS .00594 AT
4TH HIGHEST VALUE IS .00414 AT
5TH HIGHEST VALUE IS .00223 AT
6TH HIGHEST VALUE IS .00220 AT
*** RECEPTOR TYPES:
GC = GRIDCART
GP = GRIDPOLR
DC = DISCCART
DP = DISCPOLR
BD = BOUNDARY
.00,
25.00,
-25.00,
25.00,
-25.00,
-25.00,
.00,
.00,
.00,
25.00,
25.00,
-25.00,
.00,
.00,
.00,
.00,
.00,
.00,
.00)
.00)
.00)
.00)
.00)
.00)
DC
DC
DC
DC
DC
DC
D-16
-------
*** AREAST - VERSION TESTA *** *** AREA SOURCES--- 1/2 acre run ***
TEST OF ST AREA SOURCE ALGORITHM *** ***
*** MODELING OPTIONS USED: CONG RURAL FLAT DFAULT
*** Message Summary For ISC2 Model Execution ***
Summary of Total Messages
A Total of 0 Fatal Error Message(s)
A Total of 0 Warning Message(s)
A Totat of 653 Informational Message(s)
A Total of 653 Calm Hours Identified
FATAL ERROR MESSAGES
*** NONE ***
WARNING MESSAGES
*** NONE ***
*** ISCST2 Finishes Successfully ***
D-17
-------
APPENDIX E
Determination of Ground Water Dilution Attenuation
Factors
-------
DETERMINATION OF GROUNDWATER
DILUTION ATTENUATION FACTORS
FOR FIXED WASTE SITE AREAS
USING EPACMTP
BACKGROUND DOCUMENT
EPA OFFICE OF SOLID WASTE
May 11, 1994
-------
TABLE OF CONTENTS
Page
PREFACE i
ABSTRACT ii
1.0 INTRODUCTION E-l
2.0 GROUNDWATER MODEL E-2
2.1 Description of EPACMTP Model E-2
2.2 Fate and Transport Simulation Modules E-2
2.2.1 Unsaturated zone flow and transport module E-2
2.2.2 Saturated zone flow and transport module E-4
2.2.3 Model capabilities and limitations E-5
2.3 Monte Carlo Module E-7
2.3.1 Capabilities and Limitations of Monte Carlo Module E-10
3.0 MODELING PROCEDURE E-11
3.1 Modeling Approach E-l 1
3.1.1 Determination of Monte Carlo Repetition Number and
Sensitivity Analysis E-l 1
3.1.2 Analysis of DAF Values for Different Source Areas E-12
3.1.2.1 Model Options and Input Parameters E-13
4.0 RESULTS E-18
4.1 Convergence of Monte Carlo Simulation E-18
4.2 Parameter Sensitivity Analysis E-18
4.3 DAF Values as a Function of Source Area E-21
REFERENCES E-29
-------
LIST OF FIGURES
Page
Figure 1 Conceptual view of the unsaturated zone-saturated zone
system simulatedby EPACMTP E-3
Figure 2 Conceptual Monte Carlo framework for deriving probability
distribution of model output from probability distributions of
input parameters E-8
Figure 3 Flow chart of EPACMTP for Monte Carlo simulation E-9
Figure 4 Plan view and cross-section view showing location of receptor well E-15
Figure 5 Variation of DAF with size of source area for the base case
scenario (x= 25 ft. y=uniform in plume, z=nationwide
distribution) E-22
Figure 6 Variation of DAF with size of source area for the default
nationwide scenario (Scenario 2: x=nationwide distribution,
y=uniform in plume, z=nationwide distribution) E-23
Figure 7 Variation of DAF with size of source area for Scenario 3
(x = O, y =uniform within half-width of source area,
z=nationwide distribution) E-24
Figure 8 Variation of DAF with size of source area for Scenario 4
(x = 25 ft, y=uniform within half-width of source area,
z=nationwide distribution) E-25
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) E-26
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) E-27
-------
LIST OF TABLES
Page
Table 1 Parameter input values for model sensitivity analysis E-12
Table 2 Summary of EPACMTP modeling options E-13
Table 3 Summary of EPACMTP input parameters E-14
Table 4 Receptor well location scenarios E-16
Table 5 Distribution of aquifer particle diameter E-17
Table 6 Variation of DAF with number of Monte Carlo repetitions E-19
Table 7 Sensitivity of model parameters E-20
Table 8 DAF values for waste site area of 150,000 ft2 E-28
Table Al DAF values as a function of source area for base case scenario
(x=25 ft. y=uniform in plume, z-nationwide distribution) E-31
Table A2 DAF values as a function of source area for Scenario 2 (x=nationwide
distribution, y=uniform in plume, z=nationwide distribution) E-32
Table A3 DAF values as a function of source area for Scenario 3 (x=0 ft.
y=llniform within half-width of source area, z=nationwide
distribution) E-33
Table A4 DAF values as a function of source area for Scenario 4 (x=25 ft.
y=uniform within half-width of source area, z=nationwide
distribution) E-34
Table A5 DAP values as a function of source area for Scenario 5 (x=100 ft.
y=uniform within half-width of source, z=nationwide distribution) E-35
Table A6 DAF values as a function of source area for Scenario 6 (x=25 ft.
y=source width + 25 ft. z=25 It) E-36
-------
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-i
-------
ABSTRACT
The EPA Composite Model for Leachate Migration with Transformation Products (EPACMTP)
was applied to generate Dilution Attenuation Factors (DAF) for the 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 the
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 than 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 this number of iterations. Second, Monte Carlo analyses were performed to
determine DAF values as a function of the contaminated area. The effects of different placements of
the receptor well were evaluated.
E-ii
-------
1.0 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 purpose of the SSLs is to accelerate decision making concerning
contaminated soils. Generally, if contaminant concentrations in soil fall below the screening level
and the 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 of contaminants through soil to an underlying potable aquifer. Contaminant migration
through the unsaturated zone to the water table generally reduces the soil leachate concentration by
attenuation processes such as adsorption and degradation. Groundwater transport in the saturated
zone further reduces concentrations through attenuation and 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 ratio of original soil leachate concentration to the receptor point concentration.
The lowest possible value of DAF is therefore one; a value of DAF=1 means that there is no
dilution or attenuation at all; the concentration at the receptor point is the same as that in the soil
leachate. High values of DAF on the other hand correspond to a high degree of dilution and
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 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 Leachate 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
transport of contaminants after their release from the land disposal unit into the soil, 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 part of the development of Soil Screening Levels. This report describes the
application of EPACMTP for this purpose.
E-l
-------
2.0 GROUNDWATER MODEL
2.1 Description of EPACMTP Model
The EPA Composite Model for Leachate Migration with Transformation Products
(EPACMTP, EPA, 1993a, 1994) is a computer model for simulating the subsurface fate and
transport of contaminants that are released at or near the soil surface. A schematic view of the
conceptual subsurface system as simulated by EPACMTP, is shown in Figure 1. The contaminants
are initially released over a rectangular 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 aqueous phase; it migrates through
the soil 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 that 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 unsaturated zone flow and transport module
A saturated zone flow and transport 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 transport modules are
described in Section 2.2. The Monte Carlo modeling procedure is described in Section 2.3.
2.2 Fate and Transport Simulation Modules
2.2.1 Unsaturated zone flow and transport module
Details on the mathematical formulation and solution techniques of the unsaturated 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 the 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.
E-2
-------
Unsaturated
Zone
Saturated ~~
Zone
Ambient
Groundwater Flow
Leakage from
Contaminated Area
Figure 1
Conceptual View Of The Unsaturated Zone-Saturated Zone
System Simulated By EPACMTP
E-3
-------
• 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 soil solid phase only.
• Sorption of chemicals onto the soil solid phase is described by a linear or nonlinear
(Freundlich) equilibrium isotherm.
• Chemical and biological transformation process can be represented by an effective,
first-order decay coefficient.
2.2.2 Saturated zone flow and transport module
The unsaturated zone module computes the contaminant concentration arriving at the water
table, as a function of time. Multiplying this concentration by the rate of infiltration through the
unsaturated zone yields the contaminant mass flux entering the saturated zone. This mass flux is
specified as the source boundary condition for the saturated zone flow and transport module.
Groundwater flow in the saturated zone is simulated using a (quasi-) three-dimensional
steady state solution for predicting hydraulic head and Darcy velocities in 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 three-dimensional groundwater flow and contaminant fate and
transport, EPACMTP offers the option to perform quasi-3D modeling. When this option is
selected, the model ignores either the flow component in the horizontal transverse (-y) direction, or
the vertical (-z) direction. The appropriate 2D approximation is selected automatically in the code,
based on the relative significance of plume movement in the horizontal transverse versus vertical
directions. Details of this procedure are provided in the saturated zone background document
(EPA, 1993a). The switching criterion that is implemented in the code will select the 2D areal
solution for situations with a relatively thin saturated zone in 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 plume movement is more important.
The saturated zone transport module describes the advective-dispersive transport of
dissolved contaminants in a three-dimensional, constant thickness 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 saturated 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 the 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. Transport mechanisms 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 can simulate multi-species transport involving chained decay reactions. The saturated zone
transport module of EPACMTP can perform either a fully three-dimensional transport simulation,
or provide a quasi-3D approximation. The latter ignores 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.
E-4
-------
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 saturated thickness of the aquifer remains constant; mounding is represented by
the head distribution along the top boundary of the modeled saturated zone system.
• Flow is isothermal and governed by Darcy's Law.
• The chemical is dilute and present in the solution or aquifer 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 capabilities and limitations
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 detail here. The user should verify that the
assumptions are reasonable for a given application.
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 transfer processes other than 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 hydrophobic chemicals may be underpredicted by the 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
handling a wide variety of geochemical conditions is currently underway.
E-5
-------
First-Order Decay. It is assumed that the rate of contaminant loss 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 must determine
the overall, effective decay rate. In order to increase flexibility of the model, the user may instruct
the model to determine the overall decay coefficient from chemical specific 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 speciation factors
are specified by the user as constants (see EPACMTP Background Document, EPA, 1993a). In
reality, these coefficients may change as functions of aquifer conditions (temperature, pH, etc.)
and/ or concentration levels of other chemical components.
Uniform Soil. EPACMTP assumes that the unsaturated 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 transport
behavior when the waste source covers a large area.
Steady-State Flow in the Unsaturated-Zone. Flow in the unsaturated zone is always treated
as steady state, with the flow rate determined by the long term, average infiltration rate through a
disposal unit, or by the average depth of ponding in a surface impoundment. Considering the time
scale of most practical problems, assuming steady-state flow conditions in the unsaturated zone is
reasonable.
Groundwater Mounding. The saturated zone module of EPACMTP is designed to simulate
flow and transport in an unconfmed aquifer. Groundwater mounding beneath the source is
represented only by increased head values on top of the aquifer. The saturated thickness of the
aquifer remains constant in the model, and therefore the model treats the aquifer as a confined
system. This approach is reasonable as long as the mound height is small relative to the saturated
thickness of the aquifer and the thickness of the unsaturated zone. For composite modeling, the
effect of mounding is partly accounted for in the unsaturated zone module, since the soil is allowed
to become saturated. The aquifer porous material is assumed to be uniform, 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 direction, with vertical disturbance
due to recharge and infiltration from the overlying unsaturated zone and waste site (source area).
EPACMTP accounts for variable recharge rates underneath and outside the source area. It is,
however, assumed that the saturated zone has a constant thickness, which may cause inaccuracies
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 and dispersion. The aquifer is assumed to be initially contaminant free and contaminants
enter the aquifer only from the unsaturated 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
and decay reactions, both of which are modeled in the same manner as in the unsaturated zone. The
adsorption and decay coefficients are assumed to be uniform throughout saturated zone.
E-6
-------
2.3 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 this section.
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 then
statistically analyzed to yield the cumulative probability distribution of the model output. Thus, the
various steps involved in the application of the Monte Carlo simulation technique are:
(1) Selection of representative cumulative probability distribution functions for the
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 and transport simulation modules 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 simply 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 (plus 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 their 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.
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 then assigned to the model variables. Following this,
the contaminant transport fate and 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.
E-7
-------
Input
O
D
E
L
Ou f pu t
cr
O)
O
Value
Input Distributions
O
Value
Output Distribution
Figure 2
Conceptual Monte Carlo Framework For Deriving Probability Distribution
Of Model Output From Probability Distributions Of Input Parameters
E-8
-------
Read input data
and desired
modeling options
Yes
Post Processing
Perform Simulation
Print Results
Figure 3
Flow Chart Of EPACMTP For Monte Carlo Simulation
E-9
-------
2.3.1 Capabilities and Limitations of Monte Carlo Module
The Monte Carlo module in EPACMTP is implemented as a flexible module that can
accommodate a wide variety of input distributions. These include: constant, normal, lognormal,
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, the 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; it does, however, not provide a way to account or compensate for process uncertainty.
If the actual flow and transport processes that may occur at different sites, are different from those
simulated 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 bounds on the
parameters prevent unrealistically low or high values from being generated at all.
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-10
-------
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 the input
parameter values.
3.1 Modeling Approach
The overall modeling approach consisted of two stages. First, a sensitivity analysis was
performed to determine the 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 Carlo analyses were performed to determined DAF values as a
function of the size of the source area, for various scenarios of receptor well placement.
3.1.1 Determination of Monte Carlo Repetition Number 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 the number of repetitions is varied. A Monte
Carlo simulation comprising 20,000 repetitions was first made. The results from this simulation
were analyzed by calculating the 85th percentile DAP value obtained by sampling model output
sequences of different length, from 2,000 to the full 20,000 repetitions. The modeling scenario
considered in this analysis was the same as that in the base case scenario discussed 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 theEPACMTP User's Guide (EPA, 1993b).
For each parameter, the low, medium, and high values were selected, corresponding to the
15th, 50th, and 85th 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 the model
output are automatically accounted for. Each of the Monte Carlo simulations yields a probability
distribution of predicted receptor well concentrations. Evaluating the 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 2,000 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 that are
independent of contaminant-specific chemical properties, e.g., depth to water table, aquifer
thickness, receptor well distance, etc. The second category encompasses those parameters that 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
parameters has examined, by considering a non-degrading, non-sorbing contaminant. 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 the same as that entering the unsaturated
E-ll
-------
zone. The only exception to this 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 silty 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.
Table 1
Parameter input values for model sensitivity analysis.
Parameter
Low
Median
High
Source Parameters
Source Area (m2)
Infiltration Rate (m/yr)
Recharge Rate (m/yr)
4.8x104
6.0x10-4
6.0x1fJ-4
2.8x105
6.4x1 0-3
8.0x10-3
1.1 x106
1.7x10'1
1.5x10'1
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)
15.55
1.9x103
4.3 x10-3
53.2
0.374
4.2
0.53
0.026
60.8
1.5x104
1.8x10'2
404.0
0.415
12.7
1.59
0.079
159.3
5.5 X104
5.0X10-2
2883.0
0.455
98.5
12.31
0.62
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 in
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.
E-12
-------
3.1.2.1 Model Options and Input Parameters
Table 2 summarizes the EPACMTP model options used in performing the simulations. Model
input parameters used are summarized in 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.
Table 2
Summary of EPACMTP modeling options.
Option
Value Selected
Simulation Type
Number of Repetitions
Nationwide Aggregation
Source Type
Unsat. Zone Present
Sat. Zone Model
Contaminant Degradation
Contaminant Sorption
Monte Carlo
15,000
Yes
Continuous
Yes
Quasi-3D
No
No
Source Area
In the default, nationwide modeling scenario, the waste site 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 in each simulation run.
Receptor Well Location
In the default nationwide modeling scenario, the position of the nearest downgradient
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 the 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 represents 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 both 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 wells, respectively. The y-coordinate value is
determined so that the well location falls within the approximate areal 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 in Table 4.
E-13
-------
Table 3
Summary of EPACMTP input parameters.
Parameter
Value or Distribution Type
Comment
Source-Specific
Area
Infiltration Rate
Recharge Rate
Leachate Concentration
Constant
Soil-type dependent
Soil-type dependent
= 1.0
Varied in each run
default
default
default
Chemical-Specific
Hydrolysis Rate Constants
Organic Carbon Partition Coeff.
= 0.0
= 0.0
Contaminant does not degrade
Contaminant does not sorb
Unsaturated Zone Specific
Depth to Water Table
Dispersivity
Soil Hydraulic Properties
Soil Chemical Properties
Empirical
Soil-depth dependent
Soil-type dependent
Soil-type dependent
default
default
default
default
Saturated Zone Specific
Sat. Zone Thickness
Hydraulic Conductivity
Hydraulic Gradient
Seepage Velocity
Particle Diameter
Porosity
Bulk Density
Longitudinal Dispersivity
Transverse Dispersivity
Vertical Dispersivity
Receptor Well x-coordinate
Receptor Well y-coordinate
Receptor Well z-coordinate
Exponential
Derived from Part. Diam.
Exponential
Derived from Conductivity and
Gradient
Empirical
Derived from Part. Diam
Derived from Porosity
Distance-dependent
Derived from Long. Dispersivity
Derived from Long. Dispersivity
= 25 feet
Within plume
Empirical
default
default
default
default
default
default
default
default
default
default
Set to fixed value
default
default
Note: 'Default' represents default nationwide Monte Carlo scenario as presented in EPACMTP User's
Guide (EPA, 1994).
E-14
-------
GROUNDWATER FLOW
CONTAMINANTED
AREA
RECEPTOR WELL
LAND SURFACE
UNSATURATED
ZONE
GROUNDWATER FLOW
Figure 4
Plain View And Cross-Section View Showing Location Of Receptor Well
E-15
-------
Table 4
Receptor Well Location Scenarios
Scenario
1 (Base Case)
2
3
4
5
6
Xwell
25 ft from edge of source
area
Nationwide Distribution
0 ft from edge of source
area
25 ft from edge of source
area
100 ft from edge of
source area
25 ft from edge of source
area
Ywell
Monte Carlo within plume
Monte Carlo within plume
Monte Carlo within half-
width of source area
Monte Carlo within half-
width of source area
Monte Carlo within half-
width of source area
Width of source area + 25
ft
Zwell
Nationwide Distribution
Nationwide Distribution
Nationwide Distribution
Nationwide Distribution
Nationwide Distribution
25 ft below water table
Xwell = Downgradient distance of receptor well from edge of source area.
Ywell = Horizontal transverse distance from plume centerline.
Zwell = Depth of well intake point below water tablet
The base case scenario (scenario 1) involved setting the x-distance 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 plume. The depth of the well intake point (z-coordinate)
was assumed to vary within upper and lower bounds of 15 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 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, the distance between the receptor well and the source
area is varied from zero to 100 feet. In these scenarios, the ycoordinate 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.
Aquifer Particle Size Distribution
In the default Monte Carlo modeling scenario, the aquifer hydraulic conductivity, porosity,
and bulk density are determined from the mean particle diameter. The particle diameter 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 summarizes the distribution of particle size diameters used in both the default nationwide
modeling scenario and in the present analyses.
E-16
-------
Table 5
Distribution of aquifer particle diameter.
Nationwide Default
Particle Diameter Cumulative
(cm) Probability
3.9 1C'4
7.8 1C'4
1.6 1C'3
3.1 1C'3
6.3 1C'3
1.25 1C'2
2.5 1C'2
5.0 1C'2
1.0 1C'1
2.0 1C'1
4.0 1C'1
8.0 ID'1
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 Analyses
Particle Diameter Cumulative
(cm) Probability
4.0 1C'4
8.0 10-4
1.6 ID'3
3.1 ID'3
6.3 ID'3
1.25 ID'2
2.5 ID'2
5.0 ID'2
1.0 10-1
2.0 ID'1
4.0 ID'1
7.5 ID'1
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-17
-------
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 summarizes the results of this convergence analysis. It shows the variation of the
85th percentile DAF value with the number of Monte Carlo 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 2,000 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 remaining 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.
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 areal recharge rate. To
interpret these results, it should be kept in mind that 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, it directly affects the mass
flux of contaminant entering the subsurface. The size of the source would be expected to be equally
sensitive, were it not for the fact that 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 that the receptor well will be
located underneath the contaminant plume and sample uncontaminated groundwater, leading to a
E-18
-------
Table 6
Variation of DAF with number of Monte Carlo repetitions
No. of
Repetitions
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
85-th Percentile
DAF
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
-10.9
+ 17.3
+5.0
+28.2
-18.1
-0.2
+ 18.2
+0.1
+0.6
-0.7
+2.9
+2.6
Relative
Difference (%)
-3.1
+5.1
+ 1.4
+7.9
-4.7
-0.05
+4.9
+0.03
+0.15
-0.18
+0.75
+0.67
E-19
-------
Table 7
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
11.6
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 = I High-Low I /Median
high DAF value. The dilution-attenuation factors 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 that 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 also directly affects the groundwater
velocity, but was not among the sensitive parameters. This is a reflection of the narrow range of
variation assigned to this 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 depth, but is seen to have a much smaller
sensitivity. This was a result of constraining the y-location 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 that of the vertical coordinate of the well. The low relative
sensitivity of recharge rate reflects the fact that this parameter has an only indirect effect on plume
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 plume which increases the probability that a randomly located receptor well
will lie outside (underneath) the plume. Higher dispersivities will increase the chance that the well
will intercept the plume. At the same time, however, mass balance considerations dictate that in
E-20
-------
this case average concentrations inside the plume will be lower than in the low dispersivity case.
Similar reasoning 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 plume at all. At greater distances from the source, the likelihood that the well is
located inside the plume is greater, but the plume 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, OCT, compared to o^ and ocv can be contributed to the imposed
constraint that the well must always be within the areal extent of the plume.
The results of the sensitivity analysis show that the site characteristic which lends itself
best for a classification system for correlating sites to DAF values is the size of the contaminated
(or 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 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, while the 85th percentile shows the 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 that the relationship
follows the same general shape in each case, but the magnitude 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 allowed to be located
anywhere within the areal extent of the contaminant plume for a fixed x-distance 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 within the half-width of the waste source. In other words, the
well was always placed in the central portion of the contaminant plume where concentrations are
highest. As a result, these scenarios show lower DAF values then the base case scenario. The
results for Scenarios 3, 4, and 5, which differ only in the x-distance of the receptor well, show that
placement of the well at either 25 or 100 feet away from the waste source results in 85% and 90%
DAF values that are actually lower, i.e. more conservative, than placement of the well directly at
the edge of the 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 plume
below the water table. Close to the waste source, the contaminant concentrations within the plume
are highest, but the plume may not have penetrated very deeply into the saturated zone (Figure 2).
E-21
-------
1.0E+073
Ywell - Uniform within plume
1.0E+00-
1.0E+03
I i till
1.0E+04
1.0E4-05
AREA OF LANDFILL (sq. ft)
1.0E+06
85 TH -»- 90 TH
95 TH
Figure 5
Variation Of DAF With Size Of Source For The Base Case Scenario
(x=25 ft, y=uniform in plume, z=nationwide distribution)
1.0E+07
E-22
-------
1.0E + 04zr
1.0E+03;
< 1.0E+02:
Q
1.0E+01:
1.0E+00
1.0E+03
,M£ „
1.0E+04
[Xweil-^^ Nationwide'dislirlbr
Ywell • Monte Carlo within plume
| Zwell - Nationwide dlstrib,
n 1—
1.0E+05
AREA OF LANDFILL
1.0E+06
m
1.0E+07
85 TH -•- 90 TH -S- 95 TH
Figure 6
Variation Of DAF With Size Of Source Area For The Default Nationwided Scenario
(Scenario 2: x=nationwide distribution, y=uniform in plume, z=nationwide distribution)
E-23
-------
1.0E+083
1.0E+07=
1.0E+06H
1.0E+05=
< 1.0E+04;
1.0E+03d
1.0E+02=
1.0E+01 =
1.0E+00
j Yw«ll - 0 to \I2 landlill width \
1.0E+03
1.0E+04
1.0E+05
AREA OF LANDFILL
1.0E+06
1.0E+07
85 TH
90 TH
95 TH
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
E-24
-------
1.0E+06=r
1.0E+05=
1.0E+04r
< 1.0E+03z
1.0E+02=
1.0E+01 =
1.0E+00
Ywell = 0 to 1/2 landfill width
1.0E + 03
1.0E+04
1.0E+05
AREA OF LANDFILL
1.0E+06
1.0E+07
65 TH
90 TH -SK- 95 TH
Figure 8
Variation Of DAF With Size Of Source Area For Scenario 4
(x=25 fy, y=u^orm witmn half-wid«h of source area, z=nationwide distnbut.on)
E-25
-------
1.0E+05:
1.0E+04:
1.0E+03z
1.0E+02z
1.0E+01:
1.0E4-00
1.0E+03
i i ( f |
.'.'[ Xwell = 100ft j
Iff
I
Ywell = 0 to 1/2 landfill width \
.OE+04
1.0E+05
AREA OF LANDFILL (sq ft)
1.0E + 06
1.0E + 07
85 TH
90 TH ->K- 95 TH
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)
E-26
-------
1.00E+05rr
1.00E+04=
1.00E+03=
u.
<
Q
1.00E+02=
1.00E+01 =
1.00E+00
Ywell = 0 to 1/2 landfill width
1.0E4-03
I I I I I
1.0E+04
~\ T
1.0E+05
1.0E+06
AREA OF LANDFILL (sq ft)
1.0E+07
85 TH
90 TH -m- 95 TH
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)
E-27
-------
Table 8
DAF values for waste site area of 150,000 ft2.
DAF Percentile
Model Scenario 85 90 95
1 (base case)
2
3
4
5
6
237.5
300.1
158.8
132.1
98.8
94.7
26.4
114.7
17.9
16.6
15.1
25.3
2.8
26.8
1.7
1.8
2.0
4.4
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
contaminant plume is higher at close distances from the waste area. Conversely, 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 that at the 95% level, the lowest DAF is obtained in the case where the well is
located at the edge of the waste source. This reflects that 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 that the well is placed shallow enough
that it will be located inside the plume in nearly all cases, resulting in low DAF values at the 85th
and 90th percentile values. On the other hand, the well in this case is never placed immediately at
the plume centerline, so that 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.
One of the key objectives of the present analyses was to determine the appropriate
groundwater DAF value for a waste area of given size. For the base case scenario, the 90th
percentile DAF value is on the order of 100 or higher for a waste area size of 1 acre (43,560 ft2)
and less. For waste areas of 10 acres and greater, the 90th percentile DAF is 10 or less.
E-28
-------
REFERENCES
Shea, J.H., 1974. Deficiencies of elastic particles of certain sizes. Journal of Sedimentary
Petrology, 44:905-1003.
U.S. EPA, 1993a. EPA's Composite Model for Leachate 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-Dimensional Migration of Land and
Disposal Leachate with Transformation Products and Consideration of Water-Table
Mounding. Volume II: User's Guide for EPA's Composite Model for Leachate Migration
with Transformation Products (EPACMTP). Office of Solid Waste, May 1994.
E-29
-------
APPENDIX A
E-30
-------
Table A1 DAF values as a function of source area for base case
scenario (x=25 ft., y=uniform in plume, z-nationwide
distribution).
Area (sq. ft.)
1000
2000
5000
10000
30000
50000
70000
80000
150000
200000
500000
1000000
2000000
3000000
5000000
DAF
85th
1.09E+06
1.86E+05
2.91 E+04
9.31 E+03
1647.18
869.57
569.80
477.33
237.47
174.86
64.52
32.27
17.83
12.94
8.91
90th
3.76E+04
9.63E+03
2.00E+03
680.27
155.21
84.25
59.28
50.56
26.36
20.19
9.12
5.61
3.68
2.94
2.33
95th
609.01
187.69
53.02
22.57
7.82
5.41
4.34
3.97
2.77
2.37
1.61
1.32
1.16
1.11
1.06
E-31
-------
Table A2 DAF values as a function of source area for Scenario 2
(x=nationwide distribution, y=uniform in plume, z=nationwide
distribution).
Area (sq. ft.)
5000
8000
10000
45000
50000
100000
150000
220000
500000
1000000
5000000
6000000
DAF
85th
6222.78
3977.72
3215.43
817.66
745.16
424.81
300.12
218.87
110.35
63.45
21.03
19.06
90th
2425.42
1573.32
1286.01
315.06
288.27
160.82
114.71
82.30
40.10
23.75
7.85
7.01
95th
565.61
371.06
298.78
73.48
67.20
38.11
26.82
20.00
10.92
6.22
2.55
2.39
E-32
-------
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 (sq. ft.)
1000
2000
5000
10000
30000
50000
70000
80000
150000
200000
500000
1000000
2000000
3000000
DAF
85th
1.42E+07
9.19E+05
5.54E+04
1.16E+04
1.43E+03
668.45
417.19
350.39
158.76
114.63
40.55
21.13
11.58
8.66
90th
2.09E+05
2.83E+04
2.74E+ 03
644.33
120.42
60.02
37.97
33.16
17.87
12.96
5.54
3.50
2.38
1.98
95th
946.07
211.15
44.23
15.29
4.48
3.10
2.53
2.34
1.74
1.56
1.23
1.15
1.08
1.06
E-3
-------
Table A4 DAF values as a function of source area for Scenario 4 (x=25
ft, y=uniform within half-width of source area, z=nationwide
distribution).
Area (sq. ft.)
1000
2000
5000
10000
30000
50000
70000
80000
150000
200000
500000
1000000
2000000
3000000
DAF
85th
5.93E + 05
1.09E+05
1.64E + 04
4.89E+03
928.51
490.20
323.42
272.85
132.05
97.94
37.99
20.08
11.35
8.49
90th
2.07E + 04
4.92E+03
1.03E + 03
352.49
93.98
49.78
34.79
29.82
16.55
12.29
5.50
3.50
2.40
2.00
95th
348.31
118.11
29.86
13.14
4.73
3.28
2.69
2.47
1.82
1.61
1.29
1.17
1.10
1.07
E-34
-------
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
30000
50000
70000
80000
150000
200000
500000
1000000
2000000
3000000
DAF
85th
4.24E+04
1.52E + 04
4.24E+ 03
1.81 E+03
497.27
293.34
207.77
184.57
98.81
74.63
32.99
18.66
11.14
8.33
90th
3.43E+03
1.33E + 03
437.25
204.29
68.21
40.72
29.89
26.86
15.05
11.55
5.83
3.71
2.53
2.09
95th
181.88
74.79
27.23
13.09
5.10
3.71
2.96
2.73
2.03
1.82
1.40
1.26
1.16
1.13
E-35
-------
Table A6 DAF values as a function of source area for Scenario 6 (x=25
ft, y=source width + 25 ft, z=25 ft).
Area (sq. ft.)
1200
1500
5000
7500
23000
26000
29000
100000
170000
250000
800000
1800000
DAF
85th
44247.79
30759.77
4789.27
2698.33
637.76
544.66
482.63
139.66
76.69
50.40
18.10
10.26
90th
10479.98
7215.01
1273.40
725.69
155.16
135.91
121.43
35.55
21.24
15.04
6.04
3.87
95th
1004.72
744.05
140.81
82.51
21.82
18.84
16.52
5.56
3.94
3.19
1.81
1.48
E-36
-------
APPENDIX F
Dilution Factor Modeling Results
-------
Dilution Factor Model Results: DNAPL Sites
Source size (acres)
Source length (m)
Aquifer thickness (m)
0.5
45
10
201
30
349
9.1
100
636
600
1,559
Site Name State
Army Creek Landfill DE
Atlantic Wood Ind. VA
AtlasTack Corp. MA
Auburn Rd. Landfill NH
Baird & McGuire MA
Bally Groundwater PA
Beacon Hts. Landfill CT
Berks Sand Pit PA
Brodhead Creek PA
Brunswick Naval Air Sta. ME
Cannon Eng.- Bridgewater MA
Central Landfill Rl
Centre County Kepone PA
Chas.-Geo. Reclam. Trust MA
Coakley Landfill NH
Craig Farm Drum PA
Davis Liquid Waste Rl
Delaware City PVC DE
Dorney Road Landfill PA
Dover Mun. Landfill NH
DuPont-Newport DE
Dublin TCE Site PA
Durham Meadows CT
East Mt. Zion PA
Efeabethtown Landfill PA
Gallup's Quarry CT
Greenwood Chemical VA
Groveland Wells MA
Halby Chemical Co. DE
Harvey & Knott Drum DE
Havertown PCP PA
Heleva Landfill PA
Henderson Road PA
Hocomonco Pond MA
Holton Circle NH
Hunterstown Road PA
Industri-plex MA
Kane and Lombard Street MD
Kearsarge Metallurgical Corp. NH
Keefe Environmental Services NH
Kellogg-Deering Well Field CT
Kimberton Site PA
Landfill & Resource Recovery Rl
Lindane Dump PA
Infiltration by
Hyd. Region
Region
10
10
9
9
9
6
9
6
7
9
9
9
6
9
9
6
9
10
6
9
10
6
9
6
6
9
8
9
10
10
8
6
6
9
9
6
9
8
9
9
9
8
9
6
(m/yr)
0.24
0.24
0.22
0.22
0.22
0.15
0.22
0.15
0.20
0.22
0.22
0.22
0.15
0.22
0.22
0.15
0.22
0.24
0.15
0.22
0.24
0.15
0.22
0.15
0.15
0.22
0.15
0.22
0.24
0.24
0.15
0.15
0.15
0.22
0.22
0.15
0.22
0.15
0.22
0.22
0.22
0.15
0.22
0.15
Average GW
Velocity (m/yr)
Seepage
5,563
1,261
3
61
61
3,204
15
10
11,246
230
3
223
61,189
34
113
451
189
223
1,913
289
33
32
612
1,218
56
67
3
612
5
434
24
23
834
1,986
2,809
562
289
681
7
12
946
308
244
82
Darcy
1,947
442
1
21
21
1,121
5
4
3,936
81
1
78
21,416
12
40
158
66
78
670
101
12
11
214
426
20
23
1
214
2
152
9
10
292
695
983
197
101
238
2
4
331
108
86
29
Mixing
zone depth
Site size (acres)
0.5 10
5 21
5 21
10 30
5 23
5 23
5 21
6 27
6 27
5 21
5 22
11 30
5 22
5 21
6 24
5 22
5 21
5 22
5 22
5 21
5 22
6 25
5 24
5 21
5 21
5 23
5 23
10 30
5 21
9 30
5 22
6 24
5 24
5 21
5 21
5 21
5 21
5 22
5 21
8 29
7 23
5 21
5 22
5 22
5 22
30 100
37 67
37 68
46 76
40 72
40 72
37 67
44 76
44 76
37 67
38 69
46 76
38 69
37 67
42 74
39 70
37 68
38 69
38 69
37 67
38 69
42 74
41 73
37 68
37 68
39 71
40 72
46 76
37 68
46 76
37 68
41 74
41 73
37 68
37 68
37 67
37 68
38 69
37 68
46 76
45 76
37 68
37 68
38 69
39 70
600
165
166
174
173
173
165
174
174
165
168
174
168
165
174
171
166
169
169
165
168
174
173
166
165
172
172
174
166
174
167
174
173
166
165
165
166
168
166
174
174
166
167
168
170
0.5
861
197
2
12
12
793
4
4
2,085
41
2
39
15,112
8
21
113
34
36
474
51
7
10
105
303
16
13
2
105
2
69
8
9
208
336
475
141
51
170
3
4
161
78
43
22
Dilution
factor
Site size (acres)
10 30 100
370 214
85 49
1 1
5 4
5 4
341 197
2 2
2 2
895 517
18 11
1 1
17 10
6,489 3,747
3 2
9 6
49 29
15 9
16 10
204 118
22 13
3 2
4 3
45 27
130 76
7 4
6 4
1 1
45 27
1 1
30 18
4 2
4 3
89 52
145 84
204 118
61 35
22 13
73 43
2 1
2 2
69 40
34 20
19 11
10 6
118
27
1
2
2
108
1
1
284
6
1
6
2,053
2
4
16
5
6
65
8
2
2
15
42
3
3
1
15
1
10
2
2
29
46
65
20
8
24
1
1
23
11
7
4
600
49
12
1
2
2
45
1
1
116
3
1
3
839
1
2
7
3
3
27
4
1
1
7
18
2
2
1
7
1
5
1
1
12
20
27
9
4
10
1
1
10
5
3
2
F-1
-------
Dilution Factor Model Results: DNAPL Sites
Source size (acres)
Source length (m)
Aquifer thickness (m)
0.5
45
10
201
30
349
9.1
100
636
600
1,559
Site Name State
Linemaster Switch Corp. CT
Maryland Sand, Gravel & Stone MD
McKin Co. ME
Metal Banks PA
Mottolo Pig Famm NH
MW Manufacturing PA
NCR Corp. Millsboro DE
Norwood PCBS MA
Nyanza Chemicals MA
O'Conner Company ME
Old City of York Landfill PA
Old Southington Landfill CT
Old Springfield Landfill VT
Osborne Landfill PA
Otis Air Natl. Guard MA
Ottati & Goss/Kingston Drums NH
Pease Air Force Base NH
Peterson/Puritan, Inc. Rl
Picillo Farm Rl
Pinette's SalvageYard ME
PSC Resources MA
Re-Solve, Inc. MA
Recticon/Allied Steel PA
Rhinehart Tire Fire VA
Saco Tannery Waste Pits ME
SaundersSupplyCo. VA
Savage Mun. Water Supply NH
Silresim Chemical Corp. MA
Somersworth San. Landfill NH
South Municipal Water Supply NH
Southern MD Wood Treating MD
Stamina Mills, Inc. Rl
Std. Chlorine/Tyboufs Corner LF DE
Strasburg Landfill PA
Sullwan's Ledge MA
Sussex County Landfill #5 DE
Sylvester's NH
Tansitor Electronics VT
Tibbets Road NH
US Defense General Supply VA
US Dover AFB DE
US Naval Air Development PA
US Newport Nav. Educ.&Tm. Ctr. Rl
W.R.Grace & Co./Acton MA
Infiltration by
Hyd. Region
Region (m/yr)
9 0.22
8 0.15
9 0.22
6 0.15
9 0.22
7 0.20
10 0.24
9 0.22
9 0.22
9 0.22
8 0.15
7 0.20
9 0.22
7 0.20
9 0.22
9 0.22
9 0.22
9 0.22
9 0.22
9 0.22
9 0.22
9 0.22
6 0.15
6 0.15
9 0.22
10 0.24
9 0.22
9 0.22
9 0.22
9 0.22
10 0.24
9 0.22
10 0.24
8 0.15
9 0.22
10 0.24
9 0.22
9 0.22
9 0.22
8 0.15
10 0.24
6 0.15
9 0.22
9 0.22
Average GW
Velocity (m/yr)
Seepage Darcy
1,113 389
2 1
890 312
5 2
131 46
21,027 7,359
223 78
389 136
39 14
214 75
779 273
134 47
30 11
1,113 389
312 109
46 16
11 4
56 19
534 187
333 117
45 16
834 292
73 26
1,346 471
56 19
23 10
235 82
26 9
139 49
90 32
2 1
2,809 983
39 14
2,160 756
112 39
198 69
490 171
103 36
23 10
37 13
4 1
56 19
3 1
445 156
Mixing zone depth Dilution factor
Site size (acres) Site size (acres)
0.5 10 30 100 600 0.5 10 30 100 600
5 21 37 68 166
10 30 46 76 174
5 21 37 68 166
8 29 46 76 174
5 22 38 70 170
5 21 37 67 165
5 22 38 69 169
5 22 37 68 167
5 24 41 74 174
5 22 38 69 169
5 21 37 68 166
5 22 38 70 170
6 25 42 74 174
5 21 37 68 166
5 22 38 69 168
5 24 41 73 173
7 23 45 76 174
5 23 40 72 173
5 22 37 68 167
5 22 38 68 167
5 24 41 73 173
5 21 37 68 166
5 22 39 70 171
5 21 37 68 165
5 23 40 72 173
6 25 42 75 174
5 22 38 69 168
6 25 42 75 174
5 22 38 70 170
5 23 39 71 171
12 30 46 76 174
5 21 37 67 165
6 24 41 74 174
5 21 37 67 165
5 22 39 70 171
5 22 38 69 169
5 22 37 68 167
5 22 39 70 171
6 25 42 75 174
5 23 40 72 173
10 30 46 76 174
5 23 39 71 172
11 30 46 76 174
5 22 37 68 167
189 81 47 26 11
21111
152 65 38 21 9
31111
24 10 6 42
3,896 1,673 966 530 217
36 16 10 6 3
68 29 17 10 5
94321
38 16 10 6 3
194 84 49 27 12
27 12 7 42
73221
208 89 52 29 12
54 24 14 8 4
10 4 3 2 1
42111
11 5 3 2 2
92 40 23 13 6
58 25 15 9 4
94321
142 61 36 20 9
20 9 5 3 2
334 144 83 46 19
11 5 3 2 2
63221
42 18 11 6 3
63221
25 11 7 42
17 8 5 3 2
21111
475 204 118 65 27
84221
535 230 133 73 31
21 9 6 4 2
33 14 9 5 3
84 36 21 12 6
19 8 5 32
73221
11 5 3 2 2
21111
16 7 4 32
21111
77 33 20 11 5
F-2
-------
Dilution Factor Model Results: DNAPL Sites
Source size (acres)
Source length (m)
Aquifer thickness (m)
0.5
45
10
201
30
349
9.1
100
636
600
1,559
Site Name
Western Sand & Gravel
Westinghouse Elevator
Winthrop Landfill
Woodlawn County Landfill
State
Rl
PA
ME
MD
Infiltration by
Hyd. Region
Region
9
6
9
8
(m/yr)
0.22
0.15
0.22
0.15
Average GW
Velocity (m/yr)
Seepage
48
562
16
557
Darcy
17
197
6
195
Mixing zone depth
Site size (acres)
0.5 10 30
5 24 40
5 21 37
6 27 44
5 21 37
100
73
68
76
68
600
173
166
174
166
Dilution factor
0.5
10
141
5
140
Site size (acres)
10 30 100
432
61 35 20
2 2 1
60 35 20
600
1
9
1
9
F-3
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
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
(m/y)
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
Average K
(m/y)
63
946
63
946
5,676
157,680
192,370
63,072
125,829
2,759,400
126
946
1,388
1,577
1,577
23,652
31,536
95
158
2,838
315
5,992
315
315
631
1,892
4,100
11,038
16,714
107,222
Hyd. Grad.
(m/m)
3.00E-02
1 .OOE-02
8.00E-02
9.30E-02
2.00E-03
1 .OOE-04
1 .OOE-02
5.00E-03
1 .OOE-03
3.00E-02
2.00E-03
2. OOE-03
3.00E-03
1. OOE-03
5.00E-03
3.00E-03
1 .OOE-03
3.00E-04
1. OOE-03
2.00E-03
5.70E-03
1 .OOE-03
1. OOE-03
2. OOE-03
1 .OOE-02
1 .OOE-03
1 .OOE-03
2. OOE-03
4.00E-03
5.00E-03
Darcy v
(m/y)
2
9
5
88
11
16
1,924
315
126
82,782
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. Thick.
(m)
30
305
23
15
21
3
6
2
5
23
5
3
91
914
24
6
24
9
130
30
46
183
15
3
9
37
3
13
6
7
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
0.5
11
6
5
5
5
5
5
5
5
5
8
5
5
5
5
5
5
14
12
5
10
6
18
8
6
10
6
5
5
5
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
52
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
76
99
70
72
69
68
600
195
211
172
166
173
168
165
165
166
165
170
168
174
190
170
166
166
174
276
171
210
213
180
168
174
201
168
174
168
166
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
3
5
23
124
18
9
1,459
421
169
114,973
2
6
19
9
35
298
133
1
3
26
3
5
1
1
5
3
2
13
35
265
10
2
5
23
89
17
3
419
95
39
114,973
1
2
19
9
35
86
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
53
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
600
1
5
4
12
3
1
55
13
6
15,931
1
1
11
9
6
12
20
1
2
5
1
4
1
1
1
1
1
2
2
13
F-4
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
2.4
2.4
2.5
2.5
2.5
2.5
2.5
2.5
2.6
2.6
2.6
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
6.11
6.11
6.11
Infiltration
(m/y)
0.22
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
Average K
(m/y)
190,793
3,311,280
946
1,261
4,415
6,938
23,337
56,134
1,577
13,876
50,773
126
1,261
3,469
22,075
220,752
220,752
296,438
32
22
284
946
9,776
242,827
2,317,896
631
33,113
1,577
4,415
4,415
Hyd. Grad.
(m/m)
1 .OOE-03
5.00E-03
2. OOE-03
3.00E-03
7.00E-04
3.00E-03
4.00E-03
2. OOE-03
1 .OOE-03
2.80E-02
5.00E-03
2.00E-03
1 .OOE-04
2.00E-02
1 .OOE-03
1 .OOE-03
2. OOE-03
2. OOE-04
1.00E-01
2.80E-02
3.20E-03
8.00E-03
1 .30E-02
2. OOE-03
2.00E-03
3.00E-03
2.00E-06
1.00E-02
5.00E-03
1.00E-02
Darcy v
(m/y)
191
16,556
2
4
3
21
93
112
2
389
254
0.3
0.1
69
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
15
91
15
9
9
21
11
3
3
3
17
12
24
34
24
15
21
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
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
5
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
64
60
42
38
38
49
37
37
47
55
37
37
37
37
38
40
45
40
40
37
37
37
41
70
41
40
39
100
68
67
76
114
98
75
69
69
79
68
68
78
86
68
68
67
68
69
72
77
70
70
68
67
67
75
101
75
72
70
600
167
165
173
268
202
179
170
168
177
166
167
176
183
166
167
165
165
168
174
176
168
168
166
165
165
179
199
179
175
171
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
96
8,118
2
3
3
9
34
41
2
137
90
3
2
291
94
922
336
47
15
4
3
5
63
2,025
19,317
10
2
10
13
24
10
35
6,978
1
3
3
9
34
19
1
137
39
2
1
208
94
660
145
20
14
2
2
2
15
1,596
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
6,377
6
1
6
5
14
100
12
2,206
1
3
2
3
19
7
1
68
13
1
1
66
94
209
46
7
5
1
1
1
5
505
3,500
4
1
4
3
8
600
5
901
1
3
1
2
8
3
1
28
6
1
1
28
52
86
20
3
3
1
1
1
3
207
1,428
2
1
2
2
4
F-5
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
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
6.5
6.5
6.5
6.5
6.5
6.5
6.8
7.11
7.11
7.12
7.12
7.12
7.13
7.13
Infiltration
(m/y)
0.22
0.03
0.03
0.03
0.22
0.22
0.22
0.22
0.22
0.14
0.14
0.14
0.14
0.03
0.03
0.08
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
Average K
(m/y)
81,994
946
3,154
315
1,577
1,892
5,676
14,191
33,113
126
3
1,325
2,208
1,892
31,536
9,776
63
189
315
315
31,536
34,690
2,208
95
2,523
4,100
12,614
116,052
3,154
5,519
Hyd. Grad.
(m/m)
3.00E-03
8.00E-03
6.00E-03
1.70E-02
4.00E-02
2.00E-03
1.00E-03
7.00E-04
1 .OOE-02
4.00E-03
1. OOE-02
5.00E-03
3.30E-02
4.30E-02
1.40E-01
1.20E-02
4.00E-02
2.30E-02
5.00E-03
2.50E-02
5.00E-02
8.00E-03
2.50E-02
6.00E-03
2.00E-02
3.00E-03
4.90E-02
4.00E-03
1.30E-02
1. OOE-02
Darcy v
(m/y)
246
8
19
5
63
4
6
10
331
1
0.03
7
73
81
4,415
117
3
4
2
8
1,577
278
55
0.6
50
12
618
464
41
55
Aq. Thick.
(m)
9
6
3
9
8
6
6
18
23
8
5
21
30
6
3
30
20
61
21
19
6
5
2
4
3
32
6
76
17
-5
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
0.5
5
5
5
5
5
7
6
6
5
11
9
6
5
5
5
5
7
6
8
6
5
5
5
9
5
6
5
5
5
5
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
84
85
87
76
67
68
68
71
69
77
68
68
69
69
600
166
169
167
170
169
171
171
180
166
173
170
182
168
165
165
166
185
199
186
180
165
166
166
169
168
183
166
166
170
168
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (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
5
164
1
1
6
57
98
2,637
165
3
5
2
6
343
46
4
1
5
8
92
230
25
12
30
29
6
8
7
7
1
1
3
101
1
1
4
47
57
1,519
135
2
5
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
15
2
1
2
4
30
230
9
4
600
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
F-6
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
7.13
7.14
7.14
7.14
7.14
7.14
7.14
7.15
7.15
7.16
7.16
7.17
7.17
7.17
7.17
7.17
7.17
7.17
7.17
7.17
7.17
7.17
7.18
7.3
7.3
7.4
7.4
7.4
7.5
7.5
Infiltration
(m/y)
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
0.22
0.22
0.22
0.30
0.30
Average K
(m/y)
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
315
315
946
3,154
3,469
21,760
1,892
946
25,544
189
2,681
3,784
63
11,038
Hyd. Grad.
(m/m)
6.00E-03
4.90E-02
4.00E-03
2.50E-01
1.20E-02
2.00E-03
3.30E-02
6.00E-04
6.80E-03
4.00E-03
3.00E-03
8.00E-03
9.00E-03
3.00E-02
2.20E-02
1.50E-01
1.00E-03
7.00E-03
5.00E-02
1.00E-02
1.70E-02
4.00E-03
5.00E-03
5.00E-03
9.00E-04
1.20E-02
9.00E-03
4.00E-02
7.00E-03
5.00E-04
Darcy v
(m/y)
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
87
9
5
23
2
24
151
0.4
6
Aq. Thick.
(m)
8
5
8
5
18
43
18
37
11
8
9
5
3
11
3
30
12
23
14
5
55
15
1
5
4
61
2
2
518
23
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
0.5
5
5
5
5
5
5
5
8
5
9
5
10
8
10
7
5
15
7
5
5
5
5
5
6
5
9
5
5
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
25
22
38
23
22
143
30
30
37
37
40
37
38
40
37
55
38
45
41
42
40
48
40
39
49
51
38
38
38
37
38
41
39
63
39
37
230
50
100
68
68
72
67
68
72
68
93
68
75
73
73
70
78
70
72
79
86
69
69
69
68
68
72
70
106
70
68
363
85
600
167
166
172
165
168
177
166
200
168
173
173
170
168
176
168
175
177
188
169
169
169
167
166
170
168
221
167
166
618
187
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
72
109
12
921
62
14
273
3
59
2
9
1
1
2
2
16
2
4
38
24
47
68
2
5
14
3
7
25
2
4
10
27
27
5
207
53
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
600
4
4
1
28
8
4
31
1
5
1
1
1
1
1
1
4
1
1
4
2
16
7
1
1
1
1
1
2
1
1
F-7
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
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
(m/y)
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
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
0.22
Average K
(m/y)
63
126
6,623
158
8,830
631
1,892
2,208
3,879
5,676
6,307
7,253
7,884
9,776
13,245
13,876
14,822
15,768
18,922
23,967
29,959
34,374
37,843
44,150
99,654
110,376
662,256
64018030
2,523
22
Hyd. Grad.
(m/m)
7.00E-03
1 .OOE-02
2.00E-02
3.00E-03
5.00E-04
5.00E-03
3.00E-02
9.00E-04
4.00E-03
1 .OOE-03
1 .OOE-03
6.00E-04
3.00E-02
7.00E-04
6.00E-03
2. OOE-03
1 .OOE-03
1 .OOE-03
5.00E-03
2. OOE-03
4.00E-03
6.00E-03
3.00E-03
2. OOE-03
7.00E-04
4.00E-03
3.00E-03
9.00E-04
1.10E-02
4.00E-03
Darcy v
(m/y)
0.4
1
132
0.5
4
3
57
2
16
6
6
4
237
7
79
28
15
16
95
48
120
206
114
88
70
442
1,987
57,616
28
0.09
Aq. Thick.
(m)
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
6
76
6
14
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
0.5
9
9
5
9
6
7
5
9
5
6
6
7
5
6
5
5
5
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
86
89
68
78
69
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
166
180
169
177
184
179
168
171
168
167
168
168
168
166
165
165
170
179
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
1
3
102
1
5
4
30
3
10
5
5
4
75
5
41
16
9
10
48
25
61
103
58
45
36
218
976
28,244
16
1
10
1
2
102
1
5
2
30
2
4
2
5
4
18
3
23
16
9
10
18
25
53
103
25
39
12
218
280
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
600
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
F-8
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
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
Infiltration
(m/y)
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
0.08
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
Average K
(m/y)
158
315
126
126
631
4,100
631
2,208
5,046
11,038
19,237
19,237
27,752
27,752
33,113
60,864
126
284
315
8,830
25
32
126
126
158
284
315
315
631
2,208
Hyd. Grad.
(m/m)
1.20E-02
6.00E-03
5.00E-02
2.00E-02
1.50E-01
1.00E-02
1 .OOE-02
2.00E-02
3.00E-03
7.50E-02
8.00E-03
1 .30E-02
2.00E-03
2.00E-03
4.00E-04
3.00E-03
3.00E-02
1 .OOE-02
5.10E-01
4.00E-03
9.50E-03
1.70E-02
3.00E-03
2.50E-02
6.00E-04
1 .OOE-02
4.00E-03
1 .OOE-02
5.00E-03
1 .OOE-05
Darcy v
(m/y)
2
2
6
3
95
41
6
44
15
828
154
250
56
56
13
183
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
5
7
4
12
3
8
6
11
1
8
Source Length (m)
Source Area (acres)
0.5
45
10
201
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
5
6
8
5
5
9
11
9
8
8
8
10
8
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
38
45
100
70
72
72
75
68
70
77
70
75
68
69
68
71
71
79
68
79
76
68
71
72
74
71
79
70
75
73
78
68
75
600
168
170
170
173
167
169
176
168
176
166
168
167
172
172
186
167
192
174
167
172
170
172
169
177
168
173
171
176
166
173
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
2
2
4
3
22
22
4
13
7
185
55
89
21
21
7
65
7
3
81
19
1
1
1
3
1
3
2
3
1
1
10
1
1
2
1
6
7
2
4
4
42
32
45
21
21
7
65
7
2
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
5
53
7
1
14
10
1
1
1
1
1
1
1
1
1
1
100
1
1
1
1
2
3
1
2
2
14
11
15
8
8
3
30
7
1
8
6
1
1
1
1
1
1
1
1
1
1 -
600
1
1
1
1
2
2
1
1
1
6
5
7
4
4
2
13
4
1
4
3
1
1
1
1
1
1
1
1
1
1
F-9
-------
Dilution Factors (DFs) for 208 Sites in the Hydrogeologic Database (HGDB) - National Average
Hydrogeologic
Setting
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
(m/y)
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
0.08
0.08
Average K
(m/y)
2,208
3,469
4,415
4,415
19,552
607,068
315
631
4,415
631
7,569
12,614
32
284
315
315
946
1,261
1,261
1,577
2,523
3,154
8,168
13,876
176,602
309,053
5,361
7,884
Hyd. Grad.
(m/m)
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 .OOE-02
6.00E-03
5.00E-03
5.00E-03
3.00E-03
5.00E-02
1.00E-03
2.00E-04
2.00E-03
1.70E-02
2.30E-02
2.00E-03
1.50E-01
3.30E-03
2.00E-03
1 .90E-02
5.00E-04
1.00E-03
2.00E-02
Darcy v
(m/y)
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
5
473
27
28
3,355
155
5
158
Aq. Thick.
(m)
8
3
55
9
21
15
3
0
20
6
46
5
15
30
2
24
2
11
3
5
2
6
6
61
4
43
6
3
Source Length (m)
Source Area (acres)
0.5
45
10
201
Calculated Mixing Zone Depth (d)
Source Area (acres)
0.5
5
6
5
5
7
5
8
5
6
7
5
5
20
17
5
25
6
9
5
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
84
67
70
68
81
73
71
70
83
98
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
169
180
195
167
189
167
176
168
169
167
166
171
180
165
168
171
166
30
349
100
636
600
1,559
Dilution Factor (DF)
Source Area (acres)
0.5
10
3
10
23
4
424
1
1
5
4
18
22
1
2
3
2
1
3
6
13
2
166
11
12
1,045
56
9
141
10
4
1
10
10
3
303
1
1
4
2
18
6
1
1
1
1
1
1
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
28
3
12
136
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
1
1
1
1
1
1
1
1
1
7
1
5
31
15
1
5
F-10
-------
Hydrogeologic Settings for HGDB Sites
Region
Western Mountain Ranges
Alluvial Basins
Columbia Lava Plateau
Setting
Reference Number
Mountain Slopes Facing East 1.1
Mountain Flanks Facing East 1.3
Mountain Flanks Facing West 1.4
Wide Alluvial Valleys Facing East 1.6
Wide Alluvial Valleys Facing West 1.7
Alluvial Mountain Valleys Facing West 1.8
Alluvial Mountain Valleys Facing East 1.9
Coastal Beaches 1.11
Mountain Slopes 2.1
Alternating 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
Playa Lakes 2.11
Continental Deposits 2.12
Lava Flows: Hydraulically Connected 3.3
Alluvial Fans 3.5
River Alluvium 3.7
Colorado Plateau and Wyoming Basin
Resistant Ridges
Consolidated Sedimentary Rocks
Alluvium and Dune Sand
River Alluvium
4.1
4.2
4.3
4.4
High Plains
Non-Glaciated Central Region
River Alluvium with Overbank Deposits 5.2
River Alluvium without Overbank Deposits 5.3
Playa Lakes 5.7
Ogalalla 5.8
Triassic Basins
Mountain Slopes
Mountain Flanks
6.2
6.3
6.4
F-11
-------
Hydrogeologic Settings for HGDB Sites
Region
Setting
Reference Number
Non-Glaciated Central Region (cont.)
Alternating Beds of Sandstone, Limestone,
or Shale Under Thin Soil 6.5
Alternating Beds of Sandstone, Limestone,
or Shale Under Deep Regolith 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
Unconsolidated and Semi-Consolidated
Aquifers 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
River Alluvium
8.1
8.6
Northeast and Superior Uplands
Glacial Till Over Crystalline Bedrock 9.1
Glacial Lakes/Glacial Marine Deposits 9.2
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
F-12
-------
Hydrogeologic Settings for HGDB Sites
Region
Setting
Reference Number
Northeast and Superior Uplands (cont.)
River Alluvium with Overbank Deposits 9.12
River Alluvium without Overbank Deposits 9.13
Till 9.14
Atlantic 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 10.5
Southeast Coastal Plain
Solution Limestone and Shallow
Surfacial Aquifers
Swamp
Beaches and Bars
Coastal Deposits
1.11
11.2
11.3
11.4
Hawaii
Volcanic Uplands
Coastal Beaches
12.1
12.4
Alaska
Coastal Lowland Deposits 13.2
Glacial and Glacio-lacustrine Deposits of
the Interior Uplands 13.4
F-13
-------
APPENDIX G
Background Discussion for Soil-Plant-Human
Exposure Pathway
-------
APPENDIX G
Background Discussion for Soil-Plant-Human Exposure Pathway
Introduction
The U.S. Environmental 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) for 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 the 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 Calamari, 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 this 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 that use a fugacity-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 that 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 site-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
-------
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 (mg/ kg) =
(G-l)
Parameter/Definition (units) Default
Cpiant^'acceptable plant concentration (mg/kg DW) see Section G.2
Br/plant-soil bioconcentration factor (mg contaminant/kg chemical- and plant-specific
plant tissue DW)(mg contaminant/kg soil)-1 (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-l 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-l are either derived from other equations or identified from empirical
studies in the literature. Specifically, the derivation and data sources for Cpiant and Br are discussed
below.
G.2 Acceptable Concentration in Plant Tissue (Cpiant)
The acceptable contaminant concentration in plant tissues (Cpiant) in mg/kg DW for fruits and
vegetables is backcalculated using the following equation:
Acceptable Plant Concentration for Fruits and Vegetables (Cpiant)
_ IxBW (G-2)
PI,,,, p x CR
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
-------
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
Escarole
Green onions
Lettuce
Spinach green
Average for vegetables
a Plum/prune was omitted
Source: Baes et al. (1984).
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
from the average as an outlier.
Fruits
Apple
Bushberry
Cherry
Grape
Peach
Pear
Strawberry
Plum/prune
Average for fruits3
0.159
0.151
0.170
0.181
0.131
0.173
0.101
0.540
0.15
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
I =
TR x AT x 365d/yr
ED x EF x CSF
(G-3)
oral
Parameter/Definition (units)
TR/target risk level (unitless)
AT/averaging time (years)
ED/exposure duration (years)
EF/exposure frequency (d/yr)
CSForal/oral cancer slope factor (mg/kg-d)"1
Default
10-6
70
30
350
chemical-specific (see Part 2, Table 1)
G-3
-------
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
I =
HQ x RfD x AT x 365 d/yr
ED x EF
(G-4)
Parameter/Definition (units)
HQ/target hazard quotient (unitless)
AT/averaging time (years)
ED/exposure duration (years)
EF/exposure frequency (d/yr)
RfD/oral reference dose (mg/kg-d)
Default
1
30
30
350
chemical-specific (see Part 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 that 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 fresh-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, there is very little translocation of contaminants to the edible
parts of the plant. EPA recognizes that, 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
G-4
-------
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, bioconcentration factors that 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
To demonstrate how the methods described in this appendix may be used to calculate an SSL for 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 RfD for cadmium ingested in food of 1.0 x 1O3 mg/kg (the RfD is 5.0 x 1O4
in water), Equation G-4 may be solved for acceptable daily intake (I) of cadmium from a dietary
source:
HQ x RfD x AT x 365d/yr
ED x EF
1 x l.OxlO'3 mg/kg-d x 30 yr x 365d/yr
30yrs x 350 d/yr
I = l.OxlO"3 mg/kg-d
The acceptable daily intake (I) is used in Equation G-2 to estimate the acceptable contaminant
concentration in plant tissue (Cpiant). However, Equation G-2 is designed to solve for the acceptable
plant concentration (Cpiant) 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 bioconcentration 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 Level (me/kg) =
F x £(CR x Br)
G-5
-------
Table G-2. Summary Table of Empirical Bioconcentration Factors for Metals
(in mg contaminant per kg plant DW / mg contaminant per kg soil)
Bioconcentration
factors (Br)
Study
observations
pH Range
Min
Max
Geometric
Mean Br
Arsenic
grains and cereals
potatoes
leafy vegetables
legumes
root vegetables
garden fruits
sweet corn
1
8
7
7
7
5
3
7.5
5.5-7.5
5.5-7.5
NR-7.5
NR-7.5
NR-7.5
NR
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 corn
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 fruits
sweet corn
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
Nickel
grains and cereals
potatoes
leafy vegetables
legumes
root vegetables
garden fruits
sweet corn
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
Selenium
grains and cereals
potatoes
leafy vegetables
legumes
root vegetables
garden fruits
sweet corn
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
0.008
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
G-6
-------
Table G-2. (continued)
Bioconcentration
factors (Br)
Study
observations pH Range
Min
Max
Geometric
Mean Br
Zinc
grains and cereals
potatoes
leafy vegetables
legumes
root vegetables
garden fruits
sweet corn
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
0.016
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 = No data
The input parameters in Equation G-5 correspond to input parameters in Equations G-l and G-2,
with a contaminated fraction (F) of 0.4, and consumption rates (CRag and CRbg) and
bioconcentration factors (Brag 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, CRag, and CRbg
results in:
I x BW
Screening Level =
0.4 xE(CRag x Brag) + (CRbg x Brbg)
l.OxlO"3 mg/kg-d x 70kg
Screening Level =
0.4 x £(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 (Brag)
and belowground vegetables (Brbg), 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 that 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
lexicological 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
-------
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 toxicological 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 SSLs for Soil
Ingestion and LOEC Values for Phytotoxicity (all values in mg/kg)
Generic plant SSL
Soil ingestion SSL
Migration to ground
water SSLa
Phytotoxicity LOEC
Arsenic
0.4
0.4
29(1)
10
Cadmium
24
78
8(0.4)
3
Mercury
270
23
2(0.1)
0.3
Nickel
5400
1600
130(7)
30
Selenium
2400
390
5(0.3)
1
Zinc
10000
23000
12000(620)
50
a Values based on DAF of 20 (DAF of 1).
The comparison of the generic SSLs for the plant pathway with SSLs for soil ingestion and migration
to ground water suggests that 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.
Phytotoxicity - 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 that 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
-------
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/accumulation, 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, up take-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 Kow with plant bioconcentration; it is a mechanistic model that 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 that the PLANT models may be a scientifically
defensible alternative to the uptake-response slopes generated by log Kow regressions.
G.9 Conclusions and Recommendations
The comparison of generic plant SSLs with generic SSLs for soil ingestion 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 that would be harmful to humans. Although this pathway
G-9
-------
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
-------
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 I 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/TM-85/R1.
Prepared for the U.S. Department of Energy by the Environmental Sciences Division of the
Oak Ridge National Laboratory.
G-ll
-------
APPENDIX H
Evaluation of the Effect on the Draft SSLs of the
Johnson and Ettinger Model (EQ, 1994a)
-------
ENVIRONMENTAL QUALITY MANAGEMENT, INC.
MEMORANDUM
TO: JanineDinan DATE: October 7, 1994
SUBJECT: Evaluation of the Effect on the Draft SSLs FROM: Craig S. Mann
of the Johnson and Ettinger (1991) Model
for the Intrusion of Contaminant Vapors
Into Buildings
FILE: 5099-3 cc:
Under U.S. Environmental Protection Agency (EPA) Contract No. 68-D3-0035, Task
order No. 0-25, Environmental Quality Management, Inc. (EQ) was directed to evaluate the effect
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:
(vr.) = v
kv APr LD
R;
where * = Nondimensional variables
e. = Volume fraction of phase i, unitless
Q = Concentration of contaminant in phase i, g/cm3
t = Time, s
Lp = Convection path length, cm
LD = Diffusion path length, cm
P = Pressure in vapor-phase, g/cm-s2
V = Del operator, I/cm
Cv = Contaminant concentration in vapor phase, g/cm3
Deff = Effective diffusion coefficient, cm2/s
|l = Vapor viscosity, g/cm-s
H-2
-------
k^, = Soil permeability to vapor flow, cm2
A Pr = Reference indoor-outdoor pressure differential, g/cm-s2
Rj = Formation rate of contaminant in phase i, g/cm3-s
and,
r * =C/C
M M' ^r
V* =LDV
P* = P/APr
t* =t(kvAP/LDLPJu)
Ri* =RILDLPJu/CrkrAPr
where Cr, 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 unity.
The mass balance solution includes the following assumptions:
1. The soil column is isotropic 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).
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.
H-3
-------
10. The building contains no other contaminant sources or sinks, and the air volume is
well mixed.
Therefore,
o r = F
Vbuildmg ^building ^
(2)
where Qbuildmg, Cbmldmg, and E represent the volumetric flow rate or ventilation rate of the building
(cm3/s), contaminant concentration within the building (g/cm3), and rate of contaminant entry (g/s),
respectively.
Also,
a = Q /c (3)
^-buildma source V '
where Csource is the vapor-phase contaminant concentration within the soil at the source, and a
represents the attenuation coefficient. Csource is written as:
HCsp
s Kb
(4)
where H = Henry's law constant, unitless
Cs = Soil bulk concentration, g/g
pb = Soil dry bulk density, g/cm3
0W = Soil water-filled porosity, unitless
Kd = Soil-water partion coefficient, cm3/g
0a = Soil air-filled porosity, unitless.
The authors derive a solution for a 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:
a =
Deff A
V building -
x exp
- soil crack
kcrack ^
crack
Dcrack A
exp
eff
\crack
+
crack
D
^ building
D AB
Qsoii LT
exp
\crack
crack
crack
-1
(5)
H4
-------
where De
LT
v Crack
^building
= Effective diffusion coefficient, cm2/s
= Area of basement, cm2
= Source-building separation, cm
= Volumetric flow rate of soil gas into the building, cnrVs
= Building foundation thickness, cm
= Effective diffusion coefficient through crack, cm2/s (DCrack = Deff)
= Area of crack, cm2
= Building ventilation rate, cm3/s.
For quasi-steady-state conditions the long-term average attenuation coefficient is:
where
and,
Pb
AB
^-building
L °
J^T
AHC A
Vbuilding ^source T V C
: Soil dry bulk density, g/cm3
: Average contaminant level in soil, g/g
: Thickness of depth over which contaminant is distributed, cm
: Area of basement, cm2
: Building ventilation rate, cm3/s
: Vapor-phase soil concentration at source, g/cm3
: Exposure averaging period, s
: Source-building separation at t=0, cm
(6)
QSOI1
1 - exp - -^
f -f^CT,
veff
Vsoil ^ crack
Dcrack Acrack )
tf Pb CR
+ 1 (7)
(8)
H-5
-------
The time required to deplete a finite source (TD) of depth AHc is given as:
q + fl- fi2
If the exposure period (i) is greater than TD, the average emission rate into the building is
given as a simple mass balance:
= pbCRAHcAB/i (10)
and the average building concentration (Cbuildmg) is:
^building ~~ -k 'Vbmldmg (*•*•)
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 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 10"8 cm2 (1 darcy) which is
representative of silty to fine sand. Soil column-building pressure differential was set equal to 1
pascal (10g/cm-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 (1991).
In the analysis, the values for Cbuildmg (kg/m3) were calculated for the 42 chemicals in the
TBD for which human health benchmarks are available. Please note that the values of Csource and
Cbuildmg 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 106. The inverse of the value of Cbuildmg (m3/kg) was used as the
indoor volatilization factor (VFmdoor) and substituted into Equations 2-4 or 2-5 of the TBD as
appropriate to calculate the resulting carcinogenic and noncarcinogenic inhalation SSLs. SSLs were
calculated for both steadystate conditions (infinite source depth) and quasi-steady-state 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 (Cr) 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.
H-6
-------
TABLE 1.
SUMMARY OF INDOOR AND OUTDOOR INHALATION SSLs FOR
VOLATILE CONTAMINANTS
Chemical
Aldrin
Benzene
Bis(2-chloroethyl)ether
Bromoform
Carbon disulfide
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
DDT
1,2-Dichlorobenzene
1,4-Dichlorobenzene
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
1,2-Dichloropropane
1,3-Dichloropropene
Dieldrin
Ethylbenzene
Heptachlor
Heptachlor epoxide
Hexachloro-1 ,3-butadiene
Hexachlorobenzene
HCH-alpha(alpha-BHC)
HCH-beta(beta-BHC)
Hexachlorocyclopentadiene
Hexachloroethane
Methyl bromide
Methylene chloride
Nitrobenzene
Styrene
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
Indoor SSL,
infinite source
(mg/kg)
0.4
0.002
0.02
0.8
0.03
0.0007
51
0.7
0.001
5a
26
102
4
0.002
0.0001
0.06
0.0007
3
21
0.04
1
0.03
0.3
0.5
7a
0.06
0.6
0.01
0.04
9
185
0.007
0.05
6
2
6
5
0.009
0.01
64
5
0.00002
Indoor SSL,
finite source
(mg/kg)
0.4
0.02
0.05
0.9
07
0.01
53
2
0.007
5a
65
235 a
35
0.007
0.003
0.3
0.004
4
69
0.04
1
0.05
0.6
0.6
7a
0.07
0.6
0.3
0.3
25
472
0.02
0.3
28
2
9
69
0.02
0.09
94
14
0.002
Outdoor SSL,
infinite source
(mg/kg)
0.5
0.5
0.3
43
11
0.2
54
87
0.2
5a
297a
235a
939
0.3
0.04
10
0.1
2
257 a
0.3
1
1
1
0.9
7a
2
45
3
7
100
1439a
0.4
11
521 a
2a
214
980 a
1
3
190
351
0.01
1 = SSL based on C,.
H-7
-------
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 (DA) is
given here so as not to be confused with the effective diffusion coefficient (Deff) from Johnson and
Ettinger(1991):
DA = [(0;°/3 Da H + 0f Dw)/0*]/(p6 Kd + 0W + QJD (12)
where DA = Apparent diffusion coefficient, cm2/s
0a = Air-filled soil porosity, unitless
Da = Diffusivity in air, cm2/s
H = Henry's law constant, unitless
0W = Water-filled soil porosity, unitless
Dw = Diffusivity in water, cm2/s
0t = Total soil porosity, unitless
pb = Soil dry bulk density, g/cm3
Kd = Soil-water partition coefficient, cm3/g.
With all nonchemical-specific 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-specific variables, a sensitivity analysis was performed for soil
permeability to vapor flow (kj, soil-building pressure differential (A P), depth of contamination
(AHC), source-building separation at t = 0 (LT°), crack-to-total area ratio (T|), and building
ventilation rate (Qbmldmg).
H-8
-------
1.1
1. OOE-01
g 1
ng C
.00E-03
1.00E-05
1.00E-09
1.00E-0?
1.00E-<»
Apparent DlffuiIon Coefficient (cm'/i)
1.00E-03
Figure 1
Building Concentration Versus Apparent Diffusion Coefficient
1.OOE-01
H-9
-------
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 Csat is not included in limiting the value of the SSL. Depth of contamination (AHc) has the
greatest effect for contaminants with higher apparent diffusion coefficients (e.g., benzene,
chloroform, vinyl chloride, etc.), in that as AHc increases, the time required for source depletion
(TD) 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 ky or AP produces the greatest results. This is to be expected as values
of TD for these contaminants exceed the exposure duration. Table 2 also indicates that an order of
magnitude change in values Of LT° and r| produce same order of magnitude results. It must be
remembered, however, that in the case of LT°, the model assumes isotropic soil conditions from the
point of building entry to the bottom of contamination. As LT° increases, a decreases until
diffusion not convection limits the rate of contaminant vapor transport. The effect of changes in the
value of r| 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 close or directly beneath the structure.
2. The soil between the structure and the source must be very permeable (k^,, > 10~8
cm2).
3. The structure must be underpressurized.
4. The air within the structure must be well mixed (i.e., little or no soil-air boundary
layer resistance).
5. The combination of diffusion coefficient through the cracks, area of the cracks, and
building 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 building ventilation rate, source-building separation,
soil permeability to vapor flow, and source depth. If the source of contamination is relatively deep
and close to the building, and if the soil between the source and the building is very permeable,
building concentrations of contaminants with relatively high apparent diffusion coefficients will
increase dramatically.
H-10
-------
TABLE 2.
MODEL SENSITIVITY TO NONCHEMICAL SPECIFIC VARIABLES
Ratio of Variable-to-Test Condition SSL
Chemical
DDT
Dieldrin
HCH-beta(beta-BHC)
Chlordane
Heptachlor epoxide
Aldrin
HCH-alpha(alpha-BHC)
Toxaphene
2,4,6-Trichlorophenol
Hexachlorobenzene
Heptachlor
Hexachloroethane
Nitrobenzene
Bis(2-chloroethyl)ether
Hexachlorocyclo-pentadiene
1 ,2,4-Trichlorobenzene
Bromoform
Hexachloro-1 ,3-butadiene
Styrene
1 ,1 ,2,2-Tetrachloroethane
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
1 ,1 ,2-Trichloroethane
Chlorobenzene
Vinyl acetate
1 ,2-Dichloropropane
Ethylbenzene
1 ,2-Dichbropropane
Toluene
Tetrachloroethylene
1 ,3-Dichloropropene
Chloroform
1,1-Dichloroethane
Benzene
Trichloroethylene
Methylene chloride
1 ,1 ,1-Trichioroethane
Carbon tetrachloride
Carbon disulfide
1,1-Dichloroethylene
Methyl bromide
Vinyl chloride
Apparent
diffusion
coefficient,
DA
(cm /s)
1.16E-09
1 .59E-09
3.54E-09
5.63E-09
5 78E-09
1 .03E-08
2.81 E-08
3.69E-08
1.81E-07
284E-07
3.52E-07
1.80E-06
3.92E-06
5.94E-06
1.06E-05
1.89E-05
2.32E-05
6.97E-05
9.50E-05
9.89E-05
1.34E-04
1.38E-04
3.04E-04
5.18E-04
7.79E-04
8.57E-04
8.64E-04
1.24E-03
1.25E-03
1 .34E-03
1.44E-03
1.91E-03
2.08E-03
2.12E-03
2.44E-03
2.45E-03
3.79E-03
3.82E-03
5.67E-03
7.09E-03
8.56E-03
2.40E-02
Test
condition
SSL,
(mg/kg)
5a
4
T
53
1
0.4
0.6
2
94
0.6
0.04
0.6
25
0.05
0.07
9
0.9
0.05
472
0.02
65
235°
0.02
2
14
0.007
69
0.3
28
0.3
0.004
0.007
35
0.02
0.09
0.3
69
0.01
0.7
0.003
0.3
0.002
Soil vapor
permeability,
kv x 10
1
0.1
0.8
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.3
0.1
0.1
0.2
0.1
0.2
0.1
0.1
0.2
0.2
0.2
0.4
0.8
1
0.9
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Soil-bldg.
pressure
differential,
APx 10
1
0.1
0.8
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.3
0.1
0.1
0.2
0.1
0.2
0.1
0.1
02
0.2
0.2
0.4
0.8
1
0.9
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Depth to
source
lower
boundary,
AHcx 10
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
0.8
0.6
0.7
0.5
0.4
0.5
0.4
0.4
0.3
0.4
0.2
0.2
0.2
0.1
0.1
0.1
Source-
bldg.
separation
at t=0,
LT°x 10
1
1.2
1
1.3
1.3
1.2
1.2
1.1
1.1
1.1
1.3
2
1
1
1.2
1.1
1.2
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
Inverse of
crack-to-
total area
ratio,
1/11 x 10
1
1.5
1
1.3
1.5
1.5
1.5
1.1
1.5
1.5
1.5
1.4
1.5
1.5
1.5
1.5
1.5
1.5
1 5
1 5
1.5
1
1.5
1.5
1.5
1.5
1.2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Bldg.
ventilation
rate,
Qbu,,d,ngXlO
1.0
3.4
1.0
1.3
8.4
10
10
1.1
10
3.2
10
10
10
10
10
10
10
10
3.0
10
4.5
1.0
10
10
10
10
3.7
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
1 = SSL based C,.
H-ll
-------
It should be noted, however, that soil permeability, kv, is 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 soil type. Overall 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. Technol., 25(8): 1445-1452.
Jury, W. A., W. J. Farmer, and W. F. Spencer. 1983. Behavior Assessment Model for Trace
Organics in Soil, 1, Model description. J. Environ. Qual., 12:558-564.
Attachment
H-12
-------
ATTACHMENT
DETAILED MODEL EVALUATION
H-13
-------
COMPARISON OF INDOOR AND OUTDOOR INHALATION SSLs FOR VOLATILE CONTAMINANTS
Chemical
Aldrin
Benzene
Bis(2-chloroethyl) ether
Bromoform
Carbon disulfide
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
DDT
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
1,1-Dichloroethane
1 ,2-Dichloroethane
1,1-Dichloroethylene
1 ,2-Dichloropropane
1 ,3-Dichloropropene
Dieldrin
Ethyl benzene
Heptachlor
Heptachlor epoxide
Hexachloro-1 ,3-butadiene
Hexachlorobenzene
HCH-alpha(alpha-BHC)
HCH-beta(beta-BHC)
Hexachlorocyclopentadiene
Hexachloroethane
Methyl bromide
Methylene chloride
Nitrobenzene
Styrene
1,1,2,2-Tetrachloroethane
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1 ,1 ,1 -Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
NA= not applicable
" = SSL based on C,,,
CAS No.
309-00-2
71-43-2
111-44-4
75-25-2
75-15-0
56-23-5
57-74-9
108-90-7
67-66-03
50-29-3
95-50-1
106-46-7
75-34-3
107-06-2
75-35 -4
78-87-5
542-75-6
60-57-1
100-41-4
76-44-8
1024-57-3
87-68-3
118-74-1
319-84-6
319-85-7
77-47-4
67-72-1
74-83-9
75-09-2
98-95-3
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
88-06-2
108-05-4
75-01-4
Soil bulk
density,
Pu
(g/cm3)
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
Soil
moisture,
w
(g/g)
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Soil
moisutre,
e>,
(cm3/cm3)
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
0.15
Soil total
porosity,
1
(unitless)
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
0.434
Soil air-
filled
porosity,
e,
(unitless)
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
0.284
Soil
water-
filled
porosity,
&,
(unitless)
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
0.150
Diffusivity
in air,
D,
(cm2/s)
1.32E-02
8.70E 02
6.92E-02
1.49E-02
1.04E-01
7.80E-02
1.18E-02
7.30E-02
1.04E-01
1.37E-02
6.90E-02
6.90E-02
7.42E-02
1.04E-01
9.00E-02
7.82E-02
6.26E-02
1.25E-02
7.50E-02
1.12E-02
1.22E-02
5.61 E-02
5.42E-02
1.76E-02
1.76E-02
1.61 E-02
2.49E-03
7.28E-02
1.01E-01
7.60E-02
7.10E-02
7.10E-02
7.20E-02
8.70E-02
1.16E-02
3.00E-02
7.80E-02
7.80E-02
7.90E-02
3.14E-02
8.50E-02
1.06E-01
Diffusity in
water,
a
(cm2/s)
4.86E-06
9.80E-066
7.53E-06
1.03E-05
1.00E-05
8.80E-06
4.37E-06
8.70E-06
1.00E-05
4.95E-06
7.90E-06
7.90E-06
1.04E-05
9.90E-06
1.04E-05
8.73E-06
1.00E-05
4.74E4-6
7.80E-06
5.69E-06
4.68E-06
6.16E-06
5.91 E-06
5.57E-06
5.57E-06
7.21 E-06
6.80E-06
1.21E-05
1.17E-05
8.60E-06
8.00E-06
7.90E-06
8.20E-06
8.60E-06
4.34E-06
8.23E-06
8.80E-06
8.80E-06
9.10E-06
6.36E-06
9.20E-06
1.23E-05
Effective
diffusion
coeffi-
cient,
D.
(cm2/s)
1.05E-03
6.95E-03
5.53E-03
1.19E-03
8.31E-03
6.23E-03
9.43E-04
5.83E-03
8.31E-03
1.09E-03
5.51E-03
5.51E-03
5.93E-03
8.31E-03
7.19E-03
6.25E-03
5.00E-03
9.99E-04
5.99E-03
8.95E-04
9.75E-04
4.48E-03
4.33E-03
1.41E-03
1.41E-03
1 .29E-03
1 .99E-04
5.82E-03
8.07E-03
6.07E-03
5.67E-03
5.67E-03
5.75E-03
6.95E-03
9.27E-04
2.40E-03
6.23E-03
6.23E-03
6.31E-03
2.51 E-03
6.79E-03
8.47E-03
Soil vapor
permea-
bility,
K,
(cm3)
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1 .OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1. OOE-08
1 OOE-08
Soil-bldg.
pressure
differen-
tial
AP
(g/cm-s2)
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
10
10
10
10
10
10
10
10
10
10
Diffusion
path
length,
La
(cm)
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Convec-
tion path
length,
L,
(cm)
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Vapor
viscosity,
V-
(g/cm-s)
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.00E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
1.80E-04
Peclet
number,
Pe
(unitless)
0.53
0.08
0.10
0.47
0.07
0.09
0.59
0.10
0.07
0.51
0.10
0.10
0.09
0.07
0.08
0.09
0.11
0.56
0.09
0.62
0.57
0.12
0.13
0.39
0.39
0.43
2.79
0.10
0.07
0.09
0.10
0.10
0.10
0.08
0.60
0.23
0.09
0.09
0.09
0.22
0.08
0.07
Henrys
law
constant,
H
(unitless)
4.20E-03
2.20E-01
8.80E-04
2.50E-02
5.20E-01
1.20E+00
2.70E-03
1.80E-01
1.60E-01
2.20E43
8.60E-02
1.20E-01
2.40E41
5.20E-02
1.00E+00
1.20E-01
1.20E-01
1.10E-04
3.20E-01
2.40E-02
3.40E-04
9.80E-01
220E-02
2.80E-04
1.40E-05
7.10E-01
1.50E-01
5.80E-01
9.70E-02
8.40E-04
1 .40E-01
1 .50E-02
7.10E-01
2.50E-01
1.40E-04
1.10E-01
7.60E-01
4.10E-02
4.30E-01
1.70E-04
2.30E-02
3.50E+00
Organic
carbon
partition
coeffi-
cient,
K.c
(cm3/g)
4.84E+04
5.70E+01
7.60E+01
1.26E+02
5.20E+01
1.64E+02
5.13E+04
2.04E+02
5.60E+01
2.37E+05
3.76E+02
5.16E+02
5.20E+01
3.80E+01
6.50E+01
4.70E+01
2.60E+01
1.09E+04
2.21E+02
6.81E+03
7.24E+03
6.99E+03
3.75E+04
1.76E+03
2.28E+03
9.59E+03
1.83E+03
9.00E+00
1.60E+01
1.31E+02
9.12E+02
7.90E+01
3.00E+02
1.31E+02
5.01E+02
1.54E+03
9.90E+01
7.60E+01
9.40E+01
2.83E+02
5.00E+00
1.10E+01
H-14
-------
COMPARISON OF INDOOR AND OUTDOOR INHALATION SSLs FOR VOLATILE CONTAMINANTS
Chemical
Aldrin
Benzene
Bis(2-chloroethyl)ether
Bromoform
Carbon disulfide
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
DDT
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
1 ,2-Dichloropropane
1 ,3-Dichloropropene
Dieldrin
Ethylbenzene
Heptachlor
Heptachlor epoxide
Hexachloro-1 ,3-butadiene
Hexachlorobenzene
HCH-alpha(alpha-BHC)
HCH-beta(beta-BHC)
Hexachlorocyclopentadiene
Hexachloroethane
Methyl bromide
Methylene chloride
Nitrobenzene
Styrene
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1 ,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
NA= not applicable
" = SSL based on C,,,
Soil
organic
carbon
fraction,
f.c
(unitless)
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.005
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
0.006
Soil-water
partition
coefficient
K,
(cnf/g)
2.90E+02
3.42E-01
4.56E-01
7.56E-01
3.12E-01
9.84E-01
3.08E+02
1.22E+00
3.36E-01
1.42E+03
2.26E+00
3.10E+00
3.12E-01
2.28E-01
3.90E-01
2.82E-01
1.56E-01
6.54E+01
1.33E+00
4.09E+01
4.34E+01
4.19E+01
2.25E+02
1.06E+01
1.37E+01
5.75E+01
1.10E+01
5.40E-02
9.60E-02
7.86E-01
5.47E+00
4.74E-01
1.80E+00
7.86E-01
3.01E+00
9.24E+00
5.94E-01
4.56E-01
5.64E-01
1.70E+00
3.00E-02
6.60E-02
Initial soil
cone.,
C,
(mg/kg)
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00'
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
1.00E+00
Source
vapor
cone.,
Csourc,
(g/cm3)
1.45E-05
4.55E-01
1.58E-03
2.90E-02
1.02E+00
9.15E-01
8.77E-06
1.33E-01
3.43E-01
1.55E-06
3.63E-02
3.73E-02
5.25E-01
1.54E-01
1.47E+00
2.97E-01
4.31E-01
1.68E-06
2.15E-01
5.86E-04
7.81E-06
2.32E-02
9.77E-05
2.63E-05
1.02E-06
1.23E-02
1.35E-02
2.20E+00
4.53E-01
9.48E-04
2.50E-02
2.60E-02
3.49E-01
2.68E-01
4.51E-05
1.18E-02
9.07E-01
7.27E-02
5.77E-01
9.45E-05
1.71E-01
4.22E+00
Floor-wall
seam
perimeter,
xcr,cl
(cm)
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
3400
Crack
depth
below
grade,
zcrack
(cm)
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
Crack
radius,
(cm)
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
4.06
Average
vapor
flow rate
into
bldg.
Q,.,i
(cnf/s)
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
2.59
Source-
bldg.
separation
at t=0,
L,°
(cm)
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Indoor
¥
4.52E-11
9.37E-06
2.59E-08
1.02E-07
2.51E-05
1.69E-05
2.45E-11
2.29E-06
8.45E-06
5.02E-12
5.92E-07
6.09E-07
9.22E-06
3.79E-06
3.14E-05
5.49E-06
6.38E-06
4.97E-12
3.82E-06
1.55E-09
2.26E-11
3.08E-07
1.25E-09
1.09E-10
4.23E-12
4.69E-08
7.96E-09
3.79E-05
1.08E-05
1.71E-08
4.20E-07
4.37E-07
5.95E-06
5.52E-06
1.24E-10
8.35E-08
1.68E-05
1.34E-06
1.08E-05
7.03E-10
3.45E-06
1.06E-04
Area of
base-
ment,
A,
(cm2)
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
1.38E+06
Bldg.
foundation
thick-
nesses,
Lcrack
(cm)
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Crack
effective
diffusion
coeffi-
cient,
Dcr.cl
(cm2/s)
1.05E-03
6.95E-03
5.53E-03
1.19E-03
8.31E-03
6.23E-03
9.43E-04
5.83E-03
8.31E-03
1.09E-03
5.51E-03
5.51E-03
5.93E-03
8.31E-03
7.19E-03
6.25E-03
5.00E-03
9.99E-04
5.99E-03
8.95E-04
9.75E-04
4.48E-03
4.33E-03
1.41E-03
1.41E-03
1.29E-03
1.99E-04
5.82E-03
8.07E-03
6.07E-03
5.67E-03
5.67E-03
5.75E-03
6.95E-03
9.27E-04
2.40E-03
6.23E-03
6.23E-03
6.31E-03
2.51E-03
6.79E-03
8.47E-03
Crack-
to-total
area
ratio,
1
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
001
001
0.01
0.01
0.01
001
0.01
0.01
0.01
0.01
0.01
001
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Area of
crack,
Acr,Ik
(cm2)
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1 38E+04
1.38E+04
1.38E+04
1 38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
1.38E+04
Indoor
p
3.85E+01
2.48E+02
1.98E+02
4.34E+01
2.97E+02
2.23E+02
3.46E+01
2.09E+02
2.97E+02
4.00E+01
1.97E+02
1.97E+02
2.12E+02
2.97E+02
2.57E+02
2.23E+02
1.79E+02
3.66E+01
2.14E+02
3.29E+01
3.57E+01
1.61E+02
1.55E+02
5.11E+01
5.11E+01
4.68E+01
8.08E+00
2.08E+02
2.88E+02
2.17E+02
2.03E+02
2.03E+02
2.06E+02
2.48E+02
3.40E+01
8.63E+01
2.23E+02
2.23E+02
2.26E+02
9.03E+01
2.43E+02
3.02E+02
Building
ventilation
rate,
Qbulldin,
(cnf/s)
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
2.90E+04
Depth to
source
lower
boundary,
AHC
(cm)
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
200
Exposure
duration,
T
(sec)
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
9.46E+08
H-15
-------
COMPARISON OF INDOOR AND OUTDOOR INHALATION SSLs FOR VOLATILE CONTAMINANTS
Chemical
Aldrin
Benzene
Bis(2-chloroethyl)ether
Bromoform
Carbon disulfide
Carbon tetrachloride
Chlordane
Chlorobenzene
Chloroform
DDT
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
1 ,2-Dichloropropane
1 ,3-Dichloropropene
Dieldrin
Ethylbenzene
Heptachlor
Heptachlor epoxide
Hexachloro-1 ,3-butadiene
Hexachlorobenzene
HCH-alpha(alpha-BHC)
HCH-beta(beta-BHC)
Hexachlorocyclopentadiene
Hexachloroethane
Methyl bromide
Methylene chloride
Nitrobenzene
Styrene
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1 ,1-Trichloroethane
1,1,2-Trichloroethane
Trichloroethylene
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
NA= not applicable
" = SSL based on C,,,
Infinite
source
indoor
attenuation
coefficient,
a
(unitless)
9.32E-05
2.65E-04
2.21E-04
9.59E-05
3.07E-04
2.43E-04
9.11E-05
2.30E-04
3.07E-04
9.40E-05
2.20E-04
2.20E-04
2.33E-04
3.07E-04
2.72E-04
2.43E-04
2.05E-04
9.21E-05
2.35E-04
9.02E-05
9.16E-05
1.89E-04
1.84E-04
1.01E-04
1.01E-04
9.80E-05
7.81E-05
2.30E-04
3.00E-04
2.38E-04
2.25E-04
2.25E-04
2.28E-04
2.65E-04
9.08E-05
1.27E-04
2.43E-04
2.43E-04
2.45E-04
1.30E-04
2.60E-04
3.12E-04
Finite
source
indoor
attenuation
coefficient,
a
(unitless)
8.68E-05
8.32E-05
8.87E-05
8.50E-05
7.93E-05
7.78E-05
8.66E-05
8.66E-05
8.51E-05
8.69E-05
8.81E-05
8.81E-05
8.15E-05
8.71E-05
7.47E-05
8.46E-05
8.16E-05
8.67E-05
8.55E-05
8.64E-05
8.67E-05
8.81E-05
8.86E-05
8.74E-05
8.74E-05
8.64E-05
7.41E-05
6.75E-05
8.40EE-05
8.87E-05
8.83E-05
8.83E-05
8.35E-05
8.53E-05
8.65E-05
8.77E-05
7.79E-05
8.76E-05
8.13E-05
8.82E-05
8.65E-05
6.38E-05
Time for
source
depletion,
Indoor TD
(sec)
1.33E+13
3.63E+08
1.05E+11
6.51E+09
1.61E+08
1.81E+08
2.24E+13
1.25E+09
4.79E+08
1.24E+14
4.59E+09
4.46E+09
3.16E+08
1.07E+09
1.12E+08
5.59E+08
3.88E+08
1.16E+14
7.71E+08
3.39E+11
2.50E+13
7.23E+09
1.72E+12
7.03E+12
1.82E+14
1.52E+10
2.47E+10
7.55E+07
3.63E+08
1.75E+11
6.65E+09
6.39E+09
4.76E+08
6.16E+08
4.38E+12
1.49E+10
1.83E+08
2.28E+09
2.87E+08
1.84E+12
9.65E+08
3.89E+07
Exposure
duration >
time for
depletion
(yes/no)
no
yes
no
no
yes
yes
no-
no
yes
no
no
no
yes
no
yes
yes
yes
no
yes
no
no
no
no
no
no
no
no
yes
yes
no
no
no
yes
yes
no
no
yes
no
yes
no
no
yes
Infinite
source
bldg.
cone.,
Cbulldin,
(kg/nf)
1.35E-06
1.20E-01
3.50E-04
2.79E-03
3.13E-01
2.22E-01
7.99E-07
3.05E-02
1.05E-01
1.45E-07
7.99E-03
8.22E-03
1.22E-01
4.73E-02
4.01 E-01
7.21E-02
8.82E-02
1.55E-07
5.06E-02
5.29E-05
7.16E-07
4.38E-03
1.80E-05
2.65E-06
1.02E-07
1.20E-03
1.05E-03
5.05E-01
1.36E-01
2.25E-04
5.64E-03
5.86E-03
7.95E-02
7.10E-02
4.09E-06
1.49E-03
2.20E-01
1.76E-02
1.41 E-01
1.23E-05
4.45E-02
1.32E+00
Finite
source
bldg.
cone.,
Cbulldin,
(kg/nf)
1.26E-06
1.51E-02
1.40E-04
2.47E-03
1.51E-02
1.51E-02
7.59E-07
1.15E-02
1.51E-02
1.34E-07
3.19E-03
3.28E-03
1.51E-02
1.34E-02
1.51E-02
1.51E-02
1.51E-02
1.46E-07
1.51E-02
5.06E-05
6.77E-07
2.04E-03
8.66E-06
2.30E-06
8.88E-08
1.06E-03
1.00E-03
1.51E-02
1.51E-02
8.41E-05
2.21E-03
2.30E-03
1.51E-02
1.51E-02
3.90E-06
1.03E-03
1.51E-02
6.37E-03
1.51E-02
8.34E-06
1.48E-02
1.51E-02
Infinite
source
indoor
volatiliza-
tion factor
VFmj,,r
(nf/kg)
7.42E+05
8.30E+00
2.86E+03
3.59E+02
3.20E+00
4.50E+00
1.25E+06
3.28E+01
9.49E+00
6.88E+06
1.25E+02
1.22E+02
8.17E+00
2.12E+01
2.49E+00
1.39E+01
1.13E+01
6.47E+06
1.97E+01
1.89E+04
1.40E+06
2.28E+02
5.55E+04
3.78E+05
9.76E+06
8.30E+02
9.48E+02
1.98E+00
7.38E+00
4.44E+03
1.77E+02
1.71E+02
1.26E+01
1.41E+01
2.44E+05
6.70E+02
4.54E+00
5.67E+01
7.07E+00
8.13E+04
2.25E+01
7.59E-01
Finite
source
indoor
volatili-
zation
factor
VF,nd..,
(nf/kg)
7.97E+05
6.63E+01
7.13E+03
4.05E+02
6.63E+01
6.63E+01
1.32E+06
8.71E+01
6.63E+01
7.44E+06
3.13E+02
3.05E+02
6.63E+01
7.46E+01
6.63E+01
6.63E+01
6.63E+01
6.87E+06
6.63E+01
1.98E+04
1.48E+06
4.89E+02
1.16E+05
4.36E+05
1.13E+07
9.42E+02
1.00E+03
6.63E+01
6.63E+01
1.19E+04
4.53E+02
4.36E+02
6.63E+01
6.63E+01
2.56E+05
9.71E+02
6.63E+01
1.57E+02
6.63E+01
1.20E+05
6.76E+01
6.63E+01
Unit risk
factor,
URF
(Hg/nf)-1
4.90E-03
8.30E-06
3.30E-04
1.10E-06
NA
1.50E-05
6.00E-05
NA
2.30E-05
9.70E-05
NA
NA
NA
2.60E-05
5.00E-05
NA
3.70E-05
4.60E-03
NA
1.30E-03
2.60E-03
2.20E-05
4.60E-04
1.80E-03
5.30E-04
NA
4.00E-06
NA
4.70E-07
NA
NA
5.80E-05
5.80E-07
NA
3.20E-04
NA
NA
1.60E-05
1.70E-06
3.10E-06
NA
8.40E-05
Reference
cone.,
RfC
(mg/m3)
NA
NA
NA
NA
1.00E-02
NA
NA
2.00E-02
NA
NA
2.00E-01
8.00E-01
5.00E-01
NA
NA
4.00E-03
2.00E-02
NA
1.00E+00
NA
NA
NA
NA
NA
NA
7.00E-05
NA
5.00E-03
3.00E+00
2.00E-03
1.00E+00
NA
NA
4.00E-01
NA
9.00E-03
1.00E+00
NA
NA
NA
2.00E-01
NA
Infinite
source
indoor
SSL,
carcinogen
(mg/kg)
3.69E-01
2.43E-03
2.11E-02
7.94E-01
NA
7.31E-04
5.08E+01
NA
1.00E-03
1.73E+02
NA
NA
NA
1.98E-03
1.21E-04
NA
7.46E-04
3.42E+00
NA
3.54E-02
1.31E+00
2.52E-02
2.94E-01
5. 11 E-01
4.48E+01
NA
5.77E-01
NA
3.82E-02
NA
NA
7.16E-03
5.28E-02
NA
1.86E+00
NA
NA
8.62E-03
1.01E-02
6.38E+01
NA
2.20E-05
Infinite
source
indoor
SSL, non-
carcinogen
(mg/kg)
NA
NA
NA
NA
3.33E-02
NA
NA
6.83E-01
NA
NA
2.61E+01
1.02E+02
4.26E+00
NA
NA
5.79E-02
2.37E-01
NA
2.06E+01
NA
NA
NA
NA
NA
NA
6.06E-02
NA
1.03E-02
2.31E+01
9.26E+00
1.85E+02
NA
NA
5.88E+00
NA
6.29E+00
4.74E+00
NA
NA
NA
4.69E+00
NA
Finite
source
indoor
SSL,
carcinogen
(mg/kg)
3.96E-01
1.94E-02
5.26E-02
8.96E-01
NA
1.08E-02
5.34E+01
NA
7.01E-03
1.87E+02
NA
NA
NA
6.98E-03
3.23E-03
NA
4.36E-03
3.63E+00
NA
3.70E-02
1.38E+00
5.41E-02
6. 11 E-01
5.89E-01
5.17E+01
NA
6.08E-01
NA
3.43E-01
NA
NA
1.83E-02
2.78E-01
NA
1.95E+00
NA
NA
2.39E-02
9.49E-02
9.42E+01
NA
1.92E-03
Finite
source
indoor
SSL, non-
carcinogen
(mg/kg)
NA
NA
NA
NA
6. 91 E-01
NA
NA
1.82E+00
NA
NA
6.53E+01
2.54E+02
3.46E+01
NA
NA
2.76E-01
1.38E+00
NA
6.91E+01
NA
NA
NA
NA
NA
NA
6.88E-02
NA
3.46E-01
2.07E+02
2.48E+01
4.72E+02
NA
NA
2.76E+01
NA
9.11E+00
6.91E+01
NA
NA
NA
1.41E+01
NA
Infinite
soure
risk-based
indoor
SSL,
(mg/kg)
3.69E-.01
2.43E-03
2.11E-02
7.94E-01
3.33E-02
7.31E-04
5.08E+01
6.83E-01
1 OOE-03
1.73E+02
2.61E+01
1.02E+02
4.26E+00
1.98E-03
1.21E-04
5.79E-02
7.46E-04
3.42E+00
2.06E+01
3.54E-02
1.31E+00
2.52E-02
2.94E-01
5. 11 E-01
4.48E+01
6.06E-02
5.77E-01
1.03E-02
3.82E-02
9.26E+00
1.85E+02
7.16E-03
5.28E-02
5.88E+00
1.86E+00
6.29E+00
4.74E+00
8.62E-03
1.01E-02
6.38E+01
4.69E+00
2.20E-05
Finite
soure risk-
based
indoor
SSL,
(mg/kg)
3.96E-01
1.94E-02
5.26E-02
8.96E-01
6. 91 E-01
1.08E-02
5.34E+01
1.82E+00
7.01E-03
1.87E+02
6.53E+01
2.54E+02
3.46E+01
6.98E-03
3.23E-03
2.76E-01
4.36E-03
3.63E+00
6.91E+01
3.70E-02
1.38E+00
5.41E-02
6. 11 E-01
5.89E-01
5.17E+01
6.88E-02
6.08E-01
3.46E-01
3.43E-01
2.48E+01
4.72E+02
1.83E-02
2.78E-01
2.76E+01
1.95E+00
9.11E+00
6.91E+01
2.39E-02
9.49E-02
9.42E+01
1.41E+01
1.92E-03
Outdoor
apparent
diffusion
coefficient,
a
(cm2/s)
1.99E-09
5.82E-04
1.39E-06
5.13E-06
1.80E-03
9.90E-04
1.08E-09
1.12E-04
5.15E-04
2.21 E-10
2.74E-05
2.78E-05
5.94E-04
2.47E-04
2.40E-03
3.46E-04
5.01E-04
2.19E-10
1.88E-04
6.84E-08
9.93E-10
1.36E-05
5.51E-08
4.85E-09
1.87E-10
2.07E-06
3.54E-07
8.13E-03
1.06E-03
8.45E-07
1.89E-05
2.34E-05
2.95E-04
2.88E-04
5.62E-09
3.72E-06
1.04E-03
7.30E-05
6.27E-04
3.27E-08
6.78E-04
5.85E-02
Outdoor
volalitiza-
tion
factor,
VFoutdoor
(nf/kg)
9.84E+05
1.82E+03
3.72E+04
1.94E+04
1.03E+03
1.39E+03
1.34E+06
4.15E+03
1.93E+03
2.95E+06
8.38E+03
8.32E+03
1.80E+03
2.79E+03
8.95E+02
2.36E+03
1.96E+03
2.97E+06
3.20E+03
1.68E+05
1.39E+06
1.19E+04
1.87E+05
6.29E+05
3.20E+06
3.05E+04
7.37E+04
4.86E+02
1.35E+03
4.77E+04
1.01E+04
9.07E+03
2.55E+03
2.59E+03
5.85E+05
2.28E+04
1.36E+03
5.13E+03
1.75E+03
2.43E+05
1.68E+03
1.81E+02
H-16
-------
COMPARISON OF INDOOR AND OUTDOOR INHALATION SSLs FOR VOLATILE CONTAMINANTS
Chemical
Aldrin
Benzene
Bis(2-chloroethyl)ether
Bromoform
Carbon disulfide
Carbon tetrachloride
Chlordane
Chloro benzene
Chloroform
DDT
1 ,2-Dichlorobenzene
1 ,4-Dichlorobenzene
1,1-Dichloroethane
1,2-Dichloroethane
1 ,1 -Dichloroethylene
1 ,2-Dichloropropane
1 ,3-Dichloropropene
Dieldrin
Ethylbenzene
Heptachlor
Heptachlor epoxide
Hexachloro-1 ,3-butadiene
Hexachlorobenzene
HCH-alpha(alpha-BHC)
HCH-beta(beta-BHC)
Hexachlorocyclopentadiene
Hexachloroethane
Methyl bromide
Methylene chloride
Nitrobenzene
Styrene
1 , 1 ,2 ,2-Tetrachloroethane
Tetrachloroethylene
Toluene
Toxaphene
1 ,2,4-Trichlorobenzene
1,1,1-Trichloroethane
1 , 1 ,2-Trichloroethane
Trichloroethylene
2,4,6-Trichlorophenol
Vinyl acetate
Vinyl chloride
NA= not applicable
" = SSL based on C,,,
Outdoor
SSL,
carcinogen
(mg/kg)
4.89E-01
5.33E-01
2.74E-01
4.29E+01
NA
2.26E-01
5.42E+01
NA
2.04E-01
7.41E+01
NA
NA
NA
2.61E-01
4.36E-02
NA
1.29E-01
1.57E+00
NA
3.14E-01
1.30E+00
1.31E+00
9.88E-01
8.51E-01
1.47E+01
NA
4.48E+01
NA
6.97E+00
NA
NA
3.81E-01
1.07E+01
NA
4.45E+00
NA
NA
7.81 E-01
2.51E+00
1.90E+02
NA
5.25E-03
Outdoor
SSL, non-
carcinogen
(mg/kg)
NA
NA
NA
NA
1.08E+01
NA
NA
8.66E+01
NA
NA
1.75E+03
6.94E+03
9.39E+02
NA
NA
9.83E+00
4.09E+01
NA
3.33E+03
NA
NA
NA
NA
NA
NA
2.23E+00
NA
2.54E+00
4.21E+03
9.95E+01
1.05E+04
NA
NA
1.08E+03
NA
2.14E+02
1.42E+03
NA
NA
NA
3.51E+02
NA
Risk-based
outdoor SSL
(mg/kg)
4.89E-01
5.33E-01
2.74E-01
4.29E+01
1.08E+01
2.26E-01
5.42E+01
8.66E+01
2.04E-01
7.41E+01
1.75E+03
6.94E+03
9.39E+02
2.61 E-01
4.36E-02
9.83E+00
1.29E-01
1.57E+00
3.33E+03
3.14E-01
1.30E+00
1.31E+00
9.88E-01
8.51E-01
1.47E+01
2.23E+00
4.48E+01
2.54E+00
6.97E+00
9.95E+01
1.05E+04
3.81E-01
1.07E+01
1.08E+03
4.45E+00
2.14E+02
1.42E+03
7.81 E-01
2.51 E+00
1.90E+02
3.51 E+02
5.25E-03
Pure
component
solubility,
S
(mg/L)
7.84E-02
1.78E+03
1.18E+04
3.21E+03
2.67E+03
7.92E+02
2.19E-01
4.09E+02
7.96E+03
3.41 E-03
1.25E+02
7.30E+01
5.16E+03
8.31E+03
3.00E+03
2.68E+03
1.55E+03
1.87E-01
1.73E+02
2.73E-01
2.68E-01
2.54E+00
8.62E-03
2.40E+00
5.42E-01
1.53E+00
4.08E+01
1.45E+04
1.74E+04
1.92E+03
2.57E+02
3.07E+03
2.32E+02
5.58E+02
6.79E-01
3.07E+01
1.17E+03
4.40E+03
1.18E+03
7.53E+02
2.24E+04
2.73E+03
Soil
saturation
cone.,
csat
(mg/kg)
2.28E+01
8.61 E+02
6.56E+03
2.76E+03
1.36E+03
1.04E+03
6.74E+01
5.55E+02
3.71 E+03
4.85E+00
2.97E+02
2.35E+02
2.36E+03
2.81 E+03
2.04E+03
1.08E+03
4.32E+02
1.22E+01
2.57E+02
1.12E+01
1.17E+01
1.07E+02
1.94E+00
2.56E+01
7.47E+00
8.84E+01
4.53E+02
3.83E+03
3.73E+03
1.70E+03
1.44E+03
1.77E+03
4.72E+02
5.21 E+02
2.11E+00
2.87E+02
9.80E+02
2.48E+03
8.80E+02
1.35E+03
3.01 E+03
2.26E+03
Indoor SSL,
infinite
source
(mg/kg)
0.4
0.002
0.02
0.8
0.03
0.0007
51
0.7
0.001
5"
26
102
4
0.002
0.0001
0.06
0.0007
3
21
0.04
1
0.03
0.3
0.5
T
0.06
0.6
0.01
0.04
9
185
0.007
0.05
6
2
6
5
0.009
0.01
64
5
0.00002
Indoor SSL,
finite source
(mg/kg)
0.4
0.02
0.05
0.9
0.7
0.01
53
2
0.007
5"
65
235
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
28
2
9
69
0.02
0.09
94
14
0.002
Outdoor
SSL, infinite
source
(mg/kg)
0.5
0.5
0.3
43
11
0.2
54
87
0.2
5"
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
521"
2°
214
980"
1
3
190
351
0.01
H-17
-------
APPENDIX I
SSL Simulation Results
-------
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 number 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 assumed true value of the site coefficient of variation, (i.e. the true site standard
deviation divided by the true site 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 = lognormal, G = gamma, W = Weibull).
MxL, MxG, MxW =
C40L, C40G, C40W =
C30L, C30G, C30W =
C20L, C20G, C20W =
C1OL, C1OG, C1OW=
C05L, C05G, C05W =
COIL, CO1G, CO1W =
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 .30 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.
l-l
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
C=1N=4CV=1.5
C30G C20G C10G C05G
0.5
0.5
2.0
2.0
MU
0.5
0.5
0.5
2.0
2.0
2.0
MU
0.5
0 5
0 5
2.0
2.0
2 0
MU
0.5
0.5
0 5
2 0
2.0
2.0
MU
0.5
0.5
0 5
2 0
2.0
? 0
.00 .124
.50 .192
.00 750
.50 .848
MIX MxL
.00 .140
.50 .213
75 .343
.00 700
.50 753
75 .699
MIX MxL
.00 .157
50 212
85 459
.00 .642
.50 .692
85 481
MIX MxL
.00 .180
.50 .206
85 366
00 632
.50 .631
.85 .457
MIX MxL
.00 .187
.50 .210
90 312
00 616
.50 .607
.90 .346
.314
.371
.974
.863
C40L
.296
.320
.387
.919
.813
.698
C40L
.266
296
445
.857
.752
481
C40L
.280
.277
364
820
.683
.457
C40L
.267
.262
306
791
.659
.346
.246
.295
.960
.816
C30L
.227
.240
.298
.890
766
.689
C30L
.203
227
357
.816
.699
481
C30L
.234
.225
289
778
.637
.456
C30L
.221
.217
275
746
.601
.346
.161 .087 .042 .016 .099 .835 .195 .350 .276 .195 .088 .043 .011 .172 .975 .176
.182 .089 .039 .016 .297 .937 .194 .347 .270 .170 .072 .028 .013 .257 .931 .194
.926 .858 759 .426 .873 .972 757 .874 .838 775 .650 .494 .188 742 .991 760
745 .569 .361 .132 703 .947 792 .813 774 .699 .540 .334 .106 .659 .925 .825
C=1 N=4 CV=2.0
C20L C10L C05L C01L LfL LoL MxG C40G C30G C20G C10G C05G C01G LfG LoG MxW
.150 .073 .036 .007 .088 .887 .255 .338 .277 .198 .106 .046 .009 .182 .965 .219
.167 .080 .036 .010 .214 .932 .273 .343 .273 .191 .084 .031 .007 .201 .905 .241
.193 .067 .021 .004 .241 .673 .366 .407 .302 .170 .061 .023 .008 .205 .680 .377
.838 733 .593 .260 765 .964 .676 740 .698 .645 .493 .341 .072 .624 .988 .695
.688 .524 .339 .112 .680 .941 .694 712 .675 .625 .466 .291 .072 .522 .926 713
.657 .456 .176 .028 .433 704 .675 .673 .658 .615 .430 .161 .030 .406 .681 .669
C=1 N=4 CV=2.5
C20L C10L C05L C01L LfL LoL MxG C40G C30G C20G C10G C05G C01G LfG LoG MxW
.138 .072 .026 .002 .088 .900 .256 .317 .257 .197 .089 .034 .004 .183 .944 .225
165 085 038 007 186 927 267 324 260 187 093 033 003 194 868 264
208 060 014 001 103 490 449 437 350 215 063 015 006 117 486 441
.764 .648 .512 .192 .672 .962 .620 .668 .623 .562 .423 .266 .044 .526 .972 .624
.620 .468 .312 .089 .617 .930 .613 .630 .592 .537 .404 .219 .040 .496 .885 .635
480 438 082 009 421 481 474 474 474 474 425 096 008 406 474 473
C=1 N=4 CV=3.0
C20L C10L C05L C01L LfL LoL MxG C40G C30G C20G C10G C05G C01G LfG LoG MxW
.163 .091 .034 .002 .119 .912 .263 .291 .238 .178 .093 .030 .004 .164 .859 .226
.158 .074 .029 .002 .170 .941 .277 .297 .244 .179 .075 .028 .002 .142 .764 .270
207 090 023 001 135 501 316 309 265 198 075 015 002 105 450 359
724 593 429 143 631 965 566 589 552 502 377 225 030 455 921 581
.558 .434 .277 .068 .574 .939 .531 .555 .501 .447 .329 .179 .024 .388 .860 .591
.438 .336 .106 .005 .314 .459 .451 .450 .444 .422 .356 .123 .002 .348 .465 .471
C=1 N=4 CV=3.5
C20L C10L C05L C01L LfL LoL MxG C40G C30G C20G C10G C05G C01G LfG LoG MxW
.152 .085 .035 .002 .104 .905 .231 .252 .203 .149 .078 .019 .003 .124 .784 .226
.154 .082 .029 .002 .151 .897 .268 .273 .242 .186 .098 .022 .002 .145 .722 .231
199 070 021 002 071 336 301 299 266 222 101 018 000 094 341 299
674 555 397 103 580 962 489 509 466 416 311 179 017 361 862 558
.532 .403 .270 .058 .521 .945 .512 .516 .492 .445 .338 .199 .014 .376 .819 .524
.346 .310 .131 .002 .307 .346 .363 .363 .362 .356 .321 .111 .000 .311 .364 .328
.360
.373
.903
.839
C40W
.315
.330
.414
.809
747
.664
C40W
.306
310
432
.718
.672
473
C40W
.275
.302
353
657
.616
.468
C40W
.259
.266
293
631
.549
.328
.286
.294
.870
799
C30W
.261
.261
.305
770
.698
.642
C30W
.236
263
350
.671
.627
470
C30W
.226
.253
306
615
.588
.455
C30W
.214
.214
266
587
.506
.327
.209
.197
.808
729
C20W
.169
.167
.175
716
.620
.596
C20W
.168
189
213
.605
.567
464
C20W
.163
.181
235
558
.530
.435
C20W
.163
.170
216
509
.448
.324
.104
.086
707
.589
C10W
.084
.075
.055
.596
.481
.442
C10W
.078
091
064
.469
.412
419
C10W
.087
.087
098
430
.395
.353
C10W
.080
.079
096
398
.331
.291
.042
.039
.551
.370
C05W
.033
.032
.016
.433
.295
.175
C05W
.028
032
012
.331
.233
091
C05W
.029
.024
028
280
.242
.138
C05W
.034
.026
023
247
.190
.120
.009
.017
.240
.117
C01W
.006
.009
.003
.138
.073
.027
C01W
.005
006
005
.073
.036
011
C01W
.005
.003
003
048
.038
.005
C01W
.002
.004
001
037
.023
.004
.151 .945
.265 .929
768 .988
.686 .952
LfW LoW
.143 .961
.214 .902
.215 .676
.663 .995
.585 .938
.425 .694
LfW LoW
.145 .948
191 896
120 499
.558 .986
.530 .918
404 474
LfW LoW
.142 .950
.163 .872
133 480
514 986
.489 .905
.348 .491
LfW LoW
.137 .945
.141 .860
099 348
477 983
.410 .891
.289 .331
1-2
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
C40L C30L C20L C10L C05L C01L
C=1 N=4 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0.5
0 5
0 5
2.0
2.0
? n
.00
50
90
.00
.50
.90
.173 .241 .203 .141 .061 .024 .003 .091 .920 .255 .265 .234 .177 .093 .029 .002 .141
198 239 190 142 072 027 001 135 916 232 235 191 146 078 027 001 106
273 268 235 184 078 012 000 089 329 280 269 242 196 111 032 003 115
.594 .750 .708 .632 .502 .387 .104 .534 .963 .443 .450 .426 .387 .294 .162 .009 .333
.590 .655 .596 .501 .385 .239 .039 .513 .942 .453 .459 .424 .380 .283 .150 .009 .315
.354 .354 .351 .340 .305 .123 .001 .296 .355 .330 .330 .325 .314 .277 .132 .004 .270
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=1 N=6 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
2.0
? 0
00
50
.00
.50
183 338 252 180 089 039 005 099 636 250 353 270 173 098 047 006 172
233 367 285 206 100 055 014 320 981 276 371 298 206 104 049 007 300
.860 .986 .979 .971 .939 .896 .658 .945 .946 .873 .928 .902 .861 .789 .677 .356 .843
.931 .924 .901 .844 .722 .565 .242 .799 .992 .926 .899 .868 .830 .730 .573 .235 .781
C40L C30L C20L C10L C05L C01L
C=1 N=6 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
75
00
.50
.75
.220 .301 .237 .168 .081 .032 .004 .099 .728 .348 .348 .280 .189 .100 .039 .005 .182
.260 .304 .253 .176 .085 .034 .005 .223 .965 .366 .369 .281 .183 .088 .034 .002 .209
459 385 293 197 083 038 007 188 844 486 395 296 199 082 030 003 179
816 960 946 917 849 760 481 855 947 822 832 802 752 630 497 167 721
.891 .891 .842 .777 .633 .497 .189 .732 .989 .837 .818 .786 .718 .609 .453 .155 .657
.819 .796 .769 .697 .506 .353 .068 .461 .824 .804 .777 .747 .696 .518 .360 .058 .478
C40L C30L C20L C10L C05L C01L
C=1 N=6 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
0 5
2.0
2.0
? n
.00
50
85
.00
.50
.85
.267 .292 .241 .163 .095 .047 .004 .105 .796 .356 .324 .254 .172 .088 .044 .005 .162
312 333 266 190 092 035 004 219 956 371 329 252 179 073 034 002 152
591 409 300 194 082 031 003 154 631 597 434 311 200 085 027 000 154
.804 .932 .905 .868 .791 .681 .340 .801 .947 .759 .745 .705 .649 .521 .362 .086 .585
.827 .834 .789 .724 .591 .435 .151 .710 .973 .765 .734 .700 .636 .520 .367 .076 .552
.628 .628 .628 .621 .456 .221 .014 .226 .628 .596 .596 .596 .586 .431 .223 .019 .226
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=1 N=6 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.249 .282 .230 .159 .083 .034 .001 .091 .808 .380 .327 .275 .211 .101 .036 .003 .155
.305 .296 .235 .164 .083 .031 .008 .173 .950 .393 .337 .274 .208 .100 .040 .001 .161
.495 .377 .290 .188 .072 .024 .003 .118 .647 .486 .397 .329 .228 .099 .033 .005 .132
.797 .899 .865 .833 .745 .595 .261 .760 .956 .705 .669 .621 .560 .444 .307 .060 .487
.778 .780 .740 .670 .544 .379 .114 .644 .980 .690 .646 .602 .537 .429 .287 .049 .443
.637 .623 .607 .556 .403 .215 .023 .230 .638 .611 .585 .564 .524 .403 .209 .029 .239
.727
.643
.375
.753
.751
.340
LOG
.953
.971
.992
.980
LOG
.957
.931
.833
.993
.966
.812
LOG
.921
.892
.637
.980
.951
.596
LOG
.854
.844
.647
.947
.892
.627
.194
.235
.240
.511
.505
.330
MxW
.218
.320
.882
.920
MxW
.317
.387
.504
.831
.846
.786
MxW
.286
.373
.559
.781
.795
.650
MxW
.336
.337
.460
.720
.704
.583
.213
.254
.234
.579
.517
.326
C40W
.357
.418
.956
.891
C40W
.346
.372
.393
.887
.834
.768
C40W
.297
.340
.434
.815
.756
.649
C40W
.322
.296
.371
.752
.676
.560
.178
.215
.207
.526
.486
.320
C30W
.279
.303
.940
.869
C30W
.276
.289
.298
.849
.787
.738
C30W
.243
.263
.318
.772
.708
.646
C30W
.254
.239
.306
.707
.634
.536
.131
.147
.155
.466
.427
.301
C20W
.184
.208
.908
.821
C20W
.202
.217
.187
.805
.713
.665
C20W
.162
.183
.203
.715
.641
.629
C20W
.191
.183
.211
.652
.557
.500
.068
.074
.082
.342
.318
.254
C10W
.061
.107
.841
.705
C10W
.101
.114
.068
.689
.607
.473
C10W
.090
.089
.073
.599
.512
.470
C10W
.100
.094
.081
.517
.442
.406
.025
.023
.020
.213
.166
.145
C05W
.027
.045
.733
.543
C05W
.040
.039
.023
.539
.450
.313
C05W
.040
.035
.019
.474
.370
.230
C05W
.039
.035
.027
.391
.312
.203
.001
.000
.000
.018
.017
.000
C01W
.005
.007
.417
.213
C01W
.004
.003
.003
.229
.147
.058
C01W
.005
.004
.001
.150
.107
.020
C01W
.004
.007
.001
.112
.068
.023
.105
.125
.092
.445
.380
.254
LfW
.128
.304
.878
.768
LfW
.157
.243
.172
.758
.675
.423
LfW
.138
.178
.156
.664
.576
.232
LfW
.149
.144
.112
.595
.477
.224
.910
.822
.324
.979
.878
.353
LoW
.916
.965
.992
.988
LoW
.947
.958
.809
.990
.980
.804
LoW
.942
.919
.618
.990
.963
.651
LoW
.941
.877
.624
.985
.952
.604
1-3
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=6 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
LoG MxW C40W C30W C20W C10W C05W C01W LfW LoW
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
90
00
.50
.90
.240
.289
407
771
.756
.442
.251 .203 .140 .079 .028 .002 .082 .833 .361 .309 .271 .201 .102 .034 .002 .130 .785
.271 .212 .154 .080 .040 .003 .137 .931 .369 .307 .240 .164 .087 .037 .003 .103 .784
326 258 185 093 032 001 091 441 417 340 290 202 079 037 002 086 460
870 832 777 680 544 217 706 958 616 586 529 470 373 234 034 371 873
.744 .687 .629 .513 .354 .099 .598 .980 .647 .602 .564 .508 .410 .259 .038 .389 .851
.440 .440 .433 .354 .123 .014 .123 .442 .449 .445 .442 .428 .350 .140 .014 .141 .451
.296 .265 .216
.340 .287 .237
.403 .357 .311
.714 .727 .684
.672 .636 .596
.472 .468 .459
.160
.176
.223
.614
.535
.448
.085 .037 .003
.099 .040 .002
.085 .033 .000
.489 .344 .080
.433 .270 .058
.375 .143 .009
.117
.126
.084
.562
.463
.144
.927
.854
.457
.987
.936
.476
MU
MU
C40L C30L C20L C10L C05L C01L
MxG C40G
C=1 N=6 CV=4.0
C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
90
00
.50
.90
.246 .255 .200 .131 .062 .023 .004 .084 .839 .342 .275 .229 .181 .092 .028 .001 .099
.299 .271 .213 .153 .086 .037 .002 .134 .922 .315 .269 .220 .160 .084 .020 .000 .087
396 324 269 196 082 022 000 065 453 400 340 296 211 099 041 001 078
723 847 797 724 600 470 179 626 956 559 523 494 450 334 214 023 323
.745 .735 .683 .609 .482 .342 .093 .597 .968 .576 .532 .495 .429 .334 .200 .023 .299
.477 .472 .464 .439 .337 .140 .007 .145 .479 .471 .449 .437 .415 .335 .164 .013 .168
C40L C30L C20L C10L C05L C01L
MxG C40G
C=1 N=9 CV=1.5
C30G C20G C10G C05G
C01G LfG
0.5
0 5
2 0
? 0
.00
50
00
.50
.286 .335 .274 .205 .112 .057 .010 .112 .452 .363 .390 .293 .202 .089 .043 .006 .216
365 380 298 203 108 050 014 380 984 420 411 301 213 103 046 009 336
948 999 999 995 989 965 891 987 950 955 973 963 936 887 815 577 933
.983 .956 .940 .904 .820 .719 .449 .856 .995 .978 .954 .930 .900 .817 .697 .430 .841
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=1 N=9 CV=2.0
C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2.0
2.0
? n
.00
.50
75
.00
.50
.75
.312
.425
629
.913
.951
.918
.314 .247
.366 .289
418 307
.987 .983
.923 .892
.873 .826
.173 .101
.205 .092
215 107
.974 .955
.842 .756
.752 .615
.049 .005
.045 .007
053 008
.910 .714
.630 .338
.464 .138
.110 .592
.287 .962
198 840
.953 .948
.816 .989
.456 .920
.472 .369
.496 .384
642 417
.913 .897
.930 .908
.925 .872
.297
.285
324
.864
.884
.842
.199
.199
208
.828
.844
.761
.096 .034
.102 .045
102 051
.742 .619
.710 .576
.642 .469
.005 .206
.010 .250
005 187
.341 .808
.271 .743
.170 .472
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=1 N=9 CV=2.5
C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.364 .323 .256 .179 .092 .044 .006 .097 .667 .477 .346 .283 .196 .090 .046 .004 .166
.416 .315 .250 .172 .096 .042 .006 .216 .957 .525 .377 .297 .209 .101 .045 .007 .184
.706 .384 .280 .186 .083 .027 .002 .118 .744 .737 .407 .315 .197 .093 .036 .004 .118
.910 .980 .970 .954 .905 .846 .593 .912 .952 .867 .821 .782 .725 .623 .486 .191 .656
.931 .900 .868 .823 .731 .607 .322 .790 .990 .888 .823 .788 .729 .628 .489 .179 .629
.743 .741 .731 .681 .458 .367 .073 .364 .743 .764 .762 .750 .686 .432 .355 .064 .351
.725
.691
.497
.821
.792
.489
LoG
.937
.980
.999
.994
LOG
.934
.940
.841
.997
.987
.927
LOG
.864
.862
.754
.983
.961
.764
.289
.325
.350
.675
.623
.446
MxW
.336
.416
.957
.973
MxW
.397
.495
.656
.933
.938
.920
MxW
.435
.473
.719
.886
.902
.782
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.268 .216 .153 .088 .034 .002 .121 .883
.275 .226 .173 .080 .035 .001 .112 .818
.301 .254 .196 .078 .030 .003 .058 .438
.679 .634 .568 .457 .317 .057 .519 .982
.574 .531 .468 .348 .236 .032 .380 .906
.428 .411 .390 .326 .160 .013 .175 .473
C40W C30W C20W C10W C05W C01W LfW LoW
.375 .287 .188 .092 .036 .008 .168 .880
.392 .290 .201 .095 .046 .008 .335 .972
.985 .974 .965 .927 .855 .648 .950 .994
.955 .936 .907 .828 .713 .403 .863 .991
C40W C30W C20W C10W C05W C01W LfW LoW
.317 .239 .159 .087 .048 .009 .151 .907
.385 .301 .220 .112 .058 .010 .255 .940
.396 .306 .209 .102 .037 .003 .193 .840
.950 .930 .905 .826 .727 .409 .872 .989
.888 .860 .790 .696 .558 .267 .720 .986
.866 .838 .764 .609 .465 .170 .470 .923
C40W C30W C20W C10W C05W C01W LfW LoW
.350 .287 .199 .103 .049 .003 .159 .906
.341 .270 .188 .107 .049 .004 .179 .871
.406 .307 .185 .081 .025 .001 .100 .746
.892 .861 .811 .712 .601 .316 .788 .992
.834 .798 .738 .615 .489 .203 .649 .976
.773 .760 .708 .459 .370 .090 .366 .782
1-4
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=9 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
LoG MxW C40W C30W C20W C10W C05W C01W LfW LoW
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
85
00
.50
.85
.354
.443
615
888
.900
.763
.292 .218 .153 .074 .030 .006 .093 .729 .494 .338 .276 .205 .115 .049 .008 .144 .817
.313 .239 .166 .091 .041 .003 .191 .948 .507 .337 .265 .173 .082 .029 .002 .121 .801
391 310 203 103 040 003 101 686 604 363 284 200 085 030 002 080 660
953 930 895 840 739 438 834 937 851 785 731 676 548 399 099 537 955
.878 .848 .803 .693 .570 .232 .767 .983 .833 .744 .707 .644 .519 .368 .111 .485 .927
.726 .690 .618 .467 .306 .069 .287 .764 .753 .708 .682 .622 .479 .315 .089 .303 .764
C40L C30L C20L C10L C05L C01L
C=1 N=9 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
90
00
.50
.90
.353 .271 .222 .158 .085 .046 .002 .096 .742 .474 .323 .271 .198 .092 .046 .006 .112
.427 .319 .259 .186 .103 .038 .005 .179 .911 .471 .329 .265 .194 .094 .053 .002 .102
572 350 276 201 099 036 002 053 595 537 353 288 199 097 040 003 058
880 927 902 876 800 699 385 819 951 766 690 652 604 483 333 083 434
.885 .841 .795 .739 .606 .503 .198 .688 .983 .779 .699 .659 .612 .493 .335 .087 .409
.624 .620 .614 .577 .390 .195 .030 .189 .624 .610 .592 .581 .560 .401 .209 .056 .203
C40L C30L C20L C10L C05L C01L
C=1 N=9 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
0 5
2.0
2.0
? n
.00
50
90
.00
.50
.90
.385 .301 .230 .168 .097 .042 .005 .108 .755 .475 .315 .255 .182 .099 .043 .003 .079
431 307 252 181 097 033 003 152 876 444 323 270 215 116 040 004 089
520 326 267 197 099 042 007 057 546 483 324 267 191 095 037 003 044
.858 .895 .862 .809 .723 .619 .292 .740 .952 .730 .640 .596 .535 .429 .280 .056 .348
.864 .808 .774 .719 .610 .475 .173 .690 .982 .709 .608 .568 .520 .399 .257 .046 .308
.636 .609 .582 .523 .395 .223 .040 .205 .636 .594 .544 .523 .484 .385 .201 .031 .171
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=1 N=12 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
2.0
? 0
00
50
.00
.50
350
450
.985
.998
368
397
1.00
.977
276
311
1.00
.965
206 102 054 014 109 280 480 386 296 207 097 045 004 241
209 122 057 009 441 989 494 398 312 208 094 048 014 398
1.00 .999 .993 .962 .998 .955 .989 .993 .984 .971 .939 .906 .751 .967
.950 .904 .822 .598 .905 .998 .995 .981 .963 .948 .897 .815 .571 .902
C40L C30L C20L C10L C05L C01L
C=1 N=12 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.75
.00
.50
.75
.390 .330 .273 .186 .084 .041 .004 .094 .439 .576 .383 .300 .199 .101 .043 .006 .223
.496 .367 .281 .189 .093 .046 .008 .313 .964 .593 .351 .264 .178 .085 .046 .009 .231
.738 .396 .290 .181 .100 .057 .008 .171 .856 .747 .414 .320 .201 .090 .041 .011 .186
.967 .997 .995 .993 .977 .959 .865 .978 .946 .977 .952 .927 .899 .821 .746 .480 .890
.988 .967 .949 .918 .852 .763 .500 .867 1.00 .971 .933 .910 .878 .778 .682 .399 .786
.963 .907 .875 .821 .702 .580 .301 .565 .905 .972 .928 .900 .840 .724 .598 .305 .575
.817
.801
.660
.955
.927
.764
LOG
.728
.701
.560
.896
.873
.611
LOG
.673
.628
.527
.845
.822
.601
LOG
.922
.974
.998
.996
LOG
.949
.943
.822
.999
.991
.929
.426
.464
.609
.863
.855
.736
MxW
.418
.427
.553
.836
.819
.620
MxW
.408
.431
.471
.836
.797
.588
MxW
.438
.507
.987
.992
MxW
.524
.590
.741
.977
.981
.954
.313
.330
.379
.856
.805
.692
C40W
.293
.293
.383
.796
.721
.594
C40W
.267
.299
.321
.774
.733
.544
C40W
.353
.396
.996
.971
C40W
.377
.393
.402
.978
.947
.900
.256
.264
.284
.823
.761
.660
C30W
.249
.241
.322
.748
.683
.573
C30W
.211
.239
.275
.729
.674
.523
C30W
.276
.313
.990
.948
C30W
.305
.321
.302
.965
.920
.866
.172
.194
.187
.777
.700
.612
C20W
.190
.175
.214
.693
.618
.537
C20W
.154
.174
.181
.665
.603
.478
C20W
.189
.223
.986
.926
C20W
.214
.225
.198
.946
.880
.818
.079
.105
.083
.684
.597
.476
C10W
.106
.080
.108
.572
.506
.424
C10W
.076
.088
.077
.558
.488
.373
C10W
.096
.107
.972
.877
C10W
.124
.119
.085
.895
.793
.706
.038
.047
.032
.560
.446
.306
C05W
.051
.040
.043
.435
.369
.213
C05W
.038
.034
.030
.410
.330
.201
C05W
.056
.052
.948
.800
C05W
.059
.057
.047
.840
.684
.558
.003
.004
.005
.226
.166
.077
C01W
.004
.003
.003
.171
.097
.041
C01W
.002
.005
.001
.143
.089
.035
C01W
.008
.013
.808
.560
C01W
.008
.012
.007
.595
.402
.281
.133
.156
.085
.751
.610
.299
LfW
.127
.119
.059
.639
.507
.204
LfW
.118
.098
.038
.607
.477
.179
LfW
.185
.386
.980
.876
LfW
.204
.285
.181
.937
.811
.542
.850
.854
.671
.988
.958
.744
LoW
.835
.790
.584
.982
.954
.624
LoW
.798
.747
.510
.977
.924
.594
LoW
.821
.975
.990
.994
LoW
.919
.940
.809
.998
.995
.907
1-5
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=1 N=12 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
435 296 223 159 084 038 005 089 520 627 375 293 201 092 040 006 176
492 305 234 175 083 041 006 227 942 623 384 287 195 087 042 007 159
.833 .413 .312 .194 .104 .036 .003 .092 .596 .802 .411 .308 .216 .097 .046 .009 .091
.968 .991 .984 .982 .959 .921 .720 .963 .956 .939 .879 .852 .803 .703 .583 .289 .742
969 940 914 877 799 686 395 844 995 951 882 850 795 691 577 257 684
.859 .844 .812 .720 .544 .406 .116 .250 .842 .851 .835 .807 .715 .568 .455 .120 .264
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=12 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
429 311 239 162 079 042 005 092 612 600 365 283 197 099 043 004 119
510 322 253 187 109 043 005 217 930 610 365 300 211 114 059 005 148
.728 .381 .290 .205 .117 .056 .012 .088 .593 .707 .358 .283 .198 .107 .052 .004 .089
.933 .972 .959 .932 .900 .843 .605 .909 .932 .925 .821 .784 .724 .608 .492 .183 .587
955 917 883 843 748 634 318 797 991 909 826 786 719 599 458 167 520
.844 .769 .718 .631 .511 .364 .124 .252 .760 .852 .786 .745 .680 .535 .378 .130 .267
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=12 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
437 307 237 174 095 044 005 111 660 608 360 294 214 108 042 005 109
.508 .305 .238 .174 .094 .040 .003 .173 .908 .589 .325 .259 .182 .082 .046 .004 .088
.680 .377 .299 .196 .085 .037 .001 .033 .409 .666 .388 .304 .204 .100 .050 .004 .055
940 957 934 912 857 777 522 863 952 870 760 712 648 529 394 134 445
945 900 863 809 715 596 284 765 989 866 749 690 627 506 359 108 398
.723 .710 .684 .628 .420 .295 .083 .147 .703 .709 .685 .656 .598 .439 .283 .092 .158
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=12 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.470 .290 .226 .169 .090 .043 .005 .103 .699 .559 .317 .261 .197 .107 .049 .005 .081
.495 .313 .263 .173 .098 .043 .002 .152 .904 .563 .319 .260 .182 .108 .043 .004 .080
638 351 289 193 094 051 007 052 425 605 365 304 204 095 043 004 040
942 950 926 899 824 740 471 838 955 830 705 667 605 490 347 093 374
.930 .849 .819 .769 .673 .561 .227 .736 .985 .817 .689 .640 .586 .481 .317 .087 .325
.713 .672 .640 .575 .425 .286 .074 .151 .666 .710 .652 .621 .569 .437 .257 .051 .139
C40L C30L C20L C10L C05L C01L
C=1 N=16 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.438 .354 .279 .202 .102 .053 .007 .092 .161 .565 .393 .307 .216 .100 .058 .012 .293
.535 .387 .303 .205 .110 .062 .014 .482 .986 .586 .411 .291 .200 .111 .052 .004 .459
.996 1.00 1.00 1.00 1.00 1.00 .997 1.00 .957 .994 .995 .992 .989 .975 .956 .872 .989
1.00 .990 .982 .973 .950 .910 .761 .945 1.00 .999 .988 .982 .959 .925 .875 .714 .932
.902
.878
.558
.991
.976
.830
LOG
.806
.778
.540
.964
.943
.784
LOG
.735
.708
.424
.929
.889
.681
LOG
.640
.632
.442
.861
.830
.654
LOG
.900
.976
.999
.998
.530
.614
.814
.965
.954
.856
MxW
.547
.565
.713
.933
.912
.820
MxW
.505
.541
.632
.918
.886
.708
MxW
.525
.559
.585
.883
.880
.706
MxW
.542
.595
.993
.998
.340
.377
.387
.948
.894
.829
C40W
.337
.355
.404
.901
.852
.755
C40W
.316
.301
.386
.870
.783
.672
C40W
.328
.340
.343
.803
.774
.647
C40W
.371
.400
.997
.987
.273
.285
.290
.919
.860
.798
C30W
.258
.269
.315
.873
.814
.716
C30W
.260
.247
.316
.838
.737
.654
C30W
.273
.274
.282
.760
.735
.611
C30W
.287
.292
.996
.976
.190
.197
.195
.877
.824
.730
C20W
.190
.192
.225
.832
.752
.657
C20W
.194
.179
.235
.792
.668
.605
C20W
.201
.197
.196
.710
.678
.568
C20W
.196
.204
.990
.962
.088
.093
.086
.816
.736
.549
C10W
.090
.093
.118
.754
.634
.529
C10W
.110
.093
.121
.677
.572
.445
C10W
.099
.097
.093
.618
.555
.443
C10W
.111
.103
.983
.928
.038
.045
.037
.709
.624
.438
C05W
.047
.048
.053
.634
.510
.380
C05W
.046
.039
.047
.563
.438
.288
C05W
.047
.045
.049
.497
.415
.252
C05W
.054
.062
.973
.885
.005
.005
.003
.406
.307
.134
C01W
.006
.006
.004
.310
.227
.119
C01W
.004
.006
.005
.238
.176
.091
C01W
.004
.006
.002
.181
.136
.061
C01W
.010
.007
.924
.734
.164
.187
.086
.873
.731
.255
LfW
.145
.149
.090
.808
.637
.250
LfW
.139
.110
.049
.747
.561
.166
LfW
.116
.107
.046
.668
.508
.148
LfW
.223
.447
.988
.922
.898
.910
.539
.996
.983
.822
LoW
.870
.842
.591
.992
.972
.762
LoW
.841
.788
.434
.991
.963
.670
LoW
.848
.785
.412
.983
.945
.643
LoW
.801
.978
.995
.999
1-6
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=1 N=16 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.75
.00
50
.75
502 331 258 173 099 061 007 103 291 682 401 319 219 113 057 009 279
596 379 291 202 112 058 011 371 966 736 394 296 213 092 041 006 276
.804 .401 .314 .229 .098 .051 .007 .197 .805 .809 .389 .308 .206 .093 .046 .004 .173
.990 .997 .997 .996 .993 .987 .939 .992 .957 .994 .981 .976 .957 .919 .851 .623 .956
997 984 975 957 920 850 659 925 999 992 971 952 929 866 778 528 862
.992 .950 .927 .886 .805 .697 .404 .599 .946 .987 .950 .920 .873 .789 .672 .413 .563
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=16 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
541 327 265 188 110 047 009 112 420 700 375 298 204 106 057 009 195
629 364 294 209 110 053 009 300 947 725 347 276 196 107 060 014 155
.892 .405 .319 .204 .101 .043 .003 .085 .576 .898 .420 .307 .212 .099 .050 .009 .079
.984 .994 .991 .985 .978 .962 .849 .978 .953 .979 .922 .905 .864 .785 .688 .425 .813
991 965 948 923 871 799 557 885 995 972 918 899 860 784 671 384 736
.926 .899 .847 .772 .657 .489 .199 .285 .719 .935 .904 .868 .794 .710 .520 .244 .337
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=16 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.85
00
50
.85
559 322 240 173 096 036 005 094 504 705 367 267 186 096 045 007 119
.618 .330 .268 .181 .087 .039 .006 .226 .924 .716 .375 .289 .210 .113 .052 .008 .128
.808 .391 .304 .220 .123 .060 .007 .079 .543 .821 .396 .309 .203 .103 .060 .009 .076
976 986 981 966 949 907 747 949 948 973 895 868 814 710 579 294 656
982 948 929 899 831 732 457 860 995 962 874 848 790 682 550 257 596
.926 .850 .800 .734 .615 .478 .195 .285 .720 .905 .830 .773 .721 .576 .449 .193 .277
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=1 N=16 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.531 .287 .224 .156 .080 .038 .004 .084 .520 .676 .354 .286 .209 .108 .046 .007 .094
.642 .331 .273 .200 .102 .056 .007 .209 .906 .667 .352 .278 .192 .099 .048 .004 .079
787 387 302 211 097 044 004 033 415 760 390 303 197 097 048 004 034
978 984 978 961 924 880 691 938 960 935 804 769 710 599 473 221 499
.973 .932 .913 .872 .792 .677 .397 .830 .995 .915 .787 .756 .695 .582 .455 .180 .432
.803 .772 .728 .655 .506 .380 .128 .190 .504 .799 .744 .714 .639 .485 .357 .125 .168
.937
.935
.783
.998
.996
.928
LOG
.872
.834
.604
.992
.976
.755
LOG
.776
.729
.540
.979
.952
.705
LOG
.683
.636
.419
.914
.876
.493
.628
.691
.849
.997
.995
.987
MxW
.636
.693
.908
.980
.982
.928
MxW
.621
.680
.812
.968
.967
.909
MxW
.621
.675
.750
.952
.955
.813
.378
.400
.404
.995
.960
.941
C40W
.335
.392
.398
.971
.932
.896
C40W
.333
.358
.405
.943
.902
.841
C40W
.331
.359
.381
.892
.874
.758
.304
.308
.300
.991
.947
.917
C30W
.252
.305
.306
.955
.912
.861
C30W
.262
.281
.311
.925
.868
.805
C30W
.268
.291
.288
.869
.849
.725
.209
.205
.218
.980
.928
.876
C20W
.179
.194
.206
.940
.873
.787
C20W
.179
.200
.210
.890
.822
.739
C20W
.202
.201
.212
.820
.801
.665
.108
.100
.103
.961
.871
.781
C10W
.094
.096
.111
.897
.807
.675
C10W
.082
.107
.092
.824
.733
.611
C10W
.103
.113
.114
.750
.705
.510
.050
.047
.038
.925
.790
.690
C05W
.044
.046
.049
.837
.717
.478
C05W
.042
.054
.045
.739
.627
.451
C05W
.045
.053
.052
.652
.565
.360
.008
.009
.005
.753
.566
.394
C01W
.006
.004
.002
.581
.423
.201
C01W
.004
.007
.006
.467
.338
.177
C01W
.005
.004
.012
.337
.256
.111
.217
.290
.187
.982
.875
.607
LfW
.170
.193
.090
.935
.799
.297
LfW
.144
.160
.061
.877
.732
.282
LfW
.130
.123
.045
.791
.657
.167
.870
.950
.794
1.00
.997
.924
LoW
.868
.872
.600
.998
.988
.731
LoW
.845
.822
.539
.996
.974
.716
LoW
.820
.746
.399
.996
.968
.524
1-7
-------
MU
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
C40L C30L C20L C10L C05L C01L
C=1 N=16 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0.5
0 5
0 5
2.0
2.0
? n
.00
50
90
.00
.50
.90
.543
611
721
.973
.976
.810
.306 .249 .184 .112 .056 .003 .108 .576 .652 .327 .262 .190 .095 .038 .003 .058 .576
302 250 168 087 042 005 144 858 663 336 273 199 100 044 006 064 555
351 280 185 095 045 004 025 383 700 346 279 189 093 037 003 018 359
.980 .966 .944 .900 .841 .598 .905 .957 .897 .753 .709 .649 .535 .399 .127 .372 .866
.898 .869 .823 .738 .634 .352 .787 .994 .896 .742 .695 .635 .513 .393 .144 .342 .836
.745 .714 .642 .493 .346 .106 .158 .533 .784 .700 .663 .611 .456 .304 .099 .138 .504
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=4N=4CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
2 0
? 0
.00
50
00
.50
.035 .336 .261 .176 .088 .045 .018 .084 .687 .029 .403 .307 .221 .122 .082 .043 .126
026 374 280 211 113 073 033 156 852 019 390 295 204 125 073 039 177
835 1 00 1 00 1 00 997 991 890 997 960 857 994 988 984 966 924 674 977
.887 .994 .989 .978 .943 .845 .631 .989 .990 .886 .982 .976 .958 .909 .823 .572 .974
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=14N=4 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
? 0
.00
.50
75
00
.50
.75
.070 .331 .255 .188 .100 .059 .029 .106 .753 .089 .383 .292 .183 .103 .058 .020 .136
.091 .354 .292 .206 .120 .066 .024 .161 .889 .089 .385 .303 .205 .110 .064 .029 .187
055 385 293 197 117 074 036 339 939 059 394 300 205 107 074 042 358
788 999 997 993 983 964 762 983 966 820 964 952 931 860 737 436 905
.855 .976 .964 .947 .890 .791 .512 .960 .985 .870 .967 .953 .923 .837 .683 .360 .928
.884 .939 .911 .864 .742 .560 .283 .900 .987 .885 .954 .924 .861 .723 .543 .267 .914
C40L C30L C20L C10L C05L C01L
C=4 N=4 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
0 5
2.0
2.0
? n
.00
50
85
.00
.50
.85
.093 .295 .240 .172 .096 .047 .010 .101 .769 .133 .354 .283 .194 .085 .043 .018 .162
102 319 241 172 097 050 021 144 918 136 362 260 180 094 050 015 201
129 401 303 179 076 050 038 394 914 121 396 283 174 080 058 046 415
.764 .994 .992 .982 .963 .904 .653 .964 .957 .814 .945 .910 .868 .771 .611 .296 .836
.832 .966 .959 .935 .846 .735 .422 .928 .991 .808 .915 .883 .832 .720 .573 .268 .842
.893 .888 .833 .751 .525 .327 .129 .668 .922 .886 .886 .845 .757 .538 .355 .143 .677
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=4 N=4 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.107 .321 .249 .176 .090 .041 .013 .098 .844 .192 .371 .277 .181 .077 .033 .009 .194
.119 .321 .241 .165 .076 .039 .014 .134 .918 .188 .351 .273 .180 .082 .036 .011 .197
.195 .364 .279 .187 .087 .036 .010 .328 .927 .211 .404 .317 .212 .096 .052 .017 .351
.762 .987 .979 .971 .943 .872 .586 .948 .964 .762 .875 .840 .783 .656 .462 .171 .761
.791 .947 .920 .878 .784 .664 .340 .890 .990 .801 .887 .853 .793 .668 .491 .163 .810
.819 .826 .782 .710 .549 .355 .121 .668 .928 .824 .833 .796 .731 .568 .342 .111 .690
.576
.555
.359
.866
.836
.504
LOG
.814
.868
.983
.994
LOG
.931
.930
.932
.994
.996
.991
LOG
.963
.967
.907
.991
.988
.917
LOG
.975
.977
.918
.995
.997
.929
.629
.647
.707
.954
.932
.793
MxW
.030
.013
.840
.891
MxW
.072
.067
.056
.821
.839
.889
MxW
.117
.128
.126
.797
.810
.880
MxW
.149
.177
.212
.749
.762
.828
.311
.314
.370
.857
.825
.708
C40W
.377
.398
.996
.989
C40W
.381
.364
.379
.983
.974
.933
C40W
.332
.342
.378
.969
.925
.882
C40W
.328
.349
.368
.933
.897
.842
.264
.245
.296
.820
.785
.672
C30W
.291
.301
.994
.979
C30W
.288
.277
.268
.979
.963
.904
C30W
.267
.255
.274
.952
.902
.836
C30W
.258
.270
.298
.909
.863
.797
.173
.179
.221
.766
.718
.611
C20W
.204
.202
.993
.955
C20W
.195
.187
.183
.966
.927
.846
C20W
.181
.170
.160
.927
.855
.761
C20W
.178
.175
.209
.864
.817
.738
.086
.086
.095
.682
.606
.466
C10W
.095
.117
.980
.897
C10W
.104
.097
.104
.923
.832
.695
C10W
.091
.078
.081
.868
.754
.554
C10W
.090
.082
.084
.777
.697
.552
.035
.037
.046
.574
.491
.329
C05W
.059
.076
.946
.795
C05W
.054
.052
.065
.858
.704
.518
C05W
.045
.038
.046
.762
.610
.339
C05W
.041
.039
.037
.628
.511
.332
.002
.005
.011
.275
.195
.099
C01W
.031
.042
.739
.540
C01W
.015
.026
.030
.568
.408
.243
C01W
.015
.016
.036
.427
.279
.140
C01W
.014
.009
.016
.306
.211
.109
.108
.094
.029
.722
.560
.155
LfW
.107
.178
.983
.965
LfW
.126
.174
.312
.939
.930
.877
LfW
.112
.153
.371
.889
.857
.683
LfW
.122
.176
.327
.811
.820
.696
.771
.705
.356
.987
.951
.507
LoW
.797
.850
.976
.995
LoW
.871
.925
.947
.982
.992
.989
LoW
.898
.956
.906
.985
.993
.913
LoW
.929
.958
.920
.978
.995
.936
1-8
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=4 N=4 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
7 n
00
50
.90
.00
50
.90
116 279 220 154 082 038 008 091 834 242 376 295 196 090 033 008 219
140 308 239 166 083 042 009 134 934 236 362 287 201 097 042 009 239
.260 .371 .289 .188 .083 .038 .013 .315 .806 .291 .403 .304 .189 .080 .032 .010 .330
.749 .979 .974 .952 .902 .829 .517 .909 .970 .712 .808 .763 .697 .573 .406 .125 .712
753 928 901 857 756 629 306 853 988 736 806 769 714 582 371 125 720
.784 .780 .744 .675 .510 .278 .074 .574 .826 .786 .782 .732 .671 .495 .255 .067 .569
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=4 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
7 n
00
50
.90
.00
50
.90
125 263 210 148 079 038 007 090 849 277 358 291 202 096 040 007 226
162 303 240 173 085 039 008 146 945 251 339 261 182 082 034 007 205
.297 .371 .293 .196 .070 .024 .007 .276 .823 .307 .376 .288 .198 .076 .033 .009 .266
.711 .966 .953 .929 .856 .768 .441 .865 .959 .680 .751 .714 .655 .512 .330 .086 .664
737 907 871 825 728 559 238 814 990 699 753 704 640 519 319 075 657
.721 .718 .684 .626 .442 .244 .056 .544 .814 .716 .720 .686 .627 .458 .236 .047 .543
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=6 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
2.0
7 n
00
.50
.00
.50
048
.047
.928
.964
336 262 185 092 039 013 087 315 042 390 305 203 103 056 022 122
.401 .309 .209 .120 .060 .028 .205 .685 .024 .383 .308 .214 .106 .062 .030 .210
1.00 1.00 1.00 1.00 .999 .993 1.00 .965 .961 1.00 1.00 1.00 .997 .994 .918 .999
.999 .998 .995 .981 .947 .803 .997 .992 .966 .998 .996 .988 .973 .941 .783 .996
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=4 N=6 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
7 n
00
50
.75
.00
50
.75
091 368 288 193 105 058 020 111 481 106 369 285 187 089 044 013 129
089 378 293 205 108 049 011 174 805 112 382 293 206 118 073 019 231
.075 .390 .298 .209 .114 .066 .026 .457 .944 .090 .398 .311 .207 .101 .054 .022 .434
.906 1.00 1.00 1.00 .998 .996 .966 .999 .963 .930 .991 .984 .978 .960 .913 .693 .966
937 994 992 982 965 916 715 995 992 943 985 979 966 916 852 594 972
.952 .967 .951 .924 .848 .753 .429 .941 .998 .960 .973 .960 .942 .879 .770 .431 .953
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=6 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.134 .347 .274 .187 .109 .062 .009 .119 .554 .200 .385 .292 .206 .099 .040 .010 .183
.186 .383 .293 .210 .097 .048 .010 .176 .844 .212 .385 .294 .195 .095 .042 .008 .234
.175 .363 .264 .163 .087 .050 .023 .406 .970 .178 .389 .292 .204 .091 .043 .014 .435
.889 .997 .997 .997 .990 .983 .891 .993 .968 .907 .975 .957 .929 .885 .805 .514 .923
.911 .989 .983 .963 .936 .887 .642 .977 .997 .917 .963 .950 .923 .860 .771 .455 .941
.980 .941 .913 .854 .710 .534 .204 .837 .984 .953 .913 .885 .834 .685 .517 .186 .818
.975
.978
.822
.992
.994
.826
LOG
.964
.957
.800
.994
.981
.833
LOG
.604
.741
.986
.998
LOG
.843
.900
.951
.992
.997
.999
LOG
.906
.958
.975
.994
.997
.972
.171
.183
.284
.728
.750
.773
MxW
.174
.233
.307
.709
.690
.695
MxW
.043
.027
.940
.962
MxW
.113
.122
.088
.923
.947
.964
MxW
.158
.198
.196
.893
.926
.974
.303
.327
.376
.903
.871
.764
C40W
.302
.336
.382
.859
.798
.696
C40W
.391
.407
1.00
.999
C40W
.364
.390
.409
.997
.996
.971
C40W
.336
.376
.418
.993
.976
.944
.246
.255
.278
.871
.828
.730
C30W
.236
.286
.297
.832
.765
.660
C30W
.293
.306
1.00
.998
C30W
.278
.300
.303
.996
.987
.960
C30W
.268
.299
.308
.987
.967
.914
.174
.168
.173
.834
.770
.665
C20W
.173
.198
.182
.780
.699
.603
C20W
.196
.212
1.00
.995
C20W
.182
.198
.194
.991
.977
.931
C20W
.188
.203
.203
.976
.948
.869
.082
.083
.078
.715
.642
.488
C10W
.087
.108
.066
.671
.573
.445
C10W
.088
.111
1.00
.985
C10W
.091
.100
.100
.977
.946
.851
C10W
.096
.093
.107
.941
.898
.744
.038
.038
.029
.568
.488
.251
C05W
.037
.043
.029
.519
.410
.239
C05W
.045
.064
.993
.951
C05W
.046
.053
.051
.960
.884
.741
C05W
.053
.049
.048
.895
.810
.565
.008
.010
.009
.241
.193
.050
C01W
.009
.010
.009
.171
.134
.054
C01W
.018
.021
.931
.781
C01W
.008
.018
.022
.810
.642
.408
C01W
.006
.007
.019
.658
.531
.204
.126
.165
.295
.770
.757
.574
LfW
.131
.192
.265
.729
.713
.539
LfW
.098
.213
.998
.999
LfW
.111
.214
.446
.978
.984
.941
LfW
.133
.212
.445
.952
.956
.854
.932
.980
.803
.983
.996
.824
LoW
.938
.969
.813
.984
.992
.800
LoW
.572
.769
.981
.995
LoW
.738
.899
.960
.978
.996
1.00
LoW
.805
.938
.968
.981
.999
.983
1-9
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further.
MU MIX | MxL
MU
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=4 N=6 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
159 305 235 161 082 037 006 089 639 274 384 306 212 098 053 014 230
201 350 272 183 093 050 006 175 888 283 373 293 211 105 040 010 258
.270 .399 .307 .204 .102 .045 .012 .394 .972 .290 .366 .281 .182 .071 .033 .007 .338
.862 .998 .995 .994 .988 .965 .797 .988 .957 .867 .923 .901 .871 .791 .668 .328 .863
887 978 972 948 897 814 528 944 992 886 938 914 874 774 631 315 885
.924 .901 .878 .812 .668 .523 .202 .791 .978 .917 .896 .863 .805 .684 .526 .195 .774
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=6 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
182 332 260 195 110 048 007 122 687 316 359 280 193 106 051 008 221
194 308 249 182 104 050 005 158 882 315 355 283 194 085 032 008 238
.372 .378 .311 .199 .092 .041 .005 .298 .912 .387 .407 .308 .204 .090 .037 .007 .314
.852 .993 .989 .982 .970 .933 .738 .970 .959 .843 .884 .851 .799 .692 .531 .234 .808
862 968 956 929 862 774 469 937 996 886 896 867 824 722 555 212 837
.885 .844 .796 .735 .582 .408 .099 .640 .910 .880 .841 .813 .743 .584 .397 .102 .645
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=6 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
185 305 248 184 101 042 006 109 689 366 371 295 200 103 040 005 230
.208 .318 .239 .175 .087 .035 .008 .157 .899 .355 .380 .307 .210 .102 .047 .007 .258
.395 .390 .301 .200 .100 .041 .005 .289 .916 .402 .366 .298 .196 .082 .033 .005 .255
856 991 991 985 962 921 699 969 967 822 839 805 750 650 496 174 755
856 956 941 910 839 743 398 915 990 844 850 820 777 655 495 170 768
.870 .820 .774 .707 .557 .404 .113 .625 .916 .865 .828 .786 .721 .577 .411 .111 .621
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=9 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.093
.076
.981
.994
.390 .311
.400 .311
1 .00 1 .00
1 .00 1 .00
.214 .124
.202 .110
1 .00 1 .00
1 .00 .999
.071 .016 .126 .106 .074 .416 .317 .215 .112 .063 .017 .141
.069 .017 .249 .575 .047 .411 .313 .224 .116 .063 .022 .287
1.00 1.00 1.00 .963 .988 1.00 1.00 1.00 .998 .998 .990 .999
.997 .962 1.00 .999 .993 1.00 1.00 1.00 .997 .992 .951 1.00
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=9 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.75
.00
.50
.75
.153 .364 .281 .201 .111 .053 .009 .109 .239 .179 .390 .290 .206 .095 .053 .014 .165
.156 .365 .267 .184 .101 .049 .011 .213 .746 .160 .387 .288 .209 .114 .055 .015 .292
.130 .398 .312 .204 .094 .046 .011 .591 .983 .120 .386 .287 .190 .101 .047 .019 .595
.968 1.00 1.00 1.00 1.00 1.00 1.00 1.00 .954 .982 1.00 1.00 .996 .988 .972 .917 .995
.989 .999 .998 .996 .991 .982 .924 .999 .998 .990 .995 .995 .992 .978 .956 .857 .997
.991 .991 .986 .982 .956 .911 .698 .984 1.00 .992 .994 .989 .977 .946 .897 .697 .990
.977
.983
.969
.994
.997
.979
LOG
.984
.986
.934
.994
.998
.913
LOG
.975
.980
.895
.992
.994
.935
LOG
.376
.663
.989
.998
LOG
.744
.879
.973
.995
.998
1.00
.224
.236
.308
.875
.887
.923
MxW
.254
.292
.376
.854
.865
.882
MxW
.272
.302
.396
.835
.828
.830
MxW
.068
.040
.988
.990
MxW
.188
.177
.118
.978
.990
.993
.351
.339
.386
.983
.942
.898
C40W
.354
.352
.366
.953
.917
.832
C40W
.325
.342
.368
.931
.889
.791
C40W
.379
.428
1.00
1.00
C40W
.384
.369
.400
1.00
.997
.996
.273
.272
.308
.969
.921
.872
C30W
.279
.281
.273
.939
.898
.794
C30W
.245
.275
.280
.908
.855
.754
C30W
.290
.329
1.00
1.00
C30W
.311
.288
.296
.999
.997
.993
.185
.187
.197
.950
.891
.822
C20W
.207
.199
.174
.914
.860
.714
C20W
.183
.191
.180
.871
.803
.673
C20W
.199
.220
1.00
.999
C20W
.218
.190
.185
.999
.993
.988
.083
.097
.087
.908
.817
.700
C10W
.104
.115
.081
.862
.770
.572
C10W
.095
.084
.083
.789
.698
.540
C10W
.109
.113
1.00
.997
C10W
.093
.093
.084
.996
.981
.967
.046
.052
.040
.835
.716
.544
C05W
.046
.048
.036
.764
.632
.394
C05W
.040
.039
.031
.680
.583
.369
C05W
.056
.061
.999
.995
C05W
.055
.045
.046
.991
.966
.911
.010
.005
.012
.533
.393
.198
C01W
.011
.010
.006
.434
.288
.099
C01W
.004
.004
.005
.359
.259
.105
C01W
.012
.022
.997
.955
C01W
.014
.010
.010
.969
.877
.683
.125
.209
.360
.924
.894
.786
LfW
.168
.215
.258
.883
.868
.613
LfW
.145
.181
.248
.829
.819
.570
LfW
.120
.269
1.00
1.00
LfW
.125
.247
.573
.997
.997
.992
.847
.949
.963
.984
.995
.980
LoW
.896
.958
.915
.981
.997
.917
LoW
.916
.969
.871
.987
.999
.922
LoW
.287
.671
.987
.998
LoW
.571
.868
.982
.988
1.00
1.00
1-10
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=4 N=9 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
216
217
.253
.964
967
.995
348 264 180 106 051 008 109 341 278 360 270 190 090 036 003 206
368 277 200 115 067 016 216 818 280 402 296 196 092 053 017 319
.408 .302 .200 .094 .042 .008 .502 .989 .247 .383 .282 .186 .089 .046 .013 .497
1.00 1.00 1.00 1.00 .999 .991 1.00 .958 .963 .987 .979 .970 .951 .909 .753 .973
1 00 997 995 979 962 871 997 994 974 987 979 967 941 891 717 977
.977 .956 .930 .851 .746 .443 .885 .998 .992 .967 .948 .914 .841 .743 .447 .874
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=9 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.85
.00
50
.85
236 349 263 168 096 048 008 100 426 355 360 287 208 097 056 009 274
268 360 273 183 098 045 008 196 851 364 365 269 178 088 040 004 301
.380 .384 .305 .214 .110 .066 .013 .432 .989 .415 .376 .286 .177 .092 .039 .009 .406
.957 1.00 1.00 1.00 .999 .998 .960 .999 .962 .963 .981 .966 .944 .889 .827 .561 .948
966 996 993 984 964 926 781 992 993 962 961 952 932 869 792 551 953
.981 .954 .935 .905 .825 .725 .417 .879 1.00 .975 .954 .938 .901 .806 .698 .382 .868
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=9 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
253 321 266 205 111 051 012 131 471 431 366 294 209 113 048 011 281
.302 .360 .270 .181 .102 .044 .010 .194 .888 .436 .385 .302 .218 .116 .064 .011 .310
.487 .394 .304 .198 .092 .044 .004 .308 .947 .498 .372 .280 .179 .091 .040 .007 .309
930 1 00 1 00 998 989 981 917 989 951 943 946 928 905 826 734 460 914
964 992 987 978 944 909 731 983 992 933 934 912 879 797 703 411 903
.967 .916 .888 .836 .717 .584 .282 .689 .975 .972 .919 .878 .836 .712 .554 .242 .684
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=9 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.287 .344 .266 .196 .104 .048 .006 .103 .557 .476 .374 .298 .207 .103 .047 .008 .267
.274 .311 .240 .166 .083 .042 .004 .163 .881 .484 .378 .310 .222 .109 .051 .008 .276
510 377 301 212 113 058 007 309 930 578 406 307 202 105 044 007 292
958 998 998 997 991 971 897 991 964 925 901 879 834 745 620 332 839
.938 .982 .972 .958 .922 .884 .664 .970 .993 .920 .893 .873 .838 .744 .614 .326 .842
.960 .894 .872 .813 .702 .553 .239 .685 .979 .936 .865 .839 .788 .674 .541 .236 .682
C40L C30L C20L C10L C05L C01L
C=4 N=12 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.118
.066
.998
1.00
.381
.404
1.00
1.00
.300
.295
1.00
1.00
.209
.186
1.00
1.00
.107
.092
1.00
1.00
.055
.049
1.00
.999
.014
.016
1.00
.994
.108
.302
1.00
1.00
.030
.586
.974
.999
.087 .412 .311 .192 .107 .055 .013 .134
.062 .406 .300 .201 .107 .053 .012 .343
1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 1.00 1.00 1.00 .999 .992 1.00
.912
.968
.984
.993
1.00
.998
LOG
.963
.988
.979
.999
.997
.997
LOG
.979
.982
.927
1.00
.999
.987
LOG
.969
.972
.925
.996
.998
.972
LOG
.231
.669
.995
.998
.250
.268
.247
.970
.971
.991
MxW
.301
.342
.416
.957
.962
.965
MxW
.329
.371
.527
.957
.945
.963
MxW
.362
.400
.538
.937
.927
.937
MxW
.088
.048
.998
.999
.373
.374
.404
.999
.989
.971
C40W
.348
.366
.401
.993
.980
.935
C40W
.335
.377
.410
.986
.960
.899
C40W
.337
.355
.390
.978
.938
.876
C40W
.397
.392
1.00
1.00
.307
.297
.313
.997
.985
.954
C30W
.274
.295
.316
.991
.970
.918
C30W
.265
.289
.305
.983
.941
.873
C30W
.268
.278
.298
.973
.919
.845
C30W
.308
.296
1.00
1.00
.221
.192
.204
.994
.979
.931
C20W
.186
.207
.211
.982
.955
.886
C20W
.190
.192
.207
.969
.921
.804
C20W
.187
.205
.210
.959
.885
.788
C20W
.212
.203
1.00
1.00
.115
.088
.095
.986
.962
.869
C10W
.094
.107
.085
.956
.925
.796
C10W
.095
.105
.112
.932
.877
.680
C10W
.086
.109
.112
.912
.832
.681
C10W
.115
.113
1.00
1.00
.050
.040
.046
.965
.925
.760
C05W
.047
.046
.041
.921
.866
.684
C05W
.044
.048
.062
.884
.805
.547
C05W
.038
.050
.056
.837
.742
.531
C05W
.066
.062
1.00
.999
.008
.010
.008
.895
.768
.447
C01W
.004
.008
.012
.769
.636
.379
C01W
.007
.005
.009
.676
.521
.254
C01W
.009
.009
.003
.584
.474
.244
C01W
.013
.018
.999
.990
.167
.246
.529
.989
.985
.891
LfW
.146
.254
.428
.964
.964
.858
LfW
.158
.253
.332
.943
.938
.656
LfW
.165
.237
.276
.934
.906
.671
LfW
.132
.326
1.00
1.00
.716
.907
.988
.981
.998
.995
LoW
.776
.947
.972
.989
.999
.998
LoW
.825
.953
.943
.987
1.00
.975
LoW
.889
.962
.934
.986
1.00
.970
LoW
.129
.672
.987
1.00
1-11
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=4 N=12 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.75
.00
.50
.75
185
214
.172
.992
.994
.999
342
374
.422
1.00
1.00
.999
266
306
.319
1.00
1.00
.997
185 083 050 010 092 095 248 401 320 202 101 055 016 199
218 126 059 014 303 692 210 402 314 208 108 051 015 358
.219 .116 .055 .016 .739 .988 .136 .375 .286 .189 .102 .055 .016 .699
1.00 1.00 1.00 1.00 1.00 .964 .998 1.00 1.00 1.00 .999 .999 .978 1.00
1.00 .999 .997 .983 1.00 .996 .992 1.00 1.00 1.00 .994 .982 .926 1.00
.994 .987 .971 .865 .995 1.00 .998 .999 .998 .996 .989 .976 .875 .997
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=12 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
250
332
.309
.989
992
.999
368 269 193 109 057 009 112 184 354 372 277 184 084 036 007 252
389 299 203 111 063 014 245 764 346 368 280 186 097 056 012 367
.380 .286 .180 .102 .053 .016 .579 .997 .333 .385 .288 .211 .105 .046 .011 .577
1.00 1.00 1.00 1.00 1.00 .999 1.00 .962 .992 .998 .996 .994 .982 .959 .878 .993
1 00 1 00 999 995 992 963 999 995 991 996 992 984 969 954 860 994
.989 .978 .960 .918 .860 .631 .930 .997 1.00 .987 .977 .960 .909 .855 .607 .917
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=12 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2.0
2.0
? 0
00
.50
.85
.00
.50
.85
290 322 257 191 089 041 008 101 261 457 392 304 207 109 049 009 306
.358 .370 .294 .194 .106 .047 .010 .242 .821 .470 .399 .308 .217 .099 .042 .010 .377
.487 .404 .300 .218 .123 .060 .011 .506 .987 .504 .383 .294 .196 .093 .045 .010 .465
.984 1.00 1.00 1.00 1.00 .999 .991 1.00 .961 .989 .983 .980 .976 .949 .897 .744 .979
.989 .999 .999 .997 .992 .987 .922 1.00 .995 .984 .981 .977 .967 .935 .877 .683 .978
.991 .975 .964 .938 .892 .823 .586 .923 .999 .987 .972 .963 .935 .883 .813 .557 .906
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=12 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2.0
2.0
2.0
.00
.50
90
.00
.50
.90
.311 .325 .253 .183 .092 .053 .010 .095 .323 .520 .363 .276 .187 .093 .039 .008 .310
.398 .365 .285 .197 .094 .046 .006 .231 .849 .525 .373 .280 .198 .108 .047 .005 .325
597 399 309 191 094 048 007 359 954 651 401 302 203 094 045 007 327
.977 1.00 1.00 1.00 1.00 .997 .978 1.00 .959 .984 .975 .966 .948 .908 .838 .596 .961
.985 .998 .997 .990 .979 .964 .869 .996 .997 .985 .974 .962 .939 .882 .818 .588 .952
.986 .955 .939 .906 .818 .722 .429 .791 .986 .981 .942 .920 .876 .791 .695 .372 .759
.660
.879
.991
.999
1.00
1.00
LOG
.911
.968
.991
1.00
1.00
.998
LOG
.964
.990
.981
1.00
1.00
.995
LOG
.983
.979
.944
1.00
.999
.974
.228
.214
.160
.998
.996
.999
MxW
.318
.353
.320
.992
.993
.999
MxW
.373
.441
.491
.984
.985
.993
MxW
.431
.486
.619
.975
.978
.988
.385
.407
.437
1.00
1.00
.996
C40W
.361
.385
.408
1.00
.998
.989
C40W
.358
.367
.420
.998
.995
.977
C40W
.341
.374
.400
.997
.984
.959
.287
.294
.326
1.00
1.00
.995
C30W
.275
.296
.308
1.00
.997
.983
C30W
.284
.276
.314
.998
.990
.966
C30W
.267
.308
.316
.993
.974
.931
.201
.188
.213
1.00
.998
.994
C20W
.195
.212
.202
1.00
.994
.963
C20W
.197
.188
.219
.996
.980
.938
C20W
.194
.214
.205
.990
.964
.889
.108
.094
.102
1.00
.997
.984
C10W
.093
.114
.088
.998
.980
.926
C10W
.112
.099
.116
.986
.960
.881
C10W
.102
.112
.088
.972
.936
.805
.052
.049
.048
.998
.990
.967
C05W
.050
.056
.045
.994
.965
.871
C05W
.050
.056
.051
.974
.920
.806
C05W
.054
.051
.041
.953
.888
.692
.006
.009
.013
.992
.957
.871
C01W
.012
.009
.007
.971
.878
.656
C01W
.006
.012
.007
.897
.786
.561
C01W
.010
.010
.007
.833
.698
.410
.143
.321
.739
1.00
1.00
.994
LfW
.143
.324
.566
1.00
.996
.936
LfW
.181
.292
.494
.994
.995
.905
LfW
.169
.301
.369
.985
.977
.766
.396
.852
.987
.990
.998
1.00
LoW
.598
.921
.996
.992
1.00
.998
LoW
.735
.960
.990
.992
1.00
.998
LoW
.809
.966
.955
.986
1.00
.988
1-12
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
C40L C30L C20L C10L C05L C01L
C=4 N=12 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
339 321 258 190 099 050 003 102 376 536 353 284 204 094 046 008 287
398 331 252 173 096 046 005 203 852 555 378 299 198 088 049 010 297
.626 .397 .310 .212 .115 .057 .010 .321 .941 .644 .393 .292 .196 .094 .046 .006 .275
.978 1.00 .999 .999 .998 .991 .959 .999 .949 .967 .948 .937 .909 .850 .770 .492 .924
979 996 996 989 975 948 837 996 996 975 948 930 896 832 746 468 911
.978 .926 .892 .850 .770 .654 .364 .740 .981 .972 .930 .893 .844 .740 .629 .361 .708
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=16 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
2.0
2.0
00
50
.00
.50
152
096
1.00
1.00
406
383
1.00
1.00
317
287
1.00
1.00
214
207
1.00
1.00
126
105
1.00
1.00
061
053
1.00
1.00
020
015
1.00
1.00
129
397
1.00
1.00
005
644
.964
.998
112 372 288 199 099 051 013 126
073 415 327 217 116 066 014 430
.998 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 1.00 1.00 1.00 1.00 .999 1.00
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.978
.985
.924
1.00
.999
.977
LOG
.106
.672
.994
.999
.466
.498
.681
.973
.968
.969
MxW
.126
.049
1.00
.999
.365
.369
.399
.991
.971
.911
C40W
.417
.389
1.00
1.00
.288
.290
.306
.982
.958
.883
C30W
.315
.283
1.00
1.00
.201
.194
.213
.973
.935
.845
C20W
.222
.193
1.00
1.00
.103
.096
.125
.940
.892
.735
C10W
.119
.102
1.00
1.00
.046
.052
.045
.907
.811
.644
C05W
.065
.051
1.00
1.00
.007
.007
.008
.731
.587
.321
C01W
.014
.012
1.00
.998
.190
.270
.280
.959
.956
.706
LfW
.138
.415
1.00
1.00
.838
.967
.937
.992
1.00
.973
LoW
.036
.699
.995
1.00
C40L C30L C20L C10L C05L C01L
C=4 N=16 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0.5
0 5
0 5
2.0
2.0
2.0
.00
50
75
.00
.50
.75
.246
247
204
.998
1.00
1.00
.370
392
400
1.00
1.00
1.00
.269
310
305
1.00
1.00
.999
.190
210
202
1.00
1.00
.998
.098
112
091
1.00
1.00
.996
.051 .012
056 014
053 012
1.00 1.00
1.00 .996
.990 .964
.101 .025
340 664
851 998
1.00 .969
1.00 .998
.999 1.00
.285 .382 .288 .200 .110 .059 .016 .251 .602 .285 .378 .303 .210 .101 .056 .011
272 398 306 213 115 058 012 438 876 278 369 278 199 100 052 011
207 402 296 205 099 051 017 832 999 196 387 298 197 101 048 015
1.00 1.00 1.00 1.00 1.00 1.00 .995 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.998 1.00 1.00 1.00 1.00 .998 .988 1.00 1.00 .999 1.00 .999 .999 .999 .998 .985
1.00 .999 .999 .998 .996 .989 .952 .999 1.00 1.00 .999 .998 .995 .990 .984 .941
.150
400
843
1.00
1.00
.998
.252
867
1 00
.994
.999
1.00
MU MIX | MxL
MU
C40L C30L C20L C10L C05L C01L
C=4 N=16 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.85
.00
.50
.85
313
401
.436
.997
1.00
1.00
350
400
.408
1.00
1.00
.998
272
309
.306
1.00
1.00
.996
193 100 051 007 104 080 465 387 293 202 102 055 010 310
212 100 051 008 319 778 464 393 293 187 089 044 004 466
.213 .112 .058 .012 .660 .997 .410 .404 .316 .213 .107 .054 .017 .665
1.00 1.00 1.00 1.00 1.00 .961 .997 1.00 1.00 1.00 .999 .993 .957 1.00
1.00 .999 .997 .985 1.00 .999 .998 .999 .998 .997 .987 .981 .944 1.00
.993 .969 .940 .807 .968 1.00 1.00 .994 .990 .985 .972 .938 .794 .971
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=16 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.342
.463
.604
.993
.999
1.00
.344 .288 .206 .103 .052 .007 .112 .161 .576 .396 .305 .199 .107 .049 .004 .377
.381 .287 .203 .110 .054 .014 .279 .809 .575 .372 .289 .202 .109 .052 .010 .420
.434 .322 .213 .115 .059 .010 .613 .992 .602 .402 .310 .212 .108 .052 .008 .540
1.00 1.00 1.00 1.00 1.00 .999 1.00 .962 .995 .996 .991 .988 .972 .949 .866 .995
1.00 .999 .999 .997 .996 .978 1.00 .999 .997 .997 .996 .993 .979 .953 .854 .997
.993 .988 .977 .951 .906 .749 .958 .999 1.00 .988 .984 .965 .934 .889 .727 .954
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.893
.975
.999
1.00
1.00
1.00
LOG
.974
.989
.991
1.00
1.00
.999
.411
.463
.413
.998
.998
1.00
MxW
.497
.538
.600
.996
.998
1.00
.387
.408
.383
1.00
1.00
.998
C40W
.360
.393
.409
1.00
.998
.991
.293
.311
.287
1.00
.998
.994
C30W
.273
.306
.302
.999
.998
.984
.211
.196
.204
1.00
.998
.990
C20W
.175
.202
.198
.999
.994
.971
.104
.105
.102
1.00
.993
.976
C10W
.088
.098
.114
.999
.983
.932
.048
.052
.052
.999
.991
.941
C05W
.048
.045
.055
.993
.970
.881
.009
.009
.009
.991
.959
.797
C01W
.011
.004
.011
.970
.906
.716
.159
.403
.615
1.00
1.00
.972
LfW
.160
.376
.532
.998
.999
.946
.477
.920
.995
.995
.999
1.00
LoW
.660
.959
.989
.995
1.00
1.00
1-13
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=4 N=16 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
7 n
00
50
.90
.00
50
.90
392 324 249 173 088 040 015 097 195 617 399 301 189 080 042 009 374
452 355 272 189 091 047 010 261 831 663 396 298 214 094 040 009 410
.709 .409 .319 .210 .106 .046 .006 .396 .965 .745 .428 .323 .229 .117 .063 .013 .401
.993 1.00 1.00 1.00 1.00 1.00 .993 1.00 .961 .994 .990 .982 .970 .949 .901 .741 .985
997 999 999 998 992 991 961 999 999 992 986 981 963 946 903 722 983
.998 .972 .961 .937 .873 .799 .569 .832 .993 .997 .975 .957 .930 .870 .790 .566 .811
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=4 N=16 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
7 n
00
50
.90
.00
.50
.90
437 321 251 181 094 040 008 106 245 673 368 293 200 111 060 014 331
489 326 230 173 092 039 008 221 859 677 364 279 198 103 047 010 329
.700 .381 .299 .206 .108 .045 .012 .337 .947 .786 .395 .304 .207 .114 .055 .005 .338
.992 1.00 1.00 1.00 1.00 .998 .988 1.00 .962 .991 .974 .965 .942 .913 .857 .606 .962
.996 .999 .998 .994 .990 .973 .918 .999 1.00 .996 .979 .972 .957 .908 .838 .619 .960
.992 .954 .938 .911 .841 .761 .506 .805 .982 .991 .954 .929 .892 .820 .715 .459 .758
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6N=4CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
2.0
7 n
00
.50
.00
.50
028 361 278 190 097 059 024 089 572 012 368 280 196 110 074 035 109
.004 .379 .296 .198 .116 .076 .039 .120 .738 .003 .364 .283 .204 .124 .089 .044 .140
.861 1.00 1.00 1.00 1.00 1.00 .958 1.00 .975 .889 1.00 1.00 .998 .998 .981 .823 .997
.890 .999 .999 .993 .976 .922 .783 .997 .994 .893 .998 .995 .988 .969 .926 .729 .991
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=6 N=4 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
7 n
00
50
.75
.00
50
.75
033 314 237 140 075 037 012 074 688 035 377 284 198 113 072 028 134
045 370 280 186 099 071 029 127 828 031 375 289 205 123 082 039 161
.032 .396 .314 .213 .129 .086 .044 .278 .898 .022 .407 .311 .223 .132 .082 .043 .267
.819 1.00 1.00 1.00 1.00 .995 .884 .999 .952 .870 .997 .995 .989 .956 .878 .605 .971
839 998 992 986 945 884 658 973 988 859 988 983 966 915 814 535 960
.892 .982 .964 .919 .815 .682 .444 .968 .997 .892 .966 .957 .934 .826 .679 .445 .960
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=4 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.065 .321 .238 .172 .086 .050 .012 .087 .746 .086 .383 .282 .179 .092 .051 .022 .141
.070 .356 .284 .187 .099 .053 .025 .116 .858 .079 .378 .277 .210 .115 .060 .022 .185
.067 .410 .315 .209 .135 .099 .059 .419 .934 .048 .394 .302 .215 .129 .088 .043 .404
.786 1.00 1.00 1.00 .994 .973 .796 .995 .965 .832 .974 .962 .931 .863 .740 .435 .909
.832 .991 .985 .973 .923 .844 .572 .961 .984 .836 .965 .957 .919 .830 .693 .384 .906
.913 .935 .897 .843 .661 .492 .233 .854 .989 .918 .935 .908 .856 .678 .509 .257 .856
.990
.988
.967
1.00
1.00
.993
LOG
.988
.979
.935
1.00
1.00
.986
LOG
.719
.765
.982
.986
LOG
.881
.870
.907
.992
.992
.995
LOG
.920
.936
.932
.995
.997
.980
.514
.570
.749
.998
.993
1.00
MxW
.516
.593
.786
.992
.995
.992
MxW
.013
.003
.879
.895
MxW
.049
.033
.014
.836
.853
.895
MxW
.099
.087
.064
.829
.833
.906
.356
.381
.420
.999
.995
.978
C40W
.318
.372
.408
.998
.993
.944
C40W
.385
.414
1.00
.995
C40W
.358
.389
.373
.998
.990
.980
C40W
.348
.382
.410
.989
.967
.931
.286
.283
.315
.999
.992
.969
C30W
.243
.295
.325
.997
.988
.927
C30W
.310
.317
1.00
.995
C30W
.275
.289
.278
.992
.985
.971
C30W
.280
.297
.292
.986
.961
.897
.192
.191
.223
.998
.987
.941
C20W
.176
.203
.219
.992
.978
.892
C20W
.224
.196
.999
.990
C20W
.178
.215
.184
.988
.969
.933
C20W
.195
.190
.188
.975
.933
.843
.092
.101
.113
.994
.966
.889
C10W
.092
.116
.122
.985
.956
.823
C10W
.127
.098
.997
.974
C10W
.091
.113
.109
.978
.917
.844
C10W
.099
.085
.116
.945
.860
.667
.049
.047
.055
.982
.947
.809
C05W
.044
.066
.054
.961
.925
.725
C05W
.073
.066
.979
.912
C05W
.051
.066
.078
.923
.845
.670
C05W
.045
.051
.090
.884
.739
.483
.010
.005
.009
.939
.834
.559
C01W
.006
.010
.012
.862
.752
.440
C01W
.040
.033
.872
.704
C01W
.025
.033
.036
.705
.582
.434
C01W
.023
.022
.060
.573
.446
.228
.173
.339
.374
.996
.996
.831
LfW
.170
.316
.326
.988
.991
.778
LfW
.123
.123
.996
.988
LfW
.100
.140
.261
.977
.963
.971
LfW
.131
.144
.433
.951
.919
.850
.725
.963
.962
.995
1.00
.993
LoW
.778
.975
.941
.995
1.00
.980
LoW
.697
.788
.985
.987
LoW
.831
.862
.911
.977
.991
.994
LoW
.853
.921
.945
.983
.993
.979
1-14
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=6 N=4 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
081 307 244 169 090 041 011 097 766 124 346 259 160 074 038 013 144
113 371 301 210 111 051 016 137 893 145 385 290 195 097 057 022 206
.126 .366 .277 .189 .090 .052 .023 .333 .955 .125 .373 .290 .185 .096 .052 .030 .358
.766 .997 .994 .991 .972 .936 .726 .979 .958 .827 .942 .923 .879 .779 .629 .272 .862
788 977 964 947 874 777 471 933 988 831 928 908 861 759 599 279 860
.847 .894 .857 .815 .658 .445 .190 .843 .981 .860 .912 .882 .810 .654 .440 .184 .850
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=4 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
109 303 241 177 084 034 008 098 787 159 332 269 168 074 033 014 180
120 322 245 166 097 046 013 125 905 163 336 266 175 076 037 006 183
.187 .397 .293 .199 .093 .047 .022 .392 .912 .179 .402 .276 .175 .063 .029 .013 .376
.761 .996 .992 .981 .957 .896 .626 .961 .963 .780 .883 .850 .796 .676 .524 .199 .775
799 975 963 939 865 746 445 913 990 801 893 861 812 674 479 180 825
.855 .858 .827 .738 .550 .365 .143 .692 .925 .834 .840 .808 .736 .569 .356 .122 .713
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=4 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
104 296 232 165 093 041 012 107 809 212 337 277 179 087 046 010 209
.118 .278 .222 .149 .091 .048 .012 .124 .892 .227 .365 .286 .201 .093 .042 .009 .234
.226 .361 .282 .179 .076 .031 .014 .337 .923 .248 .383 .273 .174 .079 .037 .012 .324
763 991 988 973 935 864 588 938 971 761 839 806 758 642 453 150 760
745 956 941 901 819 681 357 870 988 731 817 775 714 572 388 129 721
.802 .823 .778 .698 .532 .362 .124 .683 .927 .767 .778 .744 .675 .499 .302 .069 .669
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=6 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.025
.014
.924
.965
.354 .267 .180 .089 .045 .015 .091 .218 .016 .392 .301 .208 .112 .056 .023 .115
.402 .309 .212 .119 .062 .034 .151 .457 .008 .392 .291 .200 .118 .079 .028 .152
1.00 1.00 1.00 1.00 1.00 1.00 1.00 .954 .962 1.00 1.00 1.00 1.00 .998 .976 1.00
1.00 .999 .999 .996 .991 .928 1.00 .996 .963 1.00 .998 .997 .996 .986 .905 .998
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=6 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.75
.00
.50
.75
.065 .342 .265 .184 .091 .048 .015 .093 .346 .067 .392 .297 .212 .111 .055 .022 .138
.057 .356 .266 .182 .094 .052 .013 .125 .626 .055 .409 .308 .200 .107 .055 .016 .166
.035 .387 .302 .198 .105 .062 .023 .359 .857 .033 .378 .285 .214 .110 .067 .026 .379
.934 1.00 1.00 1.00 1.00 .999 .990 1.00 .968 .945 1.00 .997 .995 .990 .971 .853 .991
.956 .998 .998 .998 .994 .981 .861 .999 .992 .958 .998 .996 .993 .978 .945 .792 .990
.965 .995 .989 .974 .952 .897 .650 .995 1.00 .972 .994 .993 .979 .947 .881 .652 .993
.970
.967
.955
.992
.998
.982
LOG
.971
.969
.923
.997
.997
.916
LOG
.970
.974
.909
.996
.992
.911
LOG
.408
.542
.985
.996
LOG
.720
.768
.848
.987
.995
.997
.099
.131
.123
.784
.793
.849
MxW
.144
.153
.172
.739
.794
.832
MxW
.148
.187
.243
.751
.761
.768
MxW
.021
.005
.960
.958
MxW
.075
.062
.026
.940
.956
.963
.318
.375
.388
.977
.939
.893
C40W
.360
.353
.367
.956
.903
.840
C40W
.331
.366
.383
.936
.895
.791
C40W
.387
.397
1.00
.999
C40W
.397
.374
.367
1.00
.997
.993
.241
.295
.283
.964
.913
.853
C30W
.281
.275
.279
.941
.887
.799
C30W
.250
.270
.296
.914
.864
.746
C30W
.294
.305
1.00
.999
C30W
.305
.294
.289
1.00
.995
.985
.172
.184
.190
.941
.888
.797
C20W
.202
.184
.171
.905
.844
.736
C20W
.184
.177
.189
.874
.818
.680
C20W
.190
.219
1.00
.998
C20W
.206
.202
.206
.999
.993
.977
.075
.102
.089
.889
.789
.643
C10W
.120
.084
.065
.815
.744
.524
C10W
.104
.092
.074
.771
.699
.518
C10W
.095
.116
1.00
.994
C10W
.102
.097
.111
.997
.980
.940
.037
.054
.059
.788
.674
.462
C05W
.053
.047
.037
.711
.590
.338
C05W
.044
.041
.034
.639
.543
.323
C05W
.053
.071
.999
.988
C05W
.054
.053
.071
.993
.953
.872
.008
.019
.023
.452
.361
.201
C01W
.009
.012
.016
.365
.275
.089
C01W
.011
.014
.015
.300
.222
.082
C01W
.017
.023
.989
.903
C01W
.016
.014
.030
.916
.792
.618
.117
.167
.346
.918
.873
.825
LfW
.147
.165
.376
.846
.828
.689
LfW
.144
.186
.340
.820
.800
.675
LfW
.098
.140
1.00
.998
LfW
.110
.157
.342
.998
.990
.991
.879
.941
.970
.980
.987
.988
LoW
.913
.961
.923
.978
.990
.923
LoW
.913
.963
.914
.977
.990
.927
LoW
.352
.547
.982
.992
LoW
.581
.770
.864
.979
.997
1.00
1-15
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=6 N=6 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
114 371 299 200 103 056 011 111 468 115 372 286 200 106 052 017 167
109 334 235 171 079 037 009 119 702 142 366 296 208 102 053 018 213
.091 .394 .298 .190 .101 .056 .027 .526 .965 .084 .390 .284 .186 .100 .061 .027 .524
.909 1.00 1.00 1.00 1.00 .997 .964 1.00 .956 .919 .989 .985 .976 .942 .901 .684 .965
919 998 998 998 986 966 842 992 982 936 991 984 972 933 866 630 974
.963 .966 .948 .920 .828 .701 .378 .933 .993 .974 .975 .956 .920 .841 .728 .378 .938
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=6 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
116
145
.174
.890
909
.938
327 256 172 092 051 008 108 532 187 365 285 195 088 041 011 202
351 275 188 097 047 009 136 777 195 395 314 204 101 050 009 232
.378 .300 .204 .094 .052 .017 .424 .979 .180 .364 .287 .188 .090 .052 .015 .441
1.00 .999 .999 .996 .992 .933 .996 .957 .918 .968 .960 .933 .881 .807 .513 .924
1 00 997 988 970 931 744 983 983 909 965 948 925 859 774 479 937
.954 .933 .889 .794 .679 .346 .916 .995 .936 .942 .925 .883 .790 .660 .322 .902
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=6 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
138 319 255 185 101 045 009 107 573 226 353 262 162 077 037 007 200
.169 .323 .259 .180 .095 .049 .007 .137 .798 .250 .362 .284 .201 .099 .039 .007 .259
.261 .375 .297 .200 .100 .051 .011 .416 .977 .270 .401 .303 .200 .088 .039 .011 .424
888 999 998 997 995 991 889 995 964 886 938 922 896 801 700 374 886
897 991 988 976 949 902 690 972 982 907 951 932 896 809 692 367 911
.944 .917 .887 .832 .708 .549 .197 .819 .982 .954 .923 .894 .837 .700 .533 .187 .827
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=6 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.157 .300 .235 .175 .096 .053 .006 .103 .594 .310 .380 .285 .202 .089 .040 .003 .253
.192 .316 .249 .177 .105 .055 .008 .143 .835 .294 .377 .306 .204 .097 .043 .008 .270
310 382 289 193 099 057 012 390 979 338 370 289 193 084 042 008 340
883 998 998 996 990 971 837 991 954 871 915 891 857 753 621 298 866
.866 .980 .969 .948 .906 .848 .582 .940 .987 .881 .901 .876 .830 .731 .591 .245 .851
.906 .878 .857 .789 .671 .518 .183 .780 .979 .892 .872 .835 .783 .664 .500 .176 .771
C40L C30L C20L C10L C05L C01L
C=6 N=9 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.047 .383
.016 .399
.980 1 .00
.994 1 .00
.295
.299
1.00
1.00
.203 .112
.194 .105
1 .00 1 .00
1 .00 1 .00
.066
.054
1.00
1.00
.010 .115 .036 .010 .402 .298 .201 .109 .062 .017 .117
.012 .153 .282 .006 .388 .296 .194 .102 .048 .015 .178
1.00 1.00 .963 .995 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.995 1.00 .992 .989 1.00 1.00 1.00 1.00 1.00 .997 1.00
.848
.919
.954
.993
.998
.998
LOG
.944
.952
.985
.996
1.00
.993
LOG
.965
.964
.985
.997
.996
.989
LOG
.980
.985
.969
.998
.999
.971
LOG
.149
.363
.993
.994
.128
.127
.085
.914
.935
.970
MxW
.198
.192
.186
.911
.918
.937
MxW
.214
.244
.278
.886
.874
.940
MxW
.216
.221
.342
.856
.877
.904
MxW
.026
.007
.987
.996
.352
.366
.418
.997
.993
.963
C40W
.373
.368
.410
.990
.990
.944
C40W
.367
.355
.422
.979
.960
.916
C40W
.361
.322
.388
.972
.953
.883
C40W
.387
.376
1.00
1.00
.268
.269
.324
.994
.991
.943
C30W
.292
.283
.313
.985
.982
.924
C30W
.294
.289
.330
.974
.941
.886
C30W
.280
.250
.288
.961
.936
.854
C30W
.292
.310
1.00
1.00
.193
.199
.227
.990
.978
.915
C20W
.205
.190
.198
.981
.968
.895
C20W
.206
.199
.204
.957
.916
.832
C20W
.186
.171
.184
.942
.902
.802
C20W
.195
.222
1.00
1.00
.102
.099
.113
.980
.952
.831
C10W
.106
.094
.089
.957
.923
.810
C10W
.102
.110
.100
.911
.864
.710
C10W
.085
.076
.067
.886
.834
.671
C10W
.095
.114
1.00
1.00
.052
.048
.065
.959
.915
.718
C05W
.047
.045
.038
.919
.857
.670
C05W
.045
.048
.052
.858
.778
.563
C05W
.034
.033
.027
.808
.734
.502
C05W
.046
.062
1.00
1.00
.015
.019
.026
.828
.700
.366
C01W
.006
.011
.013
.699
.579
.326
C01W
.008
.008
.014
.604
.486
.192
C01W
.005
.006
.006
.516
.405
.180
C01W
.013
.012
1.00
.992
.119
.177
.529
.982
.981
.929
LfW
.138
.186
.469
.965
.970
.916
LfW
.143
.203
.449
.927
.920
.817
LfW
.133
.181
.359
.912
.907
.768
LfW
.098
.170
1.00
1.00
.715
.843
.957
.981
.993
.996
LoW
.798
.925
.988
.987
.994
.999
LoW
.841
.923
.972
.980
.994
.982
LoW
.877
.944
.969
.979
.993
.983
LoW
.111
.381
.976
.997
1-16
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further.
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=6 N=9 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.75
.00
.50
.75
119
088
.057
.980
.987
.991
382
395
.408
1.00
1.00
.998
297
299
.316
1.00
1.00
.996
210 107 061 010 108 140 088 368 276 196 110 067 017 147
198 097 060 013 157 497 065 386 295 196 100 056 011 198
.218 .117 .063 .012 .529 .852 .049 .391 .286 .182 .097 .050 .014 .497
1.00 1.00 1.00 .999 1.00 .973 .988 1.00 1.00 1.00 1.00 .999 .984 1.00
1.00 1.00 .999 .983 1.00 .988 .986 1.00 .999 .999 .993 .989 .949 1.00
.993 .987 .971 .873 1.00 .999 .993 .996 .996 .996 .992 .977 .876 .997
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=9 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.85
.00
.50
.85
136
146
.114
.977
.980
.994
351 272 180 097 050 014 107 225 189 391 287 199 093 045 008 187
362 284 195 110 065 011 167 605 179 392 316 216 111 052 009 273
.363 .290 .199 .099 .055 .016 .658 .992 .136 .368 .292 .189 .098 .054 .009 .660
1.00 1.00 1.00 1.00 1.00 .999 1.00 .969 .986 .997 .996 .995 .992 .983 .912 .994
1.00 1.00 1.00 1.00 .996 .957 1.00 .994 .984 .998 .997 .992 .978 .955 .851 .995
.991 .985 .967 .937 .877 .631 .972 1.00 .992 .991 .987 .975 .943 .885 .669 .982
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=9 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2.0
2.0
? 0
00
.50
.85
.00
.50
.85
199 346 261 188 096 049 008 111 320 286 386 294 195 112 054 008 256
.204 .366 .283 .203 .109 .051 .011 .171 .664 .278 .393 .303 .196 .109 .051 .009 .309
.247 .394 .300 .199 .099 .045 .010 .583 .992 .266 .389 .314 .214 .103 .058 .007 .575
.964 1.00 1.00 1.00 1.00 1.00 .994 1.00 .949 .972 .991 .987 .978 .962 .931 .787 .974
.978 1.00 1.00 .999 .996 .985 .929 1.00 .991 .973 .989 .984 .980 .956 .905 .741 .984
.988 .978 .969 .947 .907 .846 .611 .963 .999 .990 .984 .972 .958 .915 .840 .581 .963
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=9 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2.0
2.0
2.0
.00
.50
90
.00
.50
.90
.224 .368 .289 .202 .092 .046 .004 .099 .380 .369 .402 .305 .198 .112 .064 .018 .289
.239 .354 .283 .191 .109 .054 .007 .180 .731 .342 .374 .288 .189 .091 .048 .006 .306
351 394 299 203 090 035 003 509 984 376 403 306 211 105 057 011 498
.958 1.00 1.00 1.00 1.00 .999 .989 1.00 .959 .970 .986 .976 .958 .914 .852 .633 .967
.958 .996 .992 .990 .985 .965 .889 .991 .989 .972 .980 .971 .954 .911 .838 .601 .970
.985 .966 .951 .911 .827 .710 .402 .885 .996 .990 .973 .958 .925 .845 .743 .427 .894
.534
.657
.843
.996
.997
.999
LOG
.820
.882
.986
.997
.999
1.00
LOG
.933
.955
.996
.995
1.00
.997
LOG
.966
.977
.987
.996
.998
.998
.077
.076
.042
.982
.987
.998
MxW
.180
.176
.133
.974
.985
.995
MxW
.229
.255
.268
.966
.978
.986
MxW
.265
.305
.368
.963
.967
.984
.363
.397
.367
1.00
1.00
.998
C40W
.351
.375
.402
1.00
1.00
.991
C40W
.368
.376
.409
1.00
.996
.979
C40W
.337
.372
.420
.996
.996
.949
.267
.305
.274
1.00
1.00
.996
C30W
.281
.291
.304
1.00
.999
.985
C30W
.279
.285
.323
1.00
.994
.968
C30W
.257
.289
.310
.995
.991
.924
.173
.185
.182
1.00
1.00
.994
C20W
.184
.199
.208
1.00
.995
.975
C20W
.180
.176
.214
1.00
.987
.953
C20W
.174
.202
.208
.992
.984
.884
.090
.094
.097
1.00
.998
.987
C10W
.086
.104
.084
1.00
.990
.942
C10W
.081
.092
.090
.996
.966
.907
C10W
.092
.099
.098
.982
.960
.798
.048
.050
.055
.999
.996
.972
C05W
.047
.057
.050
.997
.981
.886
C05W
.043
.050
.049
.984
.933
.842
C05W
.049
.051
.049
.952
.903
.689
.009
.010
.021
.997
.969
.871
C01W
.006
.012
.011
.958
.897
.648
C01W
.007
.008
.007
.921
.813
.588
C01W
.008
.012
.008
.841
.719
.387
.097
.201
.538
1.00
1.00
1.00
LfW
.110
.201
.670
1.00
.997
.980
LfW
.116
.206
.562
.996
.989
.962
LfW
.131
.233
.497
.988
.988
.847
.323
.656
.887
.987
.994
1.00
LoW
.543
.810
.988
.984
1.00
1.00
LoW
.655
.888
.994
.985
.999
1.00
LoW
.712
.920
.987
.993
.998
.996
1-17
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
C40L C30L C20L C10L C05L C01L
C=6 N=9 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
217 314 241 171 082 036 007 093 393 392 367 299 207 100 044 009 302
267 328 249 185 088 044 005 153 758 394 350 262 184 106 050 008 310
.430 .380 .297 .207 .113 .056 .013 .446 .987 .484 .420 .333 .222 .095 .036 .005 .449
.960 1.00 1.00 .999 .997 .993 .969 .996 .965 .954 .962 .952 .929 .867 .772 .498 .942
963 999 998 995 980 961 844 990 993 957 955 945 927 862 767 509 946
.963 .942 .916 .880 .795 .688 .394 .863 .996 .968 .947 .926 .880 .784 .646 .351 .851
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=12 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
2.0
2.0
00
50
.00
.50
070
029
.997
.998
392
386
1.00
1.00
305
297
1.00
1.00
208 102
202 103
1 .00 1 .00
1 .00 1 .00
052
060
1.00
1.00
014 108 008 020 369 279 177 086 042 008 104
012 178 228 012 404 316 204 120 064 018 193
1.00 1.00 .976 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.999 1.00 .995 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
C40L C30L C20L C10L C05L C01L
C=6 N=12 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
0 5
2.0
2.0
2.0
.00
50
75
.00
.50
.75
.132
124
071
.992
.998
1.00
.393
401
409
1.00
1.00
1.00
.313
313
322
1.00
1.00
1.00
.200 .113
218 123
214 102
1 .00 1 .00
1 .00 1 .00
1 .00 .998
.063 .008 .122 .045 .110 .377 .273 .188 .122 .060 .012 .171
060 011 190 388 089 378 276 195 098 052 017 241
049 010 698 920 071 402 290 196 102 051 015 692
1.00 1.00 1.00 .963 .997 1.00 1.00 1.00 1.00 1.00 .998 1.00
1.00 .999 1.00 .996 1.00 1.00 1.00 1.00 1.00 1.00 .995 1.00
.995 .962 1.00 1.00 1.00 1.00 1.00 1.00 .999 .995 .970 1.00
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=6 N=12 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
? n
00
50
.85
.00
.50
.85
199
222
.163
.989
.993
.999
361 284 199 110 053 012 117 098 251 395 305 222 105 054 012 243
377 279 187 112 054 013 182 502 252 419 330 222 110 068 009 331
.405 .307 .203 .097 .045 .012 .770 .998 .143 .400 .297 .187 .089 .045 .012 .789
1.00 1.00 1.00 1.00 1.00 1.00 1.00 .958 .993 1.00 .999 .999 .997 .991 .962 .998
1.00 .999 .999 .999 .998 .990 .999 .995 1.00 .999 .999 .999 .997 .992 .960 .999
.995 .992 .986 .970 .944 .800 .988 .998 1.00 .993 .991 .985 .970 .940 .807 .984
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=12 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.247 .360 .289 .200 .114 .056 .012 .121 .164 .351 .411 .311 .218 .110 .047 .014 .307
.298 .362 .284 .191 .086 .052 .010 .178 .625 .318 .367 .289 .205 .105 .056 .009 .360
.322 .401 .309 .208 .109 .054 .008 .657 .996 .332 .382 .279 .180 .096 .047 .011 .632
.992 1.00 1.00 1.00 1.00 1.00 .999 1.00 .975 .991 1.00 .998 .993 .986 .972 .904 .999
.989 .999 .998 .997 .997 .995 .984 .999 .991 .995 .999 .997 .994 .983 .966 .875 .996
.994 .992 .990 .987 .958 .925 .771 .983 .999 .999 .993 .985 .975 .949 .911 .761 .981
.977
.989
.974
.998
.999
.993
LOG
.042
.285
.993
.998
LOG
.373
.603
.912
.992
.997
1.00
LOG
.744
.837
.994
.997
.999
1.00
LOG
.894
.945
.999
.999
1.00
1.00
.295
.338
.459
.956
.956
.965
MxW
.030
.007
.995
.999
MxW
.128
.093
.061
.993
.999
.997
MxW
.210
.258
.155
.996
.996
1.00
MxW
.326
.325
.320
.992
.993
.997
.342
.353
.397
.996
.988
.941
C40W
.362
.403
1.00
1.00
C40W
.399
.383
.439
1.00
1.00
1.00
C40W
.371
.390
.393
1.00
1.00
.999
C40W
.385
.368
.388
1.00
.997
.992
.264
.270
.296
.990
.981
.920
C30W
.294
.299
1.00
1.00
C30W
.302
.294
.330
1.00
1.00
1.00
C30W
.289
.303
.301
1.00
1.00
.998
C30W
.303
.266
.291
1.00
.997
.984
.194
.197
.195
.986
.960
.877
C20W
.188
.192
1.00
1.00
C20W
.207
.209
.208
1.00
1.00
1.00
C20W
.212
.228
.208
1.00
.999
.991
C20W
.190
.183
.199
.999
.996
.975
.101
.108
.091
.973
.922
.779
C10W
.098
.104
1.00
1.00
C10W
.093
.101
.115
1.00
1.00
.997
C10W
.104
.122
.095
1.00
.997
.980
C10W
.109
.093
.084
.998
.993
.941
.045
.048
.040
.937
.857
.661
C05W
.046
.058
1.00
1.00
C05W
.053
.052
.066
1.00
1.00
.995
C05W
.044
.063
.056
1.00
.993
.949
C05W
.051
.050
.043
.996
.986
.908
.005
.006
.011
.780
.638
.343
C01W
.010
.017
1.00
1.00
C01W
.017
.014
.017
1.00
.996
.959
C01W
.011
.016
.014
.996
.969
.815
C01W
.008
.006
.008
.981
.938
.760
.160
.236
.417
.978
.962
.849
LfW
.111
.187
1.00
1.00
LfW
.119
.233
.695
1.00
1.00
1.00
LfW
.134
.268
.773
1.00
1.00
.997
LfW
.144
.254
.647
.999
.996
.982
.763
.938
.982
.981
.998
.993
LoW
.030
.299
.986
.997
LoW
.167
.573
.912
.985
.997
1.00
LoW
.358
.756
.995
.987
1.00
1.00
LoW
.550
.864
.999
.989
.999
1.00
1-18
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further.
MU MIX | MxL
MU
C40L C30L C20L C10L C05L C01L
C=6 N=12 CV=3.5
LoL MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
? n
00
50
.90
.00
.50
.90
279
326
.470
.988
.992
.998
327
352
.408
1.00
1.00
.975
269
286
.303
1.00
1.00
.969
189 105 049 009 113 216 421 364 271 188 095 051 011 327
204 109 061 011 186 642 439 392 300 202 105 050 007 376
.204 .102 .051 .009 .559 .991 .458 .384 .302 .209 .103 .050 .009 .543
1.00 1.00 1.00 .999 1.00 .963 .991 .997 .995 .985 .971 .934 .778 .990
1.00 .999 .996 .968 1.00 .993 .992 .992 .988 .984 .970 .936 .796 .989
.953 .885 .818 .570 .933 .995 .997 .981 .972 .936 .887 .807 .563 .914
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=12 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
286 333 254 169 085 035 004 092 242 506 390 295 199 095 047 008 353
313 331 251 171 087 044 012 166 696 550 415 336 231 108 058 011 402
.511 .388 .301 .210 .118 .053 .016 .487 .987 .586 .401 .309 .220 .122 .057 .008 .465
.982 1.00 1.00 1.00 1.00 1.00 .996 1.00 .962 .985 .983 .975 .956 .927 .879 .676 .974
988 999 999 998 995 989 934 998 990 981 978 969 953 910 843 643 975
.993 .974 .962 .939 .878 .791 .559 .913 .999 .991 .965 .941 .919 .858 .762 .498 .899
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.967
.981
.990
1.00
1.00
.999
LOG
.988
.994
.988
1.00
1.00
.996
.366
.386
.449
.977
.993
.996
MxW
.365
.438
.564
.978
.987
.995
.359
.369
.384
1.00
.996
.975
C40W
.322
.371
.408
.997
.994
.972
.277
.275
.286
.999
.995
.965
C30W
.259
.285
.297
.995
.992
.958
.202
.193
.196
.996
.992
.939
C20W
.173
.196
.203
.994
.983
.928
.125
.108
.090
.989
.980
.891
C10W
.081
.099
.089
.985
.969
.861
.071
.051
.049
.984
.969
.812
C05W
.040
.051
.049
.968
.937
.778
.016
.008
.011
.938
.878
.572
C01W
.008
.008
.009
.889
.800
.506
.172
.270
.502
.993
.992
.920
LfW
.152
.273
.463
.992
.989
.901
.631
.899
.988
.989
.997
.999
LoW
.696
.928
.973
.990
.998
1.00
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=6 N=16 CV=1.5
C30G C20G C10G
C05G C01G LfG
LoG | MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0.5
2.0
? 0
00
.50
.00
.50
084
.033
.999
1.00
393 302
.410 .305
1 .00 1 .00
1 .00 1 .00
206 104
.212 .112
1 .00 1 .00
1 .00 1 .00
062 006 111 000 042 398 296 197 100 049 012 121 015
.060 .012 .206 .214 .022 .373 .292 .184 .097 .057 .015 .228 .284
1.00 1.00 1.00 .977 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 .996
1.00 1.00 1.00 .997 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
044
.010
1.00
1.00
396 314
.395 .301
1 .00 1 .00
1 .00 1 .00
213 104
.185 .098
1 .00 1 .00
1 .00 1 .00
057
.036
1.00
1.00
009
.011
1.00
1.00
124 004
.224 .288
1.00 .992
1.00 .999
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
MxG C40G
C=6 N=16 CV=2.0
C30G C20G C10G
C05G C01G
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.75
.00
50
.75
180
168
.086
1.00
1 00
1.00
373 284
406 315
.414 .302
1 .00 1 .00
1 00 1 00
1 .00 1 .00
195 112
216 104
.215 .118
1 .00 1 .00
1 00 1 00
1 .00 1 .00
061 014 117 011 138 403 315 220 107 056 013 190 221
058 017 213 292 134 370 269 198 104 058 014 299 572
.060 .020 .838 .946 .066 .385 .289 .194 .105 .053 .017 .815 .956
1.00 1.00 1.00 .972 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 .999
1 00 1 00 1 00 998 999 1 00 1 00 1 00 1 00 1 00 999 1 00 1 00
1.00 .996 1.00 1.00 1.00 1.00 1.00 1.00 .999 .999 .996 1.00 1.00
182
143
.063
.999
1 00
1.00
394 308
399 297
.409 .315
1 .00 1 .00
1 00 1 00
1 .00 1 .00
201 107
199 103
.218 .094
1 .00 1 .00
1 00 1 00
1 .00 1 .00
060
065
.052
1.00
1 00
1.00
016
013
.016
1.00
1 00
.995
140 066
303 522
.812 .942
1.00 .988
1 00 1 00
1.00 1.00
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=6 N=16 CV=2.5
C30G C20G C10G
C05G C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.255
.251
.196
.996
1.00
.999
.377
.359
.392
1.00
1.00
1.00
.281 .195
.272 .187
.304 .214
1 .00 1 .00
1 .00 1 .00
.999 .999
.097
.087
.107
1.00
1.00
.994
.048 .010 .105 .029 .308 .404 .302 .203 .097 .048 .009 .263
.039 .007 .190 .387 .293 .380 .292 .196 .101 .046 .017 .364
.061 .018 .861 .999 .200 .371 .276 .180 .085 .046 .009 .871
1.00 1.00 1.00 .965 .998 1.00 1.00 1.00 1.00 .999 .994 1.00
1.00 .998 1.00 .994 .999 1.00 1.00 1.00 .999 .996 .985 1.00
.986 .944 .999 1.00 .999 1.00 1.00 .998 .991 .985 .927 .997
LoG | MxW C40W C30W C20W C10W C05W C01W LfW LoW
.666
.839
1.00
.999
1.00
1.00
.304
.291
.225
.999
.998
1.00
.393
.395
.433
1.00
1.00
.999
.316
.297
.330
1.00
1.00
.998
.209
.207
.225
1.00
1.00
.997
.109
.101
.129
1.00
.999
.992
.054
.056
.074
1.00
.999
.978
.009
.013
.015
1.00
.994
.923
.155
.316
.869
1.00
1.00
.999
.219
.717
1.00
.991
.998
1.00
1-19
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=6 N=16 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.85
.00
50
.85
318
338
.418
.999
1 00
.998
369
356
.407
1.00
1 00
.998
287
281
.298
1.00
1 00
.996
1 98 1 05
1 96 098
.205 .107
1 .00 1 .00
1 00 1 00
.993 .985
048 009 112 065 434 394 304 199 100 053 009 374
052 010 209 515 434 408 308 202 093 051 011 448
.052 .012 .772 1.00 .423 .393 .310 .214 .105 .052 .012 .748
1.00 .999 1.00 .964 .998 1.00 .998 .997 .996 .993 .967 1.00
1 00 1 00 1 00 997 999 999 998 998 995 990 955 1 00
.970 .895 .996 1.00 1.00 .997 .995 .991 .983 .957 .872 .995
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=16 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
? n
00
50
.90
.00
.50
.90
347
392
.563
.993
.999
.998
347
360
.390
1.00
1.00
.989
287
286
.275
1.00
1.00
.982
201 112 052 007 126 102 547 405 297 207 112 056 011 417
206 099 060 011 206 569 563 394 297 218 135 071 015 458
.185 .105 .050 .006 .607 .991 .546 .403 .301 .213 .109 .055 .017 .615
1.00 1.00 1.00 .998 1.00 .962 .999 .999 .999 .996 .987 .974 .899 .998
1.00 .999 .997 .989 1.00 .996 .997 .997 .996 .994 .981 .967 .891 .998
.977 .945 .911 .751 .956 .999 .999 .995 .990 .976 .945 .907 .726 .954
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=6 N=16 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2.0
2.0
2.0
00
.50
.90
.00
.50
.90
386 345 267 192 108 067 010 119 152 606 386 289 194 099 060 011 402
.404 .333 .247 .163 .091 .045 .007 .200 .595 .652 .435 .335 .230 .128 .068 .014 .481
.642 .395 .308 .205 .102 .053 .015 .546 .994 .647 .407 .308 .208 .114 .056 .013 .538
.993 1.00 1.00 1.00 1.00 1.00 .999 1.00 .958 .994 .995 .990 .979 .952 .929 .810 .992
1.00 1.00 1.00 1.00 1.00 .999 .990 1.00 1.00 .998 .990 .987 .978 .950 .923 .791 .989
.998 .986 .981 .960 .927 .872 .703 .945 1.00 .996 .982 .973 .962 .922 .860 .668 .947
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9N=4CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
? n
.00
.50
.00
.50
.006 .368 .275 .188 .107 .067 .029 .091 .468 .001 .384 .301 .187 .118 .072 .038 .115
.003 .370 .279 .180 .103 .071 .040 .095 .610 .002 .385 .290 .213 .130 .084 .043 .125
.864 1.00 1.00 1.00 1.00 1.00 .981 1.00 .972 .889 1.00 1.00 1.00 .998 .993 .905 .999
.903 .999 .998 .998 .996 .984 .910 .998 .985 .905 1.00 .999 .999 .995 .980 .869 .997
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=4 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.75
.00
.50
.75
.025
.014
.011
.845
.866
.906
.355 .267 .195 .102 .064 .028 .102 .619 .014 .382 .286 .197 .120 .069 .031 .118
.377 .295 .206 .114 .067 .029 .109 .724 .019 .398 .308 .213 .120 .081 .041 .132
.405 .297 .211 .129 .082 .045 .199 .825 .005 .395 .297 .213 .123 .075 .045 .198
1.00 1.00 1.00 1.00 1.00 .946 1.00 .968 .856 .999 .998 .997 .985 .954 .777 .994
1.00 .999 .999 .990 .967 .828 .995 .987 .869 .997 .995 .993 .982 .934 .723 .990
.993 .983 .975 .932 .856 .662 .994 .991 .903 .993 .989 .974 .924 .828 .644 .989
.896
.954
1.00
1.00
1.00
1.00
LOG
.968
.983
.998
.999
1.00
.998
LOG
.987
.995
.983
1.00
1.00
.997
LOG
.645
.660
.981
.980
LOG
.795
.809
.856
.986
.984
.993
.397
.439
.438
.998
.998
.999
MxW
.437
.491
.539
1.00
.997
.998
MxW
.496
.547
.670
.999
.996
.999
MxW
.003
.001
.881
.886
MxW
.018
.008
.006
.852
.879
.885
.381
.381
.418
1.00
1.00
.999
C40W
.386
.382
.383
1.00
.999
.986
C40W
.366
.390
.407
1.00
.999
.981
C40W
.402
.423
1.00
1.00
C40W
.368
.380
.397
.999
.998
.991
.295
.305
.334
1.00
1.00
.997
C30W
.306
.287
.296
1.00
.998
.980
C30W
.285
.303
.302
1.00
.998
.967
C30W
.292
.336
1.00
.998
C30W
.283
.287
.306
.999
.995
.985
.201
.214
.236
1.00
.999
.991
C20W
.217
.215
.198
1.00
.998
.969
C20W
.204
.217
.215
.999
.998
.956
C20W
.200
.224
1.00
.997
C20W
.204
.211
.199
.998
.989
.974
.112
.103
.115
1.00
.999
.978
C10W
.111
.101
.108
1.00
.995
.944
C10W
.095
.103
.114
.999
.989
.913
C10W
.102
.130
1.00
.985
C10W
.117
.117
.126
.998
.969
.915
.071
.053
.054
1.00
.998
.958
C05W
.056
.056
.055
.999
.989
.912
C05W
.042
.060
.059
.996
.974
.857
C05W
.062
.087
.996
.963
C05W
.070
.066
.090
.980
.930
.814
.014
.011
.009
.998
.981
.882
C01W
.015
.017
.007
.990
.946
.727
C01W
.008
.007
.012
.971
.901
.655
C01W
.028
.045
.947
.836
C01W
.026
.034
.042
.858
.722
.598
.175
.328
.762
1.00
1.00
.997
LfW
.192
.326
.608
1.00
1.00
.958
LfW
.184
.329
.511
.999
.997
.938
LfW
.092
.114
1.00
.994
LfW
.112
.119
.176
.997
.981
.988
.404
.835
1.00
.992
.999
1.00
LoW
.536
.901
.994
.995
1.00
1.00
LoW
.620
.926
.981
.993
.998
.999
LoW
.603
.710
.984
.987
LoW
.716
.803
.834
.982
.981
.994
1-20
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=9 N=4 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
053 344 265 188 107 060 022 109 683 028 379 279 184 082 051 023 111
037 359 286 198 103 056 022 102 806 039 374 284 199 101 051 019 144
.018 .366 .280 .201 .101 .060 .030 .344 .910 .014 .367 .276 .187 .107 .071 .045 .344
.824 .999 .999 .999 .999 .994 .910 .999 .970 .857 .994 .990 .984 .956 .871 .604 .968
820 998 995 990 975 914 717 984 983 873 992 986 978 930 843 583 967
.895 .976 .967 .924 .824 .620 .359 .973 .992 .881 .975 .956 .908 .807 .642 .402 .968
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=4 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
065 333 259 190 111 066 025 116 728 084 380 301 200 095 042 015 153
065 342 270 186 095 054 020 115 840 077 376 291 195 095 046 013 166
.074 .409 .297 .194 .101 .063 .029 .336 .945 .080 .406 .309 .194 .106 .064 .031 .336
.810 1.00 1.00 .998 .998 .991 .844 .997 .967 .830 .972 .959 .934 .863 .758 .465 .917
817 992 991 986 956 886 648 967 980 836 966 953 925 856 727 428 926
.846 .956 .918 .879 .761 .598 .362 .954 .991 .867 .953 .931 .885 .757 .596 .300 .949
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=4 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2 0
2 0
? 0
00
.50
.90
00
50
.90
075 332 265 181 111 063 021 118 748 126 363 281 187 100 049 019 177
.094 .343 .260 .190 .105 .055 .019 .124 .859 .117 .380 .283 .193 .103 .054 .023 .199
.114 .385 .296 .197 .105 .060 .033 .406 .963 .106 .388 .287 .184 .085 .049 .026 .417
760 1 00 1 00 999 996 974 809 997 959 818 944 926 893 816 685 331 880
800 995 993 977 937 858 575 955 988 819 941 916 879 787 627 327 892
.882 .922 .891 .826 .681 .500 .199 .874 .976 .869 .910 .879 .820 .657 .460 .186 .852
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=4 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.098 .327 .246 .166 .088 .041 .006 .101 .783 .149 .352 .262 .174 .086 .042 .010 .186
.102 .343 .259 .174 .097 .053 .017 .114 .872 .152 .368 .281 .184 .092 .043 .019 .207
151 375 282 192 091 060 027 366 963 159 394 307 194 102 058 026 383
754 1 00 999 994 985 959 738 990 958 798 913 882 848 741 592 258 842
.780 .982 .976 .951 .904 .826 .530 .934 .981 .773 .898 .862 .816 .702 .554 .247 .830
.797 .872 .835 .765 .620 .430 .164 .831 .978 .832 .872 .835 .779 .632 .434 .172 .822
C40L C30L C20L C10L C05L C01L
C=9 N=6 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.013
.007
.944
.963
.366 .275
.403 .302
1 .00 1 .00
1 .00 1 .00
.180
.204
1.00
1.00
.093
.107
1.00
1.00
.045 .018 .088 .112 .003 .412 .303 .207 .114 .071 .024 .113
.059 .017 .117 .293 .001 .404 .324 .220 .120 .072 .025 .120
1.00 1.00 1.00 .964 .964 1.00 1.00 1.00 1.00 .999 .994 1.00
1.00 .991 1.00 .990 .976 1.00 1.00 1.00 1.00 .997 .974 1.00
.877
.902
.897
.994
.992
.997
LOG
.917
.936
.958
.992
.990
.996
LOG
.950
.953
.967
.991
.997
.970
LOG
.962
.971
.963
.995
.994
.971
LOG
.240
.302
.988
.995
.061
.056
.021
.811
.844
.908
MxW
.074
.084
.064
.830
.846
.875
MxW
.115
.110
.112
.784
.802
.865
MxW
.131
.125
.149
.764
.776
.841
MxW
.006
.001
.969
.973
.384
.390
.411
.999
.992
.976
C40W
.354
.373
.361
.987
.984
.955
C40W
.346
.356
.381
.983
.968
.901
C40W
.326
.350
.365
.973
.938
.890
C40W
.380
.390
1.00
1.00
.303
.312
.313
.998
.985
.951
C30W
.280
.284
.269
.982
.980
.930
C30W
.270
.275
.275
.979
.953
.873
C30W
.260
.267
.278
.960
.909
.847
C30W
.295
.292
1.00
1.00
.212
.202
.230
.992
.977
.922
C20W
.184
.202
.180
.975
.963
.872
C20W
.181
.188
.192
.967
.928
.814
C20W
.172
.187
.179
.931
.880
.794
C20W
.199
.200
1.00
1.00
.107
.095
.135
.976
.941
.812
C10W
.091
.112
.103
.949
.915
.759
C10W
.097
.086
.083
.912
.862
.652
C10W
.077
.088
.082
.862
.797
.618
C10W
.099
.103
1.00
1.00
.065
.054
.088
.928
.863
.651
C05W
.045
.059
.069
.886
.825
.587
C05W
.049
.046
.046
.839
.728
.459
C05W
.035
.042
.043
.774
.657
.420
C05W
.062
.062
1.00
.998
.029
.024
.043
.720
.618
.373
C01W
.015
.022
.040
.633
.530
.318
C01W
.018
.021
.024
.530
.436
.175
C01W
.006
.010
.024
.443
.323
.165
C01W
.025
.024
.997
.969
.118
.130
.390
.980
.966
.968
LfW
.106
.155
.333
.954
.946
.940
LfW
.124
.143
.415
.930
.923
.855
LfW
.108
.155
.379
.889
.845
.838
LfW
.097
.104
1.00
1.00
.802
.872
.925
.978
.988
.998
LoW
.843
.905
.955
.983
.991
.995
LoW
.861
.903
.965
.967
.992
.968
LoW
.905
.943
.969
.982
.987
.983
LoW
.185
.333
.985
.997
1-21
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=9 N=6 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.75
.00
.50
.75
044
032
.006
.945
.963
.979
369 279 197 115 061 015 112 253 023 378 284 190 108 061 029 118
376 278 191 100 060 024 115 464 014 401 302 194 085 052 022 120
.399 .307 .207 .110 .069 .027 .253 .656 .012 .388 .296 .210 .111 .068 .025 .247
1.00 1.00 1.00 1.00 1.00 .998 1.00 .967 .950 1.00 1.00 1.00 .999 .999 .960 .999
1.00 1.00 1.00 1.00 .997 .969 1.00 .991 .964 1.00 .999 .999 .995 .990 .923 .999
1.00 1.00 .996 .986 .968 .837 1.00 .998 .964 .999 .997 .992 .980 .953 .819 .999
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=6 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
? n
00
50
.85
.00
50
.85
062
069
.041
.918
929
.964
327 236 164 082 038 009 086 356 051 363 291 179 099 049 010 141
371 286 191 108 052 019 121 570 059 403 294 200 101 056 016 172
.420 .325 .207 .111 .063 .028 .528 .873 .026 .419 .331 .224 .118 .075 .033 .543
1.00 1.00 1.00 1.00 1.00 .995 1.00 .959 .949 .999 .999 .995 .988 .966 .851 .993
1 00 1 00 1 00 998 990 928 999 980 949 996 994 991 979 948 781 988
.990 .983 .971 .918 .841 .581 .986 .999 .975 .992 .986 .979 .938 .862 .595 .987
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=6 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2.0
2 0
? 0
00
.50
.85
.00
50
.85
084
.101
.105
.907
932
.950
330 254 186 099 042 005 106 409 131 390 305 189 101 048 017 171
.353 .271 .192 .106 .050 .014 .128 .622 .098 .363 .283 .188 .088 .043 .017 .197
.406 .305 .194 .096 .052 .018 .458 .930 .102 .415 .331 .208 .106 .053 .019 .448
1.00 1.00 1.00 1.00 .999 .982 1.00 .955 .939 .990 .983 .978 .953 .913 .695 .967
1 00 999 999 993 978 873 998 986 941 994 990 985 955 900 669 985
.984 .971 .953 .890 .792 .521 .978 .998 .952 .983 .969 .945 .892 .802 .502 .976
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=6 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0 5
2 0
2.0
2.0
.00
.50
90
00
.50
.90
.122 .343 .278 .185 .096 .043 .005 .106 .482 .165 .364 .278 .176 .078 .037 .002 .197
.118 .353 .271 .177 .086 .038 .010 .121 .691 .171 .362 .294 .197 .104 .053 .013 .237
165 395 307 208 103 063 020 527 974 156 392 299 212 113 059 021 510
888 1 00 1 00 1 00 998 994 963 998 959 935 985 976 961 916 845 561 950
.919 .998 .998 .997 .985 .971 .867 .990 .973 .925 .978 .965 .945 .894 .815 .511 .944
.961 .965 .944 .906 .819 .710 .370 .931 .998 .950 .956 .929 .881 .799 .665 .353 .917
.511
.601
.651
.989
.994
.993
LOG
.740
.790
.856
.988
.993
.999
LOG
.856
.889
.953
.993
.996
1.00
LOG
.934
.929
.979
.997
.996
.994
.027
.014
.009
.953
.957
.972
MxW
.070
.066
.034
.936
.948
.957
MxW
.128
.122
.103
.919
.927
.956
MxW
.137
.152
.159
.910
.910
.971
.372
.371
.391
1.00
1.00
.999
C40W
.401
.385
.401
1.00
.999
.989
C40W
.367
.403
.393
.998
.996
.979
C40W
.325
.351
.386
.995
.980
.959
.285
.286
.300
1.00
1.00
.997
C30W
.316
.298
.306
1.00
.999
.981
C30W
.290
.301
.289
.997
.993
.969
C30W
.255
.274
.285
.994
.977
.930
.191
.200
.191
1.00
1.00
.996
C20W
.212
.205
.211
.999
.997
.971
C20W
.189
.196
.198
.996
.989
.946
C20W
.185
.176
.193
.989
.965
.896
.096
.109
.111
1.00
.997
.985
C10W
.110
.096
.122
.998
.987
.920
C10W
.097
.113
.091
.990
.964
.892
C10W
.085
.097
.088
.979
.933
.807
.053
.065
.070
1.00
.987
.965
C05W
.059
.053
.070
.998
.961
.851
C05W
.050
.057
.044
.978
.931
.795
C05W
.040
.054
.055
.959
.890
.670
.018
.026
.028
.989
.910
.812
C01W
.016
.017
.030
.950
.823
.588
C01W
.011
.013
.013
.879
.753
.518
C01W
.011
.012
.019
.785
.655
.314
.105
.130
.245
1.00
1.00
.999
LfW
.119
.138
.520
.998
.992
.983
LfW
.125
.156
.424
.991
.981
.974
LfW
.116
.167
.512
.984
.957
.920
.428
.559
.689
.984
.990
.996
LoW
.577
.715
.871
.982
.991
1.00
LoW
.699
.812
.956
.983
.993
1.00
LoW
.730
.864
.982
.980
.982
.997
1-22
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
C40L C30L C20L C10L C05L C01L
C=9 N=6 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
? n
00
50
.90
.00
50
.90
134 323 261 182 103 059 010 112 535 221 358 275 191 086 025 007 232
148 321 260 193 116 058 012 151 756 250 390 321 218 105 048 007 277
.229 .372 .286 .202 .088 .041 .011 .454 .986 .266 .404 .294 .192 .104 .039 .011 .452
.889 1.00 1.00 1.00 1.00 .998 .947 1.00 .956 .882 .950 .929 .901 .825 .723 .381 .906
891 998 997 994 969 944 764 983 981 911 959 945 915 856 759 447 936
.931 .941 .912 .872 .791 .672 .331 .914 .996 .932 .933 .914 .872 .762 .630 .285 .905
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=9 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
2.0
2.0
00
50
.00
.50
028
003
.991
.993
411
382
1.00
1.00
305
281
1.00
1.00
201
201
1.00
1.00
110
097
1.00
1.00
060
057
1.00
1.00
017 107 012 006 412 315 205 113 066 018 121
017 107 084 000 406 295 200 120 068 021 138
1.00 1.00 .974 .993 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 .992 .993 1.00 1.00 1.00 1.00 1.00 .999 1.00
C40L C30L C20L C10L C05L C01L
C=9 N=9 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0 5
0 5
2.0
2.0
2.0
.00
50
75
.00
.50
.75
.064
052
016
.977
.992
.994
.379
380
404
1.00
1.00
1.00
.283
294
293
1.00
1.00
1.00
.196 .112 .056 .019 .119 .065 .025 .388 .283 .187 .089 .043 .012 .114
204 105 059 014 130 242 029 397 284 195 103 061 019 162
199 109 064 017 395 598 016 397 307 210 117 068 014 384
1.00 1.00 1.00 1.00 1.00 .959 .992 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 .999 .997 1.00 .992 .994 1.00 1.00 1.00 .998 .998 .994 .999
1.00 1.00 .998 .966 1.00 .998 .991 .999 .999 .999 .998 .994 .966 1.00
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
C=9 N=9 CV=2.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.85
.00
.50
.85
089
081
.040
.979
.978
.994
361 286 191 097 052 008 103 127 076 368 270 181 095 051 012 159
361 280 175 091 049 012 127 348 080 384 282 184 104 050 024 182
.432 .330 .211 .096 .056 .016 .727 .912 .047 .388 .284 .189 .100 .062 .023 .702
1.00 1.00 1.00 1.00 1.00 .999 1.00 .968 .986 1.00 1.00 1.00 .999 .997 .977 1.00
1.00 1.00 1.00 1.00 1.00 .991 1.00 .984 .988 1.00 1.00 .999 .998 .994 .955 .998
.998 .996 .992 .981 .960 .842 1.00 .999 .994 .999 .998 .997 .982 .959 .832 .999
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=9 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.144
.140
.152
.979
.986
.993
.336 .255 .177 .089 .057 .015 .093 .191 .171 .359 .271 .196 .098 .057 .014 .214
.377 .284 .186 .095 .043 .013 .139 .484 .164 .388 .291 .194 .091 .041 .005 .228
.413 .324 .225 .114 .065 .016 .596 .933 .151 .404 .304 .201 .106 .059 .021 .609
1.00 1.00 1.00 1.00 1.00 .998 1.00 .969 .981 .998 .997 .993 .984 .972 .909 .991
1.00 1.00 1.00 .999 .998 .988 .999 .986 .988 .999 .998 .997 .992 .979 .890 .998
.999 .995 .983 .968 .927 .781 .999 1.00 .990 .997 .991 .982 .963 .931 .757 .997
.955
.968
.989
.993
.997
.997
LOG
.046
.118
.989
.992
LOG
.273
.397
.629
.993
.995
.996
LOG
.566
.654
.903
.992
.993
1.00
LOG
.788
.847
.959
.997
.999
1.00
.200
.176
.232
.887
.909
.937
MxW
.004
.002
.986
.996
MxW
.047
.023
.011
.977
.990
.995
MxW
.118
.090
.046
.987
.985
.996
MxW
.167
.176
.141
.977
.976
.989
.371
.349
.372
.991
.980
.941
C40W
.372
.388
1.00
1.00
C40W
.380
.395
.397
1.00
1.00
1.00
C40W
.381
.401
.400
1.00
1.00
.998
C40W
.367
.384
.404
1.00
1.00
.993
.289
.271
.270
.988
.971
.919
C30W
.278
.287
1.00
1.00
C30W
.284
.301
.302
1.00
1.00
1.00
C30W
.303
.304
.311
1.00
1.00
.996
C30W
.282
.278
.306
1.00
.999
.992
.194
.176
.190
.978
.957
.879
C20W
.184
.190
1.00
1.00
C20W
.191
.206
.202
1.00
1.00
1.00
C20W
.194
.212
.207
1.00
1.00
.994
C20W
.190
.189
.202
.999
.998
.979
.088
.083
.088
.952
.913
.787
C10W
.094
.102
1.00
1.00
C10W
.094
.111
.103
1.00
1.00
.999
C10W
.111
.112
.119
1.00
1.00
.982
C10W
.082
.097
.106
.997
.993
.959
.047
.031
.050
.916
.854
.639
C05W
.047
.053
1.00
1.00
C05W
.053
.060
.056
1.00
1.00
.994
C05W
.057
.062
.055
1.00
.998
.962
C05W
.036
.046
.049
.995
.989
.925
.008
.004
.010
.687
.591
.278
C01W
.024
.012
1.00
.999
C01W
.009
.021
.024
1.00
.992
.970
C01W
.014
.018
.019
.999
.973
.839
C01W
.008
.011
.017
.980
.943
.751
.129
.166
.419
.960
.948
.910
LfW
.091
.116
1.00
1.00
LfW
.101
.153
.367
1.00
1.00
1.00
LfW
.128
.185
.728
1.00
1.00
.998
LfW
.116
.186
.604
.997
.997
.994
.777
.889
.986
.975
.992
.998
LoW
.026
.136
.982
.998
LoW
.164
.328
.619
.978
.992
.999
LoW
.349
.586
.914
.987
.993
.999
LoW
.457
.727
.967
.989
.993
1.00
1-23
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=9 N=9 CV=3.5
LoL MxG C40G C30G C20G C10G C05G
C01G LfG
MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.90
.00
50
.90
156
169
.232
.964
968
.993
358 271 179 088 046 006 096 250 221 372 287 201 111 048 015 254
369 270 179 095 055 007 135 525 241 387 291 183 102 051 007 288
.378 .297 .199 .096 .046 .010 .656 .994 .241 .420 .313 .191 .099 .047 .010 .667
1.00 1.00 1.00 1.00 1.00 .996 1.00 .954 .971 .996 .995 .985 .966 .932 .802 .989
1 00 1 00 999 998 992 967 998 991 983 994 992 985 963 937 787 990
.988 .976 .963 .926 .849 .621 .974 1.00 .987 .981 .971 .959 .921 .847 .605 .966
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=9 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.90
.00
50
.90
177
186
.294
.956
959
.984
354 269 190 102 058 015 113 295 320 404 310 222 121 054 013 337
321 240 173 085 040 006 127 558 321 385 301 211 105 055 009 354
.375 .293 .202 .096 .046 .009 .592 .996 .308 .385 .286 .201 .109 .062 .015 .559
1.00 1.00 1.00 1.00 1.00 .995 1.00 .959 .965 .988 .976 .963 .938 .893 .699 .977
1 00 1 00 998 991 986 941 997 979 971 989 980 966 951 891 693 977
.983 .969 .949 .890 .809 .520 .968 1.00 .980 .966 .957 .936 .879 .796 .538 .949
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=9 N=12 CV=1.5
C30G C20G C10G
C05G C01G LfG
0 5
0.5
2.0
2.0
00
.50
.00
.50
025
.004
.997
.996
397
.380
1.00
1.00
293
.291
1.00
1.00
1 96 1 00
.197 .117
1 .00 1 .00
1 .00 1 .00
053
.063
1.00
1.00
013 104 000 005 375 280 181 102 059 018 111
.009 .136 .041 .001 .366 .262 .170 .083 .049 .017 .102
1.00 1.00 .974 .998 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 .987 .998 1.00 1.00 1.00 1.00 1.00 1.00 1.00
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
MxG C40G
C=9 N=12 CV=2.0
C30G C20G C10G
C05G C01G
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.75
.00
.50
.75
091
067
.018
.998
.997
.998
391
407
.394
1.00
1.00
1.00
293
308
.293
1.00
1.00
1.00
207 113
216 107
.197 .107
1 .00 1 .00
1 .00 1 .00
1 .00 1 .00
060
056
.049
1.00
1.00
1.00
019
017
.012
1.00
1.00
.992
120 015
140 113
.459 .625
1.00 .976
1.00 .991
1.00 1.00
049 412 324 225 115 061 016 147
036 426 321 214 107 065 016 170
.022 .371 .282 .176 .095 .058 .014 .442
.999 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.998 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.998 1.00 1.00 1.00 1.00 .999 .992 1.00
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
MxG C40G
C=9 N=12 CV=2.5
C30G C20G C10G
C05G C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.85
.00
.50
.85
.144
.134
.077
.992
.991
1.00
.391
.388
.423
1.00
1.00
1.00
.306
.290
.330
1.00
1.00
1.00
.202 .113
.198 .112
.227 .102
1 .00 1 .00
1 .00 1 .00
1 .00 1 .00
.056 .013 .120 .045 .119 .406 .317 .219 .103 .051 .009 .179
.058 .009 .151 .226 .097 .381 .273 .183 .089 .045 .009 .214
.054 .009 .870 .959 .054 .392 .307 .184 .099 .051 .010 .858
1.00 1.00 1.00 .971 .998 1.00 1.00 1.00 1.00 1.00 .999 1.00
1.00 .998 1.00 .984 .996 1.00 1.00 1.00 .999 .999 .996 1.00
.992 .953 1.00 1.00 .998 .999 .999 .997 .995 .991 .954 1.00
.887
.921
.995
.994
1.00
.999
LOG
.970
.962
.996
1.00
.999
.999
LOG
.006
.071
.989
.994
LOG
.117
.262
.624
.994
.996
1.00
LOG
.457
.584
.968
.996
.999
1.00
.192
.228
.246
.968
.968
.990
MxW
.268
.319
.308
.957
.968
.983
MxW
.010
.003
.997
.999
MxW
.057
.042
.013
.998
.996
.999
MxW
.149
.131
.069
.994
.997
.999
.357
.357
.398
1.00
.999
.984
C40W
.368
.393
.373
1.00
.998
.976
C40W
.388
.422
1.00
1.00
C40W
.384
.396
.412
1.00
1.00
1.00
C40W
.396
.384
.404
1.00
1.00
.998
.276
.282
.311
1.00
.996
.975
C30W
.291
.311
.276
.999
.996
.966
C30W
.290
.321
1.00
1.00
C30W
.284
.311
.310
1.00
1.00
1.00
C30W
.287
.297
.306
1.00
1.00
.998
.197
.201
.220
.999
.995
.948
C20W
.204
.221
.179
.993
.988
.938
C20W
.209
.200
1.00
1.00
C20W
.194
.218
.208
1.00
1.00
1.00
C20W
.201
.183
.209
1.00
1.00
.997
.092
.097
.106
.994
.983
.909
C10W
.113
.118
.074
.990
.978
.871
C10W
.096
.108
1.00
1.00
C10W
.107
.108
.107
1.00
1.00
.999
C10W
.118
.093
.108
1.00
1.00
.995
.043
.043
.055
.991
.972
.834
C05W
.049
.053
.037
.980
.955
.779
C05W
.053
.056
1.00
1.00
C05W
.060
.064
.054
1.00
1.00
.999
C05W
.059
.049
.065
1.00
1.00
.987
.010
.004
.014
.956
.898
.591
C01W
.011
.012
.010
.905
.835
.492
C01W
.010
.022
1.00
1.00
C01W
.019
.013
.016
1.00
1.00
.994
C01W
.015
.010
.014
1.00
.996
.932
.132
.196
.652
.995
.995
.966
LfW
.160
.232
.543
.991
.988
.960
LfW
.103
.136
1.00
1.00
LfW
.126
.168
.509
1.00
1.00
1.00
LfW
.137
.194
.837
1.00
1.00
.998
.594
.796
.990
.976
.993
1.00
LoW
.689
.839
.993
.989
.992
.999
LoW
.002
.075
.985
.994
LoW
.041
.219
.650
.986
.997
.999
LoW
.154
.440
.955
.984
.996
1.00
1-24
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
MU
MU
MU
MU
C40L C30L C20L C10L C05L C01L
C=9 N=12 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.85
.00
.50
.85
150
176
.174
.996
.992
.999
356 260 184 089 038 007 101 073 234 406 318 211 117 055 015 268
350 264 160 087 050 009 127 297 233 405 309 223 123 057 012 309
.413 .317 .220 .123 .068 .012 .752 .977 .188 .400 .299 .206 .111 .056 .010 .728
1.00 1.00 1.00 1.00 1.00 1.00 1.00 .968 .995 1.00 1.00 1.00 1.00 .998 .971 1.00
1.00 1.00 1.00 1.00 1.00 .999 1.00 .989 .998 .999 .999 .999 .997 .993 .967 .999
1.00 .998 .995 .983 .976 .908 .999 1.00 .995 .995 .995 .995 .989 .980 .905 .996
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=12 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.90
.00
.50
.90
215
237
.293
.988
.990
.998
378
372
.392
1.00
1.00
.989
308
288
.303
1.00
1.00
.986
208 105 047 010 122 152 338 384 308 207 102 054 008 342
201 097 048 011 160 400 303 378 293 175 090 047 006 359
.202 .109 .058 .010 .772 .997 .289 .397 .309 .203 .094 .034 .009 .753
1.00 1.00 1.00 .998 1.00 .953 .990 1.00 .999 .993 .986 .980 .917 .996
1.00 1.00 .999 .991 1.00 .990 .995 .999 .999 .996 .987 .976 .895 .995
.979 .962 .933 .787 .984 1.00 .998 .994 .989 .983 .958 .921 .766 .988
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=12 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0.5
0.5
2.0
2.0
2.0
00
.50
.90
.00
.50
.90
225
.279
.389
.987
.992
.997
334
.368
.401
1.00
1.00
.993
261
.286
.294
1.00
1.00
.986
180 087 036 004 097 149 410 397 323 219 112 059 012 374
.208 .107 .050 .006 .153 .476 .372 .371 .273 .195 .092 .041 .013 .379
.214 .111 .063 .017 .687 .997 .428 .412 .323 .212 .101 .058 .011 .664
1.00 1.00 1.00 .999 1.00 .961 .992 .997 .992 .990 .970 .944 .841 .994
1.00 1.00 .999 .994 1.00 .986 .990 .991 .990 .988 .972 .948 .832 .992
.977 .953 .903 .746 .987 1.00 1.00 .996 .990 .980 .955 .907 .712 .991
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=16 CV=1.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
2.0
2.0
.00
.50
.00
.50
.052
.010
1.00
1.00
.412
.388
1.00
1.00
.318
.286
1.00
1.00
.213 .100
.181 .102
1 .00 1 .00
1 .00 1 .00
.052
.058
1.00
1.00
.010 .110 .000 .004 .390 .313 .205 .099 .053 .008 .117
.014 .139 .038 .005 .378 .288 .193 .091 .042 .011 .146
1.00 1.00 .978 .999 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 .991 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=16 CV=2.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0.5
0.5
0.5
2.0
2.0
2.0
.00
.50
.75
.00
.50
.75
.091
.077
.020
1.00
1.00
1.00
.372
.372
.373
1.00
1.00
1.00
.283
.287
.295
1.00
1.00
1.00
.198 .103
.197 .111
.197 .105
1 .00 1 .00
1 .00 1 .00
1 .00 1 .00
.058
.062
.058
1.00
1.00
1.00
.014
.019
.014
1.00
1.00
1.00
.110 .002
.145 .059
.566 .701
1.00 .976
1.00 .989
1.00 1.00
.070 .393 .306 .191 .096 .043 .012 .137
.048 .374 .286 .187 .104 .065 .016 .204
.023 .394 .300 .208 .096 .047 .010 .567
1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
.999 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 1.00 1.00 1.00 1.00 .999 1.00
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.745
.822
.974
.997
.999
1.00
LOG
.892
.921
.998
.999
.998
1.00
LOG
.949
.967
.999
1.00
1.00
1.00
LOG
.000
.051
.990
.988
LOG
.050
.191
.702
.998
.997
1.00
.222
.237
.180
.992
.996
.998
MxW
.288
.305
.317
.994
.989
.999
MxW
.309
.360
.429
.990
.992
.994
MxW
.020
.009
.999
1.00
MxW
.064
.058
.023
1.00
1.00
1.00
.385
.382
.406
1.00
1.00
1.00
C40W
.383
.388
.409
1.00
1.00
.997
C40W
.359
.383
.415
1.00
.999
.989
C40W
.415
.397
1.00
1.00
C40W
.363
.396
.398
1.00
1.00
1.00
.302
.305
.296
1.00
1.00
.999
C30W
.284
.310
.300
1.00
.999
.993
C30W
.262
.288
.312
.999
.998
.981
C30W
.306
.304
1.00
1.00
C30W
.263
.311
.311
1.00
1.00
1.00
.205
.208
.192
1.00
.999
.996
C20W
.197
.224
.207
1.00
.999
.987
C20W
.171
.192
.217
.999
.996
.971
C20W
.212
.200
1.00
1.00
C20W
.186
.216
.213
1.00
1.00
1.00
.107
.098
.094
1.00
.998
.991
C10W
.089
.106
.113
.999
.998
.960
C10W
.084
.103
.116
.999
.995
.940
C10W
.110
.108
1.00
1.00
C10W
.091
.103
.114
1.00
1.00
1.00
.048
.053
.051
1.00
.995
.976
C05W
.043
.057
.066
.998
.995
.931
C05W
.043
.045
.057
.998
.988
.892
C05W
.050
.055
1.00
1.00
C05W
.046
.056
.066
1.00
1.00
1.00
.012
.014
.010
.998
.981
.898
C01W
.009
.012
.015
.993
.962
.772
C01W
.005
.007
.012
.972
.920
.720
C01W
.009
.015
1.00
1.00
C01W
.008
.015
.020
1.00
1.00
.998
.145
.209
.730
1.00
1.00
1.00
LfW
.144
.242
.744
1.00
.998
.994
LfW
.139
.255
.678
.999
.997
.986
LfW
.131
.148
1.00
1.00
LfW
.121
.200
.614
1.00
1.00
1.00
.349
.624
.980
.989
.998
1.00
LoW
.449
.738
1.00
.988
.994
1.00
LoW
.555
.821
.998
.994
.996
1.00
LoW
.000
.040
.986
.997
LoW
.005
.156
.735
.985
.999
1.00
1-25
-------
Appendix I. SSL Simulation Results: Estimated Probabilities of Investigating Further
MU MIX | MxL
C40L C30L C20L C10L C05L C01L
MxG C40G
C=9 N=16 CV=2.5
C30G C20G C10G
C05G C01G
0 5
0 5
0.5
2.0
2.0
2.0
00
50
.85
.00
.50
.85
163
187
.071
.998
1.00
1.00
379
396
.420
1.00
1.00
1.00
281
306
.322
1.00
1.00
1.00
198 113
209 113
.207 .098
1 .00 1 .00
1 .00 1 .00
1 .00 .998
052
056
.053
1.00
1.00
.997
008
019
.014
1.00
1.00
.982
124 012
158 121
.952 .984
1.00 .966
1.00 .995
1.00 1.00
170 397 297 199 098 054 009 216
145 388 282 190 090 048 013 269
.065 .438 .331 .232 .107 .062 .015 .963
.998 1.00 1.00 1.00 1.00 1.00 1.00 1.00
1.00 1.00 1.00 1.00 1.00 1.00 .999 1.00
1.00 1.00 1.00 1.00 1.00 .998 .989 1.00
MxW C40W C30W C20W C10W C05W C01W LfW LoW
.331
.467
.986
.996
1.00
1.00
.170
.161
.085
.998
.999
1.00
.380
.404
.406
1.00
1.00
1.00
.279
.301
.304
1.00
1.00
1.00
.186
.208
.201
1.00
1.00
1.00
.103
.105
.106
1.00
1.00
1.00
.053
.051
.058
1.00
1.00
.997
.013
.008
.018
1.00
1.00
.983
.140
.226
.942
1.00
1.00
1.00
.066
.330
.988
.986
.996
1.00
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=16 CV=3.0
MxG C40G C30G C20G C10G C05G
C01G LfG
LoG | MxW C40W C30W C20W C10W C05W C01W LfW LoW
0 5
0 5
0.5
2.0
2 0
2.0
00
50
.85
.00
50
.85
227
256
.252
.998
1 00
1.00
400 315
390 302
.387 .277
1 .00 1 .00
1 00 1 00
1 .00 1 .00
210 105
224 124
.186 .093
1 .00 1 .00
1 00 1 00
.999 .997
053 005 125 026 279 366 269 192 099 047 007 287 655
049 011 184 207 297 376 291 204 104 047 007 361 766
.050 .011 .860 .982 .225 .368 .282 .186 .093 .045 .011 .829 .991
1.00 1.00 1.00 .961 .999 1.00 1.00 1.00 .999 .998 .997 .999 .997
1 00 1 00 1 00 993 999 1 00 1 00 1 00 1 00 1 00 991 1 00 1 00
.991 .971 1.00 1.00 1.00 1.00 1.00 1.00 .998 .993 .964 1.00 1.00
278
286
.247
.999
1 00
1.00
383 292
399 299
.430 .336
1 .00 1 .00
1 00 1 00
1 .00 1 .00
205 106
192 097
.229 .135
1 .00 1 .00
1 00 1 00
1 .00 .996
057
060
.074
1.00
1 00
.995
013
017
.015
1.00
1 00
.968
1 70 1 74
271 548
.851 .987
1.00 .989
1 00 1 00
1.00 1.00
MU
MU
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=16 CV=3.5
MxG C40G C30G C20G C10G C05G
C01G LfG
0 *
0 5
0.5
2.0
2.0
2.0
00
.50
.90
.00
.50
.90
271
.319
.396
.995
.999
1.00
365
.353
.432
1.00
1.00
.997
283
.278
.338
1.00
1.00
.996
202 114
.190 .098
.224 .118
1 .00 1 .00
1 .00 1 .00
.993 .985
071 009 130 070 392 368 276 180 085 041 010 387
.058 .009 .162 .271 .382 .366 .282 .199 .097 .050 .013 .441
.058 .015 .846 1.00 .351 .387 .289 .198 .106 .044 .010 .822
1.00 1.00 1.00 .961 .999 1.00 1.00 1.00 1.00 .998 .976 1.00
1.00 .999 1.00 .991 .998 1.00 1.00 1.00 .999 .993 .968 1.00
.970 .902 .995 1.00 1.00 .999 .997 .997 .991 .975 .887 .999
MIX | MxL C40L C30L C20L C10L C05L C01L
LfL
LoL
C=9 N=16 CV=4.0
MxG C40G C30G C20G C10G C05G
C01G LfG
0 5
0 5
0.5
2.0
2.0
2.0
.00
.50
.90
.00
.50
.90
.297
.379
.481
.998
.997
1.00
.353
.393
.402
1.00
1.00
1.00
.276
.305
.294
1.00
1.00
.997
.202 .111
.209 .102
.198 .100
1 .00 1 .00
1 .00 1 .00
.991 .979
.052 .014 .125 .072 .502 .379 .282 .176 .088 .049 .010 .438
.053 .004 .181 .337 .513 .413 .325 .215 .116 .066 .015 .513
.049 .011 .793 1.00 .493 .387 .298 .201 .091 .047 .011 .738
1.00 .999 1.00 .950 .996 .998 .998 .995 .991 .981 .932 .997
1.00 .998 1.00 .990 .999 .998 .998 .995 .991 .982 .923 1.00
.952 .872 .998 1.00 .999 .998 .995 .992 .970 .934 .817 .994
LoG | MxW C40W C30W C20W C10W C05W C01W LfW LoW
.867
.918
1.00
1.00
1.00
1.00
LOG
.949
.967
.999
1.00
1.00
1.00
.348
.397
.351
.999
1.00
1.00
MxW
.414
.441
.493
.998
.997
1.00
.374
.419
.411
1.00
1.00
.998
C40W
.358
.377
.389
1.00
1.00
.998
.294
.319
.325
1.00
1.00
.996
C30W
.280
.288
.315
1.00
1.00
.995
.201
.220
.223
1.00
1.00
.993
C20W
.183
.207
.196
1.00
1.00
.986
.103
.113
.113
1.00
.999
.980
C10W
.090
.095
.102
1.00
.999
.973
.056
.052
.061
1.00
.999
.970
C05W
.039
.046
.053
1.00
.997
.939
.014
.011
.006
.999
.994
.881
C01W
.005
.010
.010
.996
.979
.820
.166
.295
.843
1.00
.999
.996
LfW
.165
.293
.773
1.00
1.00
.994
.291
.678
1.00
.991
1.00
1.00
LoW
.394
.779
1.00
.994
.999
1.00
1-26
-------
APPENDIX J
Piazza Road Simulation Results
-------
APPENDIX J
Piazza Road Simulation Results
Section 4.3.6 contains background information on the Piazza Road site and its seven exposure areas
(EAs), as well as a complete description of the simulation setup and parameters. The following
notation is used in the tables of this appendix.
EA = exposure area number (from 1 to 7)
MEAN = the true mean for the EA (an average of 96 measurements)
CV = the true value of the coefficient of variation for the EA
DES = indicator of whether compositing within strata (DES = W) or compositing
across strata (DES = X) was used
C = the number of specimens per composite
N = the number of composite samples chemically analyzed.
The remaining four variables give the estimated error rates at 0.5 SSL and 2 SSL for the Max test
(labelled as MAX 0.5SSL and MAX 2.0SSL) and the Chen test at the nominal .10 level (labelled as
CHEN 0.5SSL and CHEN 2.0SSL).
J-l
-------
Appendix J. Estimated decision error rates for Chen test at the 0.1 level, and Max test, for
compositing within sector (DES=W) or across sector (DES=X), based on simulations from Piazza
Road Data.
EA = EA # (1 to 7). MEAN and CV denote EA true mean and CV.
M = # of samples per composite. N = # of composite samples.
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.01
.00
.01
.00
.02
.00
.00
.00
.00
.00
.01
.00
.00
.00
.00
.00
.00
.00
MAX
.OSSL
.03
.12
.01
.04
.00
.01
.01
.11
.01
.04
.00
.01
.02
.09
.01
.04
.00
.02
CHEN
0.5SSL
.02
.12
.04
.11
.01
.15
.01
.13
.04
.12
.01
.12
.00
.12
.02
.13
.01
.10
CHEN
2. OSSL
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
CA-O MCAM-O A r\/--\ fi
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX MAX
0.5SSL 2
.07
.04
.09
.05
.11
.07
.04
.01
.05
.01
.06
.01
.03
.00
.04
.00
.04
.00
.OSSL
.13
.13
.04
.05
.02
.01
.13
.14
.03
.04
.01
.01
.08
.12
.02
.04
.00
.02
CHEN
0.5SSL
.08
.11
.05
.11
.05
.11
.09
.09
.04
.09
.04
.13
.09
.10
.04
.10
.04
.12
CHEN
2. OSSL
.05
.05
.02
.01
.00
.00
.02
.02
.00
.00
.00
.00
.00
.01
.00
.00
.00
.00
J-2
-------
Appendix J. Estimated decision error rates for Chen test at the 0.1 level, and Max test, for
compositing within sector (DES=W) or across sector (DES=X), based on simulations from Piazza
Road Data.
EA = EA # (1 to 7). MEAN and CV denote EA true mean and CV.
M = # of samples per composite. N = # of composite samples.
EA=3 MEAN=5 1 CV=1 1
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
—
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.01
.00
.02
.00
.06
.00
.00
.00
.01
.00
.02
.00
.00
.00
.01
.00
.03
.00
EA=4 MEAN=3.
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.01
.00
.02
.00
.02
.00
.01
.00
.00
.00
.01
.00
.00
.00
.00
.00
.00
.00
MAX
.OSSL
.03
.11
.00
.03
.00
.01
.02
.09
.00
.03
.00
.01
.02
.10
.00
.03
.00
.01
CHEN
0.5SSL
.03
.14
.00
.11
.00
.10
.01
.12
.00
.11
.00
.11
.01
.14
.00
.12
.00
.12
3p\/— 1 o
\-f V — I .£-
MAX
.OSSL
.11
.11
.04
.04
.01
.01
.10
.10
.02
.03
.01
.02
.09
.12
.03
.03
.01
.01
CHEN
0.5SSL
.07
.10
.06
.11
.05
.09
.05
.12
.04
.09
.04
.10
.04
.12
.06
.11
.03
.11
CHEN
2. OSSL
.01
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
CHEN
2. OSSL
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
J-3
-------
Appendix J. Estimated decision error rates for Chen test at the 0.1 level, and Max test, for
compositing within sector (DES=W) or across sector (DES=X), based on simulations from Piazza
Road Data.
EA = EA # (1 to 7). MEAN and CV denote EA true mean and CV.
M = # of samples per composite. N = # of composite samples.
EA=5 MEAN=9 3 CV=2 0
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
—
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
M
4
4
4
4
4
4
6
6
6
6
6
6
8
8
8
8
8
8
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.22
.03
.48
.03
.71
.05
.18
.00
.47
.00
.76
.00
.19
.00
.45
.00
.76
.00
EA=6 MEAN=15.
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.18
.07
.22
.10
.34
.13
.13
.02
.15
.05
.33
.07
.08
.01
.11
.01
.31
.01
MAX
.OSSL
.13
.17
.03
.06
.00
.03
.06
.10
.01
.02
.00
.01
.05
.08
.01
.03
.00
.01
CHEN
0.5SSL
.01
.07
.00
.08
.00
.10
.00
.10
.00
.10
.00
.11
.00
.12
.00
.14
.00
.10
8r\i—'~> i
\-f V — £-.£-
MAX
.OSSL
.16
.20
.06
.08
.03
.03
.11
.16
.04
.06
.01
.03
.09
.14
.03
.04
.01
.02
CHEN
0.5SSL
.03
.10
.03
.09
.02
.09
.02
.09
.02
.08
.02
.09
.00
.09
.01
.09
.01
.10
CHEN
2. OSSL
.12
.04
.02
.00
.00
.00
.05
.01
.00
.00
.00
.00
.03
.00
.00
.00
.00
.00
CHEN
2. OSSL
.21
.19
.07
.07
.03
.03
.11
.09
.03
.02
.01
.01
.07
.03
.01
.01
.00
.00
J-4
-------
Appendix J. Estimated decision error rates for Chen test at the 0.1 level, and Max test, for
compositing within sector (DES=W) or across sector (DES=X), based on simulations from Piazza
Road Data.
EA = EA # (1 to 7). MEAN and CV denote EA true mean and CV.
M = # of samples per composite. N = # of composite samples.
DES
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
M
4
4
4
4
4
4
6
6
6
6
6
6
EA=7 MEAN=2.
N
4
4
6
6
8
8
4
4
6
6
8
8
4
4
6
6
8
8
MAX
0.5SSL 2
.03
.00
.09
.01
.16
.01
.02
.00
.06
.00
.10
.00
.01
.00
.04
.00
.06
.00
3 CV=1 4
MAX
.OSSL
.11
.13
.03
.06
.01
.02
.08
.10
.02
.03
.00
.02
.06
.10
.01
.03
.00
.01
CHEN
0.5SSL
.02
.09
.02
.07
.02
.09
.01
.12
.01
.09
.01
.09
.01
.12
.00
.10
.00
.11
CHEN
2. OSSL
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
J-5
-------
APPENDIX K
Soil Organic Carbon (Koc) / Water (Kow) Partition
Coefficients
-------
Table K-1. Values Used for Koc / Kow Correlation
Chemical
Benzene
Bromoform
Carbon tetrachloride
Chlorobenzene
Chloroform
Dichlorobenzene, 1,2- (o)
Dichlorobenzene, 1,4- (p)
Dichloroethane, 1,1-
Dichloroethane, 1,2-
Dichloroethylene, 1,1-
Dichloroethylene, trans -1 ,2-
Dichloropropane, 1,2-
Dieldrin
Endosulfan
Endrin
Ethylbenzene
Hexachlorobenzene
Methyl bromide
Methyl chloride
Methylene chloride
Pentachlorobenzene
Tetrachloroethane, 1,1,2,2-
Tetrachloroethylene
Toluene
Trichlorobenzene, 1,2,4-
Trichloroethane, 1,1,1-
Trichloroethane, 1,1,2-
Trichloroethylene
Xylene, o-
Xylene, m-
Xylene, p-
log Kow
2.13
2.35
2.73
2.86
1.92
3.43
3.42
1.79
1.47
2.13
2.07
1.97
5.37
4.10
5.06
3.14
5.89
1.19
0.91
1.25
5.26
2.39
2.67
2.75
4.01
2.48
2.05
2.71
3.13
3.20
3.17
Calci
log Koc
1.77
1.94
2.24
2.34
1.60
2.79
2.79
1.50
1.24
1.77
1.72
1.64
4.33
3.33
4.09
2.56
4.74
1.02
0.80
1.07
4.24
1.97
2.19
2.26
3.25
2.04
1.70
2.22
2.56
2.61
2.59
lated
Koc
59
87
174
219
40
617
617
32
17
59
52
44
21,380
2,138
12,303
363
54,954
10
6
12
17,378
93
155
182
1,778
110
50
166
363
407
389
Mea
log Koc
1.79
2.10
2.18
2.35
1.72
2.58
2.79
1.73
1.58
1.81
1.58
1.67
4.41
3.31
4.03
2.31
4.90
0.95
0.78
1.00
4.51
1.90
2.42
2.15
3.22
2.13
1.88
1.97
2.38
2.29
2.49
sured
Koc (geomean)
61.7
126
152
224
52.5
379
616
53.4
38.0
65
38
47.0
25,546
2,040
10,811
204
80,000
9.0
6.0
10
32,148
79.0
265
140
1,659
135
75.0
94.3
241
196
311
Regression Statistics
Multiple R 0.9870
R Square 0.9742
Adjusted R Square 0.9733
Standard Error 0.1640
Observations 31
ANOVA
Regression
Residual
Total
df
1
29
30
SS
29.4358
0.7804
30.2161
MS
29.4358
0.0269
F Significance F
1,094 1.4032E-24
Coefficients
Std. Error
tStat
P-value
Lower 95% Upper 95%
Intercept
X Variable 1
0.0784
0.7919
0.0748
0.0239
1.0481
33.0742
0.3033
0.0000
-0.0746
0.7430
0.2314
0.8409
K-1
-------
O)
T3
C
(0
I
O)
O
0.
c
o
'•?
jC
0)
1_
o
o
^
0)
3
O)
o
CD
CD
O
CD
LO
O
O
O
CD
CO
O)
o
o
CD
C\i
tti
CD
O
CD
O
O
CD
13
03
co
CD
E
oo
1^
p
CD
+
I
^
o
CO
CD
h~
CD
O
O
D)
O
O
O
O
CD
If)
O
CD
O
CD
CO
O
CD
Cvi
O
O
O
CD
CD
O
CD
CD
-------
I/)
.0
'
E>
o
o
I
•o
o
t»
_0)
"o
o
_0)
n
Comment
0
1
o
(6Mn)">i
d
(/)
o
~ra
o
1
o
(humic acids)
RP-HPLC on PIHAC
If
Szabo (199C
Oi
i^
s
CO
co"
O)
CO
CO
CO
0
C
£
re
0
o
(humic acids)
RP-HPLC on CIHAC
'm
Szabo (199C
£
§
CD"
0
00
s
8,
RJ
g
O)
CD
00
s
C
o
•s
F
o
o
Q.
=3
t-
LO
X
O
CD
CM
'o
CO
E
TJ
CD
"Gees croft/Rot hams
£
O)
"CD
"CD
o
s
CO
CM
4 e
CD 9
N. g
"=> °
8 1
CD
Gr. Br.); 2.05% OC; pH
Batcombe silt loam (
_
Briggs(198
O)
CD
CT>_
CO
CD
to
S
00*
s,
re
g
O)
CD
in
g
00*
s
o
•s
F
o
O
humic acids
CO
O)
C
E
CD
1
CD
S
LO
-it"
I--
CM
v
^
C
0
re
•E
<
CO
CD
CD
CO
CD"
rn
soil/sediments avera
§
it
o
_c
0
1
o
CM
1
LO"
ean 5 values)
o
CD
•S
jfi
S
1
CO
O)
Landrum et
O)
CM
CM
CD
LO
o
Q.
X
cyanopropyl column
£2
CD
Hodson &V\
%
$
O)
co"
CM
C
b
s
CO
X
on, Doe Run & Hickory
avg. coarse siltfract
rn
"CD
CD
*o
0
1
,-
s
o
CD"
CM
(humic acids)
RP-HPLC on PIHAC
v
O)
Szabo et al.
CM
CO
O
CO
CD"
CM
87% sand; foe = 0.01 8"
Flint aquifer sample;
C
o
Abdul& Gib
„
CM
n 8 values)
humic acid (geomea
O)
Landrum et
S
O)
CM
CO
CO
(humic acids)
RP-HPLC on CIHAC
v
O)
Szabo et al.
CO
LO
s
CO
co"
CO
O)
CO
CM
s,
(0
s
h.
f)
1
CM
C
o
•s
F
o
O
humic acid
CO
O)
C
E
McCarthy &
CO
150,000
CO
LO
CD
LO
0
o
(0
(0
b
£
CO
humic acid
O)
Landrum et
0
199,526
s
CO
X
on, Doe Run & Hickory
avg. coarse siltfract
rn
"CD
CD
*o
0
1
CO
650,000
O
o
o
LfT
s
"CD
CD
o
1-
Q
C
m
Cl
C0~
fe
"CD
"CD
E
C7)
CM
O)
840,000
§
459,882
(0
s
LO
LO
357,537
c
o
•s
F
o
O
CD
0
CM
CD"
o
CD
n
o
Woodburn silt loam;
CO
CO
Chiou etal.
S
CO
CM
t
0)
0>
0>
m
OC; column study
forest soil with 0.2%
§
CO
"CD
"CD
Q.
'CD
C7)
CO
LO
CM
CO
CO
2.2% OC; column study
agricultural soil with
CD
"CD
"CD
Q.
'CD
C7)
s
LO
CO
0
o
0
II
s
CD
Deriments; aquifer mate
batch equilibrium ex
c^~
1
Abdul et al.
O)
CD
9
OC; column study
forest soil with 3.7%
CD
CO
O)
"CD
"CD
Q.
'CD
C7)
CO
i--
LO
CO
LO
age; shake flask UV
soils/sediments aver
§
it
o
0
1
I--
i
-------
t/)
.0
'E
n
E>
o
.0
0
Q.
2
•o
in
0
3
n
g
^
0)
•g
—
o
o
0)
.0
n
(A
•g
0>
E
o
o
0>
S
o
(/)
J
o
o
_l
K»(L/kg)
6
CASN
(0
o
0
O
c
organic carbc
§
!
CO
o
CO
CO
CO
I
OJ
c
CO
CD
°s
'E
s
^
CO
CD
CM
CO
t
T?
0 C
0 O
CO •£-
'co~
TJ
'o
CO
o
o
X
Q_
RP-HPLC on
cu
CO
lo
"CD
CO
N
CM
CO
§
"co~
TJ
0
CO
0
I
o
X
O
RP-HPLC on
CO
CO
"co
"CD
CO
N
E°
S
CO
"CD
CO
X
^
o
o
X
c
=3
C£
%
jilt fraction, DC
CD
1
O
D)
CO
.—
CO
"co
"CD
!C
o
0
1
CM
CO
CO
CO
8
CD
CM
_C
' clay loam wil
Hastings silty
o
%
"co
CD
CD
D)
O
{£.
^
^
CM
CO
CM
CO
CO
"co
"CD
>>
_^:
CO
c
TJ
'o
CO
CO
te
CO
C-
Pavlou
8*
.
o
o
55
CO
1
E
CO
_g
"o
Overton silty
o
%
"co
CD
CD
D)
O
££
g
CM
O
O
CM
CO
to
to
s,
(0
g1"
CD
^
CM
to
s
s
_o
•s
E
o
O
3.8% OC
CO
Oi
I
'~
b
c
0
CO
CO
m"
t1"-
co
"co
E
§
ID
478,947
CO
CM
CO
6
ID
0
Jl
(0
I
0
CO
humic acids
S' '
CO
"co
"CD
E
Landru
ID
CO
ID
ID
CM
CO
CO
c3
=3
average 8 val
humic acids (
1D~
CO
CO
N
CD
c
1
-C
CO
1
CO
CD
2,130,000
&
CD
1,166,733
S,
(0
g1"
g
IT)
968,774
S
_o
•s
E
o
O
o
o
55
Es
O
O
TJ"
c
CO
CO
c
I
CO
CO
"co
"CD
c
o
CO
g
§
^
CO
ID
h-
CM CO 'sj-
CO CO '^t
CO CD 'st-
ID 1^ t^-
O O t-
CM CM t-
00
E C 0
£ £ ">>
o o £
330
£ £ 5
•e * ^
•zr •§• v
CO CO CO
«
S
s,
(0
g1"
«
s
s
s
_o
•s
E
o
o
>,
I
CO
c
E
13
O
O
6"
O
55
1
CD
NY, composil
Broome Co.,
CO
CD
=3
Q
°S
"CD
1
S
•^t
h-~
CO
^
CO
0
_ro
RJ
1 2
a- S
^*^ QJ
>, ^,
x 6"
0 £
£ ^
>• >>
£ -C=
•5 9
la ^
m £.
CO
TJ
'0
CO
0
E
=3
.£=
cl
c
CD
CO
CD
Q.
C
O
"o.
O
CO
c
CD
D)
c
CO
measured ch
^
CO
CO
CO
1
°e
CO
o
ID
ID
S
CM
g
LO
114,337
S,
(0
g1"
ID
O
ID
S
E
s
o
•s
E
o
O
o
o
LD
CO
o
'o
CO
CO
1
g
cf
=3
8
s~
CO
"co
"CD
_o
"3
X
o
CM
CD
CM
•^t CM
1^ ID
CM CM
ID ID
E
(0
£
"0 -^~
E i
O (0
"o 1 ^
! 1 1
S o ~
CO CD t-
O
CM
S
s,
(0
g1"
o
CM
S
i
s
_o
•s
E
o
O
-------
0
c
TO
E>
o
0
Q
0
Q.
2
y
i
M
3
(Q
>
S
^
^
0)
t»
"o
o
c4
0)
_Q
TO
u
E
o
o
0>
3
(/)
J
O
~"
(6Mn)">i
d
z
!?
O
"re
o
E
JI
O
c
=3
O
CJ
o
O)
Broome County, NY, composite soil with
CO
O)
CD
=3
Q
"CD
CO
CO
{£.
CO
CM
-3-
CO
O)
CD"
s
LO
CO
(0
(0
Q.
&*
0
-|*
CO
CD
rn
h
H
TJ
CO
CJ
3 soils; 1.2-3.4% OC - geomean value a
§
O)
1
0
LO
O
-3-
CO
CM
LO
-3-
io
s
s,
^
•^t
to
c
s
o
•s
F
o
o
o
CO
"S
CO
^3
=3
C
sorption coefficient (assume Kom) from L
CO
-C
CJ
s
§
CM
CO
CM
CO O LO
-3- LO co
l-~ t- CM
CO 1"- LO
•n
0 O
t O
3 p
o -a -S
•g TJ 5
SEC
O O
-s -s -s
o o o
o
o
1
"co
"o
£
"x
CD
cxT
03
•^
"co
"CD
1
CD
{£.
o
CM
CM
from MS thesis
r--
£
dul eta
<
LO
CO
CM
Si
CM
O
CM
CM
Sj
s,
CO
CM
a
s
o
•s
F
o
o
CO
§
?
D)
Q.
=3
TJ
"o
X
"co
jij
CJ
1
0)
, — .
§>
^
"co
CD
C
CO
D)
O
1
LO
CD
-3-
i
O)
LO
0>
(0
•g
_o
o
humic acid polymers
O)
CO
1
g
c
O
^
-3-
s
CO
CO"
LO
^
-3-
00
R
8,
^
•^t
s
c
s
o
•s
F
o
O
CO
Woodburn silt loam; fom = 0.019; 21% c
CO
O)
"CD
=3
g
O
CM
O)
t-
co
CO
^
o
O>
s
c
c
o
_o
o
Lincoln sand; 0.087% OC
—
CO
"CD
c
o
CO
g
fe
CM
-
CO
r-
CD
n
b"
g
_CD
'co
TJ
CD
2
TJ
_CD
"o
_CD
"o
CJ
to
• — •
W
t-
"co
c3
s
ofi
CO
CD
N
^=
to
CM
CM
CM
S
system specific information not given
o
?
=
0)
O
c
^
'ci
CO
CM
£
—
CD
0)
O
°s
CO
_Q
CD
N
I
to
o
CM
O
O
CM
system specific information not given
o
&
=
CD
O
C
^
'ci
CO
CM
£
—
CD
0)
O
°s
CO
_Q
CD
N
_£=
to
s
CM
O>
CM
LO
O
O
CD
CJ
_co"
"c
CD
CD
=3
CO
E
x
'co
CD
0)
CD
CO
• — •
«
t-
"co
c3
5:
ofi
CO
^3
CD
N
I
to
^
CM
O
CD
CM
O)
O
O
calculated from field data assuming foe =
9
C71
"co
CD
_CO
o:
O)
LO
CM
OJ
S
system specific information not given
o
?
=
0)
C3
c
^
'ci
CO
CM
£
—
CD
D)
O
°s
CO
^3
CD
N
I
to
CD
CM
'
field sample; 0.08% OC
• — •
CO
t-
^
CO
CD
£
°3
^=
CJ
S
CD
N
I
to
o
CM
O
g
^
CM
§
CM
S,
LO
CO
CM
CM
S
S
o
•s
F
o
O
CO
_g
TJ
c
CO
CO
1
6"
o
CM
g-
g
^=
"co
0
,
t-
CO
CM
T co
t §
s hi.
tM co
C
(0
£
1 1
1 1
_o _o
o o
CO
CD
D)
CD
CO
ro~
«J
CD
£!
x
§
t-
'
Lincoln sand; 0.087% OC
—
CO
"CD
c
o
CO
g
CO
CO
t-
S
, — .
5
CD
CO
15
Q.
top 20 cm, Eerd soil; 4.06% OC (from un
CD~
CO
O)
CO
"CD
^=
0
^
t-
O)
LO
-------
0
E
TO
E>
o
.0
0
Q.
2
i_
t/)
•>
^
0)
fi
~
0
o
csi
«
_Q
(Q
£
o
o
0
3
in
J
O
~"
K.e(L/kg)
d
,5
O
~ra
o
E
O
i
CO
Q.
c
=3
E
2
at soil ; 11.6%OC(f
Q.
O
CM
Q.
O
CD~
CO
^T
CO
CD
-C
CD
t-
CO
CO
8
r--
CD
| 1
0 c
£ 0
O •£-
ID
h-
t-
N.
S,
CM
h-
t-
PO
E
s
o
•s
F
o
O
c3
=3
CO
T3
efrom collected mea
=3
CO
CD
C3>
CD
CO
5T
CO
O
^5
CJ
§
•^t
S
00_
C3)
^
CO
CJ
CD
ID
285,467
CO
CD
CM
O
ID
1-
Q
Q
O
s?
CM
CO
CM
CO
CO
CO
CD
-z.
q-
^
c
C3)
C3>
^
-J
CO
CJ
0
ID
496,476
diment; 4.39% OM
CO
s
c
^c
q-
^
c
D)
C3)
;=
^u
CO
CJ
^
ID
589,537
O
LD
CO
"c
CD
CO
c
^c
q-
^
c
D)
C3)
;=
^u
CO
CJ
CO
ID
747,887
5% OM
CM
'o
CO
"co
CJ
q-
^
c
C3)
C3)
^
-J
CO
CJ
ID
CD
ID
892,067
O
U3
CD
'o
CO
E
q-
^
c
C3)
C3)
^
-J
CO
CJ
CM
CD
CD
ID
S
ID
S
i
s,
CO
CO
ID
677,934
S
S
o
•s
F
o
O
OC (EPA-20)
g
CO
o
CO
g
CD
CO
CD
CD
CO
X
§
CD
CO
CD
CO
CO
^
ID
ID
565,014
CO
°
CO
ID
0
O
(0
£
•s
JH*
to"
¥
0
-Q
Q
2.28% OC (EPA-5)
c
CO
Q
-z.
§
CD
CO
CD
CD
CO
_L
CD
CO
m
CO
CO
^
CD
ID
805,292
CO
CM
Q_
O
o
CO
CO
"c
CD
0
CO
g
CD
CO
CD
CD
CO
_L
CD
CO
m
CO
CO
^
CD
ID
§
CO"
0.15%OC(EPA-8)
"c
CD
!S
g
CD
CO
"CD
CD
CO
_L
CD
CO
CO
CO
^
^
CD
1,172,847
CD
Q_
LU_
O
o
s
CM
c
CO
CJ
§
CD
CO
"CD
CD
CO
_L
CD
CO
CO
CO
^
CO
CM
CD
1,690,971
; 2.07% OC (EPA-4)
c
CD
0
o
g
CD
CO
o
CD
CO
_L
CD
CO
CO
CO
^
CO
CM
CD
1 ,687,404
Q_
O
O
CD
8
o
§
CD
CO
o
CD
CO
J_
CD
CO
CO
CO
^
CD
CD
2,277,875
c\T
CM
Q.
O
o
h-
CD
"c
CD
OJ
CO
g
CD
CO
o
CD
CO
J_
CD
CO
CO
CO
^
S
CD
2,383,765
0.72% OC (EPA-6)
Bdiment
0)
Q
g
CD
CO
o
CD
CO
_L
CD
CO
CO
CO
^
CM
CD
2,622,453
0.95%OC(EPA-15)
"c
CO
g
CD
CO
o
CD
CO
J_
CD
CO
CO
CO
^
CO
CD
2,663,317
CD
CM
Q_
O
o
CO
"c
OJ
0
CO
g
CD
CO
"CD
o
CO
_L
CD
CO
CO
CO
^
CO
CD
O
I--
CD
CD
CM"
CM
Q_
O
o
CO
CO
"c
cu
OJ
CO
g
CD
CO
o
cu
CO
_L
CD
CO
CO
CO
^
^
CD
2,962,603
Q_
O
o
CD
'o
g
CD
CO
o
cu
CO
_L
CD
CO
CO
CO
^
§
CD
3,020,262
-------
TO
E>
o
u
Q.
2
>
x°
0)
t>
"o
o
Q)
A
TO
o>
o
o
01
3
$.
i
o
I
0
z
ta
o
S~
__
1
O
KY sediment; 0.66% OC (EPA-18)
CO
Oi
^
"CD
"CD
CD
CO
CD
X
I
Means et al.
CD
LD
CM
-3-
Oi"
LD
O
co"
CO
6
h-
co
LD
E
d>
£
C
RJ
rf ¥
*rr =3
c .E
5 o
Q •£-
CO
CD
5
1"
O
c*T
|
1
CM
CD
O
i
N.
*"
C
1
O
•s
E
o
u
O
8
52
CO
o
o
N
'El
O
_£=
CD
'o
CO
"CD
n
CD
"CD
"c
=3
5T
CO
rr,
Leeetal. (1!
CO
CM
I--
Q
S
o>
^~,
o
I
c
g
o
u
Q
CN|
v"
O
o
O)
LD
CM
O
N
O
'o
CO
T3
"CD
"c
=3
5T
CO
"CD
"CD
LD
CM
O
CO
CM
CO
=3
'co
QJ
O
CD
r?
Willamette silt loam; 0.928% OC; 3.5 di
O)
r--
O)
Chiou et al. I
CM
o
CO
CD
o
5?
CM
'co
CO
CD
Woodburn silt loam; 1.9% OM; 9% san>
co~
CO
Chiou et al. I
LD
CM
s
CO
Appalachee sorbent; 1.4% OC; pH = 6.
CD
§>
&
"o
CD
Stauffer & M
a>
LD
CM
CD
s
II
I
Q.
6"
o
1
O
CO
CD
a
Q_
CD
§>
C
"o
CD
Stauffer & M
CM
CO
CO
peaty soil; 29% OM
1
Friesel et al.
oi
CD
CM
LD
CO
sediment, 0.19% OC; avg. 2 values
CO
0
OJ
OJ
c
Piwoni & Ba
o
r--
CM
r--
1
foe reported as a range; avg. of severa
i
"CD
"CD
"CD
1
LL
CM
CM
O)
CM
LD
LD
CM
0
U
U)
(0
1
CO
LD
CM
R
PO
C
(0
u
u
•c
•5
E
0
CD
O
Woodburn silt loam; fom = 0.019; 21%
CO
s
Chiou et al. (
CM
CO
r--
CM
^
CD
&
O
^-,
3
o>
0
o>
Q
"o
Q
«-"
CD
0
I
6"
o
CM
i
'o
CO
"c
8
CD~
S
OJ
1
0«
Southworth <
LD
CM
§
CM
C
o
N
'El
O
(-
CM
CD
8
CM
LD
'o
CO
c
CD
0
£="
"o
O
co~
CO
Hutzleretal.
CO
CM
S
CO
Lincoln sand; 0.087% OC
CO
Oi
Wilson et al.
o
CD
CM
CO
reported as Kom; foe given as a range
1
"CD
"CD
"CD
1
LL
CO
CD
CM
O>
CM
KS1 field sample; 0.73% OC
CO
O)
_
"co
ch&W
CD
Schwarzenb
CO
r-
CM
CO
o
CD
Apison soil with 0.1 1 % OC; 86% clay
CD
OJ
1
oK
Southworth <
CM
CO
CM
LD
CD
CD
CO
'o
CO
0
CD
CD
if
o
1
co~
CO
Oi
Hutzleretal.
LD
CO
CM
§
T3
system specific information not provide
CD
0
c
T3
'o
CO
CD
CO
O)
OJ
ch &Gi
CD
Schwarzenb
CD
CO
CM
r-.
LD
IT
0
O
II
8
CD
"CD
'5
cr
CD
"c
2
=3
CO
1
X
'co
•5
OJ
CD
CD
CO
O)
—
a
CO
ch&W
CD
Schwarzenb
r--
CO
CM
CO
CO
h-
humic acid polymers
O)
CO
O)
fii
Chin&Web(
CM
CM
CO
Fullerton soil with 0.06% OC
CD
OJ
1
oK
Southworth <
CO
CM
s
CO
c
top 20 cm of Eerd soil; 0.06% OC; colu
CD
Loch et al. ('
CD
CM
^
Charles River sediment; 8.5% OC
CD~
00
T3
C
?
Wu & Gschv
o
CO
CM
T3
system specific information not provide
I
CD
0
C
T3
'o
CO
CD
CO
O)
OJ
C3)
ch &Gi
CD
Schwarzenb
o
CO
LD
-------
(/)
O
'E
TO
E>
o
o
.Q
o
Q.
2
^
>
i_
M
0)
3
1
^^
*°
1
t3
«
O
O
c4
*
_0>
rt
^J
(Q
(A
U
E
o
O
d>
y
3
O
2
I
J
CAS No.
ro
u
E
u
£
O
I
"o
O
CD
CO
'o
CO
cT
O>
03
^7"
"co
"CD
£>
=3
CD
CD
CD
-3-
CO
•^t
CO
LD
r--
0
c
re
.c
o
j>
.c
u
a
8
*^-
q
•^t
E
£
"o
CO
c
CO
E
£
"O
_CD
"13
Q.
O
O
pr
CO
0>
^-
"co
"CD
>>
o
C£
O>
r-
CM
CD
CO
I--
S
a>
o
re
rf
CO
r--
PO
10
s
o
•s
E
o
o
T3
.ffi
O
CD
CD
CO
'o
CO
g
O)
^
"co
"CD
£>
=3
— 3
«
SI
107-06-2
g
re
£
•5
0
_o
£
u
(5
CNj^
CO
=3
'co
0
CO
CD
CD
CD
T3
O
CM
8
^
CO
CM
O)
CD
E
CO
_o
'co
£
"CD
_co
i
?
O)
^3-
"co
"CD
=3
O
^
O
CM
LD
CO
CO
O
O
N°
0^
I1--
CO
o
CD
~b"
c
CO
CO
Lincoln
1
11 '
"co
Ison e
5
CO
CO
CD
r--
s
5
8,
re
S1"
<^
CO
LD
00
PO
S
_o
•s
E
o
O
&
CO
T3
O
O
0
CO
ccT
s
^
_CD
i
^=
o
0)
CO
LD
CD
•^t
LD
CO
ID
r--
i
u
£
•s
S
°
£
u
5
^
v"
CO
U)
to
1
<
CO
Ul
to
c
re
u
u
•E
"5
E
0
E
CM
V
0
O
&
CD
"co
•E=
_CD
"co
£
CT
CO
>>
T3
1
1
CO
1—
1
o
CO
lY
=3
CO
CD
CO
CO
3
m
CO
LD
CO
CO
CM LD
d) CD
LD CD
CD CD
LD LD
0
c
u u
0 £
£ 1
U £-
S ^
1 £
" a
5 N
CN| ,
V c
V) (Q
5 £
CO
LD
S?
U
U)
re
u
5
CO
ID
00
PO
s
o
•6
E
o
at
O
CO
=3
'co
S
CO
CD
CD
D)
®
-o
O
CM
0.928% OC;
E
Willamette silt loa
O)
r--
O)
11 — '
"co
"CD
=3
O
!c
O
E?
^
LD
r-L
CO
«,
r-.
0
c
re
a.
O
a.
S
o
£
U
a
CNI
v"
E?
N.
•t
1
<
E?
N.
•t
S
O
•s
E
o
u
O
g
"Q_
tr
o
CO
CD
CO
CO
_£=
Q.
O
Q.
CO
>
mputed from
8
_co"
'o
CO
CO
en
>
CO
CO
'o
o
[-•-.
O)
2
"co
5
CO
CO
a
<*?
LD
T
CM
S
g
0)
a
0
a
0
_o
2
0
a
c\
c
g
"o.
o
CO
CD
CO
CO
Q_
o
Q.
$
E
Computed fro
co"
'o
CO
CO
di
>
CO
CO
I
o
[--»
O)
2
"co
5
5
CD
CM
S
E
CD
-C
O
Q
CD"
o
=3
O
CO
"O
CD
m. (unpublish
n
trans-, personal c
CM
g
:^
"co
"CD
CD
Q.
Waucho
LD
CM
CO
?
N.
CM
U
U)
re
u
S.
CO
•^t
N.
CM
C
re
u
u
•c
"5
E
0
o
0^
LD
CM
E
Beverly sandy loa
O)
^;
"co
"CD
°
CO
^1
0)
"
0
CO
co"
CM
te
3
^
•C
d)
Q
O
0^
r-
Plainfield sand; 0.
O)
^;
"co
"CD
Sharom
CM
^
0
CD"
CM
O
0^
CO
CM
"c
Big Creek sedime
O)
^;
"co
"CD
Sharom
|
s
CO
h-"
CM
5
^
S
ifT
CM
1
<
5
to
S
ifT
CM
S
O
•s
E
o
u
O
-------
0
C
n
E>
o
.0
0
Q.
2
•o
0)
_3
>
S
^
0)
•c
Q)
"5
o
CM
0)
.0
n
0>
E
o
o
o>
£
3
5
J
O
K»(L/kg)
d
o
"re
u
E
O
E
=3
O
o
O
O
0^
CO
ID
_£=
County, NY, composite so
O3
CD
CO
CD
^
"CD
CO
a
s
CM
CD
S
1
2
Q.
^
"5
Q
CD
d>
b"
F
NY, sand; 0.26% OC; colu
c
c
o
O
CD
CO
11
^
"CD
CO
0,
^
.
CM
rn
S
8,
CM
rn
CM
00
c
5
o
•s
F
o
O
TJ
oil with 0.2% OC; column st
CO
"co
CD
s~
CO
03
"CD
Q.
(/)
88
o
CD
CO
Y
1
re
.c
Q.
0
E
Q
TJ
CO
C
F
=3
8
6"
O
CM
CM
i
'o
CO
s
"5
o
D)
03
CD
£
03
"CD
Q.
CO
(0
CO
CM
TJ
oil with 3.7% OC; column st
CO
"co
CD
s~
CO
03
"CD
Q.
CD
m
r--
CM
h-
co
*
s,
CO
^
0)
CM
C
5
o
•s
F
o
O
E
=3
O
o
O
O
0^
CO
ID
_C
County, NY, composite so
O3
CD
CO
CD
^
"CD
CO
{£.
^
CO
s
CO
CM
^!
s
s
1
5-
o
C
Q
F
=3
8
CO
"03
D)
2-
NY, sand; 0.26% OC; adsc
Conklin
CD
c^
m
it
^
"CD
CO
££
ID
CO
ID
I--
r-
o
CO
8
S,
O
CO
N.
8
ill
s
o
•s
F
o
O
=3
03
TJ
CD
CO
3
1
2
CD
=3
03
D)
CD
5T
CO
CO
C
0
CO
CO
CM"
CM CM O h-
'M- O ^ CO
t- CM CO CM
t- CD h- ID
CM O t- t-
f- CO f~ f~
0
g g n
3 S |
o o a.
11^-1
s s S -5
.E .E o g
a a o
4 >
03
52
5
8
CD
'o
CO
'03
O
•B
untreat
CO
%
^ — '
"03
"CD
CM
s
c
N
O
•£=
m
03
O
CO
CD
3d Oshtemo soil; 0.11% OC
untreat
CO
%
^ — '
"03
"CD
CO
CM
CO
cyanopropyl column
6"
Q_
X
00"
CO
03
§
_,
C
o
CO
X
CO
ro
CM
O
CM
-------
n
E>
o
u
Q.
2
(/)
_3
>
x°
0)
t>
"o
o
•E
Commer
3
$.
J
O
5
*
d
z
U)
o
—
0
£
O
CO
s
CO
"c
CO
nent; 4.02% OC; 0.2%
Tamar estuary sedir
CO
=3
0
CD
CD
^
CM
LD
LD
^.
t
g
benzene
nued)
f S
LU >Si
s
CM
N.
1
g
CO
CM
S
S
s
_o
Hi
F
o
o>
O
\C (humic acids)
RP&HPLCon PIH/
Oi
"
0)
s
CO
CM
CD
*
CO
^
O
t
0
CM
§!
U
(0
o
3
LJ.
87% sand; foe = 0.01?
CD
Q.
CD
CO
&
=3
CT
CD
"c
LL
CD
S
c
(T
=3
T3
_Q
^
*
CO
LD
LD
\C (humic acids)
RP&HPLConCIH>
cn
"
CD
n
S
m
^~
*
S
cn
LD
cn
CD
*
3
1
g
cn
CD
*
to
§
s
o
"0
F
o
o>
O
CM
O
O
cn
d
co"
CD
=3
s
CM
ID
Borden aquifer mate
CD
cn
"— '
CD
CD
=3
T3
g
CO
cn
CO
cn
CO
h-
co
r--
CD
CO
o>
E
d>
O
3
LJ.
1
"CD
CO
D)
Flint aquifer materia
CD
rn
—
CD
CD
CD
CD
CO
LD
CD
1
"CD
CO
d>
ID
Warren aquifer mate
CO
s
—
CO
•3
LD
CO
CD
I--
LD
LD
o
humic materials (DC
CO
CO
*-
CO
°S
CD
CD
O
LD
cn
CO
CO
cn
CO
(humic acids)
RP-HPLC on CIHAC
0
cn
11 '
CD
CD
CD
N
CO
LD
^
LD
CM
(humic acids)
RP-HPLC on PIHAC
0
cn
11 '
CD
CD
CD
N
CO
CM
^
CO
LD
cn
CO
§
oo"
u
U)
(0
1
cn
CO
CO
N.
K"
c
re
u
S
u
•-
Hi
F
0
1
O
cn
CD
=3
CO
CO
CO
o
CO
CD
CO
t
_o
u
•5.
X
CO
=3
collected measured va
average value from
, ,
^
c
y
=3
o
CM
^
O
CO
o
o
^
1
0
u
U)
(0
1
CO
cn
CO
00
CM
oT
c
(0
u
u
•-
F
0
.5% OC
GA sediments; 0.5-1
LD
CO
cn
•^
o
ofi
y-
^=
CD
O
cn
^
o
s
CO
CO ^_
1 S
° ^
* 1
1 §
Q. U
^ O
3 1
Q. X
X X
o
cn
^
|
=0
u
(0
1
o
cn
^
|
=0
c
re
u
u
•-
Hi
F
0
o
55
CM
CD
1
0
is
_CD
single measuremen
cn
cn
c
CD
CD
c
=3
^
CO
03
CD
O
CO
CM
s
co "9
! I
m S
g c
.2 S
43 £
A S,
" 8
2 ° o
o ^ x
£ CB ^
g I "§-
I B A
O
g
CM
1
0
is
_CD
single measuremen
cn
cn
c
CD
m
c
=3
*=
C/J
=0
CD
O
CO
CO
LD
CM
O
CD
CM
'o
CO
CD
single measuremen
cn
cn
c
CD
m
c
=3
CO
=0
CD
CM
CO
O
CO
CO
o
CD
CM
'o
CO
>>
T3
1
single measuremen
cn
cn
c
CD
m
c
=3
CO
=0
CD
-3-
CO
CD
O
LD
d
'o
CO
"CD
=3
"CD
single measuremen
cn
cn
c
CD
m
c
=3
*=
C/J
=0
CD
cn
CO
CM
CO
LD
-------
I/)
.0
'
E>
o
u
Q.
2
U)
(1)
I
0)
t»
_0)
"o
o
_0)
n
J2
Commen
u
3
J
O
'S
^
^
0
z
^
o
—
u
E
JI
o
^
lateritic soil; 2.88% Oh
c
2
e meas
c
CO
CD
c
CO
CO
c
=3
OJ
art
TJ
I
o
CO
S
o
CM
CD
CD
CO
c
o
u
ff
o
o
•c "S
y uj
O -P
I 0
b ,£.
^
lateritic soil; 1.00% Oh
urement
e meas
c
CO
CD
c
CO
CO
c
=3
$
art
TJ
I
CO
CO
-3-
CM
O
CM
^.
O
CM
ID
ID
'o
CO
"co
jrement
e meas
C3>
C
CO
CD
C
CO
CO
c
=3
tn
ae
TJ
I
CM
CO
CO
O
CM
Kari soil; 8.21 % OM
jrement
e meas
C3>
C
CO
CD
C
CO
CO
c
=3
0
rn
TJ
I
CO
CO
CO
CO
CM
CM
•==:
lateritic soil; 0.60% Oh
c
2
e meas
C3>
C
CO
CD
C
CO
CO
c
=3
0
art
TJ
I
S
CO
CO
CD
CM
•==:
lateritic soil; 0.92% Oh
c
2
e meas
C3>
C
CO
CD
C
CO
CO
c
=3
0
rn
TJ
I
s
CO
8
CM
CM
alluvial soil; 0.70% Ol\
urement
e meas
C3>
C
CO
CD
c
CO
CO
c
=3
0
C
CO
CD
c
CO
CO
c
=3
0
art
TJ
I
g
CO
g
l-|-
ID
CO
CD
CO
0
c
s
u
o
u
o *—.
£ O
0 X
o .2
i 2
CO. £-
^
lateritic soil; 1.62% Oh
jrement
e meas
C3>
C
CO
CD
c
CO
CO
c
=3
0
(ft
art
TJ
I
r--
co
f^
Kari soil; 24.6% OM
jrement
e meas
C3>
CO
CD
c
CO
CO
c
=3
0
rn
ae
TJ
I
CO
CM
CO
CO
CD
i
0
II
c
6"
0
0
DC; 50% montmorilloni
ID
CO
CO
o
o
TJ
C
CD
CD
C3>
C3>
CD
art
I
ID
CM
CO
h-
sandysoil; 12.6% OM
jrement
e meas
C3>
CO
CD
C
CO
CO
c
=3
tn
ae
TJ
I
CD
CM
CO
CM
CO
•==:
lateritic soil; 1.00% Oh
jrement
e meas
c
CO
CD
c
CO
CO
c
=3
0
CO
CD
c
CO
CO
c
=3
0
rn
ae
TJ
I
s
CO
ID
CM
•==:
lateritic soil; 0.60% Oh
jrement
e meas
c
CO
CD
c
CO
CO
c
=3
0
rn
ae
TJ
I
^
CO
3
CM
•==:
lateritic soil; 0.92% Oh
jrement
e meas
c
CO
CD
c
CO
CO
c
=3
rn
ae
TJ
I
CO
CO
fe
CD
CM
^
lateritic soil; 2.88% Oh
jrement
e meas
c
CO
CD
c
CO
CO
c
=3
0
art
TJ
I
g
CO
CO
^
CO
^-
o
S?
CM
ID
ID
'o
CO
"co
O
Q.
jrement
e meas
C3>
CO
CD
c
CO
CO
c
=3
0
rn
ae
TJ
I
g
CO
CO
ID
CO
O
CD
O
2.8%OC;20°C; 1/n =
0
=3
"co
c
CO
CO
CD
CD
C3>
CD
art
I
ID
ID
CO
CO
CD
ID
CO
ID
CO
CO
5
*
S,
(0
g1"
CO
CO
CO
0)
f)
«
E
i
u
•s
F
o
O
CO
o
CD
II
c
o
N
O
'co
OJ
1
3
«i
CO
CD
5
§
CM
CO
CD
CD
CO
CO
ID
C
(0
u
o
8-
O
o
o ^
X -S
I .£
^- ^--
o
o
o
CM
s
CD
_co"
hree so
0
C3>
CO
CO
w
CD
CO
"co
1
£5
CM
ID
CO
.0038; couimn study
o
I
'co
CO
CJ
c\T
r-
"CD
(!)
c
OJ
=3
X
8
CM
I--
ID
II
8
CD"
CO
X
Q.
1
C
CO
Q.
o
3,
1
0
g
CD
•^
"CD
CO
§
CM
O
1°
.,_
u
CO
CO
'o
CO
o
0
g
CD
c
CO
=3
0
rn
art
TJ
I
§
CM
o
1°
CO
s
CD
II
8
O
N
'El
O
-C
5/ 'ojs eu
-------
I/)
.0
'
E>
o
o
I
0)
t»
_0)
"o
o
_0)
n
£
Commen
o>
3
J
O
3>
^
*
0
^
O
E
JI
O
Kranzburg sici, B horizon ; foe = 0.01
fe
Adams & Li (
CO
O)
CM
LO
O)
O>
CO
LO
s
o
o
o1
o
o ~
O u
Svea sil, B horizon ; foe = 0.008
fe
Adams & Li (
^
CM
CO
CO
cn
o
n
8
LO
^1
Q.
1
C
ro
^=
Q.
_g
1
TJ
C
ro
CO
rn
"ro
"S
CO
CO
CM
O
X
Muck (38% OC)
I
o
Q]
s
CM
CO
X
CO
CM
O
CD
O
c
o
N
O
.£=
'co
"3
ro
LL
ft
Adams & Li (
S
CM
LO
X
Honeywood loam (2.1% OC)
I
0
ED
s
CM
CD
X
Canisteo cl, B horizon ; foe = 0.006
ft
Adams & Li (
0
CO
LO
0
o
LO
1-^
II
T
Q.
u
Alfisol; Udalf, Para brown earth; 0.76% OC
c\T
s
ro
"3
1
Q_
i±
0
CO
o
o
Fox loamy sand (1.7% OC)
I
Kay & EIrick
0
CO
ro
O
Lester fsl, A horizon ; foe = 0.023
S
Adams & Li (
CO
0
CO
^
o
Pachappa sandy loam; foe = 0.005
c\T
O)
"ro
"3
Huggenberge
^
CO
CO
0
Beverly sand loam; 2.5% OM
§
Oi
Sharom et al
S
CO
CO
0
single measurement; sandy soil; 12.6% OC
O)
•. .
nunatha
Wahid & Set
^
CO
%
Zimmerman s, A horizon ; foe = 0.007
S
Adams & Li (
LO
0
CO
LO
Kentwood sandy loam; foe = 0.0093
c\T
O)
"ro
"S
D)
1
D)
D)
=3
HI
LO
0
CO
8
CM
CO
O
O
O
1
CD
CM
LO
II
^1
Q.
'E
_g
"c
ro
o
D)
S
"ro
"3
CO
k
CO
CO
8
CM
^
single measurement; lateritic soil; 1.27% O
Oi
— ,
c
Tunatha
Wahid & Set
CO
CO
s
CM
Brookstone sandy loam (1.9% OC)
i--
o
Q]
0,
CO
r--
CM
Brainerd fsl, A horizon ; foe = 0.026
rn
Adams & Li (
o
CO
CO
(0
CM
Bearden sil, B horizon ; foe = 0.002
rn
Adams & Li (
^_
CO
j*
CM
LO
O
o
^
CO"
LO
II
^1
Q.
'o
CO
'E
_g
o
"c
o
ro
o
D)
rn
"ro
"S
CO
^_
CO
8
CO
"ro
TJ
"ro
c
D)
'El
O
OJ
ro
OJ
1
S
CO
"o
co~
CO
O)
ro
"S
"ro
1
^_
CO
8
CO
Ann Arbor soil; 1.14%OC
S
Miller & Web
CM
CO
CO
CO
Lubbeek II soil; 0.53% OC; 10% clay
O)
i
TJ
ro
CD
c
o
CM
CO
CM
CO
Lubbeek I soil; 0.07% OC; 2% clay
O)
i
TJ
ro
CD
c
Moreale & va
CO
CO
LO
ro
CO
3.1%OC;CaVenadoclay
S
Mills &Bigga
CD
CO
CM
10
-it
Brainerd si, B horizon ; foe = 0.001
fe
Adams & Li (
CD
CO
LO
10
-it
Delta soil; 0.12% TOG
CO
O)
Miller & Web
CD
CO
CO
10
-it
Big Creek sediment; 2.8% OM
£
O)
Sharom et al
r--
CO
S
•"t
CJ
c
'E
LO
6"
O
CM
LO
LO
1
O)
•. .
c
lunatha
Wahid & Set
CO
CO
LO
LO
Pokkali soil; 5.52% OM; 1:1
O)
•. .
c
nunatha
Wahid & Set
o
CO
rn
LO
single measurement; Kari soil; 24.6% OM
O)
•. .
lunatha
Wahid & Set
CO
CO
rn
CD
^
o
1
'o
CO
0
"c
0
0
=3
CO
E
0
D)
C
'co
O)
•. .
c
nunatha
Wahid & Set
^.
CO
h-
Blue Earth sil, A horizon ; foe = 0.1 1
fe
Adams & Li (
CD
CO
ro
CO
Powerville sediment (NJ); 2-4% OC
LO
Caron et al. (
r--
CO
O)
I0
CO
^
o
1
CM
'o
CO
0
"c
OJ
OJ
=3
CO
E
OJ
D)
C
'co
O)
•. .
c
nunatha
Wahid & Set
CO
CO
rn
CO
single measurement; alluvial soil; 0.70% Oh
O)
•. .
c
nunatha
Wahid & Set
O)
CO
LO
ro
O)
Plainfield sand; 0.7% OM
§
O)
Sharom et al
O)
CO
^
O)
^
single measurement; lateritic soil; 1 .62% O
O)
•. .
c
nunatha
Wahid & Set
0
CO
s
Oi
organic soil; 75.3% OM
S
O)
Sharom et al
ro
CO
CO
I0
o
^
single measurement; Pokkali soil; 5.52% O
O)
•. .
nunatha
Wahid & Set
CM
ro
CO
rn
O
single measurement; alluvial soil; 0.75% Oh
O)
•. .
c
nunatha
Wahid & Set
CO
ro
CO
CM
-------
(/)
O
'E
n
E>
O
o
o
Q.
|
I_
I/)
0)
_3
(0
>
o
•o
Qj
O
"o
o
i
_0)
^
(Q
(A
ni
Commt
u
3
O
(0
8
0
|
J
O
z
(0
o
"J5
U
1
U
o
O
CM
CO
"ement; Kari soi
e measui
D)
c
'co
5T
CO
^1-
c
CD
(-
"CD
c
=3
^
"CD
Wahid & S.
CO
CO
CO
CO
CM
CM
CO
CO
CO
CO
ID
U
1
U
£
_o
(J
y-Hexachlorocyi
(continued)
O
Q
• r
Q-
O
CD
TJ
CD
CO
CD
Q.
O
Q.
E
2
g
s?
ID
CO
O
1
'co
_CD
O
o~
r-
co
^
To
O
Spencer &
ID
CO
CO
O
CD
CM
CM
CO
CO
O
O
II
, B horizon ; foe
Earth sil,
i
CD
?
CO
^
! Adams & L
CD
CO
CO
CO
CD
CM
CM
O
o
'o
CO
o
_2
j-r
c
CD
CD
e measui
D)
c
'co
5T
CO
^1-
c
CD
^=
"CD
c
=3
"CD
Wahid &S
CO
CO
g
CM
CM
O
S?
§
o
'o
CO
o
_o
j-r
c
CD
CD
e measui
D)
c
'co
5T
CO
^1-
c
CD
^=
"CD
c
=3
"CD
Wahid &S
CO
CO
CO
*
CM
taten peaty muc
0)
8
CO
•<-
CO
%
^
CD
n)
Mills & Big.
5
CO
s
ID
CM
O
CO
o
aywe soi
_c
o
^
°0
CO
(U
^3
m
Miller &W(
CM
CO
I
CM
CD
O
8
CO
O
'o
CO
CD
TJ
O
N
oT
CO
^3,
CD
TJ
CD
CD
c
CD
>
Moreale &
CO
CO
o
CM
c
o
•^
2
n
5 min equlil
O
S?
ID
O
'o
CO
"CD
=3
"CD
co"
CO
^1-
c
CD
(~
"CD
c
=3
"CD
Wahid &S
^
CO
CD
CM
CO
CM
CD
O
CD
CM
O
CD
1
mette sill
CD
§
°f
CO
^
Chiou etal
^
CO
CO
CO
CO
CM
CO
0
o
II
1
o
-C
CD
"co
c
_CD
^
ff
CO
^
[Adams & L
ID
CO
CM
CO
I--
CO
fc
T^
1
g
<
CO
CO
CM
*
s
s
_o
•3
o
0>
O
!•!• C\l S 0)
^- i--
c
U
'S <1>
re =
c £
Q. U Q.
— re "~~ "
S 2 tJ | <5
^ ^ ?" o ">
o o 3 2 "o
1 1 1 tl
CO
c
g
"o
*t
TJ
CD
CO
ill coarse silt
i and Hickory H
=3
tx
CD
O
Q
?
CD
CO
I--
CO
"CD
"CD
'o
^=
^:
o
•E=
CD
^.
O
CO
•*t
o
8
CO
ID
CO
Methoxychlor
o
CO
•^t
1
00
U
o
(0
1
o
CO
•*t
1
00
c
(0
U
s
U
•e
•5
E
0
_co
sral chemica
a value for sev<
CO
CD
I
"c
CD
2
Q.
CO
CO
CO
D)
D)
m
ID
CO
o
CO
CO
CO
CO
r--
Methyl bromide
(Bromomethane)
ID
CO
o
0)
U
U)
(0
1
ID
CO
O
0)
c
(0
U
s
U
£
E
0
1
'o
CO
g.
CO
£
"CD
"CD
=3
CO
O
CD
CM
CO
O
ID
r--
Methyl chloride
(chloromethane)
CO
r--
d
to
U
U)
(0
1
CO
I--
o
to
c
(0
U
s
U
£
E
0
I
a
CD
CO
ID
CO
CO
^3-
"CD
"CD
CO
CD
'c
CD
Q
O
O
t-1
o
CM
CO
cp
ID
r--
•8
•^ CD
Methylene chloi
(Dichloromethan
o
o
t-1
u
U)
(0
1
o
o
t-1
c
(0
U
s
U
£
E
0
-------
I/)
.0
'E
TO
E>
o
o
Q.
2
I
0)
t»
_0)
"o
o
_0)
ra
1
Comme
0
3
J
O
3>
5
*
d
z
^
o
"re
o
E
O
8
LO
"o
D)
s
CD
~03
"S
CO
CM
CD
CM
O
CO
CO
CO
o
CM
^L
CD
0
C
0
(0
.c
a.
z
03
CO
03
CO
si
c
CO
Tamar estuary sediment; 4.02% OC; 0.2%
r--
CD
£
Vowles & MantoL
CO
CD
CM
CO
S
CO
03
03
CO
D)
03
_co
"c
0
1
1
'o
CO
Karickhoff (1981)
^
CM
te
CO
II
I
Q.
8
0^
1
1
O
03
03
Q_
Q.
S~
CD
£
Stauffer & Macln
CO
CM
fe
o
03
03
CO
_co
"5
"c
1
X.
1
CO
15
=3
Q.
C
=3
CO
CD
CO
Hodson & Willian
CO
CM
CM
CD
Fullerton soil; 0.06% OC
CD
CD
o
Southworth & Ke
CO
CM
O
Apison soil; 0.1 1%OC
CD
CD
_gj
Southworth & Ke
8
CO
§
o
Q
Oi
Kan & Tomson (1
S
CO
§
humic acid polymers
S"
?
S
McCarthy & Jimir
S
CO
1
avg. 10 values
0
o
E
CO
CD
8
CD
CD
"o
CO
c\T
~03
"3
1
Q.
^
CO
s
CM
_CO
average of Doe Run & Hickory Hill sedime
rn
CD
Karickhoff et al. (
^
CO
s
CO
Mississippi River sediment; foe = 0.015
Karickhoff (1982)
CM
CO
CO
CO
CO
Menlo Park soil; 1.6%OC
rn
Podolletal. (198
LO
CO
8
RP-HPLC on CIHAC (humic acids)
'nT
0
Szaboetal. (199
i^)
CO
CO
O
O
1
"c
LU
c\T
~03
"3
1
Q_
CD
CO
j
RP-HPLC on PIHAC (humic acids)
'oT
0
Szaboetal. (199
CD
CO
LO
Speyer soil, 0.15-0.5 mm; 1.12%OC
c\T
CD
~03
"3
1
Q.
CM
CO
O
CO_
8
CO
CD
'o
CO
^
CD
"03
"S
1
03
CD
CM
CO
CD
Eustis sand; 0.74% OC; batch & column d
„
CD
~03
"3
E
1
CD
CM
CO
§
§
CO
K
s,
(0
>
§
CO
0)
t-
c
i
o
•s
F
o
O
forest soil with 0.2% OC; column study
E
CD
~03
"S
Q.
C7)
CD
CD
CD
CO
CO
LO
CO
CD
i
O
z
Lincoln sand; 0.087% OC
"
~03
"S
c
o
CO
g
CO
s
average for four soils; 0.6-2.5% OC
Briggs(1981)
CO
CD
§
>,
agricultural soil with 2.2% OC; column stuc
CO
S
~03
"S
Q.
C7)
LO
CD
CO
CO
CO
forest soil with 3.7% OC; column study
E
CD
~03
"S
Q.
C7)
O
CM
CO
O
CM
S
CD
II
1
1
Q
„
CD
Miller & Weber (1
LO
CM
CM
CO
s>
Gribskov B horizon soil; 2.58%OC; avg. 2
S
CD
0
O
CO
CM
CM
O
CD
Lincoln sand; 0.087% OC
"
~03
"S
c
o
CO
g
CO
CM
CM
CD
O
03
unpublished experimental results by same
CO
CD
CO
Hodson & Willian
CO
CM
CD
CM
CM
CO
CD
Gribskov C horizon soil; 1.82%OC; avg. 2
o
CO
CM
O
CM
LO
CM
^
3>
(0
>
§
CM
0)
t-
c
i
o
•s
F
o
O
Iowa soil; 2.1% OC
E
&
Wu & Gschwend
E
CO
CO
LO
CO
CD
0
CD
0
C
S
o
o
o
(0
"c
Q.
North River sediments; approx. 4.4% OC
CD
CO
Wu & Gschwend
LO
LO
LO
LO
LO"
(1
o
CM
O
S
0
o
_co
"c
0
OJ
=3
CO
E
CM
C3)
03
'5
cr
03
c
03
CO
c\T
?
"03
"S
OJ
03
CD
S
i
LO
CO"
GA sediments; 0.5-1.5% OC
03
8
CO
Karickhoff & Mor
i
S
O
o
-------
I/)
.0
'
E>
o
o
Q.
2
(1)
I
0)
t»
_0)
"o
o
_0)
n
0
£
o
o
0
3
3.
i
o
5
5
d
ff
o
~ra
o
1
o
o
55
LO
CO
X
2
Q_
Q.
CO
"c
E
E
CO
if
1
CO
O
en
TJ
C
1
0
13
„.
^
S
LO"
LO
LO
CO
en
§'
CD
0
C
1
ja
0
O T?
Z ®
RJ .—
1 1
a. &
S
^
S
to"
s,
RJ
g1"
LO
^
00
8
s
S
o
•s
E
o
O
IL sediment with 2.38% OC (EPA-23)
§' '
en
"co
"3
CO
ss
s
^
&
CO
CO
p
CD
p
en
CM
0
C
^
o.
aquifer material, 1.05% OC
r--
g
"co
"3
TJ
^3
§
^
§
r-
LO
IL sediment with 1.67% OC (EPA-22)
§' '
en
"co
•3
CO
s
S
^
CM
5
ND sediment with 2.28% OC (EPA-5)
o
CO
en
"co
"3
CO
s
o
^
g
CD
g
ND sediment with 2.07% OC (EPA-4)
o
CO
en
"co
"3
CO
s
,_
^
s
LO
1_
"co
s
"o
C
CO
_c
"3
Webster soil; 2.23% OC; 14 C; 30:70 m
en
CO
en
—
dburn et .
0
1
^
^
g
S
WV soil with 0.48% OC (EPA-14)
§' '
en
"co
CO
i
CD
^
CO
CD
LO
RP-HPLC on PIHAC (humic acids)
'nT
8
rn
"co
"3
o
_Q
CO
N
C7)
^
^
S
CO
CO
LO
IL sediment with 1.88% OC (EPA-21)
§' '
en
"co
CO
i
^
^
LO
LO
en
LO
IL soil with 1.30% OC (EPA-20)
§' '
en
"co
CO
i
s
^
I
g
CO
CD
CD
II
TJ"
C
CO
CO
55
Es
"Q.
1
'5
CO
"c
CD
cn
C
il&Gibsc
TJ
cn
^
cn
CD
GA sediment with 1.21% OC (EPA-B2)
o
CO
en
"co
"3
CO
i
g
^
g
CO
CM
CD
CO
(D
Q.
from regression of 14 sediments/soil sa
o
cn
"co
"3
"3
CO
CO
CO
X
o
^
g
CO
CD
IA loess with 0. 1 1 % OC (EPA-9)
o
CO
en
"co
"3
CO
i
CO
^
E
CO
RP-HPLC on CIHAC (humic acids)
'nT
rn
"co
"3
o
_Q
CO
N
C7)
CM
CO
^
en
CD
o
g
IL sediment with 1.48% OC (EPA-26)
§' '
en
"co
•3
CO
i
CO
CO
^
00
ES
IA sediment with 0.15% OC (EPA-8)
§' '
en
"co
•3
CO
i
CO
CO
^
ES
ES
soils/sediments average
CO
it
o
.£=
o
1
CO
CO
^
E
CD
ES
KY sediment with 0.66% OC (EPA-18)
o
CO
en
"co
"3
CO
i
CO
^
CD
CO
^
IN sediment with 0.95% OC (EPA-15)
o
CO
en
"co
"3
CO
i
CM
en
^
CM
CM
CO
CO
C
g
"o
*<—
"c
QJ
CO
'co
0
1
o
x
o
0
X
£="
=3
8
Q
D)
CO
en
en
_:
"3
"o
.£=
o
1
CM
en
^
8
o
oS
Mississippi R. sediment, 1.5% OC
CM
CO
en
it
o
.£=
o
1
CM
en
^
8
o
oS
SD sediment with 0.72% OC (EPA-6)
§' '
en
"co
•3
CO
i
CO
en
^
CM
LO
CO
, — .
Q
"o.
0
Mississippi R. sediment, 1.48% OC (de
.0
LO
CO
t-
CO
O
"o
o
1
^
^
CO
CO
CO
Es
'IT
o
e-
o
TJ
8
1
CO
"c
0
0
CO
of
O
6
LO
•j-
co
o
it
o
-C
o
1
g
^
LO
cn
CO
LO
cn
CO
CO
TJ
0
CO
0
^
o
I
'o
CO
CO
o
o
C
0
=3
cr
CJ
CO
OJ
o
=3
, — .
CD
cn
^
"co
"3
o
CO
0
CM
LO
LO
CM
CO
CO
CO
to
(D
CO
-C
"c
CM
Tamar estuary sediment; 4.02% OC; 0.
£5
cn
£
^
2
es & Mar
1
CO
LO
o
cn
CO
CO
LO
CO
^
00
§
g"
s,
(0
g1"
CO
CO
^
CM
g
rC
to
s
s
_o
•s
E
o
O
"c
2
=3
CO
CO
E
lo
"c
OJ
Q.
c\T
"co
"3
C
_co
1
C
S
'o
CO
CO
co"
cn
"co
"3
"c
QJ
TJ
OJ
CD
q
CM
CM
cn
LO
CM
8'
0
£»
W
q
CM
CM
0)
S,
(0
g1"
q
CM
CM
0)
i
s
_o
•s
E
o
O
-------
I/)
.0
'E
TO
E>
o
o
Q.
2
a)
I
•o
0)
t»
_0)
"o
o
_0)
TO
£
Commer
o>
3
i
d
O
15
o
E
.E
O
es Celsius
Willamette silt loam; 0.928% OC; 20 degre
O)
i--
O)
"co
"CD
=3
g
6
s
s
LO
S
§!
(0
£
o
_o
.E
o
(0
2
h-
CNI
'H
£•
s?
R
S,
(0
g1"
s?
R
c
1
_o
•s
E
o
o
forest soil with 0.2% OC; column stufy
CD
s
~~^
"co
"CD
Q.
'CD
C7)
LO
CM
CM
h-
h-
t
CO
CM
d>
E
0)
£
O
O
P-
O
2
"5
i-
agricultural soil wtih 2.2% OC; column stud
CD
s
~~^
"co
"CD
Q.
'CD
C7)
CO
CM
LO
o
CM
specific system information not provided
o
3
"£
CD
c
"0
'o
CO
CO
c\r
CO
^,
C3)
b
CO
_Q
c
N
0
C7)
LO
CO
CM
Si
CM
n-core sediment; foe = 0.0133
O)
CO
O)
CD
CD
c
CO
CD
°Q
'E
LO
CO
CM
Si
CM
Lincoln sand; 0.087% OC
„
£
CO
"CD
c
o
CO
g
LO
CO
CM
LO
CM
CM
1
A
-£
C3)
s
s
CD
LO
S
CD
II
T3
CD
CO
CD
8
-|j
co"
CD
=3
CO
C3)
O)
CO
O)
CD
c
CO
CD
°S
'E
E
^
CM
LO
CO
CM
; 32 soils
reported as Ksom; foe reported as a range;
„
00
O)
"co
"CD
"CD
CO
CD
LL
CO
CO
CM
I--
co
CM
LO
O
O
CD
II
8
"co
_o
"co
'5
cr
CO
_co"
"c
1
X
CD
^=
0
_Q
1
S
CO
"CD
"5
T3
^3
CM
CM
CO
CD
CM
cS
=3
CO
CM
CO
Agawam fine sandy loam soil; 2.57% OC;
g
O)
^
:=
"co
C
C3)
Q.
CO
CM
CO
CD
CM
E
CM
V
Tampa sandy aquifer material; 0.13% OC;
O)
o
CO
C£
oo
CO
CD
CO
CO
CO
CM
O)
CD
CM
top 20 cm of Podzol soil; 0.87% OC
8
O)
.
CO
"CD
CD
CM
CO
forest soil with 3.7% OC; column study
$
CO
"co
"CD
Q.
'CD
C7)
s
CM
CO
C^
coarse sand with 0.09% OC
CM
O)
O)
C-
c
CO
CO
"co
1
CO
I
o
1
LO
LO
CM
CD
LO
CO
s Celsius
Willamette silt loam; 0.93% OC; 20 degree
O)
O)
"co
"CD
=3
g
6
CO
CM
CM
0.15%OC
avg. of 6 meas. w/ different Co & sorbents;
•^
CO
O)
^,
—
CO
OJ
5
CO
c
N
1
0
C7)
te
CM
CO
CO
CM
R
CM
S,
(0
g1"
CM
CM
iS
CM
E
1
O
•5
E
o
O
agricultural soil wtih 2.2% OC; column stud
CD
s
~~^
"co
"CD
Q.
'CD
C7)
fe
i
CO
CO
CO
§
?
E
3
O
1-
co
CO
s
CO
"c
CO
Tamar estuary sediment; 4.02% OC; 0.2%
jC~
CO
O)
&
1
CO
^
ofi
o
1
8
CM
S
LO
O
o
CD
II
8
"co
'El
_CD
"co
'5
cr
CO
_co"
"c
1
X
CD
_Q
1
S
CO
"CD
"5
T3
S
CM
1^
O
o
te
b"
'CD
N
O)
c
o
_J
ofi
'E
CO
0
§
CM
CO
CM
RP-HPLC on PIHAC (avg., humic acids)
'm
g
CO
"co
"CD
CO
N
C7)
O
CM
CD
CM
forest soil with 3.7% OC; column study
$
CO
CO
"CD
Q.
'CD
C7)
CO
CM
i
Lincoln sand; 0.087% OC
„
£
CO
CD
c
o
CO
g
CO
CM
O
LO
Sapsucker Woods soil with 7.51 %C
CD
S
c
O
_J
ofi
c
CO
0
CO
CM
LO
8
LO
CD
b"
'c
D)
CO
c
o
_J
ofi
c
CO
0
CO
CM
LO
7.05%C
>
'o
CO
T3
"o
1
CD
"CD
CO
E
1
CD
O
=3
CO
C7)
O)
c
o
_J
ofi
c
CO
0
CM
CM
i
RP-HPLC on CIHAC (avg., humic acids)
'm
g
O)
"co
"CD
CO
N
C7)
CD
CM
CM
CM
CO
O
o
CD
avg. of 6 meas. w/ different Co & sorbents;
•^
CO
Oi
^,
—
CO
O
CO
_Q
c
N
1
0
O)
O)
CO
CM
CM
CD
CM
S
8,
(0
g1"
LO
CM
t
c
1
_o
•s
E
o
O
-------
E>
o
u
Q.
2
0)
t»
_0)
"o
o
Comment;
o>
o
3
O
(/)
8
U)
_J
!
d
z
o
"re
o
E
.c
O
CO
m
easured valui
average value from collected m
Gustafson (1989)
CO
co
CD
CO
LO"
CO
CM
LO
CO
8
CO
01
0
a.
re
X
£
CO
CO
to
00
S"
u
re
S
^
CO
CO
to
00
S"
c
re
u
u
•c
"5
E
0
CO
o
CM
Woodburn silt loam; fom = 0.01
Chiouetal. (1983;
CO
CM
S
CO
CM
CO
CD
CM
0
U
C
u
o
j>
u
~
1-
N
2-
Dormont soil; 1.2% OC
CD
%
0
Southworth & Kell'
LO
CO
CM
LO
CO
CO
8
CD
O
CM
'o
CO
^T
CO
CO
Scheunert et al. (1
o
CO
CO
CO
o
Fullerton soil; 0.06% OC
00
CO
0
Southworth & Kelh
^
CO
S
CO
;e alluvial soil
average of 21 values; subsurfac
LO
CO
Banerjee et al. (19
^
CO
CO
o
CO
1
1
0
C3>
CO
CO
„
Wilson etal. (1981
^
CO
O)
CO
CO
O
CD
CM
CD
'o
CO
"co
0
Q_
^^
oS
CO
"co
"0
"0
CO
0
LL
CD
CO
LO
CO
as a range
S
o
Q_
2
O
1
CO
CO
I
2
„
I
"co
"0
"0
CO
0
LL
CD
CO
!
>n; 2.59% OC
untreated Marlette soil, A horizc
Lee etal. (1989)
CO
CO
S
LO
on; 0.3% OC
N
'El
O
_£=
S
'o
CO
CD
1
CO
"co
_g
"c
=3
5T
8
"co
"CD
0
CD
CM
CO
i
KS1 field material; foe = 0.0073
1
=
"co
CD
g
Schwarzenbach &
o
CO
§
°l
provided
"o
c
c
g
o
1
Q.
CO
S
a
CO
o~
1
0
0
c
TJ
£
'o
CO
CO
c\T
CO
CO
C-
0
C3)
b
Schwarzenbach &
o
CO
LO
CO
CO
Apison soil with 0.11% OC
0?
CO
0
Southworth & Kelh
CM
CO
CM"
8
LO
& sorbents; 0
avg. of 6 meas. w/ different Co <
CO
CO
=
"co
0
5
Schwarzenbach &
r--
CO
CO
s
CM"
provided
system specific information not
8
CO
0
0
c
TJ
'o
CO
CO
C\T
CO
CO
C-
0
C3)
b
Schwarzenbach &
^
CO
o
I--
CM"
O
o
S?
CO
Droximately 8
Q.
CO
_c
"c
0
1
CO
CO
_0
CO
6
S"
CO
CO
Wu & Gschwend (
CO
CO
CO
co"
field sample; 0.8% OC
CO
CO
=
"co
0
5
Schwarzenbach &
CO
CO
LO
CM
co"
LO
CM
CO
1
u
U)
re
S
4,
CM
CM
CO
i
c
re
u
u
•c
"5
E
0
top 20 cm of Eerd soil; 4% OC
CD
"co
"0
.£=
o
CM
O
CM
CO
LO
O
LO
LO
h"
0
C
re
^
o
_o
u
•c
H
^.
^~
0
C3)
c
CO
CO
CO
reported as Ksom; foe reported
„
oS
CO
"co
"0
"0
1
LL
CO
O
CM
1
cyanopropyl column, HPLC
—v
1
^
Hodson & Williams
^
CM
CO
CM
O
O
top 20 cm of Podzol soil; 0.87%
CD
CO
CO
"co
"CD
_c
8
S
CM
CM
h-
co
=3
'co
CD
O
CO
D)
CD
TJ
LO
CO
Willamette silt loam; 0.93% OC;
Chiouetal. (1979)
LO
CM
CM
CO
^
CM
C
8,
re
s
4,
CO
CM
C
c
u
•5
o
o>
o
TJ
"co
C
forest soil with 0.2% OC; colum
Seipetal. (1986)
CO
h-
o
CD
CD
LO
8
h-
C
re
1
_o
o
^L
^~
;olumn study
agricultural soil wtih 2.2% OC; c
CD
"co
"CD
Q.
'CD
0)
o
CO
CD
Lincoln sand; 0.087% OC
„
Wilson etal. (1981
CO
S
TJ
"co
C
forest soil with 3.7% OC; colum
Seipetal. (1986)
CO
o
CM
CO
O
CO
CO
N.
N.
U
U)
re
s
4,
CO
R
C
re
u
o
•c
"5
0
-------
TO
E>
o
u
Q.
2
0)
>
x°
0)
•c
"o
o
Comments
3
$.
J
U)
O
5
J
6
z
U)
o
8
E
O
8
g
LO
"CD
CD
Q_
c\T
CO
CO
=3
o
Rutherford & Chi
CD
I--
LO
CD
O
CO
o>
C
U
t-
"m
U
Q
_O
U
•c
1-
"c
1
X
CD
CD
CO
CD
^=
Q.
a
CD
2
O
o
CXI
q
'o
CO
^-^
Smith etal. (199(
CO
CD
LO
O
O
O
II
8
"CD
JD
"CD
'5
cr
CD
_co"
"c
1
X
CD
_£=
_Q
1
„
Abdul etal. (198;
no
LO
CD
!=
Tampa sandy aquifer material; 0.13% OC; < 2mr
CO
CO
Brusseau & Rao
CO
CD
forest soil with 0.2% OC; column study
Seipetal. (1986]
CD
no
LO
CM
CM
CO
II
X
Q.
8
0^
1
O
CD
CD
Q_
Q.
CD~
no
CO
£
>>
Stauffer & Macin
CM
CO
nS
aquifer solid; 0.19%OC
S
no
CO
CD
Piwoni & Banerje
CO
CO
a
Overton silty clay loam; foc=0.018; 1/n=0.93
-P
no
CO
^
CD
c
_CD
CD
LL
CO
CD
C3)
O
{£.
-3-
co
no
Lincoln sand; 0.087% OC
"
Wilson etal. (19*
CD
CO
»
agricultural soil wtih 2.2% OC; column study
CD
8
"CD
"CD
Q.
'CD
CO
no
CO
no
LO
CO
Agawam soil; 2.57% OC (2 values)
„
I
CO
_g
"CD
c
C3>
Q.
CM
g
organic carbon soil
CO
"CD
"CD
CD
.*:
o
CD
^
c
T3
_CD
'o
CO
CD
CM~
CO
CO
Doust & Huang (
CM
o
o
_co
'o
CO
CM
CO
CD"
C3)
c
CD
CD
CO
CD
I
a
CD
8
O
CO
CD
I
O
^
8
"CD
"CD
"CD
CO
g
LL
CM
O
'top 20 cm of Eerd soil; 4% OC
'Loch etal. (1986
o
CM
CO
O
Sapsucker Woods soil with 7.51 %C
CD
CO
Garbarini & Lion
CO
o
CM
E
8
S
LO
'CD
N
CD~
1
Garbarini & Lion
LO
o
CM
^_
lignin; 65% OC
CD
1
Garbarini & Lion
no
o
CM
0
sapsucker woods ether extracted soil; 7.05% OC
CD
1
Garbarini & Lion
CO
o
CM
a
$
column; Keweenaw 7 soil; 0.85% OC (avg. 3 vak
CD
r>1
Hutzleretal. (191
CO
o
CM
CO
CM
forest soil with 3.7% OC; column study
CD
"CD
"CD
Q_
'CD
LO
CM
CM
CM
no
o
II
c
Hastings silty clay loam; foe = 0.026; Freundlich;
-P
no
CO
CD
c
CD
Rogers & McFarl
r--
CM
S
CO
CO
N.
O)
U
U)
(0
1
I--
co
c
(0
u
u
•c
•5
0
^
CO
CD
CD
CO
Tamar estuary sediment; 4.02% OC; 0.2% synth
no
CO
CD
|
CD
«8
CO
_g
>
LO
CO
CM
81
•*t CD
t- 1"-
9 t
LO LO
I"- CO
u
•a
•c
1 g
o H
">, ^
> o
RP-HPLC, CIHAC (humic acids)
<§
Szaboetal. (199
r--
co
CM
s
RP-HPLC, PIHAC (humic acids)
<§
Szaboetal. (199
o
CM
LO
Catlin sorbent at 23 degrees Celsius; 4.04% OC
r-.
no
CO
"CD
"CD
S.
•^t
CM
no
LO
no
CO
CM
?
(N|
U
U)
(0
1
no
CO
CM
?
(N|
C
(0
u
u
£
0
agricultural soil with 2.2% OC; column study
Seipetal. (1986)
o
CM
CM
no
LO
CO
c8
CO
O
c
H
X
S
LO
O
o
o
II
o
£
ID
^
CD
'5
CD
_co"
"c
CD
'El
CD
Q.
_O
"CD
_Q
E
=3
'El
.g
'5
cr
CD
^^
no
CO
"CD
"CD
|
CM
CM
CM
E
forest soil; 3.7% OC; column study
CD
8
"CD
"CD
Q.
'CD
CD
CM
CO
no
CO
CM
a
U
U)
(0
1
CO
CM
CM
to
0)
c
(0
u
u
•c
•5
0
-------
t/)
.0
'E
TO
E>
o
.0
15
0
Q.
2
^«
^,
•^
t/>
(1)
3
W
>
u
o
*:
^
0)
t»
_0>
"o
o
CM
^
0)
.0
TO
M
U
E
o
o
o>
se
3
0
(/)
J
O
o
_l
K,t(L/kg)
0
z
3
O
"re
u
u
£
O
CD
CO
CD
CD
CO
0
"CD
£
"c
to1
S?
CM
O
6"
o
52
CM
O
-^
TJ
CD
CO
^>
CD
_3
"co
CD
CD
E
i™
I—
te
O)
2
=3
|
CD
^
°e
CD
g
O
>
5
CM
O
CD
CM
CO
CM
'M-
E
U
E
U
>,
X
Q.
CO
CD
=3
CD
>
CD
D)
>
CD
6"
o
0^
ID
CD
'o
CO
^P
£
s
"co
CD
5
°e
^
CD
&
c
N
^=
0
(/)
CM
ID
CM
CO
CO
CO
•o
CD
TJ
1
Q.
"o
C
c
g
^
>:-
^C
o
^5
CD
Q.
CO
E
.2
S.
CO
o~
1
^r
E2
CD
0
c
TJ
^3
'o
CO
CD
c\T
CO
O)
^>
CD
D)
b
ofi
^=
O
s
c
N
^=
0
W
s
CM
h-
S
°.
CM
PO
PI
S,
(0
g1"
<^
?
CM
f)
S
_o
•s
E
o
O
-------
Soil/Water Partition Coefficient (Koc) Bibliography
Abdul, A.S., and T.L. Gibson. 1986. Equilibrium batch experiments with six polycyclic aromatic
hydrocarbons and two aquifer materials. Hazardous Waste & Hazardous Materials
3(2):125-137.
Abdul, A.S., T.L. Gibson and D.N. Rai. 1986. The effect of organic carbon on the adsorption of
fiuorene by aquifer materials. Hazardous Waste & Hazardous Materials 3(4):429-440.
Abdul, A.S., T.L. Gibson, and D.N. Rai. 1987. Statistical correlations for predicting the partition
coefficient for nonpolar organic contaminants between aquifer organic carbon and water.
Hazardous Waste & Hazardous Materials 4(3):211-222.
Adams, R. S., and P. Li. 1971. Soil properties influencing sorption and desorption of lindane.
Soil Science Society of America Proceedings. 35:78-81.
Banerjee, P., M.D. Piwoni, and K. Ebeid. 1985. Sorption of organic contaminants to a low
carbon subsurface core. Chemosphere. 14(8):1057-1067.
Barber, L.B., II, E.M. Thurman, and D.D. Runnells. 1992. Geochemical heterogeneity in a sand
and gravel aquifer: Effect of sediment mineralogy and particle size on the sorption of
chlorobenzen. J. Cont. Hydrol. 9:35-54.
Barrett, M.H.,DJ. Williamson, and D.N. Lerner. 1994. A continuous flow method for calculating
adsorption isotherms of organic pollutants onto aquifer materials. J. Cont. Hydrol. 15:15-
25.
Briggs, G.G. 1981. Theoretical and experimental relationships betwen soil adsorption, octanol-
water partition coefficients, water solubilities, bioconcentration factors and the parachor.
J. Agric. Food Chem. 29(5): 1050-1059.
Brusseau, M.L., and P.S.C. Rao. 1991. Influence of sorbent structure on nonequilibrium sorption
of organic compounds. Environ. Sci. Technol. 25(8):1501-1506.
Caron, G., I.H. Suffet, and T. Belton. 1985. Effect of dissolved organic carbon on the
environmental distribution of nonpolar organic compounds. Chemosphere.l4(8):993-WQQ.
Carter, C.W., and I.H. Suffet. 1983. Interactions between disolved humic and fulvic acids and
pollutants in aquatic environments. Fate of Chemicals in the Environment. R. L. Swann
and A. Eschenroeder, eds. ACS Symposium Series 225. American Chemical Society,
Washington, DC. 215-229.
Chin, Y.P, and W.J. Weber, Jr. 1989. Estimating the effects of dispersed organic polymers on the
sorption of contaminants by natural solids. 1. A predictive thermodynamic humic
substance-organic solute interaction model. Environ. Sci. Technol. 23(8):978-984.
K-20
-------
Chiou, C.T., LJ. Peters, and V.H. Freed. 1979. A physical concept of soil-water equilibria for
nonionic organic compounds. Science. 206:831-832.
Chiou, C.T., P.E. Porter, and D.W. Schmedding. 1983. Partition equilibria of nonionic organic
compounds between soil organic matter and water. Environ. Sci. Technol. 17(4):227-231.
Daniels, S. L., F. D. Hoerger, and R. J. Moolenar. 1985. Environmental exposure assessment:
experience under the Toxics Substance Control Act. Environ. Toxicol. Chem. 4:107-117.
Di Toro, D.M. 1985. A particle interaction model of reversible organic chemical sorption.
Chemosphere. 14(10): 1503-1538.
Friesel, P., G. Milde, and B. Steiner. 1984. Interactions of halogenated hydrocarbons with soils.
Fresenius Z. Anal. Chem. 319:160-164.
Garbarini, D.R., and L.W. Lion. 1986. Influence of the nature of soil organics on the sorption
of toluene and trichloroethylene. Environ. Sci. Technol. 20(12):1263-1269.
Gauthier, T.D., W.R. Seltz, and C.L. Grant. 1987. Effects of structural and compositional
variations of dissolved humic materials on pyrene Koc values. Environ. Sci. Technol.
21(3):243-248.
Gerstl, Z. 1990. Evaluating the Groundwater Pollution Hazard of Toxic Organics by Molecular
Connectivity. Final Report. Institute of Soils and Water, The Volcani Center, Bet Dagan,
Israel.
Gerstl, Z., and U. Mingelgrin. 1984. Sorption of organic substances by soils and sediments. J.
Environ. Sci. Health B19(3):297-312.
Gledhill, W.E., R.G. Kaley, WJ. Adams, O. Hicks, P. R. Michael, V.W. Saeger, and G. A.
LeBlanc. 1980. An environmental safety assessment of butyl benzyl phthalate. Environ.
Sci. Technol 14(3):301-305.
Grathwohl, P. 1990. Influence of organic matter from soils and sediments from various origins
on the sorption of some chlorinated aliphatic hydrocarbons: implications on Koc
correlations. Environ. Sci. Technol. 24(11): 1687-1693.
Hassett, J.J., J.C. Means, W.L. Banwart, and S.G. Wood. 1980. Sorption Properties of Sediments
and Energy-Related Pollutants. EPA 600/3-80-041. U.S. Environmental Protection
Agency. Athens, GA.
Hodson, J., and N.A. Williams. 1988. The estimation of the adsorption coefficient (Koc) for soils
by high performance liquid chromatography. Chemosphere. 17(l):67-77.
K-21
-------
Huggenberger, FJ. Letey, W.J. Farmer. 1972. Observed and calculated distribution of lindane in
soil columns as influenced by water movement. Soil Sci. Soc. Am. Proc. 36:544-548.
Hutzler, N.J., J.C. Crittenden, J.L. Oravitz and P.A. Schaepe. 1983. Groundwater transport of
chlorinated organic compounds. In: American Chemical Society, Div. Environ. Chem.,
186th National Meeting. 23:499-502.
Hutzler, N.J., J.C. Crittenden, and J. S. Gierke. 1986. Transport of organic compounds with
saturated groundwater flow: experimental results. Water Resources Res. 22(3):285-295.
Johnson-Logan, L.R., R.E. Broshears, and S.J. Klaine. 1992. Partitioning behavior and the
mobility of chlordane in groundwater. Environ. Sci. Technol. 26(ll):2234-2239.
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. 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 Res. 26(1): 13-20.
Kan, A.T., and M.B.Tomson. 1990. Ground water transport of hydrophobic organic compounds
in the presence of dissolved organic matter. Environ. Toxicol. Chem. 9:253-263.
Karickhoff, S.W. 1981. Semi-empirical estimation of sorption of hydrophobic pollutants on
natural sediments and soils. Chemosphere. 10(8):833-846.
Karickhoff, S.W. 1982. Sorption kinetics of hydrophobic pollutants in natural sediments.
Contaminants and Sediments. 2:193-205.
Karickhoff, S.W., D.S. Brown, and T.A. Scott. 1979. Sorption of hydrophobic pollutants on
natural sediments. Water Res. 13:241-248.
Karickhoff, S.W., and K.R.Morris. 1985a. Impact of tubificid oligochaetes on pollutant transport
in bottom sediments. Environ. Sci. Tech. 19:51-56.
Karickhoff, S.W., and K.R. Morris. 1985b. Sorption dynamics of hydrophobic pollutants in
sediment suspensions. Environ. Toxicol. Chem. 4:469-479.
Kay, B.D., and D.E. Elrick. 1967. Adsorption and movement of lindane in soils. Soil Science.
104:314.
Kishi, H., N. Kogure, and Y. Hashimoto. 1990. Contribution of soil constituents in adsorption
coefficient of aromatic compounds, halogenated alicyclic and aromatic compounds to soil.
Chemosphere. 21(7):867-876.
K-22
-------
Koch, R. 1983. Molecular connectivity index for assessing ecotoxicological behaviour of organic
compounds. Toxicol. Environ. Chem. 6:87-96.
Landrum, P.F., S.R. Nihart, B.J. Eadie, and W.S. Gardner. 1984. Reverse-phase separation
method for determining pollutant binding to aldrich humic acid and dissolved organic
carbon of natural waters. Environ. Sci. Technol. 18(3):187-192.
Lee, J.F., J.R. Crum, and S.A. Boyd. 1989. Enhances retention of organic contaminants by soils
exchanged with organic cations. Environ. Sci. Technol. 23(11): 1365-1372.
Leistra, M. 1970. Distribution of 1,3-dichloropropene over the phases in soil. J.Agr. Food Chem.
18(6):1124-1126.
Loch, J.P.G., H.J. Kool, P. Lagas, and J.H.A.M. Verheul. 1986. Removal and retention of volatile
chlorinated hydrocarbons in the soils' unsaturated zone. In: Contaminated Soil. J.W.
Assink and W.J. van der Brink, eds. Matinus Nijhoff Publishers, Dordrecht, Netherlands.
63-77.
Lokke, H. 1984. Sorption of selected organic pollutants in Danish soils. Ecotoxicol. Environ.
Safety 8:395-409.
Lord, K.A., G.G. Briggs, M.C. Neale, and R. Manlove. 1980. Uptake of pesticides from water
and soil by earthworms. Pest. Sci. 11:401-408.
Lotse, E.J., D.A. Graetz, G. Chesters, G.B. Lee, and L.W. Newland. 1968. Environ. Sci. Technol.
2:353.
Mackay, D., W. Y. Shiu, and K. C. Ma. 1992. Illustrated Handbook of Physical-Chemical
Properties and Environmental Fate for Organic Chemicals. Volume I - Monoaromatic
Hydrocarbons, Chlorobenzenes, andPCB's. Lewis Publishers, Boca Raton, FL.
Mackay, D., W. Y. Shiu, and K. C. Ma. 1993. Illustrated Handbook of Physical-Chemical
Properties and Environmental Fate for Organic Chemicals. Volume III - Volatile Organic
Chemicals. Lewis Publishers, Boca Raton, FL.
McCall, P.J., R.L. Swann, D.A. Laskowski, S.M. Unger, S.A. Vrona, and HJ. Dishburger. 1980.
Estimation of chemical mobility in soil from liquid chromatographic retention times. Bull.
Environ. Contam. Toxicol. 24:190-195.
McCall, P.J., R.L. Swann, D.A. Laskowski. 1983. Partition models for equilibrium distribution
of chemicals in environmental compartments. In: Fate of Chemicals in the Environment.
R. L. Swann and A. Eschenroeder, eds. ACS Symposium Series 225. American Chemical
Society, Washington, DC. 105-123.
K-23
-------
McCarthy, J.F., and B.D.Jimenez. 1985. Interactions between polycyclic aromatic hydrocarbons
and dissolved humic material: binding and dissociation. Environ. Sci. Technol.
19(11):1072-1076.
Means, J.C., S.G. Wood, JJ. Hassett, and W.L. Banwart, 1980. Sorption of polynuclear aromatic
hydrocarbons by sediments and soils. Environ. Sci. Technol. 14(12):1524-1528.
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, C.T., and W.J. Weber, Jr. 1986. Sorption of hydrophobic organic pollutants in saturated
soil systems. J. Cont. Hydrol. l(l/2):243-261.
Mills, A.C., and J.W. Biggar. 1969a. Adsorption of gamma-BHC from solutions on several
selected adsorbents. J. Agric. Food Chem. 17(3):604-608.
Mills, A.C., and J.W. Biggar. 1969b. Solubility-temperature effects on the adsorption of gamma-
and beta-BHC from aqueous and hexane solutions by soil materials. Soil Sci. Soc. Amer.
Proc. 33:210-216.
Moreale, A., and R. van Bladel. 1978. Adsorption et migration du lindane (1,2,3,4,5,6-
hexachlorohexane) dans les sols. Parasitica. 34(4):233-255.
Paviostathis, S.G., and K. Jaglal. 1991. Desorptive behavior of trichloroethylene in contaminated
soil. Environ. Sci. Technol. 25(2):274-279.
Paviostathis, S.G., and G.N. Mathavan. 1992. Desorption kinetics of selected volatile organic
compounds from field contaminated soils. Environ. Sci. Technol. 26(3):532-538.
Pignatello, J.J. 1990. Slowly reversible sorption of aliphatic halocarbons in soils. I. formation of
residual fractions. Environ. Toxicol. Chem. 9:1107-1115.
Pignatello, J.J. 1991. Desorption of tetrachloroethene and l,2-dibromo-3-chloropropane from
aquifer sediments. Environ. Toxicol. Chem. 10:1399-1404.
Piwoni, M.D., and P. Banerjee. 1989. Sorption of volatile organic solvents from aqueous solution
onto subsurface solids. J. Contam. Hydrol. 4(2): 163-179.
Podoll, R.T., K.C. Irwin, and H.J. Parish. 1989. Dynamic studies of naphthalene sorption on soil
from aqueous solution. Chemosphere. 18(11/12):2399-2412.
Reinbold, K.A., J.J. Hassett, J.C. Means, and W.L. Banwart. 1979. Adsorption of Energy-Related
Organic Pollutants: A Literature Review. EPA 600/3-79-086. U.S. Environmental
Protection Agency. Washington, DC.
K-24
-------
Rippen, G., M. Ilgenstein, W. Klopffer, and HJ. Poremski. 1982. Screening of the adsorption
behavior of new chemicals: natural soils and model adsorbents. Ecotoxicol. Environ.
Safety 6:236-245.
Roberts, P.V., P.L. McCarty, M. Reinhard, and J. Schreiner. 1980. Organic contaminant behavior
during groundwater recharge. Water Poll. Control Fed. J. 52:161-172.
Rogers, R.D., J.C. McFarlane, and A.J. Cross. 1980. Adsorption and desorption of benzene in
two soils and montmorillonite clay. J. Am. Chem. Soc. 14(4):457-460.
Rogers, R.D., and J.C. McFarlane. 1981. Sorption of carbon tetrachloride, ethylene dibromide,
and trichloroethylene on soil and clay. Environ. Monitoring and Assessment. 1:155-162.
Roy, W.R., I. G. Krapac, S. F. J. Chou, and R. A. Griffin. 1987. Batch-Type Adsorption
Procedures for Estimating Soil Attenuation of Chemicals. Draft Technical Resource
Document. EPA/530-SW-87-006. U. S. Environmental Protection Agency. Office of
Research and Development. Cincinnati, OH. NTIS PB87-146155.
Russell, D.J., and B. McDuffie. 1986. Chemodynamic properties of phthalate esters: partitioning
and soil migration. Chemosphere. 15(8):1003-1021.
Rutherford, D.W., and C.T. Chiou. 1992. Effect of water saturation in soil organic matter on the
partition of organic compounds. Environ. Sci. Technol. 26:965-970.
Rutherford, D.W., C.T. Chiou, and D.E. Kile. 1992. Influence of soil organic matter composition
on the partition of organic compounds. Environ. Sci. Technol. 26:336-340.
Scheunert, I., E. Topp, A. Attar, and F. Korte. 1994. Uptake pathways of chlorobenzenes in
plants and their correlation with n-octanol/water partition coefficients. Ecotoxicol.
Environ. Safety. 27:90-104.
Schwarzenbach, R.P., and J. Westall. 1981. Transport of nonpolar organic compounds from
surface water to groundwater. Laboratory sorption studies. Environ. Sci. Technol.
15(11):1360-1367.
Schwille, F. 1988. Dense Chlorinated Solvents in Porous and Fractured Media. Lewis
Publishers, Chelsea, MI.
Seip, H.M., J. Alstad, G.E. Carlberg, K. Martinsen, and R. Skanne. 1986. Measurement of
mobility of organic compounds in soils. Sci. Total Environ. 50:87-101.
Sharom, M.S., J.R.W. Miles, C.R. Harris, and F.L. McEwen. 1980. Behaviour of 12 insecticides
in soil and aqueous suspensions of soil and sediment. Water Res. 14:1095-1100.
K-25
-------
Smith, J.A., C.T. Chiou, J.A. Kammer, and D.E. Kile. 1990. Effect of soil moisture on sorption
of trichloroethene vapor to vadose-zone soil at Picatinny Arsenal, New Jersey. Environ.
Sci. Technol 24(5):676-698.
Smith, J.H., W.R. Mabey, N. Bohonos, B.R. Holt, S.S. Lee, T.-W. Chou, D.C. Bomberger, and
T. Mill. 1978. Environmental Pathways of Selected Chemicals in Freshwater Systems.
Part 2: Laboratory Studies. EPA-600/7-78-074. U.S. Environmental Protection Agency,
Athens, GA.
Southworth, G.R., and J.L. Keller. 1986. Hydrophobic sorption of polar organics by low organic
carbon soils. Water, Air, and Soil Pollution. 28:239-248.
Spencer, W.F., and M.M. Cliath. 1970. Desorption of lindane from soil as related to vapor
density. Soil Sci. Soc. Am. Proc. 34:574-578.
Stauffer, T.B., and W.G. Maclntyre. 1986. Sorption of low-polarity organic compounds on oxide
minerals and aquifer material. Environ. Toxicol. Chem. 5:949-955.
Szabo, G., S.L. Prosser, and R.A. Bulman. 1990a. Determination of the adsorption coefficient
(Koc) of some aromatics for soil by RP-HPLC on two immobilized humic acid phases.
Chemosphere. 21(6):777-788.
Szabo, G., S.L. Prosser, and R.A. Bulman. 1990b. Prediction of the adsorption coefficient (Koc)
for soil by a chemically immobilized humic acid column using RP-HPLC. Chemosphere.
21(6):729-739.
Vowles, P.O., and R.F.C. Mantoura. 1987. Sediment-water partition coefficients and HPLC
retention factors of aromatic hydrocarbons. Chemosphere. 16(1): 109-116.
Wahid, P.A., and N. Sethunathan. 1978. A simple method to study pesticide sorption in soils at
short time intervals. Soil Sci. 126(l):56-58.
Wahid, P.A., and N. Sethunathan. 1979. Sorption-desorption of a, (3 and y isomers of
hexachlorocyclohexane in soils./. Agric. Food Chem. 27(5): 1050-1053
Wahid, P. A., andN. Sethunathan. 1980. Sorption-desorption of lindane by anaerobic and aerobic
soils. J. Agric. Food Chem. 28(3):623-625.
Wauchope, R.D., T.M. Buttler, A.G. Hornsby, P.W.M. Augustijn-Beckers, and J.P. Hurt. 1992.
The SRS/ARS/CES pesticide properties database for environmental decision-making. Rev.
Environ. Contam. & Toxicol. 123:1-164.
Wilson, J.T., C.G. Enfield, WJ. Dunlap, R.L. Cosby, D.A. Foster, and L.B. Baskin. 1981.
Transport and fate of selected organic pollutants in a sandy soil. J. Environ. Qual.
10(4):501-506.
K-26
-------
Wood, A.L., D.C.Bouchard, M.L. Brusseau, and P.S.C. Rao. 1990. Cosolvent effects on sorption
and mobility of organic contaminants in soils. Chemosphere. 21(4-5):575-587.
Woodburn, K.B., L.S. Lee, P. Suresh, C. Rao, and JJ. Delfino. 1989. Comparison of sorption
energetics for hydrophobic organic chemicals by synthetic and natural sorbents from
methanol/water solvent mixtures. Environ. Sci. Technol. 23:407-413.
Wu, S., and P.M. Gschwend. 1986. Sorption kinetics of hydrophobic organic compounds to
natural sediments and soils. Environ. Sci. Technol. 20:717-725.
K-27
-------
APPENDIX L
Values for Ionizing Organics
as a Function of pH
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
a>
_3
(0
X
'o
c
a)
a.
a.
O
'5
N
C
a)
DO
CD
CD
cb
CD
CD
CD
O)
o
8
X
re
a
I
a
00000000000999999999999999999999
CO CO CN CN
OOOOOOOOOOOOOOOOOOOCD
l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^l^
oooooooooooooooo'ooooooooooooooocD
CDCOOOOI^LO-
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
tfi
a>
(0
X
'o
c
o>
a.
a.
o
c
a)
.c
a.
o
9
CM
CT>
CT>
cb
CD
CT>
LO
CT>
OOOCO'ir>ir>ir>iT>^^^
CN'CNCN'CNCNCNCNCNCN
Bo|
O)
o
re
a
I
a
OOOOOOOOOOOOOOOOCT)CT)CT)CT)CT)OOOOr^r^CDiO'
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
a)
_3
(0
X
'o
c
a)
a.
a.
o
c
a.
o
o
o
4
CM"
CD
CD
cb
CD
CD
CD
•
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
a)
_3
(0
X
'o
Q.
a.
o
c
o>
.c
a.
E
+j
'E
a
4
CM"
CD
CD
cb
CD
CD
LO
I
a.
CD
V/CO(DLOOLO(DCN|COOOOCN|^LOCOI^I^OOOO(D(D(D(DCDOOOOOOOOO
^LOLqcpl^l^l^oqoqoqCDCDCDCDCDCDCDCDCDCDCDCDCDCDppppppppp
CTT^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^T^CNiCNiCNiCNiCNiCNiCNiCNiCN
COCOCNCNCNCNt-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-T-
pppppppppppppppppppppppppppppppp
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOCDCDCD
opppppppppppppppppppppppppppppppp
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOCDCDCD
)CDCD-^fCDCDCOOOCNLOOCDCOOOOCDLO-^l-COCOCNCNT-T-T-T-T-SS2S2P^
-CDLOCNCDI^CDLO-
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
tfi
a>
(0
X
'o
c
0)
Q.
Q.
o
o>
a.
o
o
1
a>
a.
CT>
CT>
cb
CD
CT>
LO
I
a.
CT>
CN
CN
O)
o
re
Q.
I
Q.
COCOCOCOCOCOCOCOCOCOCOCOCOCOCOCNCNCNCNCNCNCNCNCNCNCNCNCNCNCNCN
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)
cocococococococococococococococococococococococococococococococo
•^•OOCOOO-^-OCDCOt-
TtcOCOCNCNCN-^-^-^
CJJr^iO-^l-COCOCNt-T-T-OOOOOOOOOOOOO
oooooooooooooooooooooooooooooooc
oooooooooooooooooooooooooooooooo
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
CT)
O T- CN
If) If) If)
CD
OO CT>
p
oo
lOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiOiO
O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)O)
O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O)
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
tfi
a)
(0
X
^
c
a)
a.
a.
o
c
a)
a.
o
o
5
?
IO_
to"
CM"
CT>
CT>
cb
CD
CT>
LO
I
a.
CT)
^J ................................
o
CDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCD
if^ {Q LO ^" CO ^"™ O5 1*^ ^" ^"™ OO CO O5 ^" OO C^ 1*^ ^~ LO C3 CO ^~ OO LO C^ ^3 OO CO LO ^f CO C^ C^
O5 O5 O5 O5 O5 OO OO OO OO 1**^ 1**^ CD CD LO LO ^" ^" CO CO C^ C^ 1^~ 1^~ 1^~ 1^~ C3 C3 C3 C3 C3 C3 C3
oooooooooooooooooooooooooooooocbcb
TOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLO
cocococococococococococococococococococococococococococococococo
&CDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCD
^p O5 (~~* ^~ OJ CO ^f LO CD 1"^ CO O5 (~~* ^~ OJ CO ^f LO CD 1"^ CO O5 (~~* ^~ OJ CO ^f LO CD 1"^ OO O5 (~~*
Q.^iOLOLOLOLOLOLOLOLOLOCDCDCDCDCDC^
CDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCD
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
(0
X
^
C
a)
a.
a.
O
c
a>
a.
o
o
5
+j
a)
to
CM"
CT>
CT>
cb
CD
CT>
LO
CT>
CO
^J ................................
o
^- ^- ^- ^- f\| f\| f\| Q5 Q5 OO P*^ P*^ 1^ f* 1^. f*. f* 1^ *—* **. f* AA. -T- _. _. .^ _. __. _
j" J" J^_ i_ * ;* j-J v;J Jr * jr* *^r • ^^ p^ LO p^ f\l LO ^T r^ fj r\l CO OO O5 CO ^~™ ^~™ CO r^ CO OO LO
Lfi Lfi CO O5 ^T O5 LO CN CN LO ^"™ ^J U*. i IT . " >r ,JL L^, jri . :rx J>C ** _•_ J^ X i ! :2L ILr
^. ^r ~^| ^ jj** ^. " v^j r^r ^_ ^^ f^ p^ ^f ^f LO OO CN OO ^T ^"™ OO CD LO ^T CO CN ^"™ ^"™ ^J ^J
^** '. jj* r** v_ ' . T^J. * . ^^ ^ * ^ * Q5 r^ co LO ^^ co co c^ CNJ c^ i^~ i^~ i^~ i^~ i^~ i^~ i^~ i^~ i^~ i^~
gcocococococococococococococococococococococococococococococococo
^ O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O) O)
co ^~™ ^~™ co ^3 r^ 05 05 r^ co oo co oo oo 05 co ^~ r^ co r^ LO co co ^~ 05 05 co c^ c^ c^ LO c^
LO co co r^ c^ c^ co co ^~ c^ c^ co co i^~ c^ co 05 r^ oo c^ ^~ 05 LO c^ 05 r^ co LO ^~ co c^ c^
r^ co co LO LO ^~ co co c^ CNJ c^ i^~ i^~ i^~ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^ c^
ooooooooooooooooooooooooooooooo
reoooooooooooooooooooooooooooooooo
cocococococococococococococococococococococococococococococococo
&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&l&lf)lf)
^P 05 c^ ^~™ c^ co ^~ LO co r^ oo 05 c^ i^~ c^ co ^~ LO co r^ oo 05 c^ i^~ c^ co ^~ LO co r^ oo 05 c^
Q.TtLOLOLOLOLOLOLOLOLOLOCDCDCDCDCDCDCDCDCDCDI^I^I^I^I^
oooooooooooooooooooooooooooooooo
CDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCDCD
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
(0
X
'o
c
a)
a.
a.
O
c
a>
a.
o
10
^"
CM"
COCOCOCNCNCNCNCN
oo
O)
O
re
a
I
a
COCOCOcOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCOCNCNCNCNCNc
cocococococococococococococococococococococococococococococococo
cocococococococococococococococococococococococococococococococo
COCN|OI^^OLOCT>T-CN|OCOCT>OOCOCOCT>CT>LOCOCOr^OCNCOCO'
-------
X
Q.
M-
o
c
o
"•5
O
C
3
LJ_
(0
(0
(0
(0
o
'E
(0
O)
6
O)
o
(0
X
^
c
a)
a.
a.
O
c
a>
a.
(O
^"
CM"
CO
CO
CD
CD
CO
LO
I
a.
CO
o
CN
CO
o o o o
o oq cq -
-------
APPENDIX M
Response to Peer-Review Comments on MINTEQA2
Model Results
-------
APPENDIX M
Response to Peer-Review Comments on MINTEQA2 Model Results
Peer review of the SSL MINTEQA2 model results 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.
The remainder of this Appendix addresses each of these issues.
Charge balance in the MINTEQA2 model runs
Although the charge imbalances (e.g., 6.8% at pH 8.0 and 54.9% at pH 4.9) are present especially at
high and low pH conditions, the conclusion that the charge imbalance makes the model results not
chemically meaningful is not warranted.
MINTEQA2 uses two primary equations to solve chemical equilibrium problems: the mass action
equation (also called the mass law equation) and the mass balance equation. MINTEQA2 does not use
the charge balance equation to obtain the mathematical solution of the equilibrium problem. This
does not mean that the charge balance equation has no meaning in MINTEQA2 calculations.
The reviewer's concern is understandable. It is logical that any chemical system whose charges are
not in balance must be incomplete or have erroneous concentrations for one or more components.
However, the systems being modeled here are not "real" systems in the sense that they physically
exist somewhere so that measurements can be made on them. Rather, they are generic,
representative systems for ground water with variable (high, medium, low) concentrations of those
parameters that most significantly impact Kd.
The modeled groundwater consists of national median concentrations of those major cations and
anions that are most likely to impact the chemistry of the trace metal of interest by: (1) their
complexation with the trace metal, (2) their competition with the trace metal for sorption sites,
and/or (3) their effect on the ionic strength of the solution and thus, the activity coefficients of all
species in solution including the trace metal. The settings of the three components of this
representative system that have the greatest impact on the calculated Kd for various trace metals are
systematically varied. The three "master variable" components are pH, iron hydroxide sorption site
concentration, and concentration of natural organic matter (particulars and dissolved).
M-l
-------
No attempt was made to reconcile charge balances at the different settings of the three master
variables. If reconciling charge balance had been attempted, it would have been accomplished by
adjusting the concentrations of relatively inert anions and cations (e.g., NO3-, Na+) as needed to
balance the charge at equilibrium. It would be unwise to adjust the concentrations of more reactive
components (e.g., CO32-, Ca2+). To do so would be inconsistent with the initial assumption that those
constituents could be adequately represented by median concentrations observed in ground water and
that variability in the system could be captured by varying the three master components.
Table M-l shows the result if the concentrations of the less reactive components NO3- and Na+ are
adjusted in the high and low pH model runs so as to give a charge imbalance of <5% at equilibrium.
The results shown pertain to the medium iron hydroxide and medium natural organic matter settings
for zinc at the pH values listed. As shown in the table, the IQ values computed differ little from
those presented in this report. The expected degree of error in the Kd values due to the many
simplifications and assumptions involved in generic modeling must surely exceed the variance due to
charge imbalance.
Table M-1. Kd values with and without counter ions (Na+ or NOs-)
added to balance charge.
pH K1 (L/kg) No Counter Ion Added K1 (L/kg) With Na+or NO "Added
4.9
8.0
1.61
16,161
1.51
16,135
1 Kd values shown correspond to the medium iron hydroxide, medium natural organic matter settings. Counter ions
were added to reduce charge imbalance to <5% at equilibrium.
Sulfate adsorption in MINTEQA2 model runs
The peer reviewer states that sulfate should not be allowed to adsorb to the iron oxide. The reviewer
concludes that by including the adsorption reactions "sulfate is removed from the aqueous phase at
pH values less than 7 and is prevented from participating in precipitation reactions at these pH
values".
The sulfate adsorption reactions on iron oxide included in the MINTEQA2 model runs were taken
from a database of adsorption reactions that has been shown to give reliable results in predicting
sulfate adsorption on pure phase iron oxide (Dzombak, 1986). The reviewer is correct in that free
sulfate concentration is enhanced at low pH in runs without sulfate adsorption relative to runs with
sulfate adsorption. However, for runs with low contaminant trace metal concentrations from which
the SSL Kd's were taken, metal-sulfate precipitates do not form regardless of whether sulfate
adsorption is included or not. Also, the Kd values over the entire range of trace metal concentrations
modeled do not differ significantly when sulfate adsorption is included versus excluded.
Test runs were conducted on barium, zinc and cadmium at various settings of the three master
variables (pH, natural organic matter (NOM) concentration, and iron oxide (FeOX) sorption site
concentration). Table M-2 shows the Kd values for the lowest and highest trace metal concentration
for model runs with and without sulfate adsorption. Results are shown for barium, zinc and cadmium
at the indicated settings of the master variables. Where results differ for the "with" and "without"
M-2
-------
sulfate adsorption cases, it is most frequently due to the formation of aqueous complexes between the
trace metal and sulfate that compete with trace metal adsorption reactions, especially at low metal
concentrations.
Table M-2. Kd values calculated with and without sulfate adsorption reactions.
Metal (settings) Kd1 (L/kg) with
Ba
Ba
Ba
Ba
(MMH)
(MMM)
(LMM)
(LLM)
Ba (LLL)
Zn
Zn
Zn
Cd
(MMH)
(LMM)
(LML)
(LMM)
2
1
0,
0,
0,
.01
.37
.59
.12
.12
1537
1
1
0,
.61
.46
.94
SO4 adsorption
-0.
-0.
-0.
-0.
-0.
10
04
04
01
01
-863
-0.
-0.
-0.
42
26
01
Kd1 (L/kg) without SO4 adsorption
2.
1.
0.
0.
0.
,01 -
,37-
,59-
,12-
,12-
0.
0.
0.
0.
0.
10
04
04
,01
,01
1478-830
1.
1.
0.
,57-
43-
,91 -
0.
0.
0.
,41
25
,01
range shown corresponds to the lowest and highest trace metal concentrations. Master variable settings are
indicated by a three letter code for each model run: the leftmost letter indicates pH, the middle letter represents the
NOM concentration, and the rightmost letter indicates the concentration of FeOX adsorption sites (eg., HLM
indicates high pH, low natural organic matter, medium and iron oxide site concentration).
Reproducing RTI results for barium and zinc
The peer reviewer had difficulty reproducing the Kd values computed for barium and zinc. The
reviewer included two sample input files for MINTEQA2 that had failed to produce results similar to
the SSL calculations.
The SSL results can be reproduced for all metals using the current version of MINTEQA2 (v3.11)
distributed by EPA. As indicated in the 1994 Technical Background Document, a modified version
of this model was used to calculate SSL Kjs. The current version can be used to calculate the same
results by performing the following steps:
1) Edit the v3.11 component database file COMP.DBS to insert a component to represent
particulate organic matter (POM). Use the 3-digit identifying number 251, a charge of -2.8, and
a molar mass of zero.
2) Edit the v3.11 file THERMO.DBS to add the metal POM reactions shown in Appendix H of the
RTI draft report. The file DATABASE.DOC included with MINTEQA2 v3.11 gives detailed
instructions for modifying the database file. After all reactions are added, del or rename the
current THERMO.UNF and TYPE6.UNF files and execute program UNFRMT (included with
v3.11) to create new *.UNF files.
M-3
-------
3) Observe that there were two modifications to v3.11 that make calculation of Kd in L/kg easier in
the version used by RTI. Since those modifications are not present in v3.11 itself, the user must
take care in computing Kd. The procedure is to first obtain the calculated concentration of the
metal of interest (say, barium) bound with POM from PART 3 of the output file. If you have set
the solid precipitation flag to print a report each time a solid precipitates or dissolves, there will
be a series of PART 3 outputs each corresponding to a precipitation or dissolution event. You
must be sure that the PART 3 output from which you obtain the metal-POM concentration is
the equilibrium output (i.e., it occurs prior to the PART 5 EQUILIBRATED MASS
DISTRIBUTION with no intervening PROVISIONAL MASS DISTRIBUTION). After obtaining
the metal-POM concentration, locate the line corresponding to the trace metal of interest, say
barium in the PART 5 EQUILIBRATED MASS DISTRIBUTION section. Obtain the total sorbed
concentration value and to this value add the concentration of metal-POM species. This is
necessary because v3.11 recognizes only components with number 811 through 859 as sorption
components. The metal-POM concentration will not have been added in the sorbed column. It
will instead have been included in the dissolved column, so subtract the metal-POM concentration
from the dissolved total. Finally, to compute Kd, take the ratio of sorbed over dissolved (after
the adjustment for metal-POM). The resulting Kd must be divided by 3.1778 kg/L (the mass of
soil that one liter of solution is equilibrated with) to express the result in L/kg.
If the above three steps are followed, the v3.11 MINTEQA2 will give the same result as in the 1994
Technical Background Document provided the data in the input file is correct. The two input files
sent designed by the reviewer did not give correct results even when these steps were followed because
of faulty values in the input file. The files supplied by the reviewer (SSLBA.INP and SSLZN.INP)
were correct in all respects except two:
1) The site concentration for the POM component at the medium setting was entered as 1.930xl03
mg/L. This value was evidently obtained from the table on page 33 of the EPA report (US EPA,
1992) after converting to mg/L. This is not the correct value for this the POM component. The
correct value is 9.31xlO-4 mol/1 and is found in the table on page 38 of EPA report.
2) The iron oxide adsorbent is represented by two site types (components 811 and 812). The high
population site has a lower affinity for the iron oxide surface for metals (expressed in a smaller
log K in the adsorption reactions involving metals). For a particular metal, say, zinc, it will be
noted that there are two reactions in the database of 42 iron oxide adsorption reactions (FEW-
DLM.DBS). This three-digit component number associated with the reaction having the smaller
log K of the two is the number to that must be used for the high population site. That is,
component 812 should be entered at the higher site concentration and 811 at the lower site
concentration. The set of site concentrations is given on page 44 of the EPA report. In the
sample input files, component 811 was associated with the high site concentration and 812 with
the lower.
After correcting these two errors and observing the special requirements of using of v3.11 as
indicated above, the Kd values obtained using MINTEQA2 v3.11 with the peer reviewer's files were
virtually identical to the SSL results.
M-4
-------
References
Dzombak, D.A., 1986. Toward a Uniform Model for the Sorption of Inorganic Ions on Hydrous
Oxides. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA.
US EPA (Environmental Protection Agency), 1992. Background Document for Finite Source
Methodology for Wastes Containing Metals. HWEP-S0040. Office of Solid Waste,
Washington, DC, 68 pages.
M-5
------- |