#tfps%
§ mm x>
uss)
MMrt^
Risk and Technology Review (RTR)
Risk Assessment Methodologies:
For Review by the EPA's Science Advisory Board
Case Studies -
MACTI Petroleum Refining Sources
Portland Cement Manufacturing
-------
EPA-452/R-09-006
June 2009
Risk and Technology Review (RTR)
Risk Assessment Methodologies:
For Review by the EPA's Science Advisory Board
Case Studies -
MACT I Petroleum Refining Sources
Portland Cement Manufacturing
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Research Triangle Park, NC
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table of Contents
1 Introduction 1-1
1.1 Purpose of this report 1-1
1.2 Nature of RTR risk management decisions 1-1
1.2.1 Questions posed by risk managers 1-2
1.2.2 Answers provided by RTR risk assessment case studies 1-3
2 Petroleum refineries case study 2-1
2.1 Introduction 2-1
2.2 Methods 2-1
2.2.1 Emissions and source data 2-1
2.2.2 Dispersion modeling for inhalation exposure assessment 2-2
2.2.3 Estimating human inhalation exposure 2-4
2.2.4 Multipathway and environmental risk screening 2-5
2.2.5 Acute Risk Screening and Refined Assessments 2-6
2.2.6 Dose-Response Assessment 2-7
2.2.7 Risk characterization 2-17
2.3 Results Summary and Risk Characterization 2-19
2.3.1 Source Category Description and Summary of Emissions 2-20
2.3.2 Source Category Inhalation Risk Assessment Results 2-24
2.3.3 Risk Characterization 2-26
2.4 General Discussion of Uncertainties 2-30
2.4.1 Exposure Modeling Uncertainties 2-30
2.4.2 Uncertainties in the Dose-Response Relationships 2-31
3 Portland cement case study 3-1
3.1 Introducti on 3-1
3.2 Source category and emissions data 3-1
3.2.1 Dioxin emissions 3-7
3.2.2 Radionuclide emissions 3-8
3.3 Risk assessment results - inhalation 3-9
3.4 Refined multipathway health risk assessment 3-12
3.4.1 Selection of HAPs for this analysis 3-12
3.4.2 Selection of facility for case study 3-12
3.4.3 Approach to exposure assessment 3-13
3.4.4 Fate and transport modeling (TRIM.FaTE) 3-13
3.4.5 Exposure assessment 3-17
3.5 Ecological risk assessment 3-20
3.5.1 Ecological risk screening 3-20
3.5.2 Refined ecological risk assessment 3-21
3.6 Risk characterization 3-23
3.6.1 Inhalation risks 3-23
3.6.2 Multipathway risks 3-25
3.6.3 Combining risks from all facilities and exposure routes 3-32
3.6.4 General discussion of uncertainties 3-33
4 Supplemental analyses and discussion of uncertainty 4-1
4.1 Corrections to the emissions inventory - data analysis 4-1
Page ii
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
4.2 Short-term emissions and exposures - data analysis 4-2
4.3 Inventory under-reporting and gaps - data analysis 4-3
4.3.1 Ambient monitor-to-model comparison for two Texas refineries 4-3
4.3.2 Comparison of RTR emissions inventory data and Refineries Emissions Model
(REM) data 4-4
4.4 Time scale of meteorological data - sensitivity analysis 4-6
4.5 Location of meteorological stations - sensitivity analysis 4-9
4.6 Atmospheric chemistry - sensitivity analysis 4-11
4.7 Deposition - sensitivity analysis 4-13
4.8 Location of receptor populations - data analysis 4-14
4.9 Population mobility - data analysis 4-16
4.10 Acute exposure - discussion of uncertainties 4-16
4.11 Dose-response assessment - discussion of uncertainties 4-17
4.11.1 Chronic dose-response 4-17
4.11.2 Acute dose-response 4-18
4.12 Compounds without dose-response assessments - sensitivity analysis 4-18
5 References 5-1
List of tables
Table 1-1. Summary risk assessment results for petroleum refineries and Portland cement source
category case studies 1-4
Table 2-1. AERMOD version 07026 model options for RTR II modeling 2-2
Table 2-2. Dose-Response Values for Chronic Inhalation Exposure to Carcinogens 2-10
Table 2-3. Dose-Response Values for Chronic Oral Exposure to Carcinogens 2-11
Table 2-4. Dose-Response Values for Chronic Inhalation Exposure to Noncarcinogens 2-12
Table 2-5. Dose-Response Values for Acute Exposure 2-16
Table 2-6. Summary of Emissions from the MACT 1 Petroleum Refining Source Category . 2-22
Table 2-7. Summary of Source Category Level Risks for Petroleum Refineries 2-25
Table 2-8. Summary of Acute Refined Results for Petroleum Refineries 2-26
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category 3-3
Table 3-2. Mean and 95% upper confidence limit (UCL) 2378-TCDD(Teq) emission factors for
Portland cement facilities, by kiln type 3-8
Table 3-3. Summary of Source Category Level Risks for Portland Cement Manufacturing 3-9
Table 3-4. Summary of Acute Screening Results for Portland Cement Manufacturing 3-11
Table 3-5. Emissions of Dioxins and Mercury from the Lafarge Facility in Ravena, NY, and
Screening Results 3-14
Table 3-6. Ingestion Exposure Scenarios 3-17
Table 3-7. Exposure Parameters Used to Derive Risk and Hazard Estimates 3-18
Table 3-8. Dose-response Values for PB-HAPs Addressed in this Assessment 3-19
Table 3-9. Summary of Wildlife TRVs ((j,g[chemical]/kg[BW]-day) for Ravena 3-21
Table 3-10. Hazard Quotients for Wildlife Exposure to Methylmercury for Ravena 3-21
Table 3-11. Hazard Quotients for Wildlife Exposure to Divalent Mercury for Ravena 3-21
Table 3-12. Hazard Quotients for Wildlife Exposure to 2,3,7,8-TCDDa for Ravena 3-22
Page iii
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-13. Estimation of Radionuclide Emissions for the Two California Facilities Using Three
Approaches 3-24
Table 3-14. Risk Calculated for Two California Portland Cement Facilities Using AERMOD
Modeling Results and Three Emission Estimation Approaches 3-25
Table 3-15. Averages of extrapolated risks for dioxins and divalent mercury emitted by the
Portland cement source category, based on emissions-to-risk ratios estimated for
the Ravena facility 3-32
Table 4-1. Comparison of Risk Assessment Results for 1-Year vs. 5-Year Meteorological Data
for a Petroleum Refinery (NEI12486) 4-8
Table 4-2. Impact of Meteorological Station Selection on Risk Assessment 4-10
Table 4-3. Effects of Exponential Decay on MIR and Incidence levels 4-12
Table 4-4. Comparison of estimated cancer MIR and incidence with and without considering
deposition and depletion at five Portland cement facilities 4-14
Table 4-5. Results of adjustment of estimated inhalation cancer risk for long-term migration
behavior for two source categories 4-16
List of figures
Figure 3-1. 2,3,7,8-TCDD Individual Lifetime Cancer Risks for Ravena 3-27
Figure 3-2. 2,3,7,8-TCDD Chronic Non-cancer Hazard Quotients for Ravena 3-29
Figure 3-3. Mercury Chronic Non-Cancer Hazard Quotients for Ravena 3-30
Figure 3-4. Mercury Chronic Non-Cancer Hazard Quotients for Ravena 3-31
Figure 4-1. MIR/Incidence Reduction as a function of Half-Life 4-13
Page iv
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
List of appendices
Appendix A: Comparison of initial risk estimates with risk estimates refined by public
comment for petroleum refineries
Appendix B: Analysis of short-term emission rates relative to long-term emission rates for
petroleum refineries in the Galveston-Houston area
Appendix C: Technical support document for TRIM-based multipathway screening scenario for
RTR: Summary of approach and evaluation
Appendix D: Detailed assessment inputs and results for petroleum refining facilities
Appendix E: Refinement of acute exposure estimates at petroleum refining facilities (El) and
Portland cement facilities (E2)
Appendix F: Development of chlorinated dibenzodioxin and -furan emissions estimates for the
Portland cement source category
Appendix G: Development of radionuclide emissions estimates for the Portland cement source
category
Appendix H: Detailed assessment inputs and results for Portland cement facilities
Appendix I: Multipathway health risk assessment case study - Lafarge Ravena Portland
cement facility
Appendix J: Ecological risk assessment case study - Lafarge Ravena Portland cement facility
Appendix K: Development of a threshold concentration for foliar damage caused by ambient
hydrogen chloride concentrations
Appendix L: Statistical comparison of monitored and modeled ambient benzene concentrations
near petroleum refineries in Texas City, TX
Appendix M: Sensitivity analysis of uncertainty in risk estimates resulting from estimating
exposures at census block centroids near petroleum refineries
Appendix N: Analysis of the effect of considering long-term mobility of receptor populations
on estimates of lifetime cancer risk
Appendix O: Potential importance of hazardous air pollutants lacking dose-response values at
Portland cement and petroleum refining facilities
Appendix P: Comparison of National Emissions Inventory data to modeled facility data for
petroleum refineries
Page v
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Index of Acronyms
ADD
Average daily dose
AEGL
Acute exposure guideline level
AERMOD
American Meteorological Society/EPA Regulatory Model
AIHA
American Industrial Hygiene Association
ANPRM
Advanced Notice of Proposed Rulemaking
ASTDR
US Agency for Toxic Substances and Disease Registry
CAA
Clean Air Act Amendments Of 1990
CalEPA
California Environmental Protection Agency
CMAQ
Community Multiscale Air Quality (model)
CTE
Central tendency exposure
ERA
Environmental risk assessment
ERPG
Emergency Response Planning Guideline
FR
Federal Register
HAP
Hazardous air pollutant
HAPEM
Hazardous Air Pollutant Exposure Model
HEM
Human Exposure Model
HHRA
Human health risk assessment
HHRAP
Human Health Risk Assessment Protocol
HI
Hazard index
HQ
Hazard quotient
IARC
International Agency for Research on Cancer
IEUBK
Integrated Exposure Uptake Biokinetic (model)
IRIS
Integrated Risk Information System
ISH
Integrated Surface Hourly (database)
LOAEL
Lowest observed adverse effect level
MACT
Maximum Achievable Control Technology
MDL
Method detection limit
MIR
Maximum individual risk
MOA
Mode of action
MRL
Minimum Risk Level (ATSDR dose-response value)
NAC
National Advisory Committee
NATA
National Air Toxics Assessment
NCDC
National Climatic Data Center
NEI
National Emissions Inventory
NORM
Naturally occurring radioactive material
NPRM
Notice of Proposed Rulemaking
NY DEC
New York Department of Environmental Conservation
OAQPS
US EPA Office of Air Quality Planning and Standards
PAH
Polycyclic aromatic hydrocarbon
PB-HAP
Persistent and bioaccumulative HAP
POM
Polycyclic organic matter
REL
Reference exposure level
RfC
Reference concentration
RfD
Reference dose
Page vi
-------
June 15,2009
RME
RTR
SF
TCDD
TOSHI
TRIM
TRV
LIRE
USGS
WHO
WOE
RTR Risk Assessment Methods for SAB Review
Reasonable maximum exposure
Risk and Technology Review
Carcinogenic slope factor (usually for oral exposure)
Tetrachlorodibenzo-/>dioxin, termed "dioxin" in this report
Target-organ-specific hazard index
Total Risk Integrated Methodology
Toxicity reference value
Unit risk estimate
US Geological Survey
World Health Organization
Weight-of-evidence for carcinogenicity in humans
Page vii
-------
1 Introduction
1.1 Purpose of this report
This report was developed to assist a panel of the EPA's Science Advisory Board (SAB) in
reviewing the risk assessment approach and methods used by the EPA Office of Air Quality
Planning and Standards (OAQPS) for its residual risk assessments in the Risk and Technology
Review (RTR) program.
Although this document illustrates various components of our approach using two case studies
from actual residual risk assessments (either previously performed or currently under
development to support residual risk rulemaking), it is intended as a description of the approach
itself. It is not intended to convey any definitive risk characterization. The case studies are
drafts that may change as input data are revised as a result of public comment, or as methods are
revised as a result of this review or our own improvement efforts. The case studies are included
for the sole purpose of clarifying our approach for technical review, and assisting reviewers in
understanding how EPA risk managers will use the information. They are not actual residual risk
assessments that may be used to support regulatory decisions, and the results of the case studies
are not the focus of this review. The final assessments for these source categories will be
published in conjunction with their respective final rulemakings. It is important to note that each
of these case study examples represents a snapshot of an analysis which is at a different stage of
development - the petroleum refinery case study has proceeded through the ANPRM stage as
well as a Notice of Proposed Rulemaking (NPRM) stage but has not yet been issued in support
of any final rulemaking. The Portland cement case study has not yet been issued through an
ANPRM, and therefore has not yet been subjected to any public scrutiny. The charge questions
to the SAB panel are intended to elicit comment on whether the details of our approach
constitute best science, and if not, how they could be improved.
In December of 2006 we obtained a consultation from a panel of the EPA Science Advisory
Board (SAB) on our "RTR Assessment Plan." In June 2007 we received a letter [7]
summarizing the key messages from that consultation on our risk assessment methods. We have
attempted to respond to these key messages in developing this report.
1.2 Nature of RTR risk management decisions
The Clean Air Act establishes a two-stage regulatory process for addressing emissions of
hazardous air pollutants (HAPs) from stationary sources. In the first stage, the Act requires the
Environmental Protection Agency (EPA) to develop technology-based standards for categories
of industrial sources (e.g., petroleum refineries, pulp and paper mills, etc.)[2], EPA has largely
completed these standards. In the second stage, EPA is required to assess the health and
environmental risks that remain after sources come into compliance with the technology-based
standards, and to develop additional standards as necessary to protect public health with an
ample margin of safety or to prevent adverse environmental effects. These risk-based standards
must be completed within eight years of the technology-based standards. Several have already
been completed.
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
In order to inform these risk-based decisions, EPA develops a risk assessment for each source
category. In developing each assessment, EPA: (1) conducts a risk assessment using currently-
available source and emissions data; (2) shares the source and emissions data and preliminary
results of the assessment with the public through an Advance Notice of Proposed Rule Making
(ANPRM) that asks for public comments on the methods and the source and emissions data; (3)
receives comments; (4) reconciles comments and corrects the source and emissions data as
appropriate, and; (5) reassesses the risks. The risk manager applies the results of the revised risk
assessment, along with other information on cost, feasibility, and other non-risk-based
information to support proposals and promulgations of technology- and risk-based regulatory
decisions for each of the categories through the regular notice-and-comment rulemaking process.
1.2.1 Questions posed by risk managers
In order to determine if additional, risk-based, standards are needed, EPA needs to assess the
"residual" risks to health and the environment that may remain after the technology-based
standards are implemented. Residual risks are assessed separately for each source category. The
Clean Air Act (CAA) requires that the EPA promulgate additional standards for a source
category "if promulgation of such standards is required to provide an ample margin of safety to
protect public health" or "to prevent, taking into consideration costs, energy, safety, and other
relevant factors, an adverse environmental effect." A key factor in this risk management
decision is the determination of the "lifetime excess cancer risk to the individual most exposed to
emissions from a source in the category," or the maximum individual risk (MIR). The CAA
specifically provides, for example, that a residual risk rulemaking is not required for a particular
source category if EPA can show that the MIR for that category is less than 1 in a million.
EPA's risk management decision framework for residual risk rulemakings was first publicized in
its finalization of the Benzene National Emission Standards for Hazardous Air Pollutants, or
NESHAP, in 1989 (see 54 FR 38044). This framework implements the determination of an
"ample margin of safety" in 2 steps. In the first step, the EPA determines "acceptable risk."
Here, the goal is to limit the MIR for the entire source category to an acceptable level, with the
proviso that the maximum limit on the acceptable MIR is ordinarily 100 in a million. EPA is
allowed to adjust this limit (up or down) by considering other risk metrics (e.g., the total
estimated cancer incidence due to emissions from the source category) and other health factors,
including the consideration of noncancer human health risks or environmental risks, as well as
uncertainties in the risk estimates. The EPA is not allowed to consider costs of regulatory
actions in this step. In the second step, the EPA determines the "ample margin of safety." Here,
EPA is allowed to factor in the costs and feasibility of controlling emissions from the source
category as it evaluates further risk reductions across the source category with the goal of
maximizing the number of persons whose lifetime cancer risks due to emissions from the source
category are less than 1 in a million.
For effects other than cancer, EPA estimates the ratios of chronic or acute exposure to
appropriate health benchmarks and considers how each benchmark was developed when
establishing a level of "acceptable risk" or determining if health is protected with an "ample
margin of safety." For environmental effects, EPA compares exposures of nonhuman receptors
to (1) human benchmarks for screening and (2) published benchmarks for similar species for
refined assessments.
Page 1-2
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
1.2.2 Answers provided by RTR risk assessment case studies
The final product of the risk assessment process is a set of overall conclusions about risk that are
complete, informative, and useful for decision makers. In general, the assessment's ability to
provide these things depends on the information available, the application of the risk information
and the resources available.
In determining whether an ample margin of safety has been achieved, and if adverse
environmental effects will not occur, EPA risk managers look to the residual risk assessment to
provide estimates of:
1) maximum individual lifetime cancer risk
2) annualized lifetime cancer incidence and/or deaths
3) distribution of lifetime cancer risk in the exposed population
4) HAPs that contribute substantially to health or environmental risk
5) maximum individual hazard quotients (HQ1) for non-cancer chronic effects
6) target organ-specific hazard indices (TOSHI2) for chronic effects other than cancer
7) maximum individual hazard quotients (HQ) for non-cancer acute effects
8) distribution of hazard index in the exposed population;
9) ecological receptors for which exposures exceed benchmarks
HAP-specific cancer risks are added across chemicals because EPA does not yet recognize any
combination of HAPs for which cancer risk is demonstrably not additive. Chemical specific
HQs are added only for chemicals having the same mechanism of action, or (in the absence of
such information) that affect the same target organ. Chronic risk estimates are based on annual
average concentrations at individual census block centroids. Acute risks are estimated using
additional protective assumptions explained in Section 2.2.5, based on maximum 1-hour
concentrations at the worst location.
The residual risk assessment must also provide a risk characterization that transparently
describes how these estimates were developed and the uncertainties associated with them. The
risk manager uses this information, along with information on costs and feasibility of reducing
emissions, legal requirements, public concern and comment, political considerations, and other
factors in developing a residual risk rule.
The residual risk assessment, therefore, represents only part of the information used in making
risk management decisions, but it is arguably the most critical element because it can determine
that no rule is needed at all, or set an upper limit on how stringent any rule might need to be.
EPA believes that it is possible to use a consistent, streamlined approach to these assessments
that is scientifically sound and that also uses time and resources efficiently. This report describes
the methods that EPA has developed to conduct these assessments through the use of two
illustrative case studies - one for the petroleum refineries source category and one for Portland
1 Hazard quotient - the ratio of an estimated exposure to an appropriate health benchmark (usually an exposure level
associated with no adverse effects).
2 Hazard index - the sum of hazard quotients for multiple chemicals. A TOSHI is a hazard index limited to
chemicals that affect the same target organ or system.
Page 1-3
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
cement manufacturers. Together, these two case studies cover the breadth of scientific issues
often addressed in our residual risk assessments.
The quantitative results of the risk characterization for each case study have been used to address
EPA risk managers' specific questions pertaining to the requirements of promulgating a residual
risk rule under the Clean Air Act. Table 1-1 provides an example of summary information that
the two example baseline risk assessments produce to support risk management decision-making.
OAQPS staff also briefs risk managers on the context of the findings, and on the uncertainties
surrounding the risk estimates. The remainder of this report provides the details of each of these
two case study risk assessments, culminating with the presentation of a summary of baseline risk
information and a characterization of the risk for each source category, including a discussion of
uncertainties and the implications of the findings. Finally, we wrap up this report by presenting a
number of sensitivity studies which address specific issues that have arisen during the process of
developing and performing these risk assessments.
Table 1-1. Summary risk assessment results for petroleum refineries and Portland cement source
category case studies.
Risk metric
Petroleum refineries
Portland cement source
source category
category
Facilities subject to MACT/modeled
156/156
118/104
Population within 50 km of modeled
facility
90 million
54 million
Lifetime inhalation cancer risk
Maximum individual cancer risk
30 in 1 million
800 in 1 million3
Population > 100 in 1 million
0
400
Population > 10 in 1 million
4000
15,000
Population > 1 in 1 million
460,000
470,000
Facilities w/ MIR > 1 in 1 million
77
29
HAP cancer risk drivers
Benzene, naphthalene,
POM, 1,3-butadiene,
Chromium (VI), arsenic,
cadmium, beryllium,
TCE
benzene
Chronic inhalation noncancer risk
Target organs/systems
Respiratory
Neurological
Respiratory
Kidney
Maximum chronic hazard index
0.3
10 Neurological
6 Respiratory
3 Kidney
Facilities w/ HI > 1.0
0
2 Neurological
3 Respiratory
1 Kidney
Population > hazard index 1.0
0
~ 3000
HAP chronic risk drivers
Diethanolamine
Manganese, chlorine,
HCI, and cadmium
Acute inhalation noncancer risk
Screening:
50 - REL for Benzene
50 - AEGL-1 for HCI
3 Does not include analysis of potential radionuclide risks in Section 3.6.1.3.
Page 1-4
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Table 1-1. Summary risk assessment results for petroleum refineries and Portland cement source
category case studies.
Risk metric
Petroleum refineries
Portland cement source
source category
category
HAP/max. acute HQ/ benchmark4
20 - REL for Hydrogen
4 - AEGL-2 for HCI
fluoride
7 - AEGL-1 for Chlorine
6 - AEGL-1 for Hydrogen
2 -AEGL-2 for Chlorine
fluoride
3-AEGL-1 for
Formaldehyde
8 - REL for Benzene
10-AEGL-1 for
Refined:
HAP/max. acute HQ/ benchmark
5 - REL for Hydrogen
fluoride
Hydrogen chloride
2-AEGL-1 for Chlorine
2 - AEGL-1 for Hydrogen
fluoride
2-AEGL-1 for
Formaldehyde
Facilities w/ HQ >1.0
20
8
Multipathway risk
Maximum individual cancer risk
200 in 1 million6
1-10 in 1 million7
HAP cancer risk drivers
N/A5
Dioxin
96
Maximum individual hazard quotient
Z.
0.087
HAP noncancer risk drivers
Dioxin8
Ecological risk
Direct contact screening:
Max. concentration/RfC ratio
0.3 (Diethanolamine)
N/A
Direct contact refined:
Max. concentration/RfC ratio
N/A
0.1 (HCI)
Multipathway screening:
Max. concentration/benchmark ratio
1.0 (PAHs in soil)
N/A
Multipathway refined:
Max. concentration/benchmark ratios
N/A
4 - Methylmercury, mink6
0.02 - Methylmercury,
mink9
4 Definitions of each benchmark appear in the glossary, with complete descriptions in Section 2.2.6.2.
5 A multipathway assessment has been developed for petroleum refineries, but it has been omitted from the case
study and from this table for brevity. Instead, we are using the Portland cement risk assessment to illustrate our
approach to multipathway health risk assessment.
6 Includes subsistence fishing in a nearby small pond at a harvest rate that is probably not sustainable.
7 Omits small pond but includes subsistence farming at nearest farm and fishing from other water bodies.
8 Methylmercury was also evaluated for noncancer effects via ingestion, but HQs did not exceed 1.0.
9 Omits small pond but includes subsistence farming at nearest farm and fishing from other water bodies.
Page 1-5
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
2 Petroleum refineries case study
2.1 Introduction
Section 2 contains the methods and the results of the baseline risk assessment performed for the
petroleum refining source category. The methods discussion includes descriptions of the
methods used to develop refined estimates of chronic inhalation exposures and human health
risks for both cancer and noncancer endpoints, as well as descriptions of the methods used to
screen for acute health risks, chronic non-inhalation health risks, and adverse environmental
effects. Since the screening assessment did not indicate any significant potential for chronic
non-inhalation health effects, or environmental impacts including effects to threatened and
endangered species, no further refinement of this assessment was performed. A screening
assessment did indicate a possible concern for acute health effects; thus, a more refined analysis
for acute exposure impacts was performed and the results are presented.
2.2 Methods
2.2.1 Emissions and source data
The 2002 National Emissions Inventory (NEI) Final Version 1 (made publicly available
February 2006) served as the starting point for this assessment. The 2002 NEI purportedly
contains information on actual emissions during the entire 2002 base year. Using the process
MACT code10, we developed a subset of this inventory that contains emissions and facility data
for the petroleum refining source category. Next, we performed an engineering review of these
using EPA engineers who were directly involved in the development of the MACT standard for
the source category, and/or who have extensive knowledge of the characteristics of this industry.
NEI data were also updated with site-specific benzene emissions data for 22 refineries as
provided by the American Petroleum Institute. The goal of the engineering review was to
identify readily-apparent limitations and issues with the emissions data (particularly those that
would greatly influence risk estimates) and to make changes to the dataset where possible to
address these issues and decrease the uncertainties associated with the assessment.
Once the dataset for the entire source category was created, it was published through an
Advanced Notice of Proposed Rulemaking (ANPRM), making it available for public comment.
After a 60-day comment period, submitted comments and corrections were evaluated for quality
and engineering consistency. Corrections we concluded were valid were incorporated into the
inventory. In August 2007, a Notice of Proposed Rulemaking (NPRM) was published making
the source category dataset available for a second 60-day comment period, which was
subsequently re-opened for another 50 days. Again the comments and corrections were
evaluated and incorporated into the inventory. The final petroleum refinery database for our case
study contains information for 156 facilities, and this is thought to represent the source category
10 The tagging of data with MACT codes allows EPA to determine reductions attributable to the MACT program.
The NEI associates MACT codes corresponding to MACT source categories with stationary major and area source
data. MACT codes may be assigned either at the process level or at the site level in the point source data (e.g., the
MACT code for municipal waste combustors (MWCs) is assigned at the site level whereas the MACT code for
petroleum refinery catalytic cracking is assigned at the process level).
Page 2-1
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
in its entirety at this time. An analysis of the revisions to the emissions inventory resulting from
this process and the associated revisions to the risk estimates is presented in Appendix A.
2.2.2 Dispersion modeling for inhalation exposure assessment
Both long- and short-term inhalation exposure concentrations and associated health risk from
each facility of interest were estimated using the Human Exposure Model in combination with
the American Meteorological Society/EPA Regulatory Model dispersion modeling system
(HEM-AERMOD, or HEM3). HEM3 performs three main operations: atmospheric dispersion
modeling, estimation of individual human exposures and health risks, and estimation of
population risks. This section focuses on the dispersion modeling component. The exposure and
risk characterization components are discussed in sections 2.2.3 and 2.2.7.
The dispersion model in the HEM3 system, AERMOD version 07026, is a state-of-the-science
Gaussian plume dispersion model that is preferred by EPA for modeling point, area, and volume
sources of continuous air emissions from facility applications [3], Further details on AERMOD
can be found in the AERMOD Users Guide [4\ The model is used to develop annual average
ambient concentration through the simulation of hour-by-hour dispersion from the emission
sources into the surrounding atmosphere. Hourly emission rates used for this simulation are
generated by evenly dividing the total annual emission rate from the inventory into the 8,760
hours of the year.
The first step in the application of the HEM3 modeling system is to predict ambient
concentrations at locations of interest. The AERMOD model options employed are summarized
in Table 2-1 and are discussed further below.
Table 2-1. AERMOD version 07026 model options for RTR II modeling
Modeling Option
Selected Parameter for chronic exposure
Type of calculations
Hourly Ambient Concentration
Source type
Point and area sources
Receptor orientation
Polar (10 rings at 10-deg)
Discrete (census block centroids)
Terrain characterization
Actual from USGS 1-degree DEM data
Building downwash
Not Included
Plume deposition/depletion
Not Included
Urban source option
No
Meteorology
1 year representative data
Meteorological data for HEM3 are selected from a list of 158 National Weather Service (NWS)
surface observation stations across the continental United States, Alaska, Hawaii, and Puerto
Rico. In most cases the nearest station is selected as representative of the conditions at the
Page 2-2
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
subject facility. Two facilities11 furnished representative meteorological datasets as part of the
ANPRM process. For these two facilities, the facility-supplied meteorological data were utilized
in place of the HEM "nearest selected" station. Ideally, when considering off-site
meteorological data most site specific dispersion modeling efforts will employ up to five years of
data to capture variability in weather patterns from year to year. However, because of the large
number of facilities in the analysis and the extent of the dispersion modeling analysis (national
scale), it was not practical to model five years of data and only the year 1991 was modeled.
While the selection of a single year may result in under-prediction of long-term ambient levels at
some locations, likewise it may result in over-prediction at others. For each facility identified by
its characteristic latitude and longitude coordinates, the closest meteorological station was used
in the dispersion modeling. The average distance between a modeled facility and the applicable
meteorological station was 40 miles (72 km). A sensitivity analysis evaluating the potential
change in risk if modeling was performed with a different meteorological station (not the nearest
one) is presented in Section 4.5 of this document.
The HEM3 system estimates ambient concentrations at the geographic centroids of census blocks
(using the 2000 Census), and at other receptor locations that can be specified by the user. In
cases where the census block centroid was found to be located on facility property (as
determined from satellite imagery) the receptor was moved to the nearest off-site location. The
model accounts for the effects of multiple facilities when estimating concentration impacts at
each block centroid. In this assessment, we combined only the impacts of facilities within the
same source category, and assessed chronic exposure and risk only for census blocks with at
least one resident {i.e., locations where people may reasonably be assumed to reside rather than
receptor points at the fenceline of a facility). Chronic ambient concentrations were calculated as
the annual average of all estimated short-term (one-hour) concentrations at each block centroid.
Possible future residential use of currently uninhabited areas was not considered. Census blocks,
the finest resolution available in the census data, are typically comprised of approximately 40
people or about ten households.
In contrast to the development of ambient concentrations for evaluating long-term exposures,
which was performed only for occupied census blocks, worst-case short-term (one-hour)
concentrations were estimated both at the census block centroids and at points nearer the facility
that represent locations where people may be present for short periods, but generally no nearer
than 100 meters from the center of the facility (note that for large facilities, this 100-meter ring
could still contain locations inside the facility property). Since short-term emission rates were
needed to screen for the potential for hazard via acute exposures, and since the NEI contains only
annual emission totals, we applied the general assumption to all source categories that the
maximum one-hour emission rate from any source was ten times the average annual hourly
emission rate for that source. Average hourly emissions rate is defined as the total emissions for
a year divided by the total number of operating hours in the year. This choice of a factor of ten
for screening was originally based on engineering judgment. To develop a more robust peak-to-
mean emissions factor, and in response to one of the key messages from the SAB consultation on
our RTR Assessment Plan, we recently performed an analysis using a short-term emissions
dataset from a number of sources located in Texas (originally reported on by Allen et al.
11 For NEI8406, data from the Fairbanks, Alaska met station from the year 2001 modeled and for NEI46556, data
from St. Croix, Virgin Islands met station from the year 2005 was utilized.
Page 2-3
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
2004)[5], In that report, the Texas Environmental Research Consortium Project compared
hourly and annual emissions data for volatile organic compounds for all facilities in a heavily-
industrialized 4-county area (Harris, Galveston, Chambers, and Brazoria Counties, TX) over an
eleven-month time period in 2001. We obtained the dataset and performed our own analysis,
focusing that analysis on sources which reported emitting high quantities of HAP over short
periods of time. Based on our analysis, ratios of short-term event release rate to long-term
release rate varied from 0.00000004 to 74. The 99th percentile ratio was 9 (i.e., an event release
rate nine times the long-term average). Only 3 events were greater than 10 times the average,
and of these, only one exceeded 11, and that single event was 74 times the average. While there
are some documented emission excursions above this level, our analysis of the data from the
Texas Environmental Research Consortium suggests that this factor should cover more than 99%
of the short-term peak gaseous or volatile emissions from typical industrial sources. Details of
this analysis are presented in Appendix B.
Census block elevations for HEM3 modeling were determined nationally from the US
Geological Survey 1-degree digital elevation model (DEM) data files, which have a spatial
resolution of about 90 meters. Polar grid elevations (used in estimating short- and long-term
ambient concentrations) were evaluated at the highest elevation of any census block in that
sector. If a sector does not contain any blocks, the model defaults to the elevation of the nearest
block. If the elevation is not provided for the emission source, the model takes the average
elevation of all sectors of the nearest model ring.
In addition to utilizing receptor elevation to determine plume height, AERMOD adjusts the
plume's flow if nearby elevated hills are expected to influence the wind patterns.
2.2.3 Estimating human inhalation exposure
For this assessment, we used the annual average ambient air concentration of each HAP at each
census block centroid as a surrogate for the lifetime inhalation exposure concentration of all the
people who reside in the census block. That is, this risk analysis did not consider either the
short-term or long-term behavior (mobility) of the exposed populations and its potential
influence on their exposure.
We did not address short-term human activity in this assessment for two reasons. First, our
experience with the 1996 and 1999 NATA assessments (which modeled daily activity using
EPA's HAPEM model ) suggests that, given our current understanding of microenvironment
concentrations and daily activities, modeling short-term activity would, on average, reduce risk
estimates about 25% for particulate HAPs; it will also reduce risk estimates for gaseous HAPs,
but typically by much less. Second, basing exposure estimates on average ambient
concentrations at census block centroids may underestimate or overestimate actual exposure
concentrations at some residences. Further reducing exposure estimates for the most highly-
exposed residents by modeling their short-term behavior could add a systematic low bias to these
results.
We did not address long-term migration in this assessment nor population growth or decrease
over 70 years, instead basing the assessment on the assumption that each person's predicted
exposure is constant over the course of their lifetime, which is assumed to be 70 years. In
Page 2-4
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
assessing cancer risk, 3 metrics are generally estimated, the maximum individual risk (MIR)
which is defined as the risk associated with a lifetime of exposure at the highest concentration,
the population risk distribution, and the cancer incidence. This assumption of not considering
short or long-term population mobility does not bias the estimate of the theoretical MIR nor does
it affect the estimate of cancer incidence since the total population number remains the same. It
does, however, affect the shape of the distribution of individual risks across the affected
population, shifting it toward higher estimated individual risks at the upper end and reducing the
number of people estimated to be at lower risks, thereby biasing the risk estimates high. In
section 4 of this report, we demonstrate a method for accounting for long-term population
mobility, and show how it affects risk estimates for this source category.
When screening for potentially significant acute exposures, we used a modeled estimate of the
highest hourly ambient concentration at any off-site location as the surrogate for the maximum
potential acute exposure concentration for any individual.
2.2.4 Multipathway and environmental risk screening
The potential for significant human health risks due to exposures via routes other than inhalation
{i.e., multipathway exposures) was screened by first determining whether any sources emitted
any hazardous air pollutants known to be persistent and bioaccumulative in the environment
(PB-HAP). There are 14 PB-HAP compounds or compound classes identified for this screening
in EPA's Air Toxics Risk Assessment Library [6], They are cadmium compounds, chlordane,
dioxins, DDE, heptachlor, hexachlorobenzene, hexachlorocyclohexane, lead compounds,
mercury compounds, methoxychlor, polychlorinated biphenyls, polycyclic organic matter
(POM), toxaphene, and trifluralin.
Emissions of one PB-HAP - polycyclic organic matter (POM) - were identified in the inventory
for some petroleum refineries. These emissions were evaluated for potential non-inhalation risks
and adverse environmental impacts using EPA's recently-developed screening scenario which
was developed for use with the TRIM-FaTE12 model. This screening scenario uses
environmental media outputs from the peer-reviewed TRIM-Fate model to estimate the
maximum potential ingestion risks for any specified emission scenario by using a generic
farming/fishing exposure scenario that simulates a subsistence environment. The screening
scenario retains many of the ingestion and scenario inputs developed for EPA's Human Health
Risk Assessment Protocols (HHRAP) for hazardous waste combustion facilities [7], In the
development of the screening scenario a sensitivity analysis was conducted to ensure that its key
design parameters were established, such that environmental media concentrations were not
underestimated and to also minimize the occurrence of false positives for human health and
ecological endpoints. See Appendix C for a complete discussion of the development and testing
of the screening scenario, which we call TRIMScreen. For the purposes of multipathway risk
screening, the levels of concern below which risks were considered insignificant were 1 in a
million for lifetime cancer risk and a hazard quotient of 1.0 for noncancer impacts.
Additionally, we evaluated the potential for significant ecological exposures to non PB-HAP
from exceedances of chronic human health inhalation thresholds in the ambient air near these
facilities. Human health dose-response threshold values are generally derived from studies
12 EPA's Total Risk Integrated Methodology (General Information) http://epa.gov/ttn/fera/trim gen.html
Page 2-5
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
conducted on laboratory animals (such as rodents) and developed with the inclusions of
uncertainty factors that could be as high as 3000. Thus, these human threshold values are often
significantly lower than the level expected to cause an adverse effect in an exposed rodent. It
should be noted that there is a scarcity of data on the direct atmospheric impact of these HAPs on
other receptors, such as plants, birds, and wildlife. Thus, if the maximum inhalation hazard in an
ecosystem is below the level of concern for humans, we have generally concluded that
mammalian receptors should be at no risk of adverse effects due to inhalation exposures from
non PB-HAP, and have assured that other ecological receptors are also similarly not at any
significant risk from direct atmospheric impact. In some isolated cases where we have data
indicating potential adverse impacts on plants, birds, or other wildlife due to the direct
atmospheric impacts of specific HAPs, we note that as an uncertainty and, where possible, refine
our analysis by comparing our modeled impacts to available threshold values from the scientific
literature. The case study for Portland cement manufacturing contains an example of such an
analysis in Section 3.5.2.2 and Appendix K.
2.2.5 Acute Risk Screening and Refined Assessments
In establishing a scientifically-defensible approach for the assessment of potential health risks
due to acute exposures to HAPs, we have followed the same general approach that has been used
for developing chronic health risk assessments under the residual risk program. That is, we
developed a tiered, iterative approach. This tiered, iterative approach to risk assessment has been
endorsed by the National Academy of Sciences in its 1993 publication "Science and Judgment in
Risk Assessment" and subsequently was endorsed in the EPA's "Residual Risk Report to
Congress" in 1999.
The assessment methodology is designed to eliminate from further consideration those facilities
for which we have confidence that no acute adverse health effects of concern will occur. To do
so, we use what is called a tiered, iterative approach to the assessment. This means that we begin
with a screening assessment, which relies on minimal data and uses conservative assumptions
that in combination approximate a worst-case exposure. The result of this screening process is
that either the facility being assessed poses no risk of acute health effects {i.e., it "screens out"),
or that it requires further, more refined, assessment. A refined assessment could utilize site-
specific data on the temporal pattern of emissions, the layout of emission points at the facility,
the boundaries of the facility, and the local meteorology. In some cases, all of these site-specific
data would be needed to refine the assessment; in others, lesser amounts of site-specific data
could be used to determine that acute exposures are not a concern, and significant additional data
collection would not be necessary. The refinement process generally continues until the acute
risk either proves to be an important part of the assessment, or it screens out.
Acute health risk screening was performed as the first step. We used conservative assumptions
for emission rates, meteorology, and exposure location. We used the following worst-case
assumptions in our screening approach:
• Peak 1-hour emissions were assumed to equal 10 times the average 1-hour emission rates.
• For facilities with multiple emission points, peak 1-hour emissions were assumed to
occur at all emission points at the same time.
Page 2-6
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
• For facilities with multiple emission points, 1-hour concentrations at each receptor were
assumed to be the sum of the maximum concentrations due to each emission point,
regardless of whether those maximum concentrations occurred during the same hour.
• Worst-case meteorology (from one year of local meteorology) was assumed to occur at
the same time the peak emission rates occurred. The recommended EPA local-scale
dispersion model, AERMOD, was used for simulating atmospheric dispersion.
• A person was located downwind at the point of maximum impact during this same 1-hour
period, but no nearer to the source than 100 meters.
• The maximum impact was compared to multiple short-term health thresholds for the
chemical being assessed to determine if a possible acute health risk might exist. These
benchmarks are described in the next section of this report.
We performed more refined acute assessments for selected facilities for which the screening
assessment showed exceedances of short-term health thresholds. In general, refined assessments
proceed stepwise through the following activities:
• Examine aerial photographs of the site to determine if the impact area of concern is
outside the facility property boundary.
• Adjust the peak one-hour emissions default (multiplier of 10) to a more source-
specific value, where data are available and indicate that such an adjustment is
appropriate.
• Perform refined modeling using site-specific information. Refined modeling can
include running AERMOD (without HEM) to estimate 1-hour concentrations that
reflect the maximum concentration due to each emission point simultaneously
emitting at its maximum assumed short-term rate.
For facilities that still show off-site acute impacts above an HQ of 1 after refining the
assessment, we present the maximum HQ values for the available acute thresholds and discuss
the possible implications of these results in light of the available health effects information and
knowledge regarding the actual facility configuration.
2.2.6 Dose-Response Assessment
2.2.6.1 Sources of chronic dose-response information
Dose-response assessment information (carcinogenic and non-carcinogenic) for chronic
exposure (either by inhalation or ingestion) for the HAPs reported in the emissions inventory
were based on the EPA Office of Air Quality Planning and Standards' existing
recommendations for HAPs [5], also used for NATA 1999 \9\ This information has been
obtained from various sources and prioritized according to (1) conceptual consistency with
EPA risk assessment guidelines and (2) level of peer review received. The prioritization
process was aimed at incorporating into our assessments the best available science with
respect to dose-response information. The recommendations are based on the following
sources, in order of priority:
1) US Environmental Protection Agency (EPA). EPA has developed dose-response
assessments for chronic exposure for many of the pollutants in this risk assessment. These
Page 2-7
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
assessments typically provide a qualitative statement regarding the strength of scientific data
and specify a reference concentration (RfC, for inhalation) or reference dose (RfD, for
ingestion) to protect against effects other than cancer and/or a unit risk estimate (URE, for
inhalation) or slope factor (SF, for ingestion) to estimate the probability of developing
cancer. The RfC is defined as "an estimate (with uncertainty spanning perhaps an order of
magnitude) of a continuous inhalation exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a
lifetime." The RfD is "an estimate (with uncertainty spanning perhaps an order of
magnitude) of a daily oral exposure to the human population (including sensitive subgroups)
that is likely to be without an appreciable risk of deleterious effects during a lifetime." The
URE is "the upper-bound excess cancer risk estimated to result from continuous lifetime
exposure to an agent at a concentration of 1 |ig/m3 in air." The SF is "an upper bound,
approximating a 95% confidence limit, on the increased cancer risk from a lifetime exposure
to an agent. This estimate, [is] usually expressed in units of proportion (of a population)
affected per mg/kg-day..." EPA disseminates dose-response assessment information in
several forms, based on the level of review. The Integrated Risk Information System (IRIS)
[70], is an EPA database that contains scientific health assessment information, including
dose-response information, that has undergone interagency review. All IRIS assessments
completed since 1996 have also undergone independent external peer review. The current
IRIS process13 includes review by EPA scientists, interagency reviewers from other federal
agencies, and the public, and peer review by a panel of independent scientists external to
EPA. Dose-response assessments for some substances were prepared by the EPA Office of
Research and Development, but not submitted for EPA consensus. EPA has assembled the
results of many such assessments in the Health Effects Assessment Summary Tables
(HEAST) [77], which this assessment uses as a source of last resort for one HAP; 1,2,4-
trichlorobenzene. EPA's science policy approach, under the current carcinogen guidelines, is
to use linear low-dose extrapolation as a default option for carcinogens for which the mode
of action (MOA) has not been identified. We expect future EPA dose-response assessments
to identify nonlinear MO As where appropriate, and we will use those analyses (once they are
peer reviewed) in our risk assessments. At this time, however, there are no available
carcinogen dose-response assessments for inhalation exposure that are based on a nonlinear
MOA.
2) California Environmental Protection Agency (CalEPA). The CalEPA Office of
Environmental Health Hazard Assessment has developed dose-response assessments for
many substances, based both on carcinogenicity and health effects other than cancer. The
process for developing these assessments is similar to that used by EPA to develop IRIS
values and incorporates significant external scientific peer review. As cited in the CalEPA
Technical Support Document for developing their chronic assessments [72]: "The guidelines
for developing chronic inhalation exposure levels incorporate many recommendations of the
U.S. EPA (1994) [73] and NAS (NRC, 1994) [7¥]." The non-cancer information includes
available inhalation health risk guidance values expressed as chronic inhalation reference
exposure levels (REL) [75], CalEPA defines the REL as "the concentration level at or below
which no adverse health effects are anticipated in the general human population". CalEPA's
13 April 10, 2008 memorandum from EPA Deputy Administrator Marcus Peacock to Assistant Administrator George
Gray, subject "Implementation of Revised IRIS Process"
Page 2-8
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
quantitative dose-response information on carcinogenicity by inhalation exposure is
expressed in terms of the URE [7(5], defined similarly to EPA's URE.
3) US Agency for Toxic Substances and Disease Registry (ATSDR). ATSDR, which is part of
the US Department of Health and Human Services, develops and publishes Minimum Risk
Levels (MRLs) [77] for inhalation and oral exposure to many toxic substances. As stated on
the ATSDR web site: "Following discussions with scientists within the Department of Health
and Human Services (HHS) and the EPA, ATSDR chose to adopt a practice similar to that of
the EPA's Reference Dose (RfD) and Reference Concentration (RfC) for deriving substance
specific health guidance levels for non neoplastic endpoints." The MRL is defined as "an
estimate of daily human exposure to a substance that is likely to be without an appreciable
risk of adverse effects (other than cancer) over a specified duration of exposure". ATSDR
describes MRLs as substance-specific estimates to be used by health assessors to select
environmental contaminants for further evaluation. Exposures above an MRL do not
necessarily represent a threat, and MRLs are therefore not intended for use as predictors of
adverse health effects or for setting cleanup levels.
In developing chronic risk estimates, we adjusted dose-response values for some HAPs based on
professional judgment, as follows:
1) In the case of HAP categories such as glycol ethers, the most conservative dose-response
value of the chemical category was used as a surrogate for other compounds in the group for
which dose-response values were not available. This was done in order to examine, under
conservative assumptions, whether these HAPs that lack dose-response values may pose an
unacceptable risk and require further examination, or screen from further assessment.
2) This assessment bases risk estimates for formaldehyde on a dose-response value published in
1999 by the CUT Centers for Health Research. EPA is currently reviewing the existing IRIS
assessment for formaldehyde.
3) A substantial proportion of POM reported to EPA's national emission inventory (NEI) were
not speciated into individual compounds. As a result, it was necessary to apply the same
simplifying assumptions to this assessment that were used for the 1999 NATA study [75],
This assessment divided POM emissions into eight categories. Categories 1 and 2 were
assigned a URE equal to 5% of that for pure benzo[a]pyrene. Categories 3-7 were composed
of emissions that were reported as individual compounds. These compounds were placed in
the category with an appropriated URE. Category 8, composed of unspeciated carcinogenic
polynuclear aromatic hydrocarbons (a subset of POM called 7-PAH), was assigned a URE
equal to 18% of that for pure benzo[a]pyrene. Details of the development of the 5% and 18%
URE estimated are available at http://www.epa.gov/ttn/atw/sab/appendix-h.pdf.
The emissions inventory for the petroleum refining source category includes emissions of 73
individual compounds comprising 54 HAP. Of the 54 HAP, 21 are classified as known,
probable, or possible carcinogens, with quantitative cancer dose-response values available. The
21 HAP, their quantitative inhalation chronic cancer dose-response values, and the source of the
value are listed below in Table 2-2. This source category emits several other HAPs {i.e., cresols,
Page 2-9
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
styrene, and vinyl acetate) for which some limited or inadequate evidence exists for determining
carcinogenicity. Because these substances lack quantitative estimates of cancer potency, we did
not estimate risks for them. The POM compounds with chronic oral cancer dose-response values
available (for which a multipathway screening assessments was performed) are listed in Table
2-3.
The emissions inventory for the petroleum refining source category includes emissions of 45
HAP with quantitative chronic noncancer threshold values available. The 45 HAP, their
threshold values, and the source of the value are listed in Table 2-4.
Table 2-2. Dose-Response Values for Chronic Inhalation Exposure to Carcinogens
URE (unit risk estimate for cancer)14 = cancer risk per |jg/mJ of average lifetime
exposure. Sources: IRIS = EPA Integrated Risk Information System, CAL = California
EPA Office of Environmental Health Hazard Assessment, EPA/OAQPS = interim value
recommended by the EPA Office of Air Quality Planning and Standards,
Pollutant
CAS
Number15
URE (1/Mg/m3)
Source
Acetaldehyde
75070
2.2E-06
IRIS
Acrylonitrile
107131
6.8E-05
IRIS
Aniline
62533
1.6E-06
CAL
Benzene16
71432
7.8E-06
IRIS
Bis(2-ethylhexyl)phthalate
117817
2.4E-06
CAL
1,3-Butadiene
106990
3.0E-05
IRIS
Carbon tetrachloride
56235
1.5E-05
IRIS
1,4-Dichloro benzene
106467
1.1E-05
CAL
1,4-Dioxane
123911
7.7E-06
CAL
Ethylene dibromide
106934
6.0E-04
IRIS
Ethylene dichloride
107062
2.6E-05
IRIS
Formaldehyde
50000
5.5E-09
EPA/OAQPS
Methyl tert-butyl ether
1634044
2.6E-07
CAL
Methylene chloride
75092
4.7E-07
IRIS
Naphthalene
91203
3.4E-05
CAL
Pentachlorophenol
87865
5.1E-06
CAL
Polycyclic Organic Matter
246
17
EPA OAQPS12
- Benzo(a)anthracene
56553
1.1E-04
EPA OAQPS12
- Benzo(a)pyrene
50328
1.1E-03
EPA OAQPS12
14 The URE is the upper-bound excess cancer risk estimated to result from continuous lifetime exposure to an agent
at a concentration of 1 |ig/m3 in air. URE's are considered upper bound estimates meaning they represent a
plausible upper limit to the true value.
15 Chemical Abstract Services identification number. For groups of compounds that lack a CAS number we have
used a surrogate 3-digit identifier corresponding to the group's position on the CAA list of HAPs.
16 The EPA IRIS assessment for benzene provides a range of plausible UREs. This assessment used the highest
value in that range, 7.8E-06 per ug/m3. The low end of the range is 2.2E-06 per ug/m3.
17 Assigned the URE associated with a mixture of POM compounds having a similar potency. Details of this
method, also used in the 1999 National Air Toxics Assessment, are available at
http://www.epa.gov/ttn/atw/natal999/99pdfs/pomapproachian.pdf
Page 2-10
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
URE (unit risk estimate for cancer)14 = cancer risk per |jg/mJ of average lifetime
exposure. Sources: IRIS = EPA Integrated Risk Information System, CAL = California
EPA Office of Environmental Health Hazard Assessment, EPA/OAQPS = interim value
recommended by the EPA Office of Air Quality Planning and Standards,
Pollutant
CAS
Number15
URE (1/Mg/m3)
Source
- Benzo(b)fluoranthene
205992
1.1E-04
EPA OAQPS12
- Benzo(k)fluoranthene
207089
1.1E-04
EPA OAQPS12
- Chrysene
218019
1.1E-05
EPA OAQPS12
- Dibenzo(a,h)anthracene
53703
1.2E-03
EPA OAQPS12
- lndeno(1,2,3-cd)pyrene
193395
1.1E-04
EPA OAQPS12
1,1,2,2-Tetrachloroethane
79345
5.8E-05
IRIS
Tetrachloroethene
127184
5.9E-06
CAL
Trichloroethylene
79016
2.0E-06
CAL
Vinyl chloride
75014
8.8E-06
IRIS
Table 2-3. Dose-Response Values for Chronic Oral Exposure to Carcinogens
SF Coral slope factor for cancer) = cancer risk per ma/ka/d of averaae lifetime exposure. Sources: IRIS
= EPA Integrated Risk Information System, CAL = California EPA Office of Environmental Health
Hazard Assessment, EPA/OAQPS
= interim value recommended by the EPA Office of Air Quality
Planning and Standards.
CAS
SF
Pollutant
Number
(1/mg/kg/d)
Source
Polycyclic organic matter (POM)
246
0.5
EPA OAQPS
- Benzo(a)anthracene
56553
1
EPA OAQPS
- Benzo(a)pyrene
50328
7
EPA OAQPS
- Benzo(b)fluoranthene
205992
1
EPA OAQPS
- Benzo(k)fluoranthene
207089
1
EPA OAQPS
- Chrysene
218019
0.1
EPA OAQPS
- Dibenz(a,h)anthracene
53703
4
EPA OAQPS
- lndeno(1,2,3-cd)pyrene
193395
1
EPA OAQPS
Page 2-11
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 2-4. Dose-Response Values for Chronic Inhalation Exposure to Noncarcinogens
RfC (or similar inhalation values) = an estimate (with uncertainty spanning perhaps an order
of magnitude) of a continuous inhalation exposure to the human population (including
sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects
during a lifetime. Sources: IRIS = EPA Integrated Risk Information System, CAL = California
EPA Office of Environmental Human Health Assessment, HEAST = EPA Health Effects
Assessment Summary Table, ATSDR = US Agency for Toxic Substances and Disease
Registry.
Pollutant
CAS Number1"
RfC (mg/m3)
Source18
Acetaldehyde
75070
0.009
IRIS - L
Acrylonitrile
107131
0.002
IRIS - M
Aniline
62533
0.001
IRIS — L
Benzene
71432
0.03
IRIS - M
Bis(2-ethylhexyl)phthalate
117817
0.01
CAL - M
1,3-Butadiene
106990
0.002
IRIS - M
Carbon disulfide
75150
0.7
IRIS - M
Carbon tetrachloride
56235
0.19
ATSDR
Chlorobenzene
108907
1
CAL
Chloroform
67663
0.098
ATSDR
Cresols (mixed)
1319773
0.6
CAL
m-Cresol19
108394
0.6
CAL
Cumene
98828
0.4
IRIS - L
p-Dichlorobenzene
106467
0.8
IRIS - M
Diethanolamine
111422
0.003
CAL
1,4-Dioxane
123911
3.6
ATSDR
Ethyl benzene
100414
1
IRIS - L
Ethylene dibromide
106934
0.009
IRIS - M
Ethylene dichloride
107062
2.4
ATSDR
Ethylene glycol
107211
0.4
CAL
Formaldehyde
50000
0.0098
ATSDR
Glycol Ethers20
171
0.02
IRIS - M
- Ethylene glycol methyl ether
109864
0.02
IRIS - M
- Methoxytriglycol15
112356
0.02
IRIS - M
n-Hexane
110543
0.7
IRIS - M
Hydrochloric acid
7647010
0.02
IRIS - L
Hydrofluoric acid
7664393
0.014
CAL
Methanol
67561
4
CAL
Methyl chloride
74873
0.09
IRIS - M
Methyl isobutyl ketone
108101
3
IRIS - L/M
Methyl tert-butyl ether
1634044
3
IRIS - M
Methylene chloride
75092
1
ATSDR
Naphthalene
91203
0.003
IRIS - M
18 The descriptors L (low), M (medium), and H (high) have been added for IRIS RfC values to indicate the overall
level of confidence in the RfC value, as reported in the IRIS file.
19 The value for cresols (mixed) was used as a surrogate.
20 The value for ethylene glycol methyl ether was used as a surrogate for all glycol ethers.
Page 2-12
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
RfC (or similar inhalation values) = an estimate (with uncertainty spanning perhaps an order
of magnitude) of a continuous inhalation exposure to the human population (including
sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects
during a lifetime. Sources: IRIS = EPA Integrated Risk Information System, CAL = California
EPA Office of Environmental Human Health Assessment, HEAST = EPA Health Effects
Assessment Summary Table, ATSDR = US Agency for Toxic Substances and Disease
Registry.
Pollutant
CAS Number10
RfC (mg/m3)
Source18
Pentachlorophenol
87865
0.1
CAL
Phenol
108952
0.2
CAL
Styrene
100425
1
IRIS - M
Tetrachloroethene
127184
0.27
ATSDR
Toluene
108883
5
IRIS - H
1,2,4-Trichlorobenzene
120821
0.2
HEAST
1,1,1-Trichloroethane
71556
1
CAL
Trichloroethylene
79016
0.6
CAL
Vinyl acetate
108054
0.2
IRIS - H
Vinyl chloride
75014
0.1
IRIS - M
Xylenes (mixed)
1330207
0.1
IRIS - M
m-Xylene21
108383
0.1
IRIS - M
o-Xylene21
95476
0.1
IRIS - M
p-Xylene21
106423
0.1
IRIS - M
2.2.6.2 Sources of acute dose-response information
Hazard identification and dose-response assessment information for acute exposure were based
on OAQPS's existing recommendations for HAPs [79], In contrast to the approach for chronic
dose-response, no prioritization has been developed for acute noncancer reference values, in
large part due to the lack of coverage across many chemicals by any one set of reference values
specifically designed for this use. We looked to reference values developed for a variety of
purposes, including Reference Exposure Levels (RELs), Acute Exposure Guideline Levels
(AEGLs), and Emergency Response Planning Guideline (ERPGs) developed for 1-hour exposure
durations.
The California Environmental Protection Agency (CalEPA) has developed acute dose-response
assessments for many substances, expressing the results as acute inhalation reference exposure
levels, or RELs.
The acute REL (http://www.oehha.ca.gov/air/pdf/acuterel.pdf) is defined by CalEPA as "the
concentration level at or below which no adverse health effects are anticipated for a specified
exposure duration [20], RELs are based on the most sensitive, relevant, adverse health effect
reported in the medical and toxicological literature. RELs are designed to protect the most
sensitive individuals in the population by the inclusion of margins of safety. Since margins
21 The RfC for mixed xylene was used as a surrogate.
Page 2-13
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
of safety are incorporated to address data gaps and uncertainties, exceeding the REL does not
automatically indicate an adverse health impact."
The National Advisory Committee for Acute Exposure Guidelines (NAC-AEGL) is a Federal
Advisory Committee Act committee consisting of representatives from multiple federal agencies,
states, industry, non-governmental organizations, and several other nations that has been
responsible for developing Acute Exposure Guideline Levels, or AEGLs. As described in their
"Standing Operating Procedures (SOP) of the National Advisory Committee on Acute Exposure
Guideline Levels for Hazardous Substances" (http://www.epa.gov/opptintr/aegl/pubs/sop.pdf),
"the NRC's previous name for acute exposure levels — community emergency exposure levels
(CEELs) — was replaced by the term AEGLs to reflect the broad application of these values to
planning, response, and prevention in the community, the workplace, transportation, the military,
and the remediation of Superfund sites." This document further states that AEGLs "represent
threshold exposure limits for the general public and are applicable to emergency exposures
ranging from 10 minute to 8 hours." The document lays out the purpose and objectives of
AEGLs by stating that "the primary purpose of the AEGL program and the NAC/AEGL
Committee is to develop guideline levels for once-in-a-lifetime, short-term exposures to airborne
concentrations of acutely toxic, high-priority chemicals." In detailing the intended application of
AEGL values, the document states that "It is anticipated that the AEGL values will be used for
regulatory and nonregulatory purposes by U.S. Federal and State agencies, and possibly the
international community in conjunction with chemical emergency response, planning, and
prevention programs. More specifically, the AEGL values will be used for conducting various
risk assessments to aid in the development of emergency preparedness and prevention plans, as
well as real-time emergency response actions, for accidental chemical releases at fixed facilities
and from transport carriers." The NAC-AEGL defines AEGL-1 and AEGL-2 as:
"AEGL-1 is the airborne concentration (expressed as ppm or mg/m3) of a substance above
which it is predicted that the general population, including susceptible individuals, could
experience notable discomfort, irritation, or certain asymptomatic nonsensory effects.
However, the effects are not disabling and are transient and reversible upon cessation of
exposure."
"AEGL-2 is the airborne concentration (expressed as ppm or mg/m3) of a substance above
which it is predicted that the general population, including susceptible individuals, could
experience irreversible or other serious, long-lasting adverse health effects or an impaired
ability to escape."
"Airborne concentrations below AEGL-1 represent exposure levels that can produce mild
and progressively increasing but transient and nondisabling odor, taste, and sensory irritation
or certain asymptomatic, nonsensory effects. With increasing airborne concentrations above
each AEGL, there is a progressive increase in the likelihood of occurrence and the severity of
effects described for each corresponding AEGL. Although the AEGL values represent
threshold levels for the general public, including susceptible subpopulations, such as infants,
children, the elderly, persons with asthma, and those with other illnesses, it is recognized that
individuals, subject to unique or idiosyncratic responses, could experience the effects
described at concentrations below the corresponding AEGL."
Page 2-14
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
The American Industrial Hygiene Association (AIHA) has developed emergency response
planning guidelines (ERPGs) [27] for acute exposures at three different levels of severity. These
guidelines represent concentrations for exposure of the general population for up to 1 hour
associated with effects expected to be mild or transient (ERPG-1), irreversible or serious (ERPG-
2), and potentially life-threatening (ERPG-3).
ERPG values (http://www.aiha.org/ldocuments/Committees/ERP-erpglevels.pdf) are described
in their supporting documentation as follows: "Emergency Response Planning Guidelines
(ERPGs) were developed for emergency planning and are intended as health based guideline
concentrations for single exposures to chemicals. These guidelines {i.e., the ERPG Documents
and ERPG values) are intended for use as planning tools for assessing the adequacy of accident
prevention and emergency response plans, including transportation emergency planning and for
developing community emergency response plans. The emphasis is on ERPGs as planning
values: When an actual chemical emergency occurs there is seldom time to measure airborne
concentrations and then to take action." ERPG-1 and ERPG-2 values are defined by AIHA as
follows:
"ERPG-1 is the maximum airborne concentration below which it is believed that nearly all
individuals could be exposed for up to 1 hour without experiencing other than mild transient
adverse health effects or without perceiving a clearly defined, objectionable odor."
"ERPG-2 is the maximum airborne concentration below which it is believed that nearly all
individuals could be exposed for up to 1 hour without experiencing or developing irreversible
or other serious health effects or symptoms which could impair an individual's ability to take
protective action."
The emissions inventory for the petroleum refining source category includes emissions of 34
HAP with relevant and available quantitative acute dose-response threshold values. These
HAPs, the acute threshold values, and the source of the value are listed below in Table 2-5.
Page 2-15
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 2-5. Dose-Response Values for Acute Exposure
Pollutant
CAS
Number10
AEGL-1
(1-hr)
(mg/m )
AEGL-2
(1-hr)
(mg/m )
ERPG-1
(mg/m3)
ERPG-2
(mg/m3)
REL
(mg/m3)
Acetaldehyde
75070
81
490
81
490
0.47
Acrylonitrile
107131
10
130
22
77
Aniline
62533
30
46
Benzene
71432
170
2600
170
2600
1.3
Biphenyl
92524
61
1,3-Butadiene
106990
1500
12000
1500
12000
Carbon disulfide
75150
40
500
40
500
6.2
Carbon tetrachloride
56235
280
1200
280
1200
1.9
Chlorobenzene
108907
46
690
Chloroform
67663
310
310
0.15
Cumene
98828
250
1500
1,4-Dioxane
123911
61
1200
3
Ethylene dibromide
106934
130
180
Ethylene dichloride
107062
200
810
Formaldehyde
50000
1.1
17
1.1
17
0.094
Glycol Ether22
171
0.093
- Ethylene glycol methyl
ether
109864
0.093
- Methoxytriglycol16
112356
0.093
n-Hexane
110543
12000
Hydrochloric acid
7647010
2.7
33
2.7
33
2.1
Hydrofluoric acid
7664393
0.82
20
0.82
20
0.24
Methanol
67561
690
2700
690
2700
28
Methyl chloride
74873
1900
1900
Methylene chloride
75092
690
1900
690
1900
14
Methyl tert-butyl ether
1634044
180
2100
Phenol
108952
58
89
58
89
5.8
Styrene
100425
85
550
85
550
21
Tetrachloroethene
127184
240
1600
240
1600
20
Toluene
108883
750
4500
750
1900
37
1,1,1-Trichloroethane
71556
1300
3300
1300
3300
68
Trichloroethylene
79016
700
2400
700
2400
Vinyl acetate
108054
24
630
18
260
Vinyl chloride
75014
640
3100
640
3100
180
m-xylene
108383
22
p-xylene
106423
22
Xylenes (mixed)
1330207
560
4000
22
22 The value for ethylene glycol methyl ether was used as a surrogate for all glycol ethers.
Page 2-16
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
2.2.7 Risk characterization
2.2.7.1 General
The final product of the risk assessment is the risk characterization, in which the information
from the previous steps is integrated and an overall conclusion about risk is synthesized that is
complete, informative, and useful for decision makers. In general, the nature of this risk
characterization depends on the information available, the application of the risk information and
the resources available. In all cases, major issues associated with determining the nature and
extent of the risk are identified and discussed. Further, the EPA Administrator's March 1995
Policy for Risk Characterization [22] specifies that a risk characterization "be prepared in a
manner that is clear, transparent, reasonable, and consistent with other risk characterizations of
similar scope prepared across programs in the Agency." These principles of transparency and
consistency have been reinforced by the Agency's Risk Characterization Handbook [23], in
2002 by the Agency's information quality guidelines [24], and in the OMB/OSTP September
2007 Memorandum on Updated Principles for Risk Analysis23, and are incorporated in these
assessments.
Estimates of health risk are presented in the context of uncertainties and limitations in the data
and methodology. Through our tiered, iterative analytical approach, we have attempted to
reduce both uncertainty and bias to the greatest degree possible in this assessment. We have
provided summaries of risk metrics for the source category (including maximum individual
cancer risks and noncancer hazards, as well as cancer incidence estimates) along with a
discussion of the major uncertainties associated with their derivation to provide decision makers
with the fullest picture of the assessment and its limitations.
For each carcinogenic HAP included in this assessment that has a potency estimate available,
individual and population cancer risks were calculated by multiplying the corresponding lifetime
average exposure estimate by the appropriate URE. This calculated cancer risk is defined as the
upper-bound probability of developing cancer over a 70-year period {i.e., the assumed human
lifespan) at that exposure. EPA's upper bound estimates represent a "plausible upper limit to the
true value of a quantity" (although this is usually not a true statistical confidence limit).24 In
some circumstances, the true risk could be as low as zero; however, in other circumstances the
risk could also be greater.
Because EPA has determined that two of the carcinogens listed in Table 2-2 {i.e., POM and vinyl
chloride) have a mutagenic mode of action, [25], EPA's Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens [26] was applied to this assessment.
This guidance has the effect of increasing the mutagens' UREs by factors of 10 (for children
aged 0-1), 3 (for children aged 2-15), or 1.6 (for 70 years of exposure beginning at birth), as
appropriate for the exposed population. In this case, this has the effect of increasing the
estimated life time risks for these pollutants by a factor of 1.6. In addition, although only a small
23 Memorandum for the Heads of Executive Departments and Agencies - Updated Principles for Risk Analysis
(September 19, 2007), From Susan E. Dudley, Administrator, Office of Information and Regulatory Affairs, Office
of Management and Budget; and Sharon L. Hays, Associate Director and Deputy Director for Science, Office of
Science and Technology Policy (http://georgewbush-whitehouse.archives.gov/omb/memoranda/fy2007/m07-
24.pdf)
24 IRIS glossary (www.epa.gov/NCEA/iris/help_gloss.htm).
Page 2-17
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
fraction of the total POM emissions were reported as individual compounds, EPA expresses
carcinogenic potency for compounds in this group in terms of benzo[a]pyrene equivalence, based
on evidence that carcinogenic POM have the same mutagenic mechanism of action as does
benzo[a]pyrene. For this reason EPA implementation policy [27] recommends applying the
Supplemental Guidance to all carcinogenic PAHs for which risk estimates are based on relative
potency. Accordingly, we have applied the Supplemental Guidance to all unspeciated POM
mixtures.
Increased cancer incidence for the entire receptor population within the area of analysis was
estimated by multiplying the estimated lifetime cancer risk for the average individual within each
census block by the number of individuals residing in that block, then summing the results for all
modeled census blocks. This lifetime population incidence estimate was divided by 70 years to
obtain an estimate of the number of cancer cases per year for the entire modeling domain.
In the case of benzene, the high end of the reported cancer URE range was used in our
assessment to provide a conservative estimate of potential cancer risks. Use of the high end of
the range provides risk estimates that are approximately 3.5 times higher than use of the equally-
plausible low end value. Use of the low end of the range and its impact on risk estimates is
included as a sensitivity analysis in the discussion of uncertainties.
Unlike linear dose-response assessments for cancer, noncancer health hazards generally are not
expressed as a probability of an adverse occurrence. Instead, "risk" for noncancer effects is
expressed by comparing an exposure to a reference level as a ratio. The "hazard quotient" (HQ)
is the estimated exposure divided by a reference level (e.g., the RfC). For a given HAP,
exposures at or below the reference level (HQ<1) are not likely to cause adverse health effects.
As exposures increase above the reference level (HQs increasingly greater than 1), the potential
for adverse effects increases. For exposures predicted to be above the RfC, the risk
characterization includes the degree of confidence ascribed to the RfC values for the
compound(s) of concern (i.e., high, medium, or low confidence) and discusses the impact of this
on possible health interpretations.
The risk characterization for chronic effects other than cancer is expressed in terms of the HQ for
inhalation, calculated for each HAP at each census block centroid. As discussed above, RfCs
incorporate generally conservative uncertainty factors in the face of uncertain extrapolations,
such that an HQ greater than one does not necessarily suggest the onset of adverse effects. The
HQ cannot be translated to a probability that adverse effects will occur, and is unlikely to be
proportional to adverse health effect outcomes in a population.
Screening for potentially significant acute inhalation exposures also followed the HQ approach.
In this case, we divided the maximum estimated acute exposure by each available short-term
threshold value to develop an array of HQ values relative to the various acute endpoints and
thresholds. In general, when none of these HQ values are greater than one, there is no potential
for acute risk. In those cases where HQ values above one are seen, additional information is
used to determine if there is a potential for significant acute risks.
2.2.7.2 Mixtures
Since most or all receptors in these assessments receive exposures to multiple pollutants rather
Page 2-18
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
than a single pollutant, we estimated the aggregate health risks associated with all the exposures
from a particular source category combined.
To combine risks across multiple carcinogens, this assessment used the EPA mixtures
guidelines' [28, 29] default assumption of additivity of effects, and combined risks by summing
them using the independence formula in the mixtures guidelines.
In assessing noncancer hazard from chronic exposures, in cases where different pollutants cause
adverse health effects via completely different modes of action, it may be inappropriate to
aggregate HQs. In consideration of these mode-of-action differences, the mixtures guidelines
support aggregating effects of different substances in specific and limited ways. To conform to
these guidelines, we aggregated non-cancer HQs of HAPs that act by similar toxic modes of
action, or (where this information is absent) that affect the same target organ. This process
creates, for each target organ, a target-organ-specific hazard index (TOSHI), defined as the sum
of hazard quotients for individual HAPs that affect the same organ or organ system. All TOSHI
calculations presented here were based exclusively on effects occurring at the "critical dose"
(i.e., the lowest dose that produces adverse health effects). Although HQs associated with some
pollutants have been aggregated into more than one TOSHI, this has been done only in cases
where the critical dose affects more than one target organ. Because impacts on organs or
systems that occur above the critical dose have not been included in the TOSHI calculations,
some TOSHIs may have been underestimated. As with the HQ, the TOSHI should not be
interpreted as a probability of adverse effects, or as strict delineation of "safe" and "unsafe"
levels. Rather, the TOSHI is another measure of the potential for adverse health outcomes
associated with pollutant exposure, and health scientists and risk managers should take care to
clearly communicate its uncertainties and limitations when characterizing risks.
Because of the conservative nature of the acute inhalation screening approach and the transient
nature of emissions fluctuations and potential exposures, acute impacts were screened on an
individual pollutant basis, not using the TOSHI approach.
2.3 Results Summary and Risk Characterization
In this section, the results of the risk assessment for the petroleum refining MACT 1 source
category are presented in terms of the following information:
1) A narrative description of the source category, including a discussion of the processes
involved and the number of facilities EPA knows or expects are affected by the petroleum
refinery MACT 1 standard;
2) A table of emissions for the entire category showing HAP emitted, total source category
emission rates for each HAP, and numbers of facilities reporting emissions of each HAP;
3) A table summarizing the chronic inhalation risk results showing the number of facilities
modeled, the number of people within 50 km, the MIR for the entire source category, the
number of facilities for which the facility-specific MIR exceeds specific cancer and
noncancer benchmarks, the number of people for whom the risks exceed the same
benchmarks, the estimated total cancer incidence, and identifying the specific HAPs
Page 2-19
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
contributing the most to those risks (HAPs identified as "drivers" include those contributing
the most to the risk metric, up to 90% of its value). In addition, this table indicates the
maximum HQ from the acute inhalation screening and an indication of how many facilities
showed HQ values above 1;
4) In those cases where the acute inhalation screening showed an HQ value greater than 1 for
any combination of source and pollutant, a table summarizing the acute screening results
showing available acute dose-response values for each affected pollutant, for three effect
levels (none, mild, and severe), if available, the maximum acute screening exposure
estimated, and the associated HQ values;
5) A narrative summarizing the risk characterization for the entire source category.
Detailed facility-level results for both chronic and acute inhalation risk assessments can be found
in Appendix D.
2.3.1 Source Category Description and Summary of Emissions
Petroleum Refineries are facilities engaged in refining and producing products made from crude
oil or unfinished petroleum derivatives including gasoline, naphtha, kerosene, jet fuels, distillate
fuel oils, residual fuel oils, and lubricants. In the list of MACT source categories (57 FR 31576,
July 16, 1992), EPA listed two separate and distinct petroleum refinery source categories:
(1) Petroleum Refineries - Catalytic Cracking (Fluid and Other) Units, Catalytic Reforming
Units, and Sulfur Plant Units and (2) Petroleum Refineries - Other Sources Not Distinctly Listed.
The MACT standard for the "Other Sources Not Distinctly Listed" source category (40 CR 63,
subpart UU) was promulgated first, on August 18, 1995 in 60 FR 43244,. Therefore, it is
commonly referred to as Petroleum Refineries MACT 1. MACT 2, which addresses the
Petroleum Refineries - Catalytic Cracking (Fluid and Other) Units, Catalytic Reforming Units,
and Sulfur Plant Units source category, was promulgated on April 11, 2002 (67 FR 17761).
Because MACT 1 and MACT 2 represent two separate and distinct source categories which were
subjected to MACT standards at different times, EPA will assess the residual risk and make
decisions on future regulations under section 112(f)(2) of the CAA independently. The data
presented in this document are only for MACT 1, the "Petroleum Refineries, Other Sources Not
Distinctly Listed" source category. Residual risk for MACT 2, Petroleum Refineries - Catalytic
Cracking (Fluid and Other) Units, Catalytic Reforming Units, and Sulfur Plant Units, will be
assessed by EPA at a later date in a later phase of RTR.
The petroleum refinery process units covered by MACT 1 include, but are not limited to, thermal
cracking, vacuum distillation, crude distillation, hydroheating and hydrorefining, isomerization,
polymerization, lube oil processing, and hydrogen production. Emissions originate from various
process vents, storage vessels, wastewater streams, loading racks, marine tank vessel loading
operations, and equipment leaks associated with refining facilities.
To create the ANPRM data set for Petroleum Refineries MACT 1, EPA started by retrieving all
facilities identified by the Petroleum Refineries Other Sources Not Distinctly Listed MACT code
(MACT Code 0503) in Version 1.0 of the 2002 NEI (February 2006). Next, we performed an
Page 2-20
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
engineering review of these facilities and updated the dataset with site-specific benzene
emissions data for 22 refineries as provided by the American Petroleum Institute. The goal of
the engineering review was to identify readily-apparent limitations and issues with the emissions
data and to make changes to the dataset where possible to address these issues and decrease the
uncertainties associated with the assessment. EPA requested comments on the adjusted 2002
NEI data as part of the Risk and Technology Review (RTR) Phase IIANPRM in March 2007
(72FR14734), making it available for a 60-day public comment period. Comments and
corrections were evaluated and incorporated into the inventory. A detailed discussion of the
changes to the inventory as a result of the ANPRM process and the risk characterization effort
are presented in the Draft Residual Risk Assessment for MACT I Petroleum Refining Sources
and the Petroleum Refineries NPRM Data Input File available in the Risk and Technology
Review Docket, ID No. EPA-HQ-OAR-2006-0895 at www.regulations.gov.
In August 2007, a NPRM was published making the source category dataset available for a
second 60-day comment period, which was subsequently re-opened for another 50 days. Again
the comments and corrections were evaluated and incorporated into the inventory. The final
petroleum refinery database contained information for 156 facilities, and this is thought to
represent the entire source category. Total HAP emissions did not change dramatically as a
result of these comments, dropping by only about 2%. Notably, emissions of metal HAP were
removed from the inventory since they cannot be emitted by the specific emission points covered
by the petroleum refinery MACT 1. Instead, these emissions are thought to be emitted by the
emission points covered by the petroleum refinery MACT 2. Details on the development of the
emissions and source data for this source category are discussed in Section 2.2.1. The emissions
data and modifications made to the NEI data are available in the Petroleum Refineries Baseline
Data Input File available in the Risk and Technology Review Docket, ID No. EPA-HQ-OAR-
2006-0895 at www.regulations.gov.
We also note that recent Canadian and European studies [30,31] indicate that emissions from
some refineries are significantly higher than amounts estimated using standard techniques such
as emission factors or AP-42 equations. This bias is apparently caused by omission (e.g.,
process leaks into cooling towers) or mischaracterization of significant emission sources, and the
same quantification issues appear to exist in the US. We have performed additional analyses
(i.e., model plant analysis and model-to-monitor comparison) in an attempt to characterize the
possible magnitude of uncertainty in emissions estimates. The model-to-monitor analysis
suggests that we may be underestimating emissions of benzene at two refineries in the Houston
area by a factor of 2. The model plant analysis suggests that we may be underestimating risk by
up to a factor of 3. Technical memoranda explaining these analyses can be found in the Docket
under "Statistical Comparison of Monitored and Modeled Ambient Benzene Concentrations
Near Two Petroleum Refineries in Texas City, TX" and "Model Plant Analysis of Residual Risk
From Petroleum Refinery Emissions."
Organic chemicals account for the majority of the total mass of HAPs emitted by MACT 1
petroleum refinery sources, with toluene, benzene, xylene,, hexane, methanol, ethyl benzene,
methyl isobutyl ketone, 2,2,4-trimethylpentane, methyl tert-butyl ether, hydrogen fluoride,
naphthalene, diethanolamine, cumene, 1,3-butadiene, carbonyl sulfide, phenol, hydrochloric
acid, hydrogen fluoride, cresols, tetrachloroethylene, ethylene glycol, chloroform,
Page 2-21
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
trichloroethylene, 16- poly cyclic aromatic hydrocarbons (PAH), and poly cyclic organic matter
accounting for 99 percent of the HAPs mass emitted across the source category. A range of
persistent and bioaccumulative HAP (PB HAP) [32] emissions were included in the NPRM
dataset, including various PAH.
The final petroleum refinery database contained information for 156 facilities, and this is thought
to represent the source category in its entirety. The emissions data and modifications made to
the NEI data are available in the Petroleum Refineries Baseline Data Input File available in the
Risk and Technology Review Docket, ID No. EPA-HQ-OAR-2006-0895 at
www.regulations.gov. Table 2-6 provides information summarizing emissions for this source
category.
Comments received on the emission inventory used for the draft baseline risk assessment for
petroleum refinery MACT 1 sources were evaluated and incorporated into the final inventory if
deemed appropriate and reasonable from an engineering standpoint. The comments covered 101
facilities, and included data provided for three facilities not contained in the original dataset.
After evaluating the comments, emissions data were corrected at 48 facilities, emission point
identifiers were corrected at 3 facilities, stack parameters were revised at 4 facilities, and location
data were corrected at 61 facilities. The final petroleum refinery emission inventory contains
information for 156 facilities representing the entire source category.
Nationwide refinery HAP emission estimates did not change dramatically as a result of the
revisions made pursuant to the public comments, dropping by only about 2 percent. In addition,
metal HAP emissions were removed from the inventory because they are not emitted by the
emission points covered by the petroleum refinery MACT 1. Metal HAPs are emitted by other
source categories in refineries, most notably by the emission points covered by the petroleum
refinery MACT 2. These metal HAP emissions will be included in the RTR assessment for that
category. Appendix A provides a comparison of the risk estimates for this source category,
before and after processing the NPRM revisions.
Table 2-6. Summary of Emissions from the MACT 1 Petroleum Refining Source Category
IIAP'
Emissions
(tpy)
Number of
Facilities
Reporting
HAP (156
facilities in
data set)
Prioritized Inhalation Dose-Response Value Identified
by OAQPSb
PB-
HAP?
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Noncancer?
Health Benchmark
Values for Acute
Noncancer?
Toluene
1,784
136
V
V
Xylenes (Mixture of o, m, and p
Isomers)
1,060
129
V
V
Hexane
1,047
130
V
V
Benzene
690
146
V
V
V
Methanol
569
61
V
V
Methyl Tert-Butyl Ether
349
45
V
V
V
p-Xylene
337
13
V
V
Ethyl Benzene
251
130
V
m-Xylene
138
17
V
V
Page 2-22
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 2-6. Summary of Emissions from the MACT 1 Petroleum Refining Source Category
IIAP'
Emissions
(tpy)
Number of
Facilities
Reporting
HAP (156
facilities in
data set)
Prioritized Inhalation Dose-Response Value Identified
by OAQPSb
PB-
HAP?
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Noncancer?
Health Benchmark
Values for Acute
Noncancer?
2,2,4-Trimethylpentane
132
47
Methyl Isobutyl Ketone
92
5
V
Naphthalene
82
104
V
V
Hydrochloric Acid
73
19
V
V
o-Xylene
72
21
V
Hydrogen Fluoride
53
34
V
V
Cumene
53
81
V
V
Diethanolamine
41
22
V
Phenol
32
42
V
V
Ethylene Glycol
22
8
V
1,3-Butadiene
17
71
V
V
V
Cresol
16
27
V
T etrachloroethy lene
15
34
V
V
V
Formaldehyde
9
28
V
V
V
16-PAH
8
2
V
Styrene
5
25
V
V
Fluoranthene
5
10
V
PAH, total
4
45
V
Polycyclic Organic Matter
4
9
V
V
Carbon Disulfide
4
15
V
V
Biphenyl
3
21
V
Carbon Tetrachloride
3
5
V
V
V
Glycol Ethers
3
4
V
V
Carbonyl Sulfide
2
16
Anthracene
1
8
V
1,1,1 -T richloroethane
1
5
V
V
Ethylene Dibromide
0.7
8
V
V
V
Ethylene Dichloride
0.7
11
V
V
V
Chloroform
0.6
7
V
V
Phenanthrene
0.6
10
V
T richloroethylene
0.6
5
V
V
V
Vinyl Acetate
0.5
3
V
V
Benzo[g,h,i,]Perylene
0.2
23
V
Methylene Chloride
0.2
4
V
V
V
Acetaldehyde
0.2
14
V
V
V
Chlorobenzene
0.1
4
V
V
Vinyl Chloride
0.1
1
V
V
V
Acetophenone
0.08
1
Quinoline
0.04
1
Page 2-23
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 2-6. Summary of Emissions from the MACT 1 Petroleum Refining Source Category
IIAP'
Emissions
(tpy)
Number of
Facilities
Reporting
HAP (156
facilities in
data set)
Prioritized Inhalation Dose-Response Value Identified
by OAQPSb
PB-
HAP?
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Noncancer?
Health Benchmark
Values for Acute
Noncancer?
p-Phenylenediamine
0.03
1
Dibenzofuran
0.03
2
V
Aniline
0.03
1
V
V
V
1,4-Dioxane
0.01
2
V
V
V
Methyl Chloride
0.01
2
V
V
Ethylene Glycol Methyl Ether
0.007
2
V
V
1,1,2,2-Tetrachloroethane
0.005
1
V
Pentachlorophenol
0.002
1
V
V
Acrylonitrile
0.002
1
V
V
V
B is (2 -Ethy lhexy l)Phthalate
0.001
3
V
V
Methoxytriglycol
0.001
1
V
V
Benzo[a]Pyrene
0.0006
4
V
V
m-Cresol
0.0005
1
V
1,4-Dichlorobenzene
0.0003
2
V
V
1,2,4-Trichlorobenzene
0.0003
1
Benzo [k]Fluoranthene
0.00005
2
V
V
Chrysene
0.00003
3
V
V
B enz [a] Anthracene
0.00002
2
V
V
Dibenzo [a,h] Anthracene
0.000004
2
V
V
Benzo [b]Fluoranthene
0.000002
3
V
V
Fluorene
0.0000007
2
V
Pyrene
0.0000002
2
V
Acenaphthene
0.0000002
1
V
Perylene
0.0000001
1
V
Indeno [ 1,2,3-c,d]Pyrene
0.00000003
2
V
a Notes for how HAP were speciated for risk assessment:
• For emissions of any chemicals or chemical groups classified as poly cyclic organic matter (POM), emissions were grouped
into POM subgroups as found on the EPA's Technology Transfer Network website for the 1999 National-Scale Air Toxics
Assessment at http://www.epa.gov/ttn/atw/natal999/nsata99.html. Those that are grouped and do not have individual dose-
response values are not checked in the table above.
• For emissions reported genetically as "Glycol Ethers" or as specific glycol ethers not found on EPA's Technology Transfer
Network website for air toxics (see footnote b), emissions will be treated as ethylene glycol methyl ether.
b Specific dose-response values for each chemical are identified in section 2.2.6 of this document and on EPA's Technology
Transfer Network website for air toxics at http://www.epa.gov/ttn/atw/toxsource/summarv.html. The acute benchmarks
considered were the REL, AEGL-1 (1-hour), ERPG-1, AEGL-2 (1-hour), and ERPG-2.
2.3.2 Source Category Inhalation Risk Assessment Results
The petroleum refining source category consists of 156 facilities, all of which were included in
this risk assessment. Refineries are located throughout the United States; we estimate that
approximately 90 million people live within 50 kilometers of at least one petroleum refinery.
Page 2-24
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Table 2-7. Summary of Source Category Level Risks
Result
Facilities in Source Category
Number of Facilities Estimated to be Subject to
MACT in Source Category
Number of Facilities Identified in NEI and Modeled
in Screening Risk Assessment
Cancer Risks
Maximum Individual Lifetime Cancer Risk (in 1
million) from any Facility in the Category
Number of Facilities with Maximum Individual Lifetime Cancer Risk:
Greater than or equal to 100 in 1 million 0
Greater than or equal to 10 in 1 million 5
156
156
30
Greater than or equal to 1 in 1 million
77
Chronic Noncancer Risks
Maximum Respiratory Hazard Index 0.3
Number of Facilities with Maximum Respiratory Hazard Index:
Greater than 1 0
Acute Noncancer Screening Results
Maximum Acute Hazard Quotient
Number of Facilities With Potential for Acute Effects
Acute Noncancer Refined Results
Maximum Acute Hazard Quotient
50,
20, 6
20
8, 0.06
5, 2, 0.06
Number of Facilities With Potential for Acute
Effects
Population Exposure
Number of People Living Within 50 Kilometers of
Facilities Modeled
Number of People Exposed to Cancer Risk:
Greater than or equal to 100 in 1 million
Greater than or equal to 10 in 1 million
Greater than or equal to 1 in 1 million
Number of People Exposed to Noncancer Respiratory Hazard Index:
Greater than 1 0
Estimated Cancer Incidence (excess cancer cases per
year)
90,000,000
0
4,000
460,000
0.03 to 0.05
for Petroleum Refineries
HAP "Drivers"
n/a
n/a
naphthalene, POM
n/a
naphthalene, POM, benzene, ethylene
dibromide, 1,3-butadiene,
tetrachloroethylene, methyl tert-butyl
ether, carbon tetrachloride
naphthalene, POM, benzene, ethylene
dibromide, 1,3-butadiene,
tetrachloroethylene, methyl tert-butyl
ether, carbon tetrachloride, ethylene
dichloride, vinyl chloride
diethanolamine
n/a
Benzene (REL)
hydrofluoric acid (REL, AEGL-1/ERPG-
1)
benzene, hydrofluoric acid,
benzene (REL, AEGL- 1/ERPG-1)
hydrofluoric acid (REL, AEGL-1/ERPG-
1, AEGL-2/ERPG-2)
benzene, hydrofluoric acid
n/a
n/a
n/a
n/a
n/a
n/a
Page 2-25
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Table 2-7. Summary of Source Category Level Risks for Petroleum Refineries
Result HAP "Drivers"
Facilities in Source Category
Contribution of HAP to Cancer Incidence
benzene
48%
n/a
naphthalene
21%
n/a
POM
15%
n/a
1,3-butadiene
5%
n/a
tetrachloroethylene
4%
n/a
Table 2-8. Summary of Acute Refined Results
or Petroleum Refineries
Refined Results
MAXIMUM ACUTE HAZARD QUOTIENTS
ACUTE DOSE-RESPONSE VALUES
HAP
Max. 1-
hr. Air
Cone.
(mg/m3)
Based on
REL
Based on
AEGL-
1/ERPG-1
Based on
AEGL-
2/ERPG-2
REL
(mg/m3)
AEGL-1
(1-hr)
(mg/m3)
ERPG-1
(mg/m3)
AEGL-2
(1-hr)
(mg/m3)
ERPG-2
(mg/m3)
benzene
10
8
0.06
0.004
1.3
170
170
2600
2600
hydrofluoric acid
1.3
5
2
0.06
0.24
0.82
0.82
20
20
Notes on Process:
1) Acute screening was performed for all emitted HAP with available acute dose-response values. Where acute screening
HQ values exceeded 1, refined analysis was performed. Only those pollutants whose refined HQs were equal to or
greater than 1 for at least one acute threshold value are shown in the table.
2) HAP with available acute dose-response values which are not in the table do not carry any potential for posing acute
health risks, based on an analysis of currently available emissions data.
Notes on Acute Dose-Response Values:
REL - California EPA reference exposure level for no adverse effects. Most, but not all RELs are for 1-hour exposures.
AEGL - Acute Exposure Guideline Levels represent exposure (1-hour) limits for the general public.
AEGL-1 is the exposure level above which it is predicted that the general population, including susceptible individuals,
could experience effects that are notable discomfort, but which are transient and reversible upon cessation of exposure.
AEGL-2 is the exposure level above which it is predicted that the general population, including susceptible individuals,
could experience irreversible or other serious, long-lasting adverse health effects or an impaired ability to escape.
ERPG - Emergency Response Program Guidelines represent emergency exposure (1-hour) limits for the general public.
ERPG-1 is the maximum level below which it is believed that nearly all individuals could be exposed for up to 1 hour
without experiencing other than mild, transient adverse health effects.
ERPG-2 is the maximum exposure below which it is believed that nearly all individuals could be exposed for up to 1
hour without experiencing or developing irreversible or other serious health effects or symptoms which could impair an
individual's ability to take protective action.
2.3.3 Risk Characterization
The maximum individual cancer risk for the petroleum refining source category is 30 in a
million. The maximum individual cancer risk for the source category as a whole is dominated by
the risks associated with emissions of naphthalene and polycyclic organic matter (POM);
however, the maximum individual cancer risk level associated with each facility and the specific
pollutants which contribute to most to that level vary significantly from facility to facility. The
total cancer incidence for the source category was estimated to be between 0.03 and 0.05 cancer
cases per year, or about 1 case in every 20 to 30 years (this range of cancer incidence depends on
the range of the IRIS cancer potency factors for benzene, each end of which is considered
equally plausible). The cancer incidence for the source category is dominated by risks associated
Page 2-26
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
with benzene and naphthalene. The estimated maximum individual cancer risk exceeded 10 in a
million at 5 facilities and exceeded 1 in a million at 77 facilities. Based on the assumption that
all individuals are exposed for 70 years, approximately 4,000 people were estimated to have
cancer risks above 10 in a million and approximately 460,000 people were estimated to have
cancer risks above 1 in a million.
Chronic noncancer inhalation risks were not identified as significant, with the maximum chronic
target organ specific hazard index associated with the cumulative impacts of all noncarcinogenic
HAP emitted by these sources being less than 1. While there were reported emissions of one
persistent HAP (polycyclic organic matter, or POM) from this source category, our multipathway
screening indicated that neither significant ingestion health risks nor environmental risks would
be anticipated to result from exposures to media concentrations associated with the deposition of
these emissions. No other potential environmental risks, including those as a direct result of
exposure of flora and fauna to ambient air concentrations, were identified.
As mentioned in the discussion of dose-response values, we calculated benzene risks throughout
this assessment using the upper end of the range of cancer unit risk estimates, or URE, identified
in IRIS. Specifically, IRIS recommends a range of URE for benzene, 2.2 x 10"6to 7.8 x 10"6 per
|ig/m3, explaining that each has equal scientific plausibility. Since benzene is an important risk
driver for many petroleum refineries, we also estimated the risk using the lower end of this range
to assess the impact of URE choice on the final results. However, we found that the maximum
individual cancer risk (MIR) is driven by pollutants other benzene. Thus, the choice of benzene
URE was seen to have little impact on the MIR for the source category. Additionally, without
re-assessing the risks for each facility, we made a very rough projection of the impact of the
benzene URE on the number of people whose individual risks are above 1 in a million, and
estimated that use of the low end URE may reduce this population from 460,000 to about
275,000. Since benzene emissions are prevalent throughout the source category, however, total
incidence estimates were seen to drop on average by about 35% (to 0.03 cases per year) when the
low end URE was chosen.
While maximum individual cancer risks vary significantly from facility to facility (see Appendix
D), they are typically dominated by risks from fugitive emissions which are responsible for about
52 percent the cancer risk. Leaks into process cooling water which are ultimately released to the
atmosphere through cooling towers were not seen to contribute significantly to the emissions
inventory or to cancer risks. However, recent studies [30, 31] suggest that these emissions,
among others, may be underestimated and underreported in current emissions inventories, but to
an unknown extent.
The initial acute screening risk calculations suggested that 20 petroleum refineries showed
potential 1-hour exposures above an acute health benchmark, but the lack of readily available
detailed property boundary information for many of the facilities evaluated made it difficult to
determine whether the points of maximum concentration were on- or off-site. The facilities that
exceeded an acute HQ of 1 were targeted for more refined evaluation. The refined evaluation
included inspecting aerial maps of the sites to see if the locations for predicted potential
exceedances occurred inside or outside the facility boundary. While exact facility boundaries
Page 2-27
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
were not visible, the aerial photographs allowed us to assess locations likely to be accessible to
the public. Results of these mapping efforts can be found in Appendix E.
Potential acute impacts of concern were identified in the acute inhalation screening assessment
for facilities emitting benzene and hydrofluoric acid. Emissions of each of these pollutants
showed the potential to create maximum offsite impacts corresponding to 50 and 20 times the
acute REL, respectively {i.e., for benzene, HQREl=50; for hydrofluoric acid, HQREl=20). One
potential exceedance of an AEGL value was identified for hydrofluoric acid (HQAegl-i=6).
Subsequent refinement to the acute analysis discussed below, indicates the potential for acute
concerns at 8 out of the 156 facilities, with maximum potential offsite impacts at 8 and 5 times
the acute reference exposure level (REL) for benzene and hydrofluoric acid {i.e., for benzene,
maximum HQREL= 8, 5 facilities with potential HQREL greater than 1; and for hydrofluoric acid,
maximum HQREi=5, 3 facilities with potential HQakgi.-i greater than 1), and a potential
exceedance of the acute exposure guideline level (AEGL-1) and the emergency response
planning guideline (ERPG-1) level for hydrofluoric acid (HQ\kgi.-i = HQERPG-i = 2) at one
facility. There were no potential exceedances of the AEGL-1 or the ERPG-1 levels for benzene
(maximum HQAEgl-i = HQ ERpg-i = 0.06). There were also no potential exceedances of the
AEGL-2 level for hydrofluoric acid (maximum HQAEgl-2 = 0.06). According to CalEPA, acute
exposure to hydrofluoric acid can be associated with eye and respiratory irritation and acute
exposure to benzene can be associated with reproductive/developmental effects (see
http://www.oehha.ca.gov/air/pdf/acuterel.pdf). Maximum predicted acute HQ values for each of
the facilities are presented in Appendix D and the refined acute analysis and results are presented
in Appendix E. We note that the number of facilities with potential acute concerns (8) is small
relative to the total number of facilities in the source category (156). The number of people
living within a mile of the 5 sites with potential acute benzene impacts is about 3000; the number
of people living within a mile of the 3 sites with potential acute hydrofluoric acid impacts is
about 8000. Concerning potential acute benzene exposures, while the maximum benzene HQREL
value is 8, the corresponding HQAEgl-i value is 0.06. This places estimated acute exposures in a
"gray area" that is well below the level "above which the general population, including sensitive
individuals, could experience notable discomfort, irritation, or certain asymptomatic, nonsensory
effects" {i.e., the AEGL-1), but still well above the level at which we can rule out the possibility
of acute health impacts {i.e., the REL). Regarding potential acute hydrofluoric acid exposures,
we note that the source of the emissions is fugitive emissions, indicating that the reported
emissions are estimates based on long-term consideration of leaking pipes, equipment, etc. In
general, such emissions do not vary dramatically in time, and our use of the emissions multiplier
of 10 in estimating acute exposures from long-term average emissions estimates is likely
conservative. We note that our screening indicates no potential to exceed the AEGL-2 level for
hydrofluoric acid, defined as an exposure level "above which the general population, including
susceptible individuals, could experience irreversible or other serious, long-lasting adverse health
effects or an impaired ability to escape." We conclude that short-term exceedances of the
AEGL-l/ERPG-1 level are possible, but unlikely for 1 facility and that HQREL values greater
than 1 may still be possible for 3 facilities, indicating that we cannot completely rule out acute
exposures of concern at these facilities.
It is important to note that acute risk estimates were based on the annual emission rate multiplied
by a factor of 10. We were not able to refine our estimates of peak emission rates beyond the
Page 2-28
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
default factor of 10 times the annual average hourly rate, nor were we able to simulate the typical
distribution of peak emission events between emission points at a facility, making the final
results of our acute assessments uncertain but probably conservative. Overall, these results
neither prove that adverse acute health effects will occur, nor do they rule out the possibility,
should all the assumed conditions of exposure (i.e., simultaneous tenfold emission rate, worst-
case meteorology, and presence of a human receptor) be met.
In addition to the inhalation risk results discussed above, human health multipathway risks were
evaluated using screening techniques for POM emissions. These results indicated that the
potential for significant cancer or noncancer human health risks due to the ingestion of these
pollutants was low. Only a small fraction of the POM mass reported for the facilities evaluated
was reported as PAH species that can be modeled individually. Consequently, the following
modeling approach was used. POM species and groups reported in NEI, including emissions
reported as mixtures of POM, were assigned to POM categories according to estimated cancer
potency (i.e., using the same methods employed to evaluate inhalation risk from POM).
Multipathway fate and transport modeling of POM emissions was conducted using
benzo[a]pyrene as a surrogate chemical for the behavior of other POM in the environment. Total
emissions of POM (including unspeciated PAHs) were evaluated using a risk screening method
based on a hypothetical ingestion exposure modeling scenario. Using this approach, we were
able to confirm for 133 of the 156 facilities that risks via ingestion exposures were well below
levels of concern. For the remaining 23 facilities, the results of the screen were less definitive,
and incremental lifetime cancer risks modeled using the hypothetical scenario were estimated to
be as high as 67 in a million at one facility. However, because we used a conservative speciation
profile to estimate the risk contribution of individual POM compounds to the total POM risks,
we believe that this screening result is highly conservative, and that actual PAH risks due to
ingestion are much lower. As a result, we did not further refine our assessment of multipathway
human health risks.
No ecological benchmarks were exceeded in our multipathway screening. Contaminant
concentrations were evaluated against ecological benchmarks for sediment, soil, and water which
were taken from the TRIM Ecological Toxicity Database [33], For PAH, the lowest, and thus
most conservative, ecological benchmark for soil (developed by the Canadian Council of
Ministries of the Environment) [34} was approximately the same as the modeled soil
concentration. This indicates little to no potential for adverse growth, reproductive effects, and
mortality in the soil community, terrestrial plants, and earthworms. This result is associated with
using a default speciation profile for assessing unspeciated PAH, as described in the previous
paragraph, and is thought to be highly conservative. No further refinement of multipathway
ecological risks was undertaken.
We also screened for potential adverse environmental impacts via direct atmospheric contact by
comparing chronic atmospheric concentrations to RfC values at locations outside estimated
facility boundaries, noting, as we have in previous residual risk assessments, that chronic human
health inhalation thresholds are generally more stringent than direct contact environmental
protection thresholds developed to date. None of the HAP emitted by petroleum refinery MACT
1 sources showed any potential for adverse environmental impacts based on this screening. We
are aware that some concerns have been expressed regarding the adequacy of this screening
Page 2-29
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
method for the pollutant hydrogen fluoride, since studies have been identified in the scientific
literature showing adverse effects on some flora at levels below human health thresholds.
Indeed, there is a significant lack of scientific understanding and assessment methodologies for
such potential adverse environmental effects. Notwithstanding these concerns, we believe that
the negative outcome of our assessment based on the chronic noncancer human health endpoint
for hydrogen fluoride (the maximum HQ for this pollutant was 0.25) provides strong support for
our conclusion that adverse environmental impacts are not expected for hydrogen fluoride
emissions from this source category.
2.4 General Discussion of Uncertainties
Uncertainty and the potential for bias are inherent in all risk assessments, including the one
performed for the petroleum refineries source category presented in this document. The primary
uncertainties in this risk characterization focus around the site-specific emissions data set (as
discussed in the previous sections and in [30, 37]) and the uncertainties in dose-response
quantification. While other aspects of the assessment, including dispersion modeling, inhalation
exposure estimates, and multi-pathway exposure modeling all bring some degree of uncertainty
to the assessment, these uncertainties are secondary if emissions and site-specific characteristics
are not represented correctly.
2.4.1 Exposure Modeling Uncertainties
Although the development of the RTR database involved quality assurance/quality control
processes, the accuracy of emissions values will vary depending on the source of the data
present, incomplete or missing data, errors in estimating emissions values, and other factors. Our
review of the data indicates that there may be a low bias in reported emissions for many
facilities, but the extent of potential underreporting is not known. It appears that data from
several processes and operations are not included in the reported emissions from many facilities.
These include exclusion of upset, malfunction, startup, and shutdown events as well as omission
of emissions sources that are unexpected, not measured, or not considered in inventories, such as
leaks in heat exchanger systems; emissions from process sewers and wastewater systems;
fugitive emissions from delayed coking units; and emissions from tank roof landings. Further,
the emissions values considered in this analysis are annual totals for a single calendar year
(2002) and do not reflect actual fluctuations during the course of the year or variations from year
to year, including plant closure or expansion. Finally, although we have performed a significant
amount of quality control on the data set, for many facilities the physical characteristics (i.e.,
stack height, physical location) of the reported sources may be inaccurate for detailed risk
characterization purposes. The following general discussion of uncertainties applies to the
remaining aspects of the risk assessment, which are thought to contribute less to overall
uncertainties in the risk results, but are nonetheless included for completeness.
The chronic exposure modeling uncertainties are considered relatively small since we are using
EPA's refined local dispersion model with site-specific parameters and reasonably representative
meteorology. If anything, the population exposure estimates are biased high by not accounting
for short- or long-term population mobility, and by neglecting processes like deposition, plume
depletion, and atmospheric degradation. Additionally, estimates of the maximum individual risk
(MIR) contain uncertainty, because they are derived at census block centroid locations rather
than actual residences. This uncertainty is known to create potential underestimates and
Page 2-30
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
overestimates of the actual MIR values for individual facilities, but, overall, it is not thought to
have a significant impact on the estimated MIR for a source category. Finally, we did not factor
in the possibility of a source closure occurring during the 70-year chronic exposure period,
leading to a potential upward bias in both the MIR and population risk estimates; nor did we
factor in the possibility of population growth or production expansion during the 70-year chronic
exposure period, leading to a potential downward bias in both the MIR and population risk
estimates.
As previously discussed in section 2.2.2, a sensitivity analysis performed for the 1999 NATA
found that the selection of the meteorology dataset location could result in a range of chronic
ambient concentrations which varied from as much as 17% below the predicted value to as much
as 84% higher than the predicted value. This variability translates directly to the predicted
exposures and risks in our assessment, indicating that the actual risks could vary from 17% lower
to 84%) higher than the predicted values.
We have purposely biased the acute screening results high, considering that they depend upon
the joint occurrence of independent factors, such as hourly emissions rates, meteorology and
human activity patterns. Furthermore, in cases where multiple acute threshold values are
considered scientifically acceptable we have chosen the most conservative of these assessments,
erring on the side of overestimating potential health risks from acute exposures. In the cases
where these results indicated the potential for exceeding short-term health thresholds, we have
refined our assessment by developing a better understanding of the geography of the facility
relative to potential exposure locations. In each of these cases, we have determined that this
refined information reduced the likelihood of acute health concerns. We were not able to refine
these assessments to incorporate the true variability of short-term emission rates; such data are
not currently available. Thus, by maintaining the peak-to-mean emission ratio of 10 even in our
refined acute assessments, we believe the results generally overstate the potential for acute
impacts. We base this conclusion on the fact that our analysis of short-term event emission data
(Appendix B) indicates that the factor of 10 covers more than 99% of all actual peak emission
events for volatile and gaseous HAPs.
2.4.2 Uncertainties in the Dose-Response Relationships
In the sections that follow, separate discussions are provided on uncertainty associated with
cancer potency factors and for noncancer reference values. Cancer potency values are derived
for chronic (lifetime) exposures. Noncancer reference values are generally derived for chronic
exposures (up to a lifetime), but may also be derived for acute (<24 hours), short-term (>24
hours up to 30 days), and subchronic (>30 days up to 10% of lifetime) exposure durations, all of
which are derived based on an assumption of continuous exposure throughout the duration
specified. For the purposes of assessing all potential health risks associated with the emissions
included in this assessment, we rely on both chronic (cancer and noncancer) and acute
(noncancer) benchmarks, which are described in more detail below.
Although every effort is made to identify peer-reviewed dose-response values for all 75 HAPs
emitted by the sources included in this assessment, some HAP have no peer-reviewed cancer
potency values or reference values for chronic non-cancer or acute effects. Since exposures to
Page 2-31
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
these pollutants cannot be included in a quantitative risk estimate, an understatement of risk for
these pollutants at environmental exposure levels is possible.
Additionally, chronic dose-response values for 26 of the compounds included in this assessment
are currently under EPA IRIS review and revised assessments may determine that these
pollutants are more or less potent than currently thought. We will re-evaluate residual risks if, as
a result of these reviews, a dose-response metric changes enough to indicate that the risk
assessment supporting today's notice may significantly mischaracterize human health risk.
Cancer assessment
The discussion of dose-response uncertainties in the estimation of cancer risk below focuses on
the uncertainties associated with the specific approach currently used by the EPA to develop
cancer potency factors. In general, these same uncertainties attend the development of cancer
potency factors by CalEPA, the source of peer-reviewed cancer potency factors used where
EPA-developed values are not yet available. To place this discussion in context, we provide a
quote from the EPA's Guidelines for Carcinogen Risk Assessment. [35] "The primary goal of
EPA actions is protection of human health; accordingly, as an Agency policy, risk assessment
procedures, including default options that are used in the absence of scientific data to the
contrary, should be health protective." The approach adopted in this document is consistent with
this approach as described in the Cancer Guidelines.
For cancer endpoints EPA usually derives an oral slope factor for ingestion and a unit risk value
for inhalation exposures. These values allow estimation of a lifetime probability of developing
cancer given long-term exposures to the pollutant. Depending on the pollutant being evaluated,
EPA relies on both animal bioassay and epidemiological studies to characterize cancer risk. As a
science policy approach, consistent with the Cancer Guidelines, EPA uses animal cancer
bioassays as indicators of potential human health risk when other human cancer risk data are
unavailable.
Extrapolation of study data to estimate potential risks to human populations is based upon EPA's
assessment of the scientific database for a pollutant using EPA's guidance documents and other
peer-reviewed methodologies. The EPA Guidelines for Carcinogen Risk Assessment describes
the Agency's recommendations for methodologies for cancer risk assessment. EPA believes that
cancer risk estimates developed following the procedures described in the Cancer Guidelines and
outlined below generally provide an upper bound estimate of risk. That is, EPA's upper bound
estimates represent a "plausible upper limit to the true value of a quantity" (although this is
usually not a true statistical confidence limit).25 In some circumstances, the true risk could be as
low as zero; however, in other circumstances the risk could also be greater.26 When developing
an upper bound estimate of risk and to provide risk values that do not underestimate risk, EPA
generally relies on conservative default approaches.27 EPA also uses the upper bound (rather
25 IRIS glossary (www.epa.gov/NCEA/iris/help_gloss.htm).
26 The exception to this is the URE for benzene, which is considered to cover a range of values, each end of which is
considered to be equally plausible, and which is based on maximum likelihood estimates.
2 7
According to the NRC report Science and Judgment in Risk Assessment (NRC, 1994) "[Default] options are
generic approaches, based on general scientific knowledge and policy judgment, that are applied to various elements
Page 2-32
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
than lower bound or central) estimates in its assessments, although it is noted that this approach
can have limitations for some uses (e.g. priority setting, expected benefits analysis).
Such health risk assessments have associated uncertainties, some of which may be considered
quantitatively, and others which generally are expressed qualitatively. Uncertainties may vary
substantially among cancer risk assessments associated with exposures to different pollutants,
since the assessments employ different databases with different strengths and limitations and the
procedures employed may differ in how well they represent actual biological processes for the
assessed substance. EPA's Risk Characterization Handbook also recommends that risk
characterizations present estimates demonstrating the impact on the assessment of alternative
choices, data, models and assumptions (U.S. EPA, 2000). Some of the major sources of
uncertainty and variability in deriving cancer risk values are described more fully below.
(1) The qualitative similarities or differences between tumor responses observed in
experimental animal bioassays and those which would occur in humans are a source of
uncertainty in cancer risk assessment. In general, EPA does not assume that tumor sites
observed in an experimental animal bioassay are necessarily predictive of the sites at which
tumors would occur in humans.28 However, unless scientific support is available to show
otherwise, EPA assumes that tumors in animals are relevant in humans, regardless of target
organ concordance. For a specific pollutant, qualitative differences in species responses can lead
to either under-estimation or over-estimation of human cancer risks.
(2) Uncertainties regarding the most appropriate dose metric for an assessment can also
lead to differences in risk predictions. For example, the measure of dose is commonly expressed
in units of mg/kg/d ingested or the inhaled concentration of the pollutant. However, data may
support development of a pharmacokinetic model for the absorption, distribution, metabolism
and excretion of an agent, which may result in improved dose metrics (e.g., average blood
concentration of the pollutant or the quantity of agent metabolized in the body). Quantitative
uncertainties result when the appropriate choice of a dose metric is uncertain or when dose
metric estimates are themselves uncertain (e.g., as can occur when alternative pharmacokinetic
models are available for a compound). Uncertainty in dose estimates may lead to either over or
underestimation of risk.
(3) For the quantitative extrapolation of cancer risk estimates from experimental animals
to humans, EPA uses scaling methodologies (relating expected response to differences in
of the risk-assessment process when the correct scientific model is unknown or uncertain." The 1983 NRC report
Risk Assessment in the Federal Government: Managing the Process defined default option as "the option chosen on
the basis of risk assessment policy that appears to be the best choice in the absence of data to the contrary" (NRC,
1983a, p. 63). Therefore, default options are not rules that bind the agency; rather, the agency may depart from them
in evaluating the risks posed by a specific substance when it believes this to be appropriate. In keeping with EPA's
goal of protecting public health and the environment, default assumptions are used to ensure that risk to chemicals is
not underestimated (although defaults are not intended to overtly overestimate risk). See EPA 2004 An Examination
of EPA Risk Assessment Principles and Practices, EPA/100/B-04/001 available at:
http://www.epa.gov/osa/pdfs/ratf-final.pdf.
28 Per the EPA Cancer Guidelines: "The default option is that positive effects in animal cancer studies indicate that
the agent under study can have carcinogenic potential in humans." and "Target organ concordance is not a
prerequisite for evaluating the implications of animal study results for humans."
Page 2-33
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
physical size of the species), which introduce another source of uncertainty. These
methodologies are based on both biological data on differences in rates of process according to
species size and empirical comparisons of toxicity between experimental animals and humans.
For a particular pollutant, the quantitative difference in cancer potency between experimental
animals and humans may be either greater than or less than that estimated by baseline scientific
scaling predictions due to uncertainties associated with limitations in the test data and the
correctness of scaled estimates.
(4) EPA cancer risk estimates, whether based on epidemiological or experimental animal
data, are generally developed using a benchmark dose (BMD) analysis to estimate a dose at
which there is a specified excess risk of cancer (called a "point of departure," or POD).
Statistical uncertainty in developing a POD using a benchmark dose (BMD) approach is
generally addressed through use of the 95% lower confidence limit on the dose at which the
specified excess risk occurs (the BMDL), decreasing the likelihood of understating risk. EPA
has generally utilized the multistage model for estimation of the BMDL using cancer bioassay
data (see further discussion below).
(5) Extrapolation from high to low doses is an important, and potentially large, source of
uncertainty in cancer risk assessment. EPA uses different approaches to low dose risk
assessment {i.e., developing estimates of risk for exposures to environmental doses of an agent
from observations in experimental or epidemiological studies at higher dose) depending on the
available data and understanding of a pollutant's mode of action {i.e., the manner in which a
pollutant causes cancer). EPA's cancer guidelines express a preference for the use of reliable,
compound-specific, biologically-based risk models when feasible; however, such models are
rarely available. The mode of action for a pollutant {i.e., the manner in which a pollutant causes
cancer) is a key consideration in determining how risks should be estimated for low-dose
exposure. A reference value is calculated when the available mode of action data show the
response to be nonlinear {e.g., as in a threshold response). A linear low-dose (straight line from
POD) approach is used when available mode of action data support a linear {e.g., nonthreshold
response) or as the most common default approach when a compound's mode of action is
unknown. Linear extrapolation can be supported by both pollutant-specific data and broader
scientific considerations. For example, EPA's Cancer Guidelines generally consider a linear
dose-response to be appropriate for pollutants that interact with DNA and induce mutations.
Pollutants whose effects are additive to background biological processes in cancer development
can also be predicted to have low-dose linear responses, although the slope of this relationship
may not be the same as the slope estimated by the straight line approach.
EPA most frequently utilizes a linear low-dose extrapolation approach as a baseline science-
policy choice (a "default") when available data do not allow a compound-specific determination.
This approach is designed to not underestimate risk in the face of uncertainty and variability.
EPA believes that linear dose-response models, when appropriately applied as part of EPA's
cancer risk assessment process, provide an upper bound estimate of risk and generally provide a
health protective approach. Note that another source of uncertainty is the characterization of
low-dose nonlinear, non-threshold relationships. The National Academy of Sciences has
encouraged the exploration of sigmoidal type functions {e.g., log-probit models) in representing
dose response relationships due to the variability in response within human populations. A
Page 2-34
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
recent National Research Council report (NRC, 2006) [36] suggests that models based on
distributions of individual thresholds are likely to lead to sigmoidal-shaped dose-response
functions for a population. This report notes sources of variability in the human population:
"One might expect these individual tolerances to vary extensively in humans depending on
genetics, coincident exposures, nutritional status, and various other susceptibility factors..."
Thus, if a distribution of thresholds approach is considered for a carcinogen risk assessment,
application would depend on ability of modeling to reflect the degree of variability in response in
human populations (as opposed to responses in bioassays with genetically more uniform
rodents). Note also that low dose linearity in risk can arise for reasons separate from population
variability: due to the nature of a mode of action and additivity of a chemical's effect on top of
background chemical exposures and biological processes.
As noted above, EPA's current approach to cancer risk assessment typically utilizes a straight
line approach from the BMDL. This is equivalent to using an upper confidence limit on the
slope of the straight line extrapolation. The impact of the choice of the BMDL on bottom line
risk estimates can be quantified by comparing risk estimates using the BMDL value to central
estimate BMD values, although these differences are generally not a large contributor to
uncertainty in risk assessment [37], It is important to note that earlier EPA assessments,
including the majority of those for which risk values exist today, were generally developed using
the multistage model to extrapolate down to environmental dose levels and did not involve the
use of a POD. Comparisons indicating that slopes based on straight line extrapolation from a
POD do not show large differences from those based on the upper confidence limit of the
multistage model [37],
(6) Cancer risk estimates do not generally make specific adjustments to reflect the
variability in response within the human population — resulting in another source of uncertainty
in assessments. In the diverse human population, some individuals are likely to be more
sensitive to the action of a carcinogen than the typical individual, although compound-specific
data to evaluate this variability are generally not available. There may also be important life
stage differences in the quantitative potency of carcinogens and, with the exception of the
recommendations in EPA's Supplemental Cancer Guidance for carcinogens with a mutagenic
mode of action, risk assessments do not generally quantitatively address life stage differences.
However, one approach used commonly in EPA assessments that may help address variability in
response is to extrapolate human response from results observed in the most sensitive species
and sex tested, resulting typically in the highest URE which can be supported by reliable data,
thus supporting estimates that are designed not to underestimate risk in the face of uncertainty
and variability.
Chronic noncancer assessment
Chronic noncancer reference values represent chronic exposure levels that are intended to be
health-protective. That is, EPA and other organizations which develop noncancer reference
values (e.g., the Agency for Toxic Substances and Disease Registry - ATSDR) utilize an
approach that is intended not to underestimate risk in the face of uncertainty and variability.
When there are gaps in the available information, uncertainty factors (UFs) are applied to derive
reference values that are intended to be protective against appreciable risk of deleterious effects.
Page 2-35
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Uncertainty factors are commonly default values29, e.g., factors of 10 or 3 used in the absence of
compound-specific data. Where data are available, uncertainty factors may also be developed
using compound-specific information. When data are limited, more assumptions are needed and
more default factors are used. Thus there may be a greater tendency to overestimate risk—in the
sense that further study might support development of reference values that are higher (i.e., less
potent) because fewer default assumptions are needed. However, for some pollutants it is
possible that risks may be underestimated.
For non-cancer endpoints related to chronic exposures, EPA derives a Reference Dose (RfD) for
exposures via ingestion, and a Reference Concentration (RfC) for inhalation exposures. These
values provide an estimate (with uncertainty spanning perhaps an order of magnitude) of daily
oral exposure (RfD) or of a continuous inhalation exposure (RfC) to the human population
(including sensitive subgroups) that is likely to be without an appreciable risk of deleterious
effects during a lifetime.30 To derive values that are intended to be "without appreciable risk,"
EPA's methodology relies upon an uncertainty factor (UF) approach (U.S. EPA, 1993, 1994)
which includes consideration of both uncertainty and variability.
EPA begins by evaluating all of the available peer-reviewed literature to determine non-cancer
endpoints of concern, evaluating the quality, strengths and limitations of the available studies.
EPA typically chooses the relevant endpoint that occurs at the lowest dose, often using statistical
modeling of the available data, and then determines the appropriate POD for derivation of the
reference value. A POD is determined by (in order of preference): (1) a statistical estimation
using the benchmark dose (BMD) approach; (2) use of the dose or concentration at which the
toxic response was not significantly elevated (no observed adverse effect level— NOAEL); or
(3) use of the lowest observed adverse effect level (LOAEL).
A series of downward adjustments using default UFs is then applied to the POD to estimate the
reference value (U.S. EPA 1994, 2002). While collectively termed "UFs", these factors account
for a number of different quantitative considerations when utilizing observed animal (usually
rodent) or human toxicity data in a risk assessment. The UFs are intended to account for: (1)
variation in susceptibility among the members of the human population (i.e., inter-individual
29 According to the NRC report Science and Judgment in Risk Assessment (NRC, 1994)
"[Default] options are generic approaches, based on general scientific knowledge and policy
judgment, that are applied to various elements of the risk-assessment process when the correct
scientific model is unknown or uncertain." The 1983 NRC report Risk Assessment in the Federal
Government: Managing the Process defined default option as "the option chosen on the basis of
risk assessment policy that appears to be the best choice in the absence of data to the contrary"
(NRC, 1983a, p. 63). Therefore, default options are not rules that bind the agency; rather, the
agency may depart from them in evaluating the risks posed by a specific substance when it
believes this to be appropriate. In keeping with EPA's goal of protecting public health and the
environment, default assumptions are used to ensure that risk to chemicals is not underestimated
(although defaults are not intended to overtly overestimate risk). See EPA 2004 An examination
of EPA Risk Assessment Principles and Practices, EPA/100/B-04/001 available at:
http://www.epa.gov/osa/pdfs/ratf-final.pdf
30 See IRIS glossary
Page 2-36
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
variability); (2) uncertainty in extrapolating from experimental animal data to humans {i.e.,
interspecies differences); (3) uncertainty in extrapolating from data obtained in a study with less-
than-lifetime exposure {i.e., extrapolating from subchronic to chronic exposure); (4) uncertainty
in extrapolating from a LOAEL in the absence of a NOAEL; and (5) uncertainty when the
database is incomplete or there are problems with applicability of available studies. When
scientifically sound, peer-reviewed assessment-specific data are not available, default adjustment
values are selected for the individual UFs. For each type of uncertainty (when relevant to the
assessment), EPA typically applies an UF value of 10 or 3 with the cumulative UF value leading
to a downward adjustment of 10-3000 fold from the selected POD. An UF of 3 is used when the
data do not support the use of a 10-fold factor. If an extrapolation step or adjustment is not
relevant to an assessment {e.g., if applying human toxicity data and an interspecies extrapolation
is not required) the associated UF is not used. The major adjustment steps are described more
fully below.
1) Heterogeneity among humans is a key source of variability as well as uncertainty.
Uncertainty related to human variation is considered in extrapolating doses from a subset or
smaller-sized population, often of one sex or of a narrow range of life stages (typical of
occupational epidemiologic studies), to a larger, more diverse population. In the absence of
pollutant-specific data on human variation, a 10-fold UF is used to account for uncertainty
associated with human variation. Human variation may be larger or smaller; however, data to
examine the potential magnitude of human variability are often unavailable. In some situations,
a smaller UF of 3 may be applied to reflect a known lack of significant variability among
humans.
2) Extrapolation from results of studies in experimental animals to humans is a necessary
step for the majority of chemical risk assessments. When interpreting animal data, the
concentration at the POD (e.g. NOAEL, BMDL) in an animal model (e.g. rodents) is
extrapolated to estimate the human response. While there is long-standing scientific support for
the use of animal studies as indicators of potential toxicity to humans, there are uncertainties in
such extrapolations. In the absence of data to the contrary, the typical approach is to use the
most relevant endpoint from the most sensitive species and the most sensitive sex in assessing
risks to the average human. Typically, compound specific data to evaluate relative sensitivity in
humans versus rodents are lacking, thus leading to uncertainty in this extrapolation. Size-related
differences (allometric relationships) indicate that typically humans are more sensitive than
rodents when compared on a mg/kg/day basis. The default choice of 10 for the interspecies UF
is consistent with these differences. For a specific chemical, differences in species responses
may be greater or less than this value.
Pharmacokinetic models are useful to examine species differences in pharmacokinetic
processing and associated uncertainties; however, such dosimetric adjustments are not always
possible. Information may not be available to quantitatively assess toxicokinetic or
toxicodynamic differences between animals and humans, and in many cases a 10-fold UF (with
separate factors of 3 for toxicokinetic and toxicodynamic components) is used to account for
expected species differences and associated uncertainty in extrapolating from laboratory animals
to humans in the derivation of a reference value. If information on one or the other of these
Page 2-37
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
components is available and accounted for in the cross-species extrapolation, a UF of 3 may be
used for the remaining component.
3) In the case of reference values for chronic exposures where only data from shorter
durations are available (e.g., 90-day subchronic studies in rodents) or when such data are judged
more appropriate for development of an RfC, an additional UF of 3 or 10-fold is typically
applied unless the available scientific information supports use of a different value.
4) Toxicity data are typically limited as to the dose or exposure levels that have been
tested in individual studies; in an animal study, for example, treatment groups may differ in
exposure by up to an order of magnitude. The preferred approach to arrive at a POD is to use
BMD analysis; however, this approach requires adequate quantitative results for a meaningful
analysis, which is not always possible. Use of a NOAEL is the next preferred approach after
BMD analysis in determining a POD for deriving a health effect reference value. However,
many studies lack a dose or exposure level at which an adverse effect is not observed (i.e., a
NOAEL is not identified). When using data limited to a LOAEL, a UF of 10 or 3-fold is often
applied.
5) The database UF is intended to account for the potential for deriving an
underprotective RfD/RfC due to a data gap preventing complete characterization of the
chemical's toxicity. In the absence of studies for a known or suspected endpoint of concern, a
UF of 10 or 3-fold is typically applied.
Acute noncancer assessment
Many of the UFs used to account for variability and uncertainty in the development of acute
reference values are quite similar to those developed for chronic durations, but more often using
individual UF values that may be less than 10. UFs are applied based on chemical-specific or
health effect-specific information (e.g., simple irritation effects do not vary appreciably between
human individuals, hence a value of 3 is typically used), or based on the purpose for the
reference value (see the following paragraph). The UFs applied in acute reference value
derivation include: 1) heterogeneity among humans; 2) uncertainty in extrapolating from
animals to humans; 3) uncertainty in LOAEL to NOAEL adjustments; and 4) uncertainty in
accounting for an incomplete database on toxic effects of potential concern. Additional
adjustments are often applied to account for uncertainty in extrapolation from observations at
one exposure duration (e.g., 4 hours) to arrive at a POD for derivation of an acute reference value
at another exposure duration (e.g., 1 hour).
Not all acute reference values are developed for the same purpose and care must be taken when
interpreting the results of an acute assessment of human health effects relative to the reference
value or values being exceeded. Where relevant to the estimated exposures, the lack of threshold
values at different levels of severity should be factored into the risk characterization as potential
uncertainties.
Page 2-38
-------
3 Portland cement case study
3.1 Introduction
This section provides documentation of our case study for the Portland Cement Manufacturing
source category. Section 3.2 provides a description of the Portland Cement Manufacturing
source category and a brief summary of the emissions data in the case study data set. Section 3.3
provides results of the inhalation risk assessment, including an assessment of the potential
inhalation risks associated with radionuclide emissions from Portland cement facilities. Section
3.4 provides the details of the multipathway exposure and risk assessment results for Portland
cement facilities, including a description of the methodologies used to refine this portion of the
assessment. Section 3.5 presents the methodologies used to assess potential ecological risks
associated with emissions from Portland cement facilities, and then provides a summary of
results and interpretation.
3.2 Source category and emissions data
The Portland Cement Manufacturing source category includes facilities that produce Portland
cement. Portland cement is a fine powder, usually gray in color, that consists of a mixture of the
minerals dicalcium silicate, tricalcium silicate, tricalcium aluminate, and tetracalcium
aluminoferrite, to which one or more forms of calcium sulfate have been added. The primary end
use of Portland cement is as the key ingredient in Portland cement concrete, which is used in
almost all construction applications.
The process of manufacturing Portland cement consists of four primary units of operation: (1)
kiln feed preparation {i.e., crushing and grinding the carefully proportioned raw materials to a
high degree of fineness); (2) firing the raw mix in a rotary kiln to produce clinker (an
intermediate product, before grinding), including fuel handling; (3) grinding the resulting clinker
to a fine powder and mixing with gypsum to produce cement; and (4) raw and finished materials
handling. As a whole, the manufacturing process is expected to result in the emission of the
following HAP: acetaldehyde, arsenic, benzene, cadmium, chromium, chlorobenzene, dioxins,
formaldehyde, hexane, hydrogen chloride, lead, manganese, mercury, naphthalene, nickel,
phenol, polycyclic organic matter, selenium, styrene, toluene, and xylene. These HAP are
associated with the emissions of specific production processes, including grinding and conveying
operation dusts, exhaust gases from the raw material dryer; kiln exhaust gases; clinker cooler
exhaust gases; and dusts from the finish grinding of clinker into cement. Emissions from the
grinding and conveying operations are essentially particulate emissions (e.g., dust from
limestone, clay, and bauxite ore) that contain HAP metals. Raw material dryers are used as part
of the feed preparation process (i.e., drying, blending, and storage), and can produce emissions in
two different ways. If the raw material dryer uses heat from a separate combustion source (fuel-
fired raw material dryer), exhaust gases can contain trace quantities of products of incomplete
combustion (PICs), HC1, and metals from the fuel. When feed materials contain organic matter,
this material may volatilize in the raw material dryer (regardless of the source of the heat) adding
organic HAPs to the dryer exhaust. Kiln exhaust emissions contain a wide variety of HAPs and
other air pollutants that originate from the fuel combustion and from the feed material. These
HAPs include gaseous organic HAPs, some of which are chlorinated, along with mercury
(emitted as either a particulate or a gas), hydrogen chloride, dioxins, and the following metal
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
HAP emissions: chromium, lead, arsenic, mercury, antimony, and manganese. Because clinker
coolers are not combustion devices, the only expected HAPs are metals associated with the
clinker cooler particulate, i.e., clinker dust. HAP metals that have been detected in clinker
include chromium, lead, nickel, arsenic, beryllium, antimony, selenium, and mercury. The finish
grinding of clinker into Portland cement produces dusts that can contain HAP metals associated
with clinker, which are listed above.
From information gathered during the MACT development and from more recent contacts with
the industry, EPA estimates that there are 104 facilities with processes belonging in the Portland
cement manufacturing source category. EPA identified each of these facilities in Version 1.0 of
the 2002 NEI (February 2006) and created a data set comprised of the HAP emissions and
emissions release parameters for Portland cement production portions of these facilities. EPA
reviewed the data set and identified processes, facilities, and chemicals that, based on SCC and
other process identifications in the NEI, were erroneously included in the Portland cement source
category, and made revisions to exclude these processes, facilities, or chemicals from the data set
for this source category. There are several factors that make the emissions dataset for the
Portland cement case study more provisional than that for the petroleum refineries case study.
First, the data are still under development, and have already been revised since this case study
was developed. Second, the data have not yet undergone public review, but will do so prior to
any regulatory action. And third, the technology-based standards that the dataset reflects may
yet be amended, with consequent reductions in emissions and risk. It is important to keep in
mind that this case study will change substantially during the RTR rule development process,
and that it is presented here only to illustrate a methodology.
Table 3-1 summarizes the emissions for the Portland Cement Manufacturing source category
data set. Based on these data, the HAP emitted in the largest quantity is hydrochloric acid, which
accounts for approximately 73 percent of the HAP mass emitted. Hydrochloric acid, along with
benzene, formaldehyde, toluene, chlorine, 1,3-butadiene, naphthalene, xylenes (mixture of o, m,
and p isomers), carbonyl sulfide, manganese, styrene, ethyl benzene, phenol, lead, manganese,
ethylene glycol, chromium, methylene chloride, carbon disulfide, acetaldehyde, chromium,
methyl chloride, lead & compounds, and hexane account for approximately 99 percent of the
HAP mass emitted across the 104 facilities. Hydrochloric acid is the HAP reported most
frequently across the source category, with reported emissions from 79 of the 104 facilities in the
data set. The dataset includes emissions of substances representing 10 of the 14 PB-HAP31
categories (mercury, lead, cadmium, POM, PCBs, hexachlorobenzene, trifluralin, methoxychlor,
heptachlor, and chlordane).
31 Persistent and bioaccumulative HAP are defined in the EPA's Air Toxics Risk Assessment Library [6].
Page 3-2
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category
HAP32
Emissions
(tpy)
Number of Facilities
Reporting HAP
(104 facilities in
ANPRM data set)
Prioritized Inhalation Dose-Response Value
Identified by OAQPS33
PB-
HAP?
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Chronic
Noncancer?
Health
Benchmark
Values for Acute
Noncancer?
Hydrochloric Acid
3,162
79
~
~
Benzene
330
56
~
~
~
Formaldehyde
155
49
~
~
~
Toluene
81
38
~
~
Chlorine
77
14
~
~
1,3-Butadiene
64
7
~
~
~
Naphthalene
55
46
~
~
Xylenes (Mixture of o, m,
and p Isomers)
50
35
~
~
Carbonyl Sulfide
48
1
Manganese
45
42
~
Styrene
30
28
~
~
Ethyl Benzene
22
33
~
Phenol
20
32
~
~
Lead
16
65
~
~
Manganese & Compounds
14
13
~
Ethylene Glycol
13
9
~
Chromium
11
36
~
~
Methylene Chloride
11
30
~
~
~
Carbon Disulfide
10
23
~
~
32
Notes for how HAP were speciated for risk assessment:
• For most metals, emissions reported as the elemental metal are combined with metal compound emissions (e.g., "cadmium"
emissions modeled as "cadmium & compounds").
• For emissions reported genetically as "chromium" or "chromium & compounds," emissions are speciated for this category
as 92 percent "chromium (III) compounds" and 8 percent "chromium (VI) compounds." Chromium speciation profiles can
be found on the EPA's Technology Transfer Network website for emissions inventories at
http://www.epa.gov/ttn/chief/net/2002inventorv.html.
• For emissions reported generically as "mercury" or "mercury & compounds," emissions are speciated for this category as 75
percent "mercury (elemental)" and 25 percent "mercuric chloride." Mercury speciation profiles can be found on the EPA's
Technology Transfer Network website for emissions inventories at http: //www, epa. gov/ttn/chief/net/2002inventory.
• For emissions of any chemicals or chemical groups classified as poly cyclic organic matter (POM), emissions will be
grouped into POM subgroups as found on EPA's Technology Transfer Network website for the 1999 National-Scale Air
Toxics Assessment at http://www.epa.gov/ttn/atw/natal999/nsata99.html.
• For emissions reported generically as "Glycol Ethers" or specific glycol ethers not found on EPA's Technology Transfer
network for air toxics (see footnote b), emissions will be treated as ethylene glycol methyl ether.
33
Specific dose-response values for each chemical are identified on EPA's Technology Transfer Network website for air toxics
at http://www.epa.gov/ttn/atw/toxsource/summarv.html.
Page 3-3
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category
Prioritized Inhalation Dose-Response Value
Number of Facilities
Identified by OAQPS33
Emissions
Reporting HAP
(104 facilities in
ANPRM data set)
PB-
HAP
(tpy)
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Chronic
Health
Benchmark
Values for Acute
HAP?
Noncancer?
Noncancer?
Acetaldehyde
9
10
~
~
~
Chromium & Compounds
9
27
~
~
Methyl Chloride
7
20
~
~
Lead & Compounds
6
28
~
~
Hexane
5
12
~
Mercury
5
53
~
~
~
1,3 -Propanesultone
4
1
~
Methanol
4
9
~
~
Phenanthrene
4
19
~
~
Acrolein
3
3
~
~
Dibenzofuran
3
8
~
1 -Chloro-2,3 -Epoxypropane
2
1
~
~
~
Acetophenone
2
4
Acrylonitrile
2
3
~
~
~
Bromoform
2
4
~
Hydrogen Fluoride
2
5
~
~
Mercury & Compounds
2
35
~
~
Methyl Bromide
2
18
~
~
Nickel & Compounds
2
12
~
~
1,4-Dichlorobenzene
1
6
~
~
Acenaphthylene
1
14
~
~
Beryllium
1
25
~
~
Beryllium & Compounds
1
13
~
~
~
Biphenyl
1
20
Chlorobenzene
1
22
~
Dibutyl Phthalate
1
19
Diethanolamine
1
1
~
Fluorene
1
16
~
~
Methyl Chloroform
1
8
~
~
Methyl Isobutyl Ketone
1
11
~
Nickel
1
30
~
~
Polycyclic Organic Matter
1
2
~
~
Selenium
1
26
~
Selenium & Compounds
1
11
~
T etrachloroethy lene
1
11
~
~
~
Vinyl Chloride
1
7
~
~
~
B is (2 -Ethy lhexy l)Phthalate
0.5
17
~
~
Cadmium
0.4
27
~
~
~
Cumene
0.4
6
~
Page 3-4
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category
Prioritized Inhalation Dose-Response Value
Number of Facilities
Identified by OAQPS33
Emissions
Reporting HAP
(104 facilities in
ANPRM data set)
PB-
HAP
(tpy)
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Chronic
Health
Benchmark
Values for Acute
HAP?
Noncancer?
Noncancer?
Glycol Ethers
0.4
4
~
N,N-Dimethyl formamide
0.4
5
~
~
Acetonitrile
0.3
7
~
~
Arsenic
0.3
25
~
~
Methyl Tert-Butyl Ether
0.3
6
~
~
m-Xylene
0.3
2
Phosphorus
0.3
5
~
Vinyl Acetate
0.3
4
~
~
Lead Compounds (Inorganic)
0.2
6
~
~
PAH, total
0.2
11
~
~
Allyl Chloride
0.1
2
~
~
~
Arsenic & Compounds
(Inorganic Including Arsine)
0.1
12
~
~
~
Benzyl Chloride
0.1
2
~
~
Cadmium & Compounds
0.1
14
~
~
~
Carbon Tetrachloride
0.1
4
~
~
~
Chromium (VI)
0.1
10
~
~
Cobalt
0.1
6
~
Dichloroethyl Ether
0.1
1
~
Ethylene Dibromide
0.1
4
~
~
Ethylene Dichloride
0.1
9
~
~
~
Ethylene Glycol Methyl
Ether
0.1
1
~
Fluoranthene
0.1
14
~
~
Methyl Iodide
0.1
1
~
o-Xylene
0.1
5
T richloroethylene
0.1
7
~
~
~
Cresol
0.05
3
~
Ethyl Chloride
0.04
3
~
Hexachlorobutadiene
0.04
3
~
~
~
Pyrene
0.04
15
~
~
1,3 -Dichloropropene
0.03
3
~
~
2,4-Dinitrophenol
0.03
2
Antimony
0.03
5
~
Asbestos
0.03
1
~
Chloroform
0.03
7
~
~
Vinylidene Chloride
0.03
3
~
1,1,2-T richloroethane
0.02
2
~
~
1,2-Epoxybutane
0.02
1
~
Page 3-5
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category
Prioritized Inhalation Dose-Response Value
Number of Facilities
Identified by OAQPS33
Emissions
Reporting HAP
(104 facilities in
ANPRM data set)
PB-
HAP
(tpy)
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Chronic
Health
Benchmark
Values for Acute
HAP?
Noncancer?
Noncancer?
2,4-Dinitrotoluene
0.02
2
~
~
Antimony & Compounds
0.02
2
~
Cellosolve Solvent
0.02
1
~
Chromium III
0.02
7
Methyl Methacrylate
0.02
6
~
~
p-Cresol
0.02
3
Pentachlorophenol
0.02
3
~
~
Vinyl Bromide
0.02
1
~
~
1,1,2,2-Tetrachloroethane
0.01
3
~
2,4,5-Trichlorophenol
0.01
2
2,4,6-Trichlorophenol
0.01
2
~
3,3 '-Dichlorobenzidene
0.01
2
~
Acrylamide
0.01
1
~
~
Benzo[a]Pyrene
0.01
18
~
~
Chrysene
0.01
18
~
~
Dibenzo [a,h] Anthracene
0.01
18
~
~
Dimethyl Phthalate
0.01
3
Ethylidene Dichloride (1,1-
Dichloroethane)
0.01
3
~
~
Hexachlorobenzene
0.01
3
~
~
~
Hexachlorocyclopentadiene
0.01
2
~
Hexachloroethane
0.01
2
~
~
Nitrobenzene
0.01
2
~
o-Cresol
0.01
1
Propylene Dichloride
0.01
2
~
~
Acenaphthene
0.005
2
~
~
B enz [a] Anthracene
0.005
18
~
~
Benzo [b]Fluoranthene
0.004
18
~
~
Ethyl Acrylate
0.004
1
~
4,6-Dinitro-o-Cresol
0.003
2
4-Nitrophenol
0.003
1
N-Nitrosodimethylamine
0.003
1
~
Pentachloronitrobenzene
0.003
1
~
Polychlorinated Biphenyls
0.003
5
~
~
4,4'-Methylenebis(2-
Chloraniline)
0.002
1
~
Trifluralin
0.002
1
~
~
1,2,4-Trichlorobenzene
0.001
1
~
l,2-Dibromo-3-
0.001
1
~
~
Page 3-6
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-1. Summary of Emissions from the Portland Cement Manufacturing Source Category
Prioritized Inhalation Dose-Response Value
Number of Facilities
Identified by OAQPS33
Emissions
Reporting HAP
(104 facilities in
ANPRM data set)
PB-
HAP
(tpy)
Unit Risk
Estimate for
Cancer?
Reference
Concentration
for Chronic
Health
Benchmark
Values for Acute
HAP?
Noncancer?
Noncancer?
Chloropropane
3,3 '-Dimethoxybenzidine
0.001
1
~
3,3 '-Dimethylbenzidine
0.001
1
~
4,4'-Methylenedianiline
0.001
1
~
~
4-Dimethylaminoazobenzene
0.001
1
~
4-Nitrobiphenyl
0.001
1
Aniline
0.001
1
~
~
~
Benzidine
0.001
1
~
~
Benzo [k]Fluoranthene
0.001
18
~
~
Hydroquinone
0.001
1
Indeno [ 1,2,3-c,d]Pyrene
0.001
18
~
~
Isophorone
0.001
~
~
N-Nitrosomorpholine
0.001
1
~
o-Anisidine
0.001
1
o-Toluidine
0.001
1
~
2-Chloroacetophenone
0.0005
1
~
4-Aminobiphenyl
0.0005
1
N,N -Dimethy laniline
0.0004
Benzo [g,h,i,]Perylene
0.0003
15
~
~
Anthracene
0.0001
~
~
m-Cresol
0.0001
1
Methoxychlor
0.00004
1
~
p-Dioxane
0.00004
1
~
~
~
Triethylamine
0.00003
1
~
Heptachlor
0.00002
1
~
~
2-Methylnaphthalene
0.00001
1
~
~
Phthalic Anhydride
0.00001
1
~
Chlordane
0.000004
1
~
~
~
3 -Methy lcholanthrene
0.0000004
1
~
~
B|j]Fluoranthene
0.0000002
1
~
~
3.2.1 Dioxin emissions
In addition to the HAPs in Table 3-1 above, this assessment also considered emissions of
chlorinated dibenzo-p-dioxins and -furans (CDD/Fs, or "dioxins"). In its dioxin inventory for
2000 [35], EPA derived a single emission factor of 0.27 ng/ kg34 clinker (expressed in terms of
34 TEQs are calculated values that allow us to combine different combinations of dioxins and dioxin-like compounds
into a single value representing the equivalent amount of a single compound, 2,3,7,8-tetrachlorodibenzo-p-dioxin.
Page 3-7
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
toxic equivalents of 2,3,7,8-tetrachl orodibenzo-/>dioxin, or 2378-TCDD(Teq)) for all non-
hazardous waste combustion units for this source category, based upon stack tests from 13 sites.
This factor was developed for all kilns regardless of type or operational parameters. For this
assessment we statistically analyzed available dioxin emission data and developed separate
emission factors for four different types of non-hazardous waste combustor kilns.
We obtained dioxin emission estimates and emission factors for 60 non-hazardous waste
combustion cement plants from 2002-2006, and calculated a mean emission factor for each of
the four facility types. An analysis of variance showed that the emission factors differed
significantly among process type (Table 3-2).
Table 3-2. Mean and 95% upper confidence limit (UCL) 2378-TCDD(TEq) emission factors for
Portland cement facilities, by kiln type
Mean emission factor
95% UCL emission factor
Kiln type
(ng/kg clinker capacity)
(ng/kg clinker capacity)
Dry
0.110
0.229
Dry with preheater and precalciner
0.170
0.614
Dry with preheater
0.168
0.377
Wet
0.768
1.877
These emission factor estimates plausibly bracket the 0.27 ng/kg estimate. Given this
plausibility, we characterized CDD/F emissions by kiln type for the Portland cement risk
assessment, and calculated plant-specific risks separately using the mean and upper confidence
limit (UCL) emissions factors. The complete analysis of dioxin emission data is described in
Appendix F.
3.2.2 Radionuclide emissions
This assessment also evaluated risks associated with radionuclides, which are regulated as HAPs
when emitted to the air. Emissions of radionuclides from industrial facilities are reported in the
2002 National Emissions Inventory (NEI) in mass-based units of US short tons per year.
However, the known hazards from radionuclides are most closely associated with the type of and
amount of radioactivity that each radioisotope releases rather than with its mass. Therefore the
practice of reporting unspeciated emissions of radioactive substances from a single facility
collectively in terms of mass, rather than individually by radioisotope in terms of radioactivity,
prevents the accurate estimation of risks posed by radionuclides emitted from industrial facilities.
As a test of possible strategies to evaluate radionuclide hazards, we identified two Portland
cement facilities in California that reported emissions in the 2002 NEI. On a mass basis,
emissions reported for these facilities are very small but still potentially important because of the
high carcinogenic potency of some radionuclides. The NEI entries did not specify which
radionuclides were emitted and how much of each was emitted, nor is it clear that the facilities
reported radionuclide emissions in a uniform manner.
We performed a more refined analysis (fully described in Appendix G) of radionuclide emissions
and risks intended to (1) improve consistency and accuracy of these emissions estimates, (2)
evaluate the utility of the NEI data for these HAPs, (3) consistently characterize actual
Page 3-8
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
emissions, and (4) attempt to quantify potential incremental inhalation cancer risks associate
with radionuclides.
We estimated radionuclide emissions for the two Portland cement sources using the NEI-
reported emissions and scaling factors developed from a "typical" Portland cement facility. We
derived the "typical" emission factors using the European Commission Radiation Protection 135
report [39], hereafter referred to as the "naturally occurring radioactive material (NORM)
report." This approach resulted in three different emission estimates for each of the two
facilities, one based on the NEI data and two based on the NORM report (one based on clinker
production and the other based on PM emissions). Estimated emissions were modeled with
HEM3 to estimate ambient concentrations, population exposures, and risks.
3.3 Risk assessment results - inhalation
This section summarizes the results of the inhalation risk assessment for the Portland Cement
Manufacturing source category. The basic risk estimates presented are the maximum individual
lifetime cancer risk, the maximum hazard index, and the cancer incidence. Also presented are
the HAP "drivers," which are the HAP that collectively contribute 90 percent of the maximum
cancer risk or maximum hazard at the highest receptor. Detailed facility-level results for both
chronic and acute inhalation risk assessments can be found in Appendix H.
Table 3-3 and Table 3-4 summarize the inhalation risk results for this source category. Acute
screening hazard quotients (HQs) were calculated for every HAP shown in Table 3-1 that has an
acute benchmark. The highest acute HQ value (and its associated HAP) is shown in Table 3-3.
Table 3-4 provides more information on the acute risk screening estimates for HAP that had an
acute HQ of greater than 1 for any benchmark. Detailed results for each facility appear in
Appendix H.
Table 3-3. Summary of Source Category Level Risks for Portland Cement Manufacturing
Result
HAP "Drivers"
Facilities in Source Category
Number of Facilities Estimated to be Subject to
MACT in Source Category in 1998, from the
Proposal Preamble (63 FR 14181, March 24,
1998)
118
n/a
Number of Facilities Identified in NEI and
Modeled in Screening Risk Assessment
104
n/a
Cancer Risks
Maximum Individual Lifetime Cancer Risk (in
1 million) from any Facility in the Category
800
chromium (VI) compounds, arsenic
compounds, cadmium compounds,
beryllium compounds
Number of Facilities with Maximum Individual Lifetime Cancer Risk:
Greater than or equal to 100 in 1 million
2
chromium (VI) compounds, arsenic
compounds, cadmium compounds,
beryllium compounds
Greater than or equal to 10 in 1 million
8
chromium (VI) compounds, cadmium
compounds, arsenic compounds, nickel
compounds, POM71002, benzene,
naphthalene, acrylamide, POM72002,
Page 3-9
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-3. Summary of Source Category Level Risks for Portland Cement Manufacturing
Result
HAP "Drivers"
beryllium compounds
Greater than or equal to 1 in 1 million
29
nickel compounds, chromium (VI)
compounds, beryllium compounds,
naphthalene, benzene, 1,3-butadiene,
1,3-propane sultone, arsenic
compounds, cadmium compounds,
POM71002, acrylamide, POM72002
Chronic Noncancer Risks
Maximum Neurological Hazard Index
10
manganese compounds
Maximum Respiratory Hazard Index
6
chlorine, hydrochloric acid
Maximum Kidney Hazard Index
3
cadmium compounds
Number of Facilities with Maximum Neurological Hazard Index:
Greater than 1
2
manganese compounds
Number of Facilities with Maximum Respiratory Hazard Index:
Greater than 1
3
chlorine, hydrochloric acid, beryllium
compounds, nickel compounds,
chromium (VI) compounds,
formaldehyde
Number of Facilities with Maximum Kidney Hazard Index:
Greater than 1
1
cadmium compounds
Acute Noncancer Screening Results
Maximum Acute Hazard Quotient
50
AEGL-1, hydrochloric acid
Number of Facilities With Potential for Acute
Effects
8
chlorine, formaldehyde, hydrochloric
acid
Population Exposure
Number of People Living Within 50 Kilometers
of Facilities Modeled
54,000,000
n/a
Number of People Exposed to Cancer Risk:
Greater than or equal to 100 in 1
million
400
n/a
Greater than or equal to 10 in 1 million
15,000
n/a
Greater than or equal to 1 in 1 million
470,000
n/a
Number of People Exposed to Noncancer Neurological Hazard Index:
Greater than 1
3,000
n/a
Number of People Exposed to Noncancer Respiratory Hazard Index:
Greater than 1
200
n/a
Number of People Exposed to Noncancer Kidney Hazard Index:
Greater than 1
170
n/a
Estimated Cancer Incidence (excess cancer
cases per year)
0.05
n/a
Contribution of HAP to Cancer Incidence
chromium (VI) compounds
61%
n/a
arsenic compounds
10%
n/a
cadmium compounds
9%
n/a
beryllium compounds
8%
n/a
benzene
5%
n/a
Page 3-10
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-4. Summary of Acute Screening Results for Portland Cement Manufacturing
MAXIMUM ACUTE HAZARD
QUOTIENTS
ACUTE DOSE-RESPONSE VALUES
Screening Results
Max. 1-hr Cone. / Min. Acute Dose-
Response Value
Mild Effects
Serious Effects
HAP
Max. 1-hr. Air
Cone, (mg/m3)
Mild Effects
Serious Effects
AEGL-1 (1-hr)
(mg/m3)
ERPG-1
(mg/m3)
AEGL-2 (1-hr)
(mg/m3)
ERPG-2
(mg/m3)
Hydrochloric acid
138
50
5
2.7
4.5
33
30
Chlorine
10
7
2
1.5
2.9
5.8
8.7
Formaldehyde
4
3
0.3
1.1
1.2
17
12
Notes on Screening Process:
1) Screening process is based on a hypothetical worst-case combination of emission rates, meteorology, and exposure location and therefore likely represents an
overestimate of actual health risk. The results are being provided only as a tool to aid in the fact-checking of the underlying emissions data and should not be interpreted
as actual health risks. A more refined analysis is needed to determine actual risks.
2) The screening was performed for all emitted HAP with available acute dose-response values. Only those pollutants whose screening HQs greater than 1 for at least one
acute threshold value are shown in the table.
3) HAP with available acute dose-response values which are not in the table do not carry any potential for posing acute health risks, based on an analysis of currently
available emissions data.
4) The acute screening risk assessment results will not be used for decision making.
Notes on Acute Dose-Response Values:
AEGL - Acute exposure guideline levels represent emergency exposure (1-hour) limits for the general public.
AEGL-1 is the exposure level above which it is predicted that the general population, including susceptible individuals, could experience effects that are notable discomfort,
but which are transient and reversible upon cessation of exposure.
AEGL-2 is the exposure level above which it is predicted that the general population, including susceptible individuals, could experience irreversible or other serious, long-
lasting adverse health effects or an impaired ability to escape.
ERPG - US DOE Emergency Removal Program guidelines represent emergency exposure (1-hour) limits for the general public.
ERPG-1 is the maximum level below which it is believed that nearly all individuals could be exposed for up to 1 hour without experiencing other than mild, transient adverse
health effects.
ERPG-2 is the maximum exposure below which it is believed that nearly all individuals could be exposed for up to 1 hour without experiencing or developing irreversible or
other serious health effects or symptoms which could impair an individual's ability to take protective action.
Page 3-11
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
3.4 Refined multipathway health risk assessment
3.4.1 Selection of HAPs for this analysis
As noted, facilities in the Portland cement manufacturing source category emit a variety of PB-
HAPs, including metals (lead, cadmium, and mercury) and organic compounds (polycyclic
organic matter and dioxins). For each facility in this source category, total emissions for each
PB-HAP were compared to de minimis levels to initially screen for the potential for non-
inhalation exposures and risks. The derivation of these de minimis emission rates is described in
Appendix C.
Emissions of every PB-HAP on EPA's list are not reported for every facility in this source
category. However, based on data from individual facilities and knowledge of the Portland
cement manufacturing process, every facility is assumed to emit dioxins, with nearly all of these
emission rates exceeding the de minimis level established for dioxin using the TRIM-based
screening scenario. More than half of the facilities also report mercury emissions, and we
consider it likely that all such facilities emit mercury. Although only one of these emission rates
exceeds the mercury de minimis level, mercury is a relatively common PB-HAP reported as
emissions from sources included in RTR. Given the potential for exposure via non-inhalation
pathways to these two PB-HAPs for RTR facilities in general, and the relatively high emissions
of dioxin (relative to the de minimis level) for Portland cement facilities in particular, mercury
and dioxin were selected as the chemicals for the case study of non-inhalation human health
risks.
3.4.2 Selection of facility for case study
To narrow the scope of the case study and enable a more in-depth evaluation, we focused on a
single Portland cement facility. We first identified Portland cement facilities that had high
emissions for both mercury and dioxins, assumed that higher emissions of the chemicals would
generally lead to higher human exposures, and began with facilities having dioxin emission rates
exceeding the de minimis levels described above, as well as facilities with relatively high
mercury emission rates. Of these facilities, we looked for one that had geographic characteristics
most similar to the two most significant basic multipathway exposure scenarios (consumption of
produce and animals and consumption of fish). Minimum requirements included (a) close
proximity to a freshwater lake of reasonable size, and (b) proximity to land used to support a
range of agricultural activities (crops and animals).
The Ravena Lafarge Portland cement facility (hereafter referred to as the Ravena facility) in
Ravena, NY, meets these criteria and was selected for evaluation in this case study. The Ravena
facility is near populated areas, several fishable water bodies, and potential farmland. Although
this facility may not necessarily represent the highest multipathway risk of all 91 Portland
cement facilities, it is useful for demonstrating the methods of the refined multipathway human
health risk assessment (HHRA), i.e., what to do when the emissions from a source category
exceed the de minimis levels. This is expected to be useful for soliciting feedback on a range of
risk assessment-related issues pertaining to EPA's RTR program.
The facility is located approximately 12 miles south of Albany, NY, in the southeastern portion
of Albany County (U.S. Census Bureau 2000; population 294,570). The population of Ravena,
Page 3-12
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
NY, located just east of the facility, is 3,369. Nearby counties include Renesselaer, Greene, and
Columbia, all in New York. In the 2002 U.S. Department of Agriculture (USDA) Census of
Agriculture, these four counties reported that livestock were raised there and crops were grown
for human and animal consumption (USDA 2002).
For the purpose of the Ravena HHRA, only dioxin and mercury emissions were evaluated. A
scenario layout for the Ravena area was created to use in TRIM.FaTE so that all relevant
ingestion pathways could be modeled.
3.4.3 Approach to exposure assessment
For this RTR case study, multipathway exposure estimates and risks were calculated for mercury
and dioxin for two basic scenarios:
A farmer scenario involving an individual living on a farm homestead in the vicinity of the
source who (a) consumes produce grown on and meat and animal products raised on the
farm, and (b) incidentally ingests surface soil at the location of the farm homestead; and
A recreational angler scenario involving an individual who regularly consumes fish caught in
freshwater lakes in the vicinity of the source of interest.
These two basic scenarios are expected to cover most of the highest possible exposures and risks.
In addition to ingestion, non-inhalation exposure to PB-HAPs can also occur by way of the
dermal pathway. However, the risk from dermal exposure is expected to be a small fraction of
the risk from inhalation exposure or ingestion exposure. Therefore, the risk from dermal
exposure was calculated as a special scenario as part of this site-specific refined analysis.
3.4.4 Fate and transport modeling (TRIM.FaTE)
Fate and transport modeling of PB-HAPs was completed using the Fate, Transport, and
Ecological Exposure Module (TRIM.FaTE) of EPA's Total Risk Integrated Methodology
(TRIM). TRIM.FaTE is a fully coupled multimedia model that estimates the flow of pollutants
through time among environmental compartments including air, soil, water, and fish. For
detailed information on TRIM.FaTE, refer to EPA's TRIM website
(http://www.epa.gov/ttn/fera/trim gen.html).
Ingestion exposures were calculated for the two exposure scenarios of interest using the
TRIM.FaTE media concentrations and typical ingestion exposure algorithms similar to those
found in the Human Health Risk Assessment Protocol [7], Chemical concentrations in
intermediate farm food types (e.g., produce, animal products) were calculated using biotransfer
factors to estimate the food chemical concentration based on the air and soil concentrations and
deposition rates from TRIM.FaTE. The RTR Multipathway Screening TSD (Appendix C)
provides details of the approach and methods used to calculate ingestion exposures. Individual
lifetime cancer risks for dioxins and chronic non-cancer hazard quotients for dioxins,
methylmercury, and divalent mercury were then calculated using oral cancer slope factors and
ingestion reference doses (RfDs).
3.4.4.1 Source characterization
For this case study, we modeled dioxin emission rates based on mean and 95th percent upper
confidence limit emission factors based on the clinker production of the facility. We present
Page 3-13
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
details about the development of these emission factors in Appendix F. The divalent and
elemental mercury emissions modeled were those reported in the 2002 NEI, and transformation
of divalent mercury into methylmercury in the sediments was included in the model. Table 3-5
presents the estimated mercury and dioxin emissions to air from the Ravena facility.
Table 3-5. Emissions of Dioxins and Mercury from the Lafarge Facility in Ravena, NY,
and Screening Results
PB-HAP
Emissions
(tons per year)
Screening Results
Dioxins3
95 percent upper confidence limit of
mean estimated emission factor
3.28E-06
Exceeds de minimis
level
Estimated mean emission factor
1.34E-06
Exceeds de minimis
level
Mercury - DivalentD [soluble fraction, likely mercuric
chloride]
5.63E-02
Screens out
Mercury - ElementalD [It is assumed that elemental
mercury is transported beyond the modeled domain.]
1.69E-01
Screens out
a Emissions estimated based on tons of clinker produced using dioxin emission factors.
b Emissions reported in 2002 National Emissions Inventory (NEI) (EPA 2002).
The modeling scenario duration was 50 years (i.e., sufficient time to achieve steady state
concentrations in the environment), and emissions of both mercury and dioxin were assumed to
be constant over the course of the simulation. TRIM.FaTE was used to estimate chemical
concentrations in air, soil, and selected surface water bodies (and their corresponding benthic
sediment layer), as well as components of a representative aquatic ecosystem in each water body
of interest for the risk assessment.
3.4.4.2 Extent and dimensions of modeled environment
The TRIM.FaTE surface parcel layout is the two-dimensional configuration of soil and water
regions included in the modeled domain; this is overlain by the air parcel layout. These layouts
provide the spatial reference for three-dimensional compartments that hold the modeled chemical
mass. The design of the modeling layout was developed based primarily on physical/geographic
characteristics of the watersheds in the Ravena area and land-use data for the region. When
designing the surface parcel layout, we sought to accurately capture the watersheds surrounding
the water bodies selected for modeling (i.e., those that contain fish that people are assumed to
eat). In pursuing this goal, parcel shapes were kept as simple as possible to reduce complexity in
the layout and the corresponding run time for the model.
The overall spatial extent of the air parcel layout is identical to that of the surface parcel layout,
and the square surface source parcel where the Ravena facility is located is identical in size,
shape and position to the air source parcel. For this assessment, the remaining air parcel layout
was designed as a radial grid centered around the source parcel, consistent with information
presented in the EPA's TRIM.FaTE Users' Guide [40]. This radial layout minimizes the
TRIM.FaTE bias for over-accumulation of mass along the axes of the grid. Overall, 31 air
parcels, including the source parcel, are included in the air parcel layout.
The overall spatial extent of the modeling scenario is a 770 km2 rectangle that captures several
significant water bodies in the area and their watersheds. Both divalent mercury and dioxins can
accumulate in the farm food chain, so the scenario layout includes two farm homesteads, on the
Page 3-14
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
east and west sides of the facility. The farm homesteads were located in areas where land use is
classified as agricultural.
Methylmercury and dioxins bioaccumulate in fish, so four freshwater water bodies were also
included in the Ravena layout to estimate exposure for the angler scenario. The Ravena area
encompasses many other water bodies including the Hudson River, but for the purposes of
TRIM.FaTE modeling, fish populations in three lakes and one pond were modeled. Alcove
Reservoir is 7 miles west of the Ravena facility and supplies drinking water to the city of
Albany. Kinderhook Lake (8 miles southeast of the facility) and Nassau Lake (11 miles
northeast) allow recreational fishing. All three of these lakes are large enough to support large
fish populations and were modeled in TRIM.FaTE. A small pond is located 2 miles southwest of
the facility. The pond was also modeled, although there is significant uncertainty whether it is
large enough to support a fishable aquatic ecosystem. The Ravena facility is within 2 miles of
the Hudson River, which was also modeled as a water body in this case study. A fish population
was not modeled in the river because of historically high pollutant levels in the river and the
difficulty in accurately modeling pollutant movement through a river.
3.4.4.3 Abiotic environment
TRIM.FaTE requires various abiotic environmental properties for each compartment that is
included in the scenario (e.g., the depth of surface soil, soil porosity and water content, erosion
and runoff rates from surface soil to water bodies, suspended sediment concentration, and
others). Where site-specific data were readily available for this assessment they were used. For
example, representative site-specific values based on available data were developed to estimate
erosion rates for each surface parcel. Rainfall/erosivity values were used from Albany County
for plots west of the Hudson River and Rensselaer County for regions east of the Hudson River
[41\. Soils data were obtained from the Soil Survey Geographic (SSURGO) database for the
counties of interest (obtained from the USD A Natural Resources Conservation Service) to
calculate site-specific soil erodibility factors. Different cover management factors were used for
farm parcels, natural forests, and grasses and herbs.
Regional or national defaults were used in numerous instances, especially for those parameters
that are not expected to influence chemical concentration dramatically. For example, a regional
pH value of 6.8 was used based on data compiled by McKone et al. [42] for use in multipathway
modeling since variation in pH is not expected to dramatically impact fate and transport of the
modeled chemicals. A complete list of TRIM.FaTE inputs for abiotic compartments is provided
in Attachment 1 to Appendix I of this document. Surface water and sediment properties for all
lakes and the river, along with the sources for these values are also listed in Attachment 1 of
Appendix I.
For the modeled water bodies, a water balance was assumed in order to estimate annual flush
rates by accounting for inputs to each water body (i.e., runoff from the surrounding watershed
and direct precipitation to the lake) and outputs from the water body (i.e., flushing through the
lake outlet and evaporation from the lake surface.) In addition, sediment inputs and outputs were
assumed to balance. The sediment balance of each watershed/water body system modeled was
estimated by accounting for sediment inputs to the lake based on the erosion calculations and the
removal of sediment from the modeled system via benthic burial and outflow of suspended
sediment in the water column.
Page 3-15
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
TRIM.FaTE uses several meteorological inputs to determine chemical transfers among the air
compartments in a scenario via advective transport {i.e., wind-driven physical movement through
the atmosphere) and from air to underlying soil or water surfaces via deposition transfers. These
processes determine the long-term spatial patterns of chemical distribution within the scenario,
and modeled concentrations are highly sensitive to the meteorological inputs used in
TRIM.FaTE.
The meteorological inputs required by TRIM.FaTE include wind speed, wind direction,
precipitation, ambient air temperature, and mixing height. For this assessment, hourly surface
meteorological data from the National Oceanic and Atmospheric Administration's National
Climatic Data Center (NCDC) Integrated Surface Hourly (ISH) Database [-/J] were obtained for
the closest meteorological station, located in Albany, NY. Three consecutive years of data (for
2001-2003) were readily available and therefore used from this data set.
3.4.4.4 Aquatic ecosystem
To estimate risks to human health for the angler scenario, site-specific models of aquatic food
webs were developed in TRIM.FaTE to represent the four modeled water bodies in the vicinity
of Ravena, NY {i.e., Nassau and Kinderhook Lakes, Alcove Reservoir, and the unnamed small
pond near the facility. Characteristics of the TRIM.FaTE fish compartments used to represent
fish in each water body were based on site-specific fish survey data, supplemented by
information from the open literature.
The development of each food web consisted of three stages. First, for the three lakes, we
collected local fish survey data for the water bodies from the New York State Department of
Environmental Conservation (NY DEC), including data on the relative abundance and
size/weight distribution of each species, to the extent available. Next, we formulated simplified
food webs for each water body, including the Ravena Pond, based on the fish surveys and other
biological and physical data for each water body. We used supplemental information on fish
feeding habits, aquatic food webs, and biomass densities for different trophic levels from the
open literature. Finally, we assigned values for the remaining parameters {e.g., individual body
weight, numeric density per unit area, lipid content) for each biotic compartment for each water
body in TRIM.FaTE from the available data. Professional judgment was used where available
data were incomplete. The process employed to configure TRIM.FaTE aquatic food webs and
set model input properties is discussed in greater detail in Addendum C of Attachment 1 of
Appendix I.
The following fish species were modeled in the TRIM.FaTE fish compartments:
Water Column Herbivore: Black crappie, common carp, fantail darter, golden shiner, and
young of the year;
• Benthic Omnivore: Bullhead and sunfish;
Water Column Omnivore: Bluegill, pumpkinseed, redbreast sunfish, rock bass, smallmouth
bass, white perch, white sucker, and yellow perch;
• Benthic Carnivore: American eel;
Water Column Carnivore'. Chain pickerel, largemouth bass, northern pike, tiger musky, and
walleye.
Page 3-16
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
3.4.5 Exposure assessment
3.4.5.1 Approach and exposure parameters
For the Ravena facility site-specific HHRA we evaluated a range of ingestion exposures for
situations that could be encountered in the vicinity of the Ravena facility. The range of
conditions considered when conceptualizing and building the scenario was chosen so that for any
given individual, a long-term exposure condition would be reasonably likely to be captured.
A summary of the sources of contaminated media for each of the three exposure scenarios
evaluated is provided in Table 3-6.
Table 3-6. Ingestion Exposure Scenarios
Scenario
Source of Ingested Media
Consumption of locally-grown
produce and animal products,
and incidental ingestion of soil
Products and soil from two locations with
agricultural land use:
o East Farm parcel
o West Farm parcel
Consumption of locally-
caught fish by sport anglers
Fish from four water bodies:
o Alcove Reservoir
o Kinderhook Lake
o Nassau Lake
o Small pond to south
Ingestion of contaminated
breast milk by infants
Breast milk; nursing mother would ingest
farm and fish media from most exposed
locations
For both the farmer and angler scenarios, we assumed that all media consumed were obtained
from locations impacted by the Ravena facility. We estimated the central tendency exposure
(CTE) using mean ingestion rates obtained primarily from EPA's Exposure Factors Handbooks
data on home-produced food consumption for adults [44] and children [45], The reasonable
maximum exposure (RME) was estimated using the 90th percentile of the distribution of national
ingestion rates from the Exposure Factors Handbook. This approach (consuming only
contaminated media and ingesting at the 90th percentile rates for all products) resulted in an
overestimate of total exposure. However, these conservative assumptions ensure that exposure
from any single food item is not underestimated. The CTE scenarios offer a less conservative
estimate of exposure.
Other characteristics of exposed individuals were also obtained primarily from EPA's Exposure
Factors Handbook. Table 3-7 summarizes the exposure parameters used in the CTE and RME
estimates.
Page 3-17
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-7. Exposure Parameters Used to Derive Risk and Hazard Estimates.
1-2 years
old
3-5 years
old
6-11 years
old
12-19
vcars old
20-69
vcars old
Body Weight
12.6 kg
18.6 kg
31.8 kg
64.2 kg
71.4 kg
Exposure Frequency
365
days/year
365
days/year
365
days/year
365
days/year
365
days/year
Exposure Period
Non-Cancer
Hazard Quotient
2 years
3 years
6 years
8 years
50 years
Cancer Risk
Lifetime cancer risk calculated with sum of risks from 5 exposure
periods above.
Averaging Period
Non-Cancer
Hazard Quotient
2 years
3 years
6 years
8 years
50 years
Cancer Risk
Lifetime cancer risk calcu
ated with sum of risks from 4 averaging
periods above.
90th Percentile
Ingestion Rates
Beef (g/kg/day)
4.5
6.7
11.4
3.53
5.39
Dairy(g/kg/day)
148
82
54.7
27.0
34.9
Other (g/kg/day)
95.6
65.7
49.7
33.3
41.1
Fish (g/day)
3.2
4.8
6.8
9.0
17
Mean Ingestion Rates
Beef (g/kg/day)
1.5
2.2
3.8
1.7
2.6
Dairy(g/kg/day)
67
37
24.8
10.9
17.1
Other (g/kg/day)
37.6
26.8
18.9
12.9
16.3
Fish (g/day)
1.4
2.0
2.7
3.9
6.9
3.4.5.2 Exposure dose estimation
Ingestion exposures for the angler and farmer scenarios for all media were calculated using the
Multimedia Ingestion Risk Calculator (MIRC) as average daily doses (ADDs), expressed in
milligrams of PB-HAP per kilogram of receptor body weight per day (mg/kg-day). Inputs used
to estimate exposure dose and risk included the following PB-HAP environmental media
concentrations from TRIM.FaTE:
Air concentrations (in |ig/m3);
Air-to-surface deposition rates for both particle and vapor phases (in (j,g/m2-yr);
Fish tissue concentrations (in mg/kg wet weight); and
Concentrations in surface soil and root zone soil (in j_ig/g dry weight).
These PB-HAP-specific values were then multiplied by empirical biotransfer factors (e.g., soil-
to-plant factors, which are the ratios of the concentrations in plants to concentrations in soil) to
calculate chemical concentrations in farm food chain media and the receptor- and exposure
scenario-specific ADDs. The equations used are presented in Appendix I, Attachment 4.
The calculated average daily doses and lifetime average daily doses were used with carcinogenic
potency slope factors (SFs) for ingestion and non-cancer oral reference doses (RfDs) for chronic
exposures to calculate individual lifetime cancer risks and hazard quotients, respectively.
Page 3-18
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
This assessment is intended to estimate the maximum individual risk for the exposure scenarios
evaluated, and the results are not intended to represent the actual exposure for a typical person
living in the vicinity of the evaluated source. Rather, we estimated the exposure for a person
who meets the criteria of the scenarios evaluated - that is, someone who consumes only produce
grown and animals raised on local farms, and/or someone who regularly consumes self-caught
fish from a local lake.
3.4.5.3 Risk Calculations
For this scenario-based risk assessment, we calculated lifetime individual cancer risks for dioxins
and non-cancer hazard quotients for dioxins, divalent mercury, and methylmercury using the
corresponding carcinogenic potency slope factors for ingestion and oral non-cancer reference
doses shown in Table 3-8.
Table 3-8. Dose-response Values for PB-HAPs Addressed in this Assessment
PB-HAP
Oral Cancer Potency
Original
Ingestion Reference
Original
Slope Factor
Source
Dose
Source
flmg/kg-dayl"1)
(mg/kg-day)
Mercury
(elemental)
NA
NA
Mercuric chloride
NA
3.0E-04
IRIS
Methylmercury
NA
1.0E-04
IRIS
2,3,7,8-TCDD
1.5E+05
EPA ORD b
1.0E-09
ATSDR
NA = not applicable. IRIS = EPA's Integrated Risk Information System; EPA ORD = EPA's Office of Research and
Development; ATSDR = U.S. Agency for Toxic Substances and Disease Registry. Values presented here are recommended by
OAQPS for evaluation of HAPs [8],
3.4.5.4 Breast milk pathway
The US EPA [46, 47] and the World Health Organization (WHO) [48, 49] have published
reports documenting the presence of environmental chemicals and contaminants in human breast
milk. These chemicals are ingested by the mother and partition into breast milk. A nursing
infant may be exposed subsequently via the mother's breast milk. The nursing infant's exposure
can be estimated from the levels of chemical concentrations in the breast milk, which in turn can
be estimated from the mother's chemical intake. Exposures can occur for infants via this
pathway for dioxins and mercury.
Exposure to dioxins and mercury via breast milk consumption during the first year of life is
expected to have a small effect on the estimated lifetime ADD and on the individual's excess
lifetime cancer risk for dioxins or the highest chronic non-cancer hazard for either chemical.
Therefore, exposures to these chemicals via the breast milk pathway were not considered in
estimating the lifetime cancer risk for dioxins or chronic non-cancer hazard quotients for
mercury or dioxins for adults. The potential for non-cancer health effects (e.g., when exposures
are compared to the ATSDR MRL, which is based on developmental effects endpoints) is of
greater concern for nursing infants exposed to either chemical during the first year of life.
The methodology and algorithms used to evaluate the breast milk consumption scenario for this
case study are presented separately in Attachment C-2 of Appendix C.
3.4.5.5 Dermal pathway
Page 3-19
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Compared to both inhalation and ingestion pathways, dermal exposure to PB-HAPs is expected
to be a minor exposure pathway. To assess the significance of the dermal exposure pathway,
dermal hazard quotients were determined in soil and water for 2,3,7,8-TCDD, divalent mercury
(Hg2+), and methylmercury (MHg) as described in EPA's dermal risk assessment guidance for
Superfund [50], These quotients were then summed in order to determine an appropriate HQ for
each chemical and age class. Site-specific soil and water concentrations from the Ravena
modeling scenario were used.
3.5 Ecological risk assessment
3.5.1 Ecological risk screening
As mentioned above, PB-HAP emissions were screened for potential multipathway human health
risks using the TRIM-based screening methodology. Emissions of any PB-HAP not passing the
initial screen for human health endpoints were assumed to also create a potential for adverse
multipathway environmental effects and subjected to more refined ecological assessment (in
addition to the human health assessment).
In addition, for both petroleum refineries and Portland cement manufacturing, the potential for
adverse ecological effects of non-PB-HAPs in air was generally screened by evaluating the
potential for chronic ambient air concentration estimates to exceed chronic human health
inhalation thresholds in the ambient air near these facilities. That is, if chronic ambient
concentrations were not estimated to exceed their respective chronic reference concentrations,
the potential for adverse environmental effects associated with direct contact with air was
considered to be insignificant. The rationale behind this thinking is that, in general, chronic
human health dose-response threshold values for HAPs are derived from studies conducted on
laboratory animals and developed with the inclusions of uncertainty factors that in some cases
aggregate as high as 3000. As a result, these human health benchmarks are often significantly
lower than levels expected or observed to cause adverse effects observed in studies with other
species. We note that there is a scarcity of data on direct atmospheric impacts of these HAPs on
other receptors, such as plants, birds, and wildlife. In those cases where the maximum predicted
inhalation hazard in an ecosystem is below the level of concern for humans, we have concluded
that mammalian receptors are unlikely to be at risk of adverse effects due to inhalation exposures
from non PB-HAPs, and have assumed that other ecological receptors are similarly not at any
significant risk.
EPA has not yet developed general criteria to select candidate HAPs for direct-contact ecological
assessments. However, the large masses of hydrogen chloride (HC1) emitted by Portland cement
facilities, and the unusually reactive and acidic nature of these emissions, suggested that HC1
should be an appropriate candidate to evaluate for potential adverse effects to ecological
receptors by direct contact {i.e., rather than by multipathway exposures). Accordingly, we
included an assessment of the threshold ambient air concentration for HCl-induced damage to
plant foliage and compared it with the threshold for chronic human health effects. As a result,
we concluded that HC1 emissions from Portland cement facilities did not pass the ecological
screening and we included them in our refined ecological assessment. Our choice of HC1 for this
case study is not meant to suggest that other HAPs do not pose similar concerns via direct
contact; rather, the case study is meant to demonstrate the refined ecological risk assessment
methodology for the purposes of review by the SAB.
Page 3-20
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
3.5.2 Refined ecological risk assessment
3.5.2.1 ERA for mercury and dioxin
A refined multipathway ecological risk assessment (ERA) was performed for the Portland
cement source category. The portion of that ERA evaluating the potential impacts of dioxin and
mercury utilized the same case study facility as in the refined human health multipathway
assessment, building off of the estimated media impacts to develop estimates of exposures for
four key ecological species (tree swallow, common merganser, bald eagle, and mink) living near
each of four bodies of water in the vicinity of the facility. The rationale for the selection of these
key species and a detailed description of the methods used to estimate their exposures (including
the sources of dietary information) are described in Appendix J. Finally, to characterize the
impacts of these exposures, they are compared to derived Toxicity Reference Values (TRV)
obtained from the literature intended for those species (Table 3-9). The comparison takes the
form of Hazard Quotient (HQ) values (Table 3-10 through Table 3-12), which are used to
determine if these exposures might be expected to result in adverse effects. The development of
the species-specific TRV values and HQs is explained in Appendix J.
Table 3-9. Summary of Wildlife TRVs (ng|"chemical"|/kg|"BW"|-day) for Ravena
Avian Values
Mink Values
Chemical
POD
(MQ/kg-
day)
UFTot
TRV (Mg/kg-
day)
POD
(Mg/kg-
day)
UFTot
TRV (Mg/kg-
day)
2,3,7,8-TCDD
14 E-03
10
1.4 E-03
1.0 E-03
10
0.10 E-3
Methylmercury
78
6
13
55
30
1.8
Divalent Mercury
N/A
N/A
Smaller birds:
26
Larger birds: 65
300
30/1.55
16
N/A = Not applicable.
Table 3-10. Hazard Quotients for Wildlife Exposure to Methylmercury for Ravena
Wildlife Species
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
0.605
0.004
0.005
0.006
Common Merganser
1.304
0.004
0.006
0.005
Bald Eagle
0.634
0.002
0.003
0.003
Mink
3.919
0.014
0.021
0.020
a Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
Table 3-11. Hazard Quotients for Wildlife Exposure to Divalent Mercury for Ravena
Wildlife Species
Water
Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
2.37
<1
<1
<1
Common Merganser
0.40
<1
<1
<1
Bald Eagle
0.04
<1
<1
<1
Mink
0.98
<1
<1
<1
a Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
bThe HQs for Hg+2 are likely to be less than 1.0 at water-bodies other Ravena Pond given that exposure
doses are more than two orders of magnitude lower for wildlife consuming prey from those water bodies.
Page 3-21
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 3-12. Hazard Quotients for Wildlife Exposure to 2,3,7,8-TCDDa for Ravena
Wildlife Species
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
0.01
0.00004
0.0001
0.0001
Common Merganser
0.70
0.001
0.01
0.002
Bald Eagle
0.77
0.001
0.01
0.004
Mink
4.27
0.003
0.03
0.01
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
b Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
The only water body for which any HQ exceeds 1 is a hypothetical worst-case farm pond with a
surface area of only 0.02 km2 and a shoreline of 0.8 km. Such a water body would probably
support very few individuals of these species, and that any adverse impacts to these few
individuals would be unlikely to adversely impact populations of those species. We conclude
that populations of piscivorous and insectivorous wildlife should not be adversely affected by
methylmercury, divalent mercury, and dioxin emissions from the facility.
3.5.2.2 ERA for hydrogen chloride
3.5.2.2.1 Local impacts
The portion of the refined ERA focusing on the potential impacts of HC1 proceeded along two
parallel paths, one aimed at evaluating the potential impact of HC1 emissions from the source
category directly on plant leaves in the vicinity of individual facilities and the other aimed at
evaluating the potential for individual facilities to cause or contribute to soil or water
acidification in their vicinities to the extent that adverse impacts on plants or animals might
result. The first path focused on evaluating estimated maximum ambient impacts from the
inhalation risk assessment and comparing them against derived benchmarks for foliar damage.
The second path revolved around identifying specific facilities in the source category with the
highest potential to cause acidification and then searching for soil and water pH data near those
facilities to see if effects can actually be detected.
Following the first path, we conducted a literature search in the attempt to locate information that
could be used in developing HC1 ecological exposure thresholds for foliar damage. Over 50
scientific databases were accessed in the literature search (described in Appendix K). Available
studies included information about gaseous HC1 injury to plants from visual observations,
photosynthetic and oxygen evolution rates, and electron microscopy of localized cellular damage
following exposure. Investigators consistently concluded that foliar damage is caused by
gaseous HC1 condensing on the leaf surface, producing an aqueous acid solution that promotes
cellular injury with degree of injury proportional to exposure to gaseous HC1. This injury is not
specific to HC1, but would be expected with exposure to any strong acid.
Available studies were all designed to determine the impact of short-term, high-concentration
exposures to gaseous HC1. While these data can provide strong support for the development of
acute ecological exposure thresholds, more uncertainty is involved in extrapolating these data to
develop chronic thresholds.
Page 3-22
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Given the limited number of available studies, however, we developed 20-minute threshold
exposure estimates (described in Appendix K) based on the lowest reported LOEL (1.5 mg/m3, at
which traces of leaf discoloration and necrosis occurred) and LOAEL (4 mg/m3, at which 25% of
leaves were necrotic), rather than the multiple-effect-level approach recommended in EPA
Guidelines for Ecological Risk Assessment. To be consistent with our dispersion model outputs,
we selected acute and chronic exposure periods of one hour and one year, respectively. We
extrapolated the LOEL and LOAEL exposures to 1-hour equivalent concentrations of 0.5 and 1
mg/m3, respectively using the common application of Haber's law, as modified by ten Berge et
al. [57], Lacking long-term study data, we applied an additional uncertainty factor of 10 to
extrapolate the lower of the two acute thresholds (0.5 mg/m3) from a 1-hour to a 1-year exposure
threshold of 0.05 mg/m3. In our refined ecological assessment for HC1 impacts on plants, these
thresholds were used to evaluate maximum ambient HC1 concentration estimates near Portland
cement facilities for potential foliar damage. It is worth noting that while the 1-year threshold
for foliar damage is greater than the RfC for health effects (0.05 vs. 0.02 mg/m3, respectively),
the 1-hour threshold for foliar damage is less than the 1-hour California REL (0.5 vs. 2.1 mg/m3,
respectively).
3.5.2.2.2 Regional impacts
Following the second path (evaluating the potential for HC1 emissions to cause acidification), we
developed a ranking procedure to determine indirect effects of HC1 deposition on ecologically
sensitive environments. Facilities were ranked according to emission rates, the pH of regional
rainfall, surface water alkalinity, and proximity to sensitive environments. Following the
identification of potential high-impact facilities, we searched for environmental measurement
data near each of the top 4 sources to determine if such measurements might corroborate or
refute the hypothesis that current emission levels are resulting in localized acidification impacts.
This ranking procedure and the subsequent data search process are described in detail in
Appendix J.
3.6 Risk characterization
3.6.1 Inhalation risks
3.6.1.1 Chronic inhalation risk assessment results
The maximum individual cancer risk (MIR) is 800 in a million, dominated by risks associated
with emissions of hexavalent chromium compounds and cadmium compounds. Out of the 104
facilities included in the assessment, 8 are associated with an MIR greater than 10 in a million
and 29 are associated with an MIR greater than 1 in a million. We estimated the total cancer
incidence attributable to the source category to be 0.05 excess cancer cases per year, with about
93% of the total contributed by hexavalent chromium compounds, arsenic compounds, cadmium
compounds, beryllium compounds, and benzene. We estimate that 15,000 people reside in areas
where the lifetime cancer risk estimate exceeds 10 in a million, and 470,000 people reside in
areas where lifetime cancer risk exceeds 1 in a million.
The maximum chronic noncancer hazard index for the Portland cement manufacturing source
category is 10, associated with potential effects of manganese compounds on the central nervous
system. Other potentially important effects include a respiratory hazard index of 6 (associated
with chlorine and hydrogen chloride), and a kidney hazard index of 3 (associated with cadmium
Page 3-23
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
compounds). We estimate that 3000 people reside in areas where the hazard index for
neurological effects exceeds 1, 200 people where the respiratory hazard index exceeds 1, and 170
people where the kidney hazard index exceeds 1.
3.6.1.2 Acute inhalation risk screening and refined results
The maximum acute screening hazard quotients (HQs) for hydrogen chloride were 4 and 50,
based on potential exceedance of the AEGL-2 and AEGL-1, respectively. Maximum HQs for
chlorine and formaldehyde were 7 and 3, respectively, also based on potential AEGL-1
exceedances. All other acute HQs were less than 1. The 8 facilities that exceeded an acute HQ
of 1 at the screening level were targeted for a more refined evaluation and are presented in
Appendix E. The refined analysis looks at the proximity of maximum predicted impacts to plant
property line. Following this refined assessment, maximum predicted acute HQ for hydrogen
chloride is 10 based on potential exceedance of the AEGL-1, and less than 1 based on the
AEGL-2. Maximum HQs for chlorine and formaldehyde are both 2, also based on potential
AEGL-1 exceedances.
When considering acute risks it is important to understand that acute health benchmarks, like any
dose-response values, are surrounded by uncertainty. For the Portland cement source category,
every acute HQ that exceeded 1 was based on the AEGL-1, a one-time mild-effect acute value.
These results suggest that (1) facilities that have one-time acute exposures above the AEGL-1 are
likely to cause increases in mild, reversible, but nevertheless adverse health effects, and (2) those
whose predicted exposures are below the AEGL-1 may or may not pose acute health risks.
3.6.1.3 Radionuclides results
As described in Section 3.2.2 and Appendix G, we tested possible strategies to evaluate
radionuclide hazards by estimating emissions for two Portland cement facilities in California that
reported emissions in the 2002 NEI. We developed one emissions estimate for each facility
using NEI-reported data, and two estimates based on the NORM report [39],
The NORM emission estimates for the facilities (i.e., those based on clinker production and PM,
respectively, shown in Table 3-13) fell nearly within the same order of magnitude, but were
many orders of magnitude less than the NEI-based emissions.
Table 3-13. Estimation of Radionuclide Emissions for the Two California Facilities
Using Three Approaches
NTI Site ID
Emissions, Based on
NEI Emissions and
Speciation
Assumptions
Emissions, Based on
Clinker Production
Scaling Factors
Emissions, Based on
PM Emission Scaling
Factors
210Po
(Ci/yr)
Rn
(Ci/yr)
210Po
(Ci/yr)
Rn
(Ci/yr)
210Po
(Ci/yr)
Rn
(Ci/yr)
NEICA1505122
3.48E+07
7.01 E+07
9.59E-01
1.93E+00
7.20E-02
1.45E-01
NEI2CA151186
6.02E+01
1.21E+02
8.13E-01
1.64E+00
1.03E-01
2.07E-01
Estimated cancer risks (Table 3-14) associated with the NEI emissions exceeded unity, but only
reached 10 in one million for the NORM-based estimates.
Page 3-24
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Table 3-14. Risk Calculated for Two California Portland Cement Facilities Using AERMOD
Modeling Results and Three Emission Estimation Approaches
NTI Site ID
Cone.
(MQ/m3)
MIR, Based on NEI
Emissions and
Speciation
Assumptions
MIR, Based on
Clinker Production
Scaling Factors
MIR, Based on PM
Emission Scaling
Factors
210Po
Rn
210Po
Rn
210Po
Rn
NEICA1505122
1.53E-03
2.94E+02
9.86E-01
8.09E-06
2.71 E-08
6.07E-07
2.04E-09
NEI2CA151186
2.53E-03
8.43E-04
2.83E-06
1.14E-05
3.82E-08
1.44E-06
4.82E-09
Using the NORM-based clinker production scaling factor, radionuclide emission rates were
extrapolated to 91 facilities modeled for the source category. Where actual clinker production
data were not available for a facility, clinker production was assumed to equal 95 percent of
clinker production capacity, based on the median actual production relative to production
capacity from all facilities having data. See Appendix G, page G-7, for details. Maximum
incremental risks were estimated for each using the HEM3 model. Of these 91 domestic
Portland cement facilities, 4 were estimated to have radionuclide-associated maximum cancer
risk higher than 100 in a million. Approximately 35 percent of the facilities (32) were estimated
to have maximum cancer risk higher than 10 in a million, and all but one facility had maximum
cancer risks higher than 1 in a million. These risk estimates, which are more or less similar in
magnitude to risks from all other HAPs combined, suggest that radionuclide emissions may be
an important source of risk for this source category. However, the extremely poor quality of
available radionuclide emissions data prompts caution in the interpretation of these risk values,
especially when comparing to better characterized risks.
In summary, using NEI mass emission estimates for radionuclides appears to result in
unrealistically high maximum incremental risk estimates, suggesting that these emissions were
reported incorrectly. However, risk estimates based on NORM-based emission rates are still
high enough to merit serious concern and to suggest that the lack of adequate radionuclide
emission data is an important gap in RTR risk assessments.
3.6.2 Multipathway risks
The results of the human health multipathway risk assessment are presented in this section.
Section 3.6.2.1 focuses on the results for 2,3,7,8-TCDD equivalence (a measure that includes all
dioxins) and Section 3.6.2.2 focuses on the results for mercury.
For both chemicals, the concentrations and human health risks estimated in this assessment are
also compared to analogous outputs estimated using the hypothetical multipathway screening
scenario developed for RTR. To accomplish this comparison, the Ravena emission rates were
modeled in the TRIM.FaTE screening scenario layout that is used in Step 1 of the multipathway
HHRA to derive the de minimis levels for screening. In addition, the results from modeling the
Ravena emissions in the screening scenario illustrate the level of conservatism associated with
the screening scenario and provide additional context for the results estimated for this site-
specific risk assessment. Throughout the multipathway HHRA discussion, the results of
modeling the Ravena emissions in the screening scenario are labeled "Screening Scenario."
In general, the presentation of results here favors those calculated using RME ingestion rates that
Page 3-25
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
are unlikely to occur but still within the bounds of what is possible. Exposures and risks
calculated using more typical CTE ingestion rates for these scenarios are presented as well in
some cases for comparison. More detailed discussion and tables of results are presented in
Appendix I.
3.6.2.1 2,3,7,8-TCDD results
For 2,3,7,8-TCDD, media concentrations and risks were estimated for two emission rates, one
based on a mean emission factor and a second rate based on the 95 percent upper confidence
limit (UCL) of the dioxin emission factor (to provide an upper bound risk estimate that takes into
account the uncertainty regarding the emissions estimate). A summary of results follows; a
complete description of the multipathway risk assessment case study can be found in Appendix I.
For this case study, we estimated individual lifetime cancer risk and non-cancer HQs for 2,3,7,8-
TCDD (assumed to be representative of risks from all emitted dioxins) for three scenarios
(farmer, angler, and breastfeeding infant) and a range of combinations involving these three
scenarios and the food source for the exposed individual. Risk estimates for two emission rates
(mean and 95 percent UCL) and two sets of ingestion rate assumptions (central tendency
exposure [CTE] and reasonable maximum exposure [RME]) were evaluated. In addition, it was
assumed that emission and ingestion rates are constant over the exposure time period for each
age group (2 to 50 years for the hazards for different ages, with the cancer risk calculated from
the sum of exposures in each age bin).
Estimated media concentrations
TRIM.FaTE results for the east and west farm parcels were similar, with air concentration and
surface soil concentration higher at the east farm and dry deposition higher at the west farm. The
concentrations in fish estimated by TRIM.FaTE were generally lower than total dioxin TEQ
concentrations measured in fish in the Hudson River and associated bays, for all water bodies
included at the Ravena site (including the pond), with a difference between the modeled and
measured values of several orders of magnitude. This outcome seems reasonable given that the
model includes a single source of chemical emissions to the air, while the reported values reflect
all local and regional sources of dioxins, the contribution of existing background concentrations
of dioxins from long-range sources, and any contributions from non-air sources (likely including
historical PCB contamination introduced to the Hudson River).
Cancer risk
Cancer risk estimates for 2,3,7,8-TCDD are summarized in Figure 3-1.
Page 3-26
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Figure 3-1. 2,3,7,8-TCDD Individual Lifetime Cancer Risks for Ravena
1.E-03
^ 1.E-04
tfl
5
0)
o
c
TO
O
o
.1 1.E-05
.0
TO
3
¦g
">
- 1.E-06
1.E-07
¦ Mean Emission Rate
¦ 95% UCL Emission Rate
Alcove
Nassau
Ravena
East Farm
West Farm
East Farm,
West Farm,
West Farm,
Screening
Reservoir
Lake
Pond
Alcove
Nassau
Ravena
Scenario
Reservoir
Lake
Pond
Angler Only
Farmer Only
Combined Scenarios
Note: Presented results assume 90th percentile ingestion rates for all age groups (RME). For the reader's reference,
the yellow lines mark a risk of 1 in 1 million (1e-6) and of 1 in 10,000 (1e-4).
In general, a lifetime individual cancer risk between 1 and 10 in a million was estimated for the
combined farmer/angler and individual (farmer or angler) scenarios, assuming RME ingestion
rates and using the 95 percent UCL dioxin emission factors, for all farm and water body
locations evaluated with the exception of the ponds. Consumption of self-caught fish (for the
angler scenario) and consumption of beef and dairy products (for the farmer scenario) are the
exposure pathways driving cancer risk estimates, with the proportional contribution of these
pathways varying by farm and lake location.
Introducing fish harvesting to the Ravena pond within the TRIM.FaTE model probably portrays
more realistic fish concentrations and reduces the estimated lifetime cancer risk from 170 in a
million to 120 in a million. However, the introduction of fish harvesting at this rate is unlikely to
be ecologically sustainable, and at a minimum proves to significantly reduce the chemical
concentrations in fish tissues in all fish types. We maintain that the water body with the second
highest cancer risks, Nassau Lake, is a more realistic upper bound on potential Ravena area
exposures for the angler scenario.
Impact of ingestion rates and dioxin emission factor
If the central tendency ingestion rates are used for produce/meat/animal products and fish or if
mean dioxin emission rates are assumed, the estimated individual cancer risk is approximately 3
in a million or less (a decrease of approximately 40 to 60 percent for scenarios excluding
Page 3-27
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
ingestion of fish from the Ravena pond). As expected, CTE ingestion rates are about 40 percent
of RME ingestion rates for the food types influencing exposures for the combined farmer/angler
scenario; the decrease in risk is proportional to the decrease in relevant ingestion rates. Using
the mean dioxin emission factor assumption decreases the risk by roughly the same proportions
(the difference between emission factors is similar to the difference in ingestion rates. In
combination, if both CTE ingestion rates and mean emission factors are assumed, the estimated
individual cancer risk for any given scenario is about 20 percent of the estimated risk when RME
ingestion rates and the UCL of the dioxin emission factor are assumed.
Comparison to RTR screening scenario
When individual cancer risks are estimated for the same emission rates, but modeled using the
generic RTR screening exposure scenario (i.e., hypothetical modeling environment and high-end
farmer and angler exposure scenarios), the risk results are 110 and 46 in a million for 95 percent
UCL and mean dioxin emission factors, respectively. These risks are between one and two
orders of magnitude higher than site-specific risk estimates for the Ravena scenarios (not
including those assuming ingestion of fish from the pond). These results provide an indication of
the degree of conservatism that the screening scenario holds, at least in comparison to the site-
specific risk assessment conducted for the Ravena site.
Dermal exposure cancer risks
Dermal exposures and associated lifetime cancer risks were estimated for soil and water
exposures (as per [50]). Despite a conservative modeling approach, dermal cancer risk varied
from 60 to 590 times less than ingestion risk under different exposure scenarios and locations.
Because dermal exposure appears to add so little to ingestion risks, we did not evaluate it further.
Chronic non-cancer hazard quotient
Chronic non-cancer HQs are shown in Figure 3-2. HQs are below 0.1 for all farmer scenarios
evaluated and all angler scenarios based on the higher dioxin emission factor, except when
consumption of fish from the pond is assumed to occur. The calculated HQ (based on adverse
liver, reproductive, developmental, endocrine, respiratory, and hematopoietic effects) for anglers
consuming fish from the pond is about 0.7 in children ages 3 to 5, and between 1.1 and 1.3 in all
other age groups, if RME ingestion rates are assumed. The estimated HQ for all age groups
drops to 1 or below if central tendency fish ingestion rates are assumed, if mean dioxin emission
factors are used, or if fish harvesting is introduced to the Ravena TRIM.FaTE modeling scenario.
By comparison, when the Ravena dioxin emissions were modeled in the RTR screening scenario,
the chronic HQs associated with the RME ingestion rates were estimated to be 0.5 to 1.5 if the
higher dioxin emission factor is used, and approximately 0.2 to 0.6 if mean dioxin emissions are
used.
Page 3-28
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Figure 3-2. 2,3,7,8-TCDD Chronic Non-cancer Hazard Quotients for Ravena
(95th Percentile UCL Emission Factor, RME Ingestion Rates)
I Child 1-2
Child 3-5
I Child 6-11
Child 12-19
I Adult 20-70
10
0.0001
0.00001
0.001
East Farm
West Farm
Alcove
Nassau Lake
Ravena Pond
East Farm,
West Farm,
West Farm,
Reservoir
Alcove
Nassau Lake
Ravena Pond
Reservoir
Farmer Only
Angler Only
Combined Scenarios
Infant exposures via breast milk
The HQs calculated for nursing infants were generally higher than HQs for the mothers by about
an order of magnitude when the same oral RfD was used to calculate HQ (as noted previously,
this analysis did not evaluate whether the RfD used for dioxins is appropriate for evaluating
chronic non-cancer hazards to nursing infants). Given the relatively low dioxin exposures
assumed for the mother, the calculated HQs for a nursing infant are still below one for all
scenario combinations evaluated except the angler-pond scenario. If the nursing mother is
assumed to consume fish from the pond, the calculated HQs for a breast-feeding infant are very
high given that the mother's HQ is calculated to be approximately 1.2, assuming RME ingestion
rates. However, as discussed above, it is unlikely that the pond provides a suitable environment
for sustained recreational fishing.
3.6.2.2 Mercury results
For this case study, we estimated individual non-cancer HQs for divalent and methylmercury for
three basic scenarios (farmer, angler, and breastfeeding infant) and a range of combinations
involving these three scenarios and the location of the exposed individual. It was assumed that
emission and ingestion rates are constant over the exposure time period for each age group (2 to
50 years for the hazards for different ages).
Page 3-29
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Estimated media concentrations
Model outputs for the east and west farm parcels were similar, with divalent mercury air
concentration and dry deposition higher at the west farm and surface soil concentrations and wet
deposition higher at the east farm. Mercury concentrations in fish estimated by TRIM.FaTE
were generally lower by several orders of magnitude than the divalent, methyl, and total mercury
concentrations measured in fish in the Hudson River and associated bays, for all water bodies
included at the Ravena site, with the exception of the pond. This difference between modeled
and measured concentrations seems reasonable given that the model includes a single source of
chemical emissions to the air, while the reported values reflect all local and regional sources of
mercury, the contribution of existing background deposition of mercury from long-range
sources, and any contributions from non-air sources (including residual mercury resulting from
historical deposition in the northeast United States).
Divalent mercury chronic non-cancer hazard quotient
For divalent mercury, chronic non-cancer HQs were below about 0.03 for all combined
farmer/angler scenarios evaluated when RME ingestion rates are assumed. As discussed
previously, the pond is not likely a plausible, viable source of fish for regular consumption by an
angler. If consumption of fish from the pond is excluded, chronic non-cancer HQs were below
0.004 for divalent mercury for all farmer and angler scenarios. When central tendency ingestion
rates were used for produce/meat/animal products and fish, the divalent mercury HQ decreased
by 50 to 75 percent.
0.00001
Figure 3-3. Mercury Chronic Non-Cancer Hazard Quotients for Ravena
Divalent Mercury
¦ Child 1-2
I Child 3-5
¦ Child 6-11
¦ Child 12-19
¦ Adult 20-70
a! 0.01
8 0.001
0.0001
Alcove
Reservoir
Kinderhook
Lake
Ravena Pond
West Farm
Alcove
Reservoir
West Farm East Farm
Farmer Only
East Farm,
Kinderhook
Lake
East Farm, Screening
Ravena Pond Scenario
Angler Only
Combined Scenarios
Note: Presented results were calculated using the 90th percentile ingestion rates (RME).
Page 3-30
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Methylmercury chronic non-cancer hazard quotient
For methylmercury, the estimated HQs ranged from very low values (less than 10"4), associated
with consumption of farm products from the west farm, up to a value of 0.2 calculated for
consumption of fish from the pond. If the scenario concerning fish consumption from the pond
was excluded, the highest estimated HQ for methylmercury was 0.002.
Figure 3-4. Mercury Chronic Non-Cancer Hazard Quotients for Ravena
Methyl Mercury
¦ Child 1-2
¦ Child 3-5
¦ Child 6-11
E Child 12-19
¦ Adult 20-70
0.1
¦5
3
¦g
n
ra 0.001
z
0.0001
0.00001
in1111
West Farm
East Farm
Alcove
Reservoir
Nassau Lake
Ravena Pond
West Farm,
Alcove
Reservoir
East Farm,
Nassau Lake
East Farm,
Ravena Pond
Farmer Only
Angler Only
Combined Scenarios
Note: Presented results were calculated using the 90th percentile ingestion rates (RME).
Comparison to RTR screening scenario
Chronic non-cancer HQs were calculated for divalent mercury using the same emission rates but
the general RTR screening exposure scenario (i.e., hypothetical modeling environment and high-
end farmer and angler exposure scenarios). These results were between 0.07 and 0.3 assuming
RME ingestion rates. These HQs are at least two orders of magnitude higher than site-specific
HQ estimates for the corresponding scenarios (with the exception of those involving
consumption of fish from the pond).
The chronic non-cancer HQs calculated for methylmercury using the screening scenario were
between 0.01 and 0.2 assuming RME ingestion rates. These results are about two orders of
magnitude higher than site-specific HQs (with the exception of those involving consumption of
fish from the pond). As noted, site-specific mercury results provide an indication of the degree
of conservatism associated with the screening scenario, at least in comparison to the case-study
Page 3-31
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
conducted for the Ravena site.
Exposures to infants via breast milk
As described in Attachment C-2 of Appendix C, there is significant uncertainty associated with
the parameters used to estimate mercury HQs for breastfeeding infants. For methylmercury, data
suggest that HQs for breastfeeding infants will be very similar to HQs for the nursing mother.
Given the high level of uncertainty, mercury exposures via breast milk were not quantitatively
modeled.
Dermal hazard quotients
Dermal hazard quotients were calculated for soil and water exposures as described in U.S. EPA
(2004). Using a highly conservative modeling approach, dermal HQs for divalent mercury were
on the same order as HQs calculated for ingestion exposures, suggesting that the methods used to
estimate dermal exposures may be overly conservative. However, because the dermal HQs were
well below levels of concern (i.e., less than 0.01 for the most conservative exposure scenario), no
additional evaluation was conducted.
3.6.3 Combining risks from all facilities and exposure routes
The multipathway risk assessment covers only a single Portland cement facility, and cannot
easily be applied to other similar facilities that may have different processes, emission
characteristics, meteorology, and surrounding populations. However, a simple extrapolation of
maximum individual risks can at least serve as a range-finding tool regarding the potential
importance of multipathway risks relative to inhalation risks for the entire source category.
With this in mind, we calculated multipathway risk-to-emission ratios for dioxin and mercury
emissions of the Ravena facility for three subsistence exposure scenarios: fishing in a nearby
lake, subsistence farming at the most contaminated nearby farm, and both. These calculations
assumed that all facilities emit mercury; where mercury emissions were not reported we
estimated them using the average mercury-to-dioxin ratio for facilities that reported both. We
then used the Ravena risk-to-emission ratios to estimate risks at the other 90 Portland cement
facilities for which we estimated dioxin emissions data. This simple extrapolation omits site-
specific variations in emission parameters (other than amount emitted), dispersion, and receptor
location and behavior, and in effect assumes that every facility differs from Ravena only in the
amounts of dioxin and mercury emitted. Since the Ravena facility was selected for
multipathway analysis in part because of its proximity to farmland and fishable water bodies, this
assumption is likely to create a high bias for risk estimates for other facilities. The results of this
extrapolation are shown in Table 3-15.
Table 3-15. Averages of extrapolated risks for dioxins and divalent
mercury emitted by the Portland cement source category, based on
emissions-to-risk ratios estimated for the Ravena facility.
Exposure scenario
Nassau Lake
West Farm
Kinderhook Lake
Dioxin
cancer risk
(avg/max)
2E-07 /1E-06
2E-07 /1E-06
Dioxin
HQ
(avg/max)
9E-03 / 6E-02
5E-03 / 3E-02
Divalent mercury
HQ
(avg/max)
3E-05 / 4E-04
Page 3-32
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Table 3-15. Averages of extrapolated risks for dioxins and divalent
mercury emitted by the Portland cement source category, based on
emissions-to-risk ratios estimated for the Ravena facility.
Dioxin Dioxin Divalent mercury
cancer risk HQ HQ
Exposure scenario (avg/max) (avg/max) (avg/max)
East Farm 1E-03 / 2E-02
Combined
fishing/farming 4E-07/3E-06 3E-02/2E-01 2E-03/3E-02
The Ravena risk and HQ estimates for dioxin were the highest for any facility in the source
category; Ravena HQ estimates for divalent mercury were about midway between the averages
and the maxima for the category. The highest lifetime cancer risk estimate for dioxin, for the
unlikely combined subsistence fishing and farming exposure scenario at the Ravena facility, was
3 in one million. The average risk for this combined scenario at all facilities was 4 in ten
million. At the average risk level it would require about 9 million subsistence fisher-farmers
living near Portland cement facilities to produce the inhalation-based incidence rate of 0.05
cancer cases per year. The highest HQs produced by any facility were 0.2 for dioxin and 0.03
for divalent mercury, suggesting no concern for noncancer hazards.
The individual risk estimates from the inhalation and multipathway assessments can be
combined by assuming that a subsistence fisher/farmer is also the person with maximum
inhalation exposure at each facility. For this source category, however, the multipathway risks
are within rounding error of the inhalation risks, and combining them would have no effect. In
summary, based on the preliminary emissions dataset used for this case study, health risks
associated with the Portland cement source category appear to derive mainly from inhalation
exposure rather than from indirect exposure.
3.6.4 General discussion of uncertainties
This risk assessment for the Portland cement source category currently exists only as a pre-
ANPRM draft. When complete, it will include a discussion of uncertainties similar to that in the
refineries case study (Section 2.4) that will appear in this section. In its current form it is subject
to the limitations and uncertainties in the following discussion. We intend to solicit public
comment about these parts of the assessment in the hope of reducing the uncertainties in the risk
estimates.
Chromium compounds were reported for about only about 30 percent of the sources in this
category, yet they dominate the cancer risk from this source category. Other HAP drivers
(e.g., cadmium, benzene, naphthalene) were also reported for less than half the facilities. It is
possible that the NEI does not have complete data for some sources that actually emit these
HAPs, and that the associated risks may therefore be biased low.
The reported speciation of chromium compounds into the most common oxidation states (III
and VI) significantly impacts predicted risk estimates. In the absence of additional
information, the default speciation profile applied to emissions reported as "chromium" or
"chromium compounds" for this source category was 92 percent chromium (III) compounds
and 8 percent chromium (VI) compounds. Because chromium (VI) compounds were a
Page 3-33
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
dominant risk driver in the assessment, risk estimates for sources with substantially different
speciation profiles may be either over- or underestimated.
It is likely that most or all facilities in this source category emit quantities of naturally
occurring radioactive materials, but only three facilities in California reported such emissions
and these emissions were not included in the ANPRM data set. The emissions appear to
have the potential to contribute substantially to total cancer risk, and risk estimates that omit
them may be biased low.
Emissions of dioxins as TCDD TEQ were estimated for every facility in this data set.
Dioxins contributed substantially to total multipathway risk in our case study, and to the
extent that our emission estimates were unrepresentative, the resulting risk estimates may be
biased either high or low.
As noted in Section 3.2, there is uncertainty in the identification of sources as major or area
in the NEI, which may have affected the risk estimates for the entire category.
Coordinates in the NEI are checked to ensure that they are generally correct (e.g., in the
correct county). However, there can still be errors in the coordinates that result in the
emission sources not being properly located on plant property. These errors have the
potential to bias the estimates of MIR either high or low.
As discussed in section 3.6.1.2, the screening assessment for acute impacts suggests that no-
effect levels for hydrogen chloride, chlorine, and formaldehyde could be exceeded under
worst-case meteorological conditions if maximum hourly emissions of these HAP exceed
their average hourly emission rate by a factor of 10. Given the generally conservative design
of our acute screening scenario, the HQ values estimated are likely greater than those which
could actually occur in the real world, but peak emissions of these compounds should be
quantified to support a more refined assessment of potential acute impacts.
Page 3-34
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
4 Supplemental analyses and discussion of uncertainty
4.1 Corrections to the emissions inventory - data analysis
In response to questions and concerns by both EPA scientists and the regulated community about
the quality of the NEI emission data used in the initial ANPRM screening stage of RTR risk
assessments, we compared the initial ANPRM assessment for petroleum refineries with the
revised NPRM assessment to determine how much the emissions estimates (and thus the
estimated cancer risks) changed as a result of public comment.35 (See Appendix A for details of
this analysis.)
The ANPRM data set for the petroleum refinery source category included 175 facilities.
Through the ANPRM process, data changes or revisions were received for 113, or 65 percent, of
the facilities. Changes to the data were supplied by EPA, State or local agencies, trade
organizations, and/or facilities themselves. Types of changes to the data included data
replacement, emissions changes, process changes, emission release point changes, and facility
changes. In addition, 30 facilities were removed and 8 facilities were added by EPA after the
screening risk assessment was conducted, resulting in 153 facilities in the NPRM data set.
The total HAP emissions included in the ANPRM data set equal 2,316 tons per year (tpy) and the
total HAP emissions included in the NPRM data set equal 2,292 tpy, for an overall reduction of
24 tpy (or 1%). These changes were evaluated by comparing the change in cancer toxicity-
weighted emissions {i.e., the emitted mass for each HAP was multiplied by its respective URE).
Overall, total toxicity-weighted emissions decreased by 12 percent from the ANPRM data set to
the NPRM data set. Toxicity-weighted benzene emissions decreased by 20%, POM emissions
by 26%, 1-3-butadiene emissions by 40%. On the other hand, toxicity-weighted naphthalene
emissions increased by 19% and nickel emissions by 150%.
Five facilities had maximum individual risk (MIR) estimates that exceeded 100 in a million in
the ANPRM assessment. During the comment period three of these facilities submitted revised
emission and stack parameters and another submitted revised emission rates only. The fifth was
determined not to be a refinery. EPA accepted these revisions, and each facility's MIR risk
estimate declined below 100 in a million for the NPRM assessment. No other facilities exceeded
this risk level in the NPRM assessment. Thirty-three facilities had MIR estimates between 10
and 100 in a million in the ANPRM assessment; 18 facilities were in this risk range in the NPRM
assessment. The comparison showed that, on average, facilities had lower MIR estimates in the
NPRM assessment. Facilities with higher MIR estimates in the ANPRM were more likely to
provide data changes, and these changes resulted in larger-than-average reductions in MIR.
Estimated overall cancer incidence for the petroleum refining category was 0.08 in the ANPRM,
and 0.05 in the NPRM. This reduction in estimated incidence was due almost entirely to data
changes for the highest-risk facilities.
35 The RTR process also allows for further refinements in the risk assessment between the NPRM and final rule, and
such refinements were made to the petroleum refineries assessment.
Page 4-1
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
4.2 Short-term emissions and exposures - data analysis
In contrast to the development of ambient concentrations for evaluating long-term exposures,
which was performed only for occupied census blocks, worst-case short-term (one-hour)
concentrations were estimated both at the census block centroids and at points nearer the facility
that represent locations where people may be present for short periods, but generally no nearer
than 100 meters from the center of the facility. For large facilities, this 100-meter ring could still
contain locations inside the facility property, which could lead to unrealistically high exposure
estimates in the acute screening. Since short-term emission rates were needed to screen for the
potential for hazard via acute exposures, and since the NEI contains only annual emission totals,
we applied the general screening-level assumption to all source categories that the maximum
one-hour emission rate from any source was ten times the average annual hourly emission rate
for that source. Average hourly emissions rate is defined as the total emissions for a year divided
by the total number of operating hours in the year (assuming either continuous operations or
more limited operating hours based on additional data). This choice of a factor of ten for
screening was originally based on engineering judgment.
Public comments on other RTR assessments have suggested that assuming a maximum hourly
emission rate equal to ten times the annualized rate may underestimate actual maximum short-
term emissions for some facilities, and thereby also underestimate maximum acute risks. To test
the conservatism of the tenfold emission rate assumption, we performed an analysis using a
short-term emissions dataset from a number of sources, several of which are refineries, located in
Texas (originally reported on by Allen et ol. (2004)[52]). In that report, the Texas
Environmental Research Consortium Project compared hourly and annual emissions data for
volatile organic compounds for all facilities in a heavily-industrialized 4-county area (Harris,
Galveston, Chambers, and Brazoria Counties, TX) over an eleven-month time period in 2001.
We obtained the dataset and performed our own analysis, focusing that analysis on sources that
reported emitting high quantities of volatile organic HAP over short periods of time (see
Appendix B, Analysis of data on short-term emission rates relative to long-term emission rates).
To evaluate the potential for release events to cause acute toxicity, we examined low-probability
events, e.g., release rates that are exceeded only one hour per year (0.011 % of the time). Ratios
of event release rate to long-term release rate varied from 0.00000004 to 74. The 99th percentile
ratio was 9 (i.e., an event release rate nine times the long-term average). Only 3 ratios exceeded
our default assumption of 10, and of these only one exceeded 11. All three with ratios greater
than 10 lasted less than one hour. The median ratio was less than two (i.e., less than twice the
annual average).
The factor of ten is intended to cover routinely variable emissions as well as startup, shutdown,
and malfunction (SSM) emissions, and although there are some documented emission excursions
above this level, our analysis suggests that this factor should cover more than 99% of the short-
term peak gaseous or volatile emissions from source categories like petroleum refineries.
Similar data were not available for particulate emissions from categories like Portland cement
manufacturing, however, and are not likely to be available for individual RTR source categories.
In summary, the tenfold ratio assumption for short-term releases appears to be reasonably
protective for the Texas VOC emitters for which data were available, but the analysis is limited
by a lack of speciated long-term release data and by an absence of data from facilities that did
Page 4-2
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
not experience a release event during the data collection period. It is also not clear whether, and
how, it is appropriate to extrapolate these results to other source categories for which short-term
release data do not exist.
4.3 In ventory under-reporting and gaps - data analysis
4.3.1 Ambient monitor-to-model comparison for two Texas refineries
As discussed in section 2.4.1 above, the development of the RTR emissions databases involved
quality assurance and quality control processes, but the accuracy of emissions values will
nevertheless vary depending on the original source of the data, the amount of incomplete or
missing data, errors in estimating emissions values, and other factors.
In order to ground-truth our facility-specific risk assessment results, we compared ambient
monitoring data for benzene from two monitoring sites to our dispersion modeling results for
those facilities (Appendix L). Benzene monitoring data were obtained from the Texas
Commission on Environmental Quality (TCEQ) for two benzene monitors in Texas City, TX.
These monitors are each located near residential areas, and within 300 meters of major industrial
sources including three large refineries (BP Refining, Marathon, and Valero Refining) and one
chemical manufacturing facility (Sterling Chemicals).
A year of hourly monitoring data for each site was paired with hourly measurements of wind
speed and wind direction. Raw hourly ambient data were evaluated and adjusted so that non-
detected (ND) values were replaced with V2 the minimum detection limits (MDLs).
Measurements that lacked matching hourly wind directions were omitted in order to support a
statistical analysis of directional source contributions at each monitor. We estimated benzene
contributions from other sources in the vicinity of each refinery using the background estimates
for the 2002 National Air Toxics Assessment (NATA). We adjusted the monitored
concentrations by subtracting these background estimates from each measurement to develop
estimates of the refinery-specific benzene contributions at each monitor.
We used AERMOD to develop modeled ambient benzene concentrations due to petroleum
refineries alone at both monitor locations using emissions data and meteorological data from the
Galveston airport that represented the same time period as the monitor data. All modeling
options were identical to those used in the baseline petroleum refinery assessment modeling.
We used analysis of variance to compare average modeled and monitored benzene
concentrations, and also the average difference between the two, among 16 wind direction
sectors. We used regression analysis to determine if a relationship exists between wind speed
and the ratio of hourly monitored to modeled benzene concentrations, a measure of model error.
Annual averages for modeled estimates, monitor data, and the difference between them all varied
significantly with wind direction at both monitors (P<0.001). Results for the monitor near the
BP facility showed a reasonable resemblance between modeled and monitored benzene levels.
The effects of the nearby refinery can be clearly seen in both sets of estimates and there appeared
to be little overall bias in the annual modeled estimate. Results for the monitor near the
Marathon facility showed that the model substantially underestimated the average measured
benzene concentrations for every wind direction, and that the difference increased substantially
Page 4-3
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
when the wind blew from the source. When winds moved from the facility toward the monitor,
measured concentrations exceeded modeled estimates by more than 2-fold. These results
suggest that the benzene emissions inventory for the BP refinery was representative of actual
emissions, but that the inventory for the Marathon refinery may have been underestimated by
more than twofold. There is no way to know which (if either) facility is representative of the
whole sector.
One EPA staff reviewer of Appendix L disagreed with some of the methods used and
conclusions reached by the authors. This reviewer's comments and suggestions are attached to
the Appendix for consideration as an alternative viewpoint.
4.3.2 Comparison of RTR emissions inventory data and Refineries Emissions
Model (REM) data
Throughout the development of the Risk and Technology Review (RTR) program, one
potentially significant area of uncertainty has been the quality of emissions data from individual
sources. While the general approach has been discussed elsewhere in this document, we note
again that there are questions as to the emissions data quality due, in part, to inconsistencies in
the values across pollutants and individual sources within a category. Emissions data are
essentially estimates since few monitored data exist. Our confidence in emissions estimates
varies depending on the original source of the data, the amount of apparent incomplete or
missing data, questionable emissions values, and other factors.
To highlight some of these uncertainties in the emissions data and their associated estimated
cancer inhalation risks, we compared two emissions datasets - the RTR inventory and an
emissions dataset developed using the Refineries Emissions Model or "REM" [53, 54] - both of
which are reasonable approaches to estimating emissions. After emissions estimates were
developed, a dispersion/risk analysis was undertaken. Chronic inhalation exposure
concentrations and associated health risks from each facility of interest were estimated using the
Human Exposure Model in combination with the American Meteorological Society/EPA
Regulatory Model dispersion modeling system (HEM-AERMOD, sometimes called HEM3).
The cancer risks associated with each facility's estimated emissions were evaluated using the
same dispersion models, exposure assumptions, and unit risk factors that were used to estimate
risks based on the RTR data. It is important to note, however, that unlike the RTR database that
sometimes includes (for less than half of emissions points) source-specific locations and release
characteristics, emissions specifications (e.g., location and release characteristics) are not
included in REM. Other limitations and uncertainties are described in Appendix P.
This analysis is not without significant uncertainties. While the purpose of the analysis is to
compare risk results from two different approaches to estimating emissions, the REM approach
did not account for specific controls at specific facilities. Except in the case of equipment leaks,
which included some information on controls due to consent decrees and state/local
requirements, the emission factor approach assumed facilities were only controlling at the
MACT level or were uncontrolled in the case of cooling towers, which currently do not have a
MACT requirement. While the RTR data can account for additional control measures, there is
an unquantifiable amount of uncertainty in how emissions are estimated and if they are estimated
correctly and completely. Secondly, differences in pollutant coverage may also contribute to
these uncertainties. REM includes the 19 HAPs that make up the vast majority of the mass of
Page 4-4
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
emissions included in the RTR database. Additionally, REM assumes these common HAPs
(e.g., benzene, naphthalene) are present at all refineries whereas none of these pollutants are
reported at all refineries in the RTR database. However, RTR includes 37 additional pollutants
reported at up to 34 facilities. While some of these may be reported in error or inappropriately
associated with this source category, others may be emitted by individual facilities in this source
category and are just not included in the REM dataset because they are not universally emitted
by all refineries. Finally, there are differences in the modeling due to different approaches to
estimating emissions (e.g., specific number, location, and height of stacks or specific size and
location of tank farms vs. default assumptions) that may influence the overall risk results.
Without more detailed analyses, we cannot quantify the impact of these uncertainties.
Given the uncertainties in this analysis, it is challenging to draw firm conclusions from these
findings. Nonetheless, we summarize the main points from the analysis.
First, across all refineries and HAPs, emission estimates are 2.6 times higher using REM; at the
facility level, differences between REM and RTR estimates can be an order of magnitude or
more.
Second, using the high-end estimate of benzene potency, the highest facility MIR, 30 in 1
million, was the same using RTR and REM data, although the highest-risk facilities were
different. The source category MIR for the RTR analysis was driven by naphthalene and POM.
The source category MIR for the REM analysis was driven by benzene, naphthalene and POM.
The MIR using the low-end estimate of benzene cancer potency is 20 in 1 million for REM and
remains 30 in 1 million for the RTR analysis.
Third, assuming the high-end benzene potency value for both analyses, the distribution of facility
MIR estimates shifted upwards using the REM data compared to the RTR data; 135 facilities in
the REM analysis have MIR estimates greater than 1 in 1 million and 41 facilities have MIR
estimates greater than 10 in 1 million, whereas 77 facilities using RTR emissions have MIR
estimates greater than 1 in a 1 million and five facilities using RTR emissions have MIR
estimates greater than 10 in 1 million. We do not know what the distribution of facility MIR
estimates is using the equally probable lower estimate of benzene potency.
Fourth, we looked at the facilities with the highest MIRs from the REM and RTR analyses, using
the higher estimate of benzene potency. The top 20 facilities with the highest MIRs based on
RTR data have REM-based MIR estimates within an order of magnitude. For the top 20 REM-
based MIR estimates, there was somewhat more variability in the magnitude of differences to
RTR-based MIR estimates; 14 of these facilities showed differences in estimates of less than an
order of magnitude, but the remainder of differences were at least a factor of 10 (and as high as
3,000-fold). Using the low-end benzene estimate may alter these differences, depending on the
relative amounts of benzene estimated at each facility.
Fifth, we note that the facilities with the highest MIRs (using the high-end benzene cancer
potency value) in either approach are generally different facilities, suggesting a more pronounced
difference in the influence of the emissions estimation approach at the facility level than in
aggregate. Additionally, the facilities with the highest MIRs in either case, with two exceptions,
are not among the facilities with the most dramatic differences in emissions. These order of
magnitude changes for facilities did not shift any individual facilities to have MIRs greater than
Page 4-5
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
or equal to 100 in 1 million, but we cannot judge how alternative emissions estimation
approaches might affect other source categories. We did not evaluate this issue using the low-
end cancer potency value.
Sixth, depending on which benzene cancer potency estimate is used, the estimate for cancer
incidence using the REM emissions estimates is three to four times higher than the incidence
estimate using the RTR emissions estimates (using the high-end benzene potency estimate, REM
incidence is 0.2 cases per year and RTR incidence is 0.05 cases per year; using the low-end
benzene potency estimate, REM incidence is 0.1 cases per year and RTR incidence is 0.03 cases
per year).
Finally, petroleum refinery emissions are thought to be relatively well-understood compared to
those for some other source categories. Therefore, the result that the MIRs are similar in this
case may be unique to this source category. It is difficult to generalize the results of this analysis
to other source categories.
4.4 Time scale of meteorological data - sensitivity analysis
Ideally, site specific dispersion modeling efforts will employ up to five years of meteorological
data to capture variability in weather patterns from year to year. However, because of the large
number of facilities in the analysis and the extent of the dispersion modeling analysis (national
scale), it was not practical to model five years of data and only the year 1991 was modeled. The
selection of a single year may result in under-prediction of long-term ambient levels at some
locations and over-prediction at others. To examine the sensitivity of ambient concentrations
(and risk estimates) to the use of single-year versus 5-year meteorological data, we ran HEM3
using single-year and 5-year meteorological datasets from four different locations for the
petroleum refinery with the highest estimated cancer risk (NEI12486). The four locations
(Lancaster, CA; Charlotte, NC; Detroit, MI; and Houston, TX) selected represent several
different climates. To determine the impact of meteorological data only we varied those data
only, but did not relocate the facility or its surrounding census block and polar receptors.
As shown in Table 4-1, we compared single-year and 5-year averages of several risk metrics,
including cancer MIR, incidence, and acute benzene concentration. For cancer MIR and
incidence, the comparison was of single-year average to 5-year average. For the acute
concentration, the comparison was of the highest hourly value in a single year to the highest
hourly value in the 5-year period. Consequently, for this comparison, the highest acute
concentration for any single year will always be less than or equal to the highest acute
concentration for a 5-year period.
The single-year and 5-year estimates of cancer MIR and incidence differed by only 7 percent on
average. The single-year average MIR and incidence differed from the 5-year average by as
much as 18 percent below to 28 percent above, but differed by less than 10 percent in 15 out of
the 20 comparisons. The highest acute concentration for a single year was, on average, 10
percent lower than the highest acute concentration for a 5-year period, with a maximum
difference of 37 percent when examining the Charlotte NC meteorological data. A closer look at
the Charlotte data found that one of the five years was significantly higher then the other 4 years.
An examination of the meteorological data for year with the highest acute concentration at
Page 4-6
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Charlotte suggests that an unlikely confluence of factors led to that value, including a very low
wind speed and a wind direction in line with the source and receptor.
In summary, the relatively small differences in risk metrics described above suggest that, in a
majority of the cases considered, that the use of meteorological data for a single year does not
introduce significant uncertainty into the risk assessment relative to other sources of uncertainty
that limit reporting risk estimates to one significant figure.
Page 4-7
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 4-1. Comparison of Risk Assessment Results for 1-Year vs. 5-Year Meteorological Data for a Petroleum Refinery (NEI12486).
Percent Difference Between 1 and 5-Year DataJb
37
Location
Data
Year
Cancer
MIR
Incidence
Acute
Benzene
Cone.
(MQ/m3)
Cancer
MIR
Incidence
Acute
Benzene
Cone.
(MQ/m3)
Lancaster, CA
90
1.43E-04
8.390E-04
3954
2
-2
-22
91
1.39E-04
8.280E-04
5053
-1
-3
0
92
1.40E-04
9.040E-04
4635
0
6
-8
93
1.39E-04
8.280E-04
5025
-1
-3
-1
94
1.39E-04
8.690E-04
4890
-1
2
-3
90-94
1.40E-04
8.540E-04
5053
NA
NA
NA
Charlotte, NC
91
1.20E-04
4.380E-04
2395
6
5
-37
92
1.12E-04
3.900E-04
2523
-1
-7
-34
93
1.09E-04
3.820E-04
3811
-4
-9
0
94
1.10E-04
3.960E-04
2546
-3
-5
-33
95
1.15E-04
4.850E-04
2532
2
16
-34
91-95
1.13E-04
4.180E-04
3811
NA
NA
NA
Detroit, Ml
02
1.93E-04
9.897E-04
7301
28
11
-6
03
1.48E-04
9.136E-04
7523
-2
2
-3
04
1.43E-04
8.198E-04
7593
-6
-8
-2
05
1.34E-04
8.661 E-04
7081
-11
-3
-9
06
1.38E-04
8.773E-04
7739
-9
-2
0
02-06
1.51E-04
8.934E-04
7739
NA
NA
NA
Houston, TX
87
8.06E-05
4.950E-04
6781
-16
17
-3
88
9.72E-05
4.740E-04
6889
2
12
-1
89
9.69E-05
3.660E-04
6959
1
-14
0
90
1.20E-04
4.390E-04
6643
26
4
-5
91
8.29E-05
3.460E-04
6779
-13
-18
-3
87-91
9.56E-05
4.241 E-04
6959
NA
NA
NA
Mean of the Absolute Values of the Percent Differences
7
7
10
36 For cancer MIR and incidence, the comparison is of single-year average to 5-year average. For acute concentration, the comparison is of the highest hourly
value in a single year to the highest hourly value in the 5-year period.
37 A negative value indicates that the 5-year value is higher than the single-year value.
Page 4-8
-------
June 15,2009
Review
RTR Risk Assessment Methods for SAB
4.5 Location of meteorological stations - sensitivity analysis
Meteorological data for HEM3 are selected from a list of 158 National Weather Service
(NWS) surface observation stations across the continental United States, Alaska, Hawaii, and
Puerto Rico. In most cases the nearest station is selected as representative of the conditions at
the subject facility. Two petroleum refinery facilities38 furnished representative
meteorological datasets as part of the ANPRM process. For these two facilities, the facility-
supplied meteorological data were utilized in place of the HEM "nearest selected" station.
For each facility, located by its characteristic latitude and longitude coordinates, the nearest
meteorological station was used in the dispersion modeling. The average distance between a
modeled facility and the nearest meteorological station was 72 km. Usually, the nearest
meteorological station is the most appropriate to use because it best represents the conditions
at the facility. However, there are situations where a more distant meteorological station may
better represent facility conditions. For example, the nearest meteorological station for an
inland facility may be on the coast, but the coastal effects on winds may make a more distant
meteorological station more appropriate to use. We performed a sensitivity analysis to
examine the variability attributable to the selection of meteorology station. We selected four
petroleum refineries in different climates (Torrance, CA; Texas City, TX; Canton, OH; and
Marcus Hook, PA) that had at least three surface meteorology stations within 200 km of the
refinery. We then ran HEM3 for each refinery and for each meteorology dataset to estimate
cancer MIR, incidence, and acute benzene concentration. The results are given in Table 4-2.
Overall, cancer MIR, incidence, and acute benzene concentration differed from the values
based on the nearest meteorological station by 26, 41, and 17 percent, respectively. Cancer
MIR varied by a much as 63 percent below to 51 percent above the value based on the nearest
meteorological station. Incidence varied by a much as 68 percent below to 120 percent above
the value based on the nearest meteorological station. The acute benzene concentration varied
by a much as 49 percent below to 21 percent above the value based on the nearest
meteorological station. In summary, in three of four cases the meteorological station nearest
the facility yielded risk estimates similar to most of the more distant stations. In the fourth
case the more distant stations yielded risk estimates that were characteristically 20 to 40
percent lower, but it is not clear that the more distant stations would be more representative.
Overall, the differences usually fall within rounding error for the 1-significant-figure
characterization of risks, and therefore appear to be relatively less important than other
sources of uncertainty, e.g., dose-response values or emission rates.
38 For NEI8406, data from the Fairbanks, Alaska met station from the year 2001 modeled and for NEI46556,
data from St. Croix, Virgin Islands met station from the year 2005 was utilized.
Page 4-9
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Table 4-2. Impact of Meteorological Station Selection on Risk Assessment.
Percent Difference With Nearest Station a
Acute
Acute
Surface
Upper Air
Distance to
Benzene
Benzene
NEI ID
Meteorological
Station
Meteorological
Station
Surface Station
(km)
Cancer
MIR
Incidence
Cone.
(Mg/m3)
Cancer
MIR
Incidence
Cone.
(Mg/m3)
Philadelphia, PA
Atlantic City, NJ
17
2.0E-5
7.8E-4
1040
0
0
0
Wilmington, DE
Atlantic City, NJ
22
2.1E-5
8.0E-4
1060
7
3
2
NEI109
Allentown, PA
Albany, NY
91
2.1E-5
9.1E-4
1080
4
17
4
Baltimore, MD
Sterling, VA
127
2.0E-5
9.4E-4
1110
-1
21
7
Sterling, VA
Sterling, VA
196
2.7E-5
8.9E-4
1100
36
14
6
Akron, OH
Pittsburgh, PA
16
5.1E-6
1.6E-4
397
0
0
0
Cleveland, OH
Pittsburgh, PA
78
7.8E-6
3.5E-4
464
51
120
17
NEI11574
Pittsburgh, PA
Pittsburgh, PA
102
4.8E-6
1.8E-4
412
-6
13
4
Columbus, OH
Dayton, OH
149
5.3E-6
2.4E-4
431
4
50
9
Erie, PA
Buffalo, NY
176
5.4E-6
2.0E-4
387
5
25
-3
Galveston, TX
Lake Charles, LA
12
1.6E-5
1.2E-3
17434
0
0
0
Houston, TX
Lake Charles, LA
80
6.0E-6
3.9E-4
9160
-63
-68
-47
NEI12044
Port Arthur, TX
Lake Charles, LA
105
8.9E-6
6.6E-4
12500
-44
-45
-28
Lake Charles, LA
Lake Charles, LA
180
1.2E-5
6.9E-4
21100
-25
-43
21
Victoria, TX
Corpus Christi, TX
201
7.6E-6
4.7E-4
8940
-53
-61
-49
Los Angeles, CA
Miramar, CA
10
8.7E-7
7.8E-4
171
0
0
0
NEI21034
San Diego, CA
Miramar, CA
162
1.1 E-6
4.7E-4
146
21
-40
-15
Daggett, CA
Desert Rock, NV
180
4.4E-7
3.4E-4
129
-49
-56
-25
Mean of the Absolute Values of the Percent Differences
26
41
17
a A negative value indicates that the value for the station is lower than the value for the station nearest the source.
Page 4-10
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
4.6 A tmospheric chemistry - sensitivity analysis
While the AERMOD model is not capable of simulating complex atmospheric chemical
reactions, such as those simulated by the CMAQ model, it does contain the option to simulate an
exponential atmospheric decay of the pollutant being modeled, a chemical transformation
process which is common to many gaseous pollutants such as VOCs and SO2. For the general
RTR dispersion modeling and risk characterization, this feature was omitted from the analysis,
under the assumption that such atmospheric decay would not occur prominently over the
transport distances and time scales typically involved in estimating maximum risk impacts. To
test this assumption, we conducted a separate modeling study to evaluate exactly how much
including the atmospheric decay in our simulations would change our estimates of maximum
individual risk (MIR), cancer incidence levels, and noncancer HQs.
For the petroleum refineries source category risk characterization, the primary risk drivers were
seen to be benzene, ethylene dibromide, naphthalene, and polycyclic organic matter (POM). In
general, none of these pollutants is expected to be highly reactive; with the exception of
naphthalene and other small POM whose average half-lives are around 10 hours, their typical
atmospheric half-lives are on the order of days or tens of days [55], Only a few of the major
pollutants emitted by petroleum refineries have atmospheric half-lives less than 12 hours. They
are (in increasing order of estimated half-life): 1,3-butadiene (ca. 1 hour), aniline (2 hours),
formaldehyde (4 hours), cresols (4-5 hours), phenol (9 hours), and acrolein (12 hours). Four
other HAPs emitted by petroleum refineries have atmospheric half-lives between 12 and 24
hours; all other HAPs have estimated half-lives greater than a day.
To simulate the effects of the exponential decay on the MIR and incidence levels, we modeled
one of the highest risk petroleum refineries (NEI7988) utilizing the AERMOD exponential decay
option using the atmospheric half-life of the risk driver, benzene, which is reported as 14 days.
Table 4-3 and Figure 4-1 depict the results of this model run, indicating that both MIR and
incidence values showed no significant changes. We also performed simulations of this source
utilizing smaller and smaller half-life values to determine how short the half-life needed to be to
effect significant reductions in these estimated risk values. These results are also presented in
Table 4-3 and Figure 4-1. For pollutants with half-lives greater than about 30 minutes, predicted
MIR values are reduced by less than a few percent, suggesting that neglecting the influence of
atmospheric decay for these pollutants is appropriate in predicting MIR estimates. For pollutants
with half-lives of about 10 minutes or less, MIR impacts are reduced by at least 10%. A review
of available literature on atmospheric half-lives for HAPs identified only one with an estimated
atmospheric half-life less than 10 minutes (N-nitrosodimethylamine, with an estimated half-life
of about 3 minutes). It is not known to be emitted by petroleum refineries. In addition, we
identified two additional HAPs with half-lives less than 1 hour (methyl hydrazine and 1,1-
dimethyl hydrazine, each with an estimated half-life of about 30 minutes). Accurately estimating
MIR values for sources of these pollutants may require including the simulation of atmospheric
decomposition.
Because cancer incidence can be contributed to by exposures significantly farther from the
facility than the MIR location, the effect of atmospheric decay can be noticed at longer half-life
values. As presented in Table 4-3 and Figure 4-1, a 5% reduction in incidence can be noted for
Page 4-11
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
pollutants with a half-life of 8 hours and a 10% reduction for pollutants with a half-life of 4
hours.
Since we report our risk results to one significant figure (in order to avoid implying greater
precision in the results than is warranted), these results suggest that including the simulation of
atmospheric decay in this type of risk assessment is only necessary for pollutants whose half-
lives are less than about 4 hours, and will not impact the estimation of MIR. For petroleum
refineries, the only pollutants meeting this criterion are 1,3-butadiene, aniline, and formaldehyde.
Since these pollutants were seen to contribute minimally to the cancer risks for petroleum
refineries, omitting their reactivity will have little impact on the chronic risk results for this
source category.
Table 4-3. Effects of Exponential Decay on MIR and Incidence levels.
Pollutant Half Life
MIR
% Reduced
Cancer
% Reduced
(pollutant)
Incidence
None (RTR Modeling)
29.3
-
0.000581
-
14 days (benzene)
29.3
0.0001
0.000580
0.1
4 8-hour
29.3
0.0004
0.000575
1.0
24-hour
29.3
0.0008
0.000569
2.0
12-hour
29.2
0.2
0.000558
3.9
8-hour
29.2
0.2
0.000548
5.7
4-hour
29.1
0.5
0.000520
10.6
2-hour
29.0
1.0
0.000474
18.5
1-hour
28.7
2.0
0.000408
29.8
30 minutes
28.2
3.9
0.000326
43.9
10 minutes
26.1
10.8
0.000190
67.2
5 minutes
23.4
20.1
0.000118
79.7
1 minute
11.3
61.4
0.000022
96.2
32 seconds (acrolein)
6.2
79.0
0.000008
98.6
Page 4-12
-------
June 15, 2009 RTR Risk Assessment Methods for SAB Review
Figure 4-1. MIR/Incidence Reduction as a function of Half-Life
Incidence
MIR
c
o
o
3
¦o
a>
0
5
10
15
20
25
30
35
40
45
50
Half-Life (hrs)
As a final note, while it may be important to include the impact of atmospheric decay in order to
accurately estimate MIR and incidence for the direct impacts of a relatively few fast-reacting
HAP, this does not account for the potential for the by-products of such decomposition to cause
any health risks. In fact, many decomposition products of rapidly-decaying HAPs are
themselves HAPs. Formaldehyde, for example, is a byproduct of the atmospheric decomposition
of acrolein, 1,3-butadiene, acetaldehyde, and propionaldehyde. While the creation of such
byproducts depends on many interactions in the atmosphere that are beyond the scope of the
typical RTR assessment, their potential formation should nonetheless be acknowledged when
characterizing the ultimate risk results.
4.7 Deposition - sensitivity analysis
While AERMOD is capable of simulating the deposition of particulate pollution, we have
generally not incorporated the simulation of particle deposition into the dispersion modeling
being performed for RTR assessments, hypothesizing that its role is relatively minor in the
calculation of risk metrics for these types of sources. Since much of the pollution arising from
Portland cement manufacturing is emitted in particle form, we decided to test the hypothesis for
this source category.
We chose the 5 highest risk facilities for the sensitivity analysis. We first simulated atmospheric
dispersion and risks without including deposition and plume depletion in the calculation and then
repeated the simulations using the deposition and depletion algorithms contained in AERMOD.
We used parameters to characterize the deposition process which were indicative of a typical fine
particle emission mixture after a particulate control device (baghouse or fabric filter, which are
Page 4-13
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
required for Portland cement facilities). The results, shown in Table 4-4, provide a comparison
of MIR values and total cancer incidence values with and without including the simulation of
deposition. For all 5 facilities, deposition was seen to decrease the estimated MIR values by less
than 3% (ranging from 0.05% to 2.27%). The estimated cancer incidence values for each facility
decreased slightly more than this when deposition was included in the simulation, but all values
decreased by less than 4% (ranging from 0.32% to 3.85%).
Table 4-4. Comparison of estimated cancer MIR and incidence with and without considering
deposition and depletion at five Portland cement facilities.
Facility ID
Maximum
Individual
Cancer Risk
(with dep and
part and vap
depletion)
Maximum
Individual
Cancer Risk
(without
deposition
or depletion)
%
Change
Incidence
(with dep
and part
and vap
depletion)
Incidence
(without
dep
or
depletion)
%
Change
PTC NEI22453
5.00E-05
5.11E-05
-2.12
8.33E-04
8.64E-04
-3.59
PTC NEI22838
2.14E-06
2.16E-06
-1.21
7.34E-06
7.62E-06
-3.64
PTC NEI2PRT14367
6.54E-06
6.57E-06
-0.44
4.32E-04
4.37E-04
-1.16
PTC NEIAL1150002
4.19E-06
4.19E-06
-0.05
7.19E-04
7.22E-04
-0.32
PTC NEIAZ0250421
1.68E-06
1.72E-06
-2.27
4.71 E-05
4.90E-05
-3.85
The results of this sensitivity analysis bore out the hypothesis that ignoring deposition for these
Portland cement sources will not significantly affect the risk results. It should be noted,
however, that this result should not be extrapolated to the simulation of sources which contain
either uncontrolled particle emissions or a significant fraction of coarse particles, as these
particles are known to deposit from the atmosphere at significantly higher rates than fine
particles.
4.8 Location of receptor populations - data analysis
The HEM3 system estimates ambient concentrations at the geographic centroids of census blocks
and other receptor locations specified by the user. In cases where the census block centroid was
found to be located on facility property (as determined from satellite imagery) the receptor is
moved to the nearest off-site location. The model accounts for the effects of multiple facilities
when estimating concentration impacts at each block centroid. In RTR risk assessments, we
combine only the impacts of facilities within the same source category, and assess chronic
exposure and risk only for census blocks with at least one resident {i.e., locations where people
may reasonably be assumed to reside rather than receptor points at the fenceline of a facility).
Chronic ambient concentrations are calculated as the annual average of all estimated short-term
(one-hour) concentrations at each block centroid. Possible future residential use of currently
uninhabited areas is not considered. Census blocks, the finest resolution available in the census
data, are typically comprised of approximately 40 people or about ten households.
Despite comments to the contrary on several residual risk rule proposals, we do not expect the
use of census block centroids as receptors for chronic exposure to introduce a low bias into the
risk assessment. However, we acknowledge that it does introduce uncertainty because the
highest residential exposure (assumed to be the residence nearest the facility in this analysis)
Page 4-14
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
may be either greater or less than the exposure at the block centroid. Large discrepancies are less
likely in highly populated areas because census blocks are typically small in such areas. In less-
populated areas census blocks are typically large, and it is possible that exposures at some
residences could vary substantially from those at the census block centroid.
To test for possible systematic bias associated with the use of census block centroids as receptors
for chronic exposure, we compared the estimated cancer MIR values at the census block centroid
and at the nearest residence for the 21 petroleum refinery facilities with cancer MIR values
greater than or equal to 10 in a million (Appendix M). We chose to refine the risk estimates for
these facilities to ensure that we captured the MIR for the source category, and we contend that
these facilities are representative of the entire source category with respect to the relative
difference in risk estimates between the census block centroid and nearest residence. Because
risk estimates are highly sensitive to the distance from source to receptor, we considered the
possibility that by selecting the facilities with the highest estimated risk, we were biasing our
sample with cases where the distance to the census block centroid is small. Larger distances
from source to census block centroid typically mean that the census block is large and more
likely to result in larger differences in estimated risk across the block. To determine if we were
biasing our sample with cases of small distance to receptor, we calculated the median and 95th
percentile values of distance from the source to the nearest census block centroid (at least one
person in the block) for the sample of 21 facilities and the entire source category of 150 facilities.
The median distance for the sample was 170 m compared to 190 m for the entire source
category. The 95th percentile distance for the sample was 530 m compared to 550 m for the
entire source category. These small differences indicate that the census blocks near facilities in
the source category are not significantly larger than those in the sample. Therefore, we believe
the sample is likely representative of the relative difference in risk estimates between census
block centroid and the nearest residence.
In eleven cases, the census blocks were small, with a typical distance from the centroid to the
block boundary less than 100 m. In these cases, we estimate that the MIR values at the census
block centroid and nearest residence are identical. There were two cases where census blocks
were relatively large, but for which the residences were located near the centroid. In these cases,
we also estimate that the MIR values at the census block centroid and nearest residence are
identical. In the remaining eight cases, the census blocks were relatively large, and the MIR
values at the centroid were higher than the values estimated at the nearest residence, with the
overestimates ranging from 40 to 2000 percent. In seven of these cases, the census blocks
overlap both facility property and adjacent residential areas. In such situations, MIR estimates at
the centroid are biased high because most of the area between the centroid and the boundary of
the block nearest the facility is not residential.
In summary, in this analysis of facility-specific MIR values, the centroid-generated values
overestimated the residence-generated values by 40 to 2000 percent in less than half the cases,
were equivalent in over half the cases, and there were no cases where the value at a residence
exceeded that at the centroid of the census block containing the residence. The MIR estimate for
the source category as a whole was the same using either methodology. While it is possible that
exposures at a residence in a large census block could be higher than at the centroid of the block,
this analysis supports the use of the centroid as a reasonable representation of the MIR for the
Page 4-15
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
nearest receptor, and it provides strong evidence that the use of the centroid is not creating a low
bias in the overall risk results, indicating, in fact, the tendency for this approach to overestimate
MIR values for the highest risk sources, and thus the MIR for the source category as a whole.
4.9 Population mobility - data analysis
The practice of omitting long-term population mobility does not bias the estimate of the
theoretical MIR, nor does it affect the estimate of cancer incidence since the total population
number remains the same. It does, however, affect the shape of the distribution of individual
risks across the affected population, shifting it toward higher estimated individual risks at the
upper end and reducing the number of people estimated to be at lower risks, thereby biasing the
risk estimates high. Therefore, although our initial refined assessments did not address long-
term activity {i.e., migration or population growth trends over 70 years), we applied an example
ex post facto adjustment for long-term population mobility to the estimates of lifetime cancer risk
for both case studies, using residence time and emigration data from the Bureau of Census
describing long-term migration patterns in the US.
As shown in Table 4-5 below, modeling long-term migration behavior can substantially reduce
the numbers of people with lifetime cancer risks above specific levels. This is offset, however,
by an increase in the number of people at lower levels of risk {i.e., those who move into an area
to replace those who leave). The estimate for total cancer incidence remains unchanged for
carcinogens that have linear low-dose relationships. Details of this mobility analysis are
provided in Appendix N.
Table 4-5. Results of adjustment of estimated inhalation cancer risk for long-term migration
behavior for two source categories.
Portland Cement
Petroleum R
clinci'ies
Cancel" Risk
I ^adjusted
Adjusted
I nadjusled
Adjusted
> 100 in a million
0
0
0
0
> 10 in a million
125
43
4,378
2,556
> 1 in a million
5,066
2,955
430,800
292,003
4.10 Acute exposure - discussion of uncertainties
We have biased the acute screening results high, considering that they depend upon the joint
occurrence of independent factors, such as peak hourly emissions rates, worst-case meteorology
{i.e., conditions that produce the highest 1-hour concentration at any modeled location), and
human presence at the point of maximum impact. Furthermore, in cases where multiple acute
threshold values are available we have chosen the most conservative of these values, thereby
likely incorporating a high bias on estimates of potential acute health risks. In the cases where
these results indicated the potential for exceeding short-term health thresholds we have refined
our assessment by developing a better understanding of the geography of the facility relative to
potential exposure locations and refining the acute multiplier based on input from industry. We
were not able to refine these assessments to incorporate the true variability of short-term
emission rates; such data for HAP emissions are seldom available. Thus, by maintaining the
peak-to-mean emission ratio of 10, even in our refined acute assessments (absent better data), we
believe the results generally overstate the potential for acute impacts.
Page 4-16
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
4.11 Dose-response assessment - discussion of uncertainties
4.11.1 Chronic dose-response
This assessment used a single set of toxicological dose-response values, typically extrapolated
from high-dose animal exposure or human occupational exposures, to estimate risk. These dose-
response values include embedded default exposure assumptions such as inhalation rate and
body mass (e.g., 70 kg), and do not explicitly take into account inter-individual variability in
health status or genetic makeup. Additional uncertainty arises from extrapolating from animals
to humans, high-level shorter-term exposures to lower-level long-term exposures and from
occupational exposures in healthy adult workers to environmental exposures to sensitive
individuals or life stages. Some of these factors may result in overestimates of risk and others in
underestimates, but in general the development of these dose-response values incorporates
modeling approaches that are biased toward overestimating rather than underestimating risk.
Consistent with EPA guidance, RfCs are developed by using quantitative factors to account for
uncertainties in developing values protective of sensitive subpopulations. The degree of
aggregate uncertainty would depend on the individual HAP.
Most of the UREs in this assessment were developed using linear low-dose extrapolation. Risks
would be overestimated if the true dose-response relationship (which is usually unknown) is
sublinear and underestimated when the dose-response curve is actually supralinear. In addition,
the extrapolation for most of the carcinogenic HAPs began with a statistical lower-bound (i.e.,
protective) estimate of the lowest tumorigenic dose, rather than the central estimate. The
exception to this is the URE for benzene, which is considered to cover a range of values
considered to be equally plausible, and which is based on maximum likelihood estimates. The
impact of selecting either end of the benzene URE range is discussed explicitly in section 2.3.3.
Extrapolation from a lower statistical limit tends to overestimate risks for carcinogens with
sparse health effects data, with the degree of overestimation decreasing as health effects data
become more robust. In general, EPA considers most UREs to be upper-bound estimates based
on the method of extrapolation, meaning they represent a plausible upper limit to the true value.
(Note that this is usually not a true statistical confidence limit.) The true risk is generally likely to
be less, could be as low as zero, but also could be greater. EPA's upper bound estimates
represent a "plausible upper limit to the true value of a quantity" (although this is usually not a
true statistical confidence limit).39 In some circumstances, the true risk could be as low as zero;
however, in other circumstances the risk could also be greater.40 When developing an upper
bound estimate of risk and to provide risk values that do not underestimate risk, EPA generally
relies on conservative default approaches.41
39 IRIS glossary (www.epa.gov/NCEA/iris/help_gloss.htm).
40 The exception to this is the URE for benzene, which is considered to cover a range of values, each end of which is
considered to be equally plausible, and which is based on maximum likelihood estimates.
41 cc
According to the NRC report Science and Judgment in Risk Assessment (NRC, 1994) [Default] options are
generic approaches, based on general scientific knowledge and policy judgment, that are applied to various elements
of the risk-assessment process when the correct scientific model is unknown or uncertain." The 1983 NRC report
Risk Assessment in the Federal Government: Managing the Process defined default option as "the option chosen on
the basis of risk assessment policy that appears to be the best choice in the absence of data to the contrary" (NRC,
1983a, p. 63). Therefore, default options are not rules that bind the agency; rather, the agency may depart from them
in evaluating the risks posed by a specific substance when it believes this to be appropriate. In keeping with EPA's
Page 4-17
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
The scientific understanding of dose-response relationships for these and other chemicals is
continually evolving. For example, 28 of the HAPs in these case studies (counting PAH and
glycol ether compounds each as a single HAP) are currently under assessments or revisions
within EPA's IRIS program. In cases where IRIS currently lacks dose-response assessments,
values were taken from other sources according to a predetermined hierarchy. In the case of
benzene where a range of UREs is presented in EPA's IRIS database, we have chosen the most
conservative of these values for these assessments, favoring to err on the side of health protection
by estimating higher potential cancer risk. Since the resulting risk results were not negligible,
we have also performed a sensitivity analysis to demonstrate the impact of choosing the lower
end of the benzene URE range on MIR and estimated cancer incidence in Section 2.3.3.
4.11.2 Acute dose-response
Many of the UFs used to account for variability and uncertainty in the development of acute
reference values are similar to those developed for chronic durations, but more often using
individual UF values less than 10. UFs are applied based on chemical-specific or health effect-
specific information (e.g., simple irritation effects do not vary appreciably between human
individuals, hence a value of 3 is typically used), or based on the purpose for the reference value
(see the following paragraph). The UFs applied in acute reference value derivation include: 1)
heterogeneity among humans; 2) uncertainty in extrapolating from animals to humans; 3)
uncertainty in LOAEL to NOAEL adjustments; and 4) uncertainty in accounting for an
incomplete database on toxic effects of potential concern. Additional adjustments are often
applied to account for uncertainty in extrapolation from observations at one exposure duration
(e.g., 4 hours) to arrive at a POD for derivation of an acute reference value at another exposure
duration (e.g., 1 hour).
Not all acute reference values are developed for the same purpose and care must be taken when
interpreting the results of an acute assessment of human health effects relative to the reference
value or values being exceeded. Where relevant to the estimated exposures, the lack of threshold
values at different levels of severity should be factored into the risk characterization as potential
uncertainties. Further, when we compare our peak 1-hour exposures against MRL values (which
are derived for 1- to 14-day exposure durations), we note that peak emission events are unlikely
to last more than an hour. As such, these comparisons are a very conservative screen which is
only useful in ruling out potential exposures of concern, limiting our ability to interpret situations
where MRL values are exceeded.
4.12 Compounds without dose-response assessments - sensitivity
analysis
Finally, many HAPs lack any dose-response values at all for cancer, chronic non-cancer and
acute effects. In some cases this reflects a relative lack of concern for the pollutant/effect in
question, but in others it may result from a lack of scientific data. This factor has the potential to
goal of protecting public health and the environment, default assumptions are used to ensure that risk to chemicals is
not underestimated (although defaults are not intended to overtly overestimate risk). See EPA 2004 An Examination
of EPA Risk Assessment Principles and Practices, EPA/100/B-04/001 available at:
http://www.epa.gov/osa/pdfs/ratf-final.pdf.
Page 4-18
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
result in an understatement of risk if there are effects associated with these HAPs at
environmental exposure levels.
In response to the SAB's previous concerns about our inability to estimate risks for HAPs that
lack peer-reviewed dose-response assessments, we conducted a "what-if' analysis based on
median and upper-bound estimates of toxic potency for these substances. Details of this analysis
are presented in Appendix O. We included in this analysis the Portland cement and petroleum
refinery source categories individually, and also all US sources combined. The analysis was
based on toxicity-weighting of the 2002 NEI, a process that provides an estimate of relative
potential cancer risk and noncancer respiratory hazard posed by each HAP. We weighted the
pollutant emissions as follows: (1) for noncancer respiratory effects, the emitted amount for each
chemical was divided by its RfC or similar chronic no-effect exposure level; (2) for cancer, the
emitted amount of each chemical was multiplied by its inhalation URE for cancer.
For HAPs that lacked an RfC or URE, we selected as surrogates the following range of values
selected from the universe of chronic RfCs and UREs in the OAQPS table of prioritized chronic
dose-response values for inhalation exposure (http://www.epa.gov/ttn/atw/toxsource/tablel.pdf):
Percentile of
RfC
i r i :
toxicity
(mu in')
(1 /ju m
5
2.28
1.0e-6
25
0.2
6.0e-6
50
0.0098
6.8e-5
75
0.00065
6.1e-4
95
0.000023
4.8e-2
All HAPs lacking an RfC were assigned this range of surrogate RfCs. Only HAPs lacking a URE
but having an EPA or IARC WOE equivalent to "possible carcinogen" or greater were assigned
the range of surrogate UREs. Toxicity-weighted emissions (TWEs) for cancer and noncancer
effects were kept separate. TWE's were normalized by dividing each score by the maximum
TWE from all chemicals that had a dose-response value.
The following compounds produced TWEs suggesting that, if their toxicity or carcinogenic
potency were found to be at the high end of the surrogate ranges, they could contribute
substantially43 to total risk:
Source caleuoiv
\oiicaivinouens
Caivinouens
Petroleum refineries
2,2,4-trimethylpentane
POMs
Biphenyl
carbonyl sulfide
chromium III
quinoline
42 Low RfCs connote high toxicity, so the RfC decreases as toxicity increases. UREs are directly proportional to
carcinogenic potency, so the URE increases as potency increases.
43 "Substantially" for this table is the 95th percentile TWE for the HAP exceeding 10% of the largest TWE for the
sources.
Page 4-19
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Source caleuorv
\oncaivinouens
Caivinouens
Portland cement facilities
carbonyl sulfide
POM
1,3-propane sultone
chromium III
bromoform
none
All NEI sources
2,2,4-trimethylpentane
carbonyl sulfide
POM
propionaldehyde
ethyl aery late
This toxicity-weighting analysis, while obviously simplistic, is nevertheless useful for
determining whether particular assessments have overlooked any potentially important
unassessed chemicals, and for informing decisions prioritizing pollutants for toxicity testing and
dose-response assessment. Similar analyses can be conducted easily on other source categories,
and with other inventory years, to identify new candidates.
Page 4-20
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
5 References
1. Morgan, G. and R. Henderson, 2007. Consultation on EPA's Risk and Technology Review (RTR)
Assessment Plan. Letter to EPA Administrator Johnson. EPA-SAB-07-009.
http://vosemite.epa.gov/sab/sabproduct.nsf/c91996cd39a82f64852574240069Q127/33152C83
D29530F08525730D006C3ABF/$File/sab-07-009.pdf
2. US EPA, 2006. Emission Standards for Hazardous Air Pollutants.
http://www.epa.gov/ttn/atw/mactfnlalph.html.
3. US EPA, 2005. Revision to the Guideline on Air Quality Models: Adoption of a Preferred General
Purpose (Flat and Complex Terrain) Dispersion Model and Other Revisions; Final Rule. 40
CFRPart 51. http://www.epa.gov/EPA-AIR/2005/November/Day-09/a21627.htm
4. US EPA, 2004. Users'guide for the AMS/EPA regulatory model - AERMOD. EPA-454/B-03-
001. http ://www.epa. gov/scramOO l/7thconf/aermod/aermodugb .pdf.
5. Allen, D., C. Murphy, Y. Kimura, W. Vizuete, T. Edgar, H. Jeffries, B.-U. Kim, M. Webster, and
M. Symons, 2004. Variable industrial VOC emissions and their impact on ozone formation in
the Houston Galveston Area. Final Report: Texas Environmental Research Consortium
Project H-l 3.
http://files.harc.edu/Proi ects/AirQualitv/Proi ects/HO 13,2003/Hl 3FinalReport.pdf.
6. US EPA, 2004. Air Toxics Risk Assessment Reference Library, Volume 1. EPA-453-K-04-
001A. http://www.epa.gov/ttn/fera/risk atra voll.html.
7. US EPA, 2006. Human Health Risk Assessment Protocol (HHRAP) for Hazardous Waste
Combustion Facilities, Final.
http://www.epa.gOv/epawaste/hazard/tsd/td/combust/risk.htm#hhrad
8. US EPA, 2005. Table 1. Prioritized Chronic Dose-Response Values (2/28/05). Office of Air
Quality Planning and Standards, http://www.epa.gov/ttn/atw/toxsource/table 1 .pdf
9. US EPA, 2005. 1999 National Air Toxics Risk Assessment.
http://www.epa. gov/ttn/atw/natal 999.
10. US EPA, 2006. Integrated Risk Information System, http J/www, epa. gov/iri s/index. html.
11. US Environmental Protection Agency. 1997. Health Effects Assessment Summary Tables
(HEAST). http://epa-heast.ornl. gov/heast/index.html.
12. California Office of Environmental Health Hazard Assessment, 2000. Air Toxics Hot Spots
Program, Risk Assessment Guidelines, Part in - Technical Support Document for the
Determination of Noncancer Chronic Reference Exposure Levels.
http://www.oehha.ca.gov/air/chronic rels/pdf/relsP32k.pdf
Page 5-1
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
13. US EPA, 1994. U.S. Environmental Protection Agency. 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.
14. NRC, 1994. National Research Council. Science and Judgment in Risk Assessment. Washington,
DC: National Academy Press.
15. California Office of Environmental Health Hazard Assessment, 2005. Chronic Reference
Exposure Levels Adopted by OEHHA as of February 2005.
http://www.oehha.ca.gov/air/chronic rels/AllChrels.html.
16. California Office of Environmental Health Hazard Assessment, 2005. Technical Support
Document for Describing Available Cancer Potency Factors, May 2005.
http://www.oehha.ca.gov/air/hot spots/mav2005tsd.html.
17. US Agency for Toxic Substances and Disease Registry. 2006. Minimum Risk Levels (MRLs)
for Hazardous Substances, http://www.atsdr.cdc.gov/mrls.html.
18. US EPA, 2006. Approach for modeling POM. Technical support information for the 1999
National Air Toxics Assessment.
http J/www, epa. gov/ttn/atw/natal 999/99pdfs/pomapproachi an.pdf.
19. US EPA, 2005. Table 2. Acute Dose-Response Values for Screening Risk Assessments
(6/02/2005). Office of Air Quality Planning and Standards.
http://www.epa.gov/ttn/atw/toxsource/table2.pdf
20. California Office of Environmental Health Hazard Assessment, 2000. All Acute Reference
Exposure Levels developed by OEHHA as of May 2000.
http J/www, oehha. ca. gov/ air/acute rels/all AcRELs .html.
21. American Industrial Hygiene Association, 2008. Current AIHA ERPG Values.
http://www.aiha.org/ldocuments/Committees/ERP-erpglevels.pdf
22. US EPA, 1995. Guidance for Risk Characterization. Science Policy Council.
http://www.epa.gov/OSA/spc/pdfs/rcguide.pdf.
23. US EPA, 2000. Risk Characterization Handbook. EPA 100-B-00-002.
24. US EPA, 2002. EPA's Guidelines for Ensuring and Maximizing the Quality, Objectivity, Utility,
and Integrity of Information Disseminated by the Environmental Protection Agency. EPA
Office of Environmental Information. EPA/260R-02-008.
http://www.epa.gov/qualitv/informationguidelines/documents/EPA InfoOualitvGuidelines.pd
f
25. US EPA, 2006. Performing risk assessments that include carcinogens described in the
Supplemental Guidance as having a mutagenic mode of action. Science Policy Council
Page 5-2
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
Cancer Guidelines Implementation Workgroup Communication II: Memo from W.H. Farland
dated 14 June 2006. http://epa.gov/osa/spc/pdfs/CGIWGCommunication Il.pdf.
26. US EPA, 2005. Supplemental guidance for assessing early-life exposure to carcinogens.
EPA/630/R-03003F. http://www.epa.gov/ttn/atw/childrens supplement final.pdf.
27. US EPA, 2005. Science Policy Council Cancer Guidelines Implementation Workgroup
Communication I: Memo from W.H. Farland dated 4 October 2005 to Science Policy
Council, http://www.epa.gov/osa/spc/pdfs/canguidl.pdf
28. US EPA, 1986. Guidelines for the Health Risk Assessment of Chemical Mixtures. EPA/630/R-
98/002. http://www.epa.gov/ncea/raf/pdfs/chem mix/chem mix 1986.pdf.
29. US EPA, 2000. Supplementary Guidance for Conducting Health Risk Assessment of Chemical
Mixtures. EPA/63 0/R-00/002.
http://www.epa.gov/ncea/raf/pdfs/chem mix/chem mix 08 2001.pdf.
30. Alberta Research Council Inc. "Refinery Demonstration of Optical Technologies for
Measurement of Fugitive Emissions and for Leak Detection" Prepared for Environment
Canada. March 26, 2006.
31. USEPA. "VOC Fugitive Losses: New Monitors, Emission Losses, and Potential Policy Gaps"
2006 International Workshop. Available at
http://www.epa.gov/ttn/chief/efpac/documents/wkshop vocfugemissions.pdf.
32. USEPA, 2004. Air Toxics Risk Assessment Reference Library, Volume 1. EPA-453-K-04-
001A. http://www.epa.gov/ttn/fera/risk atra voll.html.
33. US EPA, 2005. Background Documentation - TRIM Ecological Toxicity Database (September
2005 version), http://www.epa.gov/ttn/fera/trim risk dowm.html.
34. Canadian Council of Ministries of the Environment, 2003. Updat,e, Canadian Environmental
Quality Guidelines, http J/www, ec.gc.ca/ ceq g-rcq e/english/ceq g/ defavlt. cfm
35. US EPA, 2005. Guidelines for Carcinogen Risk Assessment (2005). U.S. Environmental
Protection Agency, Washington, DC, EPA/630/P-03/001F, 2005.
http ://cfpub. epa.gov/ncea/raf/recordisplav. cfm?deid=l 1628.
36. National Research Council, 2006. Assessing the Human Health Risks of Trichloroethylene.
National Academies Press, Washington DC.
37. Subramaniam, R.P., P. While and V. J. Cogliano. 2006. Comparison of Cancer Slope Factors
Using Different Statistical Approaches. Risk Analysis, Vol. 26, No. 3, pp. 825-830, June
2006. Available at SSRN: http://ssrn.com/abstract=943254 orDOI: 10.1111/j. 1539-
6924.2006.00769.x
Page 5-3
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
38. US EPA, 2006. An Inventory of Sources and Environmental Releases of Dioxin-Like
Compounds in the United States for the Years 1987, 1995, and 2000. National Center for
Environmental Assessment, Office of Research and Development.
http://oaspub.epa.gov/eims/eimscomm.getfile7p download id=459709.
39. Chen, Q.; Degrange, J.P; Gerchikov, M.Y.; Hillis, Z.K.; Lepicard, S.; Meijne, E.I.M.; Smith,
K.R.; and van Weers, A. 2003. Effluent and Dose Control from European Union NORM
Industries, Assessment of Current Situation and Proposal for a Harmonised Community
Approach. Volume 1: Main Report. European Commission.
40. US EPA, 2003. TRIM.FaTE User's Guide, Module 5. Accessed March 15, 2008. Available at:
http://www.epa.gOv/ttn/fera/trim_fate.html#current_user.
41. National Resource Conservation Service (NRCS), 2007. Revised Universal Soil Loss Equation,
Version 2 (RUSLE2). Last updated July 11, 2007.
42. McKone, T.E., A. Bodnar, and E. Hertwich. 2001. Development and evaluation of state-specific
landscape data sets for multimedia source-to-dose models. University of California at
Berkeley. Supported by U.S. Environmental Protection Agency (Sustainable Technology
Division, National Risk Management Research Laboratory) and Environmental Defense
Fund. July. LBNL-43722.
43. National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center
(NCDC), 2001. The FCC Integrated Surface Hourly Database, A New Resource of Global
Climate Data. Technical Report 2001-01. Nov 2001. Available at:
http://wwwl .ncdc.noaa.gOv/pub//data/techrpts/tr200101/tr2001-01.pdf
44. US EPA, 1997. Exposure Factors Handbook. National Center for Environmental Assessment.
EPA/600/P-95/002Fa, August, 1997. Available on-line at http://www.epa.gov/ncea/efh/.
45. U.S. Environmental Protection Agency (EPA). 2008. Child-Specific Exposure Factors
Handbook. Office of Research and Development, Washington, D.C. EPA/600/R-06/096F.
September, 2008. Available on-line at:
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid= 199243 .
46. US Environmental Protection Agency (EPA). 1980. Acquisition and chemical analysis of
mother's milk for selected toxic substances. Washington, DC. EPA-560/13-80-029.
December.
47. US Environmental Protection Agency (EPA). 1983. Chemicals identified in human breast milk, a
literature search. Washington, DC. EPA-560/5-83-009. October.
48. World Health Organization (WHO). 1985. The quantity and quality of breast milk. Report on the
WHO Collaborative Study on Breast-feeding. Geneva.
Page 5-4
-------
June 15,2009
RTR Risk Assessment Methods for SAB Review
49. World Health Organization (WHO). 1989. Minor and trace elements in breast milk. Report of a
joint WHO/IAEA Collaborative Study. Geneva.
50. EPA, 2004. Risk Assessment Guidance for Superfund Volume I: Human Health Evaluation
Manual (Part E, Supplemental Guidance for Dermal Risk Assessment). Office of Superfund
Remediation and Technology Innovation. Solid Waste and Emergency Response.
EPA/540/R/99/005; OSWER 9285.7-02EP; PB99-963312. July, 2004.
51. Ten Berge, W.F., A. Zwart, and L.M. Applebaum, 1986. Concentration-time mortality response
relationship of irritant and systematically acting vapours and gases. Journal of Hazardous
Materials. 13(3):301-309.
52. Allen, D., C. Murphy, Y. Kimura, W. Vizuete, T. Edgar, H. Jeffries, B.-U. Kim, M. Webster, and
M. Symons, 2004. Variable industrial VOC emissions and their impact on ozone formation in
the Houston Galveston Area. Final Report: Texas Environmental Research Consortium
Project H-l 3.
http://files.harc.edu/Proi ects/AirQualitv/Proi ects/HO 13,2003/Hl 3FinalReport.pdf.
53. RTI, 2002. Petroleum Refinery Source Characterization and Emission Model for Residual
Risk Assessment. Prepared for U.S. Environmental Protection Agency, Office of Air
Quality Planning and Standards, Research Triangle Park, NC. EPA Contract No. 68-D6-
0014. July 2, 2002.
54. Lucas, B. 2007. Memorandum from B. Lucas, EPA/SPPD, to Project Docket File (EPA
Docket No. EPA-HQ-OAR-2003 -0146). Collection of Detailed Benzene Emissions Data
from 22 Petroleum Refineries. August 20, 2007. Docket Item No. EPA-HQ-OAR-2003-
0146-0015.
55. Spicer, C.W., Gordon, S. M., Holdren, M.W., Kelly, T.J.Mukund, R. 2002. Hazardous Air
Pollutant Handbook: Measurements, Properties, and Fate in Ambient Air. CRC Press.
Page 5-5
-------
Appendix A: Comparison of initial risk estimates with risk estimates
refined by public comment for petroleum refineries
A-l
-------
Appendix A
This appendix presents the evolution of the data set used for the NPRM petroleum refinery
source category risk assessment and the amount that the estimated cancer risks changed as a
result of public comment1. This discussion includes comparisons of the maximum individual
cancer risks (MIR), cancer incidence and population exposure, HAP emissions, and toxicity-
weighted HAP emissions. In addition to a brief background, a summary of the types of changes
received for the petroleum refinery data set through the ANPRM process is also presented.
A.1 Background
A screening risk assessment was conducted for the ANPRM in September 2006. After receipt of
data revisions through the ANPRM process, a risk assessment was conducted for the NPRM in
July 2007. The HEM3 model (with AERMOD) was used for both assessments; however, several
updates to the model were made during the intervening time. Updates mainly involved the
meteorological station data library and included: the library was expanded to include additional
stations, the data was processed using a newer version of AERMET, and data was obtained for
newer years than previously used. In addition, the HAP library of dose-response values was
updated between the ANPRM and NPRM.
A.2 Summary of Data Revisions Received Through ANPRM Process
The ANPRM data set for the petroleum refinery source category included 175 facilities.
Through the ANPRM process, data changes or revisions were received for 113, or 65 percent, of
the facilities. Changes to the data were supplied by EPA, State or local agencies, trade
organizations, and/or facilities themselves. Types of changes to the data included data
replacement, emissions changes, process changes, emission release point changes, and facility
changes. Data replacement changes were those where the commenter could not match the
existing NEI data with new data they wished to provide and instead provided a complete
replacement of the entire petroleum refinery NEI data set for that facility. Emissions changes
were those related to the emissions estimates, such as an update to emissions estimates or the
removal or addition of HAP from an existing emission point. Process changes were those
changes that added or removed a process at the facility from the refinery source category, e.g., by
changing the MACT code. Emission release point changes included corrections to location
coordinates and updated stack parameters. Finally, facility changes included changes in name,
ownership, or status of the facility to major or area. A listing of these changes received for the
petroleum refinery source category can be found in the RTR docket (EPA-HQ-OAR-2006-0859),
item number 0261. In addition to those changes described above, 30 facilities were removed and
8 facilities were added by EPA after the screening risk assessment was conducted, resulting in
153 facilities in the NPRM data set. A summary of the frequency and types of changes made to
the ANPRM data set are shown in Figure 1. As can be seen in Figure 1, the most frequent types
of data changes were changes to emissions, followed by changes to emission release point
information, process changes, and the complete removal or addition of facilities from the data
set.
1 In other words, this appendix compares two drafts of the Petroleum Refineries baseline risk assessment, before and
after public comment. Subsequent changes made to the NPRM draft assessment to create the final baseline
assessment are not discussed in this appendix.
A-2
-------
Figure 1. Frequency and Type of Changes Made to ANPRM Data Set for Petroleum Refineries
60
50
48
40
41
39
38
* 5 30
20 -
10
Data Replacement Emissions Changes Process Changes Emission Release
Changes Point Changes
Type of Change Made to ANPRM Data Set
Facility Changes
Complete
Removal/Addition
A.3 Discussion of Emissions Changes and Toxicity-Weighted Emissions
Changes
As mentioned above, changes to the emissions estimates were the predominant type of change
received through the ANPRM process for petroleum refineries. Emission changes included
adding, deleting, or revising the emissions estimates for a specific point. Process changes could
also have affected the emissions estimates for a facility. For example, if a process change
removed an entire process from the data set, this would reduce the total emissions estimates for
that facility. The total HAP emissions included in the ANPRM data set equal 2,316 tons per year
(tpy), and the total HAP emissions included in the NPRM data set equal 2,292 tpy. Therefore,
there was a reduction in overall HAP emissions of 24 tpy.
A summary of emissions by HAP for the ANPRM and NPRM data sets for petroleum refineries
is presented in Figure 2. As can be seen in Figure 2, emissions of toluene (18 percent), xylenes
(17 percent), hexane (17 percent), and benzene (14 percent) made up 66 percent of the emissions
in the ANPRM data set. In the NPRM data set, the same four HAP make up 76 percent of the
total emissions in the data set and have different individual percentages: xylenes (26 percent),
toluene (21 percent), hexane (18 percent), and benzene (11 percent). However, while the overall
percentages of these four HAP increased, the mass of emissions of each of these HAP decreased
in the NPRM data set compared to the ANPRM data set.
A-3
-------
Figure 2. Percentage of Emissions by HAP for the Petroleum Refinery ANPRM and NPRM Data Sets
ANPRM Data Set NPRM Data Set
Toluene
18%
Others
26%
Xylenes
(mixed)
17%
Methanol
Ethyl benzene
n-Hexane
17%
Xylenes
(mixed)
26%
cthano1
.4%
Toluene
21%
n-Hexane
18%
Others
15%
A-4
-------
Emissions changes between the ANPRM and NPRM data sets were also evaluated by comparing
the change in cancer toxicity-weighted emissions. Toxicity-weighted emissions take into
account both the mass of the HAP emitted and its carcinogenic potency. The total emitted mass
for each HAP was multiplied by its respective cancer unit risk estimate (URE) value in order to
calculate cancer toxicity-weighted emissions. Overall, total toxicity-weighted emissions
decreased by 12 percent from the ANPRM data set to the NPRM data set. The toxicity-weighted
emissions of individual HAP generally decreased from the ANPRM data set to the NPRM data
set, however they did increase for several HAP, including naphthalene, nickel compounds, and
chromium compounds. In the ANPRM data set, 97 percent of the toxicity-weighted emissions
can be accounted for by the following HAP: benzene, naphthalene, ethylene dibromide,
POM71002, 1,3-butadiene, and POM72002. (As described in the approach for the 1999 NATA
analysis, the name "POM71002" is used to represent the following compounds: 7-PAH, total
PAH, poly cyclic organic matter, 16-PAH, and 16 PAH-7 PAH. The name "POM72002"
represents numerous compounds, including: anthracene, pyrene, benzo(g,h,i)perylene, perylene,
fluoranthene, benzofluoranthenes, acenaphthene, phenanthrene, and fluorene.) These same six
HAPs, plus nickel compounds and chromium compounds, account for almost 99 percent of the
toxicity-weighted emissions in the NPRM data set. With the exception of benzene, none of these
HAPs are those with the highest magnitude of emissions, as shown in Figure 2.
To illustrate the change in toxicity-weighted emissions between the ANPRM and NPRM data
sets, we calculated the relative percentage that each HAP in the ANPRM and NPRM data sets
contributed to the total ANPRM toxicity-weighted emissions. This comparison is presented in
Figure 3, for those HAPs that contribute 1 percent or greater of the total toxicity-weighted
ANPRM emissions. As shown in Figure 3, the toxicity-weighted ANPRM benzene emissions
account for approximately 40 percent of the total ANPRM toxicity-weighted emissions, while
the toxicity-weighted NPRM benzene emissions would relatively account for approximately 32
percent of the total ANPRM toxicity-weighted emissions. This is associated with a decrease of
64 tpy from the ANPRM to NPRM data set. Nickel compounds, on the other hand, saw a 0.4 tpy
increase in emissions from the ANPRM to the NPRM data set. As shown in Figure 3, the
toxicity-weighted ANPRM nickel compound emissions account for less than half a percent of the
total ANPRM toxicity-weighted emissions, while the toxicity-weighted NPRM nickel compound
emissions would relatively contribute slightly over 1 percent of the total ANPRM toxicity-
weighted emissions.
A-5
-------
Figure 3. Cancer Toxicity-weighted Emissions for the Petroleum Refinery ANPRM and NPRM
Source Category Data Sets
Benzene Naphthalene Ethylene POM 71002 1,3-Butadiene POM 72002 Nickel compounds
di bromide
HAP
~ ANPRM Tox-weighted Emissions
~ NPRM Tox-weighted Emissions
as a Percentage of the Total ANPRM
as a Percentage of the Total ANPRM
Tox-Weighted Emissions
Tox-Weighted Emissions
A.3.1 Comparison of ANPRM and NPRM Maximum Individual Cancer Risks
Figure 4 shows the comparison between the facility-level maximum individual cancer risk results
for the ANPRM and NPRM by providing the percentage of facilities at each cancer risk level.
As shown in Figure 4, there are no facilities in the NPRM data set with a cancer risk greater than
or equal to 100 in 1 million but three percent (5 facilities) of the facilities in the ANPRM data set
have a cancer risk greater than 100 in 1 million. There are also more facilities with a maximum
individual cancer risk estimate greater than 10 in 1 million but less than 100 in 1 million in the
ANPRM data set (19 percent or 33 facilities), compared to the NPRM data set (11 percent or
18 facilities).
A-6
-------
Figure 4. Percentage of Facilities per Cancer Risk Level Petroleum Refinery ANPRM vs.
NPRM Data Sets
60%
50%
48%
44%
^ r
40%
34%
30%
19%
20%
10%
0%
0%
MIR greater than or equal to 100 MIR greater than or equal to 10 in MIR greater than or equal to 1 in MIR less than 1 in 1 million (1E-6)
in 1 million (1E-4) 1 million (1E-5) and less than 100 1 million (1E-6) and less than 10
in 1 million (1E-4) in 1 million (1E-5)
Cancer Risk Level
~ ANPRM (175 facilities) El NPRM (153 facilities) |
Another comparison between the risk results from the ANPRM and NPRM data sets is provided
in Figure 5. In Figure 5, the ANPRM and NPRM maximum individual lifetime cancer risk for
each facility is plotted as a point and shown compared to the y = x line. The points below the
line indicate that the ANPRM risk was higher than the NPRM risk. The basic trend, shown in all
the figures, is that facilities generally have lower cancer risks after the ANPRM than before.
A-7
-------
Figure 5. NPRM Cancer Risk vs. ANPRM Cancer Risk for Petroleum Refinery Data Sets
1.00E-03
1.00E-04
1.00E-05
~ ~
a)
_i
•g 1.00E-06
E
-------
Figure 6. Percentage of Facilities (per Cancer Risk Level) With Changes to Petroleum
Refinery AN PRM Data Set
120%
100%
80%
= 60%
* 40%
20%
0%
100%
67%
57%
55%
MIR greater than or equal to 100 in 1
million (1E-4)
MIR greater than or equal to 10 in 1
million (1E-5) and less than 100 in 1
million (1E-4)
MIR greater than or equal to 1 in 1 million
(1E-6) and less than 10 in 1 million (1E-5)
MIR less than 1 in 1 million (1E-6)
Cancer Risk Level
After changes were made for 100 percent of the facilities with a cancer risk level greater than or
equal to 100 in 1 million, no facilities had risks at this level in the NPRM. However, while this
trend does continue, it is less pronounced at lower risk levels. As shown in Figure 6, while 67
percent of the facilities with an ANPRM cancer risk greater than or equal to 10 in 1 million but
less than 100 in 1 million provided data changes, there was only an 8 percent decrease in the
facilities at this risk level. The magnitude of change between the ANPRM and NPRM cancer
risk values are shown in Figure 7. (This figure only includes data for those facilities present in
both the ANPRM and NPRM data sets.) As shown in Figure 7, the NPRM cancer risk decreased
by over 50 percent from the ANPRM cancer risk for 41 percent of the facilities, while the NPRM
cancer risk changed less than 1 percent for only 3 percent of the facilities. The same information
is presented in Figure 8.
A-9
-------
Figure 7. Risk Impacts for the Facilities with Changes to Petroleum Refinery ANPRM Data Set
(not including those facilities that were completely added or removed)
45%
40%
35% -
i/i S)
.2 u> 30%
25% -
20% -
10%
5%
0%
41%
14%
14%
14%
11%
3%
3%
NPRM risk
decreased by
greater than 50%
from ANPRM risk
NPRM risk
decreased by 25
to 50% from
ANPRM risk
NPRM risk
decreased by 1 to
25% from ANPRM
risk
NPRM risk
decreased or
increased less
than 1 % from
ANPRM risk
NPRM risk NPRM risk NPRM risk
increased by 1 to increased by 25 to increased by
25% from ANPRM 50% from ANPRM greater than 50%
risk risk from ANPRM risk
Percent Difference in Cancer Risk from ANPRM to NPRM
Figure 8. Risk Impacts for the Facilities with Changes to Petroleum Refinery ANPRM Data Set
(not including those facilities that were completely added or removed)
NPRM risk increased by
greater than 50% from
ANPRM risk
14%
NPRM risk increased by 25 to
50% from ANPRM risk
3%
NPRM risk increased by 1 to
25% from ANPRM risk
14%
NPRM risk decreased by
greater than 50% from
ANPRM risk
41%
NPRM risk decreased or
increased less than 1 % from
ANPRM risk
3%
NPRM risk decreased by 1 to
25% from ANPRM risk
14%
NPRM risk decreased by 25
to 50% from ANPRM risk
11%
A-10
-------
A.3.2 Comparison ofANPRM and NPRM Cancer Incidence Values and
Population Exposure
The estimated cancer incidence value resulting from modeling the ANPRM data set is 0.08. The
estimated cancer incidence value resulting from modeling the NPRM data set is 0.05. The
incidence value is calculated for each Census block by multiplying the estimated maximum
individual lifetime cancer risk for that block by the number of people in that block and then
dividing by the estimated lifespan value of 70 years. The values for each Census block are then
summed to create a category-level incidence value. Where multiple facilities impact more than
one Census block, this is taken into account before the summation so that people are not counted
more than once. The change in the maximum individual risk estimate at each Census block, and
therefore also the change in population affected between the ANPRM and NPRM, can be seen in
Figure 9. As shown in Figure 9, there are fewer people exposed to all risk levels using the
NPRM data set.
Figure 9. Comparison of Cumulative Populations per Cancer Risk Level for the Petroleum
Refinery ANPRM and NPRM Data Sets
Maximum Individual Lifetime Cancer Risk
1.00E-03 1.00E-04 1.00E-05 1.00E-06 1.00E-07 1.00E-08 1.00E-09 1.00E-10 1.00E-11 1.00E-12 1.00E-13 1.00E-14 1.00E-15
1 i i i i i i i i i i i 1
10
oo
cn
100
~
~
o
1,000
<***>000^
Q.
o
Q_
100,000,000
10,000,000
1,000,000
100,000
10,000
o ANPRM Cumulative Population DNPRM Cumulative Population
A-l 1
-------
Appendix B
Analysis of Data on Short-Term Emission Rates
Relative to Long-Term Emission Rates
-------
Ted Palma
Roy Smith
EPA/OAQPS/SBAG
B1. Introduction
Bl.l. The problem
The process of listing hazardous air pollutants (HAPs) provided by the Clean Air Act (CAA,
section 112(b)(2)) explicitly includes acute toxicity as a listing criterion. For this reason, in
addition to chronic exposures, EPA considers acute exposures in risk-based decision-making for
the HAP regulatory program. Estimating acute exposures via dispersion modeling requires input
data on hourly meteorological conditions (available for most areas of the US) and short-term
emission rates of individual facilities (almost universally absent from the National Emissions
Inventory (NEI), the Toxic Release Inventory (TRI), and state emission databases).
Lacking short-term emission rates, we must estimate peak short-term rates based on annual
average rates, which are available. For Risk and Technology Review (RTR) rulemakings, we
have assumed that the 1-hour emission rate for each facility could exceed the annual average
hourly emission rate by as much as tenfold, and further assumed that this tenfold emission spike
could coincide with worst-case meteorological conditions and the presence of a human receptor
at the facility boundary, as a means of screening for potentially significant acute exposures.
In a consultation on the "RTR Assessment Plan", a panel of the EPA's Science Advisory Board
(SAB), several reviewers questioned the appropriateness of the factor of ten; some even
suggested that this tenfold assumption may underestimate actual maximum short-term emissions
for some facilities, and thereby also underestimate maximum acute risks. The SAB
recommended an analysis of available short-term emissions data for HAP to test this assumption.
This analysis responds to that SAB recommendation and attempts to evaluate the protectiveness
of the tenfold assumption using a database of "event emissions" collected from facilities in the
Houston-Galveston area, to compare events representative of short-term HAP releases of specific
events to long-term release rates for the entire facility. This evaluation is intended to estimate
how many short-term events might have achieved a release rate that exceeded the routine
emission rate for the entire facility during the ca. 2-year data collection period.
B2. Methods
B2.1. Texas Commission on Environmental Quality event emissions database
The Texas Commission on Environmental Quality (TCEQ) collects emissions data using online
reporting required of any facility whenever it experienced a non-routine event that released 100
pounds or more of a listed chemical (primarily ozone-forming VOCs). The TCEQ data are
intended to improve the state's knowledge of how short-term releases affect tropospheric ozone
levels in that area. The database we utilized in our analysis was a subset of the TCEQ data
covering emission events that occurred in an eight-county area in eastern Texas during a 756-day
period between January 31, 2003 and February 25, 2005.
The complete emissions event data were obtained in April 2007 from Cynthia Folsom Murphy, a
research scientist with the University of Texas at Austin (UTA) Center for Energy and
B-l
-------
Environmental Resources. The data were provided in four Excel spreadsheets generated from an
original MS Access file. We used these Excel files to reconstruct a MS Access database in order
to facilitate selection of a representative subset of records for this analysis.
Although some of the released substances were HAPs, this was incidental to the database's
primary purpose of enhancing the TCEQ's knowledge of photochemical activity. Thus, more
than 80% of the released mass was ethene and propene, neither of which are HAPs. The
database included release events caused by accidents, equipment failures, startup, shutdown, and
malfunction. It also contained facility names, information on amounts of individual compounds
released. To provide a basis for comparing the event releases with "typical" emissions, the UTA
staff included total VOC emissions data for each facility for calendar year 2004, obtained from
the EPA Toxic Release Inventory (TRI). The database did not contain any records for facilities
that did not experience any reportable events during this period.
B2.2. Data filtering
Because the event release data were intended for modeling short-term releases of ozone-
producing VOCs, the database includes releases from accidents (which are regulated under
section 112(r) of the CAA and are therefore not considered in residual risk assessments) and
releases of light hydrocarbon compounds that are not HAPs and are much more volatile than
most HAPs. This intent of our analysis, on the other hand, was to evaluate short-term releases of
HAPs due to normal process variability or scheduled startups, shutdowns, and malfunctions,
relative to long-term release rates. Because the full emission events database was not
representative of likely HAP emissions normally considered under the residual risk program, we
filtered the release data as follows in an attempt to improve its representativeness:
1. Hydrocarbons of C5 or less were dropped, except that all HAPs (including non-VOCs)
were retained regardless of volatility;
2. Releases labeled as accidental by the database were dropped, but all others (including
those labeled startup, shutdown, and malfunction) were retained;
3. Only facilities whose long-term VOC releases exceeded 0.068 tons per day (25 tons per
year) were retained, to approximate the population of facilities likely to be subject to
residual risk standards (i.e., major facilities);
4. A few release records had to be dropped because their facility numbers did not link to any
facility in the database;
5. A few facilities had to be dropped because the database did not include their 2004 TRI
VOC release information.
B2.3. Analysis
Annual VOC emissions for each facility in its entirety and release data for each event were both
converted to lb/hr. In order to conform to our atmospheric dispersion models, which estimate
ambient concentrations for periods of 1 hour or more, amounts released during events shorter
than 1 hour were assigned to the whole hour. For example, a release of 100 lb in ten minutes
was converted to 100 lb/hr. Events longer than 1 hour were converted normally, e.g., a release of
B-2
-------
100 lb in 120 minutes was converted to 50 lb/hr. The event release rates for individual
compounds were summed, yielding a total release rate for each event. This total release rate for
each event was divided by the annual VOC release rate for the entire facility to derive the ratio of
peak-to-mean emission rate for the event.
B3. Results and Discussion
B3.1. Database filtering
The original database contained 505 individual contaminants, including multiple redundancies.
These redundancies did not affect this analysis, so we did not resolve them. After filtering out
light, non-HAP VOCs, 317 contaminants remained (Table 1).
The database contained release records for 150 unique facilities. Of these, 48 facilities (Table 2)
were major VOC emitters that reported releases of at least one of the contaminants in Table 1.
The database contained 3641 individual release events reported by the original 150 facilities. Of
these, 319 events involved a Table 1 contaminant released by a Table 2 facility during startup,
shutdown, or malfunction. For evaluating short-term releases for residual risk assessments, these
319 events comprise the most representative subset of the full database.
B3.2. Descriptive statistics
For this subset of emission events, ratios of event release rate to long-term whole-facility release
rate varied from 0.00000004 to 74. Distribution statistics appear in Tables 3 and 4. The 99th
percentile ratio was 9 (i.e., an event release rate nine times the long-term average). Only 3 ratios
exceeded our default assumption of 10, and of these only one exceeded 11. The full cumulative
probability density of the ratios is shown in Figure 1.
Figure 2 shows the relationship between ratio and event duration. As expected, the ratio
declined as duration increased. Only 18 events lasted less than 2 hours, but these events
produced the three highest ratios. Figure 3 is a similar ratio vs. duration plot, but with duration
as a percentage of total time. Only 35 events exceeded 1% of the total period covered by the
database. Figure 4 shows the relationship between ratio and total amount released, and suggests
that the highest ratios were produced by facilities whose routine VOC emissions were relatively
small. Thus, the events themselves also tended to be relatively small in absolute terms.
B3.3. Discussion
These results suggest that the tenfold ratio assumption for short-term releases is protective, and
that the facilities for which it may underestimate event releases may tend to be smaller emitters.
However, this analysis is limited in the following ways by the nature of the database and the
filtering that we applied:
1. The only long-term release data available for VOCs from the database were total
emissions for 2004, and the only short-term release data were emissions of the individual
substances that triggered the data entry. Ideally, we would have preferred to have routine
release rates for each individual compound, or at least event release rates for total VOCs.
B-3
-------
However, retrieving these data from other sources and linking them to this database was
not feasible.
2. Removing VOCs that are not representative of HAPs, and comparing the releases against
all VOCs, would tend to underestimate the true ratios. This effect could be quantitatively
large.
3. Retaining HAPs that are not VOCs and including them in the total to be compared
against all VOCs, would tend to overestimate the true ratios. The size of this effect is
not known, but seems likely to be less than for (2) above.
4. The database contains only facilities that had at least one release event during the
reporting period. The number of facilities in the statistical population that did not
experience an event is not known. The lack of data for these facilities (whose ratios in
this analysis would have been zero) would cause the descriptive statistics to be skewed
toward an overestimate. The size of this effect is unknown.
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
2-Methyloctane
No
3221-61-2
90008
2-Methylpentane
No
107-83-5
43229
2-methylhexane
No
591-76-4
43263
2-Methylpentane
No
107-83-5
43229
2,2,3-Trimethylpentane
No
564-02-3
2,2,4-Trimethylpentane
Yes
540-84-1
43250
dimethyl butane
No
75-83-2
43291
2,3-Dimethylbutane
No
79-29-8
43276
2,3,4-Trimethylpentane
No
565-75-3
43252
2,3-Dimethylbutane
No
79-29-8
43276
2,4-Dimethylpentane
No
108-08-7
43247
2-methylheptane
No
592-27-8
43296
2-methylhexane
No
591-76-4
43263
2-Methylpentane
No
107-83-5
43229
3-Methylhexane
No
589-34-4
43295
3-Methylpentane
No
96-14-0
43230
3-Methylhexane
No
589-34-4
43295
3-Methylpentane
No
96-14-0
43230
3-Methylheptane
No
589-81-1
43253
3-Methylhexane
No
589-34-4
43295
3-Methylpentane
No
96-14-0
43230
Acetaldehyde
Yes
75-07-0
43503
Acetic Acid
No
64-19-7
43404
Acetonitrile
Yes
75-05-8
70016
B-4
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Acetophenone
Yes
98-86-2
Acrolein
Yes
107-02-8
43505
Acrylic acid
Yes
79-10-7
43407
Acrylonitrile
Yes
107-13-1
43704
alkylphenol
No
none
Benzene
Yes
71-43-2
45201
Benzo[a]anthracene
Yes
56-55-3
46716
Benzo[a]pyrene
Yes
50-32-8
46719
Benzo[b]fluoranthene
Yes
205-99-2
46717
Biphenyl
Yes
92-52-4
45226
Butanol
No
35296-72-1
Butyl Acrylate
No
141-32-2
43440
t-Butyl Alcohol
No
75-65-0
43309
butylcyclohexane
No
1678-93-9
90101
Butyraldehyde
No
123-72-8
43510
C9 Aromatics
No
none
Naphthalene
Yes
91-20-3
46701
Nonane
No
111-84-2
43235
C9+
No
none
Carbon tetrachloride
Yes
56-23-5
43804
Carbonyl Sulfide
Yes
463-58-1
43933
Chloral
No
75-87-6
Trichloromethane
Yes
67-66-3
43803
Chlorothalonil
No
1897-45-6
Petroleum
No
8002-05-9
Petroleum
No
8002-05-9
Cumene
Yes
98-82-8
45210
Cyclohexane
No
110-82-7
43248
Cyclohexanol
No
108-93-0
43317
Cyclohexanone
No
108-94-1
43561
Cyclohexanone
No
108-94-1
43561
Decane
No
124-18-5
43238
Decane
No
124-18-5
43238
1,2-Dichloroethane
No
107-06-2
43815
Diethylbenzene (mixture)
No
25340-17-4
45106
Methyl Ether
No
115-10-6
43350
Dimethylcyclohexane
No
27195-67-1
98059
Dimethylcyclopentane
No
28729-52-4
90064
Dimethylcyclopentane
No
28729-52-4
90064
Dimethyl formamide
Yes
68-12-2
43450
Dimethylhexane
No
28777-67-5
90067
B-5
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Dimethyl pentane
No
38815-29-1
90063
Epichlorohydrin
Yes
106-89-8
43863
Ethyl Alcohol
No
64-17-5
43302
Ethyl Acrylate
Yes
140-88-5
43438
Ethyl Alcohol
No
64-17-5
43302
Ethyl Benzene
Yes
100-41-4
45203
Ethyl Chloride
Yes
75-00-3
43812
Ethylcyciohexane
No
1678-91-7
43288
ethylacetylene
No
107-00-6
43281
Ethyl Benzene
Yes
100-41-4
45203
f"
Ethylene Oxide
Yes
75-21-8
43601
ethylmethylbenzene
No
25550-14-5
45104
formaldehyde
Yes
50-00-0
43502
Furfural
No
98-01-1
45503
straight-run middle distillate
No
64741-44-2
Gasoline
No
86290-81-5
Gasoline
No
86290-81-5
Heavy Olefins
No
none
n-Heptane
No
142-82-5
43232
n-Heptane
No
142-82-5
43232
Heptylene
No
25339-56-4
hexane
Yes
110-54-3
43231
hexane
Yes
110-54-3
43231
2-Methylpentane
No
107-83-5
43229
hexane
Yes
110-54-3
43231
Hexene
No
25264-93-1
43289
lndeno[1 ,2,3-cdlpyrene
Yes
193-39-5
46720
Isobutyraldehyde
No
78-84-2
43511
2-Methyl-1-propanol
No
78-83-1
43306
2-Methyl-1-propanol
No
78-83-1
43306
Isobutyraldehyde
No
78-84-2
43511
Isoheptanes (mixture)
No
31394-54-4
43106
2-Methylpentane
No
107-83-5
43229
2,2,4-Trimethylpentane
No
540-84-1
43250
2,2,4-Trimethylpentane
No
540-84-1
43250
Isopar E
No I I
Isoprene
No
78-79-5
43243
2-Propanol
No
67-63-0
43304
2-Propanol
No
[ 67-63-0
43304
Cumene
Yes
98-82-8
45210
Isopropylcyclohexane
No
696-29-7
90128
B-6
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Diisopropyl ether
No
108-20-3
85005
Kerosene
No
64742-81-0
Methyl ethyl ketone
No
78-93-3
43552
Methyl isobutenyl ketone
Yes
141-79-7
I"
Methanol
Yes
67-56-1
43301
Methyl Acetylene
No
74-99-7
43209
Cresol
Yes
1319-77-3
45605
Methyl Chloride
Yes
74-87-3
43801
methyl cyclohexane
No
108-87-2
43261
Methyl ethyl ketone
No
78-93-3
43552
f"
lodomethane
No
74-88-4
86025
Methyl Mercaptan
No
74-93-1
43901
methyl cyclohexane
No
108-87-2
43261
Methylcyclopentane
No
96-37-7
43262
2-Methyldecane
No
6975-98-0
98155
Methylheptane
No
50985-84-7
90045
2-methylheptane
No
592-27-8
43296
2-Methyl nonane
No
871-83-0
90047
Tert-butyl methyl ether
No
1634-04-4
43376
meta-xylene
No
108-38-3
45205
Nonane
No
111-84-2
43235
Naphtha
No
8030-30-6
45101
Naphthalene
Yes
91-20-3
46701
Naphtha
No
[ 8030-30-6
45101
Naphthalene
No
91-20-3
46701
Butyl acetate
No
123-86-4
43435
Butyraldehyde
No
123-72-8
43510
Nonane
No
111-84-2
43235
Nonane
No
111-84-2
43235
Octadecene
No
27070-58-2
n-Octane
No
111-65-9
43233
Octene (mixed isomers)
No
25377-83-7
ortho-xylene
No
95-47-6
45204
Parathion
Yes
56-38-2
4-Aminohippuric Acid
No
61-78-9
Phenol
Yes
108-95-2
45300
Silicone
No
63148-62-9
Naphtha
No
8030-30-6
45101
Naphtha
No
I 8030-30-6
45101
Polyethylene
No
9002-88-4
Poly(lsobutylene)
No
9003-27-4
B-7
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Chloromethyl pivalate
No
18997-19-8
Process fuel gas
No
none
Propionic Acid
No
79-09-4
43405
Propylene oxide
No
75-56-9
43602
I"
para-xylene
No
106-42-3
45206
Styrene
Yes
100-42-5
45220
Sulfolane
No
126-33-0
t-Butyl Alcohol
No
75-65-0
43309
t-Butyl Alcohol
No
75-65-0
43309
tert-butyl hydroperoxide
No
75-91-2
f"
Toluene
Yes
108-88-3
45202
Aqualyte(TM), LSC cocktail
No
25551-13-7
45107
1,3,4-Trimethylbenzene
No
95-63-6
45208
trimethylcyclopentane
No
30498-64-7
98058
trimethylpentane
No
29222-48-8
90092
Undecane
No
1120-21-4
43241
Vinyl acetate
Yes
108-05-4
43453
Vinyl acetate
Yes
108-05-4
43453
Vinyl chloride
Yes
75-01-4
43860
vinyl resin
No
none
Vinylcyclohexane
No
695-12-5
xylenes
Yes
1330-20-7
45102
xylenes
Yes
1330-20-7
45102
meta-xylene
Yes
108-38-3
452051
ortho-xylene
Yes
95-47-6
45204
para-xylene
Yes
106-42-3
45206
Mineral spirits
No
64475-85-0
43118
Propylene glycol
No
57-55-6
43369
Vinyl chloride
Yes
75-01-4
43860
1-Decene
No
872-05-9
90014
2-Ethyl-1-hexanol
No
104-76-7
43318
2-Pyrrolidone
No
616-45-5
Aromatic
No
none
Decene
No
25339-53-1
90014
2-N,N-Dibutylaminoethanol
No
102-81-8
86007
Diisopropanolamine
No
110-97-4
86004
N,N-Dimethylethanolamine
No
108-01-0
84004
trifluoroethane
No
27987-06-0
2.2'-Oxybisethanol
No
111-46-6
43367
Hydrocarbons
No
none
Methyl Formate
No
107-31-3
43430
B-8
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Isopropylamine
No
75-31-0
86014
n-Butanol
No
71-36-3
43305
Polypropylene glycol ether
No
N-Vinyl-2-Pyrrolidinone
No
88-12-0
1,1 -Di(t-Amylperoxy)
Cyclohexane
No
15667-10-4
1,2,3-Trimethyl-4-ethylbenzene
No
none
2-Methyldecane
No
6975-98-0
98155
2-methylheptane
No
592-27-8
43296
2-Methyl nonane
No
871-83-0
90047
2,5-Dimethylhexane-2,5-
dihydroperoxide
No
3025-88-5
Butyl ether
No
142-96-1
43372
1,2-Dichloroethane
Yes
107-06-2
43815
Hydrindene
No
496-11-7
98044
Methylheptane
No
50985-84-7
90045
methyl methacrylate
No
80-62-6
43441
Naphtha
No
8030-30-6
45101
hexane
Yes
110-54-3
43231
tert-amyl hydroperoxide
No
3425-61-4
1,3,4-Trimethylbenzene
No
95-63-6
45208
n-Butanol
No
71-36-3
43305
2-Butoxy ethanol
Yes
111-76-2
43308
hexane
Yes
110-54-3
43231
cycloheptane
No
291-64-5
43115
n-Heptane
No
142-82-5
43232
n-Octane
No
111-65-9
43233
Hexyl Carbitol
No
112-59-4
Nonene
No
27215-95-8
Silane, ethenyltrimethoxy
No
2768-02-7
tetrahydrofuran
No
109-99-9
70014
Vinyl chloride
Yes
75-01-4
43860
Methyl Formate
No
107-31-3
43430
Phenyl ether
No
101-84-81
phosgene
Yes
75-44-5]
1,2-Dichloroethane
No
107-06-2
43815
2-Butoxy ethanol
Yes
111-76-2
43308
Gasoline
No
86290-81-5
1-Tridecanol
No
112-70-9
1,2,4-Trichlorobenzene
Yes
120-82-1
45208
2- (2- B utoxy eth oxy) et h a n o I
Yes
112-34-5
43312
2,3,4-trihydroxybenzophenone
No
1143-72-21
B-9
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
Ester
I
Methyl n-amyl ketone
No
110-43-0
43562
4,4-Cyclohexylidenebisfphenoll
No
843-55-0
Anisole
No
100-66-3
2-Butoxy ethanol
Yes
111-76-2
43308
Cresol-Formaldehyde novolac
Resin
No
J proprietary
Decane
No
124-18-5
43238
gamma-Butyrolactone
No
96-48-0
Dimethyl pentane
No
i 38815-29-1
90063
Dodecyl Benzenesulfonic Acid
No
! 27176-87-0
Ethanol Amine
No
I 141-43-5
43777
ethyl lactate
No
I 687-47-81
Hexamethyldisilazane
No
999-97-3
Methyl ethyl ketone
No
¦ 78-93-3
43552
Cresol
Yes
! 1319-77-3
45605
Naphthalene Sulfonic Acid Resin
No
|
Naphthalene Sulfonic Acid Resin
No
J
n-Butanol
No
71-36-3
43305
Decane
No
124-18-5
43238
1-Methyl-2-pyrrolidinone
No
I 872-50-4
70008
Pentyl Ester Acetic Acid
No I
Phenol Formaldehyde Resin,
Novolac
No
Phenol Formaldehyde Resin,
Novolac
No
|
Propylene Glycol Monomethyl
Ether
No
I 107-98-2
70011
Pyrocatechol
No
120-80-9
Carbon Disulfide
Yes
75-15-0
43934
Hexene
No
592-41-6
43245
VOC
No
i
I none
Methacrylic acid
No
79-41-4
84009
Methyl 3-hydroxybutyrate
No
1487-49-6
t-Butyl Alcohol
No
75-65-0
43309
methyl valeraldehyde
No
123-15-9
Butyl Methacrylate
No
97-88-1
85008
dipropyl ether
No
111-43-3
n-Propanol
No
71-23-8
43303
Propyl propionate
No
106-36-5
86052
1,2-Epoxybutane
Yes
106-88-7
Methylamine
No
74-89-5
B-10
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
1,1-Dimethylcyclohexane
No
590-66-9
1,1-Dimethylcyclopentane
No
1638-26-2
2-Methylpentane
No
107-83-5
43229
dimethyl butane
No
75-83-2
43291
2,3,3-Trimethylpentane
No
560-21-4
2,3-Dimethylhexane
No
584-94-1
2,3-Dimethylpentane
No
565-59-3
2,4-Dimethylhexane
No
589-43-5
2,5-Dimethyl-hexane
No
592-13-2
2-Butoxy ethanol
Yes
111-76-2
43308
2-mercaptoethanol
No
60-24-2
Bisphenol A
No
80-05-7
straight-run middle distillate
No
64741-44-2
4-Vinylcyclohexene
No
100-40-3
straight-run middle distillate
No
64741-44-2
Allyl alcohol
No
107-18-6
xylenes
Yes
1330-20-7
45102
Naphthalene
Yes
91-20-3
46701
3-Methylethylcyclohexane
No
VOC
No
none
Gasoline
No
86290-81-5
Butyl ether
No
142-96-1
dimethyl butane
No
75-83-2
Dodecene
No
25378-22-7
Styrene
Yes
100-42-5
45220
tetrahydrofuran
No
109-99-9
70014
hexane
Yes
110-54-3
43231
2-Propanol
No
67-63-0
43304
liquified petroleum gas
No
68476-85-7
Methyl acetylene propadiene
No
methyl isobutyl ketone
Yes
108-10-1
Methyl n-amyl ketone
No
110-43-0
43562
Methylpentane
No
43133-95-5
Tert-butyl methyl ether
Yes
1634-04-4
43376
Toluene
Yes
108-88-3
45202
Mineral oil
No
8012-95-1
Gasoline
No
86290-81-5
2,2-Dimethylpropane
No
463-82-1
43222
n-propyl benzene
No
103-65-1
propylcyclohexane
No
1678-92-81
n-Octane
No
111-65-9
43233
B-ll
-------
Table 1. Event emissions in the Houston-Galveston area.
Representative contaminants included in the analysis, selected because
they are either HAPs or VOCs with more than 5 carbon atoms. (These
data were retrieved directly from the original database, which included
multiple redundancies that did not affect the analysis and were left
intact.)
Contaminant
HAP
CAS
SAROAD
ortho-xylene
No
95-47-6
45204
Gasoline
No
86290-81-5
propylenimine
No
75-55-8
Gasoline
No
86290-81-5
Technical White Oil
No
Total Alkylate - non-speciated
No
Trichloroethylene
Yes
79-01-6
Di(2-ethylhexyl)
peroxydicarbonate
No
16111-62-9
trimethylcyclopentane
No
30498-64-7
98058
Ultraformate
No
4-Vinylcyclohexene
No
100-40-3
B-12
-------
Table 2. Event emissions in the Houston-Galveston area. Major emitters
reporting at least one release event of a representative substance.
2004 VOC Emission
Company Name Rate (Ib/h)
ATOFINA PETROCHEMICALS LA PORTE PLANT 47.88
BALL METAL BEVERAGE CONTAINER CONROE 24.18
FACILITY
BASF FREEPORT SITE 46.47
BELVIEU ENVIRONMENTAL FUELS 112.3
BOC GROUP CLEAR LAKE BOC GASES PLANT 9.52
BP AMOCO CHEMICAL CHOCOLATE BAYOU PLANT 130.4
BP AMOCO CHEMICAL PASADENA PLANT 36.92
BP AMOCO POLYMERS 57.18
BP PRODUCTS NORTH AMERICA TEXAS CITY 737.4
BP TEXAS CITY CHEMICAL PLANT B 112.2
CELANESE BAY CITY PLANT 17.12
CELANESE CLEAR LAKE PLANT 53.11
CELANESE PASADENA PLANT 5.934
CHEVRON PHILLIPS CEDAR BAYOU PLANT 105.3
CHEVRON PHILLIPS CHEMICAL SWEENY COMPLEX 106.7
CHEVRON PHILLIPS HOUSTON CHEMICAL COMPLEX 215.7
CROWN BEVERAGE PACKAGING 18.05
CROWN CENTRAL PETROLEUM PASADENA PLANT 114.3
CROWN CORK & SEAL 18.10
DEER PARK LIQUID STORAGE TERMINAL 124.8
DOW CHEMICAL LA PORTE SITE 5.902
DOW TEXAS OPERATIONS FREEPORT 203.2
E I DUPONT DE NEMOURS AND COMPANY - LA 51.30
PORTE PLANT
EQUISTAR CHEMICALS CHANNELVIEW COMPLEX 275.4
EQUISTAR CHEMICALS CHOCOLATE BAYOU 84.87
COMPLEX
EQUISTAR CHEMICALS LA PORTE COMPLEX 90.97
EXXON MOBIL CHEMICAL BAYTOWN OLEFINS PLANT 84.73
EXXONMOBIL CHEMICAL BAYTOWN CHEMICAL 313.7
PLANT
EXXONMOBIL CHEMICAL MONT BELVIEU PLASTICS 40.64
PLANT
GOODYEAR HOUSTON CHEMICAL PLANT 85.68
ISP TECHNOLOGIES TEXAS CITY PLANT 22.12
KANEKA TEXAS CORPORATION 20.55
KINDER MORGAN LIQUID TERMINALS PASADENA 913.9
KINDER MORGAN LIQUIDS TERMINALS 132.7
LBC HOUSTON BAYPORT TERMINAL 12.83
LYONDELL CHEMICAL BAYPORT PLANT 30.04
LYONDELL CHEMICAL CHANNELVIEW 74.15
MARATHON ASHLAND PETROLEUM TEXAS CITY 111.8
REFINERY
MOBIL CHEMICAL HOUSTON OLEFINS PLANT 26.29
MORGANS POINT PLANT 31.03
PASADENA PLANT 13.40
B-13
-------
Table 2. Event emissions in the Houston-Galveston area. Major emitters
reporting at least one release event of a representative substance.
2004 VOC Emission
Company Name Rate (Ib/h)
SHELL OIL DEER PARK
405.2
SOLUTIA CHOCOLATE BAYOU PLANT
53.09
STOLTHAVEN HOUSTON TERMINAL
7.347
SWEENY COMPLEX
157.1
UNION CARBIDE TEXAS CITY OPERATIONS
174.4
VALERO REFINING TEXAS CITY
260.1
WHARTON GAS PLANT
7.552
Table 3. Frequency distribution for ratio of event
emission rate to long-term emission rate
Cumulative
Bin Frequency Frequency
1.00E-08
0
0
3.16E-08
0
0
1.00E-07
2
2
3.16E-07
1
3
1.00E-06
0
3
3.16E-06
2
5
1.00E-05
1
6
3.16E-05
2
8
1.00E-04
5
13
3.16E-04
9
22
1.00E-03
15
37
3.16E-03
28
65
1.00E-02
33
98
3.16E-02
41
139
1.00E-01
59
198
3.16E-01
38
236
1.00E+00
33
269
3.16E+00
31
300
1.00E+01
16
316
3.16E+01
2
318
1.00E+02
1
319
3.16E+02
0
319
B-14
-------
Table 4. Statistics for ratio of event
emission rate to long-term emission rate
Statistic for
Ratio
Value
Median
0.043923
75th %ile
0.342655
90th %ile
2.204754
95th %ile
3.344422
96th %ile
3.400832
97th %ile
3.8126
98th %ile
4.790098
99th %ile
8.973897
Max
74.37138
Average
0.815352
Figure 1. Cumulative probability density for ratio of event to routine emission rates.
Cumulative probability of event ratios
350
300
250
200
150
100
50
0 *=
1.E-07
.E-06
.E-05
.E-04
1.E-03
Ratio of event emission rate to long-term emission rate
1.E-02
1.E-01
1.E+00
1.E+01
.E+02
.E+03
.E+04
B-15
-------
Figure 2. Relationship between ratio of event to duration emission rate and emission
duration.
Event ratio vs. duration
2 1.E+01
§ 1.E+00
1.E-01
E
g* 1.E-02
O
O
Z 1.E-03
2
c
.2 1.E-04
E
1.E-05
1.E-06
£ 1.E-07
1.E-08
~
* A .
~ ~
t
~
~
~ «
5T~
~~1
4
4
A .>
4
jr*
4 ~ «
~ ~
>
*
~
;
4
4
~
~
• ~
4^f ~
>4»4»
<
~J
~
~
f
*
*
•a
~ 4«
sir
$
%
~
•
4
~<
~
tin
<
~
~
~
~
4
>*
t
:<
~
~
~
:
~~
~
~
~
4
~
~
~
~
~
*
4
4
4
~
~~
~
~
~
~
~
• ~
4
~
~
4
~
t
~
4
10
100 1,000
Event duration (min)
10,000
100,000
Figure 3. Relationship between ratio of event to duration emission rate and emission duration,
percentage of total time.
Event ratio vs. duration
i
f
1
~
A
K
~
+
~
~
X
4
*
~
i
~~
~
~
~
~
y
~
:
~
4
~
~
~
4
~
4
4
r
~
:<
•
4
4
4
4
~
4
~
~
~
4
~
>
4
~
4
~
2 1.E+01 ; y
m ' '
o
o
Z 1.E-03
.2 1.E-04
w
w
E
« 1.E-05
1.E-06
1.E-07
3% 4% 5% 6%
Event duration (as % of total time)
B-16
-------
Figure 4. Relationship between ratio of event to duration emission rate and total amount emitted
during the event.
Event ratio vs. 2004 VOC releases - by event
1.E+02
£ 1.E+01
TO
£
% 1.E+00
W
E
1.E-02
O
O
Z 1.E-03
5
c
.2 1.E-04
V)
w
E
« 1.E-05
c
>
® 1.E-06
O
S. 1E-07
1.E-08
1 10 100 1,000
Long-term VOC releases (Ib/hr)
~
*
~
4 ~
J
~
I
~
1
It'
~ ~
• r
~
1*
»
~
•
'»!* I
~
~
~
~
•
~
1
r ^
~
~
~
4
4
v *
~
~
1
~
~
~
~ ~
* ~
~
~
~
~
~
~
B-17
-------
APPENDIX C: Technical Support Document for TRIM-Based
Multipathway Screening Scenario for RTR: Summary of Approach and
Evaluation
-------
[This page intentionally left blank.]
c-i
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY viii
C-1 Introduction 1
C-1.1 Background 1
C-1.2 Organization of this Document 2
C-2 Approach for Screening Evaluation of Multipathway Exposures 2
C-2.1 Overview 2
C-2.2 Chemicals of Potential Concern 5
C-2.3 Use of De Minimis Emission Levels 6
C-2.4 The TRIM-Based Screening Scenario: Basis for De Minimis Emission Rates 7
C-2.4.1 Exposure Routes Evaluated 8
C-2.4.2 Approach to Configuration and Parameterization 9
C-2.4.3 Modeling Framework 10
C-2.4.3.1 Fate and Transport Modeling 11
C-2.4.3.2Exposure Modeling and Risk Characterization 12
C-2.5 Refined Analyses 13
C-3 Description of Modeling Scenario 13
C-3.1 TRIM.FaTE Scenario Configuration and Parameterization 13
C-3.1.1 Chemical Properties 13
C-3.1.2 Spatial Layout 13
C-3.1.3 Watershed and Water Body Parameterization 15
C-3.1.3.1 Water Balance 15
C-3.1.3.2Sediment Balance 16
C-3.1.4 Meteorology 17
C-3.1.5 Aquatic Food Web 20
C-3.1.6 Using TRIM.FaTE Media Concentrations 20
C-3.2 Exposure and Risk Calculations 22
C-3.2.1 Calculating Concentrations in Farm Food Chain Media 22
C-3.2.2 Ingestion Exposure Assessment 23
C-3.2.2.1 Ingestion Exposure Pathways and Routes of Uptake 23
C-3.2.2.2Exposure Scenarios and Corresponding Inputs 23
C-3.2.2.3Calculating Average Daily Doses 25
C-3.2.2.4lnfant Ingestion of Breast Milk 25
C-3.2.3 Risk Characterization 26
C-3.2.4 Dermal Risk Screening 26
C-3.2.4.1 Hazard Identification and Dose Response Assessment 27
C-3.2.4.2Dermal Exposure Estimation 28
Equations for Estimating Dermal Exposure 28
Exposure Factors and Assumptions 29
Receptor-Specific Parameters 29
Chemical-Specific Parameters 31
C-3.2.4.3Screening-Level Cancer Risks and Non-Cancer Hazard Quotients 32
Dermal Cancer Risk 32
C-ii
-------
Dermal Hazard Quotient
C-3.2.4.4Dermal Screening Results
C-3.3 Summary of Scenario Assumptions
C-4 Evaluation of Screening Scenario
C-4.1 Overview
C-4.2 Overall Chemical Mass Partitioning
C-4.3 Comparison to Measured Concentrations
C-4.3.1 Scope of the Evaluation
C-4.3.2 Methods and Organization of this Section
C-4.3.3 Chemical-Specific Comparisons
C-4.3.3.1 Cadmium
Behavior in the Environment
Emission Profile
Concentrations in Environmental Media
Partitioning Behavior
Concentrations in Ingestible Products
C-4.3.3.2Mercury
Behavior in the Environment
Emission Profile
Concentrations in Environmental Media
Partitioning Behavior
Concentrations in Ingestible Products
C-4.3.3.3Dioxins (2,3,7,8-TCDD)
Behavior in the Environment
Emission Profile
Concentrations in Environmental Media
Partitioning Behavior
Concentrations in Ingestible Products
C-4.3.3.4PAHs (Benzo[a]pyrene)
Behavior in the Environment
Emission Profile
Concentrations in Environmental Media
Partitioning Behavior
Concentrations in Ingestible Products
C-4.3.4 Summary
C-4.4 Sensitivity Analyses
C-4.4.1 Systematic Sensitivity Analysis
C-4.4.1.1 Methods
C-4.4.1.2Results
C-4.4.2 Evaluation of Ingestion Rate Assumptions
C-4.4.3 Evaluation of Body Weight Assumptions
C-4.4.4 Sensitivity When Accounting for Temporally Correlated Body Weight and
Ingestion Rates
C-4.4.5 Comparison of Scenarios Using Site-Specific and de minimis Meteorological
Data
C-4.5 Comparison to Other Model Results
C-4.5.1 Comparison to Preliminary RTR Screening Runs (HHRAP Approach)
C-4.5.2 Comparison of Results for Screening Scenario and Previous TRIM.FaTE
Applications
32
33
34
37
37
38
40
40
41
42
42
42
43
.43
45
46
48
48
49
.49
51
53
56
56
57
.57
59
60
62
62
63
.63
65
66
69
70
71
71
73
81
83
84
86
88
88
88
C-iii
-------
C-5 References 94
Attachment C-1 - TRIM.FaTE Inputs
Attachment C-2 - RTR Access-based Exposure and Risk Calculation Tool - Multimedia
Ingestion Risk Calculator (MIRC)
Attachment C-3 - Systematic Sensitivity Analysis Variables and Results
C-iv
-------
LIST OF EXHIBITS
Exhibit 2-1. Conceptual Decision Tree for Evaluation of Non-Inhalation Exposures of PB-HAPs 4
Exhibit 2-2. OAQPS PB HAP Compounds a 5
Exhibit 2-3. De Minimis Thresholds for Screening of Multipathway Exposures 6
Exhibit 2-4. Overview of Ingestion Exposure and Risk Screening Evaluation Method 11
Exhibit 2-5. Overview of Process Carried Out in the Multimedia Ingestion Risk Calculator 12
Exhibit 3-1. TRIM.FaTE Surface Parcel Layout 14
Exhibit 3-2. Summary of Key Meteorological Inputs 19
Exhibit 3-3. Parameters for Aquatic Biota for the Screening Scenario of TRIM.FaTE 21
Exhibit 3-4. Spatial Considerations - TRIM.FaTE Results Selected for Calculating Farm Food Chain
Media Concentrations and Receptor Exposures 22
Exhibit 3-5. Summary of Ingestion Exposure Pathways and Routes of Uptake 24
Exhibit 3-6. Overview of Exposure Factors Used for RTR Multipathway Screening a,b 25
Exhibit 3-7. Dose-response Values for PB HAPs Addressed by the Screening Scenario 26
Exhibit 3-8. Cancer Slope Factors and Reference Doses Based on Absorbed Dose 28
Exhibit 3-9. Receptor-Specific Body Surface Area Assumed to be Exposed to Chemicals 30
Exhibit 3-10. Scenario-Specific Exposure Values for Water and Soil Contact 31
Exhibit 3-11. Chemical-Specific Dermal Exposure Values for Water and Soil Contact 32
Exhibit 3-12. Summary of Dermal Non-Cancer Flazards 33
Exhibit 3-13. Summary of Dermal Cancer Risks 34
Exhibit 3-14. Summary of RTR Screening Scenario Assumptions and Associated Conservatism 35
Exhibit 4-1. Distribution of Chemical Mass in Screening Scenario 40
Exhibit 4-2. Summary of Modeled and Observed Concentrations of Cadmium in Environmental Media . 44
Exhibit 4-3. TRIM.FaTE Cadmium Concentrations in Fish and Calculated Bioaccumulation Factors with
Respect to Total Water Concentration 45
Exhibit 4-4. Fraction of Cadmium Mass Sorbed vs. Dissolved in TRIM.FaTE Compartments 46
Exhibit 4-5. Summary of Modeled and Observed Concentrations of Cadmium in Ingestible Media 47
Exhibit 4-6. Estimated Contribution of Modeled Food Types to Cadmium Ingestion Exposures and
Flazard Quotient 48
Exhibit 4-7. Summary of Modeled and Observed Concentrations of Total Mercury in Environmental
Media 50
Exhibit 4-8. TRIM.FaTE Mercury Concentrations, Speciation, and Calculated Methyl Mercury
Bioaccumulation Factors (in white boxes) in Fish Compartments 51
Exhibit 4-9. TRIM.FaTE Mercury Speciation and Partitioning in Environmental Media Compartments....52
Exhibit 4-10. Summary of Modeled and Observed Concentrations of Total Mercury in Ingestible Media54
Exhibit 4-11. Estimated Contribution of Modeled Food Types to Divalent Mercury and Methyl Mercury
Ingestion Exposures 55
Exhibit 4-12. Estimated Contribution of Summed Modeled Food Types to Divalent Mercury and Methyl
Mercury Flazard Quotients 55
C-v
-------
Exhibit 4-13. Summary of Modeled 2,3,7,8-TCDD Concentrations and Observed Total Dioxin TEQ
Concentrations in Environmental Media 58
Exhibit 4-14. TRIM.FaTE 2,3,7,8-TCDD Concentrations in Fish and Calculated Bioaccumulation Factors
with Respect to Total Water Concentration 59
Exhibit 4-15. Fraction of 2,3,7,8 - TCDD Mass Sorbed vs. Dissolved in TRIM.FaTE Compartments 59
Exhibit 4-16. Summary of Modeled 2,3,7,8-TCDD Concentrations and Observed Total Dioxin TEQ
Concentrations in Ingestible Media 61
Exhibit 4-17. Contribution of Modeled Food Types to 2,3,7,8-TCDD Ingestion Exposures (mg/kg/day) ..62
Exhibit 4-18. Summary of Modeled and Observed Concentrations of Benzo[a]pyrene in Environmental
Media 64
Exhibit 4-19. TRIM.FaTE Benzo[a]Pyrene Concentrations in Fish and Calculated Bioaccumulation
Factors with Respect to Total Water Concentration 65
Exhibit 4-20. Fraction of Benzo[a]Pyrene Mass Sorbed vs. Dissolved in TRIM.FaTE Compartments 66
Exhibit 4-21. Summary of Modeled and Observed Concentrations of Benzo[a]pyrene in Ingestible Media
68
Exhibit 4-22. Contribution of Modeled Food Types to Benzo[a]pyrene Ingestion Exposures (mg/kg/day)69
Exhibit 4-23. The 26 Variables with the Highest Elasticities for Benzo[a]Pyrene Lifetime Risk (-5%
Perturbation of Variable) 76
Exhibit 4-24. The 25 Variables with the Highest Elasticities for 2,3,7,8-TCDD Lifetime Risk (-5%
Perturbation of Variable) 77
Exhibit 4-25. The 28 Variables with the Highest Elasticities for Cadmium Hazard Quotient for Child 1-2 (-
5% Perturbation of Variable) 78
Exhibit 4-26. The 25 Variables with the Highest Elasticities for Divalent Mercury Hazard Quotient for Child
1-2 (-5% Perturbation of Variable) 79
Exhibit 4-27. The 26 Variables with the Highest Elasticities for Methyl Mercury Hazard Quotient for Child
1-2 (-5% Perturbation of Variable) 80
Exhibit 4-28. Ratio of the Modeled Total Ingestion Rates and the USEPA Total Ingestion Rates 82
Exhibit 4-29. Comparison of the Risks and Hazard Quotients in the de minimis and Alternate Ingestion
Cases 83
Exhibit 4-30. The Risk or Hazard Quotient Estimates Using Alternate Body Weight Percentiles 84
Exhibit 4-31. Comparison in the Elasticities In Lifetime Risk in the Correlated and Uncorrelated Analyses
Assuming a 5% Decrease in the Input Variables 85
Exhibit 4-32. Summary of Site-specific Meteorological Data Parameters 86
Exhibit 4-33. Percent Change in Risk or Hazard Quotient Using Site-specific Meteorological Data 87
Exhibit 4-34. The Wind Speed and the Direction Toward Which the Wind is Blowing for All Conditions for
Site 1 87
Exhibit 4-35. Emission Thresholds Derived in Preliminary HHRAP Screening Runs and in Current
Analyses 88
Exhibit 4-36. Meteorological Data Parameters for TRIM.FaTE Secondary Lead Smelting Application ....89
Exhibit 4-37. Surface Soil Parcel Spatial Layouts for New York Site Lead Smelting TRIM.FaTE
Application and Screening Scenario 90
Exhibit 4-38. Air Parcel Spatial Layouts for New York Site Lead Smelting TRIM.FaTE Application and
Screening Scenario 91
C-vi
-------
Exhibit 4-39. Comparison of Concentration Outputs: NY Site Refined TRIM.FaTE Application vs.
Screening Scenario 92
Exhibit 4-40. Comparison of Concentration Outputs Grouped By Chemical: New York Site Refined
TRIM.FaTE Application vs. Screening Scenario 93
C-vii
-------
EXECUTIVE SUMMARY
This report describes the methods used by EPA to evaluate multipathway exposures to
hazardous air pollutants and the associated human health risks. In particular, the report
explains the methods, assumptions, and input data used to develop a screening scenario that is
used in the first tier of Risk and Technology Review II (RTR II) assessments of such emissions.
This report describes the development of the screening scenario and its application in
generating qualitative, screening-level estimates of human exposure to PB-HAPs and
consequent risk.
BACKGROUND
The Clean Air Act directs the U.S. Environmental Protection Agency (EPA) to assess the
residual risk from hazardous air pollutants (HAPs) emitted by sources regulated by technology-
based standards. To evaluate multipathway exposures and human health risks, a two-tiered
approach was developed. In the first tier of the approach, a screening evaluation is conducted
that uses the identity and magnitude of HAP emissions from a source to determine whether that
source meets certain human health risk-based criteria with respect to multipathway exposures.
The purpose of this first-tier screening is to eliminate facilities from further analysis that pose no
unacceptable risk to human health, while identifying those facilities that warrant a second-tier,
more refined, site-specific analysis of residual risk.
PURPOSE OF THE SCENARIO
The approach described here for evaluating human multipathway exposures and risks consists
of an initial, screening-level tier that can be conducted quickly and efficiently to determine those
facilities for which multipathway risks are expected to be below levels of concern. The key
component of the first tier of this approach is a multipathway screening scenario based on
EPA's Total Risk Integrated Methodology (TRIM). The TRIM-based modeling scenario provides
a means for quickly and efficiently completing an initial non-inhalation exposure and risk
screening analysis of a facility. The scenario is applied for use in RTR evaluations by
calculating de minimis emission rates for selected PB-HAPs that correspond to a cancer risk of
1 in 1 million or a chronic non-cancer hazard quotient (HQ) of 1. These de minimis or threshold
emission rates then can be used in the first risk screening step without requiring additional
model runs. Sources whose emissions exceed the de minimis emission rate for any PB-HAP
would be subjected to refined evaluation(s) in a second tier analysis.
The scenario has been used to calculate numerical exposure and risk values for 4 of the 14
HAPs that OAQPS has identified as candidates for multipathway risk assessments: cadmium,
mercury, dioxins (i.e., chlorinated dibenzo-p-dioxins and -furans), and polycyclic organic matter.
These compounds were selected because they are expected, based on current knowledge of
relative emissions and toxicity, to pose a substantial share of the non-inhalation risks to humans
from air emissions at sources subject to residual risk provisions of the Clean Air Act. The
scenario is not intended to be used to produce quantitative estimates of actual or potential risk.
Rather, it provides a basis for determining if residual human health risks are of potential
concern. Such determinations can then be used to support decisions to proceed with or forego
more definitive analyses of non-inhalation exposures to HAPs and the associated risks. The
scenario does, however, provide a technically defensible starting point for additional fate and
transport and exposure/risk analyses of facility emissions that are not "screened out" in the first
tier evaluation.
C-viii
-------
OVERVIEW OF SCREENING SCENARIO
This screening scenario is intended to reduce the possibility that EPA will fail to identify
unacceptable risks. Thus, the scenario's conservative approach most likely overestimates risk,
which is appropriate for a screening assessment. Parameter values were defined carefully, and
properties having more uncertainty were assigned greater conservative bias to prevent
underestimating potential risks. The screening scenario is designed to estimate the upper end
of the range of individual, long-term, non-inhalation exposures for situations likely to be
encountered in the United States. The result of reviewing a broad range of conditions and
selecting values representative of higher exposures in conceptualizing and building the
screening tool is that the scenario is unlikely to occur at any one location but has a high
likelihood of representing the upper end of any potential exposures.
The screening scenario addresses non-inhalation exposures, which can occur through both
dermal and ingestion exposure pathways. Pathways examined include incidental ingestion of
soil; ingestion of homegrown produce, beef, cows' milk, poultry and eggs, and pork; and
ingestion of fish. Dermal absorption of chemicals that are originally airborne is generally
relatively minor, and this pathway was not included in the scenario used to calculate de minimis
emission thresholds. A highly conservative estimate of dermal exposures and risks was
calculated for comparison to ingestion exposures and risks. In addition, exposure to nursing
infants via consumption of contaminated breast milk was evaluated for dioxins as a separate
scenario.
For this approach, chemicals were modeled separately to evaluate the potential for risks, with
exposures for each PB-HAP summed across all ingestion exposure pathways. Exposures were
modeled for a hypothetical farm homestead and fishable lake near an emissions source. For
this setting, exposures were estimated for a hypothetical individual assuming subsistence
consumption of all potentially contaminated foodstuffs from the farm or lake. The scenario was
purposely designed to produce conservative (i.e., health-protective) results, and certain critical
exposure/activity assumptions, such as food ingestion rates, were selected from the upper ends
(e.g., the 90th percentile) of representative exposure parameter distributions. The
physical/chemical environment was parameterized with a mix of typical and health-protective
values. The scenario's spatial/temporal aspects and the components that influence air
concentrations were also chosen so that concentrations in environmental media would not be
underestimated given the range of possible settings and meteorological conditions that might be
encountered. Properties of the environmental media were parameterized with either typical or
conservative values, with a more protective bias introduced for properties having greater
uncertainty.
MODELING FRAMEWORK
The approach for risk evaluation of ingestion exposures and risk screening has four
components:
1. fate and transport modeling of PB-HAPs emitted to air that partition into soil, water, and
other environmental media (including fish);
2. modeling of transfer and uptake of PB-HAPs by farm food chain media from soil and air;
3. estimating ingestion exposures for the selected media contact scenarios and average
daily ingestion doses for a hypothetical human receptor; and
4. calculating lifetime cancer risk estimates or chronic non-cancer HQs for each HAP.
C-ix
-------
TRIM.FATE SCENARIO CONFIGURATION AND PARAMETERIZATION
To model chemical fate and transport in the environment, EPA's Fate, Transport, and Ecological
Exposure (TRIM.FaTE) module of the TRIM system was used. The physical configuration of
the scenario was designed to be generally conservative and the environmental and chemical-
specific properties were parameterized with either conservative or central-tendency values.
Chemical/physical properties were obtained from peer-reviewed and standard reference
sources. The spatial layout represents a farm homestead and a fishable lake near an emissions
source. The predominant wind direction is toward the farm and lake watershed, and the
downwind modeling area is symmetrical around a 10-kilometer east-west line and divided into
five pairs of parcels. The aquatic food web in the scenario is meant to represent a generic
aquatic ecosystem within a 47-hectare lake.
Fate and transport modeling outputs include average PB-HAP concentrations and deposition
rates for various media (air, soil, surface water, and fish) for each year and for each parcel of
the model scenario. TRIM.FaTE can output instantaneous chemical concentrations for a user-
specified time step and also can be configured to calculate temporal averages. For the
screening scenario, the model outputs results on a daily basis, and daily concentration results
are averaged to obtain annual average concentrations. The source is assumed to emit for 50
years.
EXPOSURE AND RISK CALCULATIONS
The Multimedia Ingestion Risk Calculator (MIRC) was developed to carry out required farm food
chain transfer, ingestion exposure, and risk calculations. Concentrations in farm food chain
media are calculated using empirical biotransfer factors (e.g., soil-to-plant factors, which are the
ratios of the concentrations in plants to concentrations in soil). Ingestion exposures based on
exposure factors, including food-type-specific ingestion rates, are calculated for a hypothetical
exposed individual. Lifetime cancer risks and the potential for chronic non-cancer effects are
estimated using chemical-specific ingestion cancer slope factors and reference doses.
Exposure pathways evaluated include incidental ingestion of soil and consumption offish,
produce, and farm animals and related products. Cancer risk estimates and HQs are calculated
separately for each PB-HAP included in an analysis.
CALCULATION OF DEMINIMIS EMISSION THRESHOLDS
After the configuration of the TRIM.FaTE and MIRC modeling scenarios was completed, de
minimis emission rate thresholds were calculated by conducting iterative model simulations to
determine emission rates for cadmium, mercury, dioxins, and polycyclic organic matter that
correspond to a cancer risk of 1 in 1 million or a chronic non-cancer hazard quotient (HQ) of 1.
Given the generally conservative nature of the scenario inputs, these thresholds are assumed to
be appropriate for screening sources emitting these HAPs.
EVALUATION OF SCREENING SCENARIO
Model evaluations serve as an important aspect of environmental risk assessments by
illustrating the performance of the model under different conditions and assumptions and
facilitating the comparison of model outputs to measurement data and other modeling results.
Evaluations thereby provide an opportunity to gain confidence in model performance and
identify and better characterize uncertainties associated with model construct and inputs. The
screening scenario was analyzed through comparisons to the literature and sensitivity analyses.
C-x
-------
Outputs from the screening scenario based on de minimis emission values were compared to
literature values for cadmium, mercury, 2,3,7,8-TCDD, and benzo[a]pyrene. In general, taking
into account the limitations of this type of comparison, the configuration of the models for
screening appear to be reasonable. Chemical partitioning predicted by the model is generally
consistent with information on environmental partitioning presented in the literature for media of
concern. Where results are not consistent with literature values, a more detailed investigation of
underlying assumptions may help to identify means of adjusting the scenario configuration.
Sensitivity analyses were performed to evaluate the influence of model parameters and to
provide information on which parameters are likely to be most influential in dictating the
uncertainty associated with the results. The sensitivity analyses conducted on the RTR
screening modeling scenario encompassed the fate and transport modeling carried out using
TRIM.FaTE and the farm food chain and ingestion exposure calculations performed using
MIRC. A systematic sensitivity analysis was conducted by varying each input independently
and calculating the resulting effect on the risk or hazard quotient estimates in order to rank the
variables from most to least sensitive. The analysis suggests that several TRIM.FaTE variables
(including wind speed, mixing height, emission rate and, for methyl mercury, sediment
deposition rate) have the largest effect on the risk and hazard estimates. The estimates of
hazard and risk are also highly sensitive to key parameters in the primary exposure pathway
(i.e., ingestion of food types resulting in the highest exposures) for each PB-HAP. Other
analyses performed indicated that accounting for temporal correlations in ingestion rates and
body weights and varying the body weight and ingestion rate percentiles used in the model
scenario have a limited effect on the risk and hazard estimates. The use of site-specific
meteorological variable values (as opposed to the generic screening scenario values) resulted
in a decrease in the risk and hazard estimates of approximately one order of magnitude.
In addition, media concentrations estimated using the screening scenario were compared to
analogous outputs estimated using site-specific TRIM.FaTE model applications configured for
two secondary lead smelting sources. The same emission rates for benzo[a]pyrene, 2,3,7,8-
TCDD, elemental mercury, and divalent mercury were entered into the screening scenario and
the site-specific model scenarios. Model results were compared for soil, water, and sediment
compartment types. In all media, the screening scenario produced higher concentrations for all
chemicals than the site-specific model applications, the expected result given that the screening
scenario is conservative (and therefore tends to result in higher media concentrations).
C-xi
-------
[This page intentionally left blank.]
C-xii
-------
C-1 Introduction
C-1.1 Background
Section 112(f)(2)(A) of the Clean Air Act (CAA) directs the U.S. Environmental Protection
Agency (EPA) to assess the risk remaining (residual risk) from hazardous air pollutants (HAPs)
that continue to be emitted from sources after application of maximum achievable control
technology (MACT) standards under section 112(d) of the CAA. Under these requirements,
EPA will promulgate additional emission standards for a source category if the MACT standards
do not provide an "ample margin of safety" for human health. One aspect of human health that
EPA must consider is the potential for exposures to HAPs via non-inhalation pathways and the
risks associated with such exposures.
As described in EPA's Risk and Technology Review (RTR) Assessment Plan (EPA 2006a),
multipathway human health risks were preliminarily evaluated in 2006. The evaluation used
draft National Emissions Inventory (NEI) data for RTR Phase II (RTR II) source categories and a
simplified multipathway exposure modeling approach (see Appendix 5 in EPA 2006a).1
However, as noted in the RTR Assessment Plan, EPA's intention was to develop an approach
that would supersede the preliminary methods used in 2006 and involve the use of EPA's Total
Risk Integrated Methodology (TRIM), a risk assessment modeling system for air toxics
developed by OAQPS. The TRIM system can be used to predict the local impacts of persistent
and bioaccumulative HAPs (PB-HAPs) from an emissions source to estimate associated human
health risk.
EPA will implement a two-tiered approach to evaluate multipathway exposures and human
health risks for RTR II. In the first tier, a screening evaluation is conducted that focuses on the
identity and magnitude of HAP emissions from a given facility to determine whether a facility
passes certain human health risk-based criteria. Sources that are "screened out" are assumed
to pose no unacceptable risks to human health and are not considered in further analyses. For
sources that do not pass the screen, more refined, site-specific multipathway assessments are
conducted as appropriate. These human health risk results are considered, in combination with
estimated inhalation human health risks, potential ecological risks, and other factors, to support
decisions about residual risk for RTR II source categories.
This current document describes the technical basis for the first, screening-level tier of EPA's
multipathway human health evaluation of RTR emission sources. Specifically, the models,
configurations, and inputs used to derive de minimis emission thresholds in the first tier of the
approach are described in detail here.2 Analyses of the screening scenario conducted to
evaluate the scenario's defensibility are also discussed. EPA expects that refined multipathway
risk assessment methods (when required) will rely on the same TRIM-based modeling approach
used to derive the de minimis thresholds used in the first screening tier. However, the details of
refined assessments will vary depending on the facility location, source category, chemicals
emitted, and other parameters, and the specific methods and processes involved in
multipathway evaluations beyond the first tier are not explored in depth in this document.
1 The preliminary evaluation conducted in 2006 relied on a simpler modeling approach that did not utilize TRIM and
involved a less rigorous analysis of parameter input values.
2 De minimis is a Latin phrase that translates to "regarding minimal things." In the current context, the term de
minimis is used in reference to human health risk that is below a level of concern (or, more specifically, a chemical
emission rate that is not expected to result in unacceptable risks). See also the definition for "risk de minimis" in the
"Glossary for Chemists of Terms Used in Toxicology" (IUPAC 1993).
C-1
-------
C-1.2 Organization of this Document
This document presents the methods, assumptions, and inputs used to develop a method for
evaluation of human multipathway exposures and health risks for RTR II risk assessments.
Section 2 presents an overview of the methods for evaluating multipathway exposures and
risks, a conceptual description of the TRIM-based scenario that is the basis for the screening-
level de minimis emission thresholds, and a brief description of how refined analyses will be
conducted for facility emissions that cannot be screened out in the first step. Section C-3
presents a technical description of the screening-level, TRIM-based modeling scenario and the
configuration of the models used to estimate the de minimis levels. Section C-4 discusses
evaluation activities conducted for this screening scenario and summarizes uncertainty.
References cited in this report are listed in Section C-5.
C-2 Approach for Screening Evaluation of Multipathway Exposures
C-2.1 Overview
As described above, EPA's method for evaluating multipathway exposures for RTR risk
assessments consists of a two-tiered process. Exhibit 2-1 diagrams the approach for evaluating
non-inhalation, multipathway exposures to PB-HAPs. The first tier of this approach is the
screening evaluation that relies on the TRIM-based "screening scenario" as the technical basis
for decisions regarding whether a facility passes the screen. Air toxics emitted by a source
under consideration are reviewed to determine first whether emissions of any PB-HAPs are
reported. If such emissions are reported, the emission rates are compared to available de
minimis threshold emission levels that have been derived using the TRIM-based screening
scenario.3 The list of chemicals that are PB-HAPs is discussed in Section C-2.2, and the use of
de minimis emission thresholds is discussed in Section C-2.3.
The TRIM-based multipathway modeling configuration, referred to in this document as the
"screening scenario," is a key component of the first tier of this approach, as this modeling
application is the technical basis for determining the levels of PB-HAP de minimis emission
thresholds. The term "screening scenario" is used in this document to refer collectively to the
specific TRIM.FaTE and exposure modeling configuration described here, including the set of
assumptions and input values associated with a hypothetical watershed and the exposure and
risk scenarios evaluated for this watershed. The screening scenario is a static configuration,
and its primary purpose is as a modeling tool to calculate the de minimis emission rates for PB-
HAPs of concern. Descriptions of the components of the screening scenario are presented in
Section C-2.4.
The two potential outcomes of the screening human health evaluation are:
. Non-inhalation exposures are unlikely to pose a human health problem; or
. The potential for unacceptable non-inhalation exposures cannot be ruled out and further
assessment is required to determine the potential for unacceptable risk.
An ideal screening approach strikes a balance between being conservative - to ensure that
unacceptable risks are identified, and being accurate, to minimize results suggesting that
additional assessment is required when in fact the actual risk is low. Typically, gains in
3 As described later in this report, to date, TRIM-based modeling has been used to calculate de minimis emission
rates only for those PB-HAPs considered as those of highest concern.
C-2
-------
accuracy in environmental modeling are accompanied by additional resource requirements.
Stated another way, a suitable approach minimizes both false negatives and false positives.
False negatives (i.e., results that suggest that the risk is acceptable when in fact the actual risk
is high) can lead to inappropriate and non-protective health or environmental policy decisions.
False positives (i.e., results that suggest more assessment is required when in fact the actual
risk is low) can result in wasted resources by leading to additional, unnecessary analysis. For
the evaluation of multipathway human health exposures to PB-HAPs, the methods for screening
described in this document are intended to achieve this balance.
Facilities whose emissions exceed the established de minimis emission rate for any PB-HAP
would be subjected to more refined evaluations. Because the initial screening evaluation
enables EPA to confidently eliminate from consideration those facilities where risks from non-
inhalation exposures are projected to be minimal, resources can be targeted toward those
facilities that do not "pass" the screening test. An overview of the anticipated approach to
refined evaluation is described in Section C-2.5.
C-3
-------
Exhibit 2-1. Conceptual Decision Tree for Evaluation of Non-Inhalation
Exposures of PB-HAPs
Diagram Key
Decision
Point
Process
^ Outcome
Decision point
regarding
additional
refinement
(based on
professional
judgment)
<
>
LU
O
LU
o:
o
CO
Evaluate HAP emissions by
facility
Are any PB-HAPs
emitted?
YES
Are PB-HAPs
of primary concern
for RTR emitted?
(Cd, dioxins, Hg,
POM, Pb)?
YES
Does the emission
rate for any PB-
HAP exceed de
minimis levels?
YES:
Conduct more refined
assessment of facility
Facility screens out; no
concern for unacceptable
multipathway risk
Unacceptable risks are not
likely, but evaluate concern on ]
a case-by-case basis
Facility passes screen; no concern for
unacceptable multipathway risk
(check emissions of PB-HAPs other than
Cd, dioxins, Hg, POM, Pb on a case-by-
case basis)
CO
z
o
I—
<
<
>
LU
Q
LU
LU
ec
m ro
Check inputs and assumptions,
-~ refine to incorporate site-
specific information
Are estimated
risks below levels
of concern?
-YES
No concern for
unacceptable multipathway)
risk
f Use risk estimates to \
( inform risk management J
decisions
C-4
-------
C-2.2 Chemicals of Potential Concern
EPA's assessment of multipathway human exposures for RTR focuses on PB-HAPs that
OAQPS has identified as candidates for multipathway risk assessments. OAQPS developed a
list of 14 chemicals and chemical groups that are PB-HAPs based on a two-step process taking
into account the following:
. their presence on three existing EPA lists of persistent, bioaccumulative, and toxic
substances, and
. a semi-quantitative ranking of toxicity and bioaccumulation potential of the entire list of
HAPs.
The list's development and utility in hazard identification for multipathway risk assessment are
further explained in Chapter 14 and Appendix D of Volume I of EPA's Air Toxics Risk
Assessment (ATRA) Reference Library (EPA 2004a). Exhibit 2-2 presents the 14 chemicals
and groups that are PB-HAPs.
The screening scenario described in this document is not configured for evaluating the risk
potential for all 14 PB-HAPs on the list. Currently, the scenario can be used to quantitatively
estimate exposures and risks for four PB-HAP compounds (indicated in bold in Exhibit 2-2).
These compounds are the focus of the current scenario because, based on current emissions
and toxicity considerations, they are expected to pose the vast majority of the non-inhalation
risks to humans from air emissions at sources subject to residual risk provisions of the CAA.4
Exhibit 2-2. OAQPS PB-HAP Compounds a
PB-HAP Compound
Addressed by Screening Scenario?
Cadmium compounds
Yes
Chlordane
No
Chlorinated dibenzodioxins and furans
Yes
DDE
No
Heptachlor
No
Hexachlorobenzene
No
Hexachlorocyclohexane (all isomers)
No
Lead compounds
No
Mercury compounds
Yes
Methoxychlor
No
Polychlorinated biphenyls
No
Polycyclic organic matter (POM)
Yes
Toxaphene
No
Trifluralin
No
a Source of list: EPA 2004a. Compounds in bold text can be evaluated using the current version of the
TRIM-based screening scenario.
4 Potential impacts on human health from non-inhalation exposures to lead are evaluated for RTR using the National
Ambient Air Quality Standard for lead, which takes into account multipathway risks. Non-inhalation exposures to the
other nine PB-HAPs not addressed by the modeling scenario discussed in this report will be evaluated on an
individual facility or source category basis as needed.
C-5
-------
C-2.3 Use of De Minimis Emission Levels
The multipathway screening evaluation for RTR compares reported air emission rates of PB-
HAPs (summed by PB-HAP for each facility) to de minimis emission rates derived using the
screening scenario described here. A de minimis emission rate is the level that, when input to a
risk model that uses emissions as a parameter, corresponds to a specified cancer risk or non-
cancer HQ that, for the purposes of the evaluation being conducted, is assumed to be below a
level of concern. De minimis rates were calculated for the screening scenario for a cancer risk
of 1 in 1 million or an HQ of 1.0 (Exhibit 2-3).5 Conceptually, a de minimis level for the RTR
multipathway screening evaluation could be obtained by back-calculating the emission rate that
results in the specified cancer risk or HQ level, taking into account the exposure and fate and
transport calculations included in the model. Because the models used in this assessment are
not designed to run "backwards," these rates were derived from regression equations
established following a series of TRIM.FaTE and exposure/risk model runs spanning a wide
range of emission rates for each chemical.
Exhibit 2-3. De Minimis Thresholds for Screening of Multipathway Exposures
Chemical
De Minimis
Emission Rate (TPY)
Basis of Threshold
(Type of Health Endpoint)
POM (as Benzo[a]pyrene toxic equivalents)
2.3E-03
Cancer
Dioxins (as 2,3,7,8-TCDD TEQ)
3.2E-08
Cancer
Divalent Mercury
1.6E-01
Non-cancer
Cadmium
6.5E-01
Non-cancer
The more probable risk for each emission rate would be lower than the level corresponding to
the de minimis risk quantities in nearly all circumstances given the conservative and highly
general nature of the screening scenario configuration. This conservatism ensures that a facility
with cancer risk greater than 1 in 1 million or a chronic HQ greater than 1.0 is very unlikely to be
omitted from refined evaluation.
Evaluation of Chemical Groups
In the screening evaluation, emissions of PB-HAPs are summed by chemical group for each
facility. The summed emission rates for each group are then compared to the de minimis
threshold corresponding to the appropriate chemical.
Emissions of polycyclic organic matter (POM, a HAP chemical group that includes polyaromatic
hydrocarbons or PAHs) are often reported in NEI as unspeciated or partially speciated groups
(such as "total PAHs" or "16-PAH") rather than as specific PAH compounds. In addition,
quantitative data are lacking for some POM compounds and groups that are suspected
carcinogens. To evaluate risks associated with exposure to emissions of the various POM
species, EPA has grouped each POM species included in NEI into categories and then defined
a cancer slope factor (CSF) for each group that can be used to estimate lifetime cancer risks.
5 For chemicals that are known to cause both cancer and chronic non-cancer impacts, and for which acceptable
quantitative dose-response values are available for both cancer and non-cancer endpoints, the endpoint that results
in the lower de minimis level will be used for screening (i.e., the threshold will be based on the effect that occurs at
the lower exposure level). For the set of PB-FIAPs for which de minimis levels have been derived, only chlorinated
dibenzo-dioxins and -furans meet both of these criteria. Because the cancer dose-response value is lower than that
for non-cancer effects, the de minimis value is based on the cancer endpoint.
C-6
-------
The CSFs used to evaluate risk for emissions of POM compounds and groups are listed in
Attachment 2.
Similarly, polychlorinated dioxins and furans are reported in NEI as individual congeners,
congener groups, or as toxic equivalents (TEQs) that are scaled relative to the toxicity of
2,3,7,8-TCDD. To evaluate cancer risks and non-cancer hazards from these compounds, EPA
has developed CSFs and reference doses (RfDs) that apply to congener groups. These values
are also presented in Attachment 2.
In the development of the RTR screening thresholds, we determined the de minimis levels for
one compound from each of the POM and dioxin groups (i.e., benzo[a]pyrene and 2,3,7,8-
TCDD). Then, to use this de minimis threshold in an evaluation of facility emissions of POMs or
dioxins, a toxicity-weighted emissions sum is calculated for each group using the toxicity of each
modeled chemical - benzo[a]pyrene and 2,3,7,8-TCDD - to scale the total POM or dioxin group
emissions. This approach avoids the need to develop de minimis emission rates for every POM
and dioxin congener (some of which are not included in the existing TRIM.FaTE algorithm
library). A consequence of this approach is that benzo[a]pyrene and 2,3,7,8-TCDD serve as
fate and transport surrogates. That is, the behavior of these two compounds is assumed to
adequately represent the behavior of all other compounds included in the POM and dioxin PB-
HAP groups, respectively.
De minimis emission thresholds were developed individually for elemental and divalent mercury.
Both were based on the lower of the thresholds associated with multipathway exposures to
divalent mercury and methyl mercury.6 However, only speciated emissions of divalent mercury
are screened because the sum of elemental mercury emissions across all NEI facilities is less
than the elemental mercury de minimis level.
C-2.4 The TRIM-Based Screening Scenario: Basis for De Minimis Emission
Rates
The TRIM-based modeling screening scenario described in this document was used to provide
a means to qualitatively estimate the potential for unacceptable non-inhalation risks for PB-
HAPs emissions from facilities in the context of residual risk assessments conducted as part of
RTR II. The screening scenario used to derive de minimis emission rates is not intended to be
representative of any particular situation. Rather, it was developed for the purpose of RTR to
portray an exposure scenario at least as conservative as any situation that might plausibly be
encountered in the United States. The range of conditions considered when conceptualizing
and building the screening scenario was chosen so that any given individual, long-term
exposure condition for a given geographic region would be reasonably likely to be captured.
These criteria were met by constructing a hypothetical scenario that would be protective in key
aspects, including spatial orientation, meteorology, types of exposures, and ingestion rates.
The overall result is a scenario that is unlikely to occur at any one location but has a high
likelihood of representing the upper end of all potential exposures. This latter aspect
accomplishes the goal of striking a balance between conservatism and accuracy called for in
the ideal screening approach.
For this approach, exposures were modeled for a hypothetical farm homestead and fishable
lake near an emissions source. The hypothetical individual exposed to PB-HAPs in this
6 Note that TRIM.FaTE models the transformation of mercury within the environment; thus, emissions of only divalent
mercury will likewise result in multipathway exposures to both elemental and methyl mercury also. Emissions of only
elemental mercury will result in multipathway exposures to both divalent and methyl mercury also.
C-7
-------
scenario was assumed to derive all potentially contaminated foodstuffs from the farm or lake.
Many of the exposure/activity assumptions were selected from the upper ends of representative
exposure parameter distributions. The physical/chemical environment was parameterized with
a mix of typical values (such as national averages) and health-protective values (i.e., values that
would tend to overestimate concentrations in media). The spatial and temporal aspects of the
scenario and the components of the scenario that influence air concentrations were chosen
from the upper ends of their possible ranges so that concentrations in the environmental media
would not be underestimated given the wide range of possible settings and meteorological
conditions that might be encountered. Chemical-specific and non-chemical-specific properties
of the environmental media were parameterized with either typical or conservative values (with
a greater conservative bias introduced for properties having greater uncertainty).
The development and application of the screening scenario for residual risk evaluations
considered EPA's technical and policy guidelines presented in the Residual Risk Report to
Congress (EPA 1999); Volumes I and II of the Air Toxics Risk Assessment Reference Library
(EPA 2004a, 2005); and other EPA publications. The scenario described in this document is
the culmination of analyses completed over the past 5 years; it provides the basis for an efficient
and scientifically defensible method for screening multipathway human health risk and is a solid
foundation for conducting more refined analyses when necessary. Nevertheless, this scenario
should not be considered "final" but rather a product that can continue to evolve based on
feedback from the scientific community and Agency reviewers, lessons learned as the scenario
is further applied for RTR, variations in EPA's needs and requirements, and other factors.
C-2.4.1 Exposure Routes Evaluated
The screening scenario is intended to address non-inhalation exposures (inhalation exposures
are being evaluated separately for RTR II using a dispersion modeling approach to estimate
ambient air concentrations). The quantitative aspects of this non-inhalation screening
evaluation for human exposures focus primarily on human exposures via the following ingestion
pathways:
. Incidental ingestion of soil,
. Ingestion of homegrown produce,
. Ingestion of homegrown beef,
. Ingestion of milk from homegrown cows,
. Ingestion of homegrown poultry and eggs,
. Ingestion of homegrown pork, and
. Ingestion of fish.
Non-inhalation exposure to PB-HAPs also can occur by way of the dermal pathway (e.g.,
through incidental contact with PB-HAP-contaminated soil). However, dermal absorption of
chemicals that are originally airborne is generally a relatively minor pathway of exposure
compared to other exposure pathways (EPA 2006, CalEPA 2000). The risk from dermal
exposure in the environmental setting from airborne toxicants is expected to be a fraction of the
risk from inhalation exposure or exposure via ingestion of contaminated crops, soil, or breast
milk, for example (CalEPA 2000). Preliminary calculations of estimated dermal exposures and
risk of PB-HAPs, presented in Section C-3.2.4, showed that the dermal exposure route is not a
significant risk pathway relative to ingestion exposures. Assessment of dermal exposure
through incidental contact with soil could be conducted on facilities that require refined
evaluation following the screening evaluation if deemed necessary. Procedures for estimating
dermal absorption from soil would be based on EPA's dermal exposure assessment principles
C-8
-------
and applications (EPA1992b) and EPA's Superfund Human Health Evaluation Manual (EPA
2004c).
Another ingestion pathway - through breast milk by nursing infants - can also be of concern for
chlorinated dibenzo-dioxins and -furans (typically referred to collectively as "dioxins,"
nomenclature that is used elsewhere in this document when referencing the collective chemical
category), and may also be of concern for mercury. Algorithms have been developed for
calculating the exposure and risk associated with dioxin contamination of breast milk and are
used to evaluate the likelihood of developmental effects resulting from exposure to dioxins via
this pathway. Assessment of breast milk exposure for nursing infants will be assessed when
refined evaluations are conducted; this exposure pathway is not incorporated in the calculation
of the de minimis levels.
One other non-inhalation exposure route discussed in ATRA Volume I of possible concern for
PB-HAPs is ingestion of drinking water from surface water sources. This exposure route,
however, is not evaluated in the current assessment. The drinking water exposure pathway is
not likely for the modeling scenario developed for this analysis because the likelihood that
humans would use a lake as a drinking water source was assumed to be low.7
C-2.4.2 Approach to Configuration and Parameterization
This screening scenario is intended to reduce the possibility that EPA would not mis-identify
unacceptable risks. Although the "conservative" approach likely overestimates risk, EPA
determined that conservatism is appropriate for the purposes of screening assessments. As in
the preliminary multipathway screening for RTR conducted in 2006 (EPA 2006a), exposures
were modeled for a hypothetical farm homestead and fishable lake located adjacent to an
emissions source. The hypothetical individual for which exposures were calculated was
assumed to derive all potentially contaminated foodstuffs from these adjacent locations, and
many of the exposure/activity assumptions (e.g., amount of food consumed per day) were
selected from the upper ends of representative exposure parameter distributions.
The physical/chemical environment represented in the screening scenario was parameterized
with two types of values. One type is typical values, such as national averages. The second
type is health-protective, conservative values, or values that would tend to overestimate
concentrations in media driving ingestion exposures for humans, based on knowledge of
exposure patterns. In general, the spatial and temporal aspects of the scenario and the
components of the scenario that influence air concentrations and deposition rates (which in turn
affect all other exposures) were defined to be relatively conservative. That is, they were chosen
from the upper ends of their respective possible ranges so that the wide range of possible
physical settings and meteorological conditions would be captured. Chemical-specific and non-
chemical-specific properties of the environmental media were parameterized with either typical
or conservative values; properties having greater uncertainty were assigned greater
conservative bias.
The spatial layout of the scenario and the meteorological data (or a combination of these two
factors) are generally more influential than physical/chemical parameters in dictating the
screening model outcomes, taking into account the potential range of variation in possible
values. For example, where and how the layout is spatially oriented relative to the dominant
7 An exception to this generality would be reservoirs used for drinking water supplies. This situation may be worthy of
additional analysis, if warranted by the characteristics of a given assessment (e.g., to estimate PB-HAP
concentrations in treated drinking water derived from reservoirs).
C-9
-------
wind direction can dramatically affect the concentrations in air, thereby driving estimated
concentrations of PB-HAPs in soil, water, and biota. In contrast, a relatively large change in soil
characteristics within the range of possible values (e.g., organic carbon content, water content)
might result in relatively small changes in outputs.
The mix of conservative and typical approaches and values is expected to result in a scenario
configuration that, on average, is likely to over-predict environmental concentrations of PB-
HAPs in media of interest for this evaluation. Given the intended application of this scenario as
a screening tool, this conservative bias was deliberate, because of the desire to ensure that
unacceptable risks are not overlooked (i.e., to minimize false negatives). Although the inclusion
of typical values where warranted is intended to minimize the number of false positives, some
false positives are to be expected from a screening scenario. These false positives would be
addressed in iterations of the refined evaluation for a particular source.
C-2.4.3 Modeling Framework
The approach for risk screening (and ingestion exposure) evaluation described here can be
divided into four steps:
1. Fate and transport modeling of PB-HAPs emitted to air by the source that partition into
soil, water, and other environmental media (including fish);8
2. Modeling of transfer and uptake of PB-HAPs into farm food chain media (produce,
livestock, dairy products) from soil and air;
3. Estimating ingestion exposures as a result of contact with the various selected media
and estimating average daily ingestion doses for a hypothetical human receptor; and
4. Calculating lifetime cancer risk estimates or chronic non-cancer HQs, as appropriate, for
each PB-HAP and comparing these metrics to selected risk management points of
departure used in the RTR II.
The relationship among these four processes is shown in Exhibit 2-4.
EPA's TRIM methodology was conceived as a comprehensive modeling framework for
evaluating risks from air toxics, and the TRIM system was designed to address each of the four
steps involved in screening ingestion risk.9 Currently, however, only one component
corresponding to the first step included in Exhibit 2-4 - the fate and transport module - is
available for application in an ingestion risk assessment. EPA has completed some
development activities for TRIM.Expo-lngestion and TRIM.Risk-Human Health, two additional
modules that cover the other three steps. Modeling software, however, is not currently available
for these modules. For the RTR screening scenario, the Multimedia Ingestion Risk Calculator
(MIRC), a Microsoft Access-based computer framework, was constructed to complete the
calculations required for estimating PB-HAP concentrations in farm food chain media, average
8 As discussed below, concentrations in fish calculated by the TRIM.FaTE model were used in the current approach
to estimate ingestion exposures for humans consuming fish. Modeling offish concentrations is therefore discussed in
this document as part of the fate and transport modeling. Uptake of PB-HAPs into all other biotic media assumed to
be ingested is modeled in the second step.
9 Information regarding the current status of TRIM modules as well as comprehensive documentation of modules that
have been developed thus far can be accessed on EPA's Technology Transfer Network (TTN) on the Fate,
Exposure, and Risk Analysis web site (http://www.epa.gov/ttn/fera/).
C-10
-------
daily ingestion doses, and cancer risks and chronic non-cancer HQs. This framework is
conceptually identical to the ingestion exposure and risk analyses that TRIM is intended to
cover.
Exhibit 2-4. Overview of Ingestion Exposure and Risk Screening Evaluation Method
Chemical Emissions to Air
'—
—\
Chemical fate and
transport:
Physical environment
and aquatic ecosystem
TRIM.FaTE
/
¦s
Uptake 8c transfer
into produce and
livestock
Human
ingestion
exposure
Risk 6t hazard
estimation
Multimedia Ingestion Risk Calculator (MIRC)
«. ji
Cancer Risk
Hazard Quotient
C-2.4.3.1 Fate and Transport Modeling
The fate and transport modeling step depicted in the first box in Exhibit 2-4 is implemented for
RTR using the Fate, Transport, and Ecological Exposure module of the TRIM modeling system
(TRIM.FaTE).10 In developing the screening scenario, Version 3.6.2 of TRIM.FaTE was used to
model the fate and transport of emitted PB-HAPs and to estimate HAP concentrations in
relevant media. Additional information about TRIM.FaTE, including support documentation,
software, and the TRIM.FaTE public reference library, is available on EPA's TTN at
http://www.epa.gov/ttn/fera/.
The algorithms used to model mercury species and PAHs are described in Volume II of the
TRIM.FaTE Technical Support Document (EPA 2002a). A comprehensive evaluation of the
performance of TRIM.FaTE for modeling mercury was documented in Volumes I and II of the
TRIM.FaTE Evaluation Report (EPA 2002b, 2005b). Algorithms specific to the fate and
transport of 14 chlorinated dibenzo-dioxin and -furan congeners were added following the
addition of those for mercury and PAHs. Documentation of the application of TRIM.FaTE for
dioxin emissions is contained in the third volume of the TRIM.FaTE Evaluation Report (EPA
2004b). Parameterization of the TRIM.FaTE library used for RTR analyses with regard to
10
TRIM.FaTE is a spatially explicit, compartmental mass balance model that describes the movement and
transformation of pollutants over time, through a user-defined, bounded system that includes both biotic and abiotic
compartments. Outputs include pollutant concentrations in multiple environmental media and biota, which provide
exposure estimates for ecological receptors (i.e., plants and animals). The output concentrations from TRIM.FaTE
are also intended to be used as inputs to a human ingestion exposure model to estimate human exposures.
Significant features of TRIM.FaTE include: (1) a fully coupled multimedia model; (2) user flexibility in defining
scenarios, in terms of the links among compartments, and number and types of compartments, as appropriate for the
application spatial and temporal scale; (3) transparent, user-accessible algorithm and input library that allows the user
to review and modify how environmental transfer and transformation processes are modeled; (4) a full accounting of
all of the pollutant as it moves among environmental compartments during simulation; (5) an embedded procedure to
characterize uncertainty and variability; and (6) the capability to provide exposure estimates for ecological receptors.
C-11
-------
dioxins is identical to the configuration described in that third evaluation report. More recently
(largely as part of this current project), the TRIM.FaTE public reference library has been
updated to include information on modeling for cadmium. In general, many of the algorithms
and properties included in the public reference library that are used to model mercury (except
for the mercury transformation algorithms) are also applicable to cadmium. Comprehensive
technical documentation of TRIM.FaTE algorithms specific to cadmium has not yet been
compiled; however, all chemical-specific properties used by TRIM.FaTE to model cadmium (as
well as PAHs, mercury, and dioxins) are documented in Attachment 1 to this document.
Parameterization of the TRIM.FaTE scenario used for RTR screening is described in more
detail in Section C-3.
C-2.4.3.2 Exposure Modeling and Risk Characterization
The algorithms included in MIRC that calculate chemical concentrations in farm food chain
media and ingestion exposures for hypothetical individuals were generally obtained from EPA's
Human Health Risk Assessment Protocol for Hazardous Waste Combustion Facilities, or
HHRAP (EPA 2005a).11 These algorithms, and the required exposure factors and other
parameter values, were compiled into a database. An overview of the computational processes
this tool carries out and the types of input data it requires is presented in Exhibit 2-5. This
exhibit demonstrates the general relationships between the relevant TRIM.FaTE outputs (i.e.,
chemical concentrations in environmental media and fish) and the ingestion exposure and risk
calculations carried out using MIRC. Additional discussion of exposure and risk calculations for
this screening scenario is presented in Section C-3.2 and Attachment 2, and all inputs required
by these calculations are documented in Attachment 2.
Exhibit 2-5. Overview of Process Carried Out in the Multimedia Ingestion Risk Calculator
TRIM.FaTE
outputs
f
X
Chemical-
specific uptake/
transfer factors
Farm Food Chain
Concentration
Calculator
FFC media
concentrations
Ingestion
Exposure
Calculator
Average daily
doses
Risk and Hazard
Calculator
Cancer risks and
hazard quotients
Ls_ ^
Plant- and
animal-specific
parameters
Human activity/
exposure factors
Ingestion
dose-response
values
Access db
Tool
Key to symbols:
process
output
Input 7
data /
11 The farm food chain calculations and ingestion exposure equations to be included in the TRIM.Expo software are
expected to be very similar to those included in HHRAP.
C-12
-------
C-2.5 Refined Analyses
For facilities that are not screened out by comparison to the de minimis emission thresholds,
additional analysis is required to evaluate the risks via multipathway exposures. EPA envisions
that the screening scenario provides a technically defensible "starting point" for additional fate
and transport and exposure/risk analyses, by substituting site-specific values for key parameters
used in the modeling scenario. As appropriate, the scenario can be further refined in iterative
steps until either the risks predicted are not of concern or sufficient information has been
gathered to inform a risk management decision. Based on the model performance evaluations
completed on the screening scenario, settings and parameters that might be appropriate for
revision in a more refined analysis include meteorological inputs and spatial configuration
assumed for TRIM.FaTE modeling, relevant ingestion exposure scenarios (based on
surrounding land-use or other characteristics), and exposure factors such as ingestion rate for
individual food types, among other inputs. The refined evaluation could eventually involve
developing a site-specific TRIM.FaTE application that incorporates significant site-specific data
(e.g., a model application that includes a site-specific spatial layout taking into account local
geographic features and environmental parameter values selected from the best available data
for that location).
C-3 Description of Modeling Scenario
C-3.1 TRIM.FaTE Scenario Configuration and Parameterization
As described in Section C-2.4.2, the physical configuration of the RTR Screening Scenario was
designed to be generally conservative, and the environmental and chemical-specific properties
were parameterized with either conservative or central-tendency values. Information regarding
the scenario configuration and important aspects of the parameterization process, justifications
for selecting particular property values, and certain uncertainties is presented in the sections
that follow. Comprehensive documentation of TRIM.FaTE property values for this scenario is
provided in Attachment 1.
C-3.1.1 Chemical Properties
The chemical/physical properties that TRIM.FaTE requires, such as Henry's law constant,
molecular weight, and other "general" parameters, were obtained from peer-reviewed and
standard reference sources. Numerous other chemical-specific properties are related more
specifically to a particular abiotic or biotic compartment type; these properties are discussed
generally in the sections that follow and are documented in Attachment 1.
C-3.1.2 Spatial Layout
For the purpose of estimating media concentrations, the TRIM.FaTE scenario is intended to
represent a farm homestead and a fishable lake (and its surrounding watershed) located near
the emissions source of interest. A diagram of the surface parcel layout is presented in Exhibit
3-1. The source parcel is parameterized as a square with sides of 250 m, which is assumed to
be a fair estimation for the size of a relatively small-to-medium facility at the fence line. With a
predominant wind direction toward the east, the modeled layout is generally symmetric about an
east-west line and is wedge-shaped to reflect Gaussian dispersion of the emission plume.
C-13
-------
Exhibit 3-1. TRIM.FaTE Surface Parcel Layout
] Vegetation (Grasses & Herbs)
| Vegetation (Coniferous Forest)
—*¦ Runoff
¦ Tilled Soil
] Vegetatioi
Source
3.5 km
10 km
Figure not drawn to scale.
A lateral, downwind distance of 10 km was established for the watershed included in the
scenario. Based on the results of dispersion modeling (results not presented here), the location
of the maximum air concentration and deposition rate would be expected to occur relatively
close to the facility (probably within a few hundred meters, with the exact location varying with
stack height and other parameters) and well within a 10-km radius. Additionally, deposition
rates for the PB-HAPs for which this screening scenario is applicable would be expected to
decrease by about two orders of magnitude relative to the predicted maximum rate within a
10-km radius. Extending the modeling layout beyond a 10-km downwind distance would
increase the amount of deposition "captured" by the modeled watershed, but the incremental
chemical mass expected to accumulate in the watershed diminishes rapidly with distance. In
addition, the impact of this additional deposited mass on ingestion exposures is expected to be
negligible.12 Given these conditions, a downwind length of 10 km was determined to be
appropriate for the screening scenario.
The north-south width of the wedge-shaped watershed was set based on the observed behavior
of chemicals emitted to the ambient air. If meteorological stability is known or can be assumed,
the lateral spread of the plume (ay, measured from the centerline) at a certain distance from the
source can be estimated using the Pasquill-Gifford curves. Turner (1970) derived the equations
for these curves, which can be found in the ISC3 Dispersion Model Manual (among other
sources).13 For a relatively neutral atmosphere (stability class D), a at 10 km is about 550 m
using this estimation. In a Gaussian distribution, about 99.6 percent of the plume spread area is
contained within 3a of the median line. Therefore, the plume a was set at 3 times 550 m, or
approximately 1.75 km from the centerline at 10 km distance. The plume width for these
conditions is expected to be about twice this distance, or 3.5 km. These dimensions were used
to define the dimensions of the overall air and surface parcel layouts for the screening scenario.
Mass deposited at the outer edge of the watershed is expected to result in only a very small increase in estimated
exposure via fish consumption by increasing the chemical mass transported to the lake through erosion and runoff.
The distance from these more distant locations to the lake would attenuate transport of chemical mass by erosion
and runoff, dampening the effect of including additional deposition beyond 10 km. (Other exposure pathways would
be largely unaffected; the soil concentrations used to calculate exposures for the farm food scenario are derived from
soil parcels located close to the source and unaffected by deposition to the far reaches of the watershed.)
13 http://www.epa.gov/scram001/userg/regmod/isc3v2.pdf
C-14
-------
The surface (land and surface water) modeling area was initially divided into five pairs of parcels
whose areas increase with distance from the source, which approximately corresponds to the
spatial gradient that is expected in the downwind direction from the source. The second north
parcel from the source was further divided into two parts, one of them tilled soil (Parcel N6) to
represent agricultural conditions.
The depth of the surface soil compartments was set to 1 centimeter (cm), except for Parcel N6,
for which the depth was set to 20 cm to simulate the effect of tillage. Characteristics of the soil
layers (e.g., organic carbon content, air and water content, and sub-soil depth) were typically set
to represent typical or national averages as summarized by McKone et al. (2001), for example.
Initial considerations when the layout was configured included the presence of a stream that ran
along the bisecting east-west line from the southwestern corner of Parcel N3 through the
eastern edge of the layout. In that configuration, the eastern extent of the pond was restricted
by a parcel (S3) directly south of Parcel N3. The stream received chemical mass from Parcels
S3, S4, and S5 and flowed directly into the pond. Preliminary modeling runs showed that the
existence of a stream somewhat decreased the concentrations of 2,3,7,8-TCDD in the pond
parcel and significantly decreased concentrations in Parcels S3 and S4. These results
indicated that the pond was receiving more 2,3,7,8-TCDD mass through surface soil transfer
than through the stream, perhaps due to a chemical sink into stream sediment. Given the goal
of creating a conservative scenario, a stream was not included in the final layout.
The overall shape and boundaries of the air parcel layout mirror those of the surface parcel
layout. A single air parcel (N2) overlies surface Parcels N6 and N7, and the air over the lake is
divided into air Parcels S2 and S3 (mirroring the analogous parcels on the north side of the
lake).
C-3.1.3 Watershed and Water Body Parameterization
Properties associated with the watershed soil and lake determine how pollutants in the system
are transported through and accumulate in various compartments. These properties describe
the physical characteristics of the environmental media included in the modeled region, as well
as the assumed connections and relationships between media types and modeled spatial
components that in turn affect chemical transport via water runoff, ground seepage, deposition
of suspended sediments in the water column, and other processes. This section presents the
justification for setting the key properties of the soil, water, and sediment compartments. Also
discussed are some of the chemical properties related to watershed and waterbody processes
(chemical-specific compartment properties in TRIM.FaTE) and the configuration of terrestrial
plants included in the scenario.
C-3.1.3.1 Water Balance
Water-related properties of the lake and related watershed characteristics (e.g., runoff rates
from each surface soil compartment) were set so that a simplified water balance is achieved.
Although TRIM.FaTE maintains a chemical mass balance, the model does not calculate or
maintain media mass balances (e.g., for water) except where specified in certain formulas. For
the screening scenario, the parameters were set to satisfy two equations relating water volume.
The first equation maintains a balance of water entering and leaving the terrestrial portion of the
scenario:
C-15
-------
[total precipitation] = [evapotranspiration] + [total runoff]
In this equation, total runoff is equal to the sum of overland runoff to the lake and seepage to the
lake via groundwater.
The second equation describes the volumetric balance of transfers of water to and from the
lake:
[total runoff] + [direct precipitation to the lake] = [evaporation from the lake surface] + [outflow
from the lake]
Note that TRIM.FaTE actually uses only some of these properties (e.g., precipitation rate and
surface runoff, but not evapotranspiration). The water characteristics assumed for the
screening scenario are meant to represent a relatively wet and moderately warm location in the
United States (USGS 1987). Following are the assumptions for this scenario:
. 35 percent of the total precipitation leaves the scenario through evapotranspiration.
. 65 percent of total precipitation remains in the modeled system and contributes to total
runoff.
. Total runoff is divided between overland runoff and seepage to groundwater as follows:
o 40 percent of total precipitation contributes to overland runoff,
o 25 percent of total precipitation seeps into the groundwater and eventually flows
into the lake.
For these calculations, the source parcel was considered to be outside the watershed and
therefore was not included in the water balance. The evaporation rate from the lake was
assumed to be 700 millimeters per year (mm/yr) based on data reported by Morton (1986) for
various lakes. This estimate is probably more representative of cooler locations [by
comparison, the overall average of evaporation rates from various reservoirs is reported by
McKone et al. (2001) to be close to 1,200 mm/yr]. The runoff rate was defined to be spatially
constant and temporally constant (i.e., it is not linked to precipitation events) throughout the
modeled domain. Based on these assumptions, the outflow of water from the lake is about 18
million m3/yr, which translates to a volumetric turnover rate of about 12.2 lake volumes per year.
Other quantitative water body and watershed characteristics TRIM.FaTE uses are listed in
Attachment 1.
C-3.1.3.2 Sediment Balance
A simplified balance of sediment transfers between the watershed and the lake was also
maintained for the screening scenario via the parameterization of sediment-related properties.
As with water, the model does not internally balance sediment mass; these calculations were
performed externally for the purposes of setting parameter values. The sediment balance
maintained is described by the following equation, where terms represent mass of sediment:
[total surface soil transfers to the lake via erosion] =
[removal of sediment from the water column via outflow] + [sediment burial]
where the second term (removal of sediment from the water column via outflow) is accounted
for in TRIM.FaTE by lake flushing rate and the third term (sediment burial) is the transfer of
sediment from the unconsolidated benthic sediment compartment to the consolidated sediment
layer.
C-16
-------
To maintain the sediment balance, erosion rates were calculated for each surface soil
compartment using the universal soil loss equation (Wischmeier and Smith 1978), assuming a
relatively high rate of erosion. The total suspended sediment concentration is assumed to
remain constant in TRIM.FaTE, and the flush rate of the lake (calculated via the water balance
approach described above) was then used to estimate the removal of sediment from the
modeling domain via lake water outflow. The difference between these sediment fluxes was
taken to be the sediment burial rate. The sediment burial rate is the rate at which sediment
particles in the unconsolidated benthic sediment layer are transported to the consolidated
sediment, where the particles can no longer freely interact with the water column.
In TRIM.FaTE, the consolidated sediment layer is represented with a sediment sink; as with all
sinks in TRIM.FaTE, chemical mass sorbed to buried sediment that is transported to the sink
cannot be returned to the modeling domain. The burial rate is a formula property calculated by
the model according to the difference between user-specified values for sediment deposition
velocity (from the water column to the benthic sediment) and sediment resuspension velocity
(back into the water column from the benthic sediments). These formula properties assume a
constant volume of particles in the sediment layer (because the densities for benthic and
suspended sediment particles were defined to the same value, the mass of particles in the
sediment is also constant).
For the screening scenario described here, the average sediment delivery rate (i.e., transfer of
sediment mass from watershed surface soil to the lake due to erosion) for the entire watershed
was estimated to be about 0.0036 kilograms per square meter per day (kg/m2-day), based on
calculations using the universal soil loss equation (USLE). The HHRAP documentation notes
that using the USLE to calculate sediment load to a lake from the surrounding watershed
sometimes leads to overestimates (EPA 2005a). For this screening scenario, however, this
approach was considered to be appropriate in that conservatism is a goal of the screening
scenario.14 Surface soil compartments adjacent to the lake are linked directly to the lake for the
purposes of estimating erosion and runoff transfers (see layout in Exhibit 3-1). Erosion and
runoff from the source parcel are linked directly to a sink and do not enter the screening
scenario lake. The transport of sediment to the lake via overland flow (e.g., by streams) is thus
assumed to be efficient. Note that erosion from parcels not directly adjacent to the lake is
assumed to be somewhat attenuated, effected by using a lower sediment delivery ratio in the
model.
Using the calculated surface soil erosion rates for the scenario, the total average daily sediment
load to the lake from the watershed is about 16,600 kg/day. About 15 percent of this load is
removed from the lake via outflow of suspended sediments (based on a calculated flush rate of
12.2 volume turnovers per year), with the remainder of the sediment input to the lake transferred
to the sediment burial sink.
C-3.1.4 Meteorology
Meteorological properties used in TRIM.FaTE algorithms include air temperature, mixing height,
wind speed and direction, and precipitation rate. These properties, which can vary significantly
among geographic locations, as well as seasonally and hourly for a single location, greatly
influence the chemical concentrations predicted in media of interest. Because the screening
scenario is intended to be generally applicable for any U.S. location, and to minimize the
14 Based on sensitivity analysis, a higher erosion rate will both increase surface water concentrations and decrease
surface soil concentrations; however, the relative impact on resulting concentrations will be proportionally greater in
the waterbody.
C-17
-------
frequency of false negatives, a conservative configuration was used. The meteorology of the
screening scenario was defined to ensure that (when used in combination with the selected
spatial layout) the maximum exposures that might be encountered for the scenarios of interest
would be encompassed (i.e., consumption of home-grown farm foodstuffs and self-caught fish,
with all farm foods and fish obtained from locations impacted by chemicals emitted from the
local source). However, ensuring that the meteorological parameters were not overly
conservative, such as always having the wind blow toward the location of interest, was also
important to avoid too many false positives.
The meteorological data for the screening scenario are intended to be representative of a
location with a low wind speed, a wind direction that strongly favors the watershed, and a
relatively high amount of total precipitation falling on the watershed. The values used were
based on actual data trends for U.S. locations as specified in Exhibit 3-2; however, an artificial
data set was compiled (for example, temporally variable meteorological parameters were made
to vary only on a daily basis). This simplified approach allowed for greater control (relative to
selecting a data set for an actual location) so that desired trends or outcomes could be
specified. Also, using a meteorological data set with values varying on a daily basis rather than
a shorter period (such as hourly, which is the typical temporal interval for meteorological
measurements) reduced required model run time. Meteorological inputs are summarized in
Exhibit 3-2.
The sensitivity of modeled 2,3,7,8-TCDD concentrations to changes in these meteorological
variables was tested. Lower wind speeds and mixing heights affected concentrations the most.
This sensitivity is not unexpected because lower wind speeds should increase pollutant
deposition into the soil and lower mixing heights should reduce the volume through which
pollutants disperse. The wind speed used for the screening scenario was 2.8 m/s, the 5th
percentile of annual average among 239 stations; by comparison, the mean annual average
wind speed is 4.0 m/s in the contiguous United States). The mixing height (mean heights from
four states) used was 710 m (the 5th percentile of annual average among all 40 states in the
SCRAM database).
C-18
-------
Exhibit 3-2. Summary of Key Meteorological Inputs
Meteorological
Property
Selected Value
Justification
Air temperature
Constant at 298 Kelvin
Recommended default value listed in HHRAP (EPA 2005a).
Value is similar to the mean daily June temperature in the
U.S. Deep South and to the mean daily July temperature in
the U.S. Central Plains.3
Mixing height
Constant at 710 meters
Value is 5th percentile of annual average mixing heights for
463 U.S. locations, using data obtained from EPA's
SCRAM Web site.b Value is the approximate U.S. median
for periods without precipitation, based on data compiled by
Holzworth (1972). Value is conservative compared to the 1-
to 2-km typical mid-latitude daytime value (Stull 1988).
Wind direction
Blows from source parcel into
scenario domain (west to
east) 3 days per week; during
other times does not blow into
domain
A wind direction that favors the location of interest (for
example, a watershed downwind of a source of concern) will
tend to result in more emitted mass accumulating in the
location of interest. For much of the U.S. mid-Atlantic and
western regions, the wind tends to favor one direction
(eastward).d For the hypothetical RTR scenario, a more
extreme example of this pattern is represented by
conditions in Yakima, Washington, where the wind blows
eastward approximately 40 percent of the time based on a
review of wind direction data compiled by the National
Weather Service (NCDC 1995). This pattern was
approximated in the RTR scenario with a configuration in
which the modeled domain is downwind of the source three
days out of seven.
Horizontal wind
speed
Constant at 2.8 meters per
second
Set to 5th percentile of annual average speed for 239
stations across the contiguous United States (about 50
years of data per station). Value is similar to the annual
average wind speeds of the U.S. Deep South.0
Precipitation
frequency
Precipitation occurs 3 days
per week; wind direction
blows into domain 2 of these
days
This value was selected so that two-thirds of the total
precipitation occurs when the domain is downwind of the
modeled source. This pattern approximates that for rainy
U.S. locations, where precipitation occurs 35 - 40% of the
time (Flolzworth 1972). These locations include parts of the
U.S. Northeast and Northwest.0
Total
Precipitation
1.5 meters per year
Assumed to represent rainy conditions for the United States.
This annual precipitation amount is experienced in parts of
the U.S. Deep South and parts of the U.S. Northwest
Coast.d Conditional precipitation rate (rainfall rate when
precipitation is occurring) is 9.59 mm per day, which is
similar to conditions along the U.S. East Coast and
Midwest.0
a National Climatic Data Center Historical Climate Series (NCDC-HCS) (2007).
http://www5.ncdc.noaa.gov/climatenormals/hcs/HCS_MAP_7100.pdf
b Support Center for Regulatory Atmospheric Modeling; http://www.epa.gov/scram001/tt24.htm.
c National Climatic Data Center CliMaps (NCDC-CliMaps) (2007). http://cdo.ncdc.noaa.gov/cgi-bin/climaps/climaps.pl
C-19
-------
C-3.1.5 A quatic Food Web
The aquatic food web is an important part of the screening scenario because the chemical
concentrations modeled in fish are used to calculate human ingestion exposure and risks
associated with eating contaminated fish. A biokinetic approach to modeling bioaccumulation in
fish is used in the RTR screening scenario. The primary producers (first trophic level) in the
TRIM.FaTE aquatic ecosystems are algae and macrophytes. The scenario includes a benthic
invertebrate compartment to represent the primary invertebrate consumers (second trophic
level) in the benthic environment, and the fish compartments represent the higher tropic levels
in the aquatic system. For TRIM.FaTE to provide reasonable predictions of the distribution of a
chemical across biotic and abiotic compartments in aquatic systems, the biomass of the aquatic
biotic compartments must represent all biota in the system and the distribution of biomass
among the trophic levels and groups must be as realistic as possible.
To support the development of a relatively generic freshwater aquatic ecosystem in which to
model bioaccumulation in fish, a literature search, review, and analysis was conducted in
support of developing and parameterizing aquatic biotic compartments for TRIM.FaTE (ICF
2005). This research demonstrated that the diversity of species and food webs across U.S.
aquatic ecosystems is substantial, reflecting the wide range of sizes, locations, and
physical/chemical attributes of both flowing (rivers, streams) and low-flow (ponds, lakes,
reservoirs) waterbodies. In general, lotic bodies of water (lakes and ponds) are at a higher risk
of accumulating contaminants in both sediments and biota than are flowing systems (rivers,
streams). Also, the previous research (ICF 2005) suggested that a lake of at least 60 hectares
(ha) likely would be sufficient to support higher trophic level predatory fish, with some fraction of
their diet comprising smaller fish.
The RTR screening scenario includes a generic aquatic ecosystem with a 47-ha lake. Although
slightly smaller than the size suggested by the previous review (ICF 2005), a 47-ha lake is large
enough to support higher trophic level fish given the appropriate conditions. Also, this size was
compatible with the overall size of the defined watershed. In the lake ecosystem defined for the
screening scenario, benthic invertebrates are an important food source for a large proportion of
the total fish biomass. The fish types, biomass, diet fractions, and body weights recommended
for fish compartments for the screening scenario of TRIM.FaTE are listed in Exhibit 3-3.
Biomass is based on an assumption that the total fish biomass for the aquatic ecosystem is 5.4
grams per square meter expressed as a wet weight (gw/m2, ICF 2005).
A strict piscivore compartment was not selected for the screening scenario because such
species are rare in lakes of small to moderate size. In general, the food web implemented in the
screening scenario is intended to be generally applicable across the United States and is
intended to be generally conservative (to simulate a food web that maximizes bioaccumulation).
C-3.1.6 Using TRIM.FaTE Media Concentrations
The screening scenario outputs include average PB-HAP concentrations and deposition rates
for each year and for each parcel of the model scenario. In each surface parcel, soil
concentrations are provided for the surface, root, and vadose zones and grass or leaf
concentrations as appropriate for the plants. Groundwater concentrations and deposition rates
to the soil are also provided. For each air parcel, air concentrations are provided. For the lake,
surface water concentrations and concentrations in the various levels of the aquatic food chain
are included. For the ingestion exposure calculations, some concentrations are used to
calculate direct exposure, and some are used to perform the farm food chain concentration
calculations in the various media that humans can ingest (see Exhibit 2-4).
C-20
-------
Exhibit 3-3. Parameters for Aquatic Biota for the Screening Scenario of TRIM.FaTE
TRIM.FaTE
Compartment
Type
Organisms
Represented by
Compartment
Biomass
Diet
Average
Body Weight
(kg)
Areal density
(gw/m2)
Fraction of
Total Fish
Biomass
Macrophyte
hydrilla, milfoil
500
--
--
--
Benthic
invertebrate
aquatic insects,
crustacean,
mollusks3
20
--
Feeds in sediment
0.000255
Water column
(WC) herbivore
young-of-the-
year, minnows
1.0
18.5%
100% algaeb
0.025
Water column
omnivore
sucker, carp
1.0
18.5%
40% WC herbivore
30% benthic invert.
30% macrophyte
0.25
Water column
carnivore
largemouth bass,
walleye
0.4
7.5%
60% WC omnivore
20% WC herbivore
20% benthic omniv.
2.0
Benthic
omnivore
small catfish,
rock bass
2.0
37%
100% benthic invert.
0.25
Benthic
carnivore
large catfish
1.0
18.5%
70% benthic invert.
30% benthic omniv.
2.0
Total Fish Biomass c
5.4
a Benthic invertebrates include aquatic insects (e.g., nymphs of mayflies, caddisflies, dragonflies, and other species
that emerge from the water when they become adults), Crustacea (e.g., amphipods, crayfish), and mollusks (e.g.,
snails, mussels).
b Algae is modeled as a phase of surface water in TRIM.FaTE.
c Total fish biomass does not include macrophytes or benthic invertebrates.
Regardless of whether the concentration and deposition values are used to calculate ingestion
directly or are used in farm food chain calculations, selecting the parcel that is the source of the
values used as inputs to succeeding calculations is necessary. The locations that determine
direct and indirect exposures were selected assuming generally conservative assumptions. In
general, decisions regarding which TRIM.FaTE outputs to use in calculating exposures for the
hypothetical scenario assume exposure at locations near the modeled source, thereby resulting
in higher exposures to emitted chemicals. These assumptions are summarized in Exhibit 3-4.
TRIM.FaTE can output instantaneous chemical concentrations for a user-specified time step
and also can be configured to calculate temporal averages (e.g., annual averages). For the
screening scenario, the model is set up to output results on a daily basis, largely because daily
is the smallest time step over which input data change (i.e., wind direction and precipitation
rate). Daily concentration results were averaged to obtain annual average concentrations. The
default assumption is annual average concentrations for media during the fiftieth year of
emissions. For the chemicals modeled in this scenario, long-term concentrations in
environmental media will be relatively constant at 50 years (most of the chemicals modeled for
RTR approach steady state well before 50 years).
C-21
-------
Exhibit 3-4. Spatial Considerations - TRIM.FaTE Results Selected for Calculating
Farm Food Chain Media Concentrations and Receptor Exposures
TRIM.FaTE Output Used in Exposure
Calculations
Representative Compartment
Concentration in air, for uptake by plants via
vapor transfer
Air compartment in air Parcel N2 (air over tilled
soil)
Deposition rates, for uptake by farm produce
Deposition to surface soil compartment in surface
Parcel N6 (tilled soil)
Concentration in surface soil, for incidental
ingestion by humans and farm animals
Surface soil compartment in surface Parcel N1
(unfilled soil, closest to facility)
Concentration in soil, for uptake by farm produce
and animal feed
Surface soil compartment in surface Parcel N6
(tilled soil)
Concentration in fish consumed by angler
Water column carnivore compartment in lake (50%
of fish consumed) and benthic omnivore in lake
(50% of fish consumed)
C-3.2 Exposure and Risk Calculations
This section describes the approach for modeling chemical concentrations in farm food chain
(FFC) media (Section C-3.2.1); estimating human exposures associated with ingestion of FFC
media, incidental ingestion of soil, consumption of fish, and infant consumption of breast milk
(Section C-3.2.2); and calculating human health screening risk metrics associated with these
exposure pathways (Section C-3.2.3). All of these calculations are conducted using the MIRC
modeling software. For this multipathway screening evaluation, partitioning into FFC media is
calculated with the same data set used to model exposure and risk, rather than as a component
of the TRIM.FaTE modeling scenario. Consequently, processes and inputs related to
estimating chemical levels in FFC media are summarized in this section and discussed in detail
in Attachment 2.
C-3.2.1 Calculating Concentrations in Farm Food Chain Media
As was shown in Exhibit 2-5, MIRC was compiled to calculate concentrations of PB-HAPs in
foodstuffs that are part of the farm food chain. The FFC media included in this screening
scenario are as follows:
. exposed and protected fruit,
. exposed and protected vegetables,
. root vegetables,
. beef,
. total dairy products,
. pork, and
. poultry and eggs.
The algorithms used in MIRC were obtained from EPA's Human Health Risk Assessment
Protocol for Hazardous Waste Combustion Facilities (HHRAP; EPA 2005a). These algorithms
model the transfer of concentrations of PB-HAPs in FFC media using empirical biotransfer
factors. As noted in Section C-1 this report, the algorithms involving ingestion exposure to be
included eventually in TRIM are expected to be very similar to those presented in HHRAP.
Environmental media concentrations (i.e., the chemical source terms in these algorithms) were
C-22
-------
obtained from TRIM.FaTE. As noted in Section C-3.1.6, the TRIM.FaTE outputs included as
inputs to MIRC are the following:
. PB-HAP concentrations in air;
. air-to-surface deposition rates for PB-HAPs in both particle and vapor phases;
. PB-HAP concentrations in groundwater (used as drinking water);
. PB-HAP concentrations in fish tissue for fish in trophic levels three and four [T3 and T4]);
and
. PB-HAP concentrations in surface soil and root zone soil.
In general, plant- and animal-specific parameter values, including chemical-specific transfer
factors for FFC media, were obtained from the Hazardous Waste Companion Database
included in HHRAP (EPA 2005a). A list of variables and PB-HAP-specific input parameters,
along with the input values used in this screening scenario, are provided in Attachment 2.
C-3.2.2 Ingestion Exposure Assessment
C-3.2.2.1 Ingestion Exposure Pathways and Routes of Uptake
MIRC was used to estimate ingestion rates as average daily doses (ADDs) normalized to body
weight for a range of exposure pathways. Exposure pathways included are incidental ingestion
of soil and consumption offish, produce, and farm animals and related products. The ingestion
exposure pathways included in the screening evaluation and the environmental media through
which these exposures occur are summarized in Exhibit 3-5.
C-3.2.2.2 Exposure Scenarios and Corresponding Inputs
Specific exposure scenarios are developed by defining the ingestion activity patterns (i.e.,
estimating how much of each medium is consumed and the fraction of the consumed medium
that is grown in or obtained from contaminated areas) and the characteristics of the hypothetical
human exposed (e.g., age and body weight). MIRC computes exposure doses and risks for
each ingestion pathway separately, enabling the pathway(s) of interest for each PB-HAP to be
determined. Data related to exposure factors and receptor characteristics were obtained
primarily from EPA's Exposure Factors Handbook (EPA 1997a).
For the screening scenario described here, exposure characteristics were selected that result in
a highly conservative estimate of total exposure. The ingestion rate for each medium was set
equal to the 90th percentile of the distribution of national data for that medium. All media were
assumed to be obtained from locations impacted by the modeled source. Although this
approach results in an overestimate of total chemical exposure for a hypothetical exposure
scenario (for example, note that the total food ingestion rate that results is extremely high for a
hypothetical consumer with ingestion rates in the 90th percentile for every farm food type), it was
selected to avoid underestimating exposure for any single farm food type. The exposure
characteristics selected for the ingestion screening scenario are summarized in Exhibit 3-6.
Quantitative input values corresponding to these parameters are presented in Attachment 2.
C-23
-------
Exhibit 3-5. Summary of Ingestion Exposure Pathways and Routes of Uptake
Inaestion Intermediate
Environmental Uptake Route
Exposure
Pathway
Medium Ingested
Exposure
Pathway - Farm
Animals a
Medium
Process b
Incidental ingestion
of soil
Surface soil
N/A
Surface
soil
Deposition; transfer through plants;
transfer via erosion and runoffc
Consumption of
fish
Fish from local water
body
N/A
Fish
tissue
Direct uptake from water and
consumption of food compartments
modeled in TRIM.FaTE c
Consumption of
breast milkd
Breast milk
N/A
Breast
milk
Ingested by mother and then
partition to breast milk
Consumption of
produce
Aboveground produce,
exposed fruits and
vegetables
N/A
Air
Air
Soil
Deposition to leaves/plants
Vapor transfer
Root uptake
Aboveground produce,
protected fruits and
vegetables
N/A
Soil
Root uptake
Belowground produce
N/A
Soil
Root uptake
Consumption of
farm animals and
related food
products
Beef
Ingestion of forage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Ingestion of silage
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Dairy (milk)
Ingestion of forage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Ingestion of silage
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Pork
Ingestion of silage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Poultry
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Eggs
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
a Calculation of intermediate exposure concentrations were required only for the farm animal/animal product
ingestion pathways.
b Process by which HAP enters medium ingested by humans.
c Modeled in TRIM.FaTE.
d The consumption of breast milk exposure scenario is discussed in Section C-3.2.2.4.
C-24
-------
Exhibit 3-6. Overview of Exposure Factors Used for RTR Multipathway Screening a'b
Exposure Factor
Selection for Screening Assessment
Age group evaluated
Infants under 1 year (breast milk only)
Children 1 to 2 years of age
Children 3 to 5 years of age
Children 6 to 11 years of age
Children 12 to 19 years of age
Adult (20 to 70 years)
Body weight (BW; varies by age)
Weighted mean of national distribution
Intake rate and ingestion rate (IR) for farm produce and animal
products (varies by age and media consumed)
90th percentile of distribution of consumers
who produce own food
Ingestion rate for fish
17 g/day (approximately 90th percentile of
general population; also equal to mean value
for anglers); lower for children
Exposure frequency (EF)
365 days/year
Exposure duration
Lifetime, for estimating cancer risk; varies by
chemical for chronic non-cancer evaluation
Fraction contaminated (FC) (varies by media consumed)c
1
Cooking loss d
Assumed to be "typical;" varies depending
on food product (see Attachment 2)
a Values for exposure characteristics are presented in Attachment 2. Exposure parameter values were based on
data obtained from the Exposure Factors Handbook (EPA 1997a). See Attachment 2 for details.
b Exposure factor inputs are used in calculating average daily dose (ADD) estimates for each exposure pathway.
ADD equations for each pathway evaluated in this screening assessment are provided in Attachment 2.
c Fraction contaminated represents the fraction of food product that is derived from the environment included in the
screening scenario (e.g., produce grown on soil impacted by PB-HAPs). This parameter is defined separately for
each FFC medium.
d Cooking loss inputs were included to simulate the amount of a food product that is not ingested due to loss during
preparation or cooking, or after cooking.
C-3.2.2.3 Calculating Average Daily Doses
MIRC calculates chemical-specific ADDs of chemicals normalized to body weight (mg PB-HAP
per kg of body weight per day). Equations used to calculate ADDs were adapted from the
algorithms provided in the technical documentation of EPA's Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) Modeling System (EPA 2003), which derived much of
its input data from the Exposure Factors Handbook (EPA 1997a). The ingestion exposure
modeling approach embodied by 3MRA is conceptually similar to that presented in HHRAP. A
discussion of exposure dose estimation is provided in Attachment 2. The equations to calculate
ADDs for each ingestion pathway are provided in Attachment 2.
C-3.2.2.4 Infant Ingestion of Breast Milk
A nursing mother exposed to contaminants by any ingestion pathway described above can pass
the contaminants on to her infant through breast milk (EPA 1998). The nursing infant's
exposure can be estimated from the levels of chemical concentrations in the breast milk, which
in turn can be estimated based on the mother's chemical intake. Exposures can occur for
infants via this pathway for dioxins and possibly also for mercury.
Exposure to dioxin and mercury via breast milk consumption during the first year of life is
expected to have a small effect on the estimated lifetime ADD and on the individual's excess
lifetime cancer risk for dioxins or the highest chronic HQ for either chemical. Therefore,
C-25
-------
exposures to these chemicals via the breast milk pathway were not considered in developing
the de minimis emission thresholds for dioxins and mercury. The potential for non-cancer
health effects (e.g., developmental effects) is of greater concern for nursing infants exposed to
either chemical during the first year of life. These exposures will be considered in more detail
for facilities and emissions that do not pass the initial screen.
C-3.2.3 Risk Characterization
MIRC was used to calculate excess lifetime cancer risk (ELCR) and non-cancer hazard
(expressed as an HQ) using the calculated ADDs and ingestion dose-response values.
Chemical dose-response data include CSFs for ingestion and non-cancer oral RfDs for chronic
exposures. The CSFs and RfDs for the PB-HAPs included in this screening scenario are
presented in Exhibit 3-7 and are discussed in more detail in Attachment 2. Equations used to
estimate cancer risk and non-cancer hazard are provided in Attachment 2.
Estimated individual cancer risks for the PAHs included in the screening scenario were adjusted
upward to account for the higher mutagenic cancer potency of these compounds during
childhood, as specified by EPA in supplemental guidance for cancer risk assessment (EPA
2005c). Specifically, cancer potency is assumed to be tenfold greater for the first 2 years of life
and threefold greater for the next 14 years for PAHs. These factors were incorporated into a
time-weighted total increase in potency over a lifetime of 70 years. The cancer potency
adjustment for chemicals with a mutagenic mode of action is discussed in detail in Attachment
2.
Exhibit 3-7. Dose-response Values for PB-HAPs Addressed by the Screening
Scenario
PB-HAP
CSF
([mg/kg-day]"1)
Source
RfD
(mg/kg-day)
Source
Inorganics
Cadmium Compounds
not available
5E-4
IRIS
Elemental Mercury
NA
not available
Divalent Mercury
not available
3E-4
IRIS
Methyl Mercury
not available
1E-4
IRIS
Organicsa
Benzo[a]pyrene
10
IRIS
not available
2,3,7,8-TCDD
1.5E+5
EPA ORD
1E-9
ATSDR
Source: EPA OAQPS 2007
CSF = cancer slope factor; RfD = reference dose; IRIS = Integrated Risk Information System; CalEPA
= California Environmental Protection Agency; ASTDR = Agency for Toxic Substances and Disease
Registry; EPA ORD = EPA's Office of Research and Development
NA = not applicable
a The CSF listed in IRIS for benzo[a]pyrene was adjusted for consistency with the overall approach for
dose-response assessment of PAHs; see Attachment 2.
C-3.2.4 Dermal Risk Screening
Non-inhalation exposure to PB-HAPs can occur by way of the dermal pathway through contact
with PB-HAP-contaminated soil and water. However, dermal absorption of chemicals that are
originally airborne is generally a relatively minor pathway of exposure compared to other
exposure pathways (EPA 2006, CalEPA 2000). This section demonstrates that for the
C-26
-------
conservative screening scenario developed for RTR multipathway evaluation, the dermal
exposure route is not a significant risk pathway when compared to the ingestion pathway. In
general, the assessment followed the protocol for evaluating a reasonable maximum exposure
as described in EPA's Risk Assessment Guidance for Superfund (RAGS), Volume I: Human
Health Evaluation Model, Part E, Supplemental Guidance for Dermal Risk Assessment (EPA
2004c).
C-3.2.4.1 Hazard Identification and Dose Response Assessment
To assess the potential contribution of dermal exposure to non-inhalation exposure, we
evaluated the potential for cancer and chronic non-cancer effects for the four PB-HAPs currently
assessed in the multipathway screening evaluation for RTR: cadmium, divalent mercury,
2,3,7,8-TCDD, and benzo(a)pyrene. EPA has not developed carcinogenic potency slope
factors (CSFs) and non-cancer reference doses (RfDs) specifically for evaluating potential
human health concerns associated with dermal exposure to PB-HAPs. Instead, dermal toxicity
values can be derived from oral toxicity values via route-to-route extrapolation by adjusting for
gastrointestinal (Gl) absorption. EPA recommends making this adjustment only when Gl
absorption of the chemical is significantly less than 100% (i.e., less than 50 percent).
Otherwise, a default value of complete (100 percent) oral absorption is assumed, and no
adjustment is made (EPA 2004c).
The absorbed cancer slope factor (CSFAbs) is based on the oral cancer slope factor (CSF0) and
the fraction of the contaminant absorbed in the gastrointestinal track (ABSGi), as follows:
CSF
CSFabs =
ABSOI
where:
CSFAbs = Absorbed slope factor (mg/kg-day)"1
CSF0 = Oral slope factor (mg/kg-day)"1
ABSGi = Fraction of chemical absorbed in gastrointestinal tract (unitless)
The absorbed reference dose (RfDABs) is based on the oral reference dose (RFD0) and the
fraction of the contaminant absorbed in the gastrointestinal track (ABSGi), as shown below.
RfDABs = RfD0 x ABSGi
where:
RfDABs = Absorbed reference dose (mg/kg-day)
RfD0 = Oral reference dose (mg/kg-day)
ABSGI = Fraction of chemical absorbed in gastrointestinal tract (unitless)
Gl absorptions for 2,3,7,8-TCDD and all polycyclic aromatic hydrocarbons (PAFIs) (which
includes benzo[a]pyrene) were estimated to be greater than 50 percent. Therefore, as shown in
Exhibit 3-8, no adjustments to the available oral CSFs were required. Similarly, no adjustment
to the oral RfD for 2,3,7,8-TCDD was required. For cadmium and divalent mercury, adjustments
C-27
-------
were made based on absorption data provided in RAGS Part E, Exhibit 4-1. The RfDs for
dermal exposure to cadmium and divalent mercury are also shown in Exhibit 3-8.
Exhibit 3-8. Cancer Slope Factors and Reference Doses Based
on Absorbed Dose
PB-HAP
ABSg,
(unitless)
CSFabs"
(mg/kg-day)"1
RfDABS a
(mg/kg-day)
Cadmium Compounds
0.05
NA
2.5E-05
Divalent Mercury
0.07
NA
2.1E-05
2,3,7,8-TCDD
No adjustment
required b
1.5E+05
1.0E-09
Benzo[a]pyrene
No adjustment
required b
1.0E+01
NA
NA = Not applicable
a Oral dose response values are presented in Exhibit 3-7. Only the resulting adjusted dose response values are
presented in this table.
According to RAGS Part E, Exhibit 4-1, Gl absorption is expected to be greater than 50%.
C-3.2.4.2 Dermal Exposure Estimation
Dermal exposures and risks resulting from absorption of the chemical through the skin from
contact with contaminated water and soil were evaluated for the RTR screening scenario.
Individuals were assumed to be exposed on a fraction of their bodies (i.e., their head, forearms,
hands, lower legs, and feet) to contaminated soil from the TRIM.FaTE surface soil parcel with
the highest concentration (N1) on a daily basis. For the water evaluation, individuals were
assumed to be exposed to contaminated surface water with the same PB-HAP concentration as
the TRIM.FaTE screening scenario lake over their entire bodies on a daily basis.
Equations for Estimating Dermal Exposure
The general equation used to estimate dermal absorbed dose (DAD) for water or soil is shown
below, and is expressed in milligrams of PB-HAP per kilogram of receptor body weight per day
(mg/kg-day). DAD is calculated separately for the water and soil pathways.
DAD =
DA
event
: EV x ED xEFxSA
BW x AT
where:
DA
event
EV
ED
EF
SA
BW
AT
Absorbed dose per event; chemical-specific; equation for DAevent also differs
depending on water or soil contact (mg/cm2-event)
Event frequency (events/day)
Exposure duration (years)
Exposure frequency (days/year)
Skin surface area available for contact (cm2)
Body weight (kg)
Averaging time; for non-cancer effects, equals ED x 365 days/year; for cancer
effects, equals 70 years x 365 days/year (days)
C-28
-------
DAevent is estimated to be the total dose absorbed through the skin at the end of exposure and
the equation for calculation is different for organic and inorganic chemicals in water and for soil.
The equations for calculating these chemical-specific DAevent values for water contact are
provided in RAGS Part E, Chapter 3 (see Equations 3.2 - 3.4). For soil, the equation for
calculating these chemical-specific DAevent values is provided in RAGS Part E, Chapter 3 (see
Equation 3.12).
Water - Organic Chemicals: DAevent = Cwx2xFAxK0 l6xTevent x tevent
7T
where:
Water - Inorganic Chemicals: DA
event ^w x ^p x ^event
Soil - All Chemicals: DAevent = CsxAFx ABS x CF
DAevent
Cw
Cs
K„
FA =
Tevent
tevent
AF
ABS
CF
Absorbed dose per event (mg/cm2-event)
Chemical concentration in water (mg/cm3) or soil (mg/kg)
Chemical-specific dermal permeability coefficient of compound in water (cm/hr)
Chemical-specific fraction absorbed; accounts for loss due to the regular
shedding of skin cells of some chemical originally dissolved into skin (unitless)
Chemical-specific lag time per event (hr/event)
Receptor-specific event duration (hr/event)
Receptor- and activity-specific adherence factor of soil to skin (mg/cm2-event)
Chemical-specific dermal absorption fraction (unitless)
Conversion factor (10"6 kg/mg)
Exposure Factors and Assumptions
The exposure parameters included in this assessment and their default and other value options
are summarized in this subsection. Default values were selected to result in a highly
conservative estimated of exposure (i.e., exposures are likely overestimated). Parameter
values were primarily obtained or estimated from RAGS Part E (EPA 2004c) and the CSEFH
(EPA 2008). Receptor-and scenario-specific exposure assumptions are discussed first, and a
discussion of chemical-specific parameters values follows. Estimated water and soil exposure
concentrations are presented at the end of this subsection.
Receptor-Specific Parameters
Dermal exposures and risks were estimated for the same age groups used in the ingestion
exposure assessment: adults (ages 20 to 70 years) and five child age groups: <1 year; 1 to 2
years; 3 to 5 years; 6 to 11 years; and 12 to 19 years. The body weight values used in the
ingestion exposure assessment were used in the dermal exposure assessment.
Body surface areas for water and soil exposures for adults were calculated using Appendix C,
Exhibit C-1, of RAGS Part E. For children, SAs for water and soil exposures for the five
children's age groups were estimated using Tables 7-1 and 7-2 of the CSEFH, respectively. For
SA (water), individuals were assumed to shower or bathe in the water with 100 percent of their
body exposed. For SA (soil), it was assumed that individuals were exposed on a fraction of
their total body, specifically their head, forearms, hands, lower legs, and feet. Based on
C-29
-------
information provided in RAGS Part E, the SA for forearms was calculated using the SA for arms
and assuming a forearm-to-arm ratio of 0.45, and the SA for lower legs was estimated using the
SA for legs and assuming a lower leg-to-leg ratio of 0.4.
Values for body SA by age group are summarized in Exhibit 3-9.
Exhibit 3-9. Receptor-Specific Body Surface Area Assumed to be
Exposed to Chemicals
Age Group3
(years)
Surface Area for
Water Exposure (cm2)
Surface Area for
Soil Exposure (cm2)
Adult 20-70
18,150 9
6,878 n
Child <1 D
3,992
1,772
Child 1-2 c
5,700
2,405
Child 3-5 a
7,600
3,354
Child 6-11 e
10,800
4,501
Child 12-19 r
17,150
6,906
a Sources for the child groups included Table 7-1 (total body surface area for SA-Water), and
Table 7-2 (fraction of total body surface area for SA-Soil) of the 2008 CSEFH.
b Represents a time-weighted average for age groups birth to <1 month, 1 to <3 months, 3 to <6
months, and 6 to <12 months.
c Represents a time-weighted average for age groups 1 to <2 years and 2 to <3 years.
d Values for age group 3 to <6 years in the 2008 CSEFH.
e Values for age group 6 to <11 years in the 2008 CSEFH. Represents a conservative (i.e.,
slightly low) estimate for ages 6 through 11 years since 11-year olds are not included in this
CSEFH age group.
f Represents a time-weighted average for age groups 11 to <16 years and 16 to <21 years. Note
that estimated values include 11-year-olds and individuals through age 20, which contributes to
uncertainty in the estimates for 12 to 19 years.
9 Represents the average total surface area of adults from Table C-1 of RAGS Part E.
h Represents the average surface area of adults for head, forearms, hands, lower legs, and feet
from Table C-1 of RAGS Part E.
Scenario-Specific Parameters
Exhibit 3-10 summarizes the exposure values related to frequency and duration of contact. In
general, these are the recommended defaults for calculating a reasonable maximum exposure
(RME) for a residential scenario as proposed by EPA in RAGS Part E, Chapter 3.
C-30
-------
Exhibit 3-10. Scenario-Specific Exposure Values for Water and Soil Contact
Exposure Parameter
Receptor
Value
Source
Water Contact
Event Duration (Went)
Child
1
Reasonable maximum exposure
scenario for showering/bathing
from RAGS Part E, Exhibit 3-2
(hr/event)
Adult
0.58
Soil Contact
Soil Adherence Factor (AF)
Child
0.2
For children, value is geometric
mean value for children playing
(wet soil) and for adults, value is
(mg/cm2)
Adult
0.1
geometric mean value for an
adult farmer from RAGS Part E,
Exhibit 3-3
Both Media
Event Frequency (EV)
(events/day)
All
1
Reasonable maximum exposure
scenario from RAGS Part E,
Exhibits 3-2 & 3-5.
Exposure Frequency (EF)
(days/year)
All
350
Child <1
1
Child 1-2
2
Represents the number of years
Exposure Duration (ED)
Child 3-5
3
included in the age group; also
(years)
Child 6-11
6
used in ingestion exposure
Child 12-19
8
calculations.
Adult 20-70
50
Averaging Time (AT)
(days)
For cancer assessment, an AT equal to a lifetime (70 years) x 365
days/year is used. Same value used in ingestion exposure calculations.
For non-cancer assessment, an AT equal to the exposure duration (ED)
x 365 days/year is used, so AT will vary by receptor group. Same value
used in ingestion exposure calculations.
Chemical-Specific Parameters
The chemical-specific parameters required to quantitatively evaluate dermal pathway exposures
are listed in Exhibit 3-11. For the water concentration in the dermal analysis, the modeled
TRIM.FaTE chemical concentration in the screening scenario pond at the de minimis emission
rate was used. For the soil concentration, the modeled TRIM.FaTE chemical concentration in
surface soil in parcel N1 of the screening scenario at de minimis emission rate. This same soil
concentration was also used in ingestion exposure calculations for soil ingestion.
Dermal absorption of chemicals in water is based on the use of a dermal permeability coefficient
(Kp), which measures the rate that a chemical penetrates the skin. Dermal absorption of soil-
bound chemicals is based on the use of a dermal absorption fraction (ABS), which is a measure
of how much of a chemical the skin absorbs through contact with soil.
C-31
-------
Exhibit 3-11. Chemical-Specific Dermal Exposure Values for Water and Soil Contact
PB-HAP
Cadmium
Divalent
Mercury
2,3,7,8-
TCDD
Benzo[a]
pyrene
Source
Chemical concentration
in Water (Cw) (mg/cm3)
4.77E-07
1.81 E-07
6.20E-17
2.03E-12
TRIM.FaTE modeled
concentration in screening
scenario pond
Chemical concentration
in Soil (Cs) (mg/kg)
1.37E+00
5.94E+00
5.36E-09
1.21E-03
TRIM.FaTE modeled
concentration in surface soil
in parcel N1 in screening
scenario
Permeability coefficient
in water (Kp) (cm/hour)
0.001
0.001
0.81
0.7
Values from RAGS Part E,
Exhibits B-3 (organics) and
B-4 (inorganics)
Fraction absorbed water
(FA) (unitless)
NA
NA
0.5
1.00
Values from RAGS Part E,
Exhibits B-3; only used for
organic chemicals
Lag time per event
(Tevent) (hr/event)
NA
NA
6.82
2.69
Values from RAGS Part E,
Exhibits B-3; only used for
organic chemicals
Dermal absorption
fraction (ABS) from soil
(unitless)
0.001
0.045 a
0.03
0.13
Values from RAGS Part E,
Exhibit 3-4, unless otherwise
noted
Value obtained from Bioavailability in Environmental Risk Assessment (Hrudey et al. 1996).
C-3.2.4.3 Screening-Level Cancer Risks and Non-Cancer Hazard Quotients
Toxicity values were used in conjunction with exposure information to evaluate the potential for
cancer risks and non-cancer health hazards. Risk estimation methods are presented below.
Dermal Cancer Risk
Cancer risk for the dermal route was calculated as the product of the age-specific DADs and the
absorbed CSF for each chemical, as follows:
Dermal Cancer Risk = DAD x CSFabs
where:
DAD = Dermal Absorbed Dose (mg/kg-day)
CSFAbs = Absorbed cancer slope factor (mg/kg-day)"1
Lifetime dermal cancer risks were calculated for 2,3,7,8-TCDD and benzo[a]pyrene. The total
risk accounts for dermal exposures that an individual might receive from these PB-HAPs in
water plus soil over his or her lifetime (70 years).
Dermal Hazard Quotient
Dermal hazard quotient (HQ) was estimated as the ratio of age-specific DADs to the absorbed
RfD for each chemical, as shown below:
Dermal HQ = DAD /RfDABs
where:
DAD = Dermal Absorbed Dose (mg/kg-day)
RfDABs = Absorbed reference dose (mg/kg-day)
C-32
-------
The aggregate HQ accounts for exposures that an individual in a receptor group may receive
from the PB-HAP in water and soil over the exposure duration. Non-cancer hazard is not
additive across the age groups evaluated here.
C-3.2.4.4 Dermal Screening Results
Exhibit 3-12 presents a summary of estimated dermal non-cancer hazards by age group. A
summary of estimated lifetime dermal cancer risks is provided in Exhibit 3-13. All HQ values
were 0.5 (representing divalent mercury exposure for children less than 1 year of age) or less, a
factor of at least two smaller than the potential ingestion hazard quotients associated with the
screening scenario. The highest estimated individual lifetime cancer risk associated with
potential dermal exposures was 3.33E-8 for benzo[a]pyrene; this value is a factor of 30 times
smaller than the ingestion risk estimated for the same de minimis emission rate.
Exhibit 3-12. Summary of Dermal Non-Cancer Hazards
2,3,7,8-TCDD
0.00012
0.0001
0.00008
0.00006
0.00004
0.00002
0
¦ Water
¦ Soil
¦ Total
I
Child Child Child Child Child Adult
<1 1-2 3-5 6-11 12-19 20-70
Receptor
Cadmium
0.012
0.01
0.008
0.006
0.004
0.002
0
¦ Water
¦ Soil
¦ Total
1M
Child
<1
Child
1-2
Child
3-5
Receptor
Child
6-11
Child Adult
12-19 20-70
Divalent Mercury
¦ Water
¦ Soil
¦ Total
O 0.3
n 0.2
Child Child 1-Child 3-Child 6- Child Adult
<1 2 5 11 12-19 20-70
Receptor
C-33
-------
Exhibit 3-13. Summary of Dermal Cancer Risks
PB-HAP
Dermal Lifetime
Cancer Risk
Magnitude of Difference
2,3,7,8-TCDD
Water
6.14E-09
>160
Soil
3.48E-10
>2800
Total
6.49E-09
>150
Benzo[a]pyrene
Water
1.06E-08
>90
Soil
2.27E-08
>40
Total
3.33E-08
>30
Based on these results and taking into consideration the extremely conservative nature of the
dermal exposure calculations, it was assumed that it is not necessary to incorporate dermal
exposures in calculating multipathway de minimis levels. Specifically, the daily exposure
durations of 0.58 hour for adults and 1 hour for children used to calculate dermal exposure from
water are highly conservative and assume that the individual is bathing in surface water taken
directly from a contaminated lake or is swimming in the lake for 350 days of the year. The
exposure frequency of 350 days and corresponding skin surface area available for contact with
contaminated soils (i.e., head, hands, arms, legs, and feet) likely also grossly overestimates
dermal exposure to soil.
C-3.3 Summary of Scenario Assumptions
As emphasized previously, the screening scenario created for evaluating PB-HAP emissions
from RTR facilities is intended to be generally conservative to prevent underestimating risk. The
overall degree of conservatism of the scenario is the sum total of the multiple assumptions that
affect the outputs of the fate and transport, exposure, and risk modeling. Exhibit 3-14
summarizes important characteristics that influence exposure and risk estimates for this
scenario and indicates the general degree of conservatism associated with the values for each
assumption. Although this summary does not provide a quantitative estimate of the output
uncertainty or the degree to which exposures and risks estimated using the scenario would be
overestimated, it does demonstrate qualitatively that the scenario generally overestimates
exposure and thus favors a health-protective risk output.
C-34
-------
Exhibit 3-14. Summary of RTR Screening Scenario Assumptions and Associated
Conservatism
Characteristic
Value
Conservatism
Comments on Assumptions
General Spatial Attributes
Farm location
375 m from source;
generally downwind
Conservative
Location dictates soil and air concentrations
and deposition rates used to calculate
chemical levels in farm produce.
Lake location
375 m from source;
generally downwind
Conservative
Location dictates where impacted fish
population is located.
Surface soil
properties
Typical values or
national averages
Neutral
Based on existing EPA documentation and
other references.
Size of farm parcel
About 4 ha
Conservative
Relatively small parcel size results in higher
chemical concentration.
Size of lake
46 ha; about 3 m
average depth
Conservative
Lake is large enough to support an aquatic
ecosystem with higher trophic level fish, but
is relatively small and shallow (thus
increasing surface area-to-volume ratio).
Meteorological Inputs
Total precipitation
1.5 m/yr
Conservative
Intended to represent rainy U.S. location;
set to highest state-wide average for the
contiguous U.S.
Precipitation
frequency (with
respect to impacted
farm/lake)
2/3 of total
precipitation fall on
farm/lake and
watershed
Conservative
Most of total precipitation occurs when the
farm/lake are downwind of the source.
Wind direction
Farm/lake are
downwind 40% of the
time
Conservative
Farm/lake located in the predominantly
downwind direction. Temporal dominance
of wind direction based on data from
Yakima, WA, where wind is predominantly
from the west.
Wind speed
2.8 m/sec
Conservative
Low wind speed (5th percentile of long-term
averages for contiguous U.S.); increases
net deposition to lake/watershed.
Air temperature
298 K
Neutral
Typical for summer temperatures in central
and southern U.S.
Mixing height
710 m
Conservative
Relatively low long-term average mixing
height (5 percentile of long-term averages
for contiguous U.S.); increases estimated
air concentration.
Watershed and Water Body Characteristics
Evaporation of lake
surface water
700 mm/yr
Neutral
Value is representative of cooler climates.
Surface runoff into
lake
Equal to 40% of total
precipitation
Conservative
Based on typical water flow in wetter U.S.
locations; higher runoff results in greater
transfer of chemical to lake.
C-35
-------
Exhibit 3-14. Summary of RTR Screening Scenario Assumptions and Associated
Conservatism
Characteristic
Value
Conservatism
Comments on Assumptions
Surface water
turnover rate in lake
About 12 turnovers
per year
Neutral
Consistent with calculated water balance;
reasonable in light of published values for
small lakes. May overestimate flush rate if
water inputs are also overestimated. Note
that after evapotranspiration, remaining
water volume added via precipitation is
assumed to flow into or through lake.
Soil erosion from
surface soil into lake
Varies by parcel;
ranges from 0.002 to
0.01 kg/m2-day
Neutral
Erosion rates were calculated using the
universal soil loss equation (USLE); inputs
to USLE were selected to be generally
conservative with regard to concentration in
the pond (i.e. higher erosion rates were
favored). May underestimate erosion for
locations susceptible to high erosion rates.
Note that higher erosion increases
concentration in lake (and fish) but
decreases levels in surface soil (and farm
products).
Aquatic food web
structure and
components
Multi-level; includes
large, upper trophic
level fish
Conservative
Inclusion of upper-trophic level fish and
absence of large-bodied
herbivore/detritivore fish favor higher
concentrations of bioaccumulative
chemicals and result in higher
concentrations in consumed fish.
Parameters for Estimating Concentrations in Farm Food Chain Media
Fraction of plants
and soil ingested by
farm animals that is
contaminated
1.0 (all food and soil
from contaminated
areas)
Conservative
Assumes livestock feed sources (including
grains and silage) are all derived from most
highly impacted locations.
Soil- and air-to-plant
transfer factors for
produce and related
parameters
Typical (see
Attachment 2 for
details)
Neutral
Obtained from peer-reviewed and standard
EPA reference sources.
Biotransfer factors
for efficiency of
uptake by animal of
chemical in food/soil
Typical (see
Attachment 2 for
details)
Neutral
Obtained from peer-reviewed and standard
EPA reference sources.
Bioavailability of
chemicals in soil (for
soil ingested by
animals)
1.0 (relative to
bioavailability of
chemical in plant
matter)
Conservative
Probably overestimates bioavailability in
soil; many chemicals are less bioavailable
in soil than in plants.
Ingestion Exposure Assumptions
Ingestion rates for all
farm
produce/livestock
types
Person obtains all
food sources from
local farm; ingestion
rate is 90th percentile
of rates for home-
produced food items
Conservative
All food derived from impacted farm; total
food ingestion rate would exceed expected
body weight-normalized ingestion rates
(prevents underestimating any individual
food type).
C-36
-------
Exhibit 3-14. Summary of RTR Screening Scenario Assumptions and Associated
Conservatism
Characteristic
Value
Conservatism
Comments on Assumptions
Fish ingestion rate
0.24 g/kg-day for
adult; between 0.14
and 0.26 g/kg-day for
children aged 1-19
Conservative
Rates are based on EPA's analysis of
freshwater and estuarine fish consumption
from the USDA's Continuing Survey of
Food Intake by Individuals (2002d). This
likely overestimates long-term fish
consumption rates for the general
population. See Attachment C-2 for a
detailed discussion.
Exposure frequency
Consumption of
contaminated food
items occurs 365
days/yr
Conservative
All meals from local farm products.
Body weight
Mean of national
distribution
Neutral
Note that this does not affect the body
weight-normalized rates for produce and
animal products.
Other Chemical-Specific Characteristics
General chemical
properties used in
fate and transport
modeling (Henry's
law, Kow, etc.)
Varies
Neutral
Obtained from peer-reviewed sources;
intended to be representative of typical
behavior and characteristics.
"General" physical
properties (plant
matter density,
aquatic life biomass,
algal growth rate,
etc.)
Varies
Neutral
Obtained from peer-reviewed sources;
intended to be representative of typical
behavior and characteristics.
Dose-response
values
Varies
Neutral to
conservative
Values used are those determined to be
appropriate for risk assessment by OAQPS;
some values may be health-protective.
C-4 Evaluation of Screening Scenario
C-4.1 Overview
To evidence our understanding of the models used, their configuration, and the total uncertainty
associated with this model application, the screening scenario developed for RTR was
subjected to a series of evaluations. These analyses were somewhat more focused on
TRIM.FaTE, given the complexity and variability associated with the fate and transport modeling
phase relative to other aspects of the screening calculations. These evaluations emphasized
the application of this approach (and especially TRIM.FaTE) in the context of ingestion
exposure and risk screening for RTR. The analyses were not intended to be general model
performance evaluations of TRIM.FaTE. EPA has conducted such analyses, which are
documented in detail in Volumes I and II of the TRIM.FaTE Evaluation Report (EPA 2002b,
2005b).
C-37
-------
The purpose of the current evaluations, however, is similar to that of the previous model
evaluations, in that both are intended to increase the confidence associated with model
performance. Although the focus of the screening scenario evaluation (assessing the utility of
the constructed modeling scenario) differs from the purpose of TRIM.FaTE model evaluations,
the underlying objectives are comparable. In summary, the current evaluations are intended to
achieve several objectives:
enhance understanding of how the scenario operates;
better characterize the uncertainty of model results;
measure model sensitivity to changes in parameter values and scenario configuration;
and
strengthen the defensibility of the scenario's application as a component of air toxics
residual risk assessment.
That these analyses do not attempt to validate or "prove" the accuracy of model results is
important to note. As specified in the TRIM.FaTE mercury test case evaluation report (EPA
2005b):
'Validation' of such models, in the classic sense (e.g., proving the model
produces accurate results across a range of input conditions), is not generally
possible, in part because there are no comprehensive data sets of measured
chemical concentrations (and associated contributing pollutant sources) for use
in such comprehensive studies, nor are there other fully validated multimedia
models against which TRIM.FaTE can be benchmarked. Thus, evaluation of
TRIM.FaTE is not a yes/no exercise but a continuing accumulation of evidence
leading to model refinement and eventually to increasing levels of confidence in
the model results.
The screening scenario was evaluated through a series of analyses:
. a general evaluation of TRIM.FaTE fate and transport modeling outputs;
. comparisons of model outputs to the literature (e.g., measured concentrations of the
chemicals evaluated, information on the expected distribution of chemical mass in the
environment);
. sensitivity analyses of TRIM.FaTE and MIRC inputs and model configurations on
endpoints of interest; and
. an evaluation of related information.
As noted previously, most of these evaluations focused on the TRIM.FaTE modeling scenario.
Methods and results are discussed in the following sections.
C-4.2 Overall Chemical Mass Partitioning
A general evaluation of TRIM.FaTE outputs from simulations run using the screening scenario
was conducted to inform the set-up and parameterization of the scenario and confirm that the
model outputs were generally reasonable. One approach for evaluating the overall performance
of the model is to review the distribution of chemical mass among the various compartment
types and chemical sinks included in the scenario. In general, many persistent and
bioaccumulative chemicals emitted to the air do not readily deposit within close range of a
source but instead are transported in the air away from the source (sometimes great distances)
before depositing on land or water surfaces or being transformed to other chemical species.
C-38
-------
This is true of the PB-HAPs that can be evaluated using the RTR screening scenario, and thus
much of the chemical mass emitted by the modeled sources, regardless of chemical, was not
expected to accumulate in the soil, water, or sediment compartments. Instead, most of the
chemical mass emitted by the source was expected to be removed through advective transport
in the air and to end up in the air sinks (the compartments that capture chemical mass blown
outside of the modeling domain). This outcome also would be expected, given the relatively
small modeling domain. For chemical mass that remains in the domain (is deposited and does
not degrade or leave the domain by other means, such as sediment burial), the overall
distribution would be expected to approximate environmental observations, with much of the
mass remaining in the soil or benthic sediment.
Exhibit 4-1 illustrates the distribution of chemical mass for the screening scenario for the PB-
HAPs of primary concern (one dioxin, one PAH, cadmium, and mercury). Results show the
distribution at the end of a 50-year simulation (or similar duration) for each chemical performed
at the de minimis emission rate (the overall distribution for each chemical is not expected to be
substantially different for other emission rates, given a relatively long simulation period). As
anticipated, much of the chemical mass emitted to the screening scenario is largely removed
from the scenario via air advection processes and transported to air sinks (blown out of the
modeled domain of the scenario). The amount remaining in the scenario varies by chemical,
with a larger fraction of mass deposited for mercury (which deposits relatively quickly as a
divalent species) and much lower deposition for dioxin. Of the chemical mass deposited to
plants, land, or water and remaining in the modeled domain, most accumulates in the soil. For
the two metals evaluated (mercury and cadmium), the sediment compartment also contains a
significant amount of mass.
The mass distribution for mercury presented in Exhibit 4-1 was calculated from a single
simulation with emissions split evenly by mass between divalent mercury and elemental
mercury. Recall that TRIM.FaTE models mercury transformation among elemental, divalent,
and methyl mercury species. Consequently, the distribution of mercury mass shown in the
middle and lower parts of the table accounts for inter-species transformation within media as
well as movement between media (and all methyl mercury in the scenario was derived from
emitted divalent or elemental mercury).
C-39
-------
Exhibit 4-1. Distribution of Chemical Mass in Screening Scenario
2,3,7,8-TCDD
Benzo[a]
pyrene
Cadmium
Mercury (Divalent Mercury Emitted)
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Distribution of Total Mass Added to Scenario from Modeled Source
Emitted chemical mass
removed from scenario and
deposited into air sinks
99.5%
81.6%
97.3%
85.4%
b
b
b
Emitted chemical mass
remaining in scenario (not in
air sinks)3
0.5%
18.4%
2.7%
14.6%
b
b
b
Distribution of Mass Remaining in Scenario0
Air
0.30%
0.2%
0.01%
0.001%
0.002%
0.0%
0.001%
Soil
89.2%
97.6%
96.5%
86.0%
91.1%
98.0%
86.3%
Plants
6.0%
1.6%
0.1%
0.7%
0.0%
0.0002%
0.7%
Surface Water
0.02%
0.01%
0.2%
0.0%
0.5%
0.03%
0.05%
Sediment
4.4%
0.6%
3.2%
13.2%
8.5%
2.0%
12.9%
Aquatic Biota
0.03%
0.001%
0.003%
0.0003%
0.0002%
0.005%
0.0004%
Groundwater
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Distribution of Mass in Soil
Surface Soil
98.8%
98.9%
30.1%
98.9%
9.3%
98.9%
97.9%
Root Zone Soil
1.2%
1.1%
69.9%
1.1%
90.7%
1.1%
2.1%
Vadose Zone Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
a Fraction includes mass transferred from air to soil or other media, but subsequently removed from the modeling
domain (e.g., accumulated in sediment burial or degradation/reaction sink).
b To calculate the de minimis level for mercury, neither elemental mercury or methyl mercury is emitted by the
modeled source.
c Fractions represent amount of mass remaining in non-sink compartments.
C-4.3 Comparison to Measured Concentrations
C-4.3.1 Scope of the Evaluation
This section presents environmental measurements of the PB-HAPs evaluated for RTR and
compares them to modeled concentrations estimated by TRIM.FaTE and the Multimedia
Ingestion Risk Calculator (MIRC). Model outputs correspond to the estimates associated with
emission rates at the de minimis thresholds for each representative PB-HAP calculated for RTR
multipathway screening. We emphasize that this analysis is not intended to "validate" the
accuracy of the models used for RTR or the specific model configurations used to calculate de
minimis emission thresholds. In general, although the comparison of environmental data to
model results is often a central component of model evaluation, deriving useful conclusions from
such comparisons is complicated by a range factors, including: fundamental differences
between the modeling scenario and the environmental system in which samples were collected;
temporal and spatial issues (e.g., model results representative of long-term average
C-40
-------
concentrations vs. measured point estimates); data quality issues; and other factors. In the
current evaluation, it is difficult to derive conclusions from model-to-measurement comparisons
given the use of a modeling scenario that is intended to be representative but hypothetical and
conservative (i.e., possessing characteristics that lead to higher exposures while not
representing any real location or specific site). In addition, the RTR modeling scenario is
intended to calculate incremental environmental concentrations and exposures without
accounting for any "background" concentrations from natural or anthropogenic sources.
In spite these complications, this comparative analysis is an appropriate—and necessary—
aspect of the model evaluation process because environmental data provide a general frame of
reference for the model outputs. Basic differences between the modeling scenario and the data
sets collected preclude the verification or validation of model configuration, performance, or
results, and the comparisons should be viewed in light of the complicating issues listed above.
However, we believe that this comparative analysis, considered in conjunction with other
aspects of the evaluation, is a useful tool in gauging the effectiveness of the screening scenario
in informing regulatory decision-making.
C-4.3.2 Methods and Organization of this Section
The TRIM.FaTE and MIRC outputs presented in this section are those used to calculate the de
minimis emission levels in the screening scenario (see Section C-2 for discussion of de minimis
emission rates). Model outputs from the source parcel are not used for comparisons; instead,
concentrations from the land parcel closest to the source (N1) are used in comparisons for soil
results, while outputs from the pond parcel are used for comparing results for water, sediment,
and fish (See Exhibit 3-1 for Surface Parcel Layout). Chemical concentrations in air are not
evaluated in this section for two reasons. First, the TRIM.FaTE model was not designed to
estimate air concentrations for the assessment of inhalation exposure, as noted in Section
2.4.1. Second, chemical concentrations in the air compartment affect exposure concentrations
less directly than deposition of the chemical. Therefore, the comparison of chemical
concentrations in air to observed values is not included in this evaluation.
The purpose of the 50-year duration of the scenario is to represent long-term input of source
emissions over a hypothetical facility lifetime (many compartments also attain an approximate
steady state by this time). For all environmental media compartments, the maximum annual
average concentrations, taken from the final year of the simulation, are used in MIRC to
determine the maximum average concentrations in ingestible products. These concentrations,
along with ingestion rates for each food product, are used to calculate average daily doses
(ADDs) for each chemical, from which the lifetime cancer risk or age-group-specific hazard
quotient (HQ) is then calculated. The model results presented in this section are the
environmental media concentrations used to calculate the de minimis emission rates (i.e., those
concentrations calculated for the 50th year of the TRIM.FaTE simulation) and the ingestible
media concentrations calculated in MIRC using those environmental media concentrations (see
Exhibit 3-5 for TRIM.FaTE outputs used in exposure calculations).
The evaluations described in this section are presented separately for each chemical for which
a screening threshold has been derived. Each chemical-specific section provides a review of
the emission profiles used in the modeling process, and a summary of the expected chemical
behavior in the environment. This is followed by comparisons between model results and
ranges of observed concentrations in environmental media, partitioning behavior, and observed
concentrations in ingestible products.
C-41
-------
In the charts displaying the comparisons between model results and observed concentrations,
the horizontal bars in the graph represent measured concentrations reported in the literature for
various environmental media. The green bars show ranges of measured concentrations that
were reported as less than a maximum concentration or detection limit, with the upper bound of
the bar representing the limit of detection. Blue bars indicate a range of values reported by a
particular source. The vertical red lines in the graph represent the modeled TRIM.FaTE
concentrations obtained using the de minimis threshold emission rates. Observed data from
areas exposed to varying ranges of chemical emissions were chosen for the comparison in
order to represent a variety of scenarios. In general, the low end of the range represents
concentrations from remote areas not located near a considerable pollution source, while the
high end of the range represents concentrations observed near industrial sources (e.g., metal
smelters, chlor-alkali plants, incineration/combustion facilities) and/or in urban areas.
The observed values provided as examples in this section include data collected from a range
of sampling techniques, locations, emissions profiles, historical contributions, and other factors
influencing concentration. They do not represent the entire range of possible measured values,
and therefore should not be combined to create a contiguous range of concentrations
representative of all patterns in the United States. Model outputs were derived using a
generally conservative set of fate and transport inputs, and they reflect the estimated
environmental levels assuming concentrations from only one source and no ambient
background concentrations.
C-4.3.3 Chemical-Specific Comparisons
C-4.3.3.1 Cadmium
Behavior in the Environment
Based upon reviewed literature, some of the largest anthropogenic sources of cadmium to air
are facilities that process, mine, or smelt cadmium-zinc or cadmium-zinc-lead ores; coal and oil-
fired boilers; other urban and industrial facilities; phosphate fertilizer manufacturing facilities;
road dust; and municipal sewage sludge incinerators (ATSDR 2008). These facilities can emit
airborne cadmium particles that are capable of traveling long distances before depositing onto
soil or water bodies via dust, rain, or snow. Cadmium adsorbs to soil particles in the surface
layers of the soil profile, but to a lesser degree than many other heavy metals (HSDB 2005a). It
may enter surface waters through atmospheric fallout, runoff, or wastewater streams, but most
cadmium compounds will be removed from the surface water compartment through adsorption
to organic matter in sediment or to other suspended compounds. Concentrations in bed
sediment are expected to be roughly an order of magnitude higher than those in overlying
surface water (HSDB 2005a).
Freshwater fish accumulate cadmium primarily through direct uptake of the dissolved form
through the gills and secondarily through diet, which plays a variable role in total cadmium
uptake (Reinfelder et al. 1998, Chen et al. 2000, Saiki et al. 1995). The degree to which
cadmium bioaccumulates in fish is largely dependent upon water pH and humus content
(ATSDR 2008). Reported bioaccumulation factors (BAFs), or bioconcentration factors (BCFs),
of 3 to 4,190 (ATSDR 2008) and 907 (EPA 2005a) have been reported for freshwater organisms
and fish. Although some biomagnification of cadmium has been reported for aquatic food
chains involving fish in saltwater systems, biomagnification in freshwater systems appears to be
present only at lower trophic levels (Chen et al. 2000) and in narrowly defined niches (e.g.,
plankton/macroinvertebrate food chains; Croteau et al. 2005). Biomagnification factors (BMFs)
C-42
-------
of less than 1 have generally been reported for fish at higher trophic levels, indicating that
cadmium concentrations generally biodiminish from lower to higher trophic levels (Chen et al.
2000, Mason et al. 2000).
Emission Profile
The cadmium de minimis emission rate derived from the RTR PB-HAP screening is 0.65 TPY
(based on an HQ of 1 in children aged 1-2). Although the annual concentration from year 50 is
used for comparison, modeled concentrations of cadmium in most compartments generally
leveled off after year 20. One notable exception was the soil compartment in parcel N6, where
the soil in the screening scenario is tilled and no plants are modeled in the TRIM.FaTE scenario.
Cadmium concentrations in this compartment continued to rise through year 50. The steady
build-up of cadmium in this parcel is likely due to the thickness of the surface soil layer in the N6
parcel. Because this parcel is tilled, the surface soil layer is thicker than in the other parcels,
and the rate of exchange between the surface soil and the root zone soil is correspondingly
lower.
Concentrations in Environmental Media
Exhibit 4-2 displays measured concentrations in environmental media with TRIM.FaTE outputs
for the screening scenario. Modeled concentrations and BAFs in fish compartments are shown
in Exhibit 4-3. Because relationships between fish compartments in the screening scenario are
established within a food web structure, there is no explicit delineation to specific trophic levels.
However, for evaluation of trophic level patterns, the water column herbivore (WCH) can be
considered to represent the lowest trophic level among fish, while the water column carnivore
(WCC) represents the highest trophic level.
The cadmium concentrations output by the RTR screening scenario were consistent with
reported values in all environmental media compartments. Consistent with trends noted in the
literature, modeled cadmium concentrations were highest in fish at the lowest trophic level
evaluated in the scenario (i.e., WCH). Though modeled concentrations did not uniformly
decrease with an increase in trophic level, the concentrations for all but the benthic carnivore
(BC) compartment were markedly lower than those in the WCH compartment. It should be
noted that BAFs for the other chemicals evaluated in this section are several orders of
magnitude greater—and span much wider ranges—than the BAFs reported for cadmium in fish.
Thus, in the context of comparison between chemicals, the difference between cadmium BAFs
in fish compartments is quite small.
The BAFs calculated using model outputs from the RTR screening scenario ranged from
approximately 200 to 1,400, which is consistent with the range of values presented in the
literature for freshwater systems. Overall, modeled concentrations and BAFs in all fish
compartments were reasonably consistent with observed trends, with cadmium levels generally
diminishing from lower to higher trophic levels, and consistent with the assumption that diet
plays a variable role in the bioaccumulation of this substance in fish.
C-43
-------
Exhibit 4-2. Summary of Modeled and Observed Concentrations of Cadmium in Environmental Media
i i Reported value was less than limit of detection (LOD = upperbound of bar)
i i Value or range from literature
^^"Modeled concentration corresponding to de minimisthreshold emission rate
Largemouth Bass, SWP, FL, Avg. (a)
I I I
Is Salmon Flesh (a)
WCC
ic] '—I
^ O*)
.<£ J= Redear Sunfish, SWP, FL, Avg. (a)
LJ- ji>
U
Catfish, Mining-Contaminated Water, OK (a)
Trout, Remote Stream, MD (a)
-g Great Lakes (b)
I
I
M Spring River Basin Mining District, KS, MO, & OK (c)
I
1
E Patroon Creek Reservoir, NY, Avg. (a)
~
~ -g Near Smelter, MT (a)
|[
I us I
I
~
o Superfund Site, KS (a)
"I
1 1
1
^ Unpolluted U.S. Soils (a)
I
1
Spring River Basin Mining District, KS, MO & OK (a)
1
:S U.S. Average (a)
o ££
Sacramento River System Acid-Mine Drainage (c)
I
in
Northeastern U.S. Lakes (d)
I I
0.0001 0.001 0.01 0.1 1 10 100 1000
Cadmium Concentration (see units at left)
WCC = Water Column Carnivore, BC = Benthic Carnivore, SWP = Stormwater Pond, TS = Tilled Soil, US = Unfilled Soil
a Source: ATSDR Toxicological Profile for Cadmium (ATSDR 2008)
b Source: Hazardous Substances Databank Record for Cadmium Compounds (HSDB 2005a)
c Source: Copper, Cadmium, and Zinc Concentrations in Aquatic Food Chains from the Upper Sacramento River and Selected Tributaries (Saiki et al. 1995)
d Source: Accumulation of Heavy Metals in Food Web Components Across a Gradient of Lakes (Chen et al. 2000)
C-44
-------
Exhibit 4-3. TRIM.FaTE Cadmium Concentrations in Fish and Calculated
Bioaccumulation Factors with Respect to Total Water Concentration
1 %
•£ 9
P
S
u_ £ II
.= '5 0.40
' " "" 666 3
s
ffi
I
Water Column Benthic Omnivore Water Column Benthic Carnivore Water Column
Herbivore Omnivore Carnivore
Partitioning Behavior
Because most cadmium emitted to the air will eventually deposit onto soils, cadmium has been
observed to partition primarily to soil when released to the environment (ATSDR 2008).
Cadmium mobility in soil depends strongly on soil pH, clay content, and availability of organic
matter—factors that determine whether the cadmium is dissolved or sorbed in surface soil. In
general, cadmium binds strongly to organic matter, rendering the metal immobile; however,
some plants are able to efficiently take up cadmium, thus providing an entry point for cadmium
into the food chain (ATSDR 2008).
Most cadmium in a natural water column is dissolved, but some adsorption to humic substances
and other organic complexing agents can occur. This behavior can be especially important in
polluted or organic-rich waters, playing a dominant role in cadmium transport (ATSDR 2008).
Concentrations in surface water tend to be lower than those in bed sediment because cadmium
readily adsorbs to mineral surfaces, hydrous metal oxides, and organic materials present in
sediment. Cadmium that has adsorbed to mineral surfaces in the sediment is not easily
bioaccumulated in the aquatic food web unless the sediment is disturbed, and the metal is
redissolved.
For the RTR screening scenario, 5.7 percent of cadmium in the pond compartment partitioned
to surface water, while 94.3 percent partitioned to sediment. The percentages of cadmium in
sorbed and dissolved states in surface water, sediment, and surface soil are presented in
Exhibit 4-4. The partitioning behavior modeled in TRIM.FaTE is consistent with the behavior of
cadmium in the natural environment. The presence of cadmium primarily in the dissolved state
in surface water and sorbed to sediment and surface soil is consistent with trends noted in
supporting literature.
C-45
-------
Exhibit 4-4. Fraction of Cadmium Mass Sorbed vs. Dissolved in
T
RIM.FaTE Compartments
Compartment
Sorbed
Dissolved
Surface Soil
100.0%
0.0%
Surface Water
4.6%
95.4%
Sediment
99.9%
0.1%
Concentrations in Ingestible Products
The major source of non-inhalation exposure to cadmium outside of occupational settings is
through dietary intake. Available data indicate that cadmium accumulates in plants, aquatic
organisms, and animals, offering multiple ingestion exposure pathways (ATSDR 2008).
However, actual cadmium levels in ingestible products can vary based upon type of food,
agricultural and cultivation practices, atmospheric deposition rates, conditions in environmental
media, and presence of other anthropogenic pollutants. General trends indicate that high levels
of cadmium can be found in green, leafy vegetables; peanuts; soybeans; and sunflower seeds.
Meat and fish generally contain lower amounts of cadmium, overall, but cadmium can be found
highly concentrated in certain organ meats, such as kidney and liver (ATSDR 2008). In a study
of cadmium concentrations in 14 food groups (including prepared foods), meat, cheese, and
fruits generally contained low levels of cadmium (ATSDR 2008).
Modeled cadmium concentrations in ingestible products are displayed in Exhibit 4-5 along with
cadmium concentrations reported in literature. In the screening scenario we assume that
individuals consume equal amounts of benthic carnivores and water column carnivores, so the
concentration for fish is given as the average of these two fish compartments. The cadmium
concentrations output by MIRC were consistent with reported values in all ingestible media
products. The products with higher reported cadmium levels in the literature, including soil,
plants, and fish, also contained the highest modeled concentrations.
To determine media types most relevant to exposure and risk, the estimated media
concentrations must be combined with ingestion exposure factors (i.e., higher concentrations do
not necessarily equal higher risk. The contribution of ingestion exposure pathways to the ADD
(and thus the HQ) for the different age categories are displayed in Exhibit 4-6. The HQ of 1 for
children aged 1-2 was used to determine the de minimis level for cadmium in the screening
scenario. The dominant exposure pathway for this age group was consumption of fruit
(protected and exposed), which comprises almost half of the total ADD. Fruit was a dominant
exposure pathway for all age groups, though less so for adults. The other dominant exposure
pathways were fish, vegetables, and soil. This trend is consistent with observed trends and
representative of the preferential bioaccumulative behavior of cadmium in the natural
environment.
C-46
-------
Exhibit 4-5. Summary of Modeled and Observed Concentrations of Cadmium in
Ingestible Media
i i Reported value was less than limit of detection (LOD = upper bound of bar)
i i Wall ip nr rang e fro m I iterature
Modeled concentration corresponding to de minimis threshold emission rate
Mean & Max, Urban Areas, Canada (a)
I I
¦5 Urban Areas, U.S. (b)
1
Dairy Products, U.S. (c)
1
„ Near Smelter, MT (b)
[
1
1
0 "3) J, Superfund Site, KS (b)
"~'"0 Unpolluted U.S. Soils (b)
1
1
¦g Carrots, Lead-Zinc Mining Area, England (a)
1
1
1
(J
^ Potatoes (Prepared) (c)
1
4-; ^ All Vegetables, Urban Areas (b)
I
1
2 ID
> Winter Squash (Prepared) (c)
1-1
All Fruits, Silver Mine-Contaminated Gardens, CO (c)
1 1
2 Cantaloupe (c)
1
0 All Fruits, Urban Areas (b)
1
Oranges (c)
I
^ Mean & Max, Meat & Poultry, Urban Areas, Canada (c)
1
1
H Meat, Poultry & Fish, Urban Areas (b)
1
^ Poultry (Prepared) (c)
I
^ Pork Products (Prepared) (c)
I
^ Meat, Poultry & Fish, Urban Areas (b)
I
Salmon Flesh (b)
1 1
Largemouth Bass, SWP, FL, Avg. (b)
~
~
.<2 Redear Sun fish, SWP, FL, Avg. (b)
Catfish, Mining-Contaminated Water, OK (a)
1
Trout. Remote Stream, MD (b)
I I
-0 Broccoli, Lead-Zinc Mining Area, England (a)
1
1
0 ¦§ Spinach (Prepared) (c)
1
1
w > All Vegetables, Urban Areas (b)
I
1
— All Fruits, SilverMine-Contaminated Gardens, CO (c)
1 1
!£ Strawberries (c)
I
1
0 All Fruits, Urban Areas (b)
~
~
uj Apples (c)
Eggs (Prepared) (c)
I
¦
1
1
S5 Urban Areas (b)
^ Beef (Prepared) (c)
I
~
m Meat, Poultry & Fish, Urban Areas (b)
1
1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1 .E+001 .E+011 .E+021 .E+03
Cadmium Concentration (mg/kg wetwt.*)
'except soil, which is expressed as mg/kg dry weight.
SWP = Stormwater Pond
a Source: Hazardous Substances Databank for Cadmium Compounds (HSDB 2005a)
b Source: ATSDR Toxicological Profile for Cadmium (ATSDR 2008)
c Source: Total Diet Study, Market Baskets 1991-3 through 2005-4 (FDA 2007)
C-47
-------
Exhibit 4-6. Estimated Contribution of Modeled Food Types to Cadmium
Ingestion Exposures and Hazard Quotient
~ Fruits 11 Vegetables ~ Fish DSoil n Meat, Dairy, & Eggs
Children
1-2
Children
3-5
Children
6-11
Children
12-19
Adults
20-70
C-4.3.3.2 Mercury
Behavior in the Environment
Mercury emitted by anthropogenic sources undergoes changes in form and species as it moves
through environmental media. The three most relevant forms of mercury in the environment are
elemental mercury, divalent inorganic mercury, and methyl mercury. Based upon reviewed
literature, some of the largest anthropogenic sources of mercury to air are facilities that process,
mine, or smelt mercury ores; industrial/commercial boilers; fossil fuel combustion activities
(primarily coal); cement production facilities; other urban and industrial facilities; and medical
and municipal waste incinerators (ASTDR 1999). Stack emissions can include a mixture of
elemental and divalent forms, mostly in the gaseous phase, with some divalent forms in particle-
bound phases (EPA 1997b). Although elemental mercury is more prevalent in the atmosphere,
divalent mercury can be found at higher concentrations near emissions sources. Elemental
mercury has a very long residence time in the atmosphere and will thus be distributed relatively
evenly on a global scale, resulting in negligible impacts from any single source on local and
regional scales. Divalent mercury, however, is removed from the atmosphere at a faster rate
than elemental mercury, and may be transferred to the surface near the emission source via wet
or dry deposition, where it appears to adsorb tightly to soil particles (EPA 1997b). Some
adsorbed divalent mercury can be reduced to elemental mercury and be revolatilized back into
the atmosphere. Methylation to methyl mercury by microbes can also occur in the soil;
however, divalent mercury is generally the dominant species in soil (EPA 1997b). Small
amounts of divalent and methyl mercury in soil can be transported to surface water through
runoff and leaching. Direct deposition to water can occur through atmospheric fallout. Once in
the water body, divalent mercury may be methylated through microbial activity, and both
divalent and methyl mercury may be further reduced to elemental mercury, which will reenter
the atmosphere. But, as in soil, divalent mercury will generally be the dominant species in both
C-48
-------
surface water and sediment. Methyl mercury is readily bioaccumulated and efficiently
biomagnified in aquatic organisms and is typically the species of greatest concern for mercury
exposure via the food chain.
Emission Profile
TRIM.FaTE assumes that only elemental mercury and divalent mercury are emitted from the
source (see Section 2.3); but once in the environment, these species may change form,
resulting in ingestion exposure to methyl mercury as well as divalent mercury. Relative to these
two species, elemental mercury does not comprise a significant amount of total mercury in
environmental media other than air. Because human health effects from exposure to methyl
mercury and divalent mercury are not additive, two de minimis rates for emissions of divalent
mercury were calculated for exposure to methyl mercury and divalent mercury, respectively, to
determine the most health-protective level. The emission rate calculated for the HQ of divalent
mercury was lower, and thus was used to define the de minimis emission threshold. At the de
minimis rate, the HQ for divalent mercury in children aged 1-2 is approximately twice as high as
that calculated for methyl mercury.
For the scale of this evaluation, we compared predicted concentrations of total mercury to
measured concentrations of total mercury in environmental media compartments and ingestible
products. In TRIM.FaTE, divalent mercury comprises from 83 to 100 percent of total mercury in
all environmental and ingestible media compartments except fish. Because methyl mercury
comprises approximately 90 percent of total mercury in fish, and this pathway is recognized as a
significant contributor to mercury exposure in humans, we also examine methyl mercury
bioaccumulation in fish and the contribution of this pathway to methyl mercury ingestion
exposure and HQ.
The de minimis emission rate used for mercury in the RTR screening scenario is 1.6E-01 TPY
of divalent mercury (based on an HQ of 1 for divalent mercury exposure for children aged 1-2).
Annual concentrations from year 50 used to calculate the de minimis threshold are presented
here, and these values are used for comparison. Modeled concentrations of mercury in all
compartments but unfilled soil continued to rise up to the end of the simulation, with the rate of
increase steadily diminishing over time.
Concentrations in Environmental Media
Exhibit 4-7 displays measured ranges of total mercury concentrations in environmental media
with TRIM.FaTE outputs for the screening scenario. The total mercury concentrations
estimated by TRIM.FaTE were generally consistent with reported values in all environmental
media compartments. Overall, modeled total mercury concentrations were generally within the
range of values reported for areas near a significant pollution point source (e.g., smelter, chlor-
alkali facility).
Methyl mercury is formed via microbial transformation of inorganic mercury in sediment, surface
water, and soil and readily accumulates in the tissues of planktivorous and piscivorous fish.
There seems to be a relationship between methyl mercury levels in fish and water pH, showing
higher levels of methyl mercury in fish found in acidic lakes (EPA 1997b). Lake alkalinity and
dissolved oxygen content may also influence the ability offish to bioaccumulate methyl mercury
(EPA 1997b). BAFs on the order of 105 to 107 have been reported for methyl mercury
concentrations in fish (Alpers et al. 2008). Modeled concentrations of mercury species in fish
compartments along with estimated methyl mercury BAFs are shown in Exhibit 4-8. The
C-49
-------
Exhibit 4-7. Summary of Modeled and Observed Concentrations of Total Mercury in Environmental Media
i i Reported value was less than limit of detection (LOD = upper bound of bar)
i i Value or range from literature
^^"Modeled concentration corresponding to de minimisthreshold emission rate
U.S.Min/Max, 1990-1995 (a)
42 Lakes and Rivers, NJ (b)
CD
CD
Paper Mills Using Chlorine (c)
Predatory Fish, ME (b)
BC | - WCC -I
Canadian Lakes Near Smelters (a)
E ^
"cs
13
CO CD
80 MN Lakes (b)
U.S. Lakes, Avg. (b)
Little Rock Lake, Wl (b)
o 5
(/> ^
8"°
CO CD
4=
,-S.
U1 c:
Near Mine/Smelter, Spain (b)
U.S. Chlor-alkali Facility, Avg. (b)
Typical U.S. (b)
TS
US I
~
CD
s
CO •
Crab Orchard Lake, IL (a)
U.S. Samples (b)
Lake Cresent, WA (b)
~
0.0001 0.001 0.01 0.1 1 10 100
Total Mercury Concentration (see units at left)
1000
BC = Benthic Carnivore, WCC = Water Column Carnivore, TS = Tilled Soil, US = Unfilled Soil
a Source: ATSDR Toxicological Profile for Mercury (ATSDR 1999)
b Source: Mercury Study Report to Congress (EPA 1997b)
c Source: National Study of Chemical Residues in Fish (EPA 1992a)
C-50
-------
Exhibit 4-8. TRIM.FaTE Mercury Concentrations, Speciation, and Calculated
Methyl Mercury Bioaccumulation Factors (in white boxes) in Fish
Compartments
0
1 | 0.15
S 2
122,2'
Water Column Benthic Omnivore Water Column Benthic Carnivore Water Column
Herbivore Omnivore Carnivore
~ Elemental
~ Divalent
~ Methyl
TRIM.FaTE outputs for the fish compartments mirror a trend noted in the literature in which
divalent mercury is observed more prominently in lower trophic level organisms, while methyl
mercury is the more prominent species in higher trophic level organisms (ATSDR 1999).
Calculated BAFs are consistent with reported values, with the higher trophic levels exhibiting
BAFs of approximately 105. The biomagnification of the methylated species in the aquatic
ecosystem can be observed in Exhibit 4-8, with modeled methyl mercury concentrations
increasing substantially from the lower to higher trophic levels.
Partitioning Behavior
Soil acts as the primary reservoir for divalent mercury emitted from anthropogenic sources. In
some cases, divalent mercury will be adsorbed onto forms of dissolved organic carbons
(DOCs), which are susceptible to runoff. However, most divalent mercury from atmospheric
deposition remains in the soil profile in the form of inorganic compounds bound to soil organic
matter. The ability of mercury to form complexes with soil organic matter is highly dependent on
soil conditions such as pH, temperature, and soil humic content. For example, mercury strongly
adsorbs to humic materials and sesquioxides in soil at pH > 4 and in soils with high iron and
aluminum content (ATSDR 1999). Although inorganic compounds containing divalent mercury
are relatively soluble, this complexing behavior with organic matter significantly limits mercury
transport. Only very small amounts of mercury present in soil are partitioned to runoff.
Elemental mercury present in soil (e.g., as a result of the reduction of divalent mercury) will
readily volatilize, especially in acidic soils, and this species therefore comprises very little of the
total mercury content in soil. Typically, methyl mercury comprises 1 to 3 percent of total
mercury in soil (EPA 1997b). For the TRIM.FaTE results in the RTR screening scenario,
divalent mercury comprised approximately 95 percent of total mercury in the surface soil
compartment; this result is consistent with environmental trends.
C-51
-------
Solubility of mercury in water depends on the species and form of mercury, water pH, chloride
ion concentrations, and other factors (ATSDR 1999). Solid forms of inorganic mercury
compounds will partition to particulates in the water column and then partition to the sediment
bed (EPA 1997b). Low pH favors the methylation of mercury in the water column, typically
performed by sulfur-reducing bacteria in anaerobic conditions. Methyl mercury generally
comprises 10 percent of total mercury observed in surface water (EPA 1997b). A considerable
amount (25 to 60 percent) of both divalent mercury compounds and methyl mercury will be
strongly particle-bound in the water column (EPA 1997b). The remaining mercury will be in a
dissolved state. Most of the elemental mercury produced as a result of reduction of divalent
mercury by humic acid will volatilize.
Screening scenario modeling results for mercury speciation in environmental media
compartments are displayed in Exhibit 4-9 along with the percent of each species that was
present in both adsorbed and dissolved states in surface soil, surface water, and sediment.
Consistent with observed trends, divalent mercury was the most prevalent species in modeled
surface soil, surface water, and sediment compartments, while methyl mercury was the
dominant species in fish. For the RTR screening scenario, 99 percent of total mercury in the
pond (i.e., surface water and sediment) was divalent mercury, while 0.7 percent was methyl
mercury and 0.2 percent was elemental mercury. The TRIM.FaTE outputs of approximately 3
percent methyl mercury in soil and 9 percent methyl mercury in surface water are also
consistent with trends noted in the literature (EPA 1997b).
Exhibit 4-9. TRIM.FaTE Mercury Speciation and Partitioning in
Environmental Media Compartments
~ Divalent ~ Elemental ~ Methyl
Surface
Soil
Surface Water
Chemical
Methy
Mercury
Elemental Mercury
Divalent Mercury
Sorbed
Dissolved
Sorbed
Dissolved
Sorbed
Dissolved
Surface soil
100.0%
0.0%
100.0%
0.0%
100.0%
0.0%
Surface Water
80.7%
19.3%
4.8%
80.7%
83.0%
17.0%
Sediment
100.0%
0.0%
100.0%
0.0%
100.0%
0.0%
Overall, the partitioning behavior in the RTR screening scenario is consistent with the behavior
of mercury in the natural environment. Within the pond compartment of the model, 99.6 percent
of divalent mercury, 98.6 percent methyl mercury, and 94.5 percent of elemental mercury
partitioned to the sediment. Though all mercury in natural soils and sediment may not be in the
C-52
-------
sorbed state, the percentage of mercury that does not complex with organic matter would likely
be negligible. The behavior in the TRIM.FaTE surface water compartment is consistent with
trends observed in natural waters in which divalent and methyl mercury species exhibit a strong
preference for adsorption to suspended matter in the water column. These species may also
form complexes with dissolved organic carbon and thus be considered as dissolved in the water
column; however, this is not as prevalent as complexation with suspended solids. Conversely,
elemental mercury does not adsorb readily to suspended particles, and is therefore found
primarily in the dissolved form in natural waters, where it is highly susceptible to revolatilization
into the atmosphere. The high percent dissolved elemental mercury estimated by TRIM.FaTE
(when compared to percent dissolved divalent or methyl mercury) in surface water is consistent
with this trend. It has been estimated that more than 97 percent of dissolved gaseous mercury
in the water column is elemental mercury (EPA 1997b).
Concentrations in Ingestible Products
Available data indicate that mercury bioaccumulates in plants, aquatic organisms, and terrestrial
animals, providing multiple ingestion exposure pathways (EPA 1997b, ASTDR 1999). Low
levels of mercury are found in plants, with leafy vegetables containing higher concentrations
than potatoes, grains, legumes, and other vegetables and fruits (ASTDR 1999, EPA 1997b).
Cattle are capable of demethylating mercury in the rumen, and therefore store little of the small
amount of mercury that is transferred to livestock via foraging or silage/grain consumption.
Mercury content in meat and cow's milk is therefore low (ASTDR 1999). Concentrations of
methyl mercury in fish are generally highest in larger, older specimens at the higher trophic
levels (EPA 1997b).
Concentrations of total mercury in ingestible products are presented in Exhibit 4-9. Though data
on mercury in foods other than fish are not abundant in the literature, total mercury
concentrations output by MIRC were generally consistent with the reported values that were
available. The exposure pathways that most influence the mercury HQs in the model can be
seen in Exhibit 4-10, which displays the contributions of ingestion exposure pathways to the
divalent mercury and methyl mercury ADDs for all age groups analyzed in the screening
scenario. The dominant exposure pathway for divalent mercury in children aged 1-2 was
ingestion of contaminated soil. Divalent mercury accumulates readily in the upper 20 cm of the
soil profile, where it is accessible to children that may frequently ingest soil (EPA 1997b, EPA
2008). Exposures to children aged 1 to 2 are 4 to 5 times higher than exposures to children and
adults aged 12 to 70 years. This is driven by a very high soil ingestion rate in children, and the
assumption that the soil consumed is next to the emission source. In older children and adults
who do not frequently ingest soil, fruits and vegetables provide the greatest exposure pathway
for divalent mercury. Though divalent mercury is not considered to be readily taken up by
plants from the soil, atmospheric deposition may figure strongly into elevated plant
concentrations. In addition to the deposition of divalent mercury directly onto plant surfaces,
divalent mercury may accumulate in edible portions of plants due to the transformation of
elemental mercury following deposition. The relatively low contributions by meat and fish
consumption are consistent with observed trends indicating that the divalent mercury biotransfer
is not high between plants and animals, and that the most persistent mercury species in fish is
methyl mercury, rather than divalent mercury (EPA 1997b, ATSDR 1999).
C-53
-------
Exhibit 4-10. Summary of Modeled and Observed Concentrations of Total Mercury in
Ingestible Media
' 1 Reported value was less than limit of detection (LOD = upper bound of bar)
i iValueorrangefrom literature
Modeled concentration corresponding to de minimis threshold emission rate
A Non-Detect With No Limit of Detection Specified
Germany and U.K. (a)
1
¦5 Milk, Germany (b)
~
~
Dairy Products, U.S. (a)
Mine/Smelter, Spain (c)
c
~
U.S. Chlor-alkali Facility (c)
Typical U.S. Soils (c)
1
Carrot/Potato Background Means, Netherlands (b)
L
II
~I
0 Root Vegetable, U.S. Avg.(a)
Potatoes, U.S. (a) L
cn All Fruits/Vegetables, Belgium, Germany & U.K. (a)
1 1
Maize (c)
[
F
O
cl Legume Vegetables, U.K. (a)
¦0 All Fruits/Vegetables, Belgium, Germany& U.K. (a)
i
oE All Fruits, U.S. (a)
~
~
cl Garden Fruits, U.S. (a)
^ Chicken, Raw & Lunch Meat, 33-80% Hg2 (c)
1
i
3 Average, Germany (b)
1
~
~
^ Chicken,Background Level, Japan (c)
^ Background Level, Japan (c)
~
Raw & Sausage Hg2 (c)
—1
1
U.S. Min/Max, 1990-1995 (b)
1
i
.c 42 Lakes and Rivers, NJ (c)
L
j
L"- Near Paper Mills Using Chlorine (d)
i
Predatory Fish, ME (c)
II
dm
]
¦o « Alfalfa (Above-Ground Parts), Near Mine/Smelter, Spain (c)
~
o ¦$ U.S. Chlor-alkali Plant (c)
i
w > Background Means, Netherlands (c)
¦o All Fruits, Belgium, Germany & U.K. (a)
1 i
o E Apples, BackgroundMean, Netherlands (c)
k
[
lu Exposed Friits (e) /
Average, Germany (b)
~
Denmark, Germany & U.K. (a)
Natural and Abnormal (a)
~
Raw Beef (c)
1
c
S Background Level, Japan (c)
Germany (b)
1
i
1.E-6 1.E-5 1.E-4 1.E-3 1.E-2 1.E-1 1.E+0 1.E+1 1.E+2
MercuryConcentration (mg/kg wetwt .*)
'except soil, which is expressed as mg/kg dry weight
a Source: Hazardous Substances Databank Report for Mercury Compounds (HSDB 2005b)
b Source: ATSDR Toxicological Profile for Mercury (ATSDR 1999)
c Source: Mercury Study Report to Congress (EPA 1997b)
d Source: National Study of Chemical Residues in Fish (EPA 1992a)
e Source: Total Diet Study, Market Baskets 1991-3 through 2005-4 (FDA 2007)
C-54
-------
The dominant exposure pathway for methyl mercury in all age groups was ingestion offish
(Exhibit 4-11). Though methyl mercury concentrations in fish were very high (approximately 88
percent of total mercury) in Exhibit 4-11, and fish consumption can contribute substantially to
methyl mercury exposure, the HQ for this species (Exhibit 4-12) is likely lower than that of
divalent mercury in children aged 1-2 due to a relatively low fish consumption rate in
combination with high exposure to divalent mercury from multiple pathways, but primarily
through ingestion of soil. The divalent and methyl mercury age-group-specific HQs are
displayed in Exhibit 4-12. In humans aged 6-70, the methyl mercury HQ is higher than that for
divalent mercury, which is likely the result of lower soil consumption rates and higher fish
consumption rates.
Exhibit 4-11. Estimated Contribution of Modeled Food Types to Divalent Mercury
and Methyl Mercury Ingestion Exposures
~ Fruits & Vegetables ~ Eggs, Pork, & Poultry
Its
Children Children Children Children Adults
~ Fruits & Vegetables
~ Eggs, Pork, & Poultry
6 1 1
¦ ¦ ¦ ¦ ¦
Exhibit 4-12. Estimated Contribution of Summed Modeled Food Types to
Divalent Mercury and Methyl Mercury Hazard Quotients
~ Divalent Mercury ~ Methyl Mercury
N-m
C-55
-------
C-4.3.3.3 Dioxins (2,3,7,8-TCDD)
Behavior in the Environment
Polychlorinated benzo-dioxins and -furans15 are reported in the National Emission Inventory
(NEI) as individual congeners, congener groups, or toxic equivalents (TEQs). The dioxin
congener 2,3,7,8-TCDD was used as a surrogate to evaluate risks for chlorinated dibenzo-p-
dioxins in the screening scenario. Of the chlorinated dioxin congeners, this compound has been
the most widely studied, and is considered to be one of the two most toxic congeners to
vertebrates (ASTDR 1998). As a result, the World Health Organization (WHO) assigned a toxic
equivalency factor (TEF) of 1 to this congener, meaning that the concentrations of the other
congeners are scaled relative to the toxicity of 2,3,7,8-TCDD (See Attachment C-2 for TEFs).
We deemed this species an appropriate surrogate to represent the fate and transport behavior
of the dioxin group.
Incineration/combustion processes are believed to be the primary emission sources for
chlorinated dioxins (ASTDR 1998). The five stationary source categories that generate the vast
majority of 2,3,7,8-TCDD emissions in the United States are municipal waste incineration,
medical waste incineration, hazardous waste kilns from Portland cement manufacturing,
secondary aluminum smelting, and biological incineration. Dioxins emitted to the atmosphere
may be transported long-range as vapors or bound to particulates, depositing in soils and water
bodies in otherwise pristine locations far from the source. Though airborne dioxins are
susceptible to wet and dry deposition, most dioxins emitted to the atmosphere through
incineration/combustion processes are not deposited close to the source (ASTDR 1998).
Dioxins strongly adsorb to organic matter in soil and show very little vertical movement,
particularly in soils with a high organic carbon content (ASTDR 1998). Most surface water
deposition occurs through dry deposition from the atmosphere and from wind-transported
eroded soil particles contaminated with dioxins. Most dioxins entering the water column will
partition to bed sediment, which appears to be the primary environmental sink for this chemical
group (EPA 2004b). Although dioxins bound to aquatic sediment will primarily be buried in the
sediment compartment, some resuspension and remobilization of congeners may occur as a
result of disturbance of sediments by benthic organisms (ASTDR 1998).
Bioaccumulation factors in fish are high for 2,3,7,8-TCDD as a result of the lipophilic nature of
chlorinated dioxins. Though the processes by which freshwater fish accumulate dioxins are not
well understood, it has been shown that fish and invertebrates both bioaccumulate congeners
when exposed to contaminated sediments, and bioconcentrate congeners dissolved in water
(EPA 2004b). However, because most dioxins in the aquatic environment are adsorbed to
suspended particles, it is unlikely that direct uptake from the water is the primary route of
exposure for most aquatic organisms at higher trophic levels (ASTDR 1998). At the lower
trophic levels, the primary route of exposure appears to be through uptake of water in
contaminated sediment pores, whereas the primary route of exposure in the higher trophic
levels appears to be through food chain transfer. Following ingestion, some fish are capable of
slowly metabolizing certain congeners, such as 2,3,7,8-TCDD, and releasing the polar
metabolites in bile. This may ultimately limit bioaccumulation at higher trophic levels (ASTDR
1998). Measured bioaccumulation factors for 2,3,7,8-TCDD are not widely reported due to the
difficulty of detecting trace levels of congeners in ambient water. Reported bioconcentration
factors, which can be measured in laboratory conditions, but do not account for exposure
15 Group is commonly referred to as "dioxins;" this short hand is used in this section.
C-56
-------
pathways other than direct uptake from water, range from 200-100,000, depending on the
species and medium (ASTDR 1998).
Emission Profile
The 2,3,7,8-TCDD de minimis emission rate used for the RTR screening scenario is 3.2E-08
TPY (based on lifetime cancer risk of 1-in-a-million). Although the annual concentration from
year 50 is used for comparison, modeled concentrations of 2,3,7,8-TCDD either reached an
approximate steady state or had begun to level off in all compartments by year 20.
Concentrations in Environmental Media
Measured ranges of total dioxin TEQ concentrations16 in environmental media are displayed
with modeled TEQ concentrations of 2,3,7,8-TCDD in Exhibit 4-13. Modeled concentrations
and BAFs in fish compartments are shown in Exhibit 4-14. The 2,3,7,8-TCDD concentrations
output by the RTR screening scenario were consistent with reported dioxin TEQs in all
environmental media compartments (Exhibit 4-13). Because dioxin congeners are present at
trace levels, analytical instruments must be very sensitive if congeners are to be detected in
environmental media. As a result, limits of detection in many instruments may not be sufficiently
sensitive to produce quantifiable results for very low levels of dioxin congeners. Many of the
concentrations output by TRIM.FaTE are low enough to be considered nondetects by most
analytical instruments.
Though the 2,3,7,8-TCDD concentrations in environmental media estimated by TRIM.FaTE are
generally low, they are not outside the range of plausible values when considered as a
surrogate for the dioxin profile. In the screening scenario, the amount of chemical mass in the
water column is small relative to the amount in the sediment. This is consistent with the trend
that most congeners are removed from water bodies through adsorption to organic matter in soil
and sediment. The screening scenario BAFs are also consistent with observed trends that
indicate that 2,3,7,8-TCDD accumulates much less readily in herbivorous fish than in
carnivorous fish that consume other contaminated organisms (ATSDR 1998).
16 It was not possible to confirm that TEQ concentrations reported in the literature and summarized here were all
estimated using the most recent TEF scheme adopted by the WHO in 1998. TEQ concentrations reported here may
represent values determined using TEFs from EPA's 1989 scheme, WHO's 1994 scheme, or WHO's 1998 scheme
(EPA 1997b).
C-57
-------
Exhibit 4-13. Summary of Modeled 2,3,7,8-TCDD Concentrations and Observed Total Dioxin TEQ Concentrations in
Environmental Media
i i Reported value was less than limit of detection (LOD = upperbound of bar)
i i Value or range from literature
^^"Modeled concentration corresponding to de minimis threshold emission rate
Great Lakes (a)
Non-background Sites, U.S. (a)
BC
Southern MS Lakes (a)
Rural NY (b)
Industrial Area, Venice Lagoon (a)
Manmade Lakes, MS (a)
E
North America (a)
CO
Polluted River, Japan (a)
Waste-to-Energy Facility, OH (a)
Industrial, Austria (a)
Urban, North America (a)
co
Rural, North America (a)
Stormwater Runoff (a)
North America (a)
TS
US
WCC
0.00001 0.0001 0.001 0.01 0.1 1 10 100 1000
2,3,7,8-TCDD Concentration (see units at left)
BC = Benthic Carnivore, WCC = Water Column Carnivore, TS = Tilled Soil, US = Unfilled Soil
a Source: Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds (EPA 2004b)
b Source: Toxicological Profile for Chlorinated Dibenzo-p-dioxins (ATSDR 1998)
C-58
-------
Exhibit 4-14. TRIM.FaTE 2,3,7,8-TCDD Concentrations in Fish and Calculated
Bioaccumulation Factors with Respect to Total Water Concentration
3.0E-08
483,900
2.5E-08
403,250
.n'
^ I
f
11
¦g g>
8|
8 g
2.0E-08
1.5E-08
1
322,600 2
241,950 1
3
E
161,300 m
Water Column Benthic Omnivore Water Column Be nth ic Carnivore Water Column
Herbivore Omnivore Carnivore
Partitioning Behavior
Chlorinated dioxins have been observed to partition mostly to soil when released to the
environment, and sorbed congeners are unlikely to leach or volatilize out of the soil profile. In
soils with a high organic carbon content, congeners below the top few millimeters of the soil are
very strongly adsorbed to organic matter and exhibit very little migration. Most dioxins
deposited onto soil are expected to remain buried in the soil profile, and erosion of
contaminated soil particles is the only significant mechanism for transport to water bodies.
Because of the hydrophobic nature of the 2,3,7,8-TCDD congener, and its affinity for organic
carbon, 70 to 80 percent of the congener is expected to bind to suspended organic particles in
natural waters. The remainder in the water column is associated with dissolved organic
substances (ASTDR 1998). Because most of the dioxins in water are in a sorbed state, the
ultimate fate of most congener-laden particles is in the bed sediment. For the RTR screening
scenario, more than 99.5 percent of 2,3,7,8-TCDD in the pond compartment was found in the
sediment. The percentages of 2,3,7,8-TCDD in sorbed and dissolved states in surface water,
sediment, and soil are presented in Exhibit 4-15.
Exhibit 4-15. Fraction of 2,3,7,8 - TCDD Mass Sorbed vs. Dissolved
in TRIM.FaTE Compartments
Compartment
Sorbed
Dissolved
Surface Soil
100.0%
0.0%
Surface Water
75.1%
24.9%
Sediment
100.0%
0.0%
C-59
-------
The partitioning behavior exhibited 2,3,7,8-TCDD in TRIM.FaTE is consistent with observed
trends. In the surface water compartment, approximately 75 percent of 2,3,7,8-TCDD was
present in the sorbed state, which is within the reported 70-80 percent range reported in the
literature. In the sediment and soil compartments, 100 percent of the congener is in the sorbed
state, which is representative of the strong preference of dioxins for adsorption to soil and
sediment organic matter.
Concentrations in Ingestible Products
The primary source of non-inhalation exposure to dioxins outside of occupational settings is
through dietary intake, which accounts for over 90 percent of daily dioxin intake (ATSDR 1998).
Available data indicate that dioxins concentrate in plants, aquatic organisms, and animals,
offering multiple ingestion exposure pathways. However, actual congener levels in ingestible
products can vary based upon type of food, agricultural and cultivation practices, atmospheric
deposition rates, conditions in environmental media, and presence of other anthropogenic
pollutants. Dioxins appear to enter the terrestrial food chain primarily through vapor phase
deposition onto plant surfaces, which are then consumed by larger animals. Another major
source of exposure to dioxins is through ingestion of contaminated soil by animals.
Observed trends indicate that meat, dairy, and fish are the dominant exposure pathways,
comprising 90 percent of dioxin dietary intake (ATSDR 1998). Though concentrations in
vegetables are generally exceptionally low, root vegetables normally contain slightly higher
concentrations of dioxins than vegetables that are affected primarily by atmospheric deposition,
such as lettuce and peas (ATSDR 1998). The 2,3,7,8-TCDD concentrations in ingestible
product compartments are displayed in Exhibit 4-16.
Data for concentrations of dioxin congeners in ingestible products are not abundant and
sophisticated analytical instruments with sufficiently sensitive limits of detection were not widely
available for older studies, which likely resulted in a greater number of nondetects in samples.
As a result, the data available for comparison was limited, but concentrations of 2,3,7,8-TCDD
in ingestible products were generally consistent with the available dioxin TEQ values (Exhibit
4-16). As noted in the literature, the concentration in the fish compartment for the screening
scenario was at least one order of magnitude greater than those in the other compartments.
Among the compartments with the lowest concentrations were fruits and vegetables, which do
not readily accumulate 2,3,7,8-TCDD in the environment. The percent contributions of ingestion
exposure pathways to the Lifetime ADD are displayed in Exhibit 4-17.
Consistent with trends reported in the literature, consumption of meat, fish, and dairy products
contribute to over 90 percent of lifetime dioxin exposure in the screening scenario. Daily intakes
of 2,3,7,8-TCDD from milk, produce, and fish have been reported in the literature to comprise 27
percent, 11 percent, and 10 percent, respectively, of the total daily intake in the general
population. However, some studies note that specific subpopulations, such as subsistence
farmers and fishers, may have very different exposure profiles in which fish, meat, and dairy
drive congener exposure (ATSDR 1998). Given the subsistence diet modeled in the RTR
screening scenario, the high exposure from consumption of fish, dairy, and beef are appropriate
within the context of this analysis.
C-60
-------
Exhibit 4-16. Summary of Modeled 2,3,7,8-TCDD Concentrations and Observed Total Dioxin TEQ
Concentrations in Ingestible Media
i iRapnrtaH value was less than limit of detection (LOD = upper bound of bar)
i i Valueor rangefrom literature
^^^Modeled concentration corresponding to de minimis thresh old emission rate
Dairy Products, Southern MS (a)
1 1
S Milk and Dairy, North America (a)
1
1
Dairy Products, Rural NY (b)
1
1
CT Waste-to-Energy Facility, OH (a)
¦
1
1 1
_ 2^ Industrial, Austria (a)
¦- -O . w
~
1
1/5 Urban, North America (a)
Rural, North America (a)
1
¦g Average, Germany (a)
~
~
a Potatoes, UK (a)
^ ¦ Above-ground Vegetables, Germany (a)
~
~
£ p
-> Non-green Vegetables, U.K. (a)
^ — All Fruits, Average, Germany (a)
1
~
^ Fresh Fruit, U.K. (a)
1
1
North America (a)
3-
1 1
3 Chicken, U.S. (a)
1
^ Chicken, MS (a)
~
North America (a)
1
1
o U.S. Samples (a)
1
1
Pork Sausage, MS (a)
~
Great Lakes (a)
1 1
^ Non-background Sites, U.S. (a)
1
1
^ Southern MS Lakes (a)
1
1
Rural NY (b)
1 1
-o ® Above-ground Vegetables, Germany (a)
¦
~
CO CD
o Non-green Vegetables, U.K. (a)
8"
^ :> Green Vegetables, U.K. (a)
1
^ ~ All Fruits, Average, Germany (a)
~
i-1-1 l*- Fresh Fruit, U.K. (a)
1
U.K. Samples (a)
~
[
~
~
5 Southern MS Samples (a)
North America (a)
Beef Products, U.S. (b)
1
1
o North America (a)
m v '
U.S. Samples (a)
~
1.E-13 1.E-12 1.E-11 1.E-10 1.E-9 1.E-8 1.E-7 1.E-6 1.E-5 1.E-4 1.E-3
2,3,7,8-TCDD Concentration (mg/kg wetwt.*)
'except for soil, which is expressed as mg/kg dry weight.
a Source: Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds (EPA 2004b)
b Source: Toxicological Profile for Chlorinated Dibenzo-p-dioxins (ATSDR 1998)
C-61
-------
Exhibit 4-17. Contribution of Modeled Food Types to 2,3,7,8-TCDD
Ingestion Exposures (mg/kg/day)
Fruits & Vegetables,
4.7E-14, soil,
IVo
2.5E-14,
<1%
Eggs, Pork, & Poultry,
1.1E-13,
2%
Beef,
6.2E-13,
9%
Dairy,
2.1E-12,
32%
Fish,
3.7E-12,
56%
C-4.3.3.4 PAHs (Benzo[a]pyrene)
Behavior in the Environment
Emissions of POMs are often reported in terms of unspeciated or partially speciated groups,
such as total PAHs, because they are often found in the environment as mixtures of two or more
compounds exhibiting comparable behavior and toxicity. Because there are relatively few data
concerning the behavior and toxicity of individual PAH compounds, EPA has proposed
assigning a relative potency factor to PAHs based on the relative toxicity of these compounds to
the most carcinogenic PAH compound(s) (EPA 1993). Like TEFs, this allows for risks from
exposure to certain PAH compounds that are likely carcinogens to be determined relative to the
toxicity of other PAH compounds that have been identified as probable carcinogens.
Benzo[a]pyrene, a high molecular weight PAH identified by EPA as a probable carcinogenic
compound (possibly the most potent carcinogen of the PAH group), was used to represent
PAHs in the screening scenario. A relative potency factor of 1 has been proposed by EPA for
this compound (EPA 1993). However, because the relative potency index has not been widely
adopted by the scientific community, modeled concentrations of benzo[a]pyrene are mostly
compared to reported concentrations of this species, rather than to total PAHs. Data for
benzo[a]pyrene comprises much of the available exposure information on carcinogenic PAHs
for the last few decades.
PAHs can enter the atmosphere as a result of a variety of combustion processes, both natural
and anthropogenic. Based upon reviewed literature, stationary emission sources account for
approximately 80 percent of total annual PAH emissions. Though the primary producer of
stationary source PAH emissions is thought to be residential wood burning, other processes
C-62
-------
such as power generation; incineration; coal tar, coke, and asphalt production; and petroleum
catalytic cracking are also major contributors (ASTDR 1995). PAHs emitted to the atmosphere
can travel long distances in vapor form or attached to particles, or they can deposit relatively
close to an emission source via wet or dry deposition onto water, soil, and vegetation. In the
atmosphere, PAHs are found primarily in the particle-bound phase, and atmospheric residence
time and transport distances are highly influenced by climatic conditions and the size of the
particles to which they are bound (ASTDR 1995).
As a result of sustained input from anthropogenic combustion sources and other sources, PAHs
are ubiquitous in soil. Like other high molecular weight PAHs, benzo[a]pyrene strongly adsorbs
to organic carbon in soil, indicating that adsorption to soil particles will limit the mobility of these
compounds following deposition to soil. Most PAHs enter the water column directly through
atmospheric fallout (ASTDR 1995). Following deposition onto surface waters, approximately
two-thirds of PAHs adsorb strongly to sediment and suspended particles, while only small
amounts revolatilize back to the atmosphere (ASTDR 1995). Aquatic organisms may
accumulate PAHs via uptake of water, sediment, or food. Though fish and other organisms
readily take up PAHs from contaminated food, biomagnification generally does not occur
because many organisms are capable of rapidly metabolizing them (ASTDR 1995). As a result,
concentrations of PAHs have generally been observed to decrease with increasing trophic
levels (ASTDR 1995). Based upon observed data, bioaccumulation factors in fish are also not
expected to be especially high because fish are able to readily metabolize the compound.
BCFs in the range of 10-10,000 have been reported for fish and crustaceans, with the higher
end of the range attributable to greater accumulation of the higher molecular weight
compounds, such as benzo[a]pyrene (ASTDR 1995). Higher BCFs have also been observed in
species at lower trophic levels, and BAFs will likely be higher in fish as a result of increased
exposure from diet. Additionally, sediment-dwelling organisms may experience increased
exposure to PAHs through association (e.g., direct uptake and/or consumption) with
contaminated sediment (ASTDR 1995).
Emission Profile
The benzo[a]pyrene de minimis emission rate used for the RTR screening scenario is 2.3E-03
TPY (based on lifetime cancer risk of 1-in-a-million). Although the annual concentration from
year 50 is used for comparison, modeled concentrations benzo[a]pyrene either reached a
steady state or had begun to level off in all compartments by year 10.
Concentrations in Environmental Media
Measured ranges of benzo[a]pyrene (and occasionally PAH-group concentrations) in
environmental media are presented in Exhibit 4-18 with concentrations from the screening
scenario. Modeled concentrations and BAFs in fish compartments are shown in Exhibit 4-19.
The benzo[a]pyrene concentrations output by the RTR screening scenario were consistently
lower than values reported in the literature. Three main factors likely contributed to this trend.
These are the high background values resulting from ubiquitous nature of PAHs in the
environment, limited availability of benzo[a]pyrene-specific data, and use of conservative
exposure factors to calculate the de minimis threshold. Firstly, due to the quantity of PAHs that
are emitted from mobile sources (-20 percent), as well as stationary synthetic and natural
sources, PAHs are typically present in the environment at relatively high background levels.
C-63
-------
Exhibit 4-18. Summary of Modeled and Observed Concentrations of Benzo[a]pyrene in Environmental Media
~ Reported value was less than limit of detection (LOD = upper bound of bar)
~ Value or range from literature
-Modeled concentration corresponding to de minimis threshold emission rate
"3> > Various Aquatic Organisms
BC
WCC
^ -g Boston Harbor
§3 (High Molecular Mass PAHs)
"a CD
CD ^
O 1=D
Great Lakes
o H
00
o
CT3 CD
GO
CD
Contaminated Sites
Urban
Agricultural
Rural
[js\
[us]
~
CD
CO
co ¦
co
Industrial Effluent
(Individual PAHs)
Urban Runoff
(Individual PAHs)
Great Lakes
1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E + 00 1.E + 01 1.E + 02 1.E + 03 1.E + 04 1.E + 05 1.E + 06
Benzo[a]pyrene Concentration* (see units at left)
* Source of literature data: ATSDR Toxicological Profile for Polycyclic Aromatic Compounds (PAHs) (ATSDR 1995). Data are for benzo[a]pyrene (BaP)
exclusively, unless otherwise noted.
C-64
-------
Exhibit 4-19. TRIM.FaTE Benzo[a]Pyrene Concentrations in Fish and Calculated
Bioaccumulation Factors with Respect to Total Water Concentration
4.5E-05
£
O)
o ^
11
If
8 g
3.0E-05
2.0E-05
1.0E-05
30,348
20,232 2
O
s
16,860 c
O
13,488
10,116 a
6,744
Water Column Benthic Omnivore Water Column Benthic Carnivore Water Column
Herbivore Omnivore Carnivore
The concentrations output from the screening scenario consider only emissions from a single
facility. As a result, it is not unreasonable that some of the TRIM.FaTE outputs were several
orders of magnitude smaller than those reported in the literature, as background exposure will
be higher than incremental exposure. Secondly, though measured data for benzo[a]pyrene
were used when available, concentrations in many media were available only for groups of
PAHs or total PAHs. Thus, those ranges are not representative of benzo[a]pyrene alone and
may contain values that are higher— and ranges that are wider— than those for a single
chemical. Thirdly, it should be noted that in order to maintain a health-protective approach to
screening emissions, a collection of moderately conservative exposure factors were used, which
likely resulted in low, but not implausible values. For these reasons, we believe that the
TRIM.FaTE outputs for concentrations of benzo[a]pyrene in environmental media are within the
range of plausible values for this chemical. However, because of this discrepancy between the
reported data and the screening scenario outputs, further investigation is necessary regarding
site-appropriate biotransfer factors for facilities that do not pass the screen as a result of PAH
emissions.
The screening scenario BAFs for benzo[a]pyrene in fish compartments were consistent with
trends reported in the literature. BAFs for all but the WCH compartment were in the range of
5,500 to 11,500. No biomagnification of this chemical was exhibited in the fish compartments in
the screening scenario, and the highest concentration of benzo[a]pyrene was in the WCH
compartment, which represents the fish with the lowest trophic level evaluated in the scenario.
The BAF for the WCH compartment in the screening scenario is approximately 28,000.
Partitioning Behavior
C-65
-------
Data suggest that benzo[a]pyrene partitions mainly to soil (82 percent) and sediment (-17
percent) following deposition to a 1km2 area adjacent to an emission source (ASTDR 1995).
Once in soil, PAHs can volatilize, undergo abiotic or biotic degradation, accumulate in plants, or
remain sorbed to soil organic matter. High molecular weight PAHs, such as benzo[a]pyrene,
tend to adsorb more strongly to organic carbon than PAHs having lower molecular weights
(ASTDR 1995). As a result of this affinity for organic carbon, volatilization is not an important
loss mechanism for benzo[a]pyrene from soils (ASTDR 1995). Because of its low solubility and
affinity for organic carbon, most benzo[a]pyrene is expected to be particle-bound in natural
waters. Less than one-third of PAHs in aquatic systems are generally present in the dissolved
phase (ASTDR 1995). The remainder may be associated with suspended particles in the water
column or particles that have settled on the bed sediment. Because most of the benzo[a]pyrene
found in natural water is in a sorbed state, the ultimate fate of most contaminant-laden particles
is burial in the bed sediment. For the RTR screening scenario, >98.8 percent of benzo[a]pyrene
in the pond compartment partitioned to the benthic sediment. The percentage of
benzo[a]pyrene in sorbed and dissolved states in soil, surface water, and sediment are
presented in Exhibit 4-19.
Exhibit 4-20. Fraction of Benzo[a]Pyrene Mass Sorbed vs.
Dissolved in TRIM.FaTE Com
partments
Compartment
Sorbed
Dissolved
Surface Soil
100.0%
0.0%
Surface Water
30.7%
69.3%
Sediment
100.0%
0.0%
The partitioning behavior of benzo[a]pyrene in the screening scenario is generally consistent
with trends reported in the literature. The presence of this chemical in the sorbed state in the
soil and sediment compartments is consistent with reported trends. A different trend was
observed in the surface water compartment where more of the chemical was estimated by
TRIM.FaTE to be dissolved. The dissolved concentration in TRIM.FaTE is affected by
suspended sediment concentration, organic carbon content, and suspended sediment
deposition and resuspension rates. Additional evaluation may be required to determine the
specific factors affecting this behavior.
Concentrations in Ingestible Products
The primary source of non-inhalation exposure to benzo[a]pyrene outside of occupational
settings is through dietary intake. Exposure may depend equally on the origin of the food
(higher values often recorded at contaminated sites) and the method of food preparation (higher
values reported for food that is smoked or grilled). PAHs have been observed to bioaccumulate
in aquatic organisms and terrestrial animals through uptake of contaminated water, soil, and
food. However, these compounds are readily metabolized by higher trophic level organisms,
including humans, so biomagnification is not considered to be significant (ASTDR 1995). Plants
accumulate PAHs primarily through atmospheric deposition, but chemical concentrations tend to
be below detection levels. In general, grains and cereals may contain slightly higher
concentrations of benzo[a]pyrene than fruits and vegetables. PAHs in meat have also been
observed at concentrations below detection levels up to higher concentrations when the meat is
smoked. Similar concentrations have been reported for fish, with smoked fish concentrations
C-66
-------
sometimes quadruple those found in terrestrial animals. Because PAH concentrations are
highest in products that are smoked or grilled, most of the available data for benzo[a]pyrene in
food is for products that have been prepared using these processes. As a result, reported
values may be significantly higher than those output by MIRC. Measured concentrations of
benzo[a]pyrene in ingestible products are presented in Exhibit 4-21 along with RTR screening
scenario concentrations.
The RTR screening scenario concentrations were generally lower than—or at the low end of—
the reported ranges for benzo[a]pyrene in ingestible products. This trend is likely the result of
background exposure in reported measurements and available data that is skewed toward
concentrations in highly contaminated products. Considering these mitigating factors, the RTR
screening scenario output concentrations are within the range of plausible values for
benzo[a]pyrene in ingestible products. The percent contributions of ingestion exposure
pathways to the lifetime ADD for benzo[a]pyrene are displayed in Exhibit 4-22.
No single exposure pathway in the RTR screening scenario appears to drive human exposure
to benzo[a]pyrene, but dairy, vegetables, and fruits are the three most dominant pathways. This
is consistent with observations indicating that only low concentrations of this chemical are
present in most ingestible products due to the ability of most plants and animals to metabolizing
it. It is also consistent with data suggesting that biotransfer factors between soil and
plants/animals are relatively low.
C-67
-------
Exhibit 4-21. Summary of Modeled and Observed Concentrations of Benzo[a]pyrene in
Ingestible Media
i i Reported value was less than limit of detection (LOD = upper bound of bar)
i 'Value or range from literature
Modeled concentration corresponding to de minimis threshold emission rate
Cheese (a
Dairy Products, Italy
Dairy Products, U.S.
~
Contaminated Sites
Urban
Rural
Agricultural
~
~
o
Potatoes, Italy
Various, U.S.
~
~
Various, U.S.
Corn, Italy
~
Citrus Fruits, Italy
Various, U.S.
~
Chicken, Italy
Smo ked Po ultry, F ranee
~
nzzi
Pork Sausage
Pork, Italy
Smoked Pork, France
~
~
Smoked Salmon
Trout and Cod (Unsmoked), Italy
Smoked Fish, France
~
~
1 1
"O ^
a) 25
in cd
§-&
rr CD
Various (No Leafy Greens), U.S.
Spinach/Kale, Near Airport Runway
Various, Italy
Leafy Greens, U.S.
~
~
Apples
Various, U.S.
~
Zl
Italian Samples
U.S. Samples
~
~
Beef Sausage
Grilled/Barbequed, U.S.
Cooked (NotSmoked orGrilled), U.S. (b
~
1.E-7 1.E-6 1.E-5 1.E-4 1.E-3 1.E-2 1.E-1 1.E+0 1.E+1 1.E+2
Benzo[a]pyrene Concentration (mg/kg wetwt.*)
* except for soil, which is expressed as mg/kg dry weight
Source: Hazardous Substances Databank Record for Benzo[a]pyrene (HSDB 2005c)
b Source: Analysis of 200 Food items for Benzo[a]pyrene and Estimation of its Intake in an Epidemiologic Study (Kazerouni et al. 2001)
cSource: ATSDR ToxicologicalProfile forPolycyclic Aromatic Compounds (PAHs) (ATSDR 1995)
C-68
-------
Exhibit 4-22. Contribution of Modeled Food Types to Benzo[a]pyrene
Ingestion Exposures (mg/kg/day)
Fish,
2.8E-09,
5%
Eggs, Pork, & Poultry,
1.8E-09,
3%
Soil,
4.2E-09,
8%
Beef,
4.2E-09,
8%
Fruit,
1.5E-08,
27%
Dairy,
1.6E-08,
29%
Vegetables,
1.1E-08,
20%
C-4.3.4 Summary
This analysis compared outputs from the RTR screening scenario (using de minimis emission
values) to observed data reported in the literature for cadmium, mercury, 2,3,7,8-TCDD, and
benzo[a]pyrene. In general, most results from TRIM.FaTE do not appear to be unreasonable
for a screening modeling approach based on this comparison to measured values. Briefly, the
results of the evaluation are as follows:
. Cadmium: Modeled concentrations in environmental media and ingestible products and
behavior with respect to partitioning and bioaccumulation appear to be reasonable.
. Mercury: Modeled concentrations in environmental media are comparable to levels for
contaminated sites, and modeled concentrations in ingestible products are generally
consistent with reported values. Speciation of mercury appears to be consistent with
observed patterns in the environment.
. 2,3,7,8-TCDD: Modeled concentrations in environmental media are consistent with
observed TEQ values, and modeled concentrations in ingestible products are slightly
lower than reported values, but still within a reasonable range. Both modeled values
and measured values were primarily located in the noise around the limit of detection,
which increases the uncertainty of the data.
. Benzo[a]pyrene: Modeled concentrations in both environmental media and ingestible
products are generally lower than those reported in the literature, in some cases by more
than two orders of magnitude. Because this chemical is found in the environment at
background levels that would far exceed concentrations resulting from single facility-
C-69
-------
emissions, a comparison to measured concentrations is not as informative for this
chemical.
Note that the exposure factors driving farm food chain ingestion rates are also conservative, and
that these parameters were not considered in this evaluation, meaning that the "actual"
concentration associated with a given risk level (if such a value could be calculated) would be
higher than the level associated with the modeled de minimis rate. Also, even "correct" model
results may mask incorrect model assumptions that combine to produce compensating errors.
Where results differ significantly from reported values, such as with benzo[a]pyrene, a more
detailed investigation of underlying assumptions may be necessary to determine appropriate
adjustments to the scenario configuration. As noted in the overview of this section, this type of
evaluation cannot be used to verify model results.
C-4.4 Sensitivity Analyses
The hypothetical subsistence farmer/fisher exposure scenario used to screen emissions of PB-
HAPs described in this document was parameterized using generally conservative inputs. The
goal was to construct a modeling scenario that is sufficiently health-protective (i.e.,
conservative) such that it can be used with confidence to screen out emissions that do not pose
unacceptable multipathway risks, while also avoiding overly conservative characteristics that
diminish the functionality of the scenario (i.e., by allowing too many "false positives," or facilities
that fail the screen for which the risks are actually acceptable). The level of conservatism of the
scenario is dictated largely by the values selected by the user for model inputs. A sensitivity
analysis is a useful method for evaluating the influence of model parameters and user
selections for parameter values. By providing quantitative information on the importance of
model parameters to a selected output, a sensitivity analysis thus also provides information on
which parameters may be most influential in dictating the uncertainty associated with the
results.
The sensitivity analyses conducted on the RTR screening modeling scenario encompassed the
fate and transport modeling carried out using TRIM.FaTE and the farm food chain (FFC) and
ingestion exposure modeling performed using MIRC.17 A systematic sensitivity analysis was
conducted on these parameters to obtain information regarding the relative importance of user
inputs (Section C-4.4.1). In addition, several other analyses were performed to evaluate model
performance with respect to parameterization of ingestion rates (Section C-4.4.2), body weight
(Section 0), assumed relationships between ingestion and body weight over time (Section C-
4.4.4), and meteorological conditions (Section C-4.4.5). Each section presents the methods
employed to conduct the sensitivity analyses, evaluation results, and notes regarding
interpretation of the results. Results of these sensitivity analyses are useful in informing the
level of uncertainty in screening results, highlighting parameters and aspects of the modeling
scenario worthy of additional research if refinement is appropriate (e.g., for evaluating sources
of PB-HAPs that do not pass the screen), and suggesting variables that are likely to be worthy
of more detailed examination if a more quantitative uncertainty analysis is desired (e.g., using
probabilistic methods).
17 PB-HAP dose-response values used for risk calculation (i.e., oral cancer slope factors and oral reference doses,
plus the mutagenicity correction factor applied to certain age groups) were not included in the sensitivity analysis.
C-70
-------
C-4.4.1 Systematic Sensitivity Analysis
C-4.4.1.1 Methods
As with all predictive models, the user-specified inputs for TRIM.FaTE and MIRC influence
model outputs and thus estimation of de minimis screening thresholds. The degree to which
model outputs diverge from a nominally "correct" answer (e.g., the actual incremental risks for
an individual exposed to PB-HAP emissions) encompasses both uncertainty (because values
are not known with perfect accuracy) and variability (because each value selected represents a
discrete choice out of a range of possible parameter values within a distribution, such as
variability in body weights among individuals in an exposed population). In this analysis, we did
not attempt to separately evaluate uncertainty and variability, and variation in model outputs is
assumed to encompass both uncertainty and variability. Representation of each parameter with
a single fixed value without representing uncertainty or variability in model calculations is a
limitation of deterministic model application.
As a first step to quantifying the combined uncertainty and variability in model outputs and the
de minimis thresholds, an analysis was performed to identify the variables to which the risk or
hazard quotient calculations are most sensitive. Model sensitivity was evaluated separately for
each of the PB-HAPs. The model output to which sensitivity was measured varied by chemical
depending on the health impacts of greatest concern. Parameter influence on cancer risk was
measured for benzo[a]pyrene and 2,3,7,8-TCDD, and influence on non-cancer hazard quotient
was measured for cadmium, divalent mercury, and methyl mercury. This part of the analysis
was conducted systematically by "perturbing," or changing, the value of each selected
parameter independently (i.e., one at a time, holding all other inputs at their base value) by a
certain percentage and calculating the corresponding percent change in the risk or hazard
quotient value. This metric is referred to as elasticity (i.e., ratio of the percent change of the
model output to percent change in the input variable), with higher elasticities corresponding to
greater influence.
This type of systematic sensitivity analysis has the advantage of focusing on a wide range of
inputs at once so that the variables can be ranked in order of importance. However, by
perturbing the variables by fixed percentages, the analysis does not necessarily focus on the
most physically relevant values of each parameter. An alternate analysis could examine each
variable independently (taking into account the plausible range of input values) and look at the
effects of using different plausible variable values on the risk or hazard quotient estimates. We
focus on a systematic sensitivity analysis here with the goal of prioritizing variables; additional
examinations may be helpful in the future to better define the uncertainty.
In the systematic sensitivity analysis, we estimated both local sensitivity, quantified as the
elasticity when a parameter value is perturbed by a small percentage of its base (or default)
value, and range sensitivity, for which parameters were varied by a larger percentage of the
base value. These two elasticities can be different if the relationship between input and output
is nonlinear (e.g., if the perturbed parameter is in the denominator of an equation in the
calculation). For the evaluation of risks associated with exposures to PB-HAPs, a large
difference between local and range sensitivity could indicate that variables that are less
important in the base scenario used to calculate de minimis thresholds scenario assumptions
may be more important in a less conservative site-specific evaluation. In this case, following
recommendations in the Risk Assessment Guidance for Superfund (EPA 2001; see Volume III,
Part A, Appendix A), we perturbed parameter values by +1-5% of the base case default value in
the local sensitivity analysis and +1-50% in the range sensitivity analysis. Thus, with some
C-71
-------
exceptions noted below, this resulted in four model simulations and corresponding elasticities
for each parameter included in the analysis for each of the PB-HAPs.
Typically, a range sensitivity analysis (also referred to as "nominal sensitivity analysis")
measures the effect on model outputs across the entire expected or plausible range of values
for a given input. This provides more comprehensive information on the behavior of the model
with respect to the input being varied by demonstrating the maximum potential influence of that
input on model outputs. In the case of the RTR modeling scenario, information on the expected
or most likely input ranges is available on only a few of the parameters defined by the user, and
there is a relatively large number of parameters to be evaluated. The selection of both 5 and 50
percent as perturbation increments was intended to be an efficient compromise between
conducting an authentic range sensitivity analysis and evaluating only more localized sensitivity.
To obtain a comprehensive estimate of the relative sensitivity of risk results across the range of
numerical user inputs, as many TRIM.FaTE and MIRC inputs as possible were included in the
analysis. To that end, all the MIRC ingestion and exposure variables were included. For
TRIM.FaTE, properties assumed a priori to have greater influence on model outputs were
included. The large number of user-specified numerical inputs to TRIM.FaTE and
computational limitations made inclusion of all TRIM.FaTE inputs in this sensitivity analysis
impractical. TRIM.FaTE inputs for this sensitivity evaluation were selected based on results
obtained from previous TRIM.FaTE evaluations (including the TRIM.FaTE mercury test case;
EPA 2005b) and professional judgment/intuition drawing on experience with the model.
The full set of inputs included in the systematic analysis is shown in Attachment C-3, Exhibit 1.
Parameters are grouped by the model in which they are used, with the MIRC variables further
divided into farm food chain and ingestion/body weight categories. In some cases, inputs
included in the analysis could not be increased by either 5% or 50% or both, since the variable
has a physical upper bound which cannot be exceeded (e.g., the number of days of exposure
per year was already set to 365, so it cannot be increased by 5% or 50%). The footnotes in
Attachment C-3, Exhibit 1 indicate the variables that could not be increased by 5% and/or 50%.
For some inputs, the ability to perturb the input depends on the PB-HAP being modeled, as
indicated. In addition, some inputs were assigned a baseline value of zero for estimating de
minimis threshold (e.g., the empirical correction factor for protected vegetables and the
exposure variables related to the water pathway). These variables are not included in the
systematic sensitivity analysis. In all, approximately 240 variables were examined.
Two types of inputs included in the systematic analysis, pH and the moisture adjustment factor
(MAF), received special treatment.
. pH, which is a user input for the surface soil, root zone soil, surface water, and sediment
compartments, is measured on a logarithmic scale, and this input is (in some cases) used
in the exponent of TRIM.FaTE algorithms. To obtain sensitivity metrics for pH inputs that
can be compared to results for other inputs with more linear relationships with risk, the
values for pH were changed to the equivalent hydrogen ion concentration (that is, 10pH)
and that value was varied by 5 or 50%. Then, the log base ten of the varied hydrogen ion
concentration was calculated to serve as the pH to use in TRIM.FaTE for the sensitivity
case.
. The moisture adjustment factor indicates the percent of a produce item that is water; this
factor is used to convert from wet weight to dry weight concentration in MIRC. This
calculation is made by finding the percent of the produce that is not water (i.e., 100 minus
MAF) and multiplying by the wet-weight basis concentration. Because the physical
C-72
-------
variable of interest is really the fraction of produce that is NOT water, this value (100
minus MAF) was used in the sensitivity analysis as the base value (rather than MAF
itself).
In addition, in two cases, known or assumed correlations between inputs were taken into
account in the systematic sensitivity analysis.
. The fractions of T3 and T4 fish that are consumed must add to one in MIRC; thus, when
one was changed by a given percentage, the other was also changed to ensure the sum
was still one.
. In parameterizing several sediment and water inputs used in TRIM.FaTE, soil/sediment
and runoff/surface water balance is assumed. These relationships are not explicitly
accounted for by TRIM.FaTE but rather were calculated off-line prior to setting the input
value in the model. TRIM.FaTE variables subject to soil or sediment balances include
erosion rate, precipitation rate, runoff rate, sediment deposition velocity, suspended
sediment concentration, sediment porosity, water retention time in the pond, and the
suspended sediment concentration. For example, increasing the sediment deposition
velocity results in an increase in the sediment resuspension velocity to ensure the
sediment mass balance is preserved (the specific relationships assumed between inputs
are described elsewhere). These correlations were respected in the systematic sensitivity
analysis such that when an independent input was perturbed, any dependent input
variables were altered correspondingly to preserve the water and sediment balances.
Other TRIM.FaTE and MIRC inputs are also related. Some of the relationships involving
ingestion rates and body weight were evaluated in the context of the sensitivity analysis, as
described in the sections that follow. In general, however, the current analysis did not endeavor
to determine additional correlations or account for them in the sensitivity calculations conducted.
C-4.4.1.2 Results
Exhibit 4-23 through Exhibit 4-27 display the variables with the highest elasticities for each PB-
HAP. The inputs were sorted according to the elasticities estimated for a decrease in input by
5%. This case (as opposed to the case where variables were increased by 5%) was chosen
because all inputs, including those for which the base case value was set at the maximum,
could be part of the sorting procedure. Parameters were sorted by absolute value of elasticity,
and the bars in the charts are color coded to distinguish whether the input is used in
TRIM.FaTE, MIRC farm food chain calculations, or MIRC ingestion exposure calculations. The
top 25 inputs are shown for all PB-HAPs; for some chemicals, additional inputs are included if
more than one input corresponded to the 25th highest elasticity.
In general, mixing height, emission rate, and horizontal wind speed always appear near the top
of the ranking for all PB-HAPs analyzed. The emission rate elasticity is always 1.00, indicating
that a 5% reduction in the emission rate gives a 5% reduction in the risk or hazard quotient.
The mixing height and horizontal wind speed elasticities are always negative and are either
above or below 1.00. When the mixing height is decreased, the PB-HAP emissions spread over
a smaller volume and more of the PB-HAP mass remains near the surface for deposition.
When the horizontal velocity decreases, the PB-HAP remains within the model domain for a
longer time and the concentrations in the model compartments increase accordingly.
A fourth input, sediment deposition/resuspension rate, has an elasticity greater than one in
magnitude for methyl mercury. This sensitivity case represents a decrease in the user-specified
C-73
-------
input for sediment deposition velocity by 5% and a concurrent 7% decrease in the resuspension
velocity (calculated to maintain the sediment balance18). This value is particularly important for
methyl mercury HQ because methylation of mercury occurs to a large degree in the sediment,
and a decrease in deposition velocity results in a larger, proportionate decrease in the
resuspension velocity, indicating a longer residence time in the sediment and a higher degree of
methylation.
Only the above four inputs (mixing height, emission rate, horizontal wind speed, and sediment
deposition) have absolute elasticities greater than or equal to one. Beyond these, the inputs
with the highest elasticities are those that influence the primary exposure pathways for each PB-
HAP. Primary ingestion exposure pathways (in this case defined as the food categories that
account for 75% to 85% of the total risk or hazard) are indicated at the top of each of the charts.
. For benzo[a]pyrene (Exhibit 4-23), the primary pathways are ingestion of fruits,
vegetables, and dairy products, and most of the inputs with the highest elasticities are
used in farm food chain and ingestion exposure calculations. Because wet deposition is
of particular importance, the exposed fruits and vegetables represent the more sensitive
pathway compared with the protected fruits and vegetables. In addition, the analysis
reveals that within the food chain diet of the dairy cows, risk is most sensitive to the inputs
used to estimate chemical transfer via the cow's forage pathway (as opposed to the silage
or grain pathways).
. Sensitivity results obtained for 2,3,7,8-TCDD (Exhibit 4-24) are consistent with the
observation that the primary pathways are fish, beef, and dairy ingestion. Results also
reinforce the conclusion that cancer risk is more sensitive to the forage pathway for dairy
and beef than the grain or silage pathways.
. Sensitivity results for cadmium (Exhibit 4-25) reflect the primary pathways of fruits,
vegetables, and fish, but in this case the hazard quotient is more sensitive to the protected
fruits and vegetables (and thus also the soil variables) than the exposed fruits and
vegetables. The HQ is also more sensitive to variables affecting the T3 fish than the T4
fish, consistent with higher cadmium concentrations estimated by TRIM.FaTE for that fish
type.
. Elasticities for divalent mercury (Exhibit 4-26) reflect the primary pathways of soil, fruits,
and vegetables through the importance of the rain and erosion variables and the protected
fruit and vegetable variables. Finally, results for methyl mercury (Exhibit 4-27) are
consistent with fish consumption as the primary pathway, with inputs specific to T4 fish
more important than those specific to T3 fish due to the higher bioaccumulation potential
(and therefore exposures) associated with higher trophic level fish.
Looking across the different PB-HAPs, several of the TRIM.FaTE variables with high elasticities
are highly dependent on assumptions made in configuring the TRIM.FaTE modeling scenario,
including mixing height, horizontal wind speed, rain rate, dry deposition velocity, surface water
retention time (correlated directly with pond depth and other inputs given the water balance that
is assumed), and surface water temperature. These inputs, which would likely be set differently
18 The sediment balance maintains a zero net flux of sediment mass into the sediment at all times (i.e., a steady
state) by balancing the deposition, resuspension, and burial fluxes. The burial flux is calculated by subtracting the
amount of sediment flushed from the pond from the amount introduced into the pond via erosion at every time step.
The deposition rate is specified. Then, the resuspension rate is calculated by adding the burial and deposition fluxes
to ensure no net flux into or out of the sediment.
C-74
-------
for a site-specific analysis, represent influential parameters worthy of additional evaluation to
quantify the conservative nature of the scenario. The area of the parcels and the temporal
pattern of the precipitation may also be influential; however, these were not evaluated
quantitatively in the systematic sensitivity analysis given the complexity involved with adjusting
these inputs.
Differences between the results of the local and range sensitivity analyses are presented in
Attachment C-3, Exhibits 2 through 6. These tables show elasticities and elasticity rankings by
PB-HAP for each input that appears in the top 25 elasticities for all four analyses (i.e.,
perturbation by -50%, -5%, 5%, and 50% of the base case value). The rankings for some inputs
are somewhat different across the four cases, but no drastic differences are noted when
comparing the range and local sensitivity analyses for any of the PB-HAPs. This suggests the
local sensitivity analysis may be sufficient for drawing some conclusions about the relative
influence of user-specified inputs on the risk and hazard quotient estimates.
The results of this sensitivity analysis indicate the 25-30 variables to which the risk and hazard
quotient estimates are most sensitive. In proceeding with a probabilistic analysis which would
quantify the uncertainty in the model estimates, these variables would be of primary importance
and should be the focus when developing input variable distributions. Further research could
also confirm that these variables are set in an appropriately conservative fashion for the
purposes of developing a screening scenario.
C-75
-------
Exhibit 4-23. The 26 Variables with the Highest Elasticities for Benzo[a]Pyrene Lifetime Risk (-5% Perturbation of Variable)
Primary Pathways: Fruits, Vegetables, and Dairy
Mxng Ht, -1.05
oriz Wind Spd, -0.9
-2
I Farm Food Chain
I TRIM
Ingestion/Body Weight
-1.5
Yield (Frge)
nterc
Biotra
Qty Frg
Pint Srf Loss (Frg
Yield (Exp
e), -0.25
Frt), -0.23
Pint Srf Loss (Ex
Yield (Exfb
p Frt), -0.19
Veg), -0.17
Emission Rate, 1.00
Annual Rain,
^vg Wet Dep Particle),
rac Wet Dep, 0.45
mmal Metab Fctr, 0.<
ept Frac (Frge), 0.34
ns Fctr (Dairy), 0.34
ctr (Dairy), 0.34
Oontam (Dairy), 0.34
Dep Particle, 0.30
e Eaten (Dairy), 0.29
e Contm (Dairy), 0
29
100-Mst
Exp Fctr
Frac Cont^i
Intercept
Fctr (Exp Frt), 0.
xp Frt), 0.22
m (Exp Frt), 0.22
ac (Exp Frt), 0.21
22
100-Mst Adj
Exp Fctr (Ex
Frac Contarr
Fctr (Exp Veg), 0.17
d Veg), 0.17
(Exp Veg), 0.17
1
1.5
C-76
-------
Exhibit 4-24. The 25 Variables with the Highest Elasticities for 2,3,7,8-TCDD Lifetime Risk (-5% Perturbation of Variable)
Primary Pathways: Fish, Beef, and Dairy
Mxng Ht, -1.05
Horiz Wind Spd, -0.9
I Farm Food Chain
I TRIM
Ingestion/Body Weight
Frac Contam (Fi
Exp Fctr (Fish), C
sh), 0.56
Wa
Body Weight, -0.4
ter Org Carb Frac, -
D.38
Frac Fish T:
-2
-1.5
Yield (Frge), -0.30
Suspended Sed Cone, -0.25
Pint Srf Loss (Frge), -0.21
-0.5
3, -0.30
Emission Rate, 1.00
Biotran
Frac C
Frac Fi
Chem Cone T4 Fish, C
g Rate (Fish), 0.43
ammal Metab Fctr, C
:er Temp, 0.39
Dep and Resusp, 0.
is Fctr (Dairy), 0.32
tr (Dairy), 0.32
ontam (Dairy), 0.32
36
5h T4, 0.30
Intercept Frac (Frge), 0.28
Avg Dry Dep Particle, 0.28
Qty Frge
Frac Frge
Eiaten (Dairy), 0.23
Contm (Dairy), 0.23
Ing Rate (D
Annual Rain.
airy), 0.20
0.20
.43
1.5
C-77
-------
Exhibit 4-25. The 28 Variables with the Highest Elasticities for Cadmium Hazard Quotient for Child 1-2 (-5% Perturbation of
Variable)
Primary Pathways: Fruits, Vegetables, and Fish
Mxng Ht, -1.03
z Wind Spd, -1.03
I Farm Food Chain
I TRIM
Ingestion/Body Weight
-1.5
Body Weig
Erosion Ra
Loss 1 (F'ro Frt), -0
Emission Rate, 1.00
Annual Rkin, 0.75
Soil Cone RoctZn Produce, 0.63
:one Fetr (Pro Frt), 0.39
Rate (Pro Frt), 0.39
Fctr (Pro Fruit), 0.3
3 Contam (Pro Frt), 0.39
-Mst Adj Fctr (Pro F"t), 0.39
Water Ti
Ing Rate (Fi
Exp Fctr (Fi
Frac Contar
Surf Soil Fra
Frac Fish T4, -0.11
-0.5
Chem Cone
Water Retent
I Sed Dep and
100-Mst Adj Fc
Ing Rate (Exp
Exp Fctr (Exp
Frac Contam
| Frac Fish T3, 0
Surf Soil Org C
smp, 0.27
sh), 0.19
sh), 0.19
(Fish), 0.19
c Air, 0.17
13
Fish, 0.15
on Time, 0.15
Resusp, 0.13
tr (Exp Veg), 0.11
Veg), 0.11
Veg), 0.11
Exp Veg), 0.11
11
arb Frac, 0.10
0.5
1.5
C-78
-------
Exhibit 4-26. The 25 Variables with the Highest Elasticities for Divalent Mercury Hazard Quotient for Child 1-2 (-5%
Perturbation of Variable)
Primary Pathways: Soil, Fruits and Vegetables
Mxng Ht, -0.91
Horiz Wind Spd, -0.91
Erosion Rc
Bod
-2
I Farm Food Chain
I TRIM
Ingestion/Body Weight
-1.5
Annual Rain, 0.74
te, -0.77
I Weight, -0.61
Soil C
Bioconc Fctr
Ing Rate (Pre
Emission Rate, 1.00
Soil Cone Human
Exp Fctr (Soil), 0
Frac Contam (Sc
Ing Rate (Soil), 0
Dnc RootZn Product
(Pro Frt), 0.15
Frt), 0.15
Ing, 0.56
100-Mst Adj Fctr (Pro Frt), 0.15
Exp Fctr (Pro
Frac Contam
Surf Soil Frac
Fruit), 0.15
(Pro Frt), 0.15
Air, 0.13
-0.5
Bioconc Fctr (Root Veg), 0.11
Ing Rate (Root Veg), 0.11
100-Mst Adj Fctr (Root Veg),
0.11
Exp Fctr (Root Veg), 0.11
Frac Contam (Root Veg), 0.11
Correct Fctr (RDOt Veg), 0.11
Soil Cone Animal Ing, 0.08
Soil Bioav Fctr (Livestock), 0.08
0.5
1.5
C-79
-------
Exhibit 4-27. The 26 Variables with the Highest Elasticities for Methyl Mercury Hazard Quotient for Child 1-2 (-5%
Perturbation of Variable)
Primary Pathway: Fish
Body Weight, -1.00
Mxng
Horiz Wind
Ht, -0.70
Spd, -0.70
zrosion Rate, -0.4
Sed Porosity, -0
I Farm Food Chain
ITRIM
Ingestion/Body Weight
amp, -0.22
sh T3, -0.19
Suspended Sed Cone, -0.18
Emission Rate, 1.
Ing Rate (Fish), 0.90
Frac Contam (Fish),
D Fctr (Fish), 0.90
Annual Rain
Chem Cone T4
0.64
Fish, 0.55
Chem Cone T3 Fish, 0
Frac Fish T4, 0.19
Surf Soil Fra
Water Retent
Fish Mass All!
Total Runoff, 0.(
Soil Cone Human
Ing Rate (Soil),
Exp Fctr (Soil),
Frac Contam (S
I Soil Cone RootZn
00
0.90
36
Sed Dep and
Resusp, 1.56
-1.5
-0.5
1.5
C-80
-------
C-4.4.2 Evaluation of Ingestion Rate Assumptions
To calculate the de minimis levels, the conservative assumption was made that a person
ingests food at the 90th percentile level for each food type included in the analysis. These
ingestion rates can then be added to get the assumed total ingestion rate.19 USEPA 2008 and
2005d provide estimates of the 90th percentile total ingestion rate for different age groups (that
is, percentiles for total amount of food ingested, rather than a sum of individual percentiles)
based on CSFII data 1994-96 and 1998 (see Exhibit 6-12 of Attachment C-2). When comparing
the two, the de minimis total ingestion rates tend to be a factor of two to three times higher than
the 90th percentile total ingestion rate (see Exhibit 7-1 of Attachment C-2). An alternative
approach that still captures the high-end would be to use the 90th percentile ingestion rates for
food types associated with highest risks for a given PB-HAP (to account for the individual who
may happen to eat higher-than-typical amounts for each of those food categories) while setting
ingestion rates for other food types to a lower percentile in order to bring the total ingestion rate
to a more realistic level (e.g., to something closer to the 90th percentile total ingestion rate).
In order to determine the effect of lowering the ingestion rates for the non-primary ingestion
pathways (and provide information on whether this refinement would be worth the additional
complexity in the model set-up), ingestion rates were kept at the 90th percentile for the primary
(risk-driving) food types for each PB-HAP but were reduced to median ingestion rates for the
same population (subsistence famers) for the other types of food. This scenario, referred to
here as the "alternate ingestion case," was evaluated for each PB-HAP. Specifically, the total
food ingestion rates was calculated for the alternate ingestion cases for each PB-HAP, and
these values were compared with the USEPA 2008 and 2005d estimates of 90th percentile total
ingestion (referred to as the "EPA total ingestion rate") via a ratio (Exhibit 4-28). Because the
EPA total ingestion rate represents the total amount of cooked food consumed, preparation
losses were applied to each of the food category-specific ingestion rates input into MIRC before
the ratio was calculated. Soil consumption was not included in these calculations. In addition,
the estimated de minimis cancer risk or HQ for each PB-HAP was compared to the recalculated
risk/HQ for the alternate ingestion cases (Exhibit 4-29).
As expected, the ratios presented in Exhibit 4-28 demonstrate that using the alternate ingestion
rate combination tends to result in lower estimated total ingestion rates, approaching the (more
realistic) USEPA total ingestion rates for the child age groups. The degree to which the total
ingestion rate is decreased depends on the number of pathways labeled "primary pathways" for
risk (as well as the mass of food represented by a given pathway). Thus, it is reasonable that
lower total ingestion rates (and lower ratios with respect to EPA total ingestion rate) are
calculated for methyl mercury, for which only one pathway (fish) drives risk, and higher rates
and ratios are calculated for benzo[a]pyrene, for which several food types are important.
Exhibit 4-29 compares the lifetime risk or child age 1-2 HQ for the alternate ingestion rate case
to the corresponding risk/HQ for the de minimis case. Overall, the changes in risk and HQ were
relatively modest, ranging between 7% and 20%. This result indicates that relaxing the
conservative assumption of high ingestion rates for all the pathways does not significantly
change the risk and hazard calculations but does drive the total ingestion rates closer to the
USEPA 90th percentile total ingestion rates. However, implementing this multi-percentile
technique requires prior knowledge of the dominant pathways for each chemical, which is
determined primarily from the model estimates themselves. Assuming 90th percentile ingestion
19 This calculated total ingestion rate will not include any other food consumed by the homestead family outside of the
MIRC food categories. In making the comparison with the total ingestion rates, then, the assumption is made that the
entire diet of the family is captured by the MIRC food categories.
C-81
-------
rates for all food categories has the advantage of simplicity (using the same set of ingestion
rates for all chemicals) and does not dramatically impact the risk and hazard quotient estimates.
Exhibit 4-28. Ratio of the Mod
Tota
eled Total Ingestion Rates and the USEPA
Ingestion Rates
Ratio of Ingestion Rates
Child
(1-2)
Child
(3-5)
Child
(6-11)
Child
(12-19)
Adult
(20-70)
Benzo[a]Pyrene Alternate Ingestion
1.7
1.3
1.4
1.4
2.6
2,3,7,8 - TCDD Alternate Ingestion
1.3
1.0
1.0
1.0
1.8
Cadmium Alternate Ingestion
1.3
1.1
1.2
1.0
1.6
Divalent Mercury Alternate Ingestion
1.3
1.1
1.1
1.0
1.6
Methyl Mercury Alternate Ingestion
0.9
0.8
0.8
0.6
0.8
de minimis Ingestion
1.8
1.4
1.6
1.5
2.8
C-82
-------
Exhibit 4-29. Comparison of the Risks and Hazard Quotients in the de minimis and
All
ternate Ingestion Cases
PB-HAP
Categories with
90th Percentile
Ingestion in
Alternate Case a
Categories with
Median Ingestion
Rates in Alternate
Case a
Case
Value
Benzo[a]pyrene
Pro. fruit, pro.
veg., exp. fruit,
exp. veg., root
veg., and dairy
Soil, fish, beef, pork
poultry, and eggs
de minimis lifetime risk
1.00E-06
Alternate ingestion case
7.98E-07
Percent Change in Risk
-20.2%
2,3,7,8 - TCDD
Fish and dairy
Soil, pro. fruit, pro.
veg., exp. fruit, exp.
veg., root veg.,
beef, pork, poultry,
and eggs
de minimis lifetime risk
1.00E-06
Alternate ingestion case
9.14E-07
Percent Change in Risk
-8.6%
Cadmium
Fish, pro. fruit,
pro. veg., exp.
fruit, exp. veg.,
and rootveg.
Soil, beef, dairy,
pork, poultry, and
eggs
de minimis child (1-2)
hazard quotient
1.00E+00
Alternate ingestion case
9.05E-01
Percent Change in Risk
-9.5%
Divalent
Mercury
Soil, pro. fruit,
pro. veg., exp.
fruit, exp. veg.,
and rootveg.
Fish, beef, dairy,
pork, poultry, and
eggs
de minimis child (1-2)
hazard quotient
1.00E+00
Alternate ingestion case
9.26E-01
Percent Change in Risk
-7.4%
Methyl Mercury
Fish
Soil, pro. fruit, pro.
veg., exp. fruit, exp.
veg., root veg.,
beef, dairy, pork,
poultry, and eggs
de minimis child (1-2)
hazard quotient
5.62E-01
Alternate ingestion case
5.19E-01
Percent Change in Risk
-7.8%
a Pro. fruit is protected fruits, pro. veg. is protected vegetables, exp. fruit is exposed fruits, exp. veg. is
exposed vegetables, and root veg. is root vegetables.
C-4.4.3 Evaluation of Body Weight Assumptions
As stated in Attachment C-2, Section 6.1.3, the de minimis rates were calculated using mean
body weight following EPA's default screening recommendation. This assumption does not
represent the most conservative assumption, however, since lower percentile body weights will
give higher risk and hazard quotient estimates. In contrast, the ingestion rates were set to the
more conservative 90th percentile level. To investigate the sensitivity of using alternate body
weight percentiles, an alternate MIRC run was performed using the 5th, 50th, 90th, and 95th
percentiles for body weight while keeping all other variable values as specified in the de minimis
calculations. These alternate risk or hazard quotient model estimates and the percent change
relative to the de minimis calculations are shown in Exhibit 4-30. Changing from the mean to
the median body weights produces only very modest changes to the risks/hazard quotients.
Changing to the 5th percentile body weights gives the most conservative model estimates, with
percent changes of up to 22% for the PB-HAPs used to set the de minimis rates and a 26%
change in methyl mercury. Using upper percentiles for body weights produces modest
decreases in risk or hazard quotient of 3% to 16% for the PB-HAPs used to set the de minimis
rates. Thus, changing the body weight to either a more conservative or less conservative
percentile produces changes in the risk or hazard quotient which are appreciable but not
C-83
-------
dramatic (i.e., typically less than 25%), suggesting the mean body weight assumption may be
an appropriate choice for the screening scenario.
Exhibit 4-30. The Risk or Hazard Quotient Estimates Using Alternate Body Weight
Percentiles
de
minimis
5th Percentile
Median
90th Percentile
95th Percentile
Risk or
HQ a
%
Ch b
Risk or
HQ a
%
Ch b
Risk or
HQ a
%
Ch b
Risk or
HQ a
%
Ch b
Benzo[a]pyrene c
1.0E-06
1.1E-06
6%
1.0E-06
0%
9.7E-07
-3%
9.6E-07
-4%
2,3,7,8-TCDD c
1.0E-06
1.2E-06
22%
1.0E-06
2%
8.8E-07
-12%
8.4E-07
-16%
Cadmium d
1.00
1.07
7%
1.00
0%
0.96
-4%
0.95
-5%
Divalent Mercury d
1.00
1.16
16%
1.00
0%
0.91
-9%
0.89
-11%
Methyl Mercury d
0.56
0.71
26%
0.57
1%
0.48
-15%
0.46
-18%
a HQ is the hazard quotient for a child aged 1 -2.
b % Ch Is the percent change in the risk or hazard quotient relative to the de minimis calculations.
c Percent change in lifetime risk.
d Percent change in the hazard quotient for a child age 1 to 2.
C-4.4.4 Sensitivity When Accounting for Temporally Correlated Body Weight and
Ingestion Rates
In calculating lifetime cancer risks, age-specific ingestion rates and body weights are used to
estimate the lifetime average daily dose. In the systematic sensitivity analysis, the body weights
and ingestion rates for each of the five age categories were changed independently to assess
the sensitivity of lifetime risk to each input separately (for example, body weight was first
perturbed only for a child 1-2 yrs old, leaving body weight unchanged for other age groups).
However, in actuality, ingestion rates (and body weights) for an individual during different age
groups will likely be correlated across that individual's lifetime. For example, a person who eats
higher-than-average amounts of poultry when they are 11 can be reasonably expected to eat
higher-than-average amounts when they are 50 due to lifetime dietary preferences. To estimate
the effect this correlation has on sensitivity of cancer risk to ingestion rate, a perturbation of -5%
for a given food type was applied to ingestion rate for all age categories and the resulting
elasticity was compared to the previous result (obtained ignoring temporal/age-group
correlations). This assumption was applied separately to each ingestion rate category, and
separately to body weight, to estimate the effect of age group correlations for these inputs
(Exhibit 4-31). Because only benzo[a]pyrene and 2,3,7,8-TCDD were evaluated for cancer
risks, this analysis was only performed for these two PB-HAPs (the other PB-HAPs use HQs
calculated separately for each age group, and the correlation analysis is not applicable).
C-84
-------
Exhibit 4-31. Comparison in the Elasticities In Lifetime Risk in the Correlated
and Uncorrelated Analyses Assuming a 5% Decrease in the Input Variables
Elasticity in the
Correlated
Analysis
Maximum
Elasticity in
Uncorrelated
Analysis
Age Group of
Maximum
Elasticity in
Uncorrelated
Analysis
Dairy Ingestion
0.34
0.11
Child 1-2
Exp. Fruit Ingestion
0.22
0.10
Adult
Body Weight
-0.17
-0.06
Child 1-2
Exp. Veg. Ingestion
0.17
0.09
Adult
Soil Ingestion
0.12
0.05
Child 1-2
Beef Ingestion
0.07
0.03
Adult
Benzo[a]pyrene
Fish Ingestion
0.04
0.02
Adult
Poultry Ingestion
0.02
0.01
Adult
Pro. Fruit Ingestion
0.01
< 0.01
Adult
Egg Ingestion
0.01
< 0.01
Adult
Pork Ingestion
< 0.01
< 0.01
Adult
Prot. Veg. Ingestion
< 0.01
< 0.01
Adult
RootVeg. Ingestion
< 0.01
< 0.01
Adult
Body Weight
-0.59
-0.45
Adult
Fish Ingestion
0.56
0.43
Adult
Dairy Ingestion
0.32
0.20
Child 1-2
Beef Ingestion
0.09
0.06
Adult
Pork Ingestion
0.01
0.01
Adult
Soil Ingestion
< 0.01
< 0.01
Adult
2,3,7,8 - TCDD
Exp. Fruit Ingestion
< 0.01
< 0.01
Adult
Exp. Veg. Ingestion
< 0.01
< 0.01
Adult
RootVeg. Ingestion
< 0.01
< 0.01
Adult
Poultry Ingestion
< 0.01
< 0.01
Adult
Egg Ingestion
< 0.01
< 0.01
Adult
Pro. Fruit Ingestion
< 0.01
< 0.01
Adult
Prot. Veg. Ingestion
< 0.01
< 0.01
Adult
The systematic sensitivity analysis suggested that cancer risks associated with benzo[a]pyrene
are relatively insensitive to body weight and any individual food ingestion rate (where both
inputs were varied for a single age group at a time). When body weight or ingestion rate for a
given food type is decreased by 5% for all age groups in concert, the corresponding elasticity
increases by a factor of two to three. This increase is significant enough to cause the dairy
ingestion, exposed fruit ingestion, body weight, and exposed vegetable ingestion variables to be
ranked among the most influential variables on the basis of absolute elasticity.
For 2,3,7,8-TCDD, the systematic sensitivity analysis illustrated that cancer risk is sensitive to
body weight, fish ingestion rate, and dairy ingestion rate. Elasticities for these inputs increase
by a factor of 1.3 to 1.6 when the inputs are kept constant across age groups. This difference
results in rise in the systematic sensitivity ranking of these variables.
C-85
-------
Overall, however, these differences represent a modest percent change in the lifetime risk
values and indicate that accounting for the assumed correlation between the body weights and
ingestion rates throughout an individual's lifetime does not significantly influence the model's
predicted risks. For the other ingestion rates, the lifetime risks are not very sensitive to the
parameters when the inputs are decreased by 5%, and accounting for the correlations between
the variables does not affect this conclusion.
C-4.4.5 Comparison of Scenarios Using Site-Specific and de minimis Meteorological
Data
For RTR, the de minimis thresholds were calculated using a conservative and hypothetical
exposure scenario for a farm homestead. If the modeling scenario has been configured as
intended, the substitution of site-specific data will nearly always result in a lower (or equal)
exposure and risk estimate. To inform the degree of conservatism associated with the baseline
meteorological inputs, three additional analyses were conducted in which observed (site-
specific) meteorology data were used in place of the meteorological values set for the de
minimis scenario. The three locations were selected to take advantage of readily-available
TRIM.FaTE meteorological data. Exhibit 4-32 compares some of the summary statistics for the
three meteorological data sets with those used for the de minimis screening scenario.20
Exhibit 4-32. Summary of Site-specific Meteorological Data Parameters
Parameter (units a)
de minimis
Scenario b
Site 1 c
Site 2 c
Site 3 c
Average air temperature (K)
298
291.9
284.7
283.5
Average horizontal wind speed
(m/s)
2.8
4.4
4
3
Annual precipitation (m/yr)
1.5
1.1
1.1
1.1
Average urban mixing height (m)
710
1,087
1,225
861
a K = Kelvin, m/s = meters/second, m/yr = meters per year, m = meters
b Values listed for the baseline scenario indicate the fixed value used in the baseline screening scenario.
c Site-specific values are arithmetic averages of single or multiyear data sets; Site 1 = 1989-1993; Site 2 = 1994;
Site 3 = 1990-1995.
The results of using these three site-specific meteorological data sets on the risk or hazard
quotient for each PB-HAP are summarized in Exhibit 4-33. The results indicate that the wind
direction - and specifically how often the wind blows from the source toward the hypothetical
watershed, or toward due east - is an important influence on the estimated media
concentrations and ingestion exposures. For the three data sets used in this analysis, the
largest decrease in risk or hazard quotient was observed for Site 1. This finding is consistent
with the underlying patterns in wind direction and their relationship to the locations where
exposure is assumed to occur. Exhibit 4-34 shows that wind direction towards due east occurs
less than two percent of the time for this location, reinforcing the conclusion that a limited wind
flow directly from the source to the watershed will decrease contamination. These results
suggest that using meteorological data more representative of a specific site will decrease the
estimated risk or hazard quotient by as much as an order of magnitude for the hypothetical
receptor represented.
20 Exhibit 4-33 does not include a measure of the frequency and average of wind directions, which can be illustrated
with a wind rose but cannot be effectively characterized with a single value.
C-86
-------
Exhibit 4-33. Percent Change in Risk or Hazard Quotient Using Site-specific
Meteorological Data
de
minimis
Site 1
Site 2
Site 3
Risk or
HQa
% Ch b
Risk or
HQa
% Ch b
Risk or
HQa
% Ch b
Benzo[a]pyrene c
1.00E-06
1.05E-07
-89%
2.43E-07
-76%
2.19E-07
-78%
2,3,7,8-TCDD c
1.00E-06
8.39E-08
-92%
3.76E-07
-62%
2.56E-07
-74%
Cadmium d
1.00
0.07
-93%
0.14
-86%
0.15
-85%
Divalent Mercury d
1.00
0.08
-92%
0.19
-81%
0.18
-82%
Methyl Mercury d
0.56
0.02
-97%
0.06
-90%
0.05
-91%
a HQ is the hazard quotient for a child aged 1 -2.
b % Ch Is the percent change in the risk or hazard quotient relative to the de minimis calculations.
c Percent change in lifetime risk.
d Percent change in the hazard quotient for a child age 1 to 2.
Exhibit 4-34. The Wind Speed and the Direction Toward Which the Wind is Blowing for
All Conditions for Site 1
20%
18%
16%
14%
12%
>
o
c
d)
= 10%
8% +-
6%
4%
2% +-
0%
~ 17-21
11-16
~ 7-10
~ 4-6
0 - 22.5 22.5-
45
45 -
67.5
67.5 -
90
90- 112.5-
112.5 135
135 - 157.5 -
157.5 180
180 - 202.5 - 225 - 247.5 - 270 - 292.5 - 315 - 337.5 -
202.5 225 247.5 270 292.5 315 337.5 360
Direction (degrees)
C-87
-------
C-4.5 Comparison to Other Model Results
C-4.5.1 Comparison to Preliminary RTR Screening Runs (HHRAP Approach)
As another component of this evaluation, de minimis emission rates calculated using the
screening scenario were compared to those calculated for the preliminary RTR screening
analyses carried out by EPA in the fall of 2006 (EPA 2006a) that use fate and transport
algorithms presented in the HHRAP documentation (EPA 2005a). Similar to the current TRIM-
based scenario, the modeling scenario evaluated in the preliminary RTR analyses was based
on a hypothetical working homestead. The homestead was adjacent to the modeled source,
with a 16-acre lake and 100 acres of tillable farm and pasture land. The lake was placed at the
location of highest total deposition and the total watershed area was just over 200 acres.
Exhibit 4-35 summarizes a comparison of the emission thresholds for each approach.
Thresholds from the current analysis are presented to two significant figures for the purposes of
comparison only. The emission threshold for benzo[a]pyrene is lower for the current scenario
than for the previous evaluation (i.e., the current scenario is more conservative), while those of
cadmium and divalent mercury are higher for the current runs. Because elemental mercury
does not readily deposit and is largely blown out of the modeling domain into the air sinks in
TRIM.FaTE, the threshold for divalent mercury is more relevant to this analysis. No threshold
was available for dioxins for the preliminary HHRAP analysis. Additional refinements, analyses
and comparisons described in this section support the current de minimis emission thresholds.
Exhibit 4-35. Emission Thresholds Derived in Preliminary HHRAP Screening Runs and in
Current Analyses
Basis of
Threshold
Emission Thresholds (TPY)
Chemical
Current
Analysis
Preliminary
RTR Analysis
Comparison
2,3,7,8-TCDD
Cancer
3.18E-08
NA
NA
Benzo[a]pyrene
Cancer
2.26E-03
2.2E-02
Preliminary threshold higher by ~10x
Cadmium
Non-cancer
6.54E-01
1.7E-01
Current threshold higher by ~4x
Divalent Mercury
Non-cancer
1.64E-01
5.4E-03
Current threshold higher by ~30x
NA = not available; threshold was not calculated for dioxins in 2006.
C-4.5.2 Comparison of Results for Screening Scenario and Previous TRIM.FaTE
Applications
To obtain another estimate of the degree of conservatism associated with results from the
screening scenario, a comparison run was performed using the TRIM.FaTE scenario developed
for a secondary lead smelting facility previously evaluated for a TRIM to I EM21 model
comparison in the state of New York (ICF 2004). Results for the current comparison were
obtained for benzo[a]pyrene; 2,3,7,8-tetrachlorodibenzo-p-dioxin; and elemental, divalent, and
methyl mercury by running the screening scenario with the emission rates for these chemicals
that were used in the previous secondary lead smelting application. The annually averaged
results for the 30th year of the screening scenario were compared to annual average
concentrations using results from years 28 through 32 from the secondary lead smelting
21 IEM is the Indirect Exposure Methodology that is now referred to as the Multiple Pathways of Exposure
Methodology (U.S. EPA 1999d)
C-88
-------
application.22 No adjustments to the screening or the New York (refined) scenario were
implemented; thus, it should be recognized that the results of this comparison are particular to
the inputs selected for the New York site TRIM.FaTE application.
Specifications of the New York site application are presented in the model comparison report
and are not included here. In general, no comparisons of the values used for the screening and
New York site TRIM.FaTE scenarios were conducted. However, because meteorological
properties are highly influential on concentrations in all compartments, the characteristics of the
New York meteorological inputs are summarized here. Five years of meteorological data
collected at a station in Allentown, Pennsylvania were used for the New York application;
average values for key properties are summarized in Exhibit 4-36.
Exhibit 4-36. Meteorological Data Parameters for TRIM.FaTE Secondary Lead
Smelting Application
Meteorological Property
Average Value Used for New York Site Application
Air temperature
284.69 K
Horizontal wind speed
3.97 m/s
Precipitation rate
1.14 m/yr
Urban mixing height
1,224 m
Wind direction (overall)
Blows predominantly from the southeast
Wind direction (during rain events)
Blows predominantly from the southwest
Media concentrations for air, surface soil, lake surface water, lake benthic sediment, and water
column fish from the two applications were compared collectively (i.e., using the mean of all
compartments of a single type) and, for soil and air, according to approximate distance from the
source. For the latter comparisons, air and soil parcels were grouped into "nearby" and "distant"
subgroups. Nearby parcels were those situated within about 2 km of the source, and the
remaining parcels are distant. Parcel groupings for air and surface soil output comparisons are
presented in the figures that follow. Results for a lake near the source in the New York
application were used for comparison to results for water, sediment, and fish in the screening
scenario.
22 Averaging results over 5 years of data minimizes bias introduced by any 1 year of meteorological conditions (5
years of repeating data were used for the New York site application).
C-89
-------
Silver Lake,
Exhibit 4-37. Surface Soil Parcel Spatial Layouts for New York Site Lead Smelting
TRIM.FaTE Application and Screening Scenario
©Q
i i Parcels
Water
Watersheds
'Nearby' parcel
'Distant' parcel
'Nearby' parcel
'Distant' parcel
10 km
C-90
-------
Exhibit 4-38. Air Parcel Spatial Layouts for New York Site Lead Smelting TRIM.FaTE
Application and Screening Scenario
NNW2 f
NNE1
NNW1
ENE2
WNW1
ENE1
^2
ESE2
WSW1
ESE1
WSW2
SSW1
SSE1
SSW2
SSE2
~ Parcels
o
2
'Nearby' parcel
I'Distant' parcel
N5
N4
N3
Source
3.5 km
S4
S5
S2
S3
10 km
'Nearby' parcel
'Distant' parcel
C-91
-------
The screening and refined results are compared in Exhibit 4-39. This table shows the ratios of
the concentrations for the screening scenario to the concentrations from the New York site
application; thus, ratios greater than 1 indicate that screening scenario concentrations are
higher than the lead smelting application concentrations. Exhibit 4-40 graphically illustrates
selected concentration ratios.
Exhibit 4-39. Comparison of Concentration Outputs: NY Site Refined TRIM.FaTE
Application vs. Screening Scenario
Outputa
Concentration Ratio of
Screening Scenario to Full Lead Smelting Run
Benzo[a]
Pyrene
2,3,7,8-
TCDD
Divalent
Mercury
Elemental
Mercury b
Methyl
Mercury
Overall Air Mean
15.1
12.8
13.0
12.6
16.4
Nearby Air Mean
15.8
13.8
13.8
13.7
17.0
Distant Air Mean
6.6
6.2
5.7
6.1
8.2
Overall Surface Soil Mean
19.4
3.9
10.4
-
10.3
Nearby Surface Soil Mean
18.9
3.6
9.1
-
9.0
Distant Surface Soil Mean
8.6
2.7
7.6
-
7.6
Lake Surface Water
17.0
6.0
12.1
28.7
5.7
Lake Sediment
52.2
10.3
6.6
12.2
7.0
Lake WC Carnivore
19.4
12.6
2.4
-
1.9
Lake WC Herbivore
10.5
0.8
2.1
-
1.4
Lake WC Omnivore
18.0
2.1
2.3
-
1.6
a "Overall Air" and "Overall Surface Soil" include all air and surface soil parcels, except for the source parcel.
"Nearby" parcels include those within about 2 km of the source; "Distant" parcels include those greater than about 2
- 3 km from the source (i.e., all non-source parcels not classified as Nearby). Refer to Exhibit 7-14 and Exhibit
7-15 for details. The source parcel was excluded from all comparisons. "Lake" is the lake for the New York site
application, and the pond for the screening scenario.
b Because elemental mercury is largely blown out of the modeling domain into the air sinks in TRIM.FaTE,
concentrations of elemental mercury are not examined here in soil or in water column fish.
In all media categories, the screening scenario produces greater concentrations than does the
lead smelting application. This result is expected because the screening scenario is designed
to be health-protective and therefore tends more towards a high-end scenario with regard to
media concentrations. The differences are generally larger for air results and smaller for
concentrations in fish.
On the whole, the greatest difference between the concentrations in the screening scenario and
those in the lead smelting application occurs with benzo[a]pyrene, especially with regard to
concentrations in fish and lake sediment. The largest difference observed across all
comparisons is the benzo[a]pyrene in lake sediment. Outputs were also substantially different
for elemental mercury in the surface water and benzo[a]pyrene in water column fish.
C-92
-------
Exhibit 4-40. Comparison of Concentration Outputs Grouped By Chemical: New York
Site Refined TRIM.FaTE Application vs. Screening Scenario
60
50
.Q 40
as
DH
B 30
¦*—I
as
-------
C-5 References
Agency for Toxic Substances and Disease Registry (ATSDR). 1995. Toxicological profile for
polycyclic aromatic hydrocarbons. Atlanta, GA: U.S. Department of Health and Human
Services, Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 1998. Toxicological profile for
chlorinated dibenzo-p-dioxins. Atlanta, GA: U.S. Department of Health and Human Services,
Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 1999b. Toxicological profile for
Mercury. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 2007. Toxicological profile for
Lead. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 2008. Toxicological profile for
Cadmium. Atlanta, GA: U.S. Department of Health and Human Services, Public Health
Service.
Alpers, C.N., A.R. Stewart, M.K. Saiki, M.C. Marvin-DiPasquale, B.R. Topping, K.M. Rider, S.K.
Gallanthine, C.A. Kester, R.O. Rye, R.C. Antweiler, and J.F De Wild. 2008. Environmental
factors affecting mercury in Camp Far West Reservoir, California, 2001-03. U.S. Geological
Survey Scientific Investigations Report 2006-5008, 358p.
Calabrese, E.J., E.J. Stanek, P. Pekow, R.M. Barnes. 1997. Soil ingestion estimates for
children residing on a Superfund site. Ecotoxicology and Environmental Safety. 36:258-268.
California Air Resources Board (CARB). 2007. CADAMP Monitoring Sites.
http://www.arb.ca.gov/pub/dioxin/cadamp.php
California Environmental Protection Agency (CalEPA) Office of Environmental Health Hazard
Assessment (OEHHA). 2000. Air Toxics Hot Spots Program Risk Assessment Guidelines; Part
IV, Exposure Assessment and Stochastic Analysis Technical Support Document. Section 6,
Dermal Exposure Assessment. September. Available at:
http://www.oehha.ca.gov/air/hot_spots/pdf/chap6.pdf.
Chen, C.Y., R.S. Stemberger, B. Klaue, J.D. Blum, P.C. Pickhardt, and C.L. Folt. 2000.
Accumulation of heavy metals in food web components across a gradient of lakes. Limnology
and Oceanography 45(7): 1525-1536.
Croteau, M., S.N. Luoma, and A.R. Stewart. 2005. Trophic transfer of metals along
freshwater food webs: Evidence of cadmium biomagnification in nature. Limnology and
Oceanography 50(5):1511-1519.
Hazardous Substances Data Bank (HSDB). 2005a. Bethesda, MD: National Library of Medicine,
U.S. [Last Revision Date 06/23/2005], Cadmium Compounds; Hazardous Substances Databank
Number: 6922. Available at: http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen7HSDB
Hazardous Substances Data Bank (HSDB). 2005b. Bethesda, MD: National Library of Medicine,
U.S. [Last Revision Date 06/23/2005], Mercury Compounds; Hazardous Substances Databank
Number: 6943. Available at: http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen7HSDB
C-94
-------
Hazardous Substances Data Bank (HSDB). 2005c. Bethesda, MD: National Library of Medicine,
U.S. [Last Revision Date 06/23/2005], Benzo(a)pyrene; Hazardous Substances Databank
Number: 2554. Available at: http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen7HSDB
Hofelt, C.S., M. Honeycutt, J.T. McCoy, and L.C. Haws. 2001. Development of a metabolism
factor for polycyclic aromatic hydrocarbons for use in multipathway risk assessments of
hazardous waste combustion facilities. Regulatory Toxicology and Pharmacology 33:60-65.
Holzworth, G.C. 1972. Mixing Heights, Wind Speeds, and Potential for Urban Air Pollution
throughout the Contiguous United States. Prepared for EPA Office of Air Programs. Research
Triangle Park, NC.
Hrudey, S.E., W. Chen and C.G. Roussex, 1996. Bioavailability in environmental risk
assessment. CRC Press, Inc, Lewis publishers.
ICF International (ICF). 2004. TRIM.FaTE Application for Two Secondary Lead Smelting
Facilities and TRIM.FaTE-IEM Model Comparison. Revised Draft Report. Submitted to Terri
Hollingsworth, Work Assignment Manager, EPA/OAQPS. June.
ICF International (ICF). 2005. Memorandum: TRIM.FaTE Screening Scenario: Aquatic Food
Web Analysis; submitted to Deirdre Murphy and Terri Hollingsworth, U.S. EPA, from Margaret
McVey and Rebecca Kauffman, ICF Consulting. October 18.
International Union of Pure and Applied Chemistry (IUPAC). 1993. Glossary for Chemists of
Terms Used in Toxicology. IUPAC, Clinical Chemistry Division, Commission on Toxicology
(IUPAC Recommendations 1993). Accessed online via the U.S. National Library of Medicine
website at http://sis.nlm.nih.gov/enviro/glossarymain.html; last accessed October 2008.
Kazerouni, N., R. Sinha, C.-H. Hsu, A. Greenberg, and N. Rothman. 2001. Analysis of 200 food
items for benzo[a]pyrene and estimation of its intake in an epidemiologic study. Food and
Chemical Toxicology 39: 423-436.
Lorber, M.N., D.L.Winters, J. Griggs, R. Cook, S. Baker, J. Ferrario, C. Byrne, A. Dupuy,
and J. Schaum. 1998. A national survey of dioxin-like compounds in the United States milk
supply. Organohalogen Compounds 38:125-129.
Mason RP, J. Laporte, and S. Andres. 2000. Factors controlling the bioaccumulation of
mercury, methylmercury, arsenic, selenium, and cadmium by freshwater invertebrates and fish.
Archives for Environmental Contamination and Toxicology 38(3):283-97.
McKone, T.E., A. Bodnar, and E. Hertwich. 2001. Development and evaluation of state-
specific landscape data sets for multimedia source-to-dose models. University of California at
Berkeley. Supported by U.S. Environmental Protection Agency (Sustainable Technology
Division, National Risk Management Research Laboratory) and Environmental Defense Fund.
July. LBNL-43722.
Morton, F.I. 1986. Practical estimates of lake evaporation. Journal of Climate and Applied
Meteorology 25:371-387.
National Climatic Data Center (NCDC). 1995. Hourly United States Weather Observations
(HUSWO) 1990-1995.
C-95
-------
Ramos, L., E. Eljarrat, L.M. Hernandez, L. Alonso, J. Rivera, and M.J. Gonzalez. 1997.
Levels of PCDDs and PCDFs in farm cow's milk located near potential contaminant sources in
Asturias (Spain). Comparison with levels found in control, rural farms and commercial
pasteurized cow's milk. Chemosphere 35(10):2167-2179.
Reinfelder, J.R., N.S. Fisher, S.N. Luoma, J.W. Nichols, and W.-X. Wang. 1998. Trace element
trophic transfer in aquatic organisms: A critique of the kinetic model approach. The Science of
the Total Environment 219: 117-135.
Saiki, M.K., D.T. Castleberry, T.W. May, B.A. Martin, F.N. Bullard. 1995. Copper, cadmium, and
zinc concentrations in aquatic food chains from the upper Sacramento River (California) and
selected tributaries. Arch. Environ. Contam. Toxicol. 29:484-491.
Schmid, P., E. Gujer, M. Zennegg, and C. Studer. 2003. Temporal and local trends of
PCDD/F levels in cow's milk in Switzerland. Chemosphere 53:129-136.
Stull, R.B. 1988. An Introduction to Boundary Layer Meteorology. Kluwer Academic
Publishers, The Netherlands. 452 pp.
U.S. Environmental Protection Agency (EPA). 1989. Review of National Ambient Air Quality
Standard for Lead: Exposure Analysis Methodology and Validation. EPA-450/2-89-011.
Research Triangle Park, NC: Office of Air Quality Planning and Standards; June. Available at:
http://nepis.epa.gov/
U.S. Environmental Protection Agency (EPA). 1992a. National Study of Chemical Residues in
Fish, Volume I. EPA 823-R-92-008a. Washington, D.C.: Office of Science and Technology;
September. Available at: http://www.epa.gov/waterscience/library/fish/residuevol1.pdf.
U.S. Environmental Protection Agency (EPA) 1992b. Dermal Exposure Assessment: Principles
and Applications. EPA/8-91/011B. Available at:
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=12188.
U.S. Environmental protection Agency (EPA). 1993. Provisional Guidance for Quantitative risk
Assessment of Polycyclic Aromatic Hydrocarbons. Office of Research and Development, Office
of Health and Environmental Assessment, Washington, DC. EPA/600/R-93/089. Available at:
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=49732.
U.S. Environmental Protection Agency (EPA). 1994. Technical Support Document: Parameters
and Equations Used in the Integrated Exposure Uptake Biokinetic Model for Lead in Children
(v.099d). EPA 540/R-94/040; OSWER-9285.7-22. Office of Solid Waste and Emergency
Response; December. Available at: http://www.epa.gov/superfund/lead/products/tsd.pdf.
U.S. Environmental Protection Agency (EPA). 1997a. Exposure Factors Handbook, Volume II,
Food Ingestion Factors. EPA/600/P-95/002Fb. Office of Research and Development,
Washington, DC. August.
U.S. Environmental Protection Agency (EPA). 1997b. Mercury Study Report to Congress.
Volume III: Fate and Transport of Mercury in the Environment. EPA-452/R-97-005. Office of Air
Quality Planning and Standards and Office of Research and Development. December.
U.S. Environmental Protection Agency (EPA). 1999. Short Sheet: IEUBK Model Soil/Dust
Ingestion Rates. EPA-540-F-00-007; OSWER-9285.7-33. Washington, D.C.: Office of Solid
C-96
-------
Waste and Emergency Response; December. Available at:
http://www.epa.gov/superfund/lead/products/ssircolo.pdf.
U.S. Environmental Protection Agency (EPA). 2001. Risk Assessment Guidance for Superfund:
Volume III - Part A, Process for Conducting Probabilistic Risk Assessment. Appendix A:
Sensitivity Analysis: How Do We Know What's Important? EPA 540-R-02-002. December.
http://www.epa.gov/oswer/riskassessment/rags3adt/pdf/appendixa.pdf.
U.S. Environmental Protection Agency (EPA). 2002a. Total Risk Integrated Methodology:
TRIM.FaTE Technical Support Document. Volume II: Description of Chemical Transport and
Transformation Algorithms. EPA-453/R-02-011 b. Office of Air Quality Planning and Standards:
Research Triangle Park, NC. September.
U.S. Environmental Protection Agency (EPA). 2002b. Evaluation of TRIM.FaTE, Volume I:
Approach and Initial Findings. EPA-453/R-02-0012. Office of Air Quality and Planning
Standards: Research Triangle Park, NC. September.
U.S. Environmental Protection Agency (EPA). 2002b. Evaluation of TRIM.FaTE, Volume III:
Model Comparison Focusing on Dioxin Test Case. EPA-453/R-04-002. Office of Air Quality
and Planning Standards: Research Triangle Park, NC. December.
U.S. Environmental Protection Agency (EPA). 2002d. Estimated Per Capita Fish Consumption
in the United States. Office of Water, Office of Science and Technology, Washington, D.C.
EPA-821- C- 02-003. August. Available at:
http://www.epa.gov/waterscience/fish/files/consumption_report.pdf.
U.S. Environmental Protection Agency (EPA). 2003. Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) Modeling System, Volume II: Site-based, Regional, and
National Data. SAB Review Draft. EP-530/D-03-001b. Office of Research and Development,
Athens, GA, and Research Triangle Park, NC, and Office of Solid Waste, Washington, DC.
July. Available at: http://www.epa.gov/epaoswer/hazwaste/id/hwirwste/risk03.htm.
U.S. Environmental Protection Agency (EPA). 2004a. Air Toxics Risk Assessment Reference
Library; Volume 1 - Technical Resource Document, Part III, Human Health Risk Assessment:
Multipathway Chapter 14, Overview and Getting Started: Planning and Scoping the
Multipathway Risk Assessment. Office of Air Quality Planning and Standards, Research
Triangle Park, NC. April. Available at:
http://www.epa.gov/ttn/fera/data/risk/vol_1/chapter_14.pdf.
U.S. Environmental Protection Agency (EPA). 2004b. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds.
Volume 2: Properties, Environmental Levels, and Background Exposures. Dioxin
Reassessment, NAS Review Draft. U.S. Environmental Protection Agency, Washington, D.C.,
EPA/600/P-00/001Cb. Available at: http://www.epa.gov/ncea/pdfs/dioxin/nas-review/
U.S. Environmental Agency (EPA). 2004c. Risk Assessment Guidance for Superfund Volume
1: Human Health Evaluation Manual (Part E, Supplemental Guidance for Dermal Risk
Assessment). EPA/540/R99/005. Available at:
http://www.epa.gov/oswer/riskassessment/ragse/index.htm.
U.S. Environmental Protection Agency (EPA). 2005a. Human Health Risk Assessment
Protocol for Hazardous Waste Combustion Facilities (including the Hazardous Waste
C-97
-------
Companion Database of chemical-specific parameter values). U.S. Environmental Protection
Agency, Office of Solid Waste and Emergency Response, Washington, DC. EPA-530-R-05-
006. September.
U.S. Environmental Protection Agency (EPA). 2005b. Evaluation of TRIM.FaTE, Volume II:
Model Comparison Focusing on Mercury Test Case. EPA-453/R-05-002. Office of Air Quality
and Planning Standards: Research Triangle Park, NC. July.
U.S. Environmental Protection Agency (EPA). 2005c. Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens. EPA-630/R-03-003F. Risk Assessment
Forum: Washington, DC. March.
U.S. Environmental Protection Agency (EPA). 2005d. Analysis of Total Food Intake and
Composition of Individual's Diet Based on the U.S. Department of Agriculture's 1994-96, 1998
Continuing Survey of Food Intakes By Individuals (CSFII) (Final). Office of Research and
Development, National Center for Environmental Assessment, Washington, D.C. EPA/600/R-
05/062F. Available at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=132173.
U.S. Environmental Protection Agency (EPA). 2006a. Risk and Technology Review (RTR)
Assessment Plan. Office of Air and Radiation. November 20, 2006. Available at:
http://www.epa.gov/sab/panels/consul_risk_and_tech_assessment_plan.htm.
U.S. Environmental Protection Agency (EPA). 2006b. Air Quality Criteria for Lead (Final).
Office of Research and Development. EPA-600/R-05-44aF and -bF. Washington, DC.
October.
U.S. Environmental Protection Agency (EPA). 2006c. Air Toxics Risk Assessment Reference
Library; Volume 3 - Community Scale Assessment; Part II, Multisource Cumulative Human
Health Assessment: Inhalation. Chapter 3, Overview of a Human Health Multisource
Cumulative Inhalation Assessment. Office of Air Quality Planning and Standards, Research
Triangle Park, NC. April. Available at:
http://www.epa.gov/ttn/fera/data/risk/vol_3/Chapter_03_April_2006.pdf.
U.S. Environmental Protection Agency (EPA). 2006d. Child-Specific Exposure Factors
Handbook, External Review Draft. EPA-600-R-06-096A. National Center for Environmental
Assessment, Office of Research and Development; September. Available at:
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=56747
U.S. Environmental Protection Agency (EPA). 2008. Child-Specific Exposure Factors
Handbook. Office of Research and Development, Washington, D.C. EPA/600/R-06/096F.
September. Available at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=199243 .
U.S. Food and Drug Administration (FDA). 2007. Total Diet Study Statistics on Elemental
Results (original samples), Revision 4.1, Market Baskets 1991-3 through 2005-4. Center for
Food Safety and Applied Nutrition, College Park, MD. December 11, 2007.
U.S. Geological Survey (USGS). 1987. National Water Summary 1987 - Hydrologic Events
and Water Supply and Use. USGS Water-Supply Paper 2350. J.E. Carr, E.B. Chase, R.W.
Paulson, and D.W. Moody, Compilers.
Wischmeier, W.H., and D. Smith. 1978. Predicting Rainfall Erosion Losses: A Guide to
Conservation Planning. USDA-ARS Agriculture Handbook No. 537, Washington, D.C. 58 pp.
C-98
-------
ATTACHMENT C-1: TRIM.FaTE Inputs
-------
[This page intentionally left blank.]
C-1-i
-------
TABLE OF CONTENTS
1 TRIM.FaTE Inputs 1
1.1 Introduction 1
Exhibit 1. TRIM.FaTE Simulation Parameters 2
Exhibit 2. Meteorological Inputs 3
Exhibit 3. Air Parameters 4
Exhibit 4. Soil and Groundwater Parameters 5
Exhibit 5. Runoff Assumptions 8
Exhibit 6. USLE Erosion Parameters 9
Exhibit 7. Terrestrial Plant Placement 10
Exhibit 8. Terrestrial Plant Parameters 11
Exhibit 9. Surface Water Parameters 14
Exhibit 10. Sediment Parameters 15
Exhibit 11. Aquatic Plant Parameters 16
Exhibit 12. Aquatic Animals Food Chain, Density, and Mass 17
Exhibit 13. Cadmium Chemical-Specific 18
Exhibit 14. Mercury Chemical-Specific Properties 29
Exhibit 15. PAH Chemical-Specific Properties 20
Exhibit 16. Dioxin Chemical-Specific Properties 21
Exhibit 17. Cadmium Chemical-Specific Properties for Abiotic Compartments 24
Exhibit 18. Mercury Chemical-Specific Properties for Abiotic Compartments 25
Exhibit 19. PAH Chemical-Specific Properties for Abiotic Compartments 29
Exhibit 20. Dioxin Chemical-Specific Properties for Abiotic Compartments 31
Exhibit 21. Cadmium Chemical-Specific Properties for Plant Compartments 35
Exhibit 22. Mercury Chemical-Specific Properties for Plant Compartments 36
Exhibit 23. PAH Chemical-Specific Properties for Plant Compartments 38
C-1-ii
-------
Exhibit 24. Dioxin Chemical-Specific Properties for Plant Compartments 39
Exhibit 25. Cadmium Chemical-Specific Properties for Aquatic Species 40
Exhibit 26. Mercury Chemical-Specific Properties for Aquatic Species 41
Exhibit 27. PAH Chemical-Specific Properties for Aquatic Species 42
Exhibit 28. Dioxin Chemical-Specific Properties for Aquatic Species 43
C-1-iii
-------
1 TRIM.FaTE Inputs
1.1 Introduction
This attachment provides the tables of the detailed modeling inputs for the TRIM.FaTE
screening scenario. Exhibit 1 presents runtime settings for TRIM.FaTE. Exhibits 2 and 3
present air parameters entered into the model. Exhibits 3 through 8 present the terrestrial
parameters. Exhibits 9 through 12 present the lake parameters, and 13 through 28 present the
chemical specific parameters.
C-1-1
-------
Exhibit 1. TRIM.FaTE Simulation Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Start of simulation
date/time
1/1/1990, midnight
Consistent with met data.
End of simulation
date/time
1/1/2040, midnight
Consistent with met data set; selected to
provide a 50-year modeling period.
Simulation time step
hr
1
Selected value.
Output time step3
hr
4
Selected value.
aOutput time step is set in TRIM.FaTE using the scenario properties "simulationStepsPerOutputStep" and
"simulationTimeStep."
C-1-2
-------
Exhibit 2. Meteorological Inputs
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Meteorological Inputs (all TRIM.FaTE scenario properties, except mixing height)
Air temperature
degrees K
298
USEPA 2005.
Horizontal wind
speed
m/sec
2.8
5th percentile annual average value for contiguous US,
calculated from 30 yrs of annual normal temperature values.
Vertical wind
speed
m/sec
0.0
Professional judgment; vertical wind speed not used by any of
the algorithms in the version of the TRIM.FaTE library used for
screening.
Wind direction
degrees clockwise
from N (blowing
from)
3-days-on
4-days-off
On is defined as time during which wind is blowing into the
model domain. A conservative estimate of time during which
wind should blow into the modeling domain was determined by
evaluating HUSWO; it was concluded that a conservative
estimate would be approximately 42% of the time.
Rainfall Rate
m3[rain]/m2[surface
area]-day
varies daily
1.5 m/yr is the maximum statewide 30-year (1971-2000)
average for the contiguous United States, excluding Rhode
Island because of extreme weather conditions on Mt.
Washington. Data obtained from the National Climatic Data
Center at
http://www.ncdc.noaa.gov/oa/climate/online/ccd/nrmpcp.txt.
The precipitation frequency was 3-days-on:4-days-off based on
data from Holzworth, 1972.
Mixing height
(used to set air
VE property
named "top")
m
710
5th percentile annual average mixing heights (calculated from
daily morning and afternoon values), for all stations on SCRM
(40 state, 70 stations).
isDay_SteadySta
te_forAir
unitless
-
Value not used in current dynamic runs (would need to be
reevaluated if steady-state runs are needed).
isDay_SteadySta
te_forOther
unitless
-
C-1-3
-------
Exhibit 3. Air Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Atmospheric dust load
kg[dust]/m3[air]
6.15E-08
Bidleman 1988
Density of air
g/cm3
0.0012
U.S. EPA 1997
Dust density
kg[dust]/m3[dust]
1,400
Bidleman 1988
Fraction organic matter on
particulates
unitless
0.2
Harnerand Bidleman 1998
Height
m
800
5th percentile for United States
C-1-4
-------
Exhibit 4. Soil and Groundwater Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Surface Soil Compartment Type
Air content
volume[airl/volume[compartmentl
0.28
McKone et al. 2001.
Average vertical velocity of
water (percolation)
m/day
8.22E-04
Assumed to be 0.2 times average
precipitation for site.
Boundary layer thickness
above surface soil
m
0.005
Thibodeaux 1996; McKone et al.
2001 (Table 3).
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2600
Default in McKone et al. 2001 (Table
3)
Thickness - unfilled3
m
0.01
McKone et al. 2001 (p. 30).
Thickness - tilled3
m
0.20
USEPA 2005.
Erosion fraction
unitless
variesb
See Erosion and Runoff Fraction
table.
Fraction of area available
for erosion
m2[area available]/m2[total]
1
Professional judgment; area
assumed rural.
Fraction of area available
for runoff
m2[area available]/m2[total]
1
Professional judgment; area
assumed rural.
Fraction of area availabe
for vertical diffusion
Fraction Sand
m2[area available]/m2[total]
1
Professional judgment; area
assumed rural.
unitless
0.25
Professional judgment.
Organic carbon fraction
unitless
0.008
U.S. average in McKone et al. 2001
(Table 16 and A-3).
PH
unitless
6.8
Professional judgment.
Runoff fraction
unitless
variesb
See Erosion and Runoff Fraction
table.
Total erosion rate
kg [soil]/m2/day
variesb
See Total Erosion Rates table.
Total runoff rate
m3[water]/m2/day
1.64E-03
Calculated using scenario-specific
precipitation rate and assumptions
associated with water balance.
Water content
volume[water]/volume[compartment]
0.19
McKone et al 2001 (Table 15).
C-1-5
-------
Exhibit 4. Soil and Groundwater Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Root Zone Soil Compartment Type
Air content
volume[air]/volume[compartment]
0.25
McKoneet al 2001 (Table 16).
Average vertical velocity of
water (percolation)
m/day
8.22E-04
Assumed as 0.2 times average
precipitation for New England in
McKone et al. 2001.
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2,600
McKone et al. 2001 (Table 3).
Fraction Sand
unitless
0.25
Professional judgment.
Thickness - unfilled3
m
0.79
McKone et al. 2001 (Tabel 16 - U.S.
Average).
Thickness - tilled3
m
0.6
Adjusted from McKone et al. 2001
(Table 16).
Organic carbon fraction
unitless
0.008
McKone et al. 2001 (Table 16 and A-
3, U.S. Average).
PH
unitless
6.8
Professional judgment.
Water content
volume[water]/volume[compartment]
0.21
McKone et al. 2001 (Table 16).
Vadose Zone Soil Compartment Type
Air content
volume[air]/volume[compartment]
0.22
McKoneet al. 2001 (Table 17).
Average vertical velocity of
water (percolation)
m/day
8.22E-04
Assumed as 0.2 times average
precipitation for New England in
McKone et al. 2001.
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2,600
Default in McKone et al. 2001 (Table
3).
Fraction Sand
unitless
0.35
Pofessional judgment.
Thickness3
m
1.4
McKoneet al. 2001 (Table 17).
Organic carbon fraction
unitless
0.003
McKone et al. 2001 (Table 16 and A-
3, U.S. Average).
PH
unitless
6.8
Professional judgment.
Water content
volume[water]/volume[compartment]
0.21
McKoneet al. 2001 (Table 17-
national average).
C-1-6
-------
Exhibit 4. Soil and Groundwater Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Ground Water Compartment Type
Thickness3
m
3
McKone et al. 2001 (Table 3).
Fraction Sand
unitless
0.4
Professional judgment.
Organic carbon fraction
unitless
0.004
Professional judgment.
PH
unitless
6.8
Professional judgment.
Porosity
volume[total pore
space l/volume[compartmentl
0.2
Default in McKone et al. 2001 (Table
3).
Density of Solid material in
aquifer
kg[soil]/m3[soil]
2,600
Default in McKone et al. 2001 (Table
3).
aSet using the volume element properties file
bSee separate tables for erosion/runoff fractions and total erosion rates.
C-1-7
-------
Exhibit 5. Runoff Assumptions
for the TRIM.FaTE Screening Scenario
Originating Compartment
Destination Compartment
Runoff/Erosion Fraction
SurfSoil N1
0.0
SurfSoil Source
SurfSoil S1
0.0
sink
1.0
SW Pond
1.0
SurfSoil Source
0.0
SurfSoil N1
SurfSoil N6
0.0
SurfSoil S1
0.0
sink
0.0
SW Pond
1.0
SurfSoil_S1
SurfSoil Source
0.0
SurfSoil N1
0.0
sink
0.0
SW Pond
1.0
SurfSoil_N6
SurfSoil N1
0.0
SurfSoil N7
0.0
sink
0.0
SW Pond
1.0
SurfSoil_N7
SurfSoil N6
0.0
SurfSoil N3
0.0
sink
0.0
SW Pond
1.0
SurfSoil_N3
SurfSoil N7
0.0
SurfSoil N4
0.0
sink
0.0
SW Pond
1.0
SurfSoil N3
0.0
SurfSoil N4
SurfSoil N5
0.0
SurfSoil S4
0.0
sink
0.0
SW Pond
1.0
SurfSoil_S4
SurfSoil N4
0.0
SurfSoil S5
0.0
sink
0.0
SW Pond
0.0
SurfSoil_N5a
SurfSoil N4
0.5
SurfSoil S5
0.5
sink
0.0
SW Pond
0.0
SurfSoil_S5a
SurfSoil N5
0.0
SurfSoil S4
1.0
sink
0.0
a Assumes that N5 is higher ground that S5, and half of the runoff flows into N4, and the other
half in S5. Assumes all runoff from S5 flows into S4.
C-1-8
-------
Exhibit 6. USLE Erosion Parameters
for the TRIM.FaTE Screening Scenario
Soil
Parcel
Area
Rainfall/
Erosivity
Index
Soil
Erodibility
Index
Length-
Slope
Factor
Land Use
Cover
Mgmt
Factor
Supporting
Practices
Factor
Unit Soil Loss
Sediment
Delivery
Ratio3
Calculated
(Adjusted)
Erosion Rate
m2
R (100 ft-
ton/ac)
K (ton/ac/(100
ft-ton/acre))
LS
(USCS)
type
C
(USCS)
P
A
(ton/ac/yr)
A
(kg/m2/d)
SDRa
calculated
(adjusted)
erosion rate
(kg/m2/d)
N1
5.8E+04
300
0.39
1.5
grass
0.1
1
17.55
0.010779
0.533
0.005740
N6
4.1E+04
300
0.39
1.5
crops
0.2
1
35.1
0.021557
0.557
0.012014
N7
7.3E+04
300
0.39
1.5
grass
0.1
1
17.55
0.010779
0.518
0.005580
N3
3.5E+05
300
0.39
1.5
grass
0.1
1
17.55
0.010779
0.385
0.004151
N4
2.0E+06
300
0.39
1.5
forest
0.1
1
17.55
0.010779
0.309
0.003331
N5
6.7E+06
300
0.39
1.5
forest
0.1
1
17.55
0.010779
0.196
0.002116
S1
5.8E+04
300
0.39
1.5
grass
0.1
1
17.55
0.010779
0.533
0.005740
S4
2.0E+06
300
0.39
1.5
forest
0.1
1
17.55
0.010779
0.309
0.003331
S5
6.7E+06
300
0.39
1.5
forest
0.1
1
17.55
0.010779
0.196
0.002116
Calculated using SD = a * (AL)"b; where a is the empirical intercept coefficient (based on the size of the watershed), AL is the total watershed area receiving
deposition (m2), and b is the empirical slope coefficient (always 0.125).
C-1-9
-------
Exhibit 7. Terrestial Plant Placement
for the TRIM.FaTE Screening Scenario
Surface Soil
Surface Soil
Coniferous
Grasses/
Volume Element
Depth (m)
Forest
Herbs
None
Source
0.01
X
N1
0.01
X
N6
0.20 (tilled)
N7
0.01
X
N3
0.01
X
N4
0.01
X
N5
0.01
X
S1
0.01
X
S4
0.01
X
S5
0.01
X
c-1-10
-------
Exhibit 8. Terrestrial Plant Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Coniferous3
Grass/Herb3
Value Used
Reference
Value Used
Reference
Leaf Compartment Type
Allow exchange
1=yes, 0=no
1
-
seasonal13
-
Average leaf area index
m2[leaf]/ m2[area]
5.0
Harvard Forest, dom. red oak and
red maple, CDIAC website
5.0
Mid-range of 4-6 for old fields, R.J.
Luxmoore, ORNL.
Calculate wet dep interception
fraction (boolean)
1=yes, 0=no
0
Professional judgment.
0
Professional judgment.
Correction exponent, octanal to
lipid
unitless
0.76
From roots, Trapp 1995.
0.76
From roots, Trapp 1995.
Degree stomatal opening
unitless
1
Set to 1 for daytime based on
professional judgment (stomatal
diffusion is turned off at night using a
different property, IsDay).
1
Set to 1 for daytime based on
professional judgment (stomatal
diffusion is turned off at night using a
different property, IsDay).
Density of wet leaf
kg/m3
820
Paterson et al. 1991.
820
Paterson et al. 1991.
Leaf wetting factor
m
3.00E-04
1E-04 to 6E-04 for different crops
and elements, Muller and Prohl 1993.
3.00E-04
1 E-04 to 6E-04 for different crops
and elements, Muller and Prohl 1993.
Length of leaf
m
0.01
Professional judgment.
0.05
Professional judgment.
Lipid content
kg/kg wet weight
0.00224
European beech, Riederer 1995.
0.00224
European beech, Riederer 1995.
Litter fall rate
1/day
0.0021
value assumes 1st-order relationship
and that 99% of leaves fall over 6
years
seasonal0
-
Stomatal area normalized
effective diffusion path length
1/m
200
Wilmer and Fricker 1996.
200
Wilmer and Fricker 1996.
Vegetation attenuation factor
m2/kg
2.9
Grass/hay, Baes et al. 1984.
2.9
Grass/hay, Baes et al. 1984.
Water content
unitless
0.8
Paterson et al. 1991.
0.8
Paterson et al. 1991.
Wet dep interception fraction
unitless
0.2
Calculated based on 5 years of local
met data, 1987-1991.
0.2
Calculated based on 5 years of local
met data, 1987-1991.
c-1-11
-------
Exhibit 8. Terrestrial Plant Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Coniferous3
Grass/Herb3
Value Used
Reference
Value Used
Reference
Wet mass of leaf per soil area
kg[fresh
leaf]/m2[area]
2.0
Calculated from leaf area index, leaf
thickness (Simonich & Hites, 1994),
density of wet foliage.
0.6
Calculated from leaf area index and
Leith 1975.
Particle on Leaf Compartment Type
Allow exchange
1=yes, 0=no
1
-
seasonal13
-
Volume particle per area leaf
m3[leaf
particles]/m2[leaf]
1.00E-09
Based on particle density and size
distribution for atmospheric particles
measured on an adhesive surface,
Coe and Lindberg 1987.
1.00E-09
Based on particle density and size
distribution for atmospheric particles
measured on an adhesive surface,
Coe and Lindberg 1987.
Root Compartment Type -
Slonwoody Only
Allow exchange
1=yes, 0=no
seasonal13
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995.
Lipid content of root
kg/kg wet weight
0.011
Calculated.
Water content of root
kg/kg wet weight
0.8
Professional judgment.
Wet density of root
kg/m3
820
Soybean, Paterson et al. 1991.
Wet mass per soil area
kg/m2
1.4
Temperate grassland, Jackson et al.
1996.
Stem Compartment Type -
Nonwoody Only
Allow exchange
1=yes, 0=no
seasonal13
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995
Density of phloem fluid
kg/m3
1,000
Professional judgment.
Density of xylem fluid
kg/cm3
900
Professional judgment.
Flow rate of transpired water
per leaf area
m3[water]/m2[leaf]
0.0048
Crank et al. 1981.
Fraction of transpiration flow
rate that is phloem rate
unitless
0.05
Paterson et al. 1991.
Lipid content of stem
kg/kg wet weight
0.00224
Leaves of European beech, Riederer
1995.
C-1-12
-------
Exhibit 8. Terrestrial Plant Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Coniferous3
Grass/Herb3
Value Used
Reference
Value Used
Reference
Water content of stem
unitless
0.8
Paterson et al. 1991
Wet density of stem
kg/m3
830
Professional judgment.
Wet mass per soil area
kg/m2
0.24
Calculated from leaf and root
biomass density based on
professional judgment.
aSee separate table for assignment of plant types to surface soil compartments.
bBegins March 9 (set to 1), ends November 7 (set to 0). Nation-wide 80th percentile.
cBegins November 7, ends December 6; rate = 0.15/day during this time (value assumes 99 percent of leaves fall in 30 days).
C-1-13
-------
Exhibit 9. Surface Water Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Algae carbon content (fraction)
unitless
0.465
APHA 1995.
Algae density in water column
g[algae]/L[water]
0.0025
Millard et al. 1996 as cited in ICF
2005.
Algae growth rate
1/day
0.7
Hudson et al. 1994 as cited in
Mason et al. 1995b
Algae radius
um
2.5
Mason et al. 1995b.
Algae water content (fraction)
unitless
0.9
APHA 1995.
Average algae cell density (per
vol cell, not water)
g[algae]/m3[algae]
1,000,000
Mason et al. 1995b, Mason et al.
1996.
Boundary layer thickness above
sediment
m
0.02
Cal EPA 1993.
Chloride concentration
mg/L
8.0
Kaushal et al. 2005.
Chlorophyll concentration
mg/L
0.0029
ICF 2005.
Depth3
m
3.18
Wl DNR 2005 - calculation based on
relationship between drainage basin
and lake area size.
Dimensionless viscous
sublayer thickness
unitless
4
Ambrose et al. 1995.
Drag coefficient for water body
unitless
0.0011
Ambrose et al. 1995.
Flush rate
1/year
12.17
Calculated based on pond
dimensions and flow calculations.
Fraction Sand
unitless
0.25
Professional judgment.
Organic carbon fraction in
suspended sediments
unitless
0.02
Professional judgment.
PH
unitless
7.3
Professional judgment.
Suspended sediment
deposition velocity
m/day
2
USEPA 1997.
Total suspended sediment
concentration
kg[sediment]/m3[water
columnl
0.05
USEPA 2005.
Water temperature
degrees K
298
USEPA 2005.
aSet using the volume element properties named "top" and "bottom."
C-1-14
-------
Exhibit 10. Sediment Parameters
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Depth3
m
0.05
McKone et al. 2001 (Table 3).
Fraction Sand
unitless
0.25
Professional judgment.
Organic carbon fraction
unitless
0.02
McKone et al. 2001 (Table 3).
Porosity of the sediment
zone
volume [total pore
space]/volume[sediment
compartmentl
0.6
USEPA 1998.
Solid material density in
sediment
kg[sediment]/m3[sediment]
2,600
McKone et al. 2001 (Table 3).
a Set using the volume e
ement properties named "top" and "bottom."
C-1-15
-------
Exhibit 11. Aquatic Plants
for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value Used
Reference
Macrophyte biomass per water area
kg/m2
0.5
Professional judgment.
Density of macrophytes
kg/L
1
Professional judgment.
C-1-16
-------
Exhibit 12. Aquatic Animals Food Chain, Density, and Mass
for the TRIM.FaTE Screening Scenario
Fraction Diet
Aquatic Biota
(Consuming
Organism)
Algae
Macrophyte
Bethic
Invertebrate
Water
Column
Herbivore
Benthic
Omnivore
Water
Column
Omnivore
Benthic
Carnivore
Water
Column
Carnivore
Biomass
(kg/m2)
Body
Weight
Reference
Benthic
Invertebrate
0%
0%
0%
0%
0%
0%
0%
0%
0.020
2.55E-04
Professional
judgment.
Water Column
Herbivore
100%
0%
0%
0%
0%
0%
0%
0%
0.001
0.025
Professional
judgment.
Benthic Omnivore
0%
0%
100%
0%
0%
0%
0%
0%
0.002
2.50E-01
Professional
judgment.
Water Column
Omnivore
0%
30%
30%
40%
0%
0%
0%
0%
0.001
0.25
Professional
judgment.
Benthic Carnivore
0%
0%
70%
0%
30%
0%
0%
0%
0.001
2.0
Professional
judgment.
Water Column
Carnivore
0%
0%
0%
20%
20%
60%
0%
0%
0.0004
2.0
Professional
judgment.
C-1-17
-------
Exhibit 13. Cadmium Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name'1
Units
Value
Reference
CAS numberb
unitless
7440-43-9
-
Diffusion coefficient in pure air
m2[air]/day
0.71
USEPA 1999 (Table A-2-35).
Diffusion coefficient in pure
water
m2[water]/day
8.16E-05
USEPA 1999 (Table A-2-35).
Henry's Law Constant
Pa-m3/mol
1.00E-37
USEPA 1999 (Table A-2-35;
assumed to be zero).
Melting point
degrees K
594
ATSDR 1999.
Molecular weight
g/mol
112.41
ATSDR 1999.
Octanol-air partition coefficient
(Koa)
m3[air]/m3[octanol]
-
-
Octanol-carbon partition
coefficient (Koc)
-
-
Octanol-water partition
coefficient (Kow)
L[water]/kg[octanol]
-
-
aAII parameters in this table are TRIM.FaTE chemical properties.
bThese CAS numbers apply to elemental Cd; however, the cations of cadmium are being modeled.
C-1-18
-------
Exhibit 14. Mercury Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Hg(0)b
Hg(2)b
MHq"
CAS number
unitless
7439-97-6
14302-87-5
22967-92-6
-
Diffusion coefficient in
pure air
m2[air]/day
0.478
0.478
0.456
USEPA 1997.
Diffusion coefficient in
pure water
m2[water]/day
5.54E-05
5.54E-05
5.28E-05
USEPA 1997.
Henry's Law constant
Pa-m3/mol
719
7.19E-05
0.0477
USEPA 1997.
Melting Point
degrees K
234
5.50E+02
443
CARB 1994.
Molecular weight
g/mol
201
201
216
USEPA 1997.
Octanol-water
partition coefficient
(Kow)
L[water]/kg[octanol]
4.15
3.33
1.7
Mason et al. 1996.
Vapor washout ratio
m3[air]/m3[rain]
1,200
1.6E+06
0
USEPA 1997, based on
Petersen et al. 1995.
aAII parameters in this table are TRIM.FaTE chemical properties.
bOn this and all following tables, Hg(0) = elemental mercury, Hg(2) = divalent mercury, and MHg = methyl mercury.
C-1-19
-------
Exhibit 15. PAH Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
BaP
BaA
BbF
BkF
Chr
DahA
IcdP
CAS number
unitless
50-32-8
56-55-3
205-99-2
207-08-9
218-01-9
53-70-3
193-39-5
-
Diffusion coefficient in
pure air
m2/day
0.188
0.213
0.197
0.197
0.214
0.156
0.164
USEPA 1998.
Diffusion coefficient in
pure water
m2/day
5.05E-05
5.37E-05
4.74E-05
4.74E-05
5.37E-05
5.19E-05
4.89E-05
USEPA 1998.
Henry's Law constant
Pa-m3/mol
8.50E-02
3.67E-01
6.26E-01
4.20E-02
1.23E-01
1.13E-03
4.92E-04
USEPA 1998.
Melting point
degrees K
452
433
441
490
531
539
437
Budavari 1996.
Molecular weight
g/mol
252.32
228.29
252.32
252.32
228.29
278.33
276.34
Budavari 1996.
Octanol-water partition
coefficient (Kow)
L[water]/L[octanol]
9.33E+05
6.17E+05
3.98E+06
6.92E+06
5.37E+05
3.16E+06
4.57E+07
Hansch et al. 1995.
C-1-20
-------
Exhibit 16. Dioxin Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Value
Parameter Name
Units
1,2,3,4,6,7,8,
9-OCDD
1,2,3,4,6,7,8,
9-OCDF
1,2,3,4,6,7,8-
HpCDD
1,2,3,4,6,7,8-
HpCDF
1,2,3,4,7,8,9-
HpCDF
1,2,3,4,7,8-
HxCDD
1,2,3,4,7,8-
HxCDF
CAS number
unitless
3268-87-9
39001-02-0
35822-46-9
67562-39-4
55673-89-7
39227-28-6
70648-26-9
Diffusion coefficient in pure
air
m2/day
0.0883
0.123
0.0925
0.129
0.129
0.0958
0.135
Diffusion coefficient in pure
water
m2/day
3.08E-06
3.15E-05
3.24E-05
3.33E-05
3.33E-05
3.43E-05
3.53E-05
Henry's Law constant
Pa-m3/mol
0.68
0.19
1.28
1.43
1.43
1.08
1.45
Melting Point
degrees K
603
259
538
236.5
222
546
499.0
Molecular weight
g/mol
460.0
443.76
425.2
409.31
409.31
391.0
374.87
Octanol-water partition
coefficient (Kow)
L[water]/L[octanol]
1.58E+08
1.00E+08
1.00E+08
2.51E+07
7.94E+06
6.31E+07
1.00E+07
C-1-21
-------
Exhibit 16. Dioxin Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Value
Parameter Name
Units
1,2,3,6,7,8-
HxCDD
1,2,3,6,7,8-
HxCDF
1,2,3,7,8,9-
HxCDD
1,2,3,7,8,9-
HxCDF
1,2,3,7,8-
PeCDD
1,2,3,7,8-
PeCDF
2,3,4,6,7,8-
HxCDF
CAS number
unitless
57653-85-7
57117-44-9
19408-74-3
72918-21-9
40321-76-4
57117-41-6
60851-34-5
Diffusion coefficient in pure
air
m2/day
0.0958
0.135
0.0958
0.135
0.101
0.142
0.135
Diffusion coefficient in pure
water
m2/day
3.43E-05
3.53E-05
3.43E-05
3.53E-05
3.65E-05
3.76E-05
3.53E-05
Henry's Law constant
Pa-m3/mol
1.08
0.74
1.08
0.74
3.33
0.5
0.74
Melting point
degrees K
558.0
506.0
517.0
509.0
513.0
499.0
512.5
Molecular weight
g/mol
390.84
374.9
390.8
374.9
356.4
340.4
374.9
Octanol-water partition
coefficient (Kow)
L[water]/L[octanol]
1.62E+08
8.24E+07
1.62E+08
3.80E+07
4.37E+06
6.17E+06
8.31E+07
C-1-22
-------
Exhibit 16. Dioxin Chemical-Specific Properties
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
2,3,4,7,8-
PeCDF
2,3,7,8-TCDD
2,3,7,8-TCDF
CAS number
unitless
57117-31-4
1746-01-6
51207-31-9
-
Diffusion coefficient in pure
air
m2/day
0.142
0.106
0.149
USEPA 1999.
Diffusion coefficient in pure
water
m2/day
3.76E-05
5.68E-05
4.04E-05
USEPA 1999.
Henry's Law constant
Pa-m3/mol
0.5
3.33
1.46
Mackay et al. 1992 as cited in USEPA 2000a. Exceptions include
1,2,3,4,6,7,8,9-OCDF; 1,2,3,4,7,8-HxCDF; and 1,2,3,6,7,8-HxCDF which
are calculated by the VP/WS Ratio Technique (USEPA 2000a)
Melting point
degrees K
469.3
578.0
500.0
Mackay et al. 2000, exceptions include USEPA 2000b (1,2,3,6,7,8-HxCDD;
1,2,3,7,8,9-HxCDF; 1,2,3,7,8-PeCDD), ATSDR 1998(1,2,3,6,7,8-HxCDF;
1,2,3,7,8-PeCDF; 2,3,4,6,7,8-HxCDF) and NLM 2002 (1,2,3,7,8,9-HxCDD)
Molecular weight
g/mol
340.4
322.0
306.0
Mackay et al. 2000, exceptions include: ATSDR 1998(1,2,3,6,7,8-HxCDF;
1,2,3,7,8,9-HxCDF; 1,2,3,7,8-PeCDD; 1,2,3,7,8-PeCDF; 2,3,4,6,7,8-
HxCDF) and NLM 2002 (1,2,3,6,7,8-HxCDD; 1,2,3,7,8,9-HxCDD)
Octanol-water partition
coefficient (Kow)
L[water]/L[octanol]
3.16E+06
6.31E+06
1.26E+06
Mackay et al. 1992a as cited in USEPA 2000a, exceptions include: USEPA
2000b (1,2,3,6,7,8-HxCDD; 1,2,3,6,7,8-HxCDF; 1,2,3,7,8,9-HxCDD;
1,2,3,7,8,9-HxCDF; 2,3,4,6,7,8-HxCDF) and Sijm et al. 1989 (1,2,3,7,8-
PeCDD)
C-1-23
-------
Exhibit 17. Cadmium Chemical-Specific Properties for Abiotic
Compartments Documentation for the TRIM.FaTE Screening
Scenario
Parameter Name
Units
Value
Reference
Air Compartment Type
Particle dry deposition
velocity
m/day
260
Calculated from
Mulbaier and Tisue
1980.
Washout Ratio
m3[air]/m3[rai
n]
200,000
MacKay et al. 1986.
Surface Soil Compartment Type
Use input characteristic
depth (boolean)
0 = no, Else =
yes
0
Professional
judgment.
Root Zone Soil Compartment Type
Use input characteristic
depth (boolean)
0 = no, Else =
yes
0
Professional
judgment.
Vadose Zone Soil Compartment Type
Use input characteristic
depth (boolean)
0 = no, Else =
yes
0
Professional
judgment.
Surface Water Compartment Type
Ratio of concentration in
water to concentration in
algae to concentration
dissolved in water
L[water]/g[alga
e wet wt]
1.87
McGeer et al. 2003.
C-1-24
-------
Exhibit 18. Mercury Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Hg(0) Hg(2) MHg
Air Compartment Type
Particle dry deposition velocity
m/day
500
500
500
CalTOX value cited in McKone et al. 2001.
Demethylation rate
1/day
N/A
N/A
0
Professional judgment.
Methylation rate
1/day
0
0
0
Professional judgment.
Oxidation Rate
1/day
0.00385
0
0
Low end of half-life range (6 months to 2 years) in
USEPA 1997.
Reduction rate
1/day
0
0
0
Professional judgment.
Washout Ratio
m3[airl/m3[rainl
200,000
200,000
200,000
Professional judgment.
Surface Soil Compartment Type
Input characteristic depth (user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate value).
Use input characteristic depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment.
Soil-water partition coefficient
L[water]/kg[soil
wet wtl
1,000
58,000
7,000
USEPA 1997.
Vapor dry deposition velocity
m/day
50
2500
0
Hg(0) - from Lindberg et al. 1992 Hg(2) - estimate by
USEPA using the Industrial Source Complex (ISC)
Model - [See Vol. Ill, App. A of the Mercury Study
Report (USEPA, 1997)1.
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and Verta 1995 is 3E-2 to
6E-2 /day; value is average maximum potential
demethylation rate constant under anaerobic
conditions.
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and Verta 1995 is 2E-4 to
1E-3 /day; value is average maximum potential
methylation rate constant under anaerobic
conditions.
Oxidation rate
1/day
0
0
0
Value assumed in USEPA 1997.
Reduction rate
1/day
0
1.25E-05
0
Value used for unfilled surface soil (2cm), 10%
moisture content, in USEPA 1997; general range is
(0.0013/day)*moisture content to
(0.0001/day)*moisture content for forested region
(Lindberg 1996; Carpi and Lindberg 1997).
C-1-25
-------
Exhibit 18. Mercury Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Hg(0) Hg(2) MHg
Root Zone Soil Compartment Type
Input characteristic depth (user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate value).
Use input characteristic depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment.
Soil-water partition coefficient
L[water]/kg[soil
wet wtl
1,000
58,000
7,000
USEPA 1997
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and Verta 1995 is 3E-2 to
6E-2 /day; value is average maximum potential
demethylation rate constant under anaerobic
conditions.
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and Verta 1995 is 2E-4 to
1E-3 /day; value is average maximum potential
methylation rate constant under anaerobic
conditions.
Oxidation rate
1/day
0
0
0
Value assumed in USEPA 1997.
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil (20cm), 10%
moisture content, in USEPA 1997 (Lindberg 1996;
Carpi and Lindberg, 1997).
Vadose Zone Soil Compartment Type
Input characteristic depth (user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate value),
Use input characteristic depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment.
Soil-water partition coefficient
L[water]/kg[soil
wet wtl
1,000
58,000
7,000
USEPA 1997.
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and Verta 1995 is 3E-2 to
6E-2 /day; value is average maximum potential
demethylation rate constant under anaerobic
conditions.
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and Verta 1995 is 2E-4 to
1E-3 /day; value is average maximum potential
methylation rate constant under anaerobic
conditions.
Oxidation rate
1/day
0
0
0
Value assumed in USEPA 1997.
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil (20cm), 10%
moisture content, in USEPA 1997 (Lindberg 1996;
Carpi and Lindberg, 1997).
C-1-26
-------
Exhibit 18. Mercury Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Hg(0) Hg(2) MHg
Ground Water Compartment Type
Soil-water partition coefficient
L[water]/kg[soil
wet wtl
1,000
58,000
7,000
USEPA 1997.
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and Verta 1995 is 3E-2 to
6E-2 /day; value is average maximum potential
demethylation rate constant under anaerobic
conditions.
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and Verta 1995 is 2E-4 to
1E-3 /day; value is average maximum potential
methylation rate constant under anaerobic
conditions.
Oxidation rate
1/day
1.00E-08
0
0
Small default nonzero value (0 assumed in USEPA
1997).
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil (20cm), 10%
moisture content, in USEPA 1997 (Lindberg 1996;
Carpi and Lindberg 1997).
Surface Water Compartment Type
Algal surface area-specific uptake rate
constant
nmol/[|jm2-day-
nmol]
0
2.04E-10
3.60E-10
Assumes radius = 2.5mm, Mason et al. 1995b,
Mason et al. 1996; Hg(0) assumed same as Hg(2).
Dow ("overall Kow")
L[water]/kg[octano
H
0
a
b
Mason et al. 1996.
Solids-water partition coefficient
L[water]/kg[solids
wet wt]
1,000
100,000
100,000
USEPA 1997.
Vapor dry deposition velocity
m/day
N/A
2500
USEPA 1997 (Vol. Ill, App. A).
Demethylation rate
1/day
N/A
N/A
0.013
Average of range of 1E-3 to 2.5E-2/day from
Gilmourand Henry 1991.
Methylation rate
1/day
0
0.001
0
Value used in EPA 1997; range is from 1E-4 to 3E-
4/day (Gilmour and Henry 1991).
Oxidation rate
1/day
0
0
0
Professional judgment.
Reduction rate
1/day
0
0.0075
0
Value used in USEPA 1997; reported values range
from less than 5E-3/day for depths greater than
17m, up to 3.5/day (Xiao et al. 1995; Vandal et al.
1995; Mason et al. 1995a; Amyot et al. 1997).
C-1-27
-------
Exhibit 18. Mercury Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Hg(0) Hg(2) MHg
Reference
Sediment Compartment Type
Solids-water partition coefficient
L[water]/kg[solids
wet wt]
3,000
50,000
3,000
USEPA 1997.
Demethylation rate
1/day
N/A
N/A
0.0501
Average of range of 2E-4 to 1E-1/day from Gilmour
and Henry 1991.
Methylation rate
1/day
0
1.00E-04
0
Value used in EPA 1997; range is from 1E-5 to 1E-
3/day,Gilmour and Henry 1991.
Oxidation rate
1/day
0
0
0
Professional judgment.
Reduction rate
1/day
0
1.00E-06
0
Inferred value based on presence of Hg(0) in
sediment porewater (USEPA 1997; Vandal et al.
1995).
aTRIM.FaTE Formula Property, which varies from 0.025 to 1.625 depending on pH and chloride concentration.
bTRIM.FaTE Formula Property, which varies from 0.075 to 1.7 depending on pH and chloride concentration.
C-1-28
-------
Exhibit 19. PAH Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
BaP BaA BbF BkF Chr DahA IcdP
Air Compartment Type
Particle dry deposition
velocity
m/day
500
500
500
500
500
500
500
McKone et al. 2001.
Half-life
day
0.046
0.125
0.596
0.458
0.334
0.178
0.262
Howard et al. 1991 / upper bound
measured or estimated value.
Washout Ratio
200,000
200,000
200,000
200,000
200,000
200,000
200,000
Mackay et al. 1986.
Surface Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
0.08
Not used (model set to calculate value).
Use input characteristic
depth (boolean)
0 = No, Else = Yes
0
0
0
0
0
0
0
Professional judgment.
Halflife
day
530
680
610
2140
1000
940
730
MacKay et al. 2000.
Root Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
0.08
Not used (model set to calculate value).
Use input characteristic
depth
0 = No, Else = Yes
0
0
0
0
0
0
0
Professional judgment.
Half-life
day
530
680
610
2140
1000
940
730
MacKay et al. 2000.
Vadose Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
0.08
Not used (model set to calculate value).
Use input characteristic
depth (boolean)
0 = No, Else = Yes
0
0
0
0
0
0
0
Professional judgment.
Half-life
day
1060
1360
1220
4280
2000
1880
1460
Howard et al. 1991 / upper bound
measured or estimated value for
groundwater.
Groundwater Compartment Type
Half-life
day
1060
1360
1220
4280
2000
1880
1460
Howard et al. 1991 / upper bound
measured or estimated value for
groundwater.
C-1-29
-------
Exhibit 19. PAH Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
BaP BaA BbF BkF Chr DahA IcdP
Surface Water Compartment Type
RatioOfConclnAlgaeToConc
DissolvedlnWater
(g[chem]/kg[algae]) /
(g[cheml/L[waterl)
3610
3610
3610
3610
3610
3610
3610
BCF data for green algae for BaP from Lu
etal. 1977
Half-life
day
0.138
0.375
90
62.4
1.626
97.8
750
Howard et al. 1991 / upper bound
measured or estimated value.
Sediment Compartment Ty
pe
Half-life
day
2290
2290
2290
2290
2290
2290
2290
Mackay et al. 1992 / PAH values are the
mean half-life of the log class that Mackay
et al. assigned for sediment, except for
BbF and IcdP, which were not on Table
2.3.
C-1-30
-------
Exhibit 20. Dioxin Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
1.2.3.4.6.7.8.
9-OCDD
1.2.3.4.6.7.8.
9-OCDF
1.2.3.4.6.7.8-
HpCDD
1.2.3.4.6.7.8-
HpCDF
1.2.3.4.7.8.9-
HpCDF
1.2.3.4.7.8-
HxCDD
Air Compartment Type
Deposition Velocity
m/day
500
500
500
500
500
500
Ha If life
day
162
321
64
137
122
42
Washout Ratio
m3[airl/m3[rainl
91000
22000
64000
32000
32000
9000
Surface Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
0
Ha If life
day
3650
3650
3650
3650
3650
3650
Root Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
0 = No, Else = Yes
0
0
0
0
0
0
Ha If life
day
3650
3650
3650
3650
3650
3650
Vadose Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
0
Ha If life
day
1008
1008
1008
1008
1008
1008
Groundwater Compartment Type
Half-life
day
1008
1008
1008
1008
1008
1008
Surface Water Compartment Type
Ratio Of Cone In Algae To
Cone Dissolved In Water
(g[chem]/g[algae])/
(g[cheml/L[waterl)
1.025
1.025
1.025
1.025
1.025
1.025
Half-life
day
0.67
0.58
47
0.58
0.58
6.3
Sediment Compartment Type
Half-life
day
1095
1095
1095
1095
1095
1095
C-1-31
-------
Exhibit 20. Dioxin Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
1.2.3.4.7.8-
HxCDF
1.2.3.6.7.8-
HxCDD
1.2.3.6.7.8-
HxCDF
1.2.3.7.8.9-
HxCDD
1.2.3.7.8.9-
HxCDF
1.2.3.7.8-
PeCDD
Air Compartment Type
Deposition Velocity
m/day
500
500
500
500
500
500
Halflife
day
78
28
55
28
51
18
Washout Ratio
m3[airl/m3[rainl
10000
9000
10000
9000
10000
18000
Surface Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
0
Halflife
day
3650
3650
3650
3650
3650
3650
Root Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
0 = No, Else = Yes
0
0
0
0
0
0
Halflife
day
3650
3650
3650
3650
3650
3650
Vadose Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
0
Halflife
day
1008
1008
1008
1008
1008
1008
Groundwater Compartment Type
Half-life
day
1008
1008
1008
1008
1008
1008
Surface Water Compartment Type
Ratio Of Cone In Algae To
Cone Dissolved In Water
(g[chem]/g[algae])/
(g[cheml/L[waterl)
1.025
1.025
1.025
1.025
1.025
1.025
Half-life
day
0.58
6.3
0.58
6.3
0.58
2.7
Sediment Compartment Type
Half-life
day
1095
1095
1095
1095
1095
1095
C-1-32
-------
Exhibit 20. Dioxin Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
1.2.3.7.8-
PeCDF
2.3.4.6.7.8-
HxCDF
2.3.4.7.8-
PeCDF
2.3.7.8-
TCDD
2,3,7,8-
TCDF
Air Compartment Type
Deposition Velocity
m/day
500
500
500
500
500
Halflife
day
31
59
33
12
19
Washout Ratio
m3[airl/m3[rainl
13000
10000
14000
18000
19000
Surface Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
Halflife
day
3650
3650
3650
3650
3650
Root Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
0 = No, Else = Yes
0
0
0
0
0
Halflife
day
3650
3650
3650
3650
3650
Vadose Zone Soil Compartment Type
Input characteristic depth
m
0.08
0.08
0.08
0.08
0.08
Use input characteristic depth
(boolean)
0 = No, Else = Yes
0
0
0
0
0
Halflife
day
1008
1008
1008
1008
1008
Groundwater Compartment Type
Half-life
day
1008
1008
1008
1008
1008
Surface Water Compartment Type
Ratio Of Cone In Algae To
Cone Dissolved In Water
(g[chem]/g[algae])/
(g[cheml/L[waterl)
1.025
1.025
1.025
1.025
1.025
Half-life
day
0.19
0.58
0.19
2.7
0.18
Sediment Compartment Type
Half-life
day
1095
1095
1095
1095
1095
C-1-33
-------
Exhibit 20. Dioxin Chemical-Specific Properties for Abiotic Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Reference
Air Compartment Type
Deposition Velocity
McKone et al. 2001
Halflife
Atkinson 1996 as cited in USEPA 2000; vapor phase reaction
with hydroxyl radical
Washout Ratio
Vulykh et al. 2001
Surface Soil Compartment Type
Input characteristic depth
Not used (model set to calculate value)
Use input characteristic depth (boolean)
Professional judgment
Halflife
Mackay et al. 2000; the degradation rate was cited by multiple
authors, value is for2,3,7,8-TCDD
Root Zone Soil Compartment Type
Input characteristic depth
not used (model set to calculate value)
Use input characteristic depth
professional judgment
Halflife
Mackay et al. 2000; the degradation rate was cited by multiple
authors, value is for2,3,7,8-TCDD
Vadose Zone Soil Compartment Type
Input characteristic depth
Not used (model set to calculate value).
Use input characteristic depth (boolean)
Professional judgment.
Halflife
Average value of the range presented in Mackay et al. 2000;
based on estimated unacclimated aerobic biodegradation half-
life, value is for 2,3,7,8-TCDD.
Groundwater Compartment Type
Half-life
Average value of the range presented in Mackay et al. 2000;
based on estimated unacclimated aerobic biodegradation half-
life, value is for 2,3,7,8-TCDD.
Surface Water Compartment Type
Ratio Of Cone In Algae To Cone Dissolved
In Water
BCF data for green algae for 2,3,7,8-TCDD from Isense
1978, at 32 days.
Half-life
Kim and O'Keefe. 1998, as cited in USEPA. 2000.
Sediment Compartment Type
Half-life
Estimation based on Adriaens and Grbic-Galic 1992,1993
and Adriaens et al. 1995 as cited in USEPA 2000.
C-1-34
-------
Exhibit 21. Cadmium Chemical-Specific Properties for Plant Compartments
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Leaf Compartment Type
Transfer factor to leaf
particle
1/day
0.002
Professional judgment.
Particle on Leaf Compartment Type
Transfer factor to leaf | 1/day | 0.200 Professional judgment.
Root Compartmenl
Type - Grasses and Herbs a
Root to Root Soil
Partition- Alpha of
Steady State
unitless
0.95
Henning et al. 2001.
Root to Root Soil
Partition- Partitioning
Coefficient
m3[bulk root
soil]/m3[root]
0.23
Nriagu 1980; based on average value calculated
from various agricultural plant species.
Root to Root Soil
Partition- Time to
Reach Alpha
day
28
Henning et al. 2001.
Stem Compartment Type - Grasses and Herbs a
Transpiration stream
concentration factor
(TSCF)
m3[soil pore
water]/m3[xylem
fluidl
0.45
Tsiros et al. 1999.
Aquatic Plants
Macrophyte Compartment Type
Water Column
Dissolved Partition-
Alpha of Equilibrium
unitless
0.95
Maine et al. 2001; based on assumption that
equilibrium was nearly reached during 21 day
experiment.
Water Column
Dissolved Partition-
Partition Coefficient
L[water]/kg[macrop
hyte wet wt]
100
Maine et al. 2001; based on calculations from
an average of four macrophyte species.
Water Column
Dissolved Partition-
Time to Reach
Equilibrium
day
21
Maine et al. 2001.
a Roots and stems are not modeled for deciduous or coniferous forest in the current version of TRIM.FaTE.
C-1-35
-------
Exhibit 22. Mercury Chemical-Specific Properties for Plant Compartments Documentation for the TRIM.FaTE
Screening Scenario
Parameter Name
Units
Value
Hg(0) Hg(2) MHg
Reference
Leaf Compartment Type
Transfer factor to leaf particle
1/day
0.002
0.002
0.002
Professional judgment (assumed 1% of
transfer factor from leaf particle to leaf).
Demethylation rate
1/day
N/A
N/A
0.03
Calculated from Bache et al. 1973.
Methylation rate
1/day
0
0
0
Assumed from Gay 1975, Bache et al. 1973.
Oxidation rate
1/day
1.0E+06
0
0
Professional judgment; assumed close to
instantaneous
Reduction rate
1/day
0
0
0
Professional judgment.
Particle on Leaf Compartment Type
Transfer factor to leaf
1/day
0.2
0.2
0.2
Professional judgment.
Demethylation rate
1/day
N/A
N/A
0
Professional judgment.
Methylation rate
1/day
0
0
0
Professional judgment.
Oxidation rate
1/day
0
0
0
Professional judgment.
Reduction rate
1/day
0
0
0
Professional judgment.
Root Compartment Type - Grasses and Herbs a
Alpha for root-root zone bulk soil
unitless
0.95
0.95
0.95
Selected value.
Root/root-zone-soil-water partition
coefficient
m3[bulk root soil]/
m3[root]
0
0.18
1.2
Hg2- geometric mean Leonard et al. 1998,
John 1972, Hogg et al. 1978; MHg-
assumed, based on Hogg et al. 1978.
t-alpha for root-root zone bulk soil
day
21
21
21
Professional judgment.
Demethylation rate
1/day
N/A
N/A
0
Professional judgment.
Methylation rate
1/day
0
0
0
Professional judgment.
C-1-36
-------
Exhibit 22. Mercury Chemical-Specific Properties for Plant Compartments Documentation for the TRIM.FaTE
Screening Scenario
Parameter Name
Units
Value
Reference
Hg(0)
Hg(2)
MHg
Oxidation rate
1/day
0
0
0
Professional judgment.
Reduction rate
1/day
0
0
0
Professional judgment.
Stem Compartment Type - Grasses and Herbs a
Transpiration stream concentration
factor (TSCF)
m3[soil pore water]/
m3[xylem fluidl
0
0.5
0.2
Calculation from Norway spruce, Scots pine,
Bishop et al. 1998.
Demethylation rate
1/day
N/A
N/A
0.03
Calculated from Bache et al. 1973.
Methylation rate
1/day
0
0
0
Professional judgment.
Oxidation rate
1/day
0
0
0
Professional judgment.
Reduction rate
1/day
0
0
0
Professional judgment.
Aquatic
3lants
Macrophyte Compartment Type
Water Column Dissolved Partition-Alpha
of Equilibrium
unitless
0.95
0.95
0.95
Selected value.
Water Column Dissolved Partition-
Partition Coefficient
L[water]/ kg[macrophyte
wet wt]
0.883
0.883
4.4
Elodea densa, Ribeyre and Boudou 1994.
Water Column Dissolved Partition-Time
to Reach Equilibrium
unitless
0.95
0.95
0.95
Selected value.
Oxidation rate
1/day
1.00E+09
0
0
Professional judgment.
t-alpha
day
18
18
18
Experiment duration from Ribeyre and
Boudou 1994.
a Roots and stems are not modeled for deciduous or coniferous forest in the current version of TRIM.FaTE.
C-1-37
-------
Exhibit 23. PAH Chemical-Specific Properties for Plant Compartments Documentation for the TRIM.FaTE Screening
Scenario
Parameter Name
Units
Value
Reference
BaP BaA BbF BkF Chr DahA IcdP
Terrestrial Plants
Leaf Compartment Type
Transfer factor to leaf
particle
1/day
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
Professional judgment.
Half-life
day
3.5
3.5
3.5
3.5
3.5
3.5
3.5
Edwards 1988 (as cited in Efroymson 1997)/
calculated from metabolic rate constant.
Particle on Leaf Compartment 1
fype
Transfer factor to leaf
1/day
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
1.00E-04
Professional judgment.
Half-life
day
2.31
1.84
3.56
17.8
4.12
17.8
17.8
Edwards 1988 (as cited in Efroymson 1997)/
calculated from metabolic rate constant
Root Compartment Type - Grasses and Herbs a
Half-life
day
34.6
34.6
34.6
34.6
34.6
34.6
34.6
Edwards 1988 (as cited in Efroymson 1997)/
calculated from metabolic rate constant.
Stem Compartment Type - Grasses and
Herbs a
Half-life
day
3.5
3.5
3.5
3.5
3.5
3.5
3.5
Edwards 1988 (as cited in Efroymson 1997)/
calculated from metabolic rate constant.
Aquatic Plants
Macrophyte Compartment Type
Half-life
days
3.5
3.5
3.5
3.5
3.5
3.5
3.5
Edwards 1988 (as cited in Efroymson 1997)/
calculated from metabolic rate constant.
a Roots and stems are not modeled for deciduous or coniferous forest in the current version of TRIM.FaTE.
C-1-38
-------
Exhibit 24. Doixin Chemical-Specific Properties for Plant Compartments Documentation for the
TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
All Dioxins
Terrestria
Plants
Leaf Compartment Type
Transfer factor to leaf particle
1/day
0.003
Calculated as 1 percent of transfer factor to leaf; highly
uncertain.
Half-life
day
70
Arjmand and Sandermann 1985, as cited in Komoba et al. 1995;
soybean root cell culture metabolism test data for DDE.
Particle on Leaf Compartment Type
Transfer factor to leaf
1/day
0.3
Professional judgment based on USEPA 2000c (an estimate for
mercury) and Trapp 1995; highly uncertain.
Half-life
day
4.4
McCrady and Maggard 1993; photodegradation sorbed to grass
foliage in sunlight; assumed 10 sunlight per day.
Root Compartment Type - Grasses and Herbs a
Half-life
day
70
Arjmand and Sandermann 1985, as cited in Komoba, et al.
1995; soybean root cell culture metabolism test data for DDE.
Root Soil Water Interaction - Alpha
unitless
0.95
Professional judgment.
Stem Compartment Type - Grasses and Herbs a
Half-life
day
70
Arjmand and Sandermann 1985, as cited in Komoba, et al.
1995; soybean root cell culture metabolism test data for DDE.
Aquatic
Plants
Macrophyte Compartment Type
Half-life
days
70
Arjmand and Sandermann 1985, as cited in Komoba et al. 1995;
soybean root cell culture metabolism test data for DDE.
a Roots and stems are not modeled for deciduous or coniferous forest in the current version of TRIM.FaTE.
C-1-39
-------
Exhibit 25. Cadmium Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Benthic Invertebrate Compart
tment Type
Sediment Partitioning - Alpha of
Equilibrium
unitless
0.95
Professional judgment
Sediment Partitioning - Partition
Coefficient
kg[bulk
sed/kg[inverte
brate wet wt]
0.27
Professional judgment
Sediment Partitioning - Time to
Reach Alpha of Equilibrium
day
21
Hare et al. 2001
Absorption Rate Constant
unitless
1.23
Calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
2.82E-03
Professional judgment
Benthic Omnivore Compartment Type
Assimilation efficiency from food
unitless
0.1
Professional judgment based on Yan and Wang
2002.
Absorption Rate Constant
unitless
1.23E+00
Calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
2.82E-03
Professional judgment
Benthic Carnivore Compartment Type
Assimilation efficiency from food
unitless
0.1
Professional judgment based on Yan and Wang
2002.
Absorption Rate Constant
unitless
6.60E-01
Calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
1.68E-03
Professional judgment
Water-column Herbivore Compartment Type
Assimilation efficiency from food
unitless
0.1
Professional judgment based on Yan and Wang
2002.
Assimilation efficiency from plants
0.1
Professional judgment based on Yan and Wang
2002.
Absorption Rate Constant
unitless
2.46
calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
5.02E-03
Professional judgment
Water-column Omnivore Compartment Type
Assimilation efficiency from food
unitless
0.1
Professional judgment based on Yan and Wang
2002.
Assimilation efficiency from plants
0.1
Professional judgment based on Yan and Wang
2002.
Absorption Rate Constant
unitless
1.232020679
Calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
2.82E-03
Professional judgment
Water-column Carnivore Compartment Type
Assimilation efficiency from food
unitless
0.1
Professional judgment based on Yan and Wang
2002.
Absorption Rate Constant
unitless
0.660223535
Calculated based on body weight from
regression in Hendriks & Heikens 2000
Elimination Rate Constant
unitless
1.68E-03
Professional judgment
C-1-40
-------
Exhibit 26. Mercury Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
Hq(0)
Hq(2)
MHq
Benthic Invertebrate Compartment Type
Alpha of equilibrium for sediment
partitioning
unitless
0.95
0.95
0.95
Selected value.
Benthic invertebrate-bulk sediment
partition coefficient
kg[bulk
sediment]/kg[invertebr
ate wet wt]
0.0824
0.0824
5.04
Hg(0) - assumed based on Hg(2) value;
Hg(2) and MHg - Saouter et al. 1991.
t-alpha for equilibrium for sediment
partitioning
day
14
14
14
Experiment duration from Saouter et al.
1991.
All Fish Compartments Types a
Elimination adjustment factor
unitless
3
3
1
Trudel and Rasmussen 1997.
Assimilation efficiency from food
unitless
0.04
0.04
0.2
Phillips and Gregory 1979.
Demethylation rate
1/day
N/A
N/A
0
Professional judgment.
Methylation rate
1/day
0
0
0
Professional judgment.
Oxidation rate
1/day
1.0E+06
0
0
Professional judgment.
Reduction rate
1/day
0
0
0
Professional judgment.
Water-column Herbivore Compartment Type
Assimilation efficiency from plants
unitless
1
1
1
Phillips and Gregory 1979.
Water-column Omnivore Compartment Type
Assimilation efficiency from food
unitless
1
1
1
Phillips and Gregory 1979.
a Screening scenario includes: Benthic Omnivore, Benthic Carnivore, Water-column Herbivore, Water-column Omnivore, and Water-column Carnivore.
C-1-41
-------
Exhibit 27. PAH Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
Reference
BaP | BaA BbF BkF Chr DahA IcdP
Benthic Invertebrate Compartment Type
Clearance constant
unitless
157.6
157.6
157.6
157.6
157.6
157.6
157.6
Stehly et al. 1990 / estimated for mayfly,
120-day-old nymphs.
alpha of equilibrium for
sediment partitioning
unitless
0.95
0.95
0.95
0.95
0.95
0.95
0.95
Professional judgment.
talpha for equilibrium for
sediment partitioning
days
14
14
14
14
14
14
14
Professional judgment.
Vd (ratio of concentration
in benthic invertebrates to
concentration in water)
ml/g
7235.0
7235.0
7235.0
7235.0
7235
7235
7235
Stehly et al. 1990 / estimated for mayfly,
120-day-old nymphs.
Half-life
day
1.5
1.5
1.5
1.5
1.5
1.5
1.5
Stehly et al. 1990 / calculated from
estimated elimination/depuration rate
constant estimated for mayfly, 120-day-
old nymphs.
All Fish Compartment Types3
Gamma fish
unitless
0.2
0.2
0.2
0.2
0.2
0.2
0.2
Thomann 1989.
Assimilation efficiency from
foodb
unitless
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Professional judgment.
Half-life
day
2.8
2.8
2.8
2.8
2.8
2.8
2.8
Spacie et al. 1983, as cited in MacKay
et al. 1992 (bluegill sunfish) for
benzo(a)pyrene.
Benthic Omnivore Compartment Type
Assimilation efficiency from
plants'3
unitless
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Professional judgment.
Water Column Omnivore Compartment Type
Assimilation efficiency from
plants'3
unitless
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Professional judgment.
a Screening scenario includes: Benthic Omnivore, Benthic Carnivore, Water-column Herbivore, Water-column Omnivore, and Water-column Carnivore.
bAII ingestion assimilation efficiencies set to 1 to be consistent with excretion rate calculations.
C-1-42
-------
Exhibit 28. Dioxin Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
1.2.3.4.6.7.8.
9-OCDD
1.2.3.4.6.7.8.
9-OCDF
1.2.3.4.6.7.8-
HpCDD
1.2.3.4.6.7.8-
HpCDF
1.2.3.4.7.8.9-
HpCDF
1.2.3.4.7.8-
HxCDD
1.2.3.4.7.8-
HxCDF
1.2.3.6.7.8-
HxCDD
1.2.3.6.7.8-
HxCDF
Benthic Invertebrate Compartment
Clearance constant
unitless
0
0
0
0
0
0
0
0
0
Sediment Partitioning Partition
Coefficient
kg/kg
0.0013
0.0017
0.0055
0.0012
0.042
0.033
0.0081
0.013
0.02
Sediment Partitioning Alpha of
Equilibrium
unitless
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
Sediment Partitioning Time to
Reach Alpha of Equilibrium
days
120
42
120
42
42
120
42
120
42
Vd (ratio of concentration
in benthic invertebrates to
concentration in water)
ml/g
0
0
0
0
0
0
0
0
0
Half-life
day
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
All Fish Compartmentsa
Assimilation efficiency from food
unitless
0.03
0.03
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Chemical Uptake Rate Via Gill
L[water]/kg[fish wet
wfl-day
142
N/A b
14
N/A b
N/A b
127
N/A b
127
N/A b
Gamma fish
unitless
N/A b
0.2
N/A b
0.2
0.2
N/A b
0.2
N/A b
0.2
Half-life
day
693.15
346.57
346.57
346.57
346.57
495.11
495.11
495.11
495.11
Water Column Herbivore Compartment
Assimilation efficiency from plants
unitless
0.03
0.03
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Water Column Omnivore Compartment
Assimilation efficiency from plants
unitless
0.03
0.03
0.5
0.5
0.5
0.5
0.5
0.5
0.5
a Screening scenario includes: Benthic Omnivore, Benthic Carnivore, Water-column Herbivore, Water-column Omnivore, and Water-column Carnivore.
b N/A = not applicable. This parameter is used in calculating the uptake when measured data are unavailable.
C-1-43
-------
Exhibit 28. Dioxin Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Value
1.2.3.7.8.9-
HxCDD
1.2.3.7.8.9-
HxCDF
1.2.3.7.8-
PeCDD
1.2.3.7.8-
PeCDF
2.3.4.6.7.8-
HxCDF
2.3.4.7.8-
PeCDF
2,3,7,8-
TCDD
2,3,7,8-
TCDF
Benthic Invertebrate Compartment
Clearance constant
unitless
0
0
0
0
0
0
0
0
Sediment Partitioning Partition
Coefficient
kg/kg
0.015
0.067
0.098
0.024
0.072
0.17
0.205
0.056
Sediment Partitioning Alpha of
Equilibrium
unitless
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
Sediment Partitioning Time to
Reach Alpha of Equilibrium
days
120
42
120
42
42
42
120
42
V_d (ratio of concentration in
benthic invertebrates to
concentration in water)
ml/g
0
0
0
0
0
0
0
0
Half-life
day
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
5776.2
All Fish Compartmentsa
Assimilation efficiency from food
unitless
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Chemical Uptake Rate Via Gill
L[water]/kg[fish wet
wfl-day
127
N/A b
700
N/A b
N/A b
N/A b
380
N/A b
Gamma fish
unitless
N/A b
0.2
N/A b
0.2
0.2
0.2
N/A b
0.2
Half-life
day
495.11
495.11
420.09
420.09
495.11
420.09
5251.1
5251.1
Water Column Herbivore Compartment
Assimilation efficiency from plants
unitless
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
Water Column Omnivore Compartment
Assimilation efficiency from plants
unitless
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
a Screening scenario includes: Benthic Omnivore, Benthic Carnivore, Water-column Herbivore, Water-column Omnivore, and Water-column
b N/A = not applicable. This parameter is used in calculating the uptake when measured data are unavailable.
C-1-44
-------
Exhibit 28. Dioxin Chemical-Specific Properties for Aquatic Species
Documentation for the TRIM.FaTE Screening Scenario
Parameter Name
Units
Reference
Benthic Invertebrate Compartment
Clearance constant
unitless
Professional judgment.
Sediment Partitioning Partition
Coefficient
kg/kg
TCDD data for sandworm in Rubenstein et al. 1990; dry weight sediment.
PeCDF: multiplied TCDD partition coefficient for sandworm by congener-specific
bioaccumulation equivalency factor in GLWQI from USEPA 1999.
Sediment Partitioning Alpha of
Equilibrium
unitless
Professional judgment.
Sediment Partitioning Time to
Reach Alpha of Equilibrium
days
TCDD: professional judgment; PeCDF: Rubinstein et al. 1990; data forTCDF in
sandworm.
V_d (ratio of concentration in
benthic invertebrates to
concentration in water)
ml/g
Professional judgment.
Half-life
day
Change source to f-pass
All Fish Compartmentsa
Assimilation efficiency from food
unitless
TCDD: calculated from data in Kleeman et al. 1986b trout data as cited in
USEPA 1993; PeCDF: used assimilation efficiency for TCDD in trout.
Chemical Uptake Rate Via Gill
L[water]/kg[fish wet
wfl-day
Muir et al. 1986.
Gamma fish
unitless
Thomann 1989
Half-life
day
Change source to f-pass
Water Column Herbivore Compartment
Assimilation efficiency from plants
unitless
TCDD: calculated from data in Kleeman et al. 1986b trout data as cited in
USEPA 1993; PeCDF: used assimilation efficiency for TCDD in trout.
Water Column Omnivore Compartment
Assimilation efficiency from plants
unitless
TCDD: calculated from data in Kleeman et al. 1986b trout data as cited in
USEPA 1993; PeCDF: used assimilation efficiency for TCDD in trout.
a Screening scenario includes: Benthic Omnivore, Benthic Carnivore, Water-column Herbivore, Water-column Omnivore, and Water-column Carnivore.
b N/A = not applicable. This parameter is used in calculating the uptake when measured data are unavailable.
C-1-45
-------
ATTACHMENT C-2: Description of Multimedia Ingestion Risk
Calculator (MIRC) Used for RTR Exposure and Risk Estimates
-------
[This page intentionally left blank.]
C-2-i
-------
TABLE OF CONTENTS
1 Introduction 1
1.1 Purpose and Overview 1
1.2 Scope of MIRC 1
1.3 Use in EPA's Air Toxics Program 2
1.4 MIRC Highlights 3
1.5 Organization of This Document 3
2 MIRC Overview 5
2.1 Software 5
2.2 Exposure Pathways 7
2.3 Receptor Groups 8
3 Exposure Algorithms 11
3.1 Farm Food Chain Algorithms 11
3.1.1 Estimating Chemical Concentrations in Produce 11
3.1.2 Estimating Chemical Concentrations in Animal Products 18
3.2 Chemical Intake Calculations for Adults and Non-Infant Children 22
3.2.1 Chemical Intake from Soil Ingestion 24
3.2.2 Chemical Intake from Fish Ingestion 24
3.2.3 Chemical Intake from Fruit Ingestion 26
3.2.4 Chemical Intake from Vegetable Ingestion 26
3.2.5 Chemical Intake from Animal Product Ingestion 28
3.2.6 Chemical Intake from Drinking Water Ingestion 30
3.3 Total Chemical Intake 30
3.4 Chemical Intake Calculations for Nursing Infants 32
3.4.1 Infant Average Daily Absorbed Dose 32
3.4.2 Chemical Concentration in Breast Milk Fat 33
3.4.3 Chemical Concentration in Aqueous Phase of Breast Milk 37
3.4.4 Alternative Model for Infant Intake of Methyl Mercury 39
4 Dose-Response Values Used for Assessment 41
5 Risk Characterization 46
5.1 Cancer Risks 46
5.2 Non-cancer Hazard Quotients 48
5.2.1 Hazard Quotients for Chemicals with a Chronic RfD 48
5.2.2 Hazard Quotients for Chemicals with RfD Based on Developmental Effects 48
5.2.3 Hazard Index for Chemicals with RfDs 48
6 Model Input Options 50
6.1 Environmental Concentrations 50
C-2-ii
-------
6.2 Farm-Food-Chain Parameter Values 51
6.2.1 List of Farm-Food-Chain (FFC) Parameters 51
6.2.2 Produce Parameter Values 52
6.2.3 Animal Product Parameter Values 59
6.3 Adult and Non-Infant Exposure Parameter Values 61
6.3.1 Body Weights 61
6.3.2 Water Ingestion Rates 62
6.3.3 Local Food Ingestion Rates 63
6.3.4 Local Fish Ingestion Rates 67
6.3.5 Soil Ingestion Rates 71
6.3.6 Total Food Ingestion Rates 72
6.4 Other Exposure Factor Values 73
6.4.1 Exposure Frequency 73
6.4.2 Fraction Contaminated 74
6.4.3 Preparation and Cooking Losses 74
6.5 Breast-Milk Infant Exposure Pathway Parameter Values 76
6.5.1 Receptor-specific Parameters 76
6.5.2 Chemical-Specific Parameter Values 80
7 Summary of MIRC Default Exposure Parameter Settings 84
7.1 Default Ingestion Rates 84
7.2 Default Screening-Level Population-Specific Parameter Values 86
7.3 Default Chemical-Specific Parameter Values for Screening Analysis 86
7.4 Screening-Level Parameter Values for Nursing Infant Exposure 88
7.4.1 Dioxins 88
7.4.2 Methyl Mercury 89
8 REFERENCES 90
C-2-iii
-------
LIST OF EXHIBITS
Exhibit 2-1. Overview of Access-based MIRC Software Application for Performing Farm-Food-Chain
Ingestion Exposure and Risk Calculations 6
Exhibit 2-2. Transfer Pathways for Modeled Farm Food Chain (FFC) Media 8
Exhibit 3-1. Chemical Transfer Pathways for Produce 12
Exhibit 3-2. Estimating Chemical Concentration in Aboveground Produce 12
Exhibit 3-3. Chemical Transfer Pathways for Animal Products 19
Exhibit 4-1. Dose-response Values for Chemicals Addressed by the Screening Scenario 44
Exhibit 6-1. MIRC Parameters Used to Estimate Chemical Concentrations in Farm Foods 51
Exhibit 6-2. Chemical-Specific Inputs for Produce Parameters 53
Exhibit 6-3. Chemical-Specific Inputs by Plant Type for Chemicals Included in MIRC ... Error! Bookmark
not defined.
Exhibit 6-4. Non-Chemical-Specific Produce Inputs Error! Bookmark not defined.
Exhibit 6-5. Animal Product Chemical-specific Inputs for Chemicals Included in MIRC 59
Exhibit 6-6. Soil and Plant Ingestion Rates for Animals 60
Exhibit 6-7. Mean and Percentile Body Weight Estimates for Adults and Children 62
Exhibit 6-8. Estimated Daily Per capita Mean and Percentile Water Ingestion Rates for Children and
Adults 63
Exhibit 6-9. Summary of Age-Group-Specific Food Ingestion Rates for Farm Food Items 64
Exhibit 6-10. Daily Mean and Percentile Per Capita Fish Ingestion Rates..Error! Bookmark not defined.
Exhibit 6-11. Daily Mean and Percentile Soil Ingestion Rates for Children and Adults 72
Exhibit 6-12. Daily Mean and Percentile Per Capita Total Food Intake 73
Exhibit 6-13. Fraction Weight Losses from Preparation of Various Foods 75
Exhibit 6-14. Scenario- and Receptor-Specific Input Parameter Values Used to Estimate Infant
Exposures via Breast Milk 76
Exhibit 6-15. Average Body Weight for Infants 77
Exhibit 6-16. Time-weighted Average Body Weight for Mothers 78
Exhibit 6-17. Infant Breast Milk Intake Rates 79
Exhibit 6-18. Chemical-specific Input Parameter Values for Breast Milk Exposure Pathway 81
Exhibit 7-1. Farm Food Category Ingestion Rates for Conservative Screening Scenario for Farming
Flouseholds 85
Exhibit 7-2. Mean Body Weight Estimates for Adults and Children 86
Exhibit 7-3. Chemical-Specific Parameter Values for Input to MIRC 87
Exhibit 7-4. Chemical and Animal-Type Specific Biotransfer Factor (Ba) Values for Input to MIRC 88
C-2-iv
-------
LIST OF EQUATIONS
Equation 3-1. Chemical Concentration in Aboveground Produce 13
Equation 3-2. Chemical Concentration in Aboveground Produce Due to Root Uptake 13
Equation 3-3. Chemical Concentration in Aboveground Produce Due to Deposition of Particle-phase
Chemical 14
Equation 3-4. Chemical Concentration in Aboveground Produce Due to 15
Equation 3-5. Conversion of Aboveground Produce Chemical Concentration from 16
Equation 3-6. Chemical Concentration in Belowground Produce: Nonionic Organic Chemicals 17
Equation 3-7. Chemical Concentration in Belowground Produce: Inorganic Chemicals 18
Equation 3-8. Conversion of Belowground Produce Chemical Concentration from 18
Equation 3-9. Chemical Concentration in Beef, Pork, or Total Dairy 19
Equation 3-10. Chemical Concentration in Poultry or Eggs 20
Equation 3-11. Incidental Ingestion of Chemical in Soil by Livestock 20
Equation 3-12. Ingestion of Chemical in Feed by Livestock 21
Equation 3-13. Chemical Concentration in Lifestock Feed (All Aboveground) 21
Equation 3-14. Chemical Concentration in Livestock Feed Due to Root Uptake 22
Equation 3-15. Average Daily Dose for Specified Age Group and Food Type 22
Equation 3-16. Chemical Intake from Soil Ingestion 24
Equation 3-17. Chemical Intake from Fish Ingestion 25
Equation 3-18. Consumption-weighted Chemical Concentration in Fish 25
Equation 3-19. Chemical Intake from Consumption of Exposed Fruits 26
Equation 3-20. Chemical Intake from Consumption of Protected Fruits 26
Equation 3-21. Chemical Intake from Exposed Vegetables 27
Equation 3-22. Chemical Intake from Protected Vegetables 27
Equation 3-23. Chemical Intake from Root Vegetables 27
Equation 3-24. Chemical Intake from Ingestion of Beef 28
Equation 3-25. Chemical Intake from Dairy Ingestion 28
Equation 3-26. Chemical Intake from Pork Ingestion 29
Equation 3-27. Chemical Intake from Poultry Ingestion 29
Equation 3-28. Chemical Intake from Egg Ingestion 30
Equation 3-29. Chemical Intake from Drinking Water Ingestion 30
Equations 3-30 to 3-35. Total Average Daily Dose of a Chemical for Different Age Groups 31
Equation 3-36. Lifetime Average Daily Dose (LADD) 31
Equation 3-37. Average Daily Dose of Chemical to the Nursing Infant 33
Equation 3-38. Chemical Concentration in Breast Milk Fat 34
Equation 3-39. Daily Maternal Absorbed Intake 35
C-2-v
-------
Equation 3-40. Biological Elimination Rate Constant for Chemicals for Non-lactating Women 36
Equation 3-41. Biological Elimination Constant for Lipophilic Chemicals for Lactating Women 36
Equation 3-42. Chemical Concentration in Aqueous Phase of Breast Milk 37
Equation 3-43. Fraction of Total Chemical in Body in the Blood Plasma Compartment 38
Equation 3-44. Biological Elimination Rate Constant for Hydrophilic Chemicals 39
Equation 3-45. Calculation of Infant Average Daily Absorbed Dose of Methyl Mercury 40
Equation 5-1. Calculation of Excess Lifetime Cancer Risk 46
Equations 5-2 to 5-8. Lifetime Cancer Risk: Chemicals with a Mutagenic MOA for Cancer 47
Equation 5-9. Hazard Quotient for Chemicals with a Chronic RfD 48
Equation 5-10. Hazard Index Calculation 49
Equation 6-1. Calculation of Age-Group-Specific and Food-Specific Ingestion Rates 67
Equation 6-2. Calculation of Alternative Age-Group-Specific Fish Ingestion Rates 69
C-2-vi
-------
1 Introduction
1.1 Purpose and Overview
This document provides a detailed description of the Multimedia Ingestion Risk Calculator
(MIRC), an Access-based tool and database designed to assist in estimating risks via multiple
ingestion pathways, particularly for food products grown or raised at home or on a farm. MIRC
was designed to estimate risks to humans from ingestion of produce or animal products, fish,
and water in the vicinity of a source of chemical emissions to air. The user can evaluate either
generalized (e.g., conservative default) or more site-specific scenarios using the same tool.
MIRC includes a database of exposure parameter values, offering the user the option of
selecting mean, median, and upper percentile values for many parameters, data permitting.
Generally conservative default values were assigned to each parameter in the tool and the
default configuration is used for initial risk screening efforts by EPA's Office of Air Quality
Planning and Standards' (OAQPS) for Risk and Technology Review (RTR) multimedia risk
assessments (the default inputs were used to calculate the de minimis screening thresholds).
MIRC also allows the user to define the farm food chain (FFC) parameter values and receptor
characteristics to better represent a site-specific scenario.
With user-input concentrations for one or more chemicals in air and soil and air-to-surface
deposition rates, MIRC calculates the chemical's concentrations in home- or farm-grown
produce and animal food products using FFC algorithms adapted from EPA's Human Health
Risk Assessment Protocol for Hazardous Waste Combustion Facilities (hereafter referred to as
HHRAP; EPA 2005a). MIRC uses these calculated concentrations, along with user-input
chemical concentrations for fish and drinking water, to estimate chemical intake rates, as
average daily doses (ADDs), for adults, children, and nursing infants. Users can obtain
chemical input concentrations and deposition rates from measurements at an actual site or from
a transport and fate model, such as TRIM.FaTE as is done for RTR risk assessment.
For a specified set of chemical concentrations and MIRC parameter options, MIRC calculates
ADDs separately for adults, four age groups of children, and infants to reflect differences in food
ingestion rates and diet at different lifestages. MIRC estimates age-specific hazard quotients
(HQs) as the ratio of age-specific ADDs to the reference dose (RfD) for a chemical. The most
appropriate HQ for a chemical depends on its toxic mode of action and the duration of exposure
required to produce an effect. MIRC also estimates average lifetime ADDs and compares those
to cancer slope factors (CSFs) to estimate cancer risks. A breast milk ingestion pathway is
included to estimate exposure and risks to nursing infants.
MIRC was developed to be a flexible, transparent application using Microsoft Access software.
The tool includes chemical transfer and ingestion exposure algorithms and a database of
parameter values, many with several options, used by these equations. The MIRC database
includes values for the relevant physiochemical properties and toxicity reference values for
more than 500 chemicals, including approximately 60 inorganics taken primarily from a
database developed for HHRAP (EPA 2005a). Although designed for OAQPS' RTR
assessments for sources of hazardous air pollutants (HAPs), the tool is flexible in its design and
can be used to assess risks in many other contexts where soil and air concentrations are
predicted or measured.
1.2 Scope of MIRC
For persistent and bioaccumulative (PB) chemicals, risks from direct inhalation of the chemical
can be much less than risks from ingestion of the chemical in water, fish, and food products
C-2-1
-------
grown in an area of chemical deposition. Vegetables and fruits in such areas can become
contaminated directly by deposition of the airborne chemical to foliage, fruits, and vegetables or
indirectly by root uptake of the chemical deposited to soils. Livestock can be exposed to the PB
chemicals via ingestion of contaminated forage and incidental ingestion of contaminated soils.
For PB chemicals, evaluation of the inhalation pathway for air pollutants may reveal only a
portion of the risk to individuals in such populations. Households that consume high quantities
of self-caught fish or locally grown produce and animal products may be particularly susceptible
to ingestion of chemicals transferred from air in the vicinity of an air emissions source. For PB
chemicals in particular, therefore, EPA developed methods of estimating risk from indirect
exposure pathways associated with the deposition of airborne chemicals to gardens and farms,
as described in HHRAP (EPA 2005a).
MIRC is an Access-based tool that facilitates calculation of risks associated with the indirect
ingestion exposure pathways for persons consuming produce and animal products grown in an
air depositional area of concern. The tool uses algorithms described in or adapted from HHRAP
to calculate exposures from the produce and animal products. Included with MIRC is a fish
ingestion pathway and drinking water ingestion pathway for scenarios in which those pathways
may be important. MIRC also includes a breast milk ingestion pathway for nursing infants
based primarily on EPA's Methodology for Assessing Health Risks Associated with Multiple
Pathways of Exposure to Combustor Emissions (hereafter MPE; EPA 1998).
1.3 Use in EPA's Air Toxics Program
MIRC was designed to help predict human health risks from PB HAPs for EPA's RTR
assessments. EPA evaluates the fate of HAP releases to air from source categories after
implementation of technology-based Maximum Achievable Control Technology (MACT)
standards. For volatile chemicals that do not partition to other environmental media and for
non-persistent chemicals that degrade relatively quickly in the environment, evaluation of health
risks from direct inhalation of the chemical in air can provide reasonable estimates of total risk.
For PB-HAPs, however, indirect exposure pathways, such as ingestion, might contribute more
to total risk than the inhalation pathway. EPA therefore developed several computer software
tools to assist in evaluating exposure and risk from non-inhalation pathways. EPA developed
the Total Risk Integrated Methodology (TRIM) Environmental Fate, Transport, and Ecological
Exposure (TRIM.FaTE) computer program to simulate the release, transport, and fate of HAPs
from a specific source throughout the area in which local (non-source) chemical deposition is
likely to be a concern. TRIM.FaTE models the transport of individual chemicals from the source
through air by advection (wind) of particle- and vapor-phase chemical and deposition of the
chemical from air to terrestrial and aquatic ecosystems by wet and dry deposition. Movement of
the chemical through a watershed via erosion and runoff, uptake by plants, and other abiotic
and biotic transfer processes also are simulated. For the chemical that reaches surface waters,
TRIM.FaTE models uptake and bioaccumulation to trophic level (TL) 3 and 4 fish (i.e., pan fish
and game fish, respectively).
MIRC was developed to process TRIM.FaTE results, in particular, air deposition rates and the
concentrations of a chemical, after a specified duration of emissions, in several spatially explicit
environmental compartments, including air, surface and root-zone soils, surface and ground
waters, and fish. MIRC uses those results to calculate exposure to the chemical through
ingestion of locally grown foods, including various types of fruits and vegetables, poultry, swine,
and dairy (and beef) cattle. MIRC also calculates the associated risks for individuals who
consume those foods. MIRC was designed to use specific TRIM.FaTE results to estimate FFC
C-2-2
-------
concentrations, ingestion exposures, and human health risks for OAQPS' RTR assessments. It
uses the same approach that OAQPS intends to implement directly in its TRIM system via three
modules beyond TRIM.FaTE: TRIM Farm Food Chain, TRIM.Expomgestion, and TRIM.Risk.1
1.4 MIRC Highlights
Although designed to assist EPA OAQPS in its RTR assessments, MIRC is a stand-alone
software application that can be used in other contexts. A user can supply either measured or
estimated chemical concentrations for soil, air, water, and fish and air deposition rates likely for
the location(s) of interest based on local meteorology. The user can accept the default values
for many exposure parameters and screen for small possibilities of risk, or the user can select
other options or overwrite parameter values to tailor the estimates to a specific scenario or
location.
MIRC complies with EPA's latest guidelines for exposure and risk assessment, including
HHRAP; the Agency's 2005 Guidelines for Carcinogen Risk Assessment (Cancer Guidelines),
Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
(Supplemental Guidance), and Guidance on Selecting Age Groups for Monitoring and
Assessing Childhood Exposures to Environmental Contaminants (EPA 2005b,c,d); and its 2008
Child-Specific Exposure Factors Handbook (EPA 2008a). In particular, MIRC provides several
important capabilities:
• When provided air and soil concentrations, the MIRC software package allows rapid
calculation of screening-level exposures and risks associated with household
consumption of locally grown/raised foods.
• MIRC can calculate exposures and risks associated with incidental ingestion of surface
soils, fish consumption, and drinking water.
• The tool calculates ADDs (i.e., chemical intake rates) for six "built-in" age groups to allow
use of age-group-specific body weights, ingestion rates, food preferences, and
susceptibility to toxic effects.
• Its database of chemical information covers plant- and animal-specific transfer factors
and other inputs that determine concentrations in farm food stuffs.
• Value options for receptor characteristics in the database include the mean and 50th,
90th, 95th, and 99th percentile values where data permit.
• For carcinogens with a mutagenic mode of action, MIRC estimates a lifetime ADD using
the three lifestages and potency adjustment factors recommended in EPA's 2005
Cancer Guidelines and Supplemental Guidance.
• The data for exposure parameters in the tool have been updated to include the latest
recommended values for children issued September 30, 2008, in the Agency's Child-
Specific Exposure Factors Handbook (CSEFH) (EPA 2008a).
1.5 Organization of This Document
Sections 2 through 5 of this document describe the exposure and risk models implemented in
MIRC. Section 2 provides an overview of the FFC exposure scenario and indicates options
1 General information about the TRIM system is available at http://www.epa.gov/ttn/fera/trim_gen.html.
C-2-3
-------
available to a user to tailor the scenario to specific applications. Section 3 describes the
exposure algorithms used in MIRC, including how ADDs are calculated. Section 4 presents the
toxicity reference values included in MIRC to calculate risks. Section 5 describes the risk
characterization algorithms in MIRC. Section 6 of this document describes data input options
for the model. Section 7 describes the default parameterization of MIRC for application to
conservative risk screening assessments, and Section 8 provides the references. Appendix A
provides guidance to users on how to set up and run MIRC for their own applications.
Note that the default parameterization described in Section 7 was used to estimate de minimis
releases of PB-HAPs from facilities assumed to pose negligible risk to subsistence communities
in the vicinity of a facility emitting the HAPs to air. Users of MIRC can modify the default values
for many of the parameters to better represent a specific exposure scenario.
C-2-4
-------
2 MIRC Overview
The Multimedia Ingestion Risk Calculator (MIRC) software package is designed to allow rapid
calculation of screening-level exposures and risks associated with a subsistence farmer/fisher
population in the vicinity of a source of chemical emissions to air. The tool allows a user to
assess human exposures via ingestion pathways, including drinking water consumption,
incidental soil ingestion, fish ingestion, and ingestion of ten types of farm food chain (FFC)
products: exposed fruits, protected fruits, exposed vegetables, protected vegetables, root
vegetables, beef, total dairy, pork, poultry, and eggs. The tool also includes a breast milk
ingestion and risk module for nursing infants. For fruits and vegetables, the terms "exposed"
and "protected" refer to whether the edible portion of the plant is exposed to the atmosphere.
The remainder of this overview consists of three sections. The first (Section 2.1) provides an
overview of the MIRC software package. The second and third sections summarize the
ingestion exposure pathways included in the tool and the "built-in" receptor age categories,
respectively (Sections 2.2 and 2.3).
2.1 Software
The MIRC application includes the following components:
• A graphical user interface through which the user locates and accesses various input
and output tables.
• Input tables in which the user can enter environmental concentrations of a chemical
estimated for air, soil, drinking water, and fish tissue.
• Internal chemical transfer and exposure algorithms and database of options for FFC
algorithm parameter value, chemical-specific inputs, and exposure factors.
• Tabulated outputs of calculated chemical concentrations in the various farm food
products (e.g., fruits, vegetables, beef, eggs) and ADDs for those foods and for water
and fish ingestion for each receptor category.
• Output tables with estimated cancer risks and non-cancer hazard estimates associated
with total ingestion exposure to each chemical for each receptor category.
Exhibit 2-1 provides a flowchart displaying the types of inputs required or optional and general
flow of calculations carried out by the tool.
C-2-5
-------
Exhibit 2-1. Overview of Access-based MIRC Software Application for
Performing Farm-Food-Chain Ingestion Exposure and Risk Calculations
User Selects Chemical(s) of Concern: For Each, MIRC Calculates Risks
User Specifies Environmental Concentrations for Chemical of Concern
:oot-zone and Surface Soils
(jish^ CPrinking Water^
Uptake by foliage / Uptake by roots:
Vegetables, l-ruits Grains Hay, (Jrass *
Animal Products -
Farm Food Chain Biotransfer Calculations
User Selects Receptor Characteristics
<^ody Weight)
From Options or Over-write
.^"Tlome Grown
Food Product
^^Jnqestion Rat'
Average Daily Dose (ADD) for Age Group y; y =1 to 5
X
X
Chemical Intake with Food/
Medium Type i; i = 1 to 10
Fish and Water
Ingestion Rates
Exposure Module
Lifetime Average
Daily Dose
Adult ADD x
absorption efficiency
User Option to Add Breast Milk Pathway
©
User Selects BMP
Parameter Values
Duration Breast Feeding; maternal
>and infant characteristics
Chemical
Toxicity
Reference
Values:
SF and RfDs
Risk Characterization Module
Age-specific & Lifetime Exposure Doses
Lifetime Cancer Risks
Age-Specific Flazard Quotients
Breast Milk Exposure
Module
ADD Maternal
[C] in milk
Infant Dose
3
C-2-6
-------
An Access form within the graphical user interface enables the user to construct specific
scenarios by choosing ingestion sources, receptor ages, and other input choices (e.g., diet
composition, body weight percentiles). This feature facilitates the analysis of various exposure
scenarios. To begin an analysis, the user must supply values for the following chemical-specific
parameters for the scenario being evaluated:
• Air concentration of total chemical,
• Fraction chemical in air in vapor-phase,
• Wet and dry deposition rates for particle-phase chemical,
• Drinking water concentration,
• Chemical concentration in surface soils (two locations; can be tilled and unfilled),
• Chemical concentration in root-zone soils (two locations; can be tilled and unfilled), and
• Chemical concentrations in pan fish and in game fish.
Users can input measured values or values estimated by TRIM.FaTE or other models for these
parameters.
The MIRC application uses the input data and a variety of empirical transfer factor values
(included in its database) to estimate chemical concentrations in nine categories of FFC food
types (Section 2.2). The FFC algorithms and transfer factor values included in MIRC are based
on those presented in Chapter 5 of EPA's Human Health Risk Assessment Protocol for
Hazardous Waste Combustion Facilities, hereafter referred to as HHRAP (EPA 2005a).
For outputs, MIRC is designed to calculate individual cancer risk and non-cancer hazard
quotients for one chemical at a time. It is up to the risk assessor to determine if cancer risks or
hazard quotients may be additive across two or more chemicals (i.e., if they cause toxic effects
in the same target organ by the same mode of action, such as multiple PAHs that are
carcinogenic by a mutagenic mode of action).
The tool assumes that the same individuals (farming family or household that gardens and
raises animals) are exposed via all of the pathways specified (i.e., pathways with non-zero
ingestion rates). The tool therefore is useful in estimating risk to the maximally exposed
individuals (MEI) in a risk assessment. To evaluate multiple populations, the user must specify
the full exposure scenario for each population separately.
2.2 Exposure Pathways
MIRC estimates the concentrations of chemicals in FFC food categories grown in an area of
airborne chemical deposition using algorithms and parameter values provided in HHRAP (EPA
2005a). FFC foods are evaluated in ten categories: exposed fruit, protected fruit, exposed
vegetables, protected vegetables, root vegetables, beef, total dairy, pork, poultry, and eggs.
Exhibit 2-2 summarizes the pathways by which chemicals are transferred to these food media.
Note that for a general screening-level assessment, all of the pathways can be modeled, as is
the case for EPA's Risk and Technology Review (RTR) calculation of de minimis emission rates
for persistent and bioaccumulative hazardous air pollutants (PB-HAPs) (EPA 2008b).
C-2-7
-------
Exhibit 2-2. Transfer Pathways for Modeled Farm Food Chain (FFC) Media
Farm Food Media
Chemical Transfer Pathways
Exposed fruit and vegetables
• Direct deposition from air of particle-bound chemical
• Air-to-plant transfer of vapor phase chemical
• Root uptake from soil
Protected fruit and vegetables
(including root vegetables)
• Root uptake from soil
Beef and total dairy
(including milk)
• Ingestion of forage, silage, and grain a
• Soil ingestion
Pork
• Ingestion of silage and grain a
• Soil ingestion
Poultry and eggs
• Ingestion of grain a
• Soil ingestion
a Chemical concentrations in forage, silage, and grain are estimated via intermediate calculations
analogous to those used for aboveground produce.
Produce types included in the FFC can accumulate chemical directly from air and/or soil. For
exposed produce, chemical mass is assumed to be transferred to plants from the air in two
ways. First, particle-bound chemical can deposit directly on the plant surface. Second, the
uptake of vapor-phase chemicals by plants through their foliage can occur. For both exposed
and protected produce, the concentration in the plant derived from exposure to the chemical in
soil is estimated using an empirical bioconcentration factor (BCF) that relates the concentration
in the plant to the concentration present in the soil. For belowground root vegetables, a root
concentration factor is applied. The algorithms used to estimate produce concentrations are
presented in Section 3.1.1.
Chemical concentrations in animal products are estimated based on the amount of chemical
consumed through the diet, including incidental ingestion of soil while grazing. The diet options
for farm animals in MIRC include forage (plants grown on-site for animal grazing, such as
grass), silage (wet forage grasses, fresh-cut hay, or other fresh plant material that has been
stored and fermented), and feed grain products grown on the farm (e.g., corn, soybeans). All
three animal feed products are assumed to accumulate chemical via root uptake from the soil.
Forage and silage also can accumulate chemical via direct deposition of particle-bound
chemical and vapor transfer.
The algorithms in MIRC are based on the assumptions that beef and dairy cattle consume all
three feed products, while pigs consume only silage and grain and chickens consume only
grain. The incidental ingestion of the chemical in soils during grazing or consumption of foods
placed on the ground is estimated using empirical soil ingestion values. For secondary animal
products (dairy products and eggs), chemical concentrations are estimated by applying a
biotransfer factor to the estimated concentration in the "source" animal (cows and chickens,
respectively). The algorithms used to estimate animal product concentrations are described in
Section 3.1.2.
2.3 Receptor Groups
As noted in EPA risk assessment guidelines (EPA 2005b,c,d, 2008a), exposures of children are
expected to differ from exposures of adults due to differences in body weights, ingestion rates,
dietary preferences, and other factors. It is important, therefore, to evaluate the contribution of
exposures during childhood to total lifetime risk using appropriate exposure factor values.
C-2-8
-------
EPA's HHRAP (Chapter 4, EPA 2005a) recommends assessing exposures for children and
adults separately, but considers all non-infant children in one category. Specifically, HHRAP
recommends eight categories of receptor: farmer, child farmer, resident, child resident, fisher,
child fisher, acute receptor, and nursing infant. Over time, different EPA programs have used
different child age groupings to evaluate body weights, ingestion rates, and other parameter
values needed to estimate chemical exposures and risks to children.
To improve the match between age groups used to estimate values across exposure
parameters, in 2005, EPA recommended a standard set of child age categories for exposure
and risk assessments (EPA 2005b). EPA recommended four age groups for infants: birth to < 1
month; 1 to < 3 months; 3 to < 6 months; and 6 to < 12 months. For young children, EPA
recommended an additional four age groups: 1 to < 2 years; 2 to < 3 years; 3 to < 6 years; and
6 to < 11 years. Two age groupings were recommended for teenagers and young adults: 11 to
< 16 years; and 16 to < 21 years. These age groupings correspond to different developmental
stages and reflect different food ingestion rates per unit body weight, with the highest ingestion
rates occurring for the youngest, most rapidly growing, age groups.
For assessment of cancer risks from early-life exposure, EPA recognizes that infants and
children may be more sensitive to a carcinogenic chemical than adults, with cancers appearing
earlier in life or with lower doses experienced during childhood (EPA 2005c,d). Thus, the
"potency" of a carcinogen might be higher for infants and children than for adults. To date,
however, data by which to evaluate the relative sensitivity of children and adults to the same
daily dose of a carcinogen remain limited. Based on analyses of radioactive and other
carcinogenic chemicals, EPA recommends evaluating two lifestages for children separately from
adults for chemicals that cause cancer by a mutagenic mode of action (MOA): from birth to < 2
years and from 2 to < 16 years (EPA 2005c,d). EPA also suggests that, as data become
available regarding carcinogens with a mutagenic MOA, further refinements of these age
groupings may be considered.
For purposes of RTR assessment using MIRC, the selection of age categories is limited by the
categories for which most of the FFC food ingestion rates have been calculated. In Chapter 13
of both its Exposure Factors Handbook (EFH; EPA 1997a) and its Child-Specific Exposure
Factors Handbook (CSEFH; EPA 2008a), EPA summarized home-grown/raised food ingestion
rates for four children's age groups: 1 to < 3 years; 3 to < 6 years; 6 to < 12 years; and 12 to <
20 years. Intake rates were not calculated for children younger than 1 year because infants are
unlikely to consume those foods. They are more likely to be nursing or to be fed formula and
other commercial baby-food products.
Although the age groupings used to estimate FFC ingestion rates do not match precisely the
groupings that EPA recommended in 2005 for Agency exposure assessments (EPA 2005b),
they are the only age-groupings for which such data are available. The U.S. Department of
Agriculture's (USDA's) 1987-1988 Nationwide Food Consumption Survey (USDA 1992, 1993,
1994) remains the most recent survey of ingestion rates for home-grown foods, and EPA's
analysis of those data, published in its 1997 EFH, remains the most recently published major
analysis of those data. Because ingestion of home-grown produce and animal products are the
primary exposure pathways for which MIRC was developed, those are the age groupings used
for all child parameter values used to estimate exposure and risk in MIRC.
C-2-9
-------
Thus, in MIRC, values for each exposure parameter were estimated for adults (20 to 70 years)
and five children's age groups:
• infants under 1 year (i.e., 0 to < 1 year);
• children ages 1 through 2 years (i.e., 1 to < 3 years);
• children ages 3 through 5 years (i.e., 3 to < 6 years);
• children ages 6 through 11 years(i.e., 6 to < 12 years) and
• children ages 12 through 19 years (i.e., 12 to < 20 years).
Exposure and risks to infants under 1 year of age are estimated only for the breast-milk-
ingestion pathway.
For assessing risks from exposures to carcinogenic chemicals that act via a mutagenic MOA,
the two early lifestages recommended by EPA (EPA 2005c,d) also are included in MIRC:
• children under the age of 2 years (i.e., 0 to < 2 years); and
• children from 2 through 15 years (i.e., 2 to < 16 years).
Different age groupings are needed for the assessment of risks from carcinogenic chemicals
with a mutagenic MOA and other carcinogens with other or unknown MOAs. Currently in MIRC,
the only PB-HAPs with a mutagenic mode of carcinogenesis are some of the PAHs.
C-2-10
-------
3 Exposure Algorithms
The exposure algorithms in MIRC are described below in four sections. Section 3.1 presents
the algorithms used to estimate chemical concentrations in FFC foods from chemical
concentrations in soil and air. Pathway-specific algorithms used to estimate chemical intakes by
adults and non-infant children are described in Section 3.2, and total chemical intake
calculations are described in Section 3.3. Finally, the sets of algorithms used to estimate
chemical intake via consumption of breast milk by nursing infants are described in Section 3.4.
As noted previously, the exposure algorithms used in MIRC are based on those presented in
HHRAP (EPA 2005b). Any differences between MIRC and HHRAP are explained in this
section.
3.1 Farm Food Chain Algorithms
The algorithms and parameters used to estimate chemical concentrations in produce and
animal products are described in Sections 3.1.1 and 3.1.2, respectively. Discussions of the
parameter value options and the values selected as defaults in MIRC for RTR risk assessment
are provided in Section 6.2. The use of TRIM.FaTE to model chemical fate and transport in the
environment prior to FFC calculations drives the most significant difference between the FFC
algorithms included in HHRAP and the equations used for RTR. The approach in HHRAP uses
estimated ambient air concentrations and deposition rates from dispersion model simulations
that use unitized emission rates. Chemical-specific emission rates (adjusted for vapor and
particle-bound fractions) are then incorporated into some of the HHRAP FFC algorithms to
calculate concentrations in FFC media. Soil concentrations are calculated using a similar
approach in HHRAP. For assessment of multipathway exposures for RTR, TRIM.FaTE is used
to estimate air concentrations, air-to-surface deposition rates, and soil concentrations, and
these outputs are used in the FFC algorithms.
3.1.1 Estimating Chemical Concentrations in Produce
Produce (vegetables and fruits) can become contaminated directly by deposition of airborne
chemicals to foliage and fruits or indirectly by uptake of chemicals deposited to the soil. Given
these two contamination processes, produce is divided into two main groups: aboveground and
belowground produce. Aboveground produce is divided into fruits and vegetables. These
groups are further subdivided into "exposed" and "protected" depending on whether the edible
portion of the plant is exposed to the atmosphere or is protected by a husk, hull, or other outer
covering.
Exhibit 3-1 lists the pathways by which chemicals are transferred to the FFC produce
categories. Note that for a general screening-level assessment, all of the pathways can be
modeled, as was done for EPA's calculation of de minimis emission rates for PB-HAPs in its
RTR assessments (EPA 2008b), and as described in the "Technical Support Document for TRIM-
Based Multipathway Screening Scenario for RTR". Sections 3.1.1.1 and 3.1.1.1 describe the
transfer pathways and algorithms for aboveground and belowground produce, respectively.
C-2-11
-------
Exhibit 3-1. Chemical Transfer Pathways for Produce
Farm Food Media
Chemical Transfer Pathways
Aboveground
Produce
Exposed fruits and
vegetables
• Direct deposition from air of
particle-bound chemical
• Air-to-plant transfer of vapor
phase chemical
• Root uptake from soil
Protected fruits and
vegetables
• Root uptake from soil
Belowground
Produce
Root vegetables
• Root uptake from soil
Exhibit 3-2. Estimating Chemical
Concentration in Aboveground Produce
¦#
Deposition
of Particles
(Pd)
Vapor
Transfer
(Pv)
Root Uptake
from Soil
(PrAG-produce)
Aboveground Produce
For aboveground exposed produce,
chemical mass is assumed to be transferred
to plants from the air in three ways, as
illustrated in Exhibit 3-2. First, particle-
bound chemical can deposit directly on the
plant surface via deposition (Pd). The
amount of chemical accumulated is
estimated based on the areal fraction of
chemical deposition intercepted by the plant
surface, minus a loss factor that is intended
to account for removal of deposited
chemical by wind and rain and changes in
concentration due to growth dilution.
Second, for chemical present in air in the
vapor phase, the concentration of chemical
accumulated by the plant's foliage is
estimated using an empirical air-to-plant
biotransfer factor (Pv). Third, the chemical
concentration in the plant due to root uptake from the soil (PrAG.pr0duce) is estimated using an
empirical bioconcentration factor (BrAG.produce) that relates the chemical concentration in the plant
to the average chemical concentration in the soil at the root-zone depth in the produce-growing
area ( Csroot-zone_produce) ¦
The edible portions of aboveground protected produce are not subject to contamination via
particle deposition (Pd) or vapor transfer (Pv). Therefore, root uptake of chemicals is the
primary mechanism through which aboveground protected produce becomes contaminated.
The chemical concentration in the aboveground plant due to root uptake from soil (PrAG.pr0duce-
DW) is estimated using an empirical bioconcentration factor (BrAG.prodUce-Dw) that relates the
chemical concentration in the plant to the average chemical concentration in the soil at the root-
zone depth in the produce-growing area (Csr0ot-zone_produce) ¦
Chemical Concentration in
Aboveground Produce
C-2-12
-------
Equation 3-1. Chemical Concentration in Aboveground Produce
CAG-produce-DW(i) ~ ^^AG-produce-DW(i)^^^ (i) ^^(i)
where:
_ _ Concentration of chemical in edible portion of aboveground produce type /',
AG-produce-Dwo) - eXp0sec| or protected, on a dry-weight (DW) basis (mg/kg produce DW)
Chemical concentration in edible portion of aboveground produce type /' due to
Pd(i) = deposition of particles (mg/kg produce DW); for protected aboveground
produce, Pd equals zero
Chemical concentration in edible portion of aboveground produce type /',
Pi~AG-Produce-DW(i) = exposed or protected, due to root uptake from soil at the root-zone depth of the
produce growing area (mg/kg produce DW)
Chemical concentration in edible portion of aboveground produce type /' due to
Pv0) = air-to-plant transfer (|jg/g [or mg/kg] produce DW); for protected aboveground
produce, Pv equals zero
Equation 3-2. Chemical Concentration in Aboveground Produce Due to Root Uptake
^^AG-produce-DW(i) ~ ^^root-zone_produce x ®^AG-produce-DW(i)
where:
Concentration of chemical in edible portion of aboveground produce type /',
PrAG-produce-DW(i) = exposed or protected, due to root uptake from soil at root-zone depth in the
produce-growing area, on a dry-weight (DW) basis (mg/kg produce DW)
r _ Average chemical concentration in soil at root-zone depth in produce-growing
osro0f_Z0ne produce - area (mg/kg soj| DW)
Chemical-specific plant/soil chemical bioconcentration factor for edible portion
BrAG-produce-DW(i) = of aboveground produce type /', exposed or protected (g soil DW / g produce
DW)
C-2-13
-------
Equation 3-3. Chemical Concentration in Aboveground Produce Due to Deposition of
Particle-phase Chemical
_ 1,000 X (Drdp + (Fw x Drwp)) x Rp{i) x (1 - e(-kp(i)"Tp(i)))
YP(i) x kp(i)
Chemical concentration in aboveground produce type /' on a dry-weight (DW)
basis due to particle deposition (mg/kg produce DW); set equal to zero for
protected aboveground produce
Average annual dry deposition of particle-phase chemical (g/m2-yr)
Fraction of wet deposition that adheres to plant surfaces; 0.2 for anions, 0.6 for
cations and most organics (unitless)
Average annual wet deposition of particle-phase chemical (g/m2-yr)
Interception fraction of the edible portion of plant type /' (unitless)
Plant surface loss coefficient for plant type /' (yr 1)
Length of exposure to deposition in the field per harvest of the edible portion of
plant type /' (yr)
Yield or standing crop biomass of the edible portion of plant type /' (kg produce
DW/m2)
Note that Equation 3-3 differs from Equation 5-14 in HHRAP, from which it is derived. In
HHRAP, Equation 5-14 includes the term Q x (1 - Fv) to indicate the emissions rate, in g/sec, of
chemical from the source and the proportion of the chemical that remains in, or partitions to, the
particle-phase in the air. Also in HHRAP, the dry and wet particle phase deposition rates, Dydp
and Dywp, respectively, are normalized to the emission rate and are expressed in units of
sec/m2-yr.
With MIRC, the user inputs both the dry and wet particle-phase deposition rates, Drdp and
Drwp, respectively, in units of g/m2-yr for a specific location relative to an emissions source.
Those deposition rates might be values measured near that location or estimated using a fate
and transport model, such as TRIM.FaTE, in conjunction with local meteorological information
and emissions rate data. The chemical emissions term used in HHRAP, Q, therefore, is not
used in MIRC's Equation 3-3. In addition, in MIRC, Drdp and Drwp, the average annual dry-
and wet-particle-phase deposition rates, respectively, are in units of g/m2-yr. Users of
TRIM.FaTE should note that the dry- and wet-particle-deposition rates output from TRIM.FaTE
are in units of g/m2-day; therefore, users must adjust the TRIM.FaTE output values to units of
g/m2-yr (i.e., multiply by 365 days/yr) before inputting values for Drdp and Drwp into MIRC.
where:
Pd,
0)
Drdp =
Fw
Drwp =
RP(o =
kPo) =
TPc) =
YPO) =
C-2-14
-------
Equation 3-4. Chemical Concentration in Aboveground Produce Due to
Air-to-Plant Transfer of Vapor-phase Chemical
CaxFv x Bvag^ x VGAG(j)
Concentration of chemical in edible portion of aboveground produce type /'
= from air-to-plant transfer of vapor-phase chemical on a dry-weight (DW) basis
(|jg/g produce DW); set equal to zero for protected aboveground produce
= Average annual total chemical concentration in air (g/m3)
= Fraction of airborne chemical in vapor phase (unitless)
Air-to-plant biotransfer factor for aboveground produce type /' for vapor-phase
chemical in air ([mg/g produce DW] / [mg/g air], i.e., g air/ g produce DW)
Empirical correction factor for aboveground exposed produce type /' to address
= possible overestimate of the diffusive transfer of chemical from the outside to
the inside of bulky produce, such as fruit (unitless)
= Density of air (g/m3)
Note that Equation 3-4 differs from Equation 5-18 in HHRAP, from which it is derived. In
HHRAP, Equation 5-18 includes the term Qx Fvto indicate the emissions rate, in g/sec, of
chemical from the source and the fraction of the chemical in vapor phase in the air. HHRAP
also includes the parameter Cyv, or the unitized yearly average air concentration of vapor-
phase chemical in units of |jg-sec/g-m3. For MIRC, the user inputs the average annual total air
concentration of the chemical, Ca, for a specific location relative to the source in units of g/m3;
MIRC includes a chemical-specific default value for Fv for chemicals included in its database.
The air concentration might be a value measured near that location or a value estimated by a
fate and transport model such as TRIM.FaTE. Users of TRIM.FaTE should note that the
average annual concentration of the total chemical in air (i.e., total of both vapor and particulate
phases), Ca, output from TRIM.FaTE is in units pg/m3; therefore, the user must adjust the value
to units of g/m3 (i.e., divide by 1,000 pg/g) before entering it in MIRC.
The calculations of chemical concentration in aboveground produce, (CAG-Produce-Dw), shown
above, are on a dry-weight (DW) basis. The family FFC food ingestion rates, on the other hand,
are on a fresh- or wet-weight (WW) basis. MIRC therefore calculates the concentration in
aboveground produce on a wet-weight basis, CAG-Produce-ww, using Equation 3-5 and the moisture
content (MAF) of the FFC food category.
where:
Pv0)
Ca
Fv
BVAG(i)
VGAG(i)
Pa
C-2-15
-------
Equation 3-5. Conversion of Aboveground Produce Chemical Concentration from
Dry- to Wet-Weight Basis
^AG-produce-WW(i) ^AG-produce-DW(i) x
((100 -MAFns)^
100
where:
Ca G-produce- WW(i)
Cfi G-produce-D W(i)
Chemical concentration in edible portion of aboveground produce type /' on a
wet-weight (WW) basis (mg/kg produce WW)
Chemical concentration in edible portion of aboveground produce type /' on a
dry-weight (DW) basis (mg/kg produce DW)
Moisture adjustment factor for aboveground produce type /' to convert the
MAF(0 = chemical concentration estimated for dry-weight produce to the corresponding
chemical concentration for full-weight fresh produce (percent water)
3.1.1.1 Belowground Produce
The equations by which chemical concentrations are estimated in belowground produce are
different for nonionic organic chemicals than for inorganic chemicals and ionic organic
chemicals.
Nonionic Organic Chemicals
For belowground produce and for nonionic organic chemicals, the concentration in the tuber or
root vegetable derived from exposure to the chemical in soil is estimated using an empirical root
concentration factor (RCF) and the average chemical concentration in the soil at the root-zone
depth in the produce-growing area (Csr0ot-zone_produce), as shown in Equation 3-6. The RCF
relates the chemical concentration in the plant on a wet-weight basis to the average chemical
concentration in the root-zone soil (Csr0ot-zone_produce) on a dry-weight basis. Belowground
produce (i.e., tubers or root vegetables) are protected from the deposition and vapor transfer by
being covered by soil. Therefore, root uptake of chemicals is the primary mechanism through
which belowground produce becomes contaminated.
C-2-16
-------
Equation 3-6. Chemical Concentration in Belowground Produce: Nonionic Organic
Chemicals
_ C^root zone _ produce x RCF X r00fveg
BG-produce-WW " KdS X UCF
Concentration of chemical in belowground (BG) produce (i.e., tuber or root
vegetable) on a wet-weight (WW) basis (mg chemical/kg produce WW) *
Average chemical concentration in soil at root-zone depth in produce-growing
area, on a dry-weight (DW) basis (mg chemical/kg soil DW)
Chemical-specific root concentration factor for tubers and root produce (L soil
pore water/kg root WW) *
Empirical correction factor for belowground produce (i.e., tuber or root
vegetable) to account for possible overestimate of the diffusive transfer of
chemicals from the outside to the inside of bulky tubers or roots (based on
carrots and potatoes) (unitless) *
Chemical-specific soil/water partition coefficient (L soil pore water/kg soil DW)
Units conversion factor of 1 kg/L
* Note that there is only one type of BG produce; hence there are no plant-type-specific subscripts
The RCF, as developed by Briggs et al. (1982), is the ratio of the chemical concentration in the
edible root on a wet-weight basis to its concentration in the soil pore water. RCFs are based on
experiments with growth solutions (hydroponic) instead of soils; therefore, it is necessary to
divide the soil concentration by the chemical-specific soil/water partition coefficient (Kds). There
is no conversion of chemical concentrations in belowground produce from DW to WW because
the values are already on a WW basis.
For nonionic organic chemicals, it is possible to predict RCF values and Kds values (for a
specified soil organic carbon content) from an estimate of the chemical's Kow from empirically
derived regression models. Those models are shown in HHRAP Appendix A-2, Equations A-2-
14 and A-2-15 (RCF) and in Equations A-29 and A-2-10 (Kds). The RCF and Kds values so
calculated for many of the chemicals in HHRAP are included in the MIRC database (including
the values for PAHs and dioxins).
Inorganic Chemicals
For inorganic chemicals and ionized organic chemicals, it is not possible to predict RCF or Kds
values from Kow. For inorganic chemicals, one must use empirical values for the root/soil
bioconcentration factor measured for specific chemicals. The root/soil bioconcentration factor,
now specified as BrBG-Produce-Dw, must be obtained from the literature for each inorganic chemical
on a DW basis. For inorganic chemicals, therefore, Equation 3-7 is used instead of
Equation 3-6.
where:
C BG-produce-WW
CSroot-zone_produce
RCF
VG,
rootveg
Kds
UCF
C-2-17
-------
Equation 3-7. Chemical Concentration in Belowground Produce: Inorganic Chemicals
yB G-produce-D 1/1/
CSroot-zone _ produce x ^^BG-produce-DW *VGrootveg
where:
^BG-produce-DW
CSroot-zone_produce
BrsG -produce-DW
VG rootveg
Concentration of chemical in edible portion of aboveground produce, due to
root uptake from soil at root-zone depth in the produce-growing area, on a dry-
weight (DW) basis (mg/kg produce DW)
Average chemical concentration in soil at root-zone depth in produce-growing
area (mg/kg soil DW)
Chemical-specific root/soil chemical bioconcentration factor for edible portion
of belowground produce (g soil DW / g produce DW)
Empirical correction factor for belowground produce (as in Equation 3-6)
(unitless)
As for the aboveground produce, the DW estimate of concentration of chemical in the root
vegetables must be transformed to a WW estimate, as shown in Equation 3-8.
Equation 3-8. Conversion of Belowground Produce Chemical Concentration from
Dry- to Wet-Weight Basis
where:
^BG-produce-WW ^BG-produce-DW x
(100 - MAFgG)
100
'B G-produce- WW
*BG-produce-DW
MAF,
¦(BG)
Chemical concentration in edible portion of belowground produce on a weight-
weight (WW) basis (mg/kg produce WW)
Concentration of chemical in edible portion of aboveground produce, due to root
uptake from soil at root-zone depth in the produce-growing area, on a dry-
weight (DW) basis (mg/kg produce DW)
Moisture adjustment factor (as in Equation 3-5, but single value for below
ground produce) (percent water)
3.1.2 Estimating Chemical Concentrations in Animal Products
Chemical concentrations in animal products are estimated based on the amount of chemical
consumed by each animal group m through each plant feed type /' (PlantCh-mtake(i,m>) and
incidental ingestion of soil for ground-foraging animals (SoilCh-intake(m))¦ Exhibit 3-3 summarizes
the pathways by which chemicals are transferred to these home- or farm-raised animal food
products. Note that for a general screening-level assessment, all of the pathways can be
modeled, as is done for EPA's RTR calculation of de minimis emission rates for PB-HAPs (EPA
2008b).
The feed options for farm animals in MIRC include forage (plants grown on-site for animal
grazing, such as grass), silage (wet forage grasses, fresh-cut hay, or other fresh plant material
that has been stored and fermented), and grain products grown on the farm. As seen in Exhibit
3-3, the algorithms in MIRC for chemical intake with plant feeds (Plantch-mtakeo.m)) are based on
the assumptions that beef and dairy cattle consume all three plant feed products, while pigs
consume only silage and grain, and chickens consume only grain.
C-2-18
-------
Exhibit 3-3. Chemical Transfer Pathways for Animal Products
Farm Food Media
Chemical Transfer Pathways
Animal Products
Beef and total dairy
(including milk)
. Ingestion of forage, silage, and grain a
. Incidental soil ingestion
Pork
. Ingestion of silage and grain a
. Incidental soil ingestion
Poultry and eggs
. Ingestion of grain a
. Incidental soil ingestion
a Chemical concentrations in plant feed (i.e., forage, silage, and grain) are estimated via intermediate
calculations (see Equations 3-13, 3-14, 3-3, and 3-4).
Forage and silage are exposed to the air and can accumulate chemical via direct deposition of
particle-bound chemical and transfer of vapor-phase chemical, while all animal feed grains are
assumed to be protected from the air by a husk or pod (e.g., corn, soybeans). All three animal
feed products are assumed to accumulate chemical via root uptake.
Chemical concentrations are estimated for animal feeds using algorithms analogous to those for
aboveground farm produce described above. MIRC uses Equation 3-9 to calculate the
concentration of chemical in beef, pork, or total dairy and Equation 3-10 to calculate the
concentration of chemical in poultry or eggs. The chemical concentration in mammalian farm
animals (i.e., beef and pigs) is adjusted using a metabolism factor (MF) that accounts for
endogenous degradation of the chemical (see Equation 3-9). MF is set to 1.0 for chemicals that
are not metabolized and for chemicals for which the metabolic degradation rate is unknown.
Although other vertebrates, including birds, are likely to have similar metabolic pathways for
most chemicals, the conservative assumption is that birds do not metabolize any chemicals;
therefore, the MF is omitted from Equation 3-10 for poultry and eggs.
Equation 3-9. Chemical Concentration in Beef, Pork, or Total Dairy
f n \
Soilrh_i„takolm\ + PlStltch lntake(i,m)
^mammal(m) ®®(m) x /WF X
Ch-lntake(m)
V '=1
where:
smammal(m)
Ba,
(m)
MF =
Soil,
Ch -lntake(m)
Plant,
Ch-lntake(i,m)
Concentration of chemical in mammalian animal product m, where m = beef,
pork, or total dairy (mg chemical/kg animal product WW)
Chemical-specific biotransfer factor for chemical in diet to chemical in animal
food product m, where m = beef, pork, or total dairy ([mg chemical/kg animal
product WW] / [mg chemical intake/day] or day/kg WW)
Chemical-specific mammalian metabolism factor that accounts for endogenous
degradation of the chemical (unitless)
Incidental ingestion of chemical in surface soils by livestock type m during
grazing or consumption of foods placed on the ground (mg/day); see Equation
3-11 below
For livestock (animal product) type m, ingestion of chemical from plant feed
type /' (mg chemical/kg livestock WW); see Equation 3-12 below
(If m = beef or total dairy, then n = 3 and /' = forage, silage, and grain;
m = pork, then n = 2 and /' = silage and grain;
m = poultry, then n = 1 and I = grain.)
C-2-19
-------
Equation 3-10. Chemical Concentration in Poultry or Eggs
poultry(m) ~ ®®(m) x {Soilch-lntake(m) Pl&HtCh-lntake(i,m))
Concentration of chemical in food product m, where m = poultry or eggs (mg
chemical/kg animal product WW)
Chemical-specific biotransfer factor for food product m, where m = poultry or
eggs (day/kg animal product WW)
Incidental ingestion of chemical in surface soils by consumption of food on the
ground (mg chemical/day) where m = poultry; see Equation 3-11
For poultry (and eggs), animal m, ingestion of the chemical in plant feed type /'
(mg chemical/day), which for poultry is limited to grain; see Equation 3-12
In MIRC, the incidental ingestion of the chemical in soils by livestock during grazing or
consumption of feed placed on the ground (SoilCh-mtake(m>) is estimated using empirical soil
ingestion rates (Qs) and a soil bioavailability factor for livestock (Bs), as shown in
Equation 3-11. At this time, the default value for Bs in MIRC for all chemicals is 1.0 (i.e., the
chemical in soil is assumed to be 100 percent bioavailable to the animal). This assumption may
be reasonably accurate for the soil surface to which airborne chemical is deposited. MIRC
allows the user to enter a surface soil concentration for areas where livestock forage, CsS-nVestock,
that is distinct from the surface soil concentration input for areas where produce may be grown
and where humans might incidentally ingest soils (see Section 6.1).
Equation 3-11. Incidental Ingestion of Chemical in Soil by Livestock
Soilch Intake(m) ~ x S-livestock x
where:
Incidental ingestion of the chemical in surface soils by livestock type m during
grazing or consumption of foods placed on the ground (mg chemical/day)
Quantity of soil eaten by animal type m each day (kg soil DW/day)
Chemical concentration in surface soil in contaminated area where livestock
feed (mg chemical/kg soil DW)
Soil bioavailability factor for livestock (unitless) (assumed to be the same for
birds and mammals)
Animal ingestion of the chemical in feed is calculated for each type of livestock based on their
assumed diets. For m = beef and dairy cattle, chemical intake is estimated for all three feed
types: /' = forage, silage, and grain. For pork, chemical intake is estimated only for silage and
grain. The chemical intake for poultry is based on grain consumption only. The intake of
chemical with each feed type, /', PlantCh-mtake(i,m), is calculated separately according to Equation
3-12. Note that the animal feed ingestion rates are on a dry-weight (DW) basis; hence, no DW
to wet weight (WW) conversion is needed.
where:
Cpoultry(m)
Bd(m)
Soilch -intake(m)
Plantch -intake(i,m)
Soilch -lntake(m) ~
QS(m) =
CSs-livestock =
Bs =
C-2-20
-------
Equation 3-12. Ingestion of Chemical in Feed by Livestock
Pl3ntch-intake(i,m) = i,m) x QP(/,m) x Cfeed(i)
where:
Ingestion of chemical in plant feed type /' (mg chemical/day), where /' = forage,
silage, or grain, for livestock type m
Fraction of plant feed type /' obtained from contaminated area used to grow
animal feed, where I = forage, silage, or grain (unitless) for lifestock type m
Quantity of plant feed type /' consumed per animal per day (kg plant feed
DW/day), where /' = forage, silage, or grain, for livestock type m
Concentration of chemical in ingested plant feed type /' (mg chemical/kg plant
feed DW), where /' = forage, silage, or grain
The concentrations of chemical in the three different types of plant feeds for livestock are
calculated according to Equation 3-13. The equation is the same as that for aboveground
produce in Equation 3-1, with the exception that the concentrations are for plants used as
animal feeds (not produce consumed by humans) and all types of plant feed (i.e., forage, silage,
and grain) are aboveground.
Chemical Concentration in Lifestock Feed (All Aboveground)
Cfeed(i) = Prfeed(i) + Pd(i) + PV(i)
Concentration of chemical in plant feed type /' on a dry-weight (DW) basis (mg
chemical/kg plant feed DW), where /' = forage, silage, or grain
Concentration of chemical in plant feed type /' due to root uptake from soil
(mg/kg DW), where /' = forage, silage, or grain; see Equation 3-14 below
Concentration of chemical in plant feed type /' due to wet and dry deposition of
particle-phase chemical (mg/kg DW), where /' = forage, silage, or grain; when /'
= grain, the Pd term equals zero
Concentration of chemical in plant feed type /' due to air-to-plant transfer of
vapor-phase chemical (|jg/g [or mg/kg] DW) where /' = forage, silage, or grain;
when /' = grain, the Pd term equals zero
MIRC calculates the chemical concentration in animal feed due to root uptake from the soil
using Equation 3-14. The equation is the same as Equation 3-2, except that a Br value
appropriate to grasses is used and MIRC allows for different soil concentrations in the area
used to grow animal feed than in the area used to grow produce for human consumption (see
Section 6.1, user inputs). Note that for feed type /' = grains, the Pd and Pv terms do not apply
(are set to zero), because the feed products (i.e., corn kernels, soy beans) are protected from
the air (i.e., by husks, pods).
Plant Ch-lntake(i,m) —
F(i,m) =
QP(i,m)
Cfeed(i) =
Equation 3-13.
where:
Cfeed(i) =
Prfeed(i) =
Pd«, =
Pv,
0)
C-2-21
-------
Equation 3-14. Chemical Concentration in Livestock Feed Due to Root Uptake
^^feed(i) — root-zone_ feed(i) x ^^feed(i)
where:
Concentration of chemical in plant feed type /' due to root uptake from soil on a
Pffeedo) = dry-weight (DW) basis (mg chemical/kg plant feed DW), where /' = forage,
silage, or grain
£s _ Average chemical concentration in soil at root-zone depth in area used to grow
root-zonejeed(i) - p|ant feec| type / (mg chemical/kg soil DW), where /' = forage, silage, or grain
gr _ Chemical-specific plant-soil bioconcentration factor for plant feed type /' (kg soil
feed(o - Dw/kg plant feed DW), where /' = forage, silage, or grain
The algorithms used to calculate Pd(i) and Pv(i) when plant feed type /' = forage and silage are
identical to those used to calculate Pd(i) and Pv(i) for aboveground exposed produce (i.e.,
Equations 3-3 and 3-4, respectively).
There are no conversions of DW feed to WW feed, because all feed ingestion rates for livestock
are based on DW feed.
3.2 Chemical Intake Calculations for Adults and Non-Infant Children
MIRC calculates human chemical intake rates from the ingestion of home-grown foods as
average daily doses (ADDs) normalized to body weight for each age group, chemical, and food
type separately. ADDs, calculated using Equation 3-15, are expressed in milligrams of chemical
per kilogram of receptor body weight per day (mg/kg-day).
Equation 3-15. Average Daily Dose for Specified Age Group and Food Type
Cm x IR,wii x FCm x ED,wi Y EF,„
ADD(y,i) ~
'W A " (y,i) A ' ^(j) A "-"(y)
BW(y) x AT(y)
(y)
where:
365 days
ADD - Average d3ily dose for age group y from food type or ingestion medium /' (mg
(yJ> ~ chemical/kg body weight-day)
£ _ Concentration of chemical in food type /' harvested from the contaminated area
0> (mg chemical/kg food or mg food/L water)
IR(yj) = Ingestion rate for age group y of food type /' (kg/day or L/day)
FC0) = Fraction of food type /' that was harvested from contaminated area (unitless)
ED(y) = Exposure duration for age group y (years)
BW(y) = Body weight for age group y (kg)
AT _ Averaging time for calculation of daily dose (years) for age group y, set equal
(y) ~ to ED in MIRC
EF(y) = Annual exposure frequency for age group y (days)
Equation 3-15 takes into account the chemical concentration in each food type /' (or in water),
the quantity of food brought into the home for consumption, the loss of some of the mass of the
foods due to preparation and cooking, how much of the food is consumed per year, the amount
C-2-22
-------
of the food obtained from contaminated areas, and the consumer's body weight (EPA 1997a,
2003a). In MIRC, ADDs are calculated separately for each chemical, home-grown food type,
and consumer age group.
ADD values, expressed as intakes, not absorbed doses, are appropriate for comparison with
RfDs and for use with cancer slope factors (CSFs) to estimate risk, as discussed in Section 5.
An exception is for the breast-milk exposure pathway, where the dose absorbed by the mother
is relevant to calculating the dose available to and absorbed by her nursing infant, as discussed
in Section 3.4.
MIRC evaluates only one contaminated area (set of environmental concentrations), or exposure
scenario, at a time. For screening level assessments, all components of this equation are
assumed to remain constant for consumers in a given age group over time (e.g., seasonal and
annual variations in diet are not explicitly taken into account). To calculate an ADDW) from the
contaminated area for food group /' over an entire lifetime of exposure, age-group-specific
ingestion rates and body weights are used for the age groups described in Section 2.3. In
MIRC, the averaging time used to calculate the daily dose for an age group (ATy) is equal to the
exposure duration for that group (EDy); therefore these variables drop out of Equation 3-15.
For each chemical included in a screening scenario, total average daily exposure for age
group y (ADD(y)) is estimated as the sum of chemical intake from all ingestion pathways
combined:
Incidental soil ingestion;
Ingestion of fish;
Ingestion of homegrown fruits (exposed and protected);
Ingestion of homegrown vegetables (exposed, protected, and root);
Ingestion of animal products from home-raised animals:
o Milk and other dairy products from cows,
o Beef products,
o Pork products, and
o Poultry and eggs;
• Ingestion of drinking water from specified source; and
• Ingestion of breast milk by infants.
Note that the last exposure pathway is limited to infants.
The algorithms for the first six exposure pathways listed above are described in Sections 3.2.1
through 3.2.6. The algorithms for the breast-milk ingestion pathway are described in Section
3.4.
C-2-23
-------
3.2.1 Chemical Intake from Soil Ingestion
Equation 3-16 shows the equation used to estimate chemical intake through incidental ingestion
of soil.
Equation 3-16. Chemical Intake from Soil Ingestion
ADD
Soil(y)
CSoil x IRsoil(y) x FCSoil X 0.001
mg
V9
BW,
(y)
EF
365 days
where:
ADD,
Soil(y)
Csoil
IRsoil(y)
FCsoil
BW(y)
EF
Average daily chemical intake from incidental ingestion of soil or ingestion by
child in age group y (mg chemical/kg body weight-day)
Concentration of chemical in soil from contaminated area on a dry-weight
(DW) basis (|jg/g soil DW)
Soil ingestion rate for age group y (g DW/day)
Fraction of soil ingested that is from contaminated area (unitless)
Body weight for age group y (kg)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Note: MIRC saves soil ingestion rates in units of mg/day (not g/day); therefore, there is an additional
0.001 g/mg conversion unit in the actual MIRC algorithm than shown here.
3.2.2 Chemical Intake from Fish Ingestion
Ingestion of locally caught fish is included as a possible exposure pathway in MIRC
(Equation 3-17). Two types offish are included in the exposure algorithm: trophic level 3 (T3)
fish, equivalent to small "pan" fish such as bluegill, and trophic level 4 (T4) fish, equivalent to
game fish such as trout and walleye. The chemical concentration in fish in Equation 3-17 is
estimated as the consumption-weighted chemical concentration using Equation 3-18.
C-2-24
-------
Equation 3-17. Chemical Intake from Fish Ingestion
ADDpjshfy) - (1 LAFish)x(} ^Fish)'
'Fish "^Fish(y)
g
'Fish
BW,
(y)
EF
365 days
where:
Equation 3-18. Consumption-weighted Chemical Concentration in Fish
CFish = (PFishT3 X ^T3 ) + {^FishTA X ^Ta)
ADD,
Fish(y)
L1
Fish
Average daily chemical intake from ingestion of local fish for age group y
(mg/kg-day)
Weight of fish brought into home that is discarded during preparation (e.g.,
head, bones, liver, other viscera, belly fat, skin with fat) (unitless)
Loss of weight during cooking, such as evaporation and loss of fluids into pan
(unitless)
Chemical concentration in whole fish for trophic level 3 (T3) fish on a wet-
weight (WW) basis (mg/kg WW)
Chemical concentration in whole fish for trophic level 4 (T4) fish on a wet-
weight (WW) basis (mg/kg WW)
Fraction offish intake that is from T3 (unitless)
Fraction offish intake that is from T4 (unitless)
Consumption-weighted mean chemical concentration in total fish (i.e., as
specified by Equation 3-18) (mg/kg WW)
Fraction of local fish consumed derived from contaminated area (unitless)
Body weight for age y (kg)
Local fish ingestion rate for age y (g WW/day)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
* Parameter values must be internally consistent. In contrast to the tables included in MIRC for ingestion
rate options for homegrown food products, which are based on the products as brought into the home
from the field (see Section 6.3.3), the tables of fish ingestion rate options included in MIRC are from
CSFII data (see Section 6.3.4) and, therefore, are on an "as consumed" basis (i.e., after preparation and
cooking losses), and L1 and L2 therefore are set equal to zero. If the user wishes to enter local fish
ingestion rates on an "as harvested" basis, the user also should enter L1 and L2 values as specified in
Section 6.4.3.
L2Fish*
CFishT3
CFishT4
FT3
Ft4
CFish
FCpiSh
BW(y)
I f^Fish(y)
EF =
When whole fish are prepared for cooking, it is usual for the viscera, head, and fins to be
removed, particularly for larger fish. Many persons also remove (or do not eat) the skin, bones,
and belly fat. EPA has, therefore, estimated the proportion of the weight of whole fish that tends
to be lost during preparation and cooking across a variety of fish species (Exposure Factors
Handbook Table 13-5, EPA 1997a) and included those losses in its HHRAP algorithms for
chemical intake from fish (LI Fish and L2FiSh in Equation 3-17).
C-2-25
-------
3.2.3 Chemical Intake from Fruit Ingestion
Average daily doses of a chemical from homegrown exposed fruits are calculated separately for
exposed and protected fruits (Equations 3-19 and 3-20, respectively).
Equation 3-19. Chemical Intake from Consumption of Exposed Fruits
0 LAExpFruit) X 0 ^ExpFruit) x
kg
A
ADDExpFnijt(y) ~ \) L"\ExpFruit )XV 1-2 ExpFrujt )x | CExpFrujt X IRExpFniit(y) X 0.001 X FC ExpFruit
y
r
EF
365 days
Equation 3-20. Chemical Intake from Consumption of Protected Fruits
ADD,
ProFruit(y)
— (l U\ProFrujt)x
f
kg
CproFruit x ^ProFruit(y) x 0-001 ~ x ^^ProFruit
\ f
X
EF
365 days
where:
ADDExPFruit(y) _ Average daily chemical intake from ingestion of exposed fruit or protected fruit
ADDproFruit(y) ~ (depending on subscript) (mg chemical/kg body weight-day)
Mean reduction in fruit weight resulting from removal of skin or peel, core or
L1 ExpFruit = pit, stems or caps, seeds and defects, and from draining liquids from canned
or frozen forms (unitless)
Mean reduction in fruit weight that results from paring or other preparation
techniques for protected fruits (unitless)
Mean reduction in fruit weight that results from draining liquids from cooked
forms of the fruit (unitless)
Chemical concentration in whole exposed fruits or whole protected fruits
(depending on subscript) on a wet-weight (WW) basis (mg chemical/kg
exposed fruit WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
FCExPFruit _ Fraction of exposed fruits or protected fruits (depending on subscript) obtained
FCProFruit ~ from contaminated area (unitless)
IRExPFruit(y) _ Ingestion rate of home-grown exposed fruits or protected fruits (depending on
IRproFruit(y) ~ subscript) for age y (g WW/kg body weight-day)
LIproFruit
L2ExpFruit
CExpFruit
CproFruit
EF =
Fruit ingestion rates in the survey were based on weights of unprepared fruits (e.g., one apple;
one pear) or the weight of a can of fruit (e.g., 8 oz can). The weight of the fruit ingested is less
than the initial weight owing to common preparation actions (L7£xpF™fand L1Pr0Fruit; e.g., coring
apples and pears; peeling apples; pitting cherries). Cooking of exposed fruit (e.g., berries,
apples, peaches) often results in further weight loss that results from liquids lost during cooking
and drained from the cooking vessel (L2ExpFrwt)¦ EPA has assumed that cooking of protected
fruit results in no loss of weight for the fruit.
3.2.4 Chemical Intake from Vegetable Ingestion
MIRC includes three separate algorithms for homegrown vegetables adapted from EPA's
HHRAP Modeling System (EPA 2005a): one for exposed vegetables such as asparagus,
broccoli, lettuce, and tomatoes (although they are actually a fruit); one for protected vegetables
such as corn, cabbage, soybeans, and peas; and one for root vegetables such as carrots,
beets, and potatoes (see Equations 3-21, 3-22, and 3-23, respectively).
C-2-26
-------
Equation 3-21. Chemical Intake from Exposed Vegetables
ADD,
ExpVeg(y)
0 l-\ExpVeg )x I ^ExpVeg x ^ExpVeg(y) x 0-001 ^ x FC
Exp Veg
f_EF_\
365 days
Equation 3-22. Chemical Intake from Protected Vegetables
ADD,
Pro Veg (y)
0 l-\ProVeg )x I ^ProVeg x ^ProVeg(y) x 0.001 ^ X FC,
ProVeg
y ef }
365 days
Equation 3-23. Chemical Intake from Root Vegetables
^ DDRoof \/eg fyj
where:
0 ^RootVeg )x 0 ^-^-RootVeg )x
L1
L1
L1
kg
^ Root Veg x ^RootVeg(y) x 0.001—— X FCpootVeg
y
A f
x
EF
365 days
^ DDexP Veg(y)
A DDprQ Veg(y)
A DDRoot Veg(y)
ExpVeg
ProVeg
RootVeg
L2p;00t\/eg
C ExpVeg
C ProVeg
CRoot Veg
EF =
FC^xpVgg
FOprQ\/eg
FCftootVeg
I f^Exp Veg(y)
IRproVeg(y)
I^RootVeg(y)
Average chemical intake from ingestion of exposed vegetables, protected
vegetables, or root vegetables (depending on subscript) for age group y (mg
chemical/kg body weight-day)
Mean net preparation and cooking weight loss for exposed vegetables
(unitless); includes removing stalks, paring skins, discarding damaged leaves
Mean net cooking weight loss for protected vegetables (unitless); includes
removing husks, discarding pods of beans and peas, removal of outer leaves
Mean net cooking weight loss for root vegetables (unitless); includes losses
from removal of tops and paring skins
Mean net post cooking weight loss for root vegetables from draining cooking
liquids and removal of skin after cooking (unitless)
Chemical concentration in exposed vegetables, protected vegetables, or root
vegetables (depending on subscript) on a wet-weight (WW) basis (mg
chemical/kg vegetable WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Fraction of exposed vegetables, protected vegetables, or root vegetables
(depending on subscript) obtained from contaminated area (unitless)
Ingestion rate of exposed vegetables, protected vegetables, or root vegetables
(depending on subscript) for age group y (g vegetable WW/kg body weight-
day)
C-2-27
-------
3.2.5 Chemical Intake from Animal Product Ingestion
Calculations of chemical intake from the consumption of farm animals and related food products
are provided below in Equations 3-24 through 3-28 for homegrown beef, dairy (milk), pork,
poultry, and eggs, respectively.
Equation 3-24. Chemical Intake from Ingestion of Beef
( Ien \ ( EF
ADDBeef(y) - (l LABeef) y- (l
Beef )
where:
Kq
^ ^Beef ^ ^Beef(y) x 0-001 x FCBeef
g
365 days
„nn Average daily chemical intake from ingestion of beef for age group y (mg/kg-
AUUBeef(y) ~ fjay)
Llseef = Mean net cooking loss for beef (unitless)
L2Seef = Mean net post cooking loss for beef (unitless)
Cseef = Concentration of contaminant in beef (mg/kg WW))
_ Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
IRBeem = Ingestion rate of contaminated beef for age group y (g WW/kg-day)
FC,
Beef
Fraction of beef consumed raised on contaminated area or fed contaminated
silage and grains (unitless)
Equation 3-25. Chemical Intake from Dairy Ingestion
( \er\ \ ( EF
X
ADD,
where:
Dairy(y)
CDairy x ^Dairy(y) x 0.001 ^ X FCDajry
y
365 days
ADDoairyfy)
CDairy
EF
If^Dairy(y)
FC Dairy
Average daily chemical intake from ingestion of total dairy for age group y
(mg/kg-day)
Average concentration of contaminant in total dairy (mg/kg WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Ingestion rate of contaminated total dairy for age group y (g WW/kg-day)
Fraction of total dairy products from contaminated area (unitless)
C-2-28
-------
Equation 3-26. Chemical Intake from Pork Ingestion
where:
ADDpork(y) - 0 ^Pork)x L2pori<)>
f kq ^ ( EF
Cpork x IRPork(y) x 0 001 ~ x FCPork x 355 (jays
ADDPork(y)
L1 Pork
L2pork
Cpork
EF
IRpork(y)
FCpork
Average daily chemical intake from ingestion of pork for age group y (mg/kg-
day)
Mean net cooking loss for pork (unitless); includes dripping and volatile losses
during cooking; averaged over various cuts and preparation methods
Mean net post cooking loss for pork (unitless); includes losses from cutting,
shrinkage, excess fat, bones, scraps, and juices; averaged over various cuts
and preparation methods
Concentration of contaminant in pork (mg/kg WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Ingestion rate of contaminated pork for age y (g WW/kg-day)
Fraction of pork obtained from contaminated area (unitless)
The reduction in the weight of pork during and after cooking may correlate with an increase or
decrease in the concentration of the chemical in the pork as consumed depending on the
chemical and depending on the cooking method.
Equation 3-27. Chemical Intake from Poultry Ingestion
ADD,
Poultry(y)
0 LA poultry )x (l L2Poultry)x\ CPoultry x IRPoultry(y) x0.00'\ ^xFC}
where:
Poultry
f EF ^
365 days
A DD Poultry (y)
I-1 Poultry
L2p0ultry
^Poultry
EF
IF Poultry (y)
FCpoultry
Average daily dose (chemical intake) from ingestion of poultry (mg/kg-day)
Mean net cooking loss for poultry (unitless)
Mean net post cooking loss for poultry (unitless)
Concentration of chemical in poultry (mg/kg WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Ingestion rate of poultry for age group y (g WW/kg-day)
Fraction of poultry from contaminated area or fed contaminated grains
(unitless)
C-2-29
-------
where:
Equation 3-28. Chemical Intake from Egg Ingestion
ADD
Egg(y)
CEgg X IREgg(y) X 0.001 kg9 X FCEgg ) X f 365£^ays
ADDEgg(y)
^Egg
EF
IREgg(y)
FCEgg
Average daily chemical intake from ingestion of eggs for age group y (mg/kg-
day)
Concentration of contaminant in eggs (mg/kg WW)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
Ingestion rate of contaminated eggs for age group y (g WW/kg-day)
Fraction of eggs obtained from contaminated area (unitless)
3.2.6 Chemical Intake from Drinking Water Ingestion
If the user chooses to evaluate chemical ingestion via drinking water, the user specifies a
chemical concentration in g/L (equivalent to mg/mL) based on their particular scenario. The
chemical concentration could represent water from groundwater wells, community water, nearby
surface waters, or other source. For this exposure pathway, ingestion rates are in units of
milliliters of water per day (mL/day).
Equation 3-29. Chemical Intake from Drinking Water Ingestion
where:
ADD,
DW(y)
CdW x ^DW(y) x FCdw
BW,
\ r
X
(y)
EF
J
\
365 days
ADD,
DW(y)
Cdw
IRdwm
FCdw
BW(y)
EF =
Average daily chemical intake from ingestion of drinking water from local
residential water source for age group y (mg/kg-day)
Concentration of contaminant in drinking water (g/L)
Drinking water ingestion rate for age group y (mL/day)
Fraction of drinking water obtained from contaminated area (unitless)
Body weight of age group y (kg)
Exposure frequency; number of days per year of exposure for family(ies) as
specified for scenario (< 365 days)
3.3 Total Chemical Intake
To estimate the total ADD, or intake of a chemical from all of the exposure media that a single
individual in each age group is expected to contact (e.g., soil, local fish, five types of home-
grown produce, and five types of home-raised animals or animal products), the media-specific
chemical intakes are summed for each age group. Total average daily exposure for a particular
age group y (ADD(y)) is estimated as the sum of chemical intake from all ingestion pathways
combined, as illustrated in Equations 3-30 through 3-35 below.
C-2-30
-------
Equations 3-30 to 3-35. Total Average Daily Dose of a Chemical for Different Age Groups
Equation 3-30. ADD( = L=/DZW/)
Equation 3-35. ADD(adult> = Z/=i ADD^uit,i)
where /' represents the ith food type or ingestion medium and n equals the total number of food
types or ingestion media, and ADD parameters are defined below:
Total average daily dose of chemical for infants less than one year from
AUU^I) = : c , / ,i._
ingestion of breast milk (mg/kg-day)
_ Total average daily dose of chemical from all ingestion sources for children
<1'2> ~ ages 1 through 2 years (mg/kg-day)
ADD(3-5) = Total average daily dose for children ages 3 through 5 years (mg/kg-day)
ADD(e-u) = Total average daily dose for children ages 6 through 11 years (mg/kg-day)
ADD(12-19) = Total average daily dose for children ages 12 through 19 years (mg/kg-day)
ADD(adun) = Total average daily dose for adult age 20 up to 70 years (mg/kg-day)
The lifetime average daily dose (LADD) is calculated as the time-weight average of the ADD
values for each age group (Equation 3-36).
Equation 3-36. Lifetime Average Daily Dose (LADD)
LADD^ADD^j + ADD„.2)^ + + ADDm1v[±) + ADD„[
The time-weighting factors simply equal the duration of exposure for the specified age category
in years divided by the total lifespan, assumed to be 70 years. For risk assessments for
chemicals with a subchronic RfD or for developmental effects in children, ADD(y) values for the
child age groups are compared with the RfD (see Section 5).
C-2-31
-------
3.4 Chemical Intake Calculations for Nursing Infants
The scientific literature indicates that infants can be exposed to some chemicals via their
mothers' breast milk. The magnitude of the exposure can be estimated from information on the
mother's exposure, data on the partitioning of the chemical into various compartments of the
mother's body and into breast milk, and information on the infant's consumption of milk and
absorption of the chemical. To add this exposure pathway to the MIRC application, we adapted
exposure algorithms and default assumptions from EPA's Methodology for Assessing Health
Risks Associated with Multiple Pathways of Exposure to Combustor Emissions (EPA 1998),
hereafter referred to as MPE, as explained below.
Note that this pathway generally is of most concern for lipophilic bioaccumulative chemicals
(e.g., dioxins) that can cause developmental effects. The period of concern for the more
hydrophilic chemicals that cause developmental effects generally is earlier, that is, from
conception to birth. Hydrophilic chemicals generally exchange well between the maternal and
fetal blood supplies at the placenta.
3.4.1 Infant Average Daily Absorbed Dose
The average daily dose of chemical absorbed by the infant (DAIint) is estimated in MIRC with
Equation 3-37. This basic exposure equation relies on the concentration of the chemical in the
breast milk, the infant's breast-milk ingestion rate (IRmnk), the absorption efficiency of the
chemical by the oral route of exposure (AEint), the bodyweight of the infant (BWinf), and the
duration of breast feeding (ED). Equation 3-37 is EPA's (EPA 1998) modification of an average
daily dose for the infant model first published by Smith (1987) and includes variables for both
the concentration of the chemical in the breast milk fat (Cm„Waf) and the concentration of the
chemical in the aqueous phase of breast milk (Caqueous)¦ The remainder of the DAIinrassociated
equations assume that most chemicals of concern will partition either to the lipid phase or to the
aqueous phase of breast milk, although some chemicals may partition significantly to both
phases of milk. Thus, the remaining equations in MIRC assume that either Cm//waf or Caqueous is
equal to zero and hence drops out of the equation.
For the parameters in Equation 3-37 (and the equations that follow) that are not calculated from
another equation, an EPA default value and options for other values available in MIRC for the
infant breast-milk-exposure pathway are described in Section 6.4. The user also can overwrite
those parameter values with a different value from the literature as appropriate.
C-2-32
-------
where:
Equation 3-37. Average Daily Dose of Chemical to the Nursing Infant
[(^m//f
-------
Equation 3-38. Chemical Concentration in Breast Milk Fat
'milkfat
DAImat x ff
kelim x ffm
where:
*e//m
1
fat _elac
kfat _ elac x ^bf
—k t
-j 0 Ae//'mlpn
*e//m
fat _elac
^1 0 ^fat elac^bf ^
_ Concentration of chemical in lipid phase of maternal milk (ma chemical/kq
Cmilkfat = |ipjd)
_ Daily absorbed maternal chemical dose (mg chemical/kg maternal body
mat ~ weight-day; calculated using Equation 3-39)
Fraction of total maternal body burden of chemical that is stored in maternal fat
ff = (mg chemical in body fat / mg total chemical in whole body; value from
literature or EPA default - see Section 6.5)
Chemical-specific total elimination rate constant for elimination of the chemical
keiim = by non-lactating women (per day; e.g., via urine, bile to feces, exhalation;
value from literature or calculated using Equation 3-40)
ffm = Fraction of maternal body weight that is fat stores (unitless)
Chemical-specific rate constant for total elimination of chemical in the lipid
kfat_eiac = phase of milk during nursing (per day; value from literature or calculated using
Equation 3-41)
tbf = Duration of breast feeding (days)
fpn
Duration of mother's exposure prior to parturition and initiation of breast
feeding (days)
Equation 3-38 relies on the daily maternal absorbed intake (DAImat) to determine the
concentration of the chemical in the breast milk fat. DAImat is multiplied by the fraction of the
chemical that is stored in maternal fat (fr) to determine the amount (i.e., mass) of chemical in the
fat. This product, divided by the chemical-specific elimination rate constant (kenm) for non-
lactating adult women and the fraction of the mother's weight that is fat (ffm), represents the
maximum theoretical steady-state concentration of the chemical in an adult woman. If used
alone to estimate the chemical concentration in breast milk fat, the equation as explained thus
far is likely to overestimate the chemical concentration in milk fat because it does not account
for losses due to breast feeding. Alone, this term (DAImatffl kenm ffm) also assumes that the
biological half-life of the chemical in the mother's breast milk fat is small relative to the duration
of the mother's exposure. However, for chemicals with half-lives that are longer than the
exposure duration, which are the chemicals of concern in the applications of MIRC to date, an
additional term is needed to determine the average concentration in the milk fat over the
duration of her exposure.
To account for breast feeding losses and longer chemical half-lives in the mother than the
exposure duration, an additional term is included in Equation 3-38. This term includes a fraction
dependent on two rate constants, keiim and the elimination constant for a lipophilic chemical in
lactating women via the lipid phase of breast milk (kfat_eiac), the duration of the mother's
chemical exposure prior to nursing (fpn), and the duration of breast feeding (tbf). The whole body
concentration (DAImatffl ke,jm ffm), the maximum theoretical steady-state concentration, is
multiplied by the rate of elimination averaged over the duration of the mother's exposure,
including her exposure prior to and during lactation. To review the derivation of Equation 3-38,
see Appendix B of MPE (EPA 1998).
C-2-34
-------
To estimate an average daily dose absorbed by an infant's mother, or DAImat, the average daily
dose (ADD) (in mg/kg-day) for the chemical from all sources that MIRC calculates for adults
(ADD(aduit), described in Section 3.3, Equation 3-35), is multiplied by an absorption efficiency
(AEmat) or fraction of the chemical absorbed by the oral route of exposure, as shown in Equation
3-39. The value for AEmat can be estimated from absorption efficiencies for adults in general.
Available data for some chemicals, in particular some inorganic compounds, indicate AE values
for ingestion exposures of substantially less than 100 percent. For a few of these chemicals,
data also indicate lower AEs for the chemical when ingested in food or in soil than when
ingested in water (e.g., cadmium). For a screening level assessment, however, it is reasonable
to either assume 100 percent for the AEmat or to use the higher AEmat of the food and water
AEmat values if available; hence, a single AEmat parameter is included in Equation 3-39.
Equation 3-39. Daily Maternal Absorbed Intake
DAImat = ADD(adult) xAE
mat
= Daily maternal dose of chemical absorbed from medium /' (mg/kg-day)
_ Average daily dose to the mother (mg/kg-day) (calculated by MIRC - see
Section 3.3, Equation 3-35)
Absorption efficiency of the chemical by the oral route of exposure (i.e.,
= chemical-specific fraction of ingested chemical that is absorbed) by the mother
(unitless) (value from literature or EPA default - see Section 6.4)
Equation 3-35, used to calculate ADD(aM)t is based on many medium-specific ingestion rates
that are normalized to body weight. The adult body weights to which the homegrown food
ingestion rates are normalized are the body weights of the consumers in the original USDA
survey (see Section 6.3.3), which included both males and females. An assumption in the
breast-milk exposure pathway is that those ingestion rates also are applicable to nursing
mothers. The original data for ingestion rates for soil, drinking water, and fish are on a per
person basis for males and females combined. MIRC divides those chemical intakes by an
adult body weight for males and females combined as specified by the user (e.g., 71.4 kg mean
value) to estimate the ADD normalized to body weight from those sources. If the user finds that
those exposure media contribute the majority of the chemical intake for the risk scenario under
consideration, the user may use alternative ingestion rates for those media and alternative body
weights for nursing women, as described in Section 6.5.
Elimination rates for chemicals often are reported as the half-life of the chemical in the body
following a known dose of chemical. Many chemicals exhibit a two-phase elimination process,
the first being more rapid than the second. For screening risks for persistent and
bioaccumulative chemicals, the half-life of the slower phase of elimination, presumably from
non-blood compartments of the body, is the more important of the two. Assuming first-order
kinetics, Equation 3-40 is used to convert a measured half-life for elimination of a chemical for
adults or non-lactating women to an elimination rate constant (EPA 1998). The equation can be
used to estimate any kind of chemical loss rate constant from a measured chemical half-life.
where:
DAImat
ADD(aduit)
AEmat
C-2-35
-------
Equation 3-40. Biological Elimination Rate Constant for Chemicals for Non-lactating
Women
In 2
K —
elim ^
where:
telim
Chemical-specific elimination rate constant for elimination of the chemical for
non-lactating women (per day; e.g., via urine, bile to feces, exhalation)
In2 = Natural log of 2 (unitless constant)
. _ Chemical-specific biological half-life of chemical for non-lactating women
(days)
For chemicals transferred from the body of lactating women to breast milk, the rate of chemical
elimination is augmented by the rate of chemical loss via the milk. The total elimination rate for
lactating women sometimes is measured directly and reported in the literature. Where direct
measurements are not available, and for chemicals that partition predominantly to the lipid-
phase of milk, EPA has used Equation 3-41 to estimate the total chemical elimination rate for
lactating women, kfat_eiac{£PI\ 1998).
Equation 3-41. Biological Elimination Constant for Lipophilic Chemicals for Lactating
Women
/. I* i l^milk **ffX fmbm
fat_elac ~ elim
where:
Vaf elac —
telim
ffm x BWmat
Rate constant for total elimination of chemical during nursing (per day);
accounts for both elimination by adults in general and the additional chemical
elimination via the lipid phase of milk in nursing women
Elimination rate constant for chemical from adults, including non-lactating
women (per day; e.g., via urine, bile to feces, exhalation; chemical-specific;
value from literature or calculated from half-life using Equation 3-40)
IRmiik = Infant milk ingestion rate over the duration of nursing (kg/d)
Fraction of total maternal body burden of chemical that is stored in maternal fat
ff = (mq chemical in body fat / mq chemical total in body; value from literature or
EPA default)
fmbm = Fraction of fat in breast milk (unitless)
ffm = Fraction of maternal body weight that is fat stores (unitless)
gl/l/ _ Maternal body weight over the entire duration of the mother's exposure to the
mat ~ chemical including during pregnancy and lactation (kg)
Equation 3-41 is based on a model from Smith (1987) and accounts for the additional
elimination pathway for lipophilic chemicals via the breast milk fat. The term Kfal_eiac is
estimated by adding an estimate of the first-order elimination constant for breast feeding losses
to keiim, which is the chemical-specific total elimination rate constant for non-lactating women.
The breast feeding losses are estimated from the infant's intake rate of breast milk (IRmiik), the
fraction of the total maternal body burden of the chemical that is stored in maternal body fat (ff),
C-2-36
-------
the fraction of the mother's breast milk that consists of fat (lipids) (fmbm), the mother's body
weight (BWmat), and the fraction of the mother's weight that is body fat (ffm). In Equation 3-41,
the value for the mother's body weight should be specific to women of child-bearing age, as
opposed to a body weight value for both males and females that is used to estimate an adult
average daily dose and the mother's absorbed daily intake in Equation 3-39. Body weight
values for the mother are described in Section 6.5. Smith's (1987) model assumes that the
chemical partitions to the lipid-phase of breast milk to the same degree that it partitions into the
mother's body fat. For highly lipophilic compounds, losses from breast feeding can be larger
than losses by all other pathways (EPA 1998).
3.4.3 Chemical Concentration in Aqueous Phase of Breast Milk
When developing MPE (EPA 1998), EPA also considered models to estimate chemical
concentrations in the aqueous phase of breast milk (CaqUeous)- EPA adapted Smith's (1987)
steady state concentration model for estimating Cmi|kfatand developed the Caqueous model shown
in Equation 3-42 (EPA 1998). Chemicals that would partition to the aqueous phase of human
milk include water-soluble chemicals, such as salts of metals, and other hydrophilic chemicals
that may be in equilibrium with bound forms of the chemical in different tissues. The CaqUeous
equation assumes that the chemical concentration in the aqueous phase of milk is directly
proportional to the chemical concentration in the mother's blood plasma. The portion of
chemical sequestered in red blood cells (e.g., bound to RBC proteins) is assumed to be
unavailable for direct transfer to breast milk.
Equation 3-42. Chemical Concentration in Aqueous Phase of Breast Milk
DAImat x fn. x Pch
C
'mat ~ 'pi bm
aqueous . ,
aq_elac pm
where:
Caqueous = Concentration of chemical in aqueous phase of maternal milk (mg/kg)
n.. _ Daily absorbed maternal chemical dose (mg/kg-day; calculated by Equation
UAIma, - 3 3g)
Fraction of chemical in the body (based on absorbed intake) that is in the
fpi = blood plasma compartment (unitless; value from literature or calculated by
Equation 3-43)
pc _ Partition coefficient for chemical between the plasma and breast milk in the
bm ~ aqueous phase (unitless); assumed to equal 1.0
Chemical-specific rate constant for total elimination of chemical in the aqueous
kaq_eiac = phase of milk during nursing (per day; value from literature or calculated in
Equation 3-44)
fpm = Fraction of maternal weight that is blood plasma (unitless)
Equation 3-42 is a steady-state concentration model that, like the Equation 3-38 for Cmmat, is
dependent on the maternal absorbed daily intake (DAImat). In Equation 3-42, DAImat is multiplied
by the fraction of the absorbed chemical that is circulating in the blood plasma compartment (fpt)
and a partitioning coefficient for the chemical between plasma and the aqueous phase of breast
milk (Pcbm). For highly water-soluble chemicals that are not transported via special carrier
molecules, the chemical is assumed to diffuse passively from the mother's blood serum to the
aqueous phase of her milk, in which case Pcbm would equal 1.0. The denominator includes the
biological elimination constant for the chemical in the aqueous phase of breast milk in lactating
C-2-37
-------
women (kaq_eiac) and the fraction of the mother's weight that is plasma (fp/). Because the model
assumes steady-state, it does not account for chemical species with long half-lives in the body
or for body burden losses due to lactation. These factors are important for highly lipophilic
chemicals and for non-lipophilic chemicals such as methyl mercury, lead, and cadmium that
partition into body compartments such as red blood cells and bone. While these latter
chemicals or forms of these chemicals are water-soluble when free, they have relatively long
half-lives because they are in equilibrium with the chemical bound to macromolecules in some
tissue compartments. Lead is of particular concern because it can be released from the bone
into the blood during lactation, and thus into the breast milk (EPA 2001a). Due to this limitation,
the model may over- or underestimate exposure to the infant.
Because Equation 3-42 is based on the relationship between the chemical concentrations in the
aqueous phase of breast milk and the blood plasma, a value for the fraction of the chemical in
the mother's blood plasma (fpt) is required. Ideally, an empirical value for fpt should be used. If
empirical values are not available, fpi can be estimated from Equation 3-43, provided that an
empirical value can be found for the fraction of the chemical in the body that is in the mother's
whole blood compartment (fft,; EPA 1998).
Equation 3-43. Fraction of Total Chemical in Body in the Blood Plasma Compartment
^ _ ^bl x ^bp
?bp + PCRBC ~~ ^bp )
where:
Fraction of chemical in body (based on absorbed intake) that is in the blood
plasma compartment (unitless); chemical-specific
Fraction of chemical in body (based on absorbed intake) in the whole blood
compartment (unitless); chemical-specific
Fraction of whole blood that is plasma (unitless)
Partition coefficient for chemical between red blood cells and plasma
(unitless); chemical-specific
If the fraction of the total chemical in the body that is in the whole blood compartment (fbj) is
known for a given chemical, then the fraction of that chemical that is in blood plasma depends
only on the partition coefficient for the chemical between the red blood cells and the plasma
(Pcrbc) and the fraction of whole blood that is plasma (fbp).
Another parameter for which a value is needed to solve Equation 3-42 is the total chemical
elimination rate for lactating women for hydrophilic chemicals, kaq_e,ac. As for kfat_eiac for lipophilic
chemicals, kaq_eiac for hydrophilic chemicals would be equal to kenm plus the loss rate for the
chemical in the aqueous phase of breast-milk during lactation. In the case of hydrophilic
chemicals, EPA has yet to propose a term for the additional elimination of a chemical in the
aqueous phase of milk from breast feeding. Given basic physiological mechanisms, we assume
that chemical loss rates via urine are likely to be significantly higher than loss rates from
nursing, however. This is because the counter-current anatomy of kidney tubules allows
substantial concentration of chemicals in the tubules for elimination in urine compared with the
concentration in circulating blood and because of active secretion of some chemicals into urine.
Therefore, the best estimation of elimination of hydrophilic chemicals by lactating women is
simply keum, the elimination of the chemical from a non-lactating woman, as shown in Equation
fpi -
fbi =
fbp =
PcRBC =
C-2-38
-------
3-40. The extent to which keiim is an underestimate of kaq_eiac for a given chemical will determine
the extent of conservative bias in kaq_eiac-
Equation 3-44. Biological Elimination Rate Constant for Hydrophilic Chemicals
Is — Is
rxaq_elac Aelim
where:
Chemical-specific rate constant for total elimination of chemical by lactating
women for hydrophilic chemicals (per day)
Chemical-specific rate constant for total elimination of chemical by non-
lactating women (per day; e.g., via urine, bile to feces, exhalation; value from
literature or calculated from half-life using Equation 3-40)
3.4.4 Alternative Model for Infant Intake of Methyl Mercury
In this version of MIRC, we were unable to fully parameterize the aqueous model for mercury. In
particular, no empirical value could be found for the steady-state fraction of total hydrophilic
chemical body burden in the mother that is in the blood plasma (fpi, see Exhibit 6-20). This
parameter could be estimated using Equation 3-43 if a suitable chemical-specific fraction of
chemical in the body that is in the whole blood (fbi) could be found. However, the value found for
fbi is based on a single-dose study and is not considered reliable for use in chronic exposure
calculations.
We therefore conducted a literature search to identify existing physiologically based
toxicokinetic (PBTK) models of lactational transfer of methylmercury (MeHg) in humans. Most
PBTK models that we identified focused on gestational transfer of mercury between mother and
fetus, including a PBTK dynamic compartmental model for gestational transfer of MeHg in
humans developed by Gearhart et al. (1995, 1996), and reparameterized by Clewell et al.
(1999).
We did find, however, that Byckowski and Lipscomb (2001) had added a lactational transfer
module to the Clewell et al. (1999) model. Byckowski and Lipscomb compared their model's
predictions to epidemiological data from mother-nursing-infant pairs obtained following an
accidental high-dose poisoning in Iraq (Amin-Zaki et al. 1976) and from 34 mother-nursing-
infant pairs examined in a low-dose, chronic exposure environment (Fujita and Takabatake
1977). Using data from the Iraq incident, Byckowski and Lipscomb (2001) found good
agreement between their model's predictions and the clinical data relating MeHg concentrations
in breast milk to MeHg concentrations in infant's blood with time following the poisoning. To
compare their model's predictions to data from chronic exposure to low doses of MeHg,
Byckowski and Lipscomb (2001) simulated MeHg intake for 500 days prior to conception,
continued through gestation, and 6.5 months (200 days) of lactation. Their model's predictions
were consistent with Fujita and Takabatak's (1977) study, although use of hair/blood partition
coefficients based on the results of the 1977 study precluded use of this comparison as model
validation. Both the model predictions and the mean values from the 1977 data indicated that
the concentration of MeHg in the blood of nursing infants was close to the MeHg concentration
in their mothers' blood (approximately 0.025 to 0.027 mg/L, Figure 4 of report). At those blood
concentrations, the PBTK model estimated the average maternal intake of MeHg to be 0.68 ±
0.33 (SD) |jg/kg-day and the average infant intake of MeHg to be 0.80 ± 0.38 |jg/kg-day.
Therefore, for purposes of MIRC, the DAIinf of MeHg is estimated to be the same as the
maternal intake per unit body weight (Equation 3-42).
kaq_elac
kelim
C-2-39
-------
Equation 3-45. Calculation of Infant Average Daily Absorbed Dose of Methyl Mercury
where:
DAIinf_MeHg DAImat_MeHg
DAIinf_MeHg = Average daily dose ofMeHg absorbed by infant from breast milk (mg/kg-day)
n.. _ Average daily dose of methyl mercury absorbed by the mother, predominantly
UAImat_MeHg - f|.om fjsh (mg/kg_day)
C-2-40
-------
4 Dose-Response Values Used for Assessment
Chemical dose-response values included in MIRC include carcinogenic potency slope factors
for ingestion and non-cancer oral reference doses (RfDs) for chronic exposures. The cancer
slope factors (CSFs) and RfDs for chemicals used to calculate PB-HAP emission thresholds are
provided in Exhibit 4-3. Dose-response values in MIRC that are used for EPA's Risk and
Technology Review (RTR) evaluations are consistent with dose-response data that the
Agency's Office of Air Quality Planning and Standards (OAQPS) uses for risk assessments of
hazardous air pollutants (HAPs) (EPA 2007). In general, OAQPS chose these values based on
the following hierarchy of sources: EPA's Integrated Risk Information System (IRIS); the
Centers for Disease Control's Agency for Toxic Substances and Disease Registry (ATSDR);
and the California Environmental Protection Agency (CalEPA).
Exhibit 4-1. Oral Dose-response Values for PB-HAP Chemicals Used to Calculate RTR
De Minimis Thresholds
Cancer Slope Factor
Reference Dose
Chemical
CAS No.
f mg T1
Source
f mg ]
Source
^kg-day J
^ kg - day J
Inorganics
Cadmium compounds in food
7440439
not available
1.0E-03
IRIS
Mercury (elemental)
7439976
NA
not available
Mercuric chloride
7487947
not available
3.0E-04
IRIS
Methyl mercury (MeHg)
22967926
not available
1.0E-04
IRIS
Organics
Benzo(a)pyrene
50328
1.0E+01
EPA OAQPS a
not available
2,3,7,8-TCDD
1746016
1.5E+05
EPA ORD
1.0E-09
ATSDR
ATSDR = Agency for Toxic Substances and Disease Registry IRIS = Integrated Risk Information System
EPA OAQPS = EPA's Office of Air Quality Planning and Standards NA = not applicable
EPA ORD = EPA's Office of Research and Development
aThe method to assign oral cancer slope factors for polycyclic organic matter (POM) is the same as was used in the
1999 National Air Toxics Assessment (EPA 1999b). A complete description of the methodology is available at:
http://www.epa.gov/ttn/atw/nata1999/99pdfs/pomapproachjan.pdf.
Cadmium
EPA has developed two chronic RfDs for cadmium, one for food and one for water, based on
data in IRIS indicating a lower absorption efficiency of cadmium from food than from water. The
default RfD set in MIRC is the higher RfD for cadmium compounds in food (no drinking water is
assumed to occur when calculating de minimis thresholds). Users of MIRC who assess
exposures via drinking water may need to use the RfD for Cd compounds in water (i.e., 5.0E-4
mg/kg-day).
Mercury
EPA's RfD for MeHg of 1E-04 mg/kg-day is based on a Benchmark Dose Lower Limit (BMDL) to
dose-response data from an epidemiological study of neurobehavioral effects in children for
which mercury concentrations had been measured in cord blood at birth. The island
populations included in the study had been exposed for many years to MeHg in their seafoods.
The RfD applies to the pregnant mother as well as young children. EPA has not specified the
C-2-41
-------
minimum exposure duration at the RfD level of exposure that is appropriate to use in
characterizing risk; we assume ten years for women of child-bearing age and 1 year for infants.
We note that human exposures to MeHg are primarily through the consumption offish and
shellfish (EPA 2001 b). EPA found that on average, approximately 76 percent of the exposure to
MeHg for women of childbearing age could be attributed to ingestion of mercury in freshwater
and estuarine fish and shellfish, with the remaining 24 percent derived from marine fish and
shellfish. Other sources accounted for less than 0.06 percent of total exposures (EPA 2001 b).
Dioxins (2,3,7,8-TCDD)
For chemicals for which the critical health effect is developmental, either in utero and/or during
the first months or years of life, the exposure duration and timing of exposure for comparison
with the RfD (or comparable value) require special consideration. The most sensitive health
endpoints for both mercury and 2,3,7,8-TCDD are neurological effects during development that
have long-lasting effects on learning and social behaviors. To ensure a protective risk
characterization for these chemicals, it is important to use the shortest exposure duration
appropriate, at the appropriate life stage, for comparison with the toxicity reference values. This
approach avoids "dilution" of an estimated average ADD that would result from averaging the
lower daily chemical intake rates normalized to body weight for older children and adults with
the potentially higher daily intake rates of infants over a longer exposure averaging period.
For 2,3,7,8-TCDD, although exposures may start in utero, a period of special concern is the
nursing stage, because the highly lipophilic chemical is effectively transferred to the infant in the
lipid phase of its mother's milk. ATSDR has established a minimal risk level (MRL) of 1E-09
mg/kg-day for exposures of the mother or infant of 365 days or longer to 2,3,7,8-TCDD. The
MRL is based on a behavioral study of offspring of female rhesus macaques that were exposed
prior to conception, during gestation, and while nursing (ATSDR 1998). In the critical study, all
mothers were exposed for seven months prior to opportunities for mating; however, dates of
conception ranged over a five month period (some females did not conceive for several cycles).
When the offspring were born, their mothers had been exposed for an average of 16.2 months.
Exposure continued for the 4-month lactation period, after which the offspring were weaned and
tested for non-social and social behavioral deficits (ATSDR 1998, Bowman et al. 1989, Schantz
and Bowman 1989, Schantz et al. 1992). It is not known whether the behavior deficits resulted
from pre- or post-natal exposures or both. In this case, it is appropriate to compare the 365-day
MRL to the ADD for women of child-bearing age and to the ADD for a nursing infant less than 1
year of age during risk characterization (see Section 5).
The convention for assessing risk from mixtures of dioxins is by application of a toxic
equivalency factor (TEF) to dioxin concentrations, which are then expressed as toxic
equivalents (TEQs). Of the dioxin congeners, 2,3,7,8-TCDD is the most widely studied, and
considered to be one of the two most toxic congeners. It is therefore assigned a TEF of one,
with the other dioxin congener TEQ concentrations scaled relative to 2,3,7,8-TCDD
concentrations on the basis of toxicity. The World Health Organization (WHO) 2005 TEFs
presented in Exhibit 4-2 are used for risk assessment of dioxins for RTR.
C-2-42
-------
Exhibit 4-2. WHO 2005 Toxic Equivalency Factors (TEFs) for Dioxins
Dioxin Congener
CAS No.
WHO 2005 Toxic
Equivalency Factor
1,2,3,4,6,7,8-Heptachlorodibenzofuran
67562394
0.01
1,2,3,4,7,8,9-Heptachlorodibenzofuran
55673897
0.01
1,2,3,4,7,8-Hexachlorod-benzofuran
70648269
0.1
1,2,3,6,7,8-Hexachlorodibenzofuran
57117449
0.1
1,2,3,7,8,9-Hexachlorodibenzofuran
72918219
0.1
2,3,4,6,7,8-Hexachlorodibenzofuran
60851345
0.1
1,2,3,4,7,8-Hexachlorodibenzo-p-dioxin
39227286
0.1
1,2,3,6,7,8-Hexachlorodibenzo-p-dioxin
57653857
0.1
1,2,3,7,8,9-Hexachlorodibenzo-p-dioxin
19408743
0.1
1,2,3,4,6,7,8-Heptachlorodibenzo-p-dioxin
35822469
0.01
1,2,3,4,6,7,8,9-Octachlorodibenzofuran
39001020
0.0003
1,2,3,4,6,7,8,9-Octachlorodibenzo-p-dioxin
3268879
0.0003
1,2,3,7,8-Pentachlorodibenzofuran
57117416
0.03
2,3,4,7,8-Pentachlorodibenzofuran
57117314
0.3
1,2,3,7,8-Pentachlorodibenzo-p-dioxin
40321764
1
2,3,7,8-Tetrachlorodibenzofuran
51207319
0.1
2,3,7,8-Tetrachlorodibenzo-p-dioxin
1746016
1
Source: van den Berg et al. 2006
Polycyclic Organic Matter
Previously, for risk assessment of inhalation exposures to polycyclic organic matter (POM) for
EPA's National Air Toxics Assessments (NATA) and for RTR, OAQPS developed an approach
for characterizing risks associated with the individual POM species and POM groups reported in
NEI. Individual PAHs were assigned to one of eight POM groups according to cancer potencies
derived by EPA for IRIS and by CalEPA, and based on assumptions regarding relative
carcinogenicity. OAQPS then estimated an inhalation CSF for each group. The same approach
was used to derive oral CSFs for POMs for use in multipathway risk assessment for RTR. POM
groups (with their member POM species reported in NEI) and the corresponding CSFs used for
RTR risk assessment are presented in Exhibit 4-3. As noted in the main TSD, a de minimis
threshold for non-inhalation risk was derived only for benzo[a]pyrene, and facility emissions
were then screened by comparing the total POM emissions to this threshold (where a toxicity-
weighted sum of POM emissions is calculated for each evaluated source).
C-2-43
-------
Exhibit 4-3. Oral Dose-response Values for Polycyclic Organic Matter
(POM) Groups
Cancer Slope Factor1
Individual POM or POM group
CAS No.
fms r
l^kg-day J
POM Group 71002
Benz(a)Anthracene/Chrysene (7PAH)
103
Total PAH
234
Polycyclic Organic Matter
246
0.5
16-PAH
40
16PAH-7PAH
75040
POM Group 72002
Anthracene
120127
Pyrene
129000
Benzo[g,h,i,]perylene
191242
Benzo[e] pyrene
192972
Benzo(c)phenanthrene
195197
Perylene
198550
Benzo(g,h,i)Fluoranthene
203123
Benzo(a)fluoranthene
203338
Fluoranthene
206440
Acenaphthylene
208968
1-Methylpyrene
2381217
12-Methylbenz(a)Anthracene
2422794
0.5
Methylbenzopyrenes
247
Methylchrysene
248
Methylanthracene
26914181
Benzofluoranthenes
56832736
9-Methylbenz(a) Anthracene
779022
1 -Methylphenanthrene
832699
Acenaphthene
83329
Phenanthrene
85018
Fluorene
86737
2-Methylnaphthalene
91576
2-Chloronaphthalene
91587
POM Group 73002
7,12-Dimethylbenz[a]anthracene
57976
1000
POM Group 74002
Dibenzo[a,i]pyrene
189559
100
D[a,h]pyrene
189640
C-2-44
-------
Exhibit 4-3, continued. Oral Dose-response Values for Polycyclic Organic
Matter (POM) Groups
Individual PAH or PAH group
CAS No.
Cancer Slope Factora
f mg T
l^kg-day J
POM Group 75002
3-Methylcholanthrene
56495
10
D[a,e]pyrene
192654
5-Methylchrysene
3697243
Benzo[a]pyrene
50328
Dibenzo[a,h]anthracene
53703
POM Group 76002
Benzo[b+k]fluoranthene
102
1
lndeno[1,2,3-c,d]pyrene
193395
B[j]fluoranthen
205823
Benzo[b]fluoranthene
205992
Benzo[k]fluoranthene
207089
D[a,j]acridine
224420
Benz[a]anthracene
56553
POM Group 77002
Chrysene
218019
0.1
POM Group 77002
7-PAH
75
0.5
a The method to assign oral cancer slope factors for polycyclic organic matter (POM) is the same as
was used in the 1999 National Air Toxics Assessment (EPA 1999b). A complete description of the
methodology is available at:
http://www.epa.gov/ttn/atw/nata1999/99pdfs/pomapproachjan.pdf.
C-2-45
-------
5 Risk Characterization
For persistent and bioaccumulative hazardous air pollutants (BP-HAPs), risks from inhalation of
a chemical directly from air generally will be negligible compared with risks from ingestion of the
chemical with foodstuffs grown in an area subject to air deposition of the chemical. For other
(non-PB) HAPs, inhalation risks can be estimated separately and compared with risks
associated with ingestion exposure to determine the focus of subsequent tiers of the risk
assessment. Risk characterization for carcinogens with a linear mode of action at low doses is
described in Section 5.1. Risk characterization for chemicals likely to exhibit a threshold for
response (e.g., non-cancer hazards) is described in Section 5.2.
5.1 Cancer Risks
The estimated risk of developing cancer from exposure to a chemical from a specified source is
characterized as the excess lifetime cancer risk (ELCR). The ELCR represents the incremental
probability of an individual developing cancer over a lifetime as a result of lifetime exposure to
the chemical. For a known or suspected carcinogen with a low-dose linear mode of action, the
estimated ELCR is calculated as the product of the lifetime average daily dose (LADD) and the
cancer slope factor (SF):
Equation 5-1. Calculation of Excess Lifetime Cancer Risk
ELCR = LADD x CSF
where:
Estimated excess lifetime cancer risk from a chemical summed across all
exposure pathways and media (unitless)
Lifetime average total daily dose from all exposure pathways and media
(mg/kg-day)
Oral carcinogenic potency slope factor for chemical (per mg/kg-day)
As described in Section 3.3, the LADD (in mg/kg-day) for a chemical is calculated to reflect age-
related differences in exposure rates that are experienced by a hypothetical individual
throughout his or her lifetime of exposure. The total chemical intake is normalized to a lifetime,
which for the purposes of this assessment is assumed to be 70 years.
EPA considers the possibility that children might be more sensitive than adults to toxic
chemicals, including chemical carcinogens (EPA 2005b,c). Where data allow, EPA
recommends development of lifestage-specific cancer potency CSFs. To date, EPA has
developed a separate slope factor for early lifestage exposure for only one chemical (i.e., 1,1,1-
trichloroethane; EPA 2007b), and current data availability for most chemicals preclude this
approach. EPA has, therefore, examined options for default adjustments of the CSF to protect
children. To date, the only mode of action (MOA) for carcinogenesis for which EPA has
adequate data to develop a reasonable quantitative default approach is mutagenesis (EPA
2005b,c). For carcinogens with a mutagenic MOA for cancer, EPA concluded that the
carcinogenic potency of a chemical may be approximately tenfold greater for the first 2 years of
life (i.e., birth up to second birthday) and threefold greater for the next 14 years of life (i.e., ages
2 through 15) than for adults (EPA 2005c). These conclusions are represented by age-
dependent adjustment factors (ADAFs) of 10, 3, and 1 for the first two lifestages and for adults,
respectively.
ELCR =
LADD =
CSF =
C-2-46
-------
These three lifestages do not match the age categories for the home-grown food ingestion
rates, the age categories in MIRC. As a consequence, ADAFs for the age groups in MIRC are
adapted as time-weighted average values as follows:
ADAF,
()
In other words, Equation 5-8 indicates that the total extra lifetime cancer risk (ELCR) equals the
sum of the age-group-specific risks estimated by Equations 5-2 through 5-7, where:
Risk(<1)
Risk(i-2)
Riskp- 5)
Risk(6-ii}
Risk(12-19)
Risk.;. ;
ADD(<1)
ADD (1.2)
ADD(3- 5)
ADD (6.11)
ADD (12-19)
ADDfadult)
CSF
Risk0)
ELCR
n
Risk from chemical ingestion in first year of life
Risk from chemical ingestion from first birthday through age 2 years
Risk from chemical ingestion from age 3 through 5 years of age
Risk from chemical ingestion from age 6 through 11 years of age
Risk from chemical ingestion from age 12 through 19 years of age
Risk from chemical ingestion from age 20 to 70 years age
Average daily dose for infants under one year of age (mg/kg-day)
Average daily dose from first birthday through age 2 years of age (mg/kg-day)
Average daily dose from age 3 through 5 years of age (mg/kg-day)
Average daily dose from age 6 through 11 years of age (mg/kg-day)
Average daily dose from age 12 through 19 years of age (mg/kg-day)
Average daily dose for adults age 20 to 70 years of age (mg/kg-day)
Oral carcinogenic potency slope factor for chemical (per mg/kg-day)
Risk from chemical ingestion for the th age group
Total extra lifetime cancer risk (incremental or extra risk)
Number of age groups (i.e., 6)
C-2-47
-------
5.2 Non-cancer Hazard Quotients
Non-cancer risks are presented as hazard quotients (HQs), that is, the ratio of the estimated
daily intake (i.e., ADD) to the reference dose (e.g., chronic RfD). If the HQ for a chemical is
equal to or less than 1, EPA believes that there is no appreciable risk that non-cancer health
effects will occur. If the HQ is greater than 1, however, EPA cautions that adverse health
effects are possible, although an HQ above 1 does not indicate an effect will definitely occur.
This is because of the margin of safety inherent in the derivation of all RfD values. The larger
the HQ value, the more likely it is that an adverse effect may occur.
5.2.1 Hazard Quotients for Chemicals with a Chronic RfD
For chemicals with a chronic RfD, MIRC calculates an HQ for each age group separately using
Equation 5-9 to indicate the potential for adverse health effects associated with chronic
exposure via ingestion pathways. The HQ is the ratio of a long-term, daily average exposure
normalized to the receptor's body weight (i.e., ADD) to the RfD for that chemical.
Equation 5-9. Hazard Quotient for Chemicals with a Chronic RfD
where:
Hazard quotient for chemical (unitless)
Average daily ingested dose of chemical (mg/kg-day) from all food types and
ingested media for the age group
Chronic oral reference dose for chemical (mg/kg-day)
HQ =
ADD =
RfD =
5.2.2 Hazard Quotients for Chemicals with RfD Based on Developmental Effects
For chemicals for which the toxicity reference value is an RfD based on developmental effects in
infants, children, or young animals, a shorter exposure duration (ED) and averaging time (AT)
may be required. For this type of chemical (e.g., methylmercury, 2,3,7,8-TCDD), the
appropriate ED/AT and sensitive lifestage for exposure may need to be estimated from the
information provided in the critical developmental study(ies) from which the RfD was derived
(e.g., in consultation with the RfD documentation in EPA's IRIS or in a toxicological profile
developed for the chemical). For screening-level risk assessments, however, a conservative
approach is to compare the highest ADD from among the child age categories provided in MIRC
to the RfD.
5.2.3 Hazard Index for Chemicals with RfDs
When conducting screening-level assessments for multiple chemicals, it can be informative to
calculate a hazard index (HI) for toxicologically similar chemicals (EPA 2000). The HI is the
sum of HQs across chemicals as shown in Equation 5-12. As with the HQ, if the HI value is less
than 1, adverse health effects are not expected for that suite of chemicals. If the screening level
HI exceeds 1, however, the risk assessor is advised to evaluate the assumptions of the
screening-level assessment to determine if more realistic local values are available for
parameters that drive risk. In addition, the risk assessor may need to examine the mode of
action (MOA) and target organ(s) for the chemicals with the highest HQs to develop an
appropriate approach to assessing their potential joint action.
C-2-48
-------
where:
Equation 5-10. Hazard Index Calculation
HI = HQ1+HQ2... HQn
HI = Hazard index (unitless)
HQi = Hazard Quotient for chemical 1 (unitless)
HQ2 = Hazard Quotient for chemical 2 (unitless)
HQn = Hazard Quotient for chemical n (unitless)
The HI approach can be appropriate for chemicals with the same MOA and same target organ;
however, MOA often is difficult to determine. An HI usually is "developed for each exposure
route of interest, and for a single toxic effect or for toxicity to a single target organ" (EPA 2000; p
79). If a receptor is exposed to multiple chemicals that affect different target organs or that
operate by different MOAs, and if more than one HQ is close to 1, the risk assessor is advised
to perform a follow-on evaluation of assumptions and to consider whether chemical interactions
may play a role in chemical toxicity (EPA 2000). Exposures to more than one chemical can
result in a greater or lesser toxic response than might be predicted on the basis of one or the
other chemical acting alone (toxicologically independent) or acting in concert (toxicologically
similar chemicals). Users are referred to EPA's Supplementary Guidance for Conducting Health
Risk Assessment of Chemical Mixtures for approaches to assessing the potential for adverse
health effects from exposure to multiple chemicals (EPA 2000).
Note that users of MIRC are responsible for determining how to interpret HQs for multiple
chemicals.
C-2-49
-------
6 Model Input Options
This section describes the input options currently included in MIRC. Required user inputs for
environmental media concentrations and air deposition rates, such as those predicted by
(output of) TRIM.FaTE, are described in Section 6.1. Values for farm-food-chain (FFC)
parameters for specific types of produce and animal products are discussed in Section 6.2.
Options for parameterizing receptor characteristics are described in Section 6.3, including age-
group-specific values for body weight, water ingestion, and food ingestion by food type. Options
for other exposure parameter values in MIRC, such as exposure frequency and loss of chemical
during food preparation and cooking, are provided in Section 6.4.
Where values for chemical-specific parameters are presented, values are presented only for
PB-HAP chemicals currently evaluated using the TRIM-based RTR screening scenario. The
database included with MIRC contains chemical-specific parameter values for a large number of
chemicals, because all of the chemical-specific input data compiled by EPA for use in HHRAP
were uploaded into MIRC. However, only chemicals that are PB-HAPs evaluated for RTR are
discussed in this document, and the HHRAP inputs provided for other chemicals have not been
evaluated. The data presented in this chapter were reviewed and used to develop the set of
modeling defaults used to calculate de minimis screening thresholds for RTR. Note that the
default values used to estimate RTR screening thresholds, and the justification for selecting a
specific value from the data sets described in this chapter, are discussed in Chapter 7.
6.1 Environmental Concentrations
As noted in Section 2, MIRC is intended to estimate exposures and risks to self-sufficient
farming families from airborne chemicals. The tool analyzes one exposure scenario at a time;
therefore, it is best used to evaluate a maximally exposed individual (MEI) or family when MIRC
is used to screen for possible risks.
The following values specific to the air pollutant of concern are required inputs to MIRC:
• a single air concentration (in g/m3);
• the fraction of chemical in the air that is in the vapor phase;
• air-to-surface deposition rates for both vapor- and particle-phase chemical in the air (in
g/m2-yr);
• two fish tissue concentrations, one each for forage and game fish (i.e., fish in TL 3 and
TL 4) (in mg/kg wet weight);
• concentrations in drinking water (in g/L); and
• four chemical concentrations in soil (in |jg/g dry weight), one each for:
1. surface soil in produce growing area,
2. surface soil where livestock feed,
3. root-zone soil in produce growing area, and
4. root-zone soil in livestock feed growing area.
The MIRC software is configured to estimate ingestion exposures via drinking water for a
specified chemical concentration in the drinking water source (e.g., groundwater well).
The user must provide the inputs listed above; no default values are included for these
parameters in MIRC. Media concentrations output by TRIM.FaTE can be entered into the tool
manually from model output files or can be imported. For RTR evaluations, a tool to facilitate
this process was developed using a Microsoft Excel routine written in Visual Basic.
C-2-50
-------
6.2 Farm-Food-Chain Parameter Values
Using the chemical information specified in Section 6.1 above as inputs, MIRC calculates
chemical concentrations in foods that are commonly grown or raised on family farms: exposed
and protected fruits; exposed and protected vegetables; root vegetables; beef; total dairy
products; pork; and poultry and eggs.
6.2.1 List of Farm-Food-Chain (FFC) Parameters
MIRC estimates chemical concentrations in the produce identified above using algorithms from
HHRAP (EPA 2005a) as described in Section 3.2. Parameter values required for these HHRAP
algorithms, including chemical-specific media transfer factors (e.g., soil-to-plant transfer
coefficients) and plant- and animal-specific properties (e.g., plant interception fraction, quantity
of forage consumed by cattle), are included in tables in MIRC. As described in Section 7, the
HHRAP-recommended parameter values are the default values in MIRC; however, these and
other inputs in MIRC can be edited as needed. Exhibit 6-1 describes the parameters that are
included in the algorithms used to estimate chemical concentrations in the farm food categories.
The parameter names and symbols are referenced in this section for plants/produce and animal
products.
Exhibit 6-1. MIRC Parameters Used to Estimate Chemical Concentrations in Farm Foods
Parameter
Description
Units
Plants/Produce
BrAG-produce-DW(i)
Chemical-specific plant/soil chemical bioconcentration
factor for edible portion of aboveground produce type i,
exposed or protected
Unitless (g soil DW / g
produce DW)
BVAG(i)
Chemical-specific air-to-plant biotransfer factor for
aboveground produce type /' for vapor-phase chemical in
air
Unitless ([mg chemical / g
DW plant] / [mg chemical /
9 air])
Fw
Fraction of wet deposition that adheres to plant surfaces;
0.2 for anions, 0.6 for cations and most organics
Unitless
Kds
Chemical-specific soil/water partition coefficient
L soil pore water / kg soil
DW
kP(0
Plant-specific surface loss coefficient for aboveground
exposed produce and animal forage and silage
-1
yr
MAFf),
Moisture adjustment factor for aboveground produce type /'
to convert the chemical concentration estimated for dry-
weight produce to the corresponding chemical
concentration for full-weight fresh produce
Percent water
RCF
Chemical-specific root concentration factor for tubers and
root produce on a wet-weight (WW) basis
L soil pore water/ kg root
WW
RP(i)
Plant-specific interception fraction for the edible portion of
aboveground exposed produce or animal forage and silage
Unitless
TPfi)
Length of plant exposure to deposition per harvest of the
edible portion of aboveground exposed produce or animal
forage and silage
Year
VGAG(i)
Empirical correction factor for aboveground exposed
produce type /' to address possible overestimate of the
diffusive transfer of chemical from the outside to the inside
of bulky produce, such as fruit
Unitless
C-2-51
-------
Exhibit 6-1, continued. MIRC Parameters Used to Estimate Chemical Concentrations in
Farm Foods
Parameter
Description
Units
VGrootveg
Empirical correction factor for belowground produce (i.e.,
tuber or root vegetable) to account for possible
overestimate of the diffusive transfer of chemicals from the
outside to the inside of bulky tubers or roots (based on
carrots and potatoes)
Unitless
YPO)
Plant-specific yield or standing crop biomass of the edible
portion of produce or animal feed
kg produce DW/m2
Animal Products
Bs
Soil bioavailability factor for livestock
Unitless
MF
Chemical-specific mammalian metabolism factor that
accounts for endogenous degradation of the chemical
Unitless
Bdfbeef)
Chemical-specific biotransfer factor for chemical in diet of
cow to chemical in beef on a fresh-wet (FW; equivalent to
WW) basis
mg chemical/kg FW
tissue/mg chemical/day
or day/kg FW tissue
Bd (dairy)
Biotransfer factor in dairy
day/kg FW tissue
Bd(pork)
Biotransfer factor in pork
day/kg FW tissue
Bd (poultry)
Biotransfer factor in poultry
day/kg FW tissue
Bd(eggs)
Biotransfer factor in eggs
day/kg FW tissue
QS(m)
Quantity of soil eaten by animal type m each day
kg/day
QP(i,m)
Quantity of plant feed type /' consumed per animal type m
each day
kg/day
Source: EPA Source: EPA 2005a
DW = dry weight; FW = fresh weight; WW = wet weight
6.2.2 Produce Parameter Values
Exhibit 6-2 and Error! Reference source not found, provide the chemical-specific input values
that are the current defaults for produce FFC food types in MIRC. Exhibit 6-4 presents
additional non-chemical-specific input values for parameters used in the algorithms that
calculate chemical concentrations in produce. Unless otherwise noted, the default parameter
values were obtained from HHRAP. Options for other parameter values are not included in
MIRC at this time; however, the user can overwrite values if appropriate. Refer to HHRAP (EPA
2005a, Chapter 5 and associated appendices) for detailed descriptions of these parameters and
documentation of input values.
C-2-52
-------
Exhibit 6-2. Chemical-Specific Inputs for Produce Parameters
for Chemicals Included in MIRC
Chemical
Fraction of Wet
Deposition (Fw)
(unitless)3
Root
Concentration
Factor (RCF)
(belowground)
(l_/kg)b
Soil-Water
Partition
Coefficient
(Kds)
(L/kg)c
Chemical Air-to-
Plant
Biotransfer
Factor (BvAGm)
(unitless)d
Inorganics
Cadmium compounds
0.6
NA
7.5E+01
NA e
Mercury (elemental)
0.6
NA
1.0E+03
0f
Mercuric chloride
0.6
NA
5.8E+04
1.8E+03
Methyl mercury
0.6
NA
7.0E+03
0f
Organics
Benzo(a)anthracene
0.6
5.7E+03
6.0E+04
1.9E+04
Benzo(a)pyrene
0.6
9.7E+03
1.6E+05
1.2E+05
Benzo(b)fluoranthene
0.6
1.2E+04
1.0E+04
1.7E+03
Benzo(k)fluoranthene
0.6
1.2E+04
1.9E+05
2.1E+05
Chrysene
0.6
5.7E+03
6.0E+04
6.9E+02
Dibenz(a,h)anthracene
0.6
2.3E+04
5.8E+05
3.1E+07
lndeno(1,2,3-cd)
pyrene
0.6
2.8E+04
5.3E+05
3.7E+05
2,3,7,8-TCDD
0.6
4.0E+04
3.9E+04
6.6E+04
Source: EPA 2005a. NA = not applicable.
a 6E-01 is the value for cations and most organic chemicals. As described in HHRAP (EPA 2005a), Appendix B
(available at http://www.epa.gov/osw/hazard/tsd/td/combust/finalmact/ssra/05hhrapapb.pdf), EPA estimated this
value (EPA 1994a, 1995a) from a study by Hoffman et al. (1992) in which soluble gamma-emitting radionuclides
and insoluble particles tagged with gamma-emitting radionuclides were deposited onto pasture grass via simulated
rain. Note that the values developed experimentally for pasture grass may not accurately represent all
aboveground produce-specific values. Also note that values based on the behavior of insoluble particles tagged
with radionuclides may not accurately represent the behavior of organic compounds under site-specific conditions.
b For nonionic organic chemicals, as described in HHRAP (EPA 2005a), Appendix A (available at
http://www.epa.gov/osw/hazard/tsd/td/combust/finalmact/ssra/05hhrapapa.pdf), RCF is used to calculate the
below-ground transfer of contaminants from soil to a root vegetable on a wet-weight basis as shown in Equation
3-6. EPA estimated chemical-specific values for RCF from empirical regression equations developed by Briggs et
al. (1982) based on their experiments measuring uptake of compounds into barley roots from growth solution.
Briggs' regression equations allow calculation of RCF values from log Kow. For metals and mercuric compounds,
empirical values for soil to root vegetable transfer on a dry-weight basis are available from EPA or other sources.
c As discussed in HHRAP (EPA 2005a), Appendix A, Kds describes the partitioning of a compound between soil
pore-water and soil particles and strongly influences the release and movement of a compound into the
subsurface soils and underlying aquifer. Kds values for mercuric compounds were obtained from EPA (1997b).
For all PAHs except for benzo(b)fluoranthene, Kds values were obtained from EPA 2004a. For
benzo(b)fluoranthene and 2,3,7,8-TCDD, Kds values were calculated usinq the correlation equation provided in
EPA 1993.
d As discussed in HHRAP (EPA 2005a), Appendix A, the value for mercuric chloride was obtained from EPA
1997b. Bvag(i) values for PAHs were calculated using the correlation equation derived for azalea leaves as cited in
Bacci et al. (1992), then reducing this value by a factor of 100, as suggested by Lorber (1995), who concluded that
the Bacci factor reduced by a factor of 100 was similar to his own observations in various studies. The 2,3,7,8-
TCDD value was obtained from Lorber and Pinsky (2000).
e It is assumed that metals, with the exception of vapor-phase elemental mercury, do not transfer significantly from
air into leaves.
f Speciation and fate and transport of mercury from emissions suggest that Bvag0 values for elemental and methyl
mercury are likely to be zero (EPA 2005a).
C-2-53
-------
Exhibit 6-3. Chemical-Specific Inputs by Plant Type for Chemicals in MIRC
Compound Name
Plant Part
Plant-Soil Bio-
Concentration
Factor
(BfAG-produce-DWfi))
(unitless)a
Empirical Correction
Factor- Belowground
Produce
(VGrootveg) (unitless)b
Empirical Correction
Factor- Aboveground
Produce
(VGAgo)) (unitless)c
Inorganics
Cadmium compounds
Exp. Fruit
1.3E-01
NA
1.0E+00
Exp. Veg.
1.3E-01
NA
1.0E+00
Forage
3.6E-01
NA
1.0E+00
Grain
6.2E-02
NA
NA
Prot. Fruit
1.3E-01
NA
NA
Prot. Veg.
1.3E-01
NA
NA
Root
6.4E-02
1.0E+00
NA
Silage
3.6E-01
NA
5.0E-01
Mercury (elemental)
Exp. Fruit
NA
NA
1.0E+00
Exp. Veg.
NA
NA
1.0E+00
Forage
NA
NA
1.0E+00
Grain
NA
NA
NA
Prot. Fruit
NA
NA
NA
Prot. Veg.
NA
NA
NA
Root
NA
1.0E+00
NA
Silage
NA
NA
5.0E-01
Mercuric chloride
Exp. Fruit
1.5E-02
NA
1.0E+00
Exp. Veg.
1.5E-02
NA
1.0E+00
Forage
0
NA
1.0E+00
Grain
9.3E-03
NA
NA
Prot. Fruit
1.5E-02
NA
NA
Prot. Veg.
1.5E-02
NA
NA
Root
3.6E-02
1.0E+00
NA
Silage
0
NA
5.0E-01
Methyl mercury
Exp. Fruit
2.9E-02
NA
1.0E-02
Exp. Veg.
2.9E-02
NA
1.0E-02
Forage
0
NA
1.0E+00
Grain
1.9E-02
NA
NA
Prot. Fruit
2.9E-02
NA
NA
Prot. Veg.
2.9E-02
NA
NA
Root
9.9E-02
1.0E-02
NA
Silage
0
NA
5.0E-01
C-2-54
-------
Exhibit 6-3, continued. Chemical-Specific Inpu
ts by Plant Type for Chemicals in MIRC
Compound Name
Plant Part
Plant-Soil Bio-
Concentration
Factor
(Bra G-produce-DW(i))
(unitless)8
Empirical Correction
Factor- Belowground
Produce
(VGrootveg) (unitless)b
Empirical Correction
Factor- Aboveg round
Produce
(VGag(o) (unitless)c
Organics
Benzo(a)anthracene
Exp. Fruit
2.0E-02
NA
1.0E-02
Exp. Veg.
2.0E-02
NA
1.0E-02
Forage
2.0E-02
NA
1.0E+00
Grain
2.0E-02
NA
NA
Prot. Fruit
2.0E-02
NA
NA
Prot. Veg.
2.0E-02
NA
NA
Root
9.5E-02
1.0E-02
NA
Silage
2.0E-02
NA
5.0E-01
Benzo(a) pyrene
Exp. Fruit
1.3E-02
NA
1.0E-02
Exp. Veg.
1.3E-02
NA
1.0E-02
Forage
1.3E-02
NA
1.0E+00
Grain
1.3E-02
NA
NA
Prot. Fruit
1.3E-02
NA
NA
Prot. Veg.
1.3E-02
NA
NA
Root
6.1E-02
1.0E-02
NA
Silage
1.3E-02
NA
5.0E-01
Benzo(b)fluoranthene
Exp. Fruit
1.1E-02
NA
1.0E-02
Exp. Veg.
1.1E-02
NA
1.0E-02
Forage
1.1E-02
NA
1.0E+00
Grain
1.1E-02
NA
NA
Prot. Fruit
1.1E-02
NA
NA
Prot. Veg.
1.1E-02
NA
NA
Root
1.2E+00
1.0E-02
NA
Silage
1.1E-02
NA
5.0E-01
Benzo(k)fluoranthene
Exp. Fruit
1.2E-02
NA
1.0E-02
Exp. Veg.
1.2E-02
NA
1.0E-02
Forage
1.2E-02
NA
1.0E+00
Grain
1.2E-02
NA
NA
Prot. Fruit
1.2E-02
NA
NA
Prot. Veg.
1.2E-02
NA
NA
Root
6.1E-02
1.0E-02
NA
Silage
1.2E-02
NA
5.0E-01
Chrysene
Exp. Fruit
2.0E-02
NA
1.0E-02
Exp. Veg.
2.0E-02
NA
1.0E-02
Forage
2.0E-02
NA
1.0E+00
Grain
2.0E-02
NA
NA
Prot. Fruit
2.0E-02
NA
NA
Prot. Veg.
2.0E-02
NA
NA
Root
9.5E-02
1.0E-02
NA
Silage
2.0E-02
NA
5.0E-01
C-2-55
-------
Exhibit 6-3, continued. Chemical-Specific Inputs by Plant Type for Chemicals in MIRC
Compound Name
Plant Part
Plant-Soil Bio-
Concentration
Factor
(Bra G-produce-DWfi))
(unitless)a
Empirical Correction
Factor- Belowground
Produce
(VGrootveg) (unitless)b
Empirical Correction
Factor- Aboveground
Produce
(VGago)) (unitless)c
Dibenz(a,h)anthracene
Exp. Fruit
6.8E-03
NA
1.0E-02
Exp. Veg.
6.8E-03
NA
1.0E-02
Forage
6.8E-03
NA
1.0E+00
Grain
6.8E-03
NA
NA
Prot. Fruit
6.8E-03
NA
NA
Prot. Veg.
6.8E-03
NA
NA
Root
4.1E-02
1.0E-02
NA
Silage
6.8E-03
NA
5.0E-01
lndeno(1,2,3-cd)pyrene
Exp. Fruit
5.9E-03
NA
1.0E-02
Exp. Veg.
5.9E-03
NA
1.0E-02
Forage
5.9E-03
NA
1.0E+00
Grain
5.9E-03
NA
NA
Prot. Fruit
5.9E-03
NA
NA
Prot. Veg.
5.9E-03
NA
NA
Root
5.3E-02
1.0E-02
NA
Silage
5.9E-03
NA
5.0E-01
2,3,7,8-TCDD
Exp. Fruit
4.6E-03
NA
1.0E-02
Exp. Veg.
4.6E-03
NA
1.0E-02
Forage
4.6E-03
NA
1.0E+00
Grain
4.6E-03
NA
NA
Prot. Fruit
4.6E-03
NA
NA
Prot. Veg.
4.6E-03
NA
NA
Root
1.0E+00
1.0E-02
NA
Silage
4.6E-03
NA
5.0E-01
Source: EPA 2005a. NA = not applicable.
a As discussed in HHRAP (EPA 2005a), the BrAG-produce-owo) for aboveground produce and forage accounts for the
uptake from soil and the subsequent transport of contaminants through the roots to the aboveground plant parts.
For organics, correlation equations to calculate values for Br on a dry weight basis were obtained from Travis and
Arms (1988). For cadmium, Srvalues were derived from uptake slope factors provided in EPA 1992. Uptake
slope is the ratio of contaminant concentration in dry weight plant tissue to the mass of contaminant applied per
hectare soil. Br aboveground values for mercuric chloride and methyl mercury were calculated using methodology
and data from Baes, et al. (1984). Br forage values for mercuric chloride and methyl mercury (on a dry weight
basis) were obtained from EPA 1997b. The FIFIRAP methodology assumes that elemental mercury doesn't
deposit onto soils. Therefore, it's assumed that there is no plant uptake through the soil.
b As discussed in FIFIRAP (EPA 2005a), Appendix B, VGrootveg represents an empirical correction factor that
reduces produce concentration. Because of the protective outer skin, size, and shape of bulky produce, transfer of
lipophilic chemicals (i.e., log Kow greater than 4) to the center of the produce is not likely. In addition, typical
preparation techniques, such as washing, peeling, and cooking, further reduce the concentration of the chemical in
the vegetable as consumed by removing the high concentration of chemical on and in the outer skin, leaving the
flesh with a lower concentration than would be the case if the entire vegetable were pureed without washing. For
belowground produce, FIFIRAP (EPA 2005a) recommends using a VGr00tveg value of 0.01 for PB-FIAP with a log
Kow greater than 4 and a value of 1.0 for PB-FIAP with a log Kow less than 4 based on information provided in
EPA 1994b. In developing these values, EPA (1994b) assumed that the density of the skin and the whole
vegetable are equal (potentially overestimating the concentration of PB-FIAP in belowground produce due to root
uptake).
C-2-56
-------
Exhibit 6-3, continued. Chemical-Specific Inputs by Plant Type for Chemicals in MIRC
c As discussed in HHRAP (EPA 2005a), Appendix B, VGag represents an empirical correction factor that reduces
aboveground produce concentration and was developed to estimate the transfer of PB-HAP into leafy vegetation
versus bulkier aboveground produce (e.g., apples). Because of the protective outer skin, size, and shape of bulky
produce, transfer of lipophilic PB-HAP (log Kow greater than 4) to the center of the produce is not likely. In
addition, typical preparation techniques, such as washing, peeling, and cooking, further reduces residues. For
aboveground produce, HHRAP (EPA 2005a) recommends using a VGag value of 0.01 for PB-HAP with a log Kow
greater than 4 and a value of 1.0 for PB-HAP with a log Kow less than 4 based on information provided in EPA
1994b. In developing these values, EPA (1994b) assumed the following: (1) translocation of compounds
deposited on the surface of aboveground vegetation to inner parts of aboveground produce would be insignificant
(potentially underestimating the concentration of PB-HAP in aboveground produce due to air-to-plant transfer); (2)
the density of the skin and the whole vegetable are equal (potentially overestimating the concentration of PB-HAP
in aboveground produce due to air-to-plant transfer); and (3) the thickness of vegetable skin and broadleaf tree
skin are equal (effects on the concentration of PB-HAP in aboveground produce due to air-toplant transfer
unknown).
For forage, HHRAP recommends a VGag value of 1.0, also based on information provided in EPA 1994b.
A VGag value for silage is not provided in EPA 1994b; the VGag value for silage of 0.5 was obtained from NC
DEHNR (1997); however, NC DEHNR does not present a specific rationale for this recommendation. Depending
on the composition of the site-specific silage, this value may under- or overestimate the actual value.
C-2-57
-------
Exhibit 6-4. Non-Chemical-Specific Produce Inputs
Plant Part
Interception
Fraction
(RPm)
(unitless)a
Plant
Surface
Loss
Coefficient
(frPrn) h
(1/year)b
Length of
Plant
Exposure to
Deposition
(TPm)
(year)c
Yield or
Standing
Crop
Biomass
(Yp(n)
(kg/m )d
Plant Tissue-
Specific
Moisture
Adjustment
Factor (MAFm)
(percent)6
Exposed Vegetable
0.982
18
0.16
5.66
92
Protected Fruit
NA
NA
NA
NA
90
Protected
Vegetable
NA
NA
NA
NA
80
Forage (animal
feed)
0.5
18
0.12
0.24
92
Exposed Fruit
0.053
18
0.16
0.25
85
Root Vegetables
NA
NA
NA
NA
87
Silage (animal feed)
0.46
18
0.16
0.8
92
Grain (animal feed)
NA
NA
NA
NA
90
Source: EPA 2005a. NA = not applicable.
a Baes et al. (1984) used an empirical relationship developed by Chamberlain (1970) to identify a correlation
between initial Rp values and pasture grass productivity (standing crop biomass [Yp]) to calculate Rp values for
exposed vegetables, exposed fruits, forage, and silage. Two key uncertainties are associated with using these
values for Rp-. (1) Chamberlain's(1970) empirical relationship developed for pasture grass may not accurately
represent aboveground produce. (2) The empirical constants developed by Baes et al. (1984) for use in the
empirical relationship developed by Chamberlain (1970) may not accurately represent the site-specific mixes of
aboveground produce consumed by humans or the site-specific mixes of forage or silage consumed by livestock.
b The term kp is a measure of the amount of chemical that is lost to natural physical processes (e.g., wind, water)
over time. The HHRAP-recommended value of 18 yr1 (also recommended by EPA 1994a and 1998) represents
the midpoint of a range of values reported by Miller and Hoffman (1983). There are two key uncertainties
associated with using these values for kp\ (1) The recommended equation for calculating kp includes a conservative
bias in that it does not consider chemical degradation processes. (2) Given the reported range of kp values from
7.44 to 90.36 yr"1, plant concentrations could range from about 1.8 times higher to about 5 times lower than the
plant concentrations estimated in FFC media using the midpoint kp value of 18.
c HHRAP (EPA 2005a) recommends using a Tp value of 0.16 years for aboveground produce and cattle silage.
This is consistent with earlier reports by EPA (1994a, 1998) and NC DEHNR (1997), which recommended treating
Tp as a constant based on the average period between successive hay harvests. Belcher and Travis (1989)
estimated this period at 60 days. Tp is calculated as 60 days 365 days/year = 0.16 years. For forage, the
average of the average period between successive hay harvests (60 days) and the average period between
successive grazing (30 days) is used (that is, 45 days), and Tp is calculated as (60 days + 30 days)/ 2 365
days/yr = 0.12 yr. Two key uncertainties are associated with use of these values for Tp-. (1) The average period
between successive hay harvests (60 days) may not reflect the length of the growing season or the length between
successive harvests for site-specific aboveground produce crops. The concentration of chemical in aboveground
produce due to direct (wet and dry) deposition (Pd) will be underestimated if the site-specific value of Tp is less than
60 days, or overestimated if the site-specific value of Tp is more than 60 days.
d Yp values for aboveground produce and forage were calculated using an equation presented in Baes et al. (1984)
and Shor et al. (1982): Yp = Yhil Ahi, where Yhi= Flarvest yield of crop (kg DW) and Ahi = Area planted to ith crop
(m2), and using values for Yh and Ah from USDA (1994b and 1994c). A production-weighted U.S. average Yp of
0.8 kg DW/m2 for silage was obtained from Shor et al. 1982.
e MAF represents the plant tissue-specific moisture adjustment factor to convert dry-weight concentrations into wet-
weight concentrations (which are lower owing to the dilution by water compared with dry-weight concentrations).
Values obtained from Chapter 10 of EPA's 2003 SAB Review materials for 3MRA Modeling System, Volume II,
"Farm Food Chain and Terrestrial Food Web Data" (EPA 2003a), which references EPA 1997c. Note that the value
for grain used as animal feed is based on corn and soybeans, not seed grains such as barley, oats, or wheat.
C-2-58
-------
6.2.3 Animal Product Parameter Values
MIRC also requires chemical-specific inputs for many of the animal product algorithms. The
relevant values are shown in Exhibit 6-5 for the chemicals included in MIRC to date. The
HHRAP algorithms require additional inputs for the animal products calculations that are not
specific to PB-HAPs, but are specific to the animal and animal product type. The soil and plant
ingestion rates recommended in HHRAP for beef cattle, dairy cattle, swine, and chicken are
provided in Exhibit 6-6.
Exhibit 6-5. Animal Product Chemical-specific Inputs for Chemicals Included in MIRC
Soil Bio-
_ ... Availability
Compound Name Factor(Bs»
Biotransfer Factors (Bam) (day/kg FW tissue)a
and Metabolism Factors (MF) (unitless) b
Mammal
Non-mammal
(umtiess)
Beef
[BSbeef)
Dairy
{Bd faiiy)
Pork
(SSporit)
MF
Eggs
{BdeqqS)
Poultry
(Bdpoultrv)
MF
Cadmium compounds
1
1.2E-04
6.5E-06
1.9E-04
1
2.5E-03
1.1E-01
NA
Mercury (elemental)
1
0
0
0
1
0
0
NA
Mercuric chloride
1
1.1E-04
1.4E-06
3.4E-05
1
2.4E-02
2.4E-02
NA
Methyl mercury
1
1.2E-03
1.7E-05
5.1E-06
1
3.6E-03
3.6E-03
NA
Benzo(a)anthracene
1
4.0E-02
8.4E-03
4.8E-02
0.01
1.7E-02
2.9E-02
NA
Benzo(a)pyrene
1
3.8E-02
7.9E-03
4.5E-02
0.01
1.6E-02
2.8E-02
NA
Benzo(b)fluoranthene
1
3.6E-02
7.6E-03
4.4E-02
0.01
1.5E-02
2.7E-02
NA
Benzo(k)fluoranthene
1
3.6E-02
7.7E-03
4.4E-02
0.01
1.5E-02
2.7E-02
NA
Chrysene
1
4.0E-02
8.4E-03
4.8E-02
0.01
1.7E-02
2.9E-02
NA
Dibenz(a,h)anthracene
1
3.1E-02
6.5E-03
3.7E-02
0.01
1.3E-02
2.3E-02
NA
lndeno(1,2,3-cd)
pyrene
1
2.9E-02
6.2E-03
3.6E-02
0.01
1.2E-02
2.2E-02
NA
2,3,7,8-TCDD
1
2.6E-02
5.5E-03
3.2E-02
1
1.1E-02
1.9E-02
NA
Source: EPA 2005a, unless otherwise indicated. NA = not applicable.
a As discussed in HHRAP (EPA 2005a), Appendix A, biotransfer factors for mercury compounds were obtained
from EPA 1997b. Considering speciation, fate, and transport of mercury from emission sources, elemental
mercury is assumed to be vapor-phase and hence is assumed not to deposit to soil or transfer into aboveground
plant parts. As a consequence, there is no transfer of elemental mercury into animal tissues. Biotransfer factors
for cadmium compounds were obtained from EPA 1995b. Biotransfer factors for 2,3,7,8-TCDD and PAHs were
calculated from chemical octanol-water partitioning coefficients (Kow values) using the correlation equation from
RTI (2005) and assuming the following fat contents: milk - 4%; beef -19%; pork - 23%; poultry -14%; and eggs -
8%.
b As discussed in HHRAP (EPA 2005a), EPA (1995c) recommends using a metabolism factor (MF) to account for
metabolism of PAHs by mammals to offset the amount of bioaccumulation suggested by biotransfer factors. EPA
has recommended an MF of 0.01 for bis(2-ethylhexyl)phthalate (BEHP) and 1.0 for all other chemicals (EPA
1995d). For MIRC, an MF of 0.01 is also used to calculate concentrations of PAHs in food products from
mammalian species based on the work of Hofelt et al. (2001). This factor takes into account the P450-mediated
metabolism of PAHs in mammals; applying this factor in our approach reduced the concentrations of chemicals in
beef, pork, and dairy by two orders of magnitude.
C-2-59
-------
Exhibit 6-6. Soil and Plant Ingestion Rates for Animals
Animal
Soil Ingestion Rate -
QS(m) (kg/day)a
Plant Part Consumed
by Animal
Plant Ingestion Rate -
QPo.m) (kg/day)
Beef cattleb
0.5
Silage
2.5
Forage
8.8
Grain
0.47
Dairy cattle c
0.4
Silage
4.1
Forage
13.2
Grain
3.0
Swined
0.37
Silage
1.4
Grain
3.3
Chicken (eggs)e
0.022
Grain
0.2
Source: EPA 2005a HHRAP (Chapter 5).
a Beef cattle: NC DEHNR (1997) and EPA (1994b) recommended a soil ingestion rate for subsistence beef cattle
of 0.5 kg/day based on Fries (1994) and NAS (1987). As discussed in HHRAP, Fries (1994) reported soil
ingestion to be 4 percent of the total dry matter intake. NAS (1987) cited an average beef cattle weight of 590 kg,
and a daily dry matter intake rate (non-lactating cows) of 2 percent of body weight. This results in a daily dry
matter intake rate of 11.8 kg DW/day and a daily soil ingestion rate of about 0.5 kg/day.
Dairy cattle: NC DEHNR (1997) and EPA (1994b) recommended a soil ingestion rate for dairy cattle of 0.4 kg/day
based on Fries (1994) and NAS (1987). As discussed in HHRAP, Fries (1994) reported soil ingestion to be 2
percent of the total dry matter intake. NAS (1987) cited an average beef cattle weight of 630 kg and a daily dry
matter intake rate (non-lactating cows) of 3.2 percent of body weight. This resulted in a daily dry matter intake
rate of 20 kg/day DW, and a daily soil ingestion rate of approximately 0.4 kg/day. Uncertainties associated with
Qs include the lack of current empirical data to support soil ingestion rates for dairy cattle and the assumption of
uniform contamination of soil ingested by cattle.
Swine: NC DEHNR (1997) recommended a soil ingestion rate for swine of 0.37, estimated by assuming a soil
intake that is 8% of the plant ingestion rate of 4.3 kg DW/day. Uncertainties include the lack of current empirical
data to support soil ingestion rates and the assumption of uniform contamination of the soil ingested by swine.
Chicken: HHRAP (EPA 2005a) assumes that chickens consume 10 percent of their total diet (which is
approximately 0.2 kg/day grain) as soil, a percentage that is consistent with the study from Stephens et al. (1995).
Uncertainties include the lack of current empirical data to support soil ingestion rates for chicken and the
assumption of uniform contamination of soil ingested by chicken.
b The beef cattle ingestion rates of forage, silage, and grain are based on the total daily intake rate of about 12 kg
DW/day (based on NAS [1987] reporting a daily dry matter intake that is 2 percent of an average beef cattle body
weight of 590 kg) and are supported by NC DEHNR (1997), EPA (1994b and 1990), and Boone et al. (1981). The
principal uncertainty associated with these Qp values is the variability between forage, silage, and grain ingestion
rates for cattle.
c The dairy cattle ingestion rates of forage, silage, and grain are based on the total daily intake rate of about 20 kg
DW/day (NAS 1987; EPA 1992) as recommended by NC DEHNR (1997). Uncertainties include the proportion of
each food type in the diet, which varies from location to location. Assuming uniform contamination of plant
materials consumed by cattle also introduces uncertainty.
d Swine are not grazing animals and are assumed not to eat forage (EPA 1998). EPA (1994b and 1998) and NC
DEHNR (1997) recommended including only silage and grains in the diet of swine. EPA (1995c) recommended
an ingestion rate of 4.7 kg DW/day for a swine, referencing NAS (1987). Assuming a diet of 70 percent grain and
30 percent silage (EPA 1990), HHRAP estimated ingestion rates of 3.3 kg DW/day (grain) and 1.4 kg DW/day
(silage). Uncertainties associated with Qp include variability of the proportion of grain and silage in the diet, which
varies from location to location.
e Chickens consume grain provided by the farmer. The daily quantity of grain feed consumed by chicken is
assumed to be 0.2 kg/day (Ensminger [1980], Fries [1982], and NAS [1987]). Uncertainties associated with this
variable include the variability of actual grain ingestion rates from site to site. In addition, assuming uniform
contamination of plant materials consumed by chicken introduces some uncertainty.
C-2-60
-------
6.3 Adult and Non-Infant Exposure Parameter Values
The exposure parameters included in MIRC and their default and other value options are
summarized in the following subsections. The default values were selected to result in a highly
conservative screening scenario. Parameter value options were primarily obtained or estimated
from EPA's Exposure Factors Handbook (EFH; EPA 1997a) and Child-Specific Exposure
Factors Handbook (CSEFH; EPA 2008a). Where values were reported for age groupings other
than those used in MIRC (see Section 2.3 above for MIRC age groups), time-weighted average
values were estimated for the MIRC age groups from the available data.
In MIRC, ingestion rates for home-produced farm food items are included for exposed fruit,
protected fruit, exposed vegetables, protected vegetables, root vegetables, beef, total dairy,
pork, poultry, and eggs. Those ingestion rates are already normalized to body weight (i.e., gwet
weight/kg-day), as presented in the original data analysis (EPA 1997a). The body weight
parameter values presented in Exhibit 6-7, therefore, are not applied in the chemical intake
(ADD) equations for these food types.
In MIRC, ingestion rates also are included for drinking water (mL/day), soil (mg/day), and fish
(g/day). These ingestion rates, however, are on a per person basis (i.e., not normalized for
body weight). The body weight parameter values presented in Exhibit 6-7, therefore, are
applied in the chemical intake (ADD) equations for these media.
6.3.1 Body Weights
Body weight (BW) options included in MIRC include mean, 5th, 10th, 50th, 90th, and 95th
percentiles for adults and the five children's age groups. For its default screening assessment,
EPA uses the mean BW for each age group. The BWs currently in the MIRC database are
listed in Exhibit 6-7. For adults, BW represents the weighted average of male and female mean
body weights for all races, ages 18-74 years, from EPA's 1997 EFH (EPA 1997a; Tables 7-4
and 7-5). In general, BW values for the five children's age groups were calculated from the
summary data provided in Table 8-3 of EPA's 2008 CSEFH. For purposes of comparison,
alternative BW values for children ages 12 through 19 years also were estimated using data
from Portier et al. (2007). These values are listed in the last row of Exhibit 6-7, but are not
included in MIRC. The means calculated using the two methods for children ages 12 through
19 years were essentially identical at 64 kg. The other percentile values differed by
approximately 10 percent or less.
C-2-61
-------
Exhibit 6-7. Mean and Percentile Body Weight Estimates
for Adults and Children
Lifestage
Duration
Body Weight (kg)
(years)
(years)
Mean
5tn
10tn
50tn
90tn
95tn
Adulta (20-70)
50
71.4
52.9
56.0
69.3
89.7
97.6
Child < 1 b
1
7.83
6.03
6.38
7.76
9.24
9.66
Child 1-2 c
2
12.6
9.9
10.4
12.5
14.9
15.6
Child 3-5 d
3
18.6
13.5
14.4
17.8
23.6
26.2
Child 6-11 e
6
31.8
19.7
21.3
29.3
45.6
52.5
Child 12-19 f
8
64.2
39.5
45.0
64.2
83.5
89.0
[Child 12-19 9
8
64.3
41.1
44.6
60.9
88.5
98.4]
a BW represents the sample-size weighted average of male and female mean body weights (all races, 18-74
years) from EPA's 1997 EFH (Tables 7-4 for males and 7-5 for females). Note that these weights include the
weight of clothing, estimated to range from 0.09 to 0.28 kg. Although the 18 to 74 year age category in EPA's EFH
does not match exactly the age 20 to 70 year categorization of adults in MIRC, the magnitude of error in the mean
and percentile body weights is likely to be very small (i.e., less than 1 %).
b Each BW represents a time-weighted average of body weights for age groups birth to <1 month, 1 to <3 months,
3 to <6 months, and 6 to <12 months from Table 8-3 of the 2008 CSEFH. Original sample sizes for each of these
age groups can also be found in Table 8-3.
c Each BW represents a time-weighted average of body weights for age groups 1 to <2 years and 2 to <3 years
from Table 8-3 of the 2008 CSEFH. Original sample sizes for each of these age groups can also be found in
Table 8-3.
d BWs obtained directly from Table 8-3 of the 2008 CSEFH (age group 3 to <6 years).
e BWs obtained directly from Table 8-3 of the 2008 CSEFH (age group 6 to <11 years). This value represents a
conservative (i.e., slightly low) estimate of BW for ages 6 through 11 years since 11 -year olds are not included in
this CSEFH age group.
fMean BW estimated using Table 8-22 of the 2008 CSEFH, which is based on NHANES IV data as presented in
Portier et al. (2007). This estimate was calculated as the average of the 8 single-year age groups from 12 to 13
years through 19 to 20 years. Values for the other percentiles were estimated using Portier et al., 2007.
9 Each BW represents a time-weighted average of body weights for age groups 11 to <16 years and 16 to <21
years from Table 8-3 of the 2008 CSEFH. Note that estimated values include 11-year-olds and individuals through
age 20, which contributes to uncertainty in the estimates for 12 to 19 years. Those values are provided for
comparison purposes only and are not included in MIRC.
6.3.2 Water Ingestion Rates
MIRC also includes the option of calculating chemical ingestion via drinking water obtained from
surface-water sources or from wells (i.e., from groundwater) in the contaminated area. Users
have the option in MIRC to set drinking water ingestion rates to zero or to revise the drinking
water ingestion rates in MIRC to better reflect site-specific water uses. The 2008 CSEFH
recommends values for drinking water ingestion rates for children based on a study reported by
Kahn and Stralka (2008). Table 3-4 of the CSEFH provides per capita estimates of community
water ingestion rates by age categories. Community water ingestion includes both direct and
indirect ingestion of water from the tap. Direct ingestion is defined as direct consumption of
water as a beverage, while indirect ingestion includes water added during food or beverage
preparation. The source of these data is the 1994-1996 and 1998 U.S. Department of
Agriculture's (USDA's) Continuing Survey of Food Intakes by Individuals (CSFII) (USDA 2000).
Exhibit 6-8 includes the drinking water ingestion rates for children that are included in MIRC.
Mean and percentile adult drinking water ingestion rates were obtained from EPA (2004b),
which presents estimated per capita water ingestion rates for various age categories based on
C-2-62
-------
data collected by the USDA's 1994-1996 and 1998 CSFII (USDA 2000). Adult ingestion rates,
presented in Exhibit 6-8, represent community water ingestion, both direct and indirect as
defined above, for males and females combined, ages 20 years and older.
Exhibit 6-8. Estimated Daily Per Capita Mean and Percentile Water Ingestion Rates for
Children and Adults a
Lifestage (years)
Ingestion Rates, Community Water (mL/day)
Mean
50th
90th
95th
99th
Child <1 D
324
146
866 *
1,011 *
1,377 *
Child 1-2 c
294
217
654
857
1,290 *
Child 3-5 0
380
291
834
1,078
1,654
Child 6-11 e
447
350
980
1,235
1,870 *
Child 12-19 r
697
516
1,537
2,022 *
3,195 *
Adult9
1,098
920
2,224
2,801
4,488
Sources: EPA 2004b, 2008a
* The sample size does not meet minimum reporting requirements as described in EPA 2008a. For some of these
MIRC age groupings, the values are based on the time-weighted average value for 2 or more age ranges from
CSEFH Table 3-4. One or more age ranges within the group may not meet the minimum reporting requirements,
but not necessarily all of them fall within this category.
a Source is Kahn and Stralka 2008, also presented in the CSEFH (EPA 2008a).
b Each IR represents a time-weighted average of ingestion rates for age groups birth to <1 month, 1 to <3 months,
3 to <6 months, and 6 to <12 months from Table 3-4 of the 2008 CSEFFI.
c Each IR represents a time-weighted average of ingestion rates for age groups 1 to <2 years and 2 to <3 years
from Table 3-4 of the 2008 CSEFFI.
d Each IR represents the ingestion rate for age group 3 to <6 years from Table 3-4 of the 2008 CSEFFI.
e Each IR represents the ingestion rate for age group 6 to <11 years from Table 3-4 of the 2008 CSEFFI. This
value represents a conservative (i.e., slightly low) estimate of IR for ages 6 through 11 years since 11 -year olds
are not included in this CSEFFI age group.
f Each IR represents a time-weighted average of ingestion rates for age groups 11 to <16 years, 16 to <18, and 18
to <21 years from Table 3-4 of the 2008 CSEFFI. Note that estimated values include 11-year-olds and individuals
through age 20, which contributes to uncertainty in the estimates for 12 to 19 years.
9 Adult drinking water ingestion rates were obtained from EPA (2004b), Appendix E, Part I, Table A1 for community
water, both sexes (ages 20+), direct plus indirect water ingestion.
6.3.3 Local Food Ingestion Rates
MIRC includes mean, median, 90th, 95th, and 99th percentile food-specific ingestion rates (IRs)
for consumers-only of farm food chain (FFC) media for adults and children. The mean and
percentile values are from EPA's analysis of data from the USDA's 1987 to 1988 Nationwide
Food Consumption Survey (NFCS) (USDA 1993), as presented in Chapter 13 of the Agency's
Exposure Factors Handbook (i.e., Intake Rates for Various Home Produced Food Items) (EPA
1997a). Consumers-only means that individuals who did not report eating a specified type of
food during the three-day period covered by the food ingestion part of the survey were not
included in the analysis of ingestion rates for that food type. The questionnaire included the
options for a household to self-identify in one or more of five categories: as a household that
gardens, raises animals, hunts, fishes, or farms. As of September, 2008, that survey was the
most recent NFCS available (EPA 2008a, CSEFH), and we are not aware of any that might be
more recent.2
For the adult age group in MIRC, we compiled data on food-specific IRs separately for two types
of households as indicated in the "Response to Questionnaire" (EPA 1997a, Chapter 13): (1)
households that farm (F) and (2) households that garden or raise animals (HG for homegrown).
2 Note that EPA's 2008 CSEFH does not distinguish between exposed and protected fruits and vegetables when
recommending food ingestion rates based on the same data set for the same age categories. EPA's 1997 analysis
for its EFH therefore remains the most appropriate data source for use in MIRC.
C-2-63
-------
This division reflects EPA's data analysis. EPA tabulated IRs for fruits and vegetables for
households that that farm and for households that garden. EPA tabulated IRs for animals and
animal products for households that farm and for households that raise animals. Thus, the first
type of household, F, represents farmers who may both grow crops and raise animals and who
are likely to consume more home grown/raised foods than the second type of household. The
second type of household, HG, represents the non-farming households that may consume lower
amounts of home-grown or raised foods (i.e., HG encompasses both households that garden
and households that raise animals).
The food-specific ingestion rates are based on the amount of each food type that households
that farm (F) or households that garden and raise animals (HG) produced and brought into their
homes for consumption and the number of persons consuming the food. EPA averaged the
actual consumption rate for home-grown foods over the 1-week survey period.
The default food-specific ingestion rates in MIRC for adults are those for farming households (F)
in Exhibit 6-9. The user can specify use of the generally less conservative, non-farming
household (HG) ingestion rates if they are more appropriate for the user's exposure scenario
(second column of IR values under Adults in Exhibit 6-9).
Exhibit 6-9. Summary of Age-Group-Specific Food Ingestion Rates for Farm Food
Items
Product
Ingestion Rate by Age Group (g/kg-day)
Children (Farm and Homegrown)e
Adults
Child
<1
Child
1-2
Child
3-5
Child
6-11
Child
12-19
Farm f
Homegrown 9
Mean
Beef a
NA
1.49
2.21
3.77
1.72
2.63
2.66
Dairyd
NA
67
37
24.79
10.90
17.1
15.9
Eggsd
NA
2.5
1.4
0.86
0.61
0.90
0.75
Exposed Fruitb
NA
1.8
2.6
2.52
1.33
2.32
1.55
Exposed Vegetable
NA
3.5
1.7
1.39
1.07
2.17
1.57
Porkd
NA
2.2
2.1
1.49
1.17
1.30
1.34
Poultryd
NA
3.6
3.4
2.13
1.59
1.54
1.58
Protected Fruitd
NA
19
13
8.13
5.44
5.19 c
5.9
Protected Vegetable
NA
2.5
1.3
1.1
0.776
1.3
1.01
Root Vegetable
NA
2.5
1.3
1.32
0.937
1.39
1.15
C-2-64
-------
Exhibit 6-9, continued. Summary of Age-Group-Specific Food Ingestion Rates for Farm
Food Items
Product
Ingestion Rate by Age Group (g/kg-day)
Children (Farm and Homegrown)e
Adults
Child
<1
Child
1-2
Child
3-5
Child
6-11
Child
12-19
Farm f
Homegrown 9
50w Percentile
Beef a
NA
0.84
1.23
2.11
1.51
1.64
1.83
Dairyd
NA
102
60
39
14
12.1
10.8
Eggsd
NA
1.5
0.79
0.56
0.49
0.67
0.48
Exposed Fruitb
NA
1.2
1.82
1.11
0.609
1.3
0.88
Exposed Vegetable
NA
1.9
1.2
0.643
0.656
1.38
0.89
Porkd
NA
1.8
1.4
1.02
1.02
0.924
0.97
Poultryd
NA
2.9
2.7
1.4
1.4
1.06
1.37
Protected Fruitd
NA
10.2
7.6
4.2
2.3
2.08 c
2.42
Protected Vegetable
NA
1.94
1.04
0.791
0.583
0.599
0.64
Root Vegetable
NA
0.92
0.46
0.523
0.565
0.88
0.67
90tn Percentile
Beef a
NA
4.5
6.7
11.4
3.53
5.39
5.39
Dairyd
NA
148
82
54.67
26.98
34.9
34.9
Eggsd
NA
5.1
2.8
1.8
1.34
1.65
1.36
Exposed Fruitb
NA
3.7
5.4
6.98
3.41
5
3.41
Exposed Vegetable
NA
10.7
3.47
3.22
2.35
6.01
3.63
Porkd
NA
4.5
4.4
3.04
2.65
3.08
2.9
Poultryd
NA
7.4
6.8
4.58
3.28
3.47
2.93
Protected Fruitd
NA
53
36
24.14
16.19
15.14 c
16
Protected Vegetable
NA
3.9
2.5
2.14
1.85
3.55
2.32
Root Vegetable
NA
7.3
4.3
3.83
2.26
3.11
2.81
95w Percentile
Beef a
NA
5.0
7.3
12.5
3.57
7.51
7.51
Dairyd
NA
139
75
52
27
44
44
Eggsd
NA
5.5
3.7
2.4
1.5
1.85
1.85
Exposed Fruitb
NA
4.1
6.1
12
4.8
6.12
5.0
Exposed Vegetable
NA
11.9
6.29
5.5
3.8
6.83
5.45
Porkd
NA
6.2
6.0
4.7
3.3
3.7
3.4
Poultryd
NA
8.2
7.2
5.3
3.7
4.8
3.3
Protected Fruitd
NA
59
42
28
20
19.16 c
19.1
Protected Vegetable
NA
9.4
5.1
3.12
2.2
5.4
3.05
Root Vegetable
NA
10.4
4.73
5.6
3.3
4.6
3.64
C-2-65
-------
Exhibit 6-9, continued. Summary of Age-Group-Specific Food Ingestion Rates for Farm
Food Items
Product
Ingestion Rate by Age Group (g/kg-day)
Children (Farm and Homegrown)e
Adults
Child
<1
Child
1-2
Child
3-5
Child
6-11
Child
12-19
Farm f
Homegrown 9
99m Percentile
Beef a
NA
5.3
7.8
13.3
4.3
11
12.5
Dairyd
NA
113
56
37
24
80
80
Eggsd
NA
16
12
8.6
5.0
6.6
6.6
Exposed Fruitb
NA
22
32.5
16
5.9
16
12.9
Exposed Vegetable
NA
12
7.4
13
5.7
10
10
Porkd
NA
9.1
9.9
6.3
4.2
4.9
4.3
Poultryd
NA
10
10
6.4
4.8
6.2
5.3
Protected Fruitd
NA
113
81
57
45
34.42 c
47.3
Protected Vegetable
NA
9.4
5.3
5.4
2.69
9.2
6.49
Root Vegetable
NA
10.4
4.7
7.5
5.1
7.5
7.5
Source: EPA 1997a (Chapter 13), unless otherwise noted.
NA = not applicable
a No data are available for Child 1 -2 or Child 3-5. The value for Child 6-11 was used, scaled down by the ratio of
the mean body weight for Child 1 -2 or Child 3-5, as appropriate, to the mean body weight of Child 6-11.
b No data are available for Child 1 -2. The value for this age group is the IR for Child 3-5, scaled down by the ratio
of the mean body weight for Child 1 -2 to the mean body weight for Child 3-5.
c These values represent a time-weighted average IR for two age groups, using exposure duration (ED) for the 20-
39 (ED=20 years) and 40-69 year age groups (ED=30 years).
d In many cases, intake rates for children were not available in EPA's 1997 EFH. Intakes for these receptor
groups were calculated using the methodology recommended in HHRAP (EPA 2005a), Section 6.2.2.2. Sources
to develop these values included EPA 1997a and EPA 2003b.
e In Chapter 13 of the 1997 EPA EFH, age group-specific IRs are provided for home produced items as a whole;
separate IRs are not presented for children from households that raise animals and households that farm.
f These values represent the IRs for "households who farm."
9 These values represent the IRs for "households who raise animals."
For children, EPA estimated food-specific IRs for four age categories (EPA 1997a): 1 to 2
years, 3 to 5 years, 6 to 11 years, and 12 to 19 years. Sample sizes were insufficient to
distinguish IRs for children in different types of households; hence, for children, a single IR
value represents both F and HG households for a given food type and age category (Exhibit
6-9). For some food types and age categories, there were insufficient data for EPA to provide
consumer-only intake rates (i.e., data set for the subpopulation consisted of fewer than 20
observations). The HHRAP methodology, Section 6.2.2.2, recommends a method by which to
calculate the "missing" age-specific consumer-only ingestion rates, as explained below. Food-
specific intake rates (IRs) for those child age groups and food items not included in Chapter 13
of the 1997 EFH, that is IRage_group_x, were derived using the following information:
• Mean or percentile-specific consumer-only intake of the farm food item, as brought into
the home, for the total NFCS survey population (from EFH Chapter 13) - IRcojotab
• Mean or percentile-specific per capita intake of the food type from all sources, as
consumed, for the specific child age group, from Chapter 3 of the CSFII Analysis of Food
Intake Distributions (EPA 2003c) - IRPC, age_grouP_x, and
• Mean or percentile-specific per capita intake of the farm food item for the total CSFII
survey population (from Chapter 3 of EPA 2003c) - IRPc_totai-
C-2-66
-------
The ratio of IRPC, age_grouP_x to IRpcjotai from the CSFII data shows the consumption rate of a
particular food type by a specific age group relative to the consumption rate for that food type for
the population as a whole. The ratio of IRCo, age_grouP_x to IRcojotai, that is the consumption rate
of a particular food type by a specific age group (consumers only) relative to the consumption
rate for that food type for the NFCS survey population as a whole (consumers only), should be
approximately the same. Given the assumption that the two ratios are equal, Equation 6-2 was
used to calculate the "missing" age-specific consumer-only IRs:
Equation 6-1. Calculation of Age-Group-Specific and Food-Specific Ingestion Rates
lD _ COJtotaI x I^PC, age_group_x
'~CO, age_group_x ~ 7^
'Kpcjotal
Mean or percentile-specific consumer-only intake of the food type from all
sources, as consumed, for the specific child age group X
Mean or percentile-specific consumer-only intake of the farm food item, as
brought into the home, for the total NFCS survey population
Mean or percentile-specific per capita intake of the food type from all sources,
as consumed, for the specific child age group X from the CSFII
Mean or percentile-specific per capita intake of the farm food item for the total
CSFII survey population
In this discussion, per capita (as opposed to consumer-only) indicates the intake rates are
based on the entire population rather than the subset of the population that ingests the
particular food category (i.e., consumers). Flere, the use of per capita ingestion rates are
recommended by the FIFIRAP methodology because no consumer-only percentile-specific
intakes are provided for the different age groups.
The above calculation implicitly assumes that the distribution of the consumption rate for a food
type for a specific age group (consumers only) has the same shape as the distribution of the
consumption rate for a food type for a specific age group in the general population (per capita).
Otherwise, the separate calculation of each percentile might yield intake estimates that
decrease as the percentile increases. This calculation artifact could occur if the shapes of the
two distributions differ in the upper percentiles (or "tails") of the distributions.
In the instances where the above calculations were used to fill data gaps in the above exhibit,
only the dairy child-specific age group intake estimates are not strictly increasing with increasing
percentile. The distributions likely track better (and thus the above assumption of equal ratios is
more reasonable) for lower percentiles, with deviations occurring due to outlier ingestion rates
based on only a few respondents in the tails of the distributions. The MIRC defaults use the 90th
percentile ingestion estimates, which are likely more reliable than the 95th or 99th percentile
estimates in this particular calculation.
6.3.4 Local Fish Ingestion Rates
The USDA's 1987 to 1988 Nationwide Food Consumption Survey (NFCS) (USDA 1993, 1994a),
as presented in Chapter 13 of the Agency's Exposure Factors Flandbook (i.e., Intake Rates for
Various Flome Produced Food Items) (EPA 1997a), includes family-caught fish ingestion rates
by age category. There are several disadvantages, however, to using that data source to
where:
IRCO, age_group_x =
IRcojotai =
IRpc, age_group_x
IR " =
C-2-67
-------
estimate fish ingestion rates. First, due to inadequate sample sizes, EPA did not report fish IRs
for children less than 6 years of age. Second, the NFCS data were collected approximately two
decades ago. Third, the reported fish IRs are for ages 6 to 11 and 12 to 19 and are based on
29 and 21 individuals in each age category, respectively (EPA 1997a, Table 13-23). Finally, the
IRs from NFCS data are based on total weight offish as brought into the home, and do not
include losses from preparation of the fish (i.e., removal of inedible parts and, possibly, the
skin). Estimates of preparation losses for fish intended to apply to the NFCS fish IR data are
very uncertain and are based on a wide variety of freshwater, estuarine, and marine fish, and
squid (EPA 1997a, Table 13-5). Therefore, a more recent survey was sought that included
larger sample sizes, data for children younger than six years, and IRs for the parts offish
actually consumed.
EPA's (2002) analysis freshwater and estuarine fish consumption from the USDA's Continuing
Survey of Food Intake by Individuals (CSFII) for 1994-96 and 1998 was chosen to provide per
capita fish IR options by age category in MIRC. Although the fish consumption rates reported in
the CSFII include all sources, commercial and self-caught, for purposes of screening level risk
assessments, it was assumed that all freshwater and estuarine fish consumed are self-caught.
The inclusion of commercially obtained and estuarine fish will overestimate locally caught
freshwater fish IRs for many rural populations in the United States; however, it also may
underestimate locally caught fish IRs for some populations (e.g., Native Americans, Asian and
Pacific Island communities, rural African American communities). Because consumption of
locally caught fish varies substantially from region to region in the United States and from one
population or ethnic group to the next, users of MIRC are encouraged to use more locally
relevant data whenever available.
For children ages 3 to 17 years and for adults, MIRC includes values for the mean and the 90th,
95th, and 99th percentile fish ingestion rates (freshwater and estuarine fish only) based on
EPA's analysis of 1994-96 and 1998 CSFII data (EPA 2002, 2008a). As shown in EPA's
2008(a) CSEFH, Table 10-7, the 90th percentile per capita ingestion rates estimated from the
two-day CSFII recall period are zero for some child age groups. Although not presented in
CSEFH Table 10-7, median ingestion rates for all child age groups would be zero (considering
the "consumer only" sample sizes [CSEFH Table 10-9] relative to the "per capita" sample sizes
in Table 10-7).
The high percentile fish IRs that are zero result from the short duration of the CSFII survey (two
days) compared with the averaging time of interest (a year) and the relatively infrequent
consumption of fish (e.g., on the order of once a week to once a month or less) compared with
the near daily ingestion of other types of food products (e.g., dairy, produce, meat). Use of zero
for fish IRs, however, is not useful in MIRC. As a result, an alternative method was used to
estimate fish ingestion rates for children and adults that could provide reasonable, non-zero
values for all age groups and percentiles.
The alternative, age-group-specific fish ingestion rates were derived using values for each age
group, y:
• Mean or other appropriate percentile consumer-only fish ingestion rates for age group y,
IRco.y, from EPA's Estimated Per Capita Fish Consumption in the United States (EPA
2002), Section 5.2.1.1, Table 5, for freshwater/estuarine habitat.3
3 Most of these data also are provided in Table 10-9 of the CSEFH; the median values, however, are not presented in
the CSEFH, and values for the mean and all other percentiles are slightly different due to rounding.
C-2-68
-------
• Fraction of the population consuming freshwater/estuarine fish, FPCy, calculated as
consumer-only sample size / U.S. population sample for age group y. The data to
calculate these fractions are available in the 2008 CSEFH and EPA 2002.
Equation 6-2 was used to calculate the alternative, per capita fish ingestion rates by age group
(IRpc.yY
Equation 6-2. Calculation of Alternative Age-Group-Specific Fish Ingestion Rates
where:
'RpC.y ~ '^CO.y x Fpc.y
IRpc.y = Per capita fish ingestion rate for age group y (g/day)
_ Consumer-only fish ingestion rates for age group y (g/day) (EPA 2002, Section
COy ~ 5.2.1.1, Table 5, for freshwater/estuarine habitat)
Fraction of the population consuming freshwater/estuarine fish, calculated as
Fpcy = consumer-only sample size I total U.S. population sample size for age group y
(unitless) (2008 CSEFH, EPA 2002)
In the above, per capita (as opposed to consumer-only) indicates the intake rates are based on
the entire population rather than the subset of the population that ingests the particular food
category. Here, per capita ingestions are recommended by the HHRAP methodology because
no consumer-only percentile-specific intakes are provided for the different age groups.
However, over 90% of the respondents consumed milk products.
The mean and percentile consumer-only fish ingestion rates for children and adults and the
fraction of the population consuming freshwater/estuarine fish used in calculating long-term per
capita fish ingestion rates by age group are presented in Error! Reference source not found,
and Exhibit 6-11. The mean and percentile per capita fish ingestion rates estimated using this
methodology are summarized in Exhibit 6-12 and are available in MIRC.
The fish ingestion rates provided in Exhibit 6-12 and included in MIRC are intended to represent
the harvest and consumption of fish in surface waters in a hypothetical depositional area. For
site-specific application of this tool, users should consider using more localized survey data to
estimate more appropriate fish ingestion rates. The fishing season varies substantially across
the United States by latitude, and fish consumption patterns also vary by type of water body
(e.g., ponds, lakes, rivers, streams, estuaries, coastal marine), cultural heritage, and general
geographic area. Therefore, use of more localized information is encouraged.
As noted in Section 6.4.3, if the user overwrites the fish IRs shown in Exhibit 6-12 with fresh-
weight as caught values (e.g., values obtained from a local creel survey), the user is advised to
set non-zero values for the preparation and cooking loss factors L1 and L2 in Equation 3-15.
Suggested values are presented in Section 6.4.3.
C-2-69
-------
Exhibit 6-10. Daily Mean and Percentile Consumer-Only Fish Ingestion Rates
for Children and Adults (IRCo.v)a
Lifestage (years)
Ingestion Rates, All Fish (g/day)
Mean
50th
90th
95th
99th
Child <1
NA
NA
NA
NA
NA
Child 1-2 b
27.31
15.61
64.46
87.60
138.76 *
Child 3-5 c
40.31
23.04
95.16
129.31
204.84 *
Child 6-11 d
61.49
28.46
156.86 *
247.69 *
385.64 *
Child 12-19 e
79.07
43.18
181.40 *
211.15 *
423.38 *
Adultf
81.08
47.39
199.62 *
278.91
505.65 *
Sources: EPA 2002, 2008a
NA = not applicable; it is assumed that children < 1 year of age do not consume fish.
* Indicates that the sample size does not meet minimum reporting requirements as described in EPA 2002. Owing
to the small sample sizes, these upper percentiles values are highly uncertain.
a Per capita fish ingestion (Fl) rates for children by age group are available from Chapter 10 of the CSEFH (EPA
2008a); however, all 50th and some 90th percentile ingestion rates are zero. Per capita Fl rates were therefore
estimated as described in Equation 6-2 to provide reasonable, non-zero values for all age groups and percentiles.
b A fish IR for ages 1 -2 years was not available. The value represents the consumer-only fish ingestion rate for
ages 3 to 5 from EPA (2002) (Section 5.2.1.1 Table 5 [freshwater/estuarine habitat]), scaled down by the ratio of
the mean Child 1-2 body weight to the mean Child 3-5 body weight.
c These values represent the consumer-only fish ingestion rate for ages 3 to 5 from EPA (2002), Section 5.2.1.1
Table 5 (freshwater/estuarine habitat). Sample size = 442.
d These values represent the consumer-only fish ingestion rate for ages 6 to 10 from EPA (2002), Section 5.2.1.1
Table 5 (freshwater/estuarine habitat). Sample size = 147.
e These values represent the time-weighted average per capita fish ingestion rate for ages 11 to 15 and 16 to 17
years from EPA (2002), Section 5.1.1.1 Table 5 (freshwater/estuarine habitat); the value may underestimate
ingestion rate for ages 12 to 19 years. Sample size = 135.
f These values represent the consumer-only fish ingestion rate for individuals 18 years and older from EPA (2002),
Section 5.2.1.1 Table 4 (freshwater/estuarine habitat). Sample size = 1,633.
Exhibit 6-11. Fraction of Population Consuming Freshwater/Estuarine Fish on a
Single Day (FPC y)
Lifestage (years)
Fraction Consuming Fish
Child 3-5
0.0503a
Child 6-11
0.0440b
Child 12-19
0.0493c
Adult
0.08509d
Sources: EPA 2002, 2008a
a This value was calculated using the ages 3 to 5 sample size for consumers only divided by the sample
size for the U.S. population divided by 2 to represent the proportion consuming fish on a single day (the
consumers-only group includes individuals who consumed fish on at least one of two survey days) to
match the one-day ingestion rate.
b As in footnote a, the value was calculated using the ages 6 to 10 sample size for consumers only divided
by the sample size for U.S. population divided by 2.
c The value was calculated by summing the ages 11 to 15 and 16 to 17 sample sizes for consumers only
and dividing by both by the sum of the sample sizes for U.S. population and by a factor of 2.
d The value was calculated using the ages 18 and older sample size for consumers only divided by the
sample size for U.S. population from Section 5.1.1.1 Table 4. The result was divided by 2 to represent a
one-day sampling period in order to match the one-day ingestion rate.
C-2-70
-------
Exhibit 6-12. Calculated Long-term Mean and Percentile Per capita Fish Ingestion
Rates for Children and Adults (IRPC.V)
Lifestage (years)
Ingestion Rates, All Fish (g/day)
Mean
50th
90th
95th
99th
Child <1
NA
NA
NA
NA
NA
Child 1-2 a
1.37
0.79
3.24
4.41
6.98
Child 3-5 b
2.03
1.16
4.79
6.51
10.3
Child 6-11c
2.71
1.25
6.90
10.9
17.0
Child 12-19 d
3.90
2.13
8.95
10.4
20.9
Adulte
6.90
4.03
16.99
23.73
43.02
Sources: EPA 2002, 2008a
NA = not applicable; it is assumed that children < 1 year of age do not consume fish.
a Values were calculated as (consumer-only IR for Child 1 -2) x (fraction of population consuming fish for Child 3-5).
b Values were calculated as (consumer-only IR for Child 3-5) x (fraction of population consuming fish for Child 3-5).
c Values were calculated as (consumer-only IR for Child 6-11) x (fraction of population consuming fish for Child 6-
11).
d Values were calculated as (consumer-only IR estimated for Child 12-19) x (fraction of population estimated to
consume fish for Child 12-19).
e Values were calculated as (consumer-only IR for Adults) x (fraction of population consuming fish for Adults).
Applications to date of MIRC have used whole fish concentrations estimated by TRIM.FaTE.
The proportion lipid in TL3 and TL4 fish in TRIM.FaTE is assumed to be 5.7 percent (by weight)
for the whole fish, based on information provided by Thomann (1989). The lipid content of the
part(s) of the fish normally consumed is likely to be less than 5.7 percent. For example, EPA
estimated a consumption-weighted mean lipid value for fillets offish from TL3 to be 2.6 percent
and from TL4 to be 3.0 percent (Table 6-9 in EPA 2003b). If a user of MIRC wishes to account
for reduced chemical concentration in fillet compared with whole fish for lipophilic chemicals, the
user can specify a "preparation" loss of chemical (see Section 6.4).
For lipophilic chemicals (e.g., log Kow greater than 4), which partition primarily into the fatty
tissues offish, much of the higher concentration tissues might be stripped from the fish during
preparation (e.g., belly fat, viscera which includes fat in liver, etc, fat under skin). The degree to
which the concentration of chemical in a fillet is less than the average total concentration in the
whole fish is chemical specific. Assuming that the chemical concentration in the fillet is the
same as in the whole fish may result in a conservative bias for highly lipophilic chemicals. For
persons who prefer to consume fillets with the skin on and do not discard belly fat, assuming the
same concentration of chemical in the fish consumed as in the whole fish is protective.
6.3.5 Soil Ingestion Rates
Adult gardeners may incidentally ingest soils from gardening activities, and gardening and
farming families might ingest soil particles that adhere to exposed fruits and exposed and
belowground vegetables. Soils that are re-suspended in the air by wind can resettle on
exposed fruits and vegetables. Children may incidentally ingest soils in those ways, but in
addition, children playing outdoors may ingest soils directly or by hand-to-mouth activities during
play. MIRC includes soil ingestion rate options by age group for these types of exposures.
MIRC does not include options for children who may exhibit pica, or the recurrent ingestion of
unusually high amounts of soil (i.e., on the order of 1,000 - 5,000 mg/day or more) (EPA 2008a).
Data on soil ingestion rates are sparse; the soil ingestion rates listed in Exhibit 6-13 and
included in MIRC are based on very limited data, as is evident from the values listed. The
studies evaluated by EPA for children generally focused on children between the ages of 1 and
C-2-71
-------
3 to 6 years and were not specific to families that garden or farm. The default ingestion rates in
MIRC are the 90th percentile values, as for other ingestion rate parameters.
Exhibit 6-13. Daily Mean and Percentile Soil Ingestion Rates for Children and Adults
Age Group
(years)
Soil Ingestion Rate (mg/day)
Mean a
50tn a
90tn
95tn
99tn
Child < 1
NA
Child 1-2
50
50
400 b
400 b
400 b
Child 3-5
50
50
400 b
400 b
400 b
Child 6-11
50
50
201 c
331 d
331 d
Child 12-19
50
50
201 c
331 d
331 d
Adult 20-70
50
50
201 c
331 d
331 d
Sources: EPA 1997a, 2008a
a For the mean and 50th percentile soil ingestion rates for children, value represents a "central tendency" estimate
from EPA's 2008 CSEFH, Table 5-1. For adults, value is the recommended mean value for adults from EPA's
1997 EFH, Chapter 4, Table 4-23.
b These values are the recommended "upper percentile" value for children from EPA's 1997 EFH, Chapter 4,
Table 4-23. The 2008 CSEFFI included a high-end value associated with pica only.
c These values are 90th percentile adult ingestion rates calculated in Stanek et al. 1997, and they are used to
represent older children and adults.
d These values are 95th percentile adult ingestion rates calculated in Stanek et al. 1997, and they are used to
represent older children and adults.
6.3.6 Total Food Ingestion Rates
Although not included in MIRC for deterministic screening-level exposure and risk assessments,
total food ingestion rates would be included in any probabilistic module developed for MIRC.
The total food ingestion rates presented in Exhibit 6-14 will be used to normalize or to truncate
the sum of food-specific ingestion rates to reasonable values. This procedure is particularly
important when chemical intake from multiple upper-percentile food ingestion rates for different
types of food are added together. Individuals representing the upper percentile ingestion rate
for one food category might not be the same individuals who reported high percentile ingestion
rates for one or any of the other food categories.
C-2-72
-------
Exhibit 6-14. Daily Mean and Percentile Per Capita Total Food Intake for Children and
Adults
Lifestage (years)
Percent of Group
Consuming Food
Mean
50th
90th
95th
99th
Total Food Intake (g/day, as consumed)
Child < 1 a
67.0% - 99.7% h
322
270
599
779
1152
Child 1-2 b
100%
1,032
996
1537
1703
2143
Child 3-5 c
100%
1,066
1,020
1,548
1,746
2,168
Child 6-11 d
100%
1,118
1,052
1,642
1,825
2,218
Child 12-19 e
100%
1,197
1,093
1,872
2,231
2,975
Adultf
100%
1,100
1,034
1,738
2,002
2,736
Total Food Intake (g/kg-day, as consumed)
Child < 1 a
67.0% - 99.7% h
39
34
72
95
147
Child 1-2 b
100%
82
79
125
144
177
Child 3-5 c
100%
61
57
91
102
132
Child 6-11 d
100%
40
38
61
70
88
Child 12-19 e
100%
21
19
34
40
51
Adult9
100%
14.8
13.9
23.7
27.6
35.5
Sources: EPA 2005e, 2008a
a These values represent a time-weighted average for age groups birth to <1 month (N=88), 1 to <3 months
(N=245), 3 to <6 months (N=411), and 6 to <12 months (N=678) from Table 14-3 of the 2008 CSEFH.
b These values represent a time-weighted average for age groups 1 to <2 years (N=1,002) and 2 to <3 years
(N=994) from Table 14-3 of the 2008 CSEFH.
c These values were obtained from Table 14-3 of the 2008 CSEFH (age group 3 to <6 years, N=4,112).
d These values were obtained from Table 14-3 of the 2008 CSEFH (age group 6 to <11 years, N=1,553). These
values represents a conservative (i.e., slightly low) estimate for ages 6 through 11 years since 11-year olds are not
included in this CSEFH age group.
e These values represent a time-weighted average for age groups 11 to <16 years (N=975) and 16 to <21 (N=743)
years from Table 14-3 of the 2008 CSEFH. Note that estimated values include 11-year-olds and individuals
through age 20, which contributes to uncertainty in the estimates.
f These values represent a time-weighted average for age groups 20 to 39 years (N=2,950) and 40 to 69 years
(N=4,818) from Table 5B of the 2005 EPA analysis of CSFII.
9 These values represent a time-weighted average for age groups 20 to 39 years (N=2,950) and 40 to 69 years
(N=4,818) from Table 5A of the 2005 EPA analysis of CSFII.
h Percents consuming foods from Table 14-3 of the 2008 CSEFH include: 67.0% (birth to <1 month); 74.7% (1 to
<3 months); 93.7% (3 to <6 months); and 99.7% (6 to <12 months). Infants under the age of 1 that consume
breast milk are classified as "non-consumers" of food.
6.4 Other Exposure Factor Values
The other exposure parameters included in the MIRC algorithms are exposure frequency
(Section 6.4.1), fraction of the food type obtained from the contaminated area (Section 6.4.2),
and reduction in the weight of the food types during preparation and cooking (Section 6.4.3).
For the breast milk ingestion pathway, additional exposure parameters are included in the FFC
algorithms (Section 6.5).
6.4.1 Exposure Frequency
The exposure frequency (EF) represents the number of days per year that an individual
consumes home-produced food items that are contaminated with the chemical being evaluated.
In MIRC, the default value for EF is 365 days/year for all exposure sources and all potential
C-2-73
-------
receptors. This assumption is consistent with the food ingestion rates used in MIRC (i.e., daily
intake rates equivalent to annual totals divided by 365 days) and does not imply that residents
necessarily consume home-produced food products every day of the year.
If the user wishes to evaluate daily intake rates based on shorter averaging times, the user can
overwrite both the food-specific ingestion rates and the EF for each home-grown food product.
Users of MIRC might want to specify a lower EF values for various food types where residents
obtain some of their diet from commercial sources and where consumption of home grown
produce is seasonal.
6.4.2 Fraction Contaminated
The fraction contaminated (FC) represents the fraction of each food product consumed that is
contaminated by the chemical at a level consistent with environmental concentrations in the
area of concern (e.g., area with maximum deposition rates). Obviously, the most conservative
assumption is that all food products consumed (i.e., 100 percent) are from the location
represented by the chemical concentrations input into MIRC.
For non-infant children and the adult age cohorts, MIRC includes the default FC of 1, assuming
that 100 percent of the food product consumed is produced by households that farm, garden, or
raise animals. The user can vary this default FC value for individual food products to tailor the
assessment to a particular exposure scenario.
6.4.3 Preparation and Cooking Losses
Food preparation and cooking losses are included in the FFC exposure calculations to account
for the amount of a food product as brought into the home that is not ingested due to loss during
preparation, cooking, or post-cooking. These losses need to be accounted for in the ADD
equations because the food ingestion rates calculated from the USDA 1987 to 1988 NFCS are
based on the weight of home grown produce and animal products brought from the field into the
house prior to any type of preparation. Not all of the produce or products were eventually
ingested. In general, some parts of the produce and products are discarded during preparation
while other parts might not be consumed even after cooking (e.g., bones). Thus, the actual food
ingested is generally less than the amount brought into the home.
Three distinct types of preparation and cooking losses are included in the ingestion exposure
algorithms in MIRC: (1) loss of parts of the food type from paring (i.e., removing the skin from
vegetables and fruits) or other types of preparation (e.g., removing pits, coring, deboning), (2)
additional loss of weight for the food type during cooking (e.g., evaporation of water), and (3)
post-cooking losses (e.g., non-consumption of bones, draining cooking liquid [e.g., spinach]).
MIRC includes mean values for these three types of preparation and cooking losses for
exposed fruit, protected fruit, exposed vegetables, protected vegetables, root vegetables, beef,
pork, poultry, and fish. Different types of losses apply to different types of foods. Therefore, the
losses can be represented by only two parameters, L1 and L2, the definitions of which vary
according to the food type as explained in the endnotes in Exhibit 6-15. All preparation and
cooking loss parameter values were estimated as specified in the Exhibit's endnotes from data
presented in Chapter 13 of the EPA's 1997 EFH.
C-2-74
-------
Exhibit 6-15. Fraction Weight Losses from Preparation of Various Foods
Product
Mean Cooking, Paring, or
Preparation Loss
(Cooking Loss Type 1 [L1])
(unitless)a
Mean Net Post Cooking
(Cooking Loss Type 2 [L2])
(unitless) b
Exposed Fruitc
0.244
0.305
Exposed Vegetable
0.162 d
NA
Protected Fruit
0.29 e
NA
Protected Vegetable
0.088 f
NA
Root Vegetable 9
0.075
0.22
Beef
0.27
0.24
Pork
0.28
0.36
Poultry
0.32
0.295 h
Fish'
0.0
0.0
Source: EPA 1997a (Chapter 13; Tables 13-5 [meats], 13-6 [fruits], and 13-7 [vegetables])
NA = Not Available
a For fruits, includes losses from draining cooked forms. For vegetables, includes losses due to paring, trimming,
flowering the stalk, thawing, draining, scraping, shelling, slicing, husking, chopping, and dicing and gains from the
addition of water, fat, or other ingredients. For meats, includes dripping and volatile losses during cooking.
b For fruits, includes losses from removal of skin or peel, core or pit, stems or caps, seeds and defects; may also
include losses from removal of drained liquids from canned or frozen forms. For vegetables, includes losses from
draining or removal of skin. For meats, includes losses from cutting, shrinkage, excess fat, bones, scraps, and
juices.
c These values represent averages of means for all fruits with available data (except oranges) (Table 13-6).
d This value represents an average of means for all exposed vegetables with available data (Table 13-7). Exposed
vegetables include asparagus, broccoli, cabbage, cucumber, lettuce, okra, peppers, snap beans, and tomatoes.
e This value was set equal to the value for oranges (Table 13-6).
f This value represents an average of means for all protected vegetables with available data (Table 13-7).
Protected vegetables include pumpkin, corn, peas, and lima beans.
9 These values represent averages of means for all root vegetables with available data (Table 13-7). Root
vegetables include beets, carrots, onions, and potatoes.
h This value represents an average of means for chicken and turkey (Table 13-5).
' If the user changes fish ingestion rates to match a survey of the whole weight of fish brought into the home from
the field (divided by the consumers of the fish), an appropriate value for L1 would be 0.3 (EPA 1997a, Table 13-5).
For volatile or water soluble chemicals, a non-zero value for L2 also may be appropriate. Although EPA (1997a)
recommended 0.11 for L2, it varies substantially by chemical.
There are substantial uncertainties associated with the L1 and L2 parameters, including the
wide variation in values across produce types that were averaged together to recommend a
central tendency value for each. For example, the L2 factor does not distinguish between
weight loss during cooking by water evaporation, which might leave the chemical in the fruit,
and pouring the cooking liquid down the drain (chemical lost) or using the liquid to create a
sauce (chemical not lost). In addition, the concentration of chemical might be highest in the
skin, which often is discarded, and lower in the consumed portion of many bulky fruits and
vegetables. Finally, the data EPA used to evaluate L1 included negative losses (i.e., weight
gains) due to hydration of dried vegetables (e.g., peas and lima beans), which increases the
range of L1 values across different vegetables.
Note that the default L1 and L2 values for fish are set to zero. That is because the data source
for the fish ingestion rates is not the USDA's 1987 to 1988 NFCS (USDA 1993, 1994) as
reported in EPA's EFH, which reported food as brought into the home, as is the case for the
other food categories. Instead, the fish IR data included in MIRC are from a more recent and
larger survey, EPA's (2002) analysis of freshwater and estuarine fish consumption from the
USDA's 1994-96 and 1998 CSFII. That survey reports ingestion rates offish parts actually
C-2-75
-------
consumed, and so no loss processes for preparation are needed. The zero value for L2
assumes that no chemical is lost by volatilization and that pan juices are consumed. The user
may reset that value where chemical-specific data are available.
If the user manually changes fish ingestion rates to match a local survey of the whole weight of
fish brought into the home from the field (divided by number of persons consuming the fish), the
user should also set the L1 and L2 parameter to non-zero values. An appropriate value for L1
would be 0.3 (EPA 1997a, Table 13-5). For volatile or water soluble chemicals, a non-zero
value for L2 also may be appropriate. Although EPA recommended 0.11 for L2 (EPA 1997a,
Table 13-5), it varies substantially by chemical.
6.5 Breast-Milk Infant Exposure Pathway Parameter Values
Values used for parameters in the breast-milk exposure pathway algorithms (Section 3.4) can
be scenario-specific, receptor-specific, and/or chemical-specific and might be empirically
derived or estimated by an appropriate model. For parameters that are scenario-specific or for
which empirical values are required, the default values provided in MIRC are listed. For
parameters for which MIRC calculates values, the appropriate equation is listed. Scenario- and
receptor-specific parameters are discussed in Section 6.5.1 and chemical-specific parameters
are discussed in Section 6.5.2.
6.5.1 Receptor-specific Parameters
Receptor-specific values are needed for parameters that describe the characteristics or
activities of the exposed individual. In this context, there are two relevant receptors: the mother
and the infant. Exhibit 6-16 lists the parameters and their default values. The text that follows
describes the input value or value options for each exposure parameter required by MIRC to
calculate the infant absorbed chemical intake rate, or DAIinf. For parameter values that can be
estimated when empirical values are not available, see the equation description in Section 3.4.
Exhibit 6-16. Scenario- and Receptor-Specific Input Parameter Values Used to
Estimate Infant Exposures via Breast Milk
Parameter
Description
Default Value
AT
Averaging time for infant's exposure via breast milk, i.e., duration of
nursing (days)
= ED
BWjnf
Body weight of infant (kg) averaged over duration of nursing exposure
7.8
BWmat
Body weight of mother (kg) averaged over duration of mother's
exposure
66
DAImat
Daily absorbed intake of chemical by mother (mg/kg-day)
Equation 3-36
ED
Exposure duration for infant, i.e., duration of breast feeding (days)
=AT
AT/ED
Averaging time divided by exposure duration
1.0
fbp
Fraction of mother's whole blood that is plasma (unitless)
0.65
ffm
Fraction of mother's body weight that is fat (unitless)
0.30
fmbm
Fraction of fat in mother's breast milk (unitless)
0.04
fpm
Fraction of mother's body weight that is plasma (unitless)
0.046
IRmilk
Mean infant milk ingestion rate over duration of nursing (kg/day)
0.709
tbf
Duration of breast feeding (days)
365
tpn
Duration of maternal chemical exposure prior to nursing (days)
3285
C-2-76
-------
Averaging time (AT) and exposure duration (ED). AT refers to the time over which the infant's
exposure to the chemical of concern is averaged. ED refers to the duration of the infant's
exposure. For the exposure scenario considered for this age group, both AT and ED equal the
duration of the nursing period, and they therefore cancel each other out in the infant average
daily dose equation.
Infant body weight (BWinf). The user selects a value for BWmf, the time-weighted average body
weight of the infant over the entire duration of breast feeding, based on the age at which the
infant stops breast feeding. For example, if the infant breast feeds for one year, the user should
select the body weight for an infant that is averaged from birth to the first birthday. Similarly, if
an infant breast feeds for 6 months, the user should select the body weight for an infant that is
averaged from birth to six months. Because the default breast feeding duration (tbf) is one year
(i.e., 365 days), the default infant body weight is 7.8 kg, which is the time-weighted average for
the mean infant body weight between birth and its first birthday from EPA's (2008) Child Specific
Exposure Factors Handbook (CSEFH; EPA 2008a). Exhibit 6-17 presents additional values for
the infant body weight parameter that the user can select instead of the MIRC default.
Exhibit 6-17. Average Body Weight for Infants
Statistic
0 to < 6 months
(kg)
0 to < 12 months
(kg)
0 to < 18 months
(kg)
0 to < 24 months
(kg)
Mean
6.5
7.8a
9.0
9.6
5th percentile
5.0
6.0
7.0
7.5
10th percentile
5.3
6.4
7.4
7.8
15th percentile
5.5
6.7
7.7
8.2
25th percentile
5.8
7.0
8.1
8.7
50th percentile
6.4
7.8
8.9
9.5
75th percentile
7.1
8.6
9.9
10.5
85th percentile
7.4
9.0
10.3
11.0
90th percentile
7.7
9.2
10.6
11.3
95th percentile
8.0
9.7
11.1
11.8
Source: EPA 2008a; each value is the time-weiqhted averaqe from the data summaries presented in the CSEFH,
Table 8-3.
a MIRC default
Maternal body weight (BWmat). This parameter represents the body weight of the mother
averaged over the entire duration of the mother's exposure to the chemical of concern. The
maternal body weight is needed to calculate the biological elimination constant for the lipophilic
chemical in lactating women (kfat_eiac)¦ MIRC assumes that the mother will be pregnant for 9
months (i.e., 0.75 year) and will be lactating for 1 year. The MIRC default maternal body weight
also assumes that the mother has been exposed for 10 years total. For 8.25 years, she is not
pregnant or lactating, for 0.75 year she is pregnant, and for 1 year she is lactating. The MIRC
default BWmat of 66 kg is based on CSFII data compiled by EPA for non-lactating and non-
pregnant women between the ages of 15 and 44 (i.e., women of child-bearing age), lactating
women, and pregnant women (EPA 2004b). Exhibit 6-18 presents additional values for the
maternal body weight parameter which the user may choose to use instead of the MIRC default.
The BWmat value is not the value that MIRC uses to estimate the mother's absorbed daily intake
(DA I mat) ¦ The daily ingestion rates for home-grown/raised food products in MIRC are for men
and women combined, with the rates normalized to body weight. The ingestion rates for soil,
C-2-77
-------
water, and fish included in MIRC are not normalized to body weight but are based on both men
and women. For those ingestion rates, MIRC uses an average body weight value for males and
females to estimate the average daily dose (intake) of the chemical in mg/kg-day. These values
are subject to the assumption that the body-weight normalized ingestion rates and resulting
ADD values are applicable to nursing mothers.
Exhibit 6-18. Time-weighted Average Body Weight for Mothers
Statistic
Weight (kg)
Mean
66.0a
5tn
47.1
10tn
50.2
25tn
54.3
50tn
62.0
75tn
72.0
90tn
85.7
95tn
97.0
Source: EPA 2004b
a MIRC default value
Exposure duration (ED). See discussion of AT and ED above.
Fraction of mother's whole blood that is plasma (fhp). Steinbeck (1954) reported that plasma
volume accounts for approximately 60 percent of the total blood volume in non-lactating human
females (EPA 1998). Harrison (1967) and Ueland (1976) reported plasma volumes between 63
to 70 percent in postpartum women (EPA 1998). The default value in MIRC of 65 percent (0.65)
is the value recommended by EPA in its Methodology for Assessing Health Risks Associated
with Multiple Pathways of Exposure to Combustor Emissions (MPE, EPA 1998).
Fraction of mother's body weight that is fat (frm). A limitation of using a steady-state, instead of
a dynamic partitioning, model for lactational transfer of chemicals is that several key parameters
change over the course of exposure. For example, Equation 3-38, used to estimate the
concentration of a lipophilic chemical in breast milk fat, assumes that the mother's body fat will
remain constant over the entire duration of breast feeding (tbf), which is unlikely to be true (EPA
2001a). Another limitation of the single analytic model is that chemical transfer rates from blood
to milk are unlikely to be the same as the rate of mobilization of the chemical from fat stores to
the blood (EPA 2001a). Studies cited in ATSDR's toxicological profile for chlorinated dibenzo-p-
dioxins show a correlation between percent body fat and the elimination rate of dioxins, with
longer half-lives for dioxins in individuals with a higher proportion of fat in their bodies (ATSDR
1998). In the context of a screening model, however, EPA recommends a default value for the
fraction of a mother's body comprised of fat of 0.3 based on data and discussions presented by
Smith (1987) and Sullivan et al. (1991) (EPA 1998). A fraction of 0.3 indicates that 30 percent
of the mother's body weight is fat, which is a conservative value (EPA 2001a). To establish a
conservative screening scenario, the MIRC default value for ffm is 0.30.
Fraction of fat in mother's breast milk (fmhm). The Cmmat model (Equation 3-38) assumes that a
constant fraction of breast milk is fat, even though there is evidence that indicates variation in
the fat content of breast milk throughout lactation (Sim and McNeil 1992). Different studies
suggest a fat content of breast milk in humans of between 1 and 5 percent (Jensen 1987,
Schecter et al. 1994, Hong et al. 1994, McLachlan 1993, Bates et al. 1994, NAS 1991, Butte et
al. 1984, Maxwell and Burmaster 1993, EPA 1997a, Smith 1987, Sullivan et al. 1991). The
C-2-78
-------
MIRC default value for fmbm of 0.04 (i.e., 4 percent) is the value EPA recommended for MPE
(EPA 1998).
Fraction of maternal weight that is plasma (fpm). Altmann and Dittmer (1964) estimated that
plasma volume for adult women ranged from 37 to 60 mL/kg of body weight and averaged
about 45 mL/kg. Ueland (1976) observed that the average plasma volume of women 6 weeks
postpartum was 45 mL/kg of body weight. Using a value of 1.026 for the specific gravity of
plasma from Conley (1974), EPA estimated a value of 0.046 for the fraction of maternal weight
that is plasma (EPA 1998). The MIRC default for fpm therefore is 0.046.
Infant breast milk ingestion rate (IRmiik). Milk ingestion rates vary with several factors, including
the age and size of the infant and use of other foods such as formula. Based on its review of a
several studies, EPA recommended time-weighted average and upper percentile milk ingestion
rates for infants that nurse for six and for twelve months (EPA 1997a, Chapter 14, Table 14-15).
To estimate an "average" value, EPA first estimated study-sample-size weighted average
values for 1, 3, 6, 9, and 12 months of age and then developed time-weighted average milk
ingestion rates from those (EPA 1997a). EPA estimated an upper percentile (upper bound)
value as the mean plus two standard deviations. MIRC converts the ingestion rates measured
volumetrically (mL/day) to mass-based estimates (kg/day) assuming the density of human milk
to be 1.03 g/mL (reported by NAS 1991 and recommended by EPA 1997a). The resulting
values are shown in the first two rows of Exhibit 6-19. The MIRC screening-level default of 980
mL/day is an upper-bound estimate based on a one-year nursing period.
Exhibit 6-19 also includes the recommended values for four non-overlapping age categories
from the CSEFH (EPA 2008, Table 15-1). The values demonstrate that although infants grow
substantially from birth to one year of age, the "upper bound" estimates of their milk ingestion
rates are very close to 1 liter per day at all stages of development in the first year.
Exhibit 6-19. Infant Breast Milk Intake Rates
Age Category
Average
(mL/d)
Average
(kg/d)
"Upper Bound"
(mL/d)
"Upper
Bound"
(kg/d)
Reference
1 to 6 months
742
0.764
1,033
1.064
EPA 1997a1
0 to < 1 2
months
688
0.709
980a
1.01a
EPA 1997a1
0 to < 1 month
510
0.525
950
0.979
EPA 2008tt
1 to < 3 months
690
0.711
980
1.01
EPA 20081
3 to < 6 months
770
0.793
1,000
1.03
EPA 20081
6 to < 1 2
months
620
0.639
1,000
1.03
EPA 20081
a MIRC default;f Based on review of multiple studies; n Based on a single study
Duration of breast feeding (thf). This parameter is equal to the infant's exposure duration (ED)
and the infant's averaging time (AT). In its MPE Methodology, EPA asserts a conservative
value for the duration of breast feeding of 1 year (i.e., 365 days) and a central tendency
estimate of 6 months (180 days) (EPA 1998). Reviewers of MPE noted that 365 days may be
overly conservative, given than only 20 percent of infants are breast fed for 6 months, at which
point alternative foods are introduced, at least in addition to breast milk (EPA 2001a).
Nonetheless, to establish a conservative screening scenario, the MIRC default for tbf is 365
days.
C-2-79
-------
Duration of the mother's exposure to the chemical of concern prior to nursing (tpn). The model
shown as Equation 3-38 includes this parameter to reduce the over-estimate of chemical
concentration in milk fat that occurs if the model is applied to a chemical with a long biological
half-life (e.g., many years). The factor is needed for applications of the model to scenarios with
a brief exposure duration (e.g., beginning a few months prior to the start of nursing) relative to
the chemical half life. As the duration of an exposure scenario increases to meet and exceed
the chemical half life, however, the overestimate that occurs without this parameter is reduced.
For example, assume a chemical biological half-life of 8 years and a nursing period of 1 year. If
exposure of the mother starts at the beginning of nursing, using Equation 3-38 without the tpn
term results in an over-estimate of the concentration of the chemical in breast milk by a factor of
28.1 compared with the prediction using Equation 3-38 with the tpn term (EPA 1998, Table 9-6).
However, at longer pre-natal exposures of the mother, the magnitude of the over-estimate is
reduced: for a 10-year exposure, the magnitude of the overestimate without the tpn term is 2.28,
and for a 30-year exposure, the overestimate is reduced to 1.39.
For purposes of the screening-level of assessment for dioxins, we assume an exposure duration
equal to the half-life of the chemical, or 10 years. Only 3285 days of that period are pre-natal
(i.e., 3650 minus 365 days, assuming 1 year lactation period). Although longer exposure
periods are possible for the screening scenario, there is sufficient uncertainty in the model to
merit accepting a conservative bias for this parameter value.
6.5.2 Chemical-Specific Parameter Values
The chemical-specific parameters in the breast-milk pathway in MIRC are listed in Exhibit 6-20.
Note that the parameters for which values are needed are different for the lipophilic chemicals
(i.e., dioxins), for which lactational transfer is assumed to occur via milk fat, and inorganic
chemicals, for which the transfer is assumed to occur via the aqueous phase of breast milk (i.e.,
mercury).
Absorption efficiency of the chemical by the oral route of exposure for the infant (AEinf). The
models included in MIRC assume that the AEinf from the lipid phase of breast milk is equal to the
AE^f from the aqueous phase of the milk. Reviewers of the model stated that this assumption
may not be valid and that ideally, the equation DAIinf would include variables for the AEmf from
the breast milk fat and the AEmf from the aqueous phase of breast milk (EPA 2001a). However,
since the MIRC assumption is that chemicals will partition to either the lipid or aqueous phase of
milk, it is not necessary at this time to have multiple AEinf values for a given chemical. If data on
the AE from the mother or an adult but not for the infant are available, data for the adult may be
used for AEinf. Reviewers also recommended that chemical-specific values come from studies
that account for absorption of the chemical from milk, because absorption from other matrices
(e.g., solid foods) may not be relevant (EPA 2001a). If chemical-specific data are not available
for adults or infants, a conservative default value for AEmf for a screening level assessment is
1.0, which assumes 100 percent absorption (EPA 1998).
The default value for AEinf in Ml RC for both MeHg and dioxin is 1.0. For ingested lipophilic
chemicals, it is reasonable to assume that absorption will be high (EPA 2004c). ATSDR (1998)
reported that dioxins are well absorbed by the oral route of exposure, with one human
experiment indicating more than 86 percent absorption. It is EPA policy to assume 100 percent
absorption for chemicals with reported AEs of 50 percent or higher (EPA 2004c). MeHg also is
well absorbed, with measured values as high as 95 percent, and so a value of 100 percent is
used in MIRC (EPA 2001b).
C-2-80
-------
Exhibit 6-20. Chemical-specific Input Parameter Values for Breast Milk Exposure
Pathway
Parameter and Description
2,3,7,8-
TCDD
MeHg
AEinf
Infant absorption efficiency of the chemical by the
oral route of exposure (i.e., fraction of ingested
chemical that is absorbed by the infant; unitless)
1.0 (default)
1.0 (default)
AEmat
Maternal absorption efficiency of the chemical by
the oral route of exposure (i.e., fraction of ingested
chemical that is absorbed by the mother; unitless)
1.0 (default)
1.0 (default)
ft.
Fraction of steady-state total body burden of
hydrophilic chemical in mother that is in the
mother's whole blood compartment (unitless)
NA
0.059 (Kershaw et
al. 1980)a
ff
Fraction of steady-state lipophilic chemical body
burden in mother that is stored in body fat
(unitless)
> 0.90 (ATSDR
1992)
NA
fp/
Fraction of steady-state total hydrophilic chemical
body burden in mother that is in the blood plasma
compartment (unitless)
NA
Not yet identified b
h
Biological half-life for chemical in non-lactating
women (days)
3650 (EPA
1994c)
50 (Sherlock et al.
1984)
kaq elac
Rate constant for total elimination of hydrophilic
chemicals by lactating women (per day)
NA
= kelim
kelim
Rate constant for elimination of chemical for non-
lactating women (per day; related to chemical half-
life)
1.9E-04b
1.4E-02 c
kfat_elac
Rate constant for total elimination of lipophilic
chemicals by lactating women (per day)
Est. using
Equation 3-41
NA
Partition coefficient for hydrophilic chemical
between maternal blood plasma and aqueous
phase of breast milk (g milk/g plasma; model
assumption)
NA
1.0 (model
assumption)
Pcrbc
Partition coefficient for hydrophilic or protein-
bound chemical between red blood cells (RBC)
and plasma in maternal blood (ml_ whole blood/mL
RBC)
NA
40 (Hollins et al.
1975)
NA = not applicable. ND = not yet determined from literature.
a This value is based on a single-dose study and may not be appropriate for a chronic exposure model.
b An empirical value for this variable is currently missing for application of model.
d This value was calculated from biological half-life (h) using Equation 3-40.
Note that AE values for some inorganic compounds are substantially less than 1.0. For
cadmium, for example, AEs for adults of 0.025 to 0.05 have been reported (EPA 2004c, Exhibit
B-4).
Absorption efficiency of the chemical by the oral route of exposure for the mother (AEmat). The
default value for both dioxins and MeHg is 1.0, as described in the previous paragraph.
Fraction of total maternal chemical body burden that is in the whole blood (fhi). The default
value for MeHg in MIRC, 0.059, is from Kershaw et al. (1980), which reported kinetics of blood
deposition and clearance of MeHg in humans. Individuals consumed one meal of fish that
contained between 18 and 22 pg Hg/kg body weight. The fraction of the dose deposited in the
blood volume after mercury was fully distributed in tissues was 5.9 percent or 0.059. This study
used a single-dose and thus may not be appropriate for a chronic exposure analysis.
C-2-81
-------
Fraction of total maternal chemical body burden that is in body fat (ff). Based on ATSDR's
Toxicological Profile for Selected PCBs (ATSDR 1992) and Sullivan et al. (1991), EPA
concluded that the "fraction of ingested contaminant stored in fat may be >90%" for lipophilic
chemicals such as PCBs and dioxins (EPA 1998). This statement was interpreted to mean that
90 percent of the maternal body burden of chemical at "steady state" is located in body fat for
dioxins at steady state.
Fraction of total maternal chemical body burden that is in blood plasma (fni). For hydrophilic
chemicals, this parameter represents the steady-state fraction of the total chemical in the body
that is circulating in the blood plasma. Values for fpt may be available for some chemicals in the
scientific literature. No value for this parameter for methyl mercury has been identified from the
literature at this time. A value can be calculated using Equation 3-43. However, this equation
requires a reliable value for fbi, and the value found for mercury may not be appropriate for a
chronic exposure analysis (see above).
Chemical half-life in non-lactatinq women (h). In general, highly lipophilic chemicals tend to
have relatively long biological half-lives. EPA estimates that the half-life for dioxins is between 7
and 10 years (EPA 1994a). ATSDR estimates that the half-life for 2,3,7,8-TCDD in particular
may be as long as 12 years (ATSDR 1998). To establish a conservative screening scenario,
the MIRC default half-life for dioxins is set to 10 years or 3650 days.
The half-life for methylmercury is on the order of weeks, not years. Greenwood et al. (1978)
measured blood clearance rates for MeHg in lactating Iraqi women exposed accidentally to
MeHg via bread prepared from wheat treated with a fungicide that contained MeHg. The data
indicated a mean half-life for MeHg of approximately 42 days. Sherlock et al. (1984) reported
an average measured half-life for MeHg of 50 days with a range of 42-70 days. The MIRC
default for MeHg is set to the longer average half life of 50 days.
Chemical elimination rate constant for lactating women - aqueous (kaq_Piar). The parameter
kaq_eiac is equal to keiim plus the loss rate for the chemical in the aqueous phase of breast-milk
during lactation. EPA has yet to propose a term for the additional elimination of a chemical in
the aqueous phase of milk from breast feeding. In the absence of empirical values, a
reasonable assumption for water soluble chemicals is that kaq_e,ac is equal to keiim as discussed
for Equation 3-43. The extent to which kenm is an underestimate of kaq_eiac for a given chemical
will determine the extent of conservative bias in /fac/_e/ac-
Chemical elimination rate constant for non-lactatinq women (kPlim). Although values for this
parameter often are reported directly in the literature, MIRC estimates kenm from chemical half-
life assuming first-order kinetics as shown in Equation 3-40. For example, for a biological half-
life of 3,650 days for dioxins, keiim is estimated to be 1.9E-04 per day. Assuming a biological
half-life of 50 days for MeHg, the value for keiim is estimated to be 0.014 per day.
Rate constant for total elimination of lipophilic chemicals by lactating women (kfa, ). Although
values for this parameter might be found in the scientific literature for some chemicals, in MIRC,
kfat_eiac for dioxins is calculated from Equation 3-41. When the parameters in that equation are
set to the default values in MIRC for dioxins, MIRC estimates a value of 0.0015 per day for
kfat_elac-
Partition coefficient for chemical between maternal blood plasma and aqueous phase of breast
milk (PChm). The aqueous model, presented in Equation 3-42, assumes that the concentrations
in the plasma and aqueous phase of breast milk are directly proportional (EPA 1998).
Therefore, the default value for this parameter for MeHg in MIRC is 1.0.
C-2-82
-------
Partition coefficient for chemical between red blood cells and plasma in maternal blood (PcPBr).
Chemical-specific values for this parameter should be located in the scientific literature. If
chemical-specific values are unavailable and it is assumed that there is equal distribution of the
chemical in the plasma and red blood cells, EPA suggests a default value of 1.0 (EPA 1998).
For MeHg, MIRC includes a value of 40 based on Hollins et al. (1975) study of cats exposed to
MeHg, which reported a ratio of radio-labeled mercury in red blood cells to plasma of 97.7 to 2.3
(i.e., ratio of 42.5).
C-2-83
-------
7 Summary of MIRC Default Exposure Parameter Settings
The default settings included in MIRC are intended to be characteristic of a conservative (but
plausible) exposure scenario that results in a negligible or extremely low chance of
underestimating risk to farming households in an area with chemical concentrations and air
deposition rates as specified by the user. These default parameter values were used to derive
the de minimis emission rates used for screening emissions of PB-HAPs from sources included
in RTR risk assessments. These values are the default, or initial setting, for parameter values in
MIRC as described in Section 6. This section summarizes the default parameter values used to
calculate screening thresholds.
This chapter is organized to present the chemical- and scenario-specific inputs to MIRC by data
type. The screening-level analysis uses 90th percentile ingestion rates, presented in Section
7.1, and population-specific characteristic assumptions, presented in Section 7.2, that are
generally conservative in nature. De minimis thresholds were derived for five RTR chemical
species: benzo(a)pyrene, cadmium, mercuric chloride, methyl mercury, and 2,3,7,8-TCDD;
Section 7.3 presents chemical-specific parameter inputs for these five chemicals. Finally,
Section 7.4 presents default parameter values for the nursing infant exposure scenario, which
applied only to dioxin and methyl mercury.
7.1 Default Ingestion Rates
The screening-level (or default) values for ingestion rates for soil, fish, breast milk, and for each
farm food type are equal to the 90th percentile of the distribution of national data for that
ingestion medium (Exhibit 7-1). The default settings also assume that all food types are
obtained from the area of chemical deposition specified by the user (i.e., fraction of food from
contaminated area = 1.0).
For estimates of de minimis emission rates for PB-HAPS, environmental concentrations and air
deposition rates were estimated using TRIM.FaTE for the area of maximal deposition in the
vicinity of a hypothetical facility, and thus represent risks estimated for a maximally exposed
individual/farm/family.
Exhibit 7-1 also includes a sum of the 90th percentile ingestion rates for homegrown food
categories and fish ingestion (preceding rows) to show the implied total food ingestion rate
associated with setting multiple food-type-specific ingestion rates at a 90th percentile. Because
the 90th percentiles for each farm food category are likely to reflect different individuals, it is
likely that addition of multiple 90th percentile intake values will exceed the total food ingestion
rates likely for the general population.
The final row in Exhibit 7-1 lists the likely magnitude of the overestimates by age category. The
preceding row includes the 90th percentile of the distribution of individual total food ingestion
rates from the USDA's 1994-96 and 1998 Continuing Survey of Food Intakes by Individuals
(CSFII) (USDA 2000) data sets, as analyzed by EPA (EPA 2005e). The total ingestion rate for
the farming households takes into account the cooking losses typical of each food category to
provide a better comparison with the 90th percentile individual total food ingestion rates (which
are based on consumption of prepared foods).
C-2-84
-------
Exhibit 7-1. Farm Food Category Ingestion Rates for Conservative Screening
Scenario for Farming Households
Product
90tn Percentile Consumer Ingestion Rate
Units a
Infants
< 1 yr
Child
1 -2 yrs
Child
3-5
yrs
Child
6-11
yrs
Child
12-19
yrs
Adult
Farm Food Item
Beef
NA
4.5
6.7
11.4
3.53
5.39
g/kg-day
Dairyc
NA
148
82
54.7
27.0
34.9
g/kg-day
Eggsc
NA
5.1
2.8
1.80
1.34
1.65
g/kg-day
Exposed Fruit
NA
3.7
5.4
6.98
3.41
5
g/kg-day
Exposed Vegetable
NA
10.7
3.5
3.22
2.35
6.01
q/kq-day
O
CL
NA
4.5
4.4
3.04
2.65
3.08
q/kq-day
Poultryc
NA
7.4
6.8
4.58
3.28
3.47
q/kq-day
Protected Fruitc
NA
53
36
24.1
16.2
15.1 D
q/kq-day
Protected Vegetable
NA
3.9
2.5
2.14
1.85
3.55
q/kq-day
Root Vegetable
NA
7.3
4.3
3.83
2.26
3.11
q/kq-day
Other
Breast milk0
1.01
NA
NA
NA
NA
NA
kq/day
Soil (dry)
NA
400e
400e
201 r
201 r
201 r
mq/day
Water
NA
654
834
980
1537
2224
mL/day
Fish (per individual)9
NA
3.24
4.79
6.9
8.95
17
q/day
Fish (per kg BW)n
NA
0.26
0.26
0.22
0.14
0.24
q/kq-day
Total Food Ingestion Rates for Comparison Only (not in MIRC; excludes soil and water)
Total Food:
Flomegrown only'
NA
219
131
95
52
67
g/kg-day
Total Food: All
Sourcesj
NA
125
91
61
34
23.7
g/kg-day
Overestimate (ratio of
Homegrown/Total)k
NA
1.8
1.4
1.6
1.5
2.8
(unitless)
Sources: EPA 1997a (Chapter 13), unless otherwise noted.
NA = not applicable
a As indicated by the units, the ingestion rates for produce and animal products are already normalized to
consumer body weight. Ingestion rates for soil (mg/day) and water (mL/day) are not normalized to body
weight. Soil is reported as dry weight, water as volume, and the remaining values on a wet-weight basis.
b This value represents a weighted average for the 20-39 and 40-69 age groups.
c For several farm food categories, ingestion rates were not available in EPA's 1997 EFH or 2008 CSEFH
(EPA 1997a, 2008a). Ingestion rates for these child age categories were calculated using the methodology
recommended in HHRAP (EPA 2005a), Section 6.2.2.2, as described in Section 6.3.3 of this document.
Sources to develop these values included EPA 1997a and EPA 2003c.
d Infants are assumed to consume only breast milk for one year.
e This value represents an estimated "upper percentile" for children (EPA 1997a).
f These values represent soil ingestion rates for individuals who consume homegrown food products from
Stanek et al. 1997.
9 90th percentile adult fish ingestion rates are based on data from 1995-1996 and 1998 CSFII as
summarized in EPA 2002; child fish ingestion rates are based on the same survey data, but estimated by
multiplying average two-day consumption rate for children who consumed fish on one or both days of the
survey by the frequency of fish consumption (i.e., proportion of children that reported consuming fish out of
all children sampled).
h Fish ingestion rates, original data in g/day, have been normalized to body weight in this table to allow
addition into total food estimate using the mean body weight for each age category.
' Sum of 90th percentile post-cooking food food ingestion rates. This estimate is calculated by multiplying the
food ingestion rates on previous rows (excluding soil and water) by (1 -Li)x(1 -L2), where Li and L2 are the
loss rates from Exhibit 6-15. The rows are then summed to get the total post-cooking ingestion rate.
j 90th percentile total food intake rates from EPA 2008a and 2005e based on CSFII data 1994-96 and 1998;
see Exhibit 6-14 of this document.
C-2-85
-------
The final row of Exhibit 7-1 is the ratio of the two preceding rows. The magnitude of the
overestimate that results from adding 90th percentile values for 10 different categories of
homegrown foods for children is a factor of 1.8. The overestimate of the 90th percentile total
food ingestion rate for adults is larger, a factor of 2.8. This bias may be considered when
evaluating the cancer risks and noncancer hazard quotients estimated by MIRC.
7.2 Default Screening-Level Population-Specific Parameter Values
The screening-level values for body weights (BWs) for the RTR de minimis analysis, which
serve as the default values in MIRC, are mean values and are presented in Exhibit 7-2. As
stated in Section 6, EPA recommends using the mean BW for each age group when using
upper (90th) percentile values for medium ingestion rates. Use of the mean body weights
introduces no bias toward over- or underestimating risk.
Exhibit 7-2. Mean Body Weight Estimates for Adults and
Children a
Lifestage (years)
Duration (years)
Mean Body Weight (kg)
Adultb (20-70)
50
71.4
Child < 1 c
1
7.83
Child 1-2 c
2
12.6
Child 3-5 d
3
18.6
Child 6-11e
6
31.8
Child 12-19 f
8
64.2
a Sources: EPA 1997, 2008a
b These values were calculated from data presented in EPA's (1997a) Exposure
Factors Handbook.
0 These values were calculated as time-weighted average body weight (BW)
from data presented in Table 8-3 of EPA's (2008a) Child-Specific Exposure
Factors Handbook (CSEFH).
d These values were obtained directly from Table 8-3 of the 2008 CSEFH.
e These values were obtained directly from Table 8-3 of the 2008 CSEFH for
age group 6 to <11 years. The values represents a slight underestimate of BW
for ages 6 through 11 years, since 11 -year olds are not included in this CSEFH
age group.
f These values were calculated as time-weighted average BW for age groups 11
to <16 years and 16 to <21 years from Table 8-3 of the 2008 CSEFH. The
direction of the possible bias is unknown. The values match the estimate based
on Table 8-22 of the NHANES IV data as presented by Portier et al. (2007).
7.3 Default Chemical-Specific Parameter Values for Screening Analysis
Exhibit 7-3 presents chemical-specific parameter values for input to MIRC for the screening-
level analysis. Values for reference dose (RfD), cancer slope factor (CSF), bioavailability when
ingested in soil (6s), mammalian metabolism factors (MF), correction factors for belowground
produce (VGrootveg), wet deposition fractions (Fw), air to plant transfer factors (BvAG), root
concentration factors (RCF), and soil-water partition coefficient (Kds) are presented in Exhibit
7-3.
C-2-86
-------
Exhibit 7-3. Chemical-Specific Parameter Values for Input to MIRC
Parameter
Description
Benzo(a)-
pyrene
Cadmium
Mercuric
chloride
Methyl
mercury
2,3,7,8-
TCDD
Units
RfD
Reference Dose for Chemical
0
0.0005
0.0003
0.0001
1E-09
mg/kg-day
CSF
Oral carcinogenic potency slope factor
for chemical
10
0
0
0
150,000
mg/kg-day
Bs
Soil bioavailability factor for livestock
1
1
1
1
1
unitless
MF
Mammalian metabolism factor
0.01
1
1
1
1
unitless
VGrootveg
Empirical correction factor for
belowground produce, i.e., tuber or
root vegetable, to account for possible
overestimate of the transfer of
chemicals from the outside to the
inside of bulky tubers or roots (based
on carrots and potatoes)
0.01
1
1
0.01
0.01
unitless
Fw
Fraction of wet deposition that adheres
to plant surfaces; 0.2 for anions, 0.6 for
cations and most organics
0.6
0.6
0.6
0.6
0.6
unitless
BVag
Air-to-plant biotransfer factor for
aboveground produce for vapor-phase
chemical in air
124,742
0
1,800
0
65,500
[mg/g produce DW]
/ [mg/g air]
RCF
Chemical-specific root concentration
factor for tubers and root produce
9,684
0
0
0
39,999
L soil pore water/kg
root WW
Kds
Chemical-specific soil/water partition
coefficient
160,000
75
58,000
7,000
38,904.51
L soil pore water/kg
soil DW
Values presented in this Exhibit are previously presented in Exhibit 4-1, Exhibit 6-2, Error! Reference source not found., and Exhibit 6-5. However,
exact values used in the analysis are presented here, rather than values restricted by significant figures.
C-2-87
-------
Only single estimates were developed for each of these parameters for HHRAP (EPA 2005a),
and the potential direction and magnitude of bias toward over- or underestimating risks were not
investigated in this analysis. The inputs that are both chemical-specific and plant-type-specific,
as presented in Error! Reference source not found., are not repeated here. Again, only
single estimates were developed for these parameters and the potential direction and
magnitude of bias toward over- or underestimating risks were not investigated. Finally, Exhibit
7-4 presents biotransfer factors for each of the chemicals and animal types assessed in the
screening level assessment.
Exhibit 7-4. Chemical and Animal-Type Specific Biotransfer Factor (Ba) values for Input
to MIRC
Chemical
Beef
Dairy
Pork
Poultry
Eggs
Benzo(a)pyrene
0.03756242
0.00790788
0.0454703
0.02767758
0.01581576
Cadmium
0.00012
0.0000065
0.00019149
0.10625
0.0025
Mercuric chloride
0.000105
0.00000143
0.00003393
0.023925
0.023925
Methyl mercury
0.00124
0.0000169
0.00000507
0.003575
0.003575
2,3,7,8-TCDD
0.02612123
0.00549921
0.03162044
0.01924722
0.01099841
Note: Exact values used in the analysis are presented here, rather than values restricted by significant figures.
Sensitivity analyses were conducted to identify which, if any, of these parameter values
significantly influence risk estimates from MIRC; results are presented elsewhere. Further
evaluation of the possible range and distribution of values can be conducted for any parameter
that appears important to the model outputs.
7.4 Screening-Level Parameter Values for Nursing Infant Exposure
EPA also included an assessment of risk to nursing infants exposed to dioxins and to
methylmercury (MeHg) in their mother's milk for a family farming and catching fish in the area of
maximal air deposition of chemical. Input values were summarized in Section 6.5.
7.4.1 Dioxins
For dioxins, chemical intake via breast milk by nursing infants was estimated using the model
presented in EPA's Methodology for Assessing Health Risks Associated with Multiple Pathways
of Exposure to Combustor Emissions (MPE, EPA 1998a). The assumption that lactational
transfer of dioxins to the infant occurs via the lipid-phase of milk appears reasonable. The
following screening-level assumptions used in that model should bias the results toward some
overestimate of risks.
• Duration of nursing is a full year and no other foods or liquids are consumed by the
infant; a more "typical" value would be six months.
• Absorption efficiency of dioxin in food or milk by mother and infant are 100 percent; this
assumption might overestimate absorption but probably by no more than 15 percent.
• The fat content of human milk is assumed to be 4 percent, a value toward the high end
of the reported range of values (1 to 5 percent).
• The maternal chemical intake is estimated using 90th percentile ingestion rates for the
different homegrown foods (see discussion for Exhibit 7-1); this assumption might
overestimate total ingestion of homegrown foods by a factor of more than 3.
C-2-88
-------
• If the fraction of the maternal body burden of dioxin that is in the body fat compartment is
greater than 90 percent, as suggested by ATSDR (1998), then actual exposures of the
infant may be less than estimated.
There also are parameter values and model assumptions for the lipid-phase breast-milk
pathway for which possible bias is unknown.
• The accuracy of the model is unknown; it has not been verified or validated with
empirical data.
• Using a half-life of 10 years for dioxins may over- or under-estimate risks.
Finally, there is one assumption that might possibly introduce some bias toward underestimating
risks. The model results are sensitive to the biological half-life of the chemical in the mother
relative to the length of her exposure prior to the lactation period. Using an exposure duration
for the mother equal to the assumed half-life for dioxins, 10 years, may underestimate the
duration of exposure of the mother.
7.4.2 Methyl Mercury
For MeHg, empirical data from a single human study (Fujita and Takabatake 1977) was used in
conjunction with a physiologically based pharmacokinetic (PBPK) model of lactational transfer of
MeHg developed and partially validated by Byckowski and Lipscomb (2001) to support a very
simple predictive model. Both the human data and the PBPK model indicated that for relatively
low MeHg exposures, the concentration of MeHg in the nursing infant's blood is similar to its
concentration in the mother's blood. The PBPK model suggested in addition that the average
daily dose of MeHg absorbed from milk by the nursing infant (DAIinf) is indistinguishable from the
dose of MeHg absorbed by its mother from her food (DAImat). The data are limited, and the
model includes various assumptions; however, there is no known directional bias in the
estimates.
C-2-89
-------
8 REFERENCES
Altman, P.L., and D.S. Dittmer, Eds. 1964. Biology Data Book. Volume 1. Bethesda, MD. 263-
264 (As cited in U.S. Environmental Protection Agency (EPA) 1998).
Amin-Zaki, L., S. Elhassani, M.A. Majeed, T.W. Clarkson, R.A. Doherty, M.R. Greenwood, and
T. Giovanoli-Jakubczak. 1976. Perinatal methylmercury poisoing in Iraq. Am. J. Diseases in
Children 130: 1070-1076 (As cited in Byckowski and Lipscomb 2001).
Agency for Toxic Substances and Disease Registry (ATSDR). 1992. Toxicological profile for
selected PCBs (Arochlor- 1260, 1254, 1248, 1242, 1232, 1221, and 1016). Atlanta, GA: U.S.
Department of Health and Human Services, Public Health Service (As cited in U.S.
Environmental Protection Agency (EPA) 1998).
Agency for Toxic Substances and Disease Registry (ATSDR). 1998. Toxicological profile for
chlorinated dibenzo-p-dioxins. Atlanta, GA: U.S. Department of Health and Human Services,
Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 1999. Toxicological profile for
Mercury. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.
Agency for Toxic Substances and Disease Registry (ATSDR). 2000. Toxicological profile for
selected PCBs. Atlanta, GA: U.S. Department of Health and Human Services, Public Health
Service.
Bacci E., M. Cerejeira, C. Gaggi, G. Chemello, D. Calamari, and M. Vighi. 1992. Chlorinated
dioxins: Volatilization from soils and bioconcentration in plant leaves. Bull. Environ. Contam.
Toxicol. 48: 401-408.
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. ORNL-
5786. Oak Ridge National Laboratory. Oak Ridge, Tennessee. September.
Bates, M.N., D.S. Hannah, S.J. Buckland, J.A. Taucher, and T. van Mannen. 1994.
Chlorinated organic contaminants in breast milk of New Zealand women. Environmental Health
Perspectives 102(Supplement 1): 211-217.
Belcher, G.D., and C.C. Travis. 1989. Modeling support for the RURA and municipal waste
combustion projects: Final report on sensitivity and uncertainty analysis for the terrestrial food
chain model. Interagency Agreement No. 1824-A020-A1, Office of Risk Analysis, Health and
Safety Research Division, Oak Ridge National Laboratory. Oak Ridge, Tennessee. October.
Boone, F.W., Y.C. Ng, and J.M. Palm. 1981. Terrestrial pathways of radionuclide particulates.
Health Physics 41:735-747.
Bowman, R.E., S.L. Schantz, M.L. Gross, et al. 1989. Behavioral effects in monkeys exposed
to 2,3,7,8-TCDD transmitted maternally during gestation and for four months of nursing.
Chemosphere 18: 235-242 (As cited in ATSDR 1998).
Briggs, G.G., R.H. Bromilow, and A.A. Evans. 1982. Relationships between lipophilicity and
root uptake and translocation of non-ionized chemicals by barley. Pesticide Science 13: 495-
504 (As cited in U.S. Environmental Protection Agency (EPA) 2005a, Appendix A-2).
C-2-90
-------
Butte, N.F., C. Garza, E.O. Smith, and B.L. Nichols. 1984. Human milk intake and growth in
exclusively breast-fed infants. The Journal of Pediatrics 104(2): 187-195.
Byczkowski, J.Z., and J.C. Lipscomb. 2001. Physiologically based pharmacokinetic modeling
of the lactational transfer of methylmercury. Risk Analysis 21 (5): 869-882.
Chamberlain, A.C. 1970. Interception and retention of radioactive aerosols by vegetation.
Atmospheric Environment 4: 57-78.
Clewell, H.J, J.M. Gearhart, R.P. Gentry, T.R Covington, C.B. VanLandingham, C.B. Crump,
and A.M. Shipp. 1999. Evaluation of the uncertainty in an oral reference dose for
methylmercury due to interindividual variability in pharmacokinetics. Risk Analysis 19: 547-558
(As cited in Byckowski and Lipscomb 2001).
Conley, C.L. 1974. The blood. Medical Physiology 13(2). V.B. Moutcasle, Ed. 1997. The C.V.
Mosby Company, St. Louis, MO. 1027-1046 (As cited in U.S. Environmental Protection Agency
(EPA) 1998).
Ensminger, M.E. 1980. Poultry Science. Interstate Printers and Publishers, Inc. Danville,
Illinois.
Fries, G. F. 1982. Potential polychlorinated biphenyl residues in animal products from
application of contaminated sewage sludge to land. J. Environ. Qual. 11:14.
Fries, G.F. 1994. Personal communication between G.F. Fries, U.S. Department of Agriculture,
and Glenn Rice and Jennifer Windholtz, U.S. Environmental Protection Agency, Office of
Research and Development. Agricultural Research Service. March.
Fujita, M., and E. Takabatake. 1977. Mercury levels in human maternal and neonatal blood,
hair and milk. Bull. Environ. Contam. Toxicol. 18(2): 205-209.
Gearhart, J.M., H.J. Clewell III, K.S. Crump, A.M. Shipp, and A. Silvers. 1995. Pharmacokinetic
dose estimates of mercury in children and dose-response curves of performance tests in a large
epidemiological study. Water, Air, Soil Pollut. 80: 49-58 (As cited in Byckowski and Lipscomb
2001).
Gearhart, J., T. Covington, and H. Clewell III. 1996. Application of a physiologically based
pharmacokinetic model for MeHg in a dose reconstruction of the Iraqi accidental exposures.
Paper presented at the Fourth International Conference on Mercury as a Global Pollutant,
Congress Centre, Hamburg, Germany; August (As cited in Byckowski and Lipscomb 2001).
Greenwood, M.R., T.W. Clarkson, R.A. Doherty, A.H. Gates, L. Amin-Zaki, S. Elhassani, and
M.A. Majeed. 1978. Blood clearance half-times in lactating and nonlactating members of a
population exposed to mercury. Environ. Res. 16: 48-54.
Harrison, K.A. 1967. Blood volume changes in severe anemia in pregnancy. Lancet
1 (7480):20-25.
Hofelt, C.S., M. Honeycutt, J.T. McCoy, and L.C. Haws. 2001. Development of a metabolism
factor for polycyclic aromatic hydrocarbons for use in multipathway risk assessments of
hazardous waste combustion facilities. Reg. Toxicol. Pharmacol. 33:60-65.
C-2-91
-------
Hoffman, F.O., K.M. Thiessen, M.L. Frank, and B.G. Blaylock. 1992. Quantification of the
interception and initial retention of radioactive contaminants deposited on pasture grass by
simulated rain. Atmospheric Environ. 26a(18): 3313-3321.
Hollins, J.G., R.F. Willes, F.R. Bryce, S.M. Charbonneau, and I.C. Munro. 1975. The whole
body retention and tissue distribution of Hg methylmercury in adult cats. Toxicol. App.
Pharmacol. 33: 438-449.
Hong, C.S., J. Xiao, A.C. Casey, B. Bush, E.F. Fitzgerald, and S.A. Hwang. 1994. Mono-ortho
and non-ortho-substituted polychlorinated biphenyls in human milk from Mohawk and control
women: Effects of maternal factors and previous lactation. Arch. Environ. Contam. Toxicol.
27(3): 431-437.
Jensen, A.A. 1987. Polychlorinated biphenyls (PCBs), polychlorodibenzo-p-dioxins (PCDDs)
and polychlorodibenzofurans (PCDFs) in human milk, blood and adipose tissue. Sci. Total
Environ. 64(3): 259-293.
Kahn, H., and K. Stralka. 2008. Estimated daily average per capita water ingestion by child
and adult age categories based on USDA's 1994-96 and 1998 continuing survey of food intakes
individuals. J. Expo. Anal. Environ. Epidemiol. 1-9.
Kershaw, T.G., T.W. Clarkson , and P.H. Dhahir. 1980. The relationship between blood levels
and dose of methylmercury in man. Arch. Environ. Health 35(1): 28-36.
Lorber, M. 1995. Development of an air-to-plant vapor phase transfer for dioxins and furans.
Presented at the 15th International Symposium on Chlorinated Dioxins and Related
Compounds. August 21-25, 1995 in Edmonton, Canada. Abstract in Organohalogen
Compounds 24:179-186.
Lorber, M., and P. Pinsky. 2000. An evaluation of three empirical air-to-leaf models for
polychlorinated dibenzo-p-dioxins and dibenzofurans. Chemosphere 41 (6):931 -41.
Maxwell, N.I., and D.E. Burmaster. 1993. A simulation model to estimate a distribution of lipid
intake from breast milk during the first year of life. Journal of Exposure Analysis and
Environmental Epidemiology 3(4): 383-406.
McLachlan, M.S. 1993. Digestive tract absorption of polychlorinated dibenzo-p-dioxins,
dibenzofurans and biphenyls in a nursing infant. Toxicology and Applied Pharmacology 123(1):
68-72.
Miller, C.W., and F.O. Hoffman. 1983. An examination of the environmental half-time for
radionuclides deposited on vegetation. Health Physics 45(3): 731-744.
National Academy of Sciences (NAS). 1987. Predicting Feed intake of Food-Producing
Animals. National Research Council, Committee on Animal Nutrition, Washington, D.C.
National Academy of Sciences (NAS). 1991. Nutrition During Lactation. National Academies
Press. Washington, DC.
North Carolina Department of Health, Environment, and Natural Resources (NC DEHNR).
1997. North Carolina Protocol for Performing Indirect Exposure Risk Assessments for
Hazardous Waste Combustion Units. January.
C-2-92
-------
Portier K., J. Tolson, and S. Roberts. 2007. Body weight distributions for risk assessment.
Risk Anal. 27(1): 11-26.
Prentice, A.M., G.R. Goldberg, and A. Prentice. 1994. Body mass index and lactation
performance. European Journal of Clinical Nutrition 48(Supplement 3): S78-S86.
RTI. 2005. Methodology for predicting cattle biotransfer factors. Prepared for U.S.
Environmental Protection Agency (EPA) Office of Solid Waste. EPA Contract No. 68-W-03-042.
September. Available at: http://www.epa.gov/waste/hazard/tsd/td/combust/
finalmact/ssra/btfreportfull05.pdf.
Schantz, S., and R.E. Bowman. 1989. Learning in monkeys exposed perinatally to 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD). Neurotox. Teratol. 11: 13-19.
Schantz, S.L., S.A. Ferguson, and R.E. Bowman. 1992. Effects of 2,3,7,8-tetrachlorodibenzo-
p-dioxin on behavior of monkeys in peer groups. Neurotox. Teratol. 14: 433-446.
Schecter, A., P. Furst, C. Furst, O. Papke, M. Ball, J. Ryan, H. Cau, L. Dai, H. Quynh, H.Q.
Cuong, N. Phuong, P. Phiet, A. Beim, J. Constable, J. Startin, M. Samedy, and Y. Seng. 1994.
Chlorinated dioxins and dibenzofurans in human tissue from general populations: A selective
review. Environmental Health Perspectives 102(Supplement 1): 159-171.
Sherlock, J., D. Hislop, G. Newton, G. Topping, and K. Whittle. 1984. Elevation of mercury in
human blood from controlled ingestion of methylmercury in fish. Human Toxicology 3: 117-131
(As cited in USFDA 2009).
Shor, R.W., C.F. Baes, and R.D. Sharp. 1982. Agricultural production in the United States by
county: A compilation of information from the 1974 census of agriculture for use in terrestrial
food-chain transport and assessment models. Oak Ridge National Laboratory Publication.
ORNL-5786
Sim, M.R. and J.J. McNeil. 1992. Monitoring chemical exposure using breast milk: A
methodological review. American Journal of Epidemiology 136(1): 1-11.
Smith, A.H. 1987. Infant exposure assessment for breast milk dioxins and furans derived from
waste incineration emissions, risk analysis 7:347-353.
Stanek, E.J., E.J. Calabrese, R. Barnes, P. Pekow. 1997. Soil ingestion in adults - results of a
second pilot study. Toxicol. Environ. Safety 36:249-257.
Steinbeck, A.W. 1954. Plasma and blood volumes of normal Australian females. Australian
Journal of Experimental Biology and Medical Science 32(1): 95-9.
Stephens, R.D., M. Petreas, and G.H. Hayward. 1995. Biotransfer and bioaccumulation of
dioxins and furans from soil: Chickens as a model for foraging animals. The Science of the
Total Environment 175: 253-273. July 20.
Sullivan, M.J., S.R. Custance, and C.F. Miller. 1991. Infant exposure to dioxin in mother's milk
resulting from maternal ingestion of contaminated fish. Chemosphere 23(8-10): 1387-1396.
Thomann, R.V. 1989. Bioaccumulation model of organic-chemical distribution in aquatic food-
chains. Environ. Sci. Technol. 23(6): 699-707.
C-2-93
-------
Travis, C.C., and A.D. Arms. 1988. Bioconcentration of organics in beef, milk, and vegetation.
Environ. Sci. Technol. 22:271-274.
Ueland, K. 1976. Maternal cardiovascular dynamics. VII. Maternal cardiovascular dynamics.
Intrapartum blood volume changes. Am. J. Obstetrics Gynecol. 126(6): 671-677.
U.S. Department of Agriculture (USDA). 1980. Food and Nutrient Intakes of Individuals in One
Day in the United States, Spring 1977. Nationwide Food Consumption Survey 1977-1978.
Preliminary Report No. 2.
U.S. Department of Agriculture (USDA). 1992. Changes in Food Consumption and
Expenditures in American Households during the 1980's. USDA, Washington, D.C. Statistical
Bulletin o. 849. (As cited in U.S. ENVIRONMENTAL PROTECTION AGENCY (EPA) 1997)
U.S. Department of Agriculture (USDA). 1993. Food and Nutrient Intakes by Individuals in the
United States, 1 Day, 1987-88. Nationwide Food Consumption Survey 1987-88, NFCS Report
No. 87-1-1. (As cited in U.S. ENVIRONMENTAL PROTECTION AGENCY (EPA) 1997)
U.S. Department of Agriculture (USDA). 1994a. Food Consumption and Dietary Levels of
Households in the United States, 1987-88. Agricultural Research Service, Report No. 87-H-1.
(As cited in U.S. ENVIRONMENTAL PROTECTION AGENCY (EPA) 1997)
U.S. Department of Agriculture (USDA). 1994b. Vegetables 1993 Summary. National
Agricultural Statistics Service, Agricultural Statistics Board. Washington, D.C. Vg 1-2 (94). Jan.
U.S. Department of Agriculture (USDA). 1994c. Noncitrus Fruits and Nuts 1993 Summary.
National Agricultural Statistics Service, Agricultural Statistics Board, Washington, D.C. Fr Nt 1-3
(94).
U.S. Department of Agriculture (USDA). 2000. 1994-96, 1998 Continuing Survey of Food
Intakes by Individuals (CSFII). CD-ROM. Agricultural Research Service, Beltsville Human
Nutrition Research Center, Beltsville, MD. Available from the National Technical Information
Service, Springfield, VA, Accession Number PB-2000500027. (As cited in U.S.
ENVIRONMENTAL PROTECTION AGENCY (EPA) 2008a, Chapter 14)
U.S. Environmental Protection Agency (EPA). 1989. Risk Assessment Guidance for
Superfund. Volume I. Human Health Evaluation Manual (Part A). Office of Emergency and
Remedial Response (OERR), Washington, D.C. December. OERR 9200 6-303-894. Available
at: http://www.epa.gov/oswer/riskassessment/ragsa/index.htm.
U.S. Environmental Protection Agency (EPA). 1990. Interim Final Methodology for Assessing
Health Risks Associated with Indirect Exposure to Combustor Emissions. Environmental
Criteria and Assessment Office. ORD.EPA-600-90-003. January.
U.S. Environmental Protection Agency (EPA). 1992. Technical Support Document for the Land
Application of Sewage Sludge: Volumes I and II. EPA 822/R-93-001a. Office of Water.
Washington, D.C.
U.S. Environmental Protection Agency (EPA). 1993. Review Draft Addendum to the
Methodology for Assessing Health Risks Associated with Indirect Exposure to Combustor
Emissions. Office of Health and Environmental Assessment. Office of Research and
Development. EPA-600-AP-93-003. November 10.
C-2-94
-------
U.S. Environmental Protection Agency (EPA). 1994a. Revised Draft Guidance for Performing
Screening Level Risk Analysis at Combustion Facilities Burning Hazardous Wastes.
Attachment C, Draft Exposure Assessment Guidance for RCRA Hazardous Waste Combustion
Facilities. Office of Emergency and Remedial Response. Office of Solid Waste. December 14.
U.S. Environmental Protection Agency (EPA). 1994b. Estimating Exposure to Dioxin-Like
Compounds. Volume II: Properties, Sources, Occurrence, and Background Exposures.
External Review Draft. Office of Research and Development. Washington, DC. EPA/600/6-
88/005Cc. June.
U.S. Environmental Protection Agency (EPA). 1994c. Estimating Exposure to Dioxin-Like
Compounds. External Review Draft. Office of Research and Development, Washington, D.C.
EPA/600/6-88/005Cb. June. Available at:
http://oaspub.epa. gov/eims/eimscomm.getfile?p_download_id=438673.
U.S. Environmental Protection Agency (EPA). 1995a. Review Draft Development of Human
Health-Based and Ecologically-Based Exit Criteria for the Hazardous Waste Identification
Project. Volumes I and II. Office of Solid Waste. March 3.
U.S. Environmental Protection Agency (EPA). 1995b. Memorandum Regarding Further
Studies for Modeling the Indirect Exposure Impacts from Combustor Emissions. From Mathew
Lorber, Exposure Assessment Group, and Glenn Rice, Indirect Exposure Team, Environmental
Criteria and Assessment Office. Washington, D.C. January 20.
U.S. Environmental Protection Agency (EPA). 1995c. Further Issues for Modeling the Indirect
Exposure Impacts from Combustor Emissions. Office of Research and Development.
Washington, D.C. January 20.
U.S. Environmental Protection Agency (EPA). 1995d. Waste Technologies Industries
Screening Human Health Risk Assessment (SHHRA): Evaluation of Potential Risk from
Exposure to Routine Operating Emissions. Volume V. External Review Draft. U.S. EPA
Region 5, Chicago, Illinois.
U.S. Environmental Protection Agency (EPA). 1997a. Exposure Factors Handbook. Volumes
I, II, and III. Office of Research and Development, Washington, D.C. EPA-600-P-95-002Fa,b,c.
August. Available at: http://www.epa.gov/nceawww1/efh/.
U.S. Environmental Protection Agency (EPA). 1997b. Mercury Study Report to Congress.
Volume III: Fate and Transport of Mercury in the Environment. Office of Air Quality Planning
and Standards and Office of Research and Development. EPA-452/R-97-005. December.
U.S. Environmental Protection Agency (EPA). 1997c. Parameter Guidance Document.
National Center for Environmental Assessment, NCEA-0238.
U.S. Environmental Protection Agency (EPA). 1998. Methodology for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions. National Center for
Environmental Assessment, Cincinnati, OH. EPA-600-R-98-137. Available at:
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=55525.
U.S. Environmental Protection Agency (EPA). 1999a. Background Document for the Human
Risk Module for HWIR99 Multimedia, Multipathway, and Multireceptor Risk Assessment (3MRA)
C-2-95
-------
Model. Office of Solid Waste, Washington, D.C. October. Available at:
http://epa.gov/epawaste/hazard/wastetypes/wasteid/hwirwste/pdf/risk/modules/s0054.pdf.
U.S. Environmental Protection Agency (EPA). 1999b. 1999 National-Scale Air Toxics
Assessment Results; Approach for Modeling POM. Available at:
http://www.epa.gov/ttn/atw/nata1999/nsata99.html.
U.S. Environmental Protection Agency (EPA). 2000. Supplementary Guidance for Conducting
Health Risk Assessment of Chemical Mixtures. Risk Assessment Forum, Washington D.C.
EPA/630/R-00/002. August. Available at:
http://www.epa.gov/ncea/raf/pdfs/chem_mix/chem_mix_08_2001 .pdf.
U.S. Environmental Protection Agency (EPA). 2001a. Peer Review of EPA's Hazardous Waste
Identification Rule Risk Assessment Model: Breast milk exposure model for the HWIR 3MRA
Model. Prepared by Eastern Research Group for EPA Office of Solid Wastes. 68-W5-0057.
U.S. Environmental Protection Agency (EPA). 2001b. Water Quality Criterion for the Protection
of Human Health: Methylmercury. Office of Water, Office of Science and Technology.
Washington D.C. EPA-823-R-01-001. January. Available at:
http://www.epa.gov/ncea/raf/pdfs/chem_mix/chem_mix_08_2001 .pdf.
U.S. Environmental Protection Agency (EPA). 2002. Estimated Per capita Fish Consumption in
the United States. Office of Water, Office of Science and Technology, Washington, D.C. EPA-
821-C- 02-003. August. Available at:
http://www.epa.gov/waterscience/fish/files/consumption_report.pdf.
U.S. Environmental Protection Agency (EPA). 2003a. Chapter 10 In: Multimedia, Multipathway,
and Multireceptor Risk Assessment (3MRA) Modeling System, Volume II: Site-based, Regional,
and National Data. SAB Review Draft. EP-530/D-03-001b. Office of Research and
Development, Athens, GA, and Research Triangle Park, NC, and Office of Solid Waste,
Washington, D.C. July. Available at:
http://www.epa.gov/osw/hazard/wastetypes/wasteid/hwirwste/risk03.htm.
U.S. Environmental Protection Agency (EPA). 2003b. Methodology for Deriving Ambient Water
Quality Criteria for the Protection of Human Health (2000): Technical Support Document.
Volume 2: Development of National Bioaccumulation Factors. Office of Water, Office of
Science and Technology, Washington, D.C. EPA-822-R-03-030. December. Available at:
http://www.epa.gov/waterscience/criteria/humanhealth/method/tsdvol2.pdf.
U.S. Environmental Protection Agency (EPA). 2003c. CSFII Analysis of Food Intake
Distributions. Office of Research and Development, National Center for Environmental
Assessment, Washington, D.C. EPA-600-R-03-29. Available at:
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=56610.
U.S. Environmental Protection Agency (EPA). 2004a. Superfund Chemical Data Matrix.
OERR. Washington, D.C. January.
U.S. Environmental Protection Agency (EPA). 2004b. Estimated Per capita Water Ingestion
and Body Weight in the United States - An Update. Office of Water, Office of Science and
Technology, Washington, D.C. EPA-822-R-00-001. October. Available at:
http://www.epa.gov/waterscience/criteria/drinking/percapita/2004.pdf
C-2-96
-------
U.S. Environmental Protection Agency (EPA). 2004c. Risk Assessment Guidance for
Superfund. Volume I: Human Health Evaluation Manual (Part E, Supplemental Guidance for
Dermal Risk Assessment), Final. Office of Superfund Remediation and Technology Innovation,
Washington, D.C. EPA/540/R/99/005; OSWER 9285.7-02EP; NTIS PB99-963312. July.
U.S. Environmental Protection Agency (EPA). 2005a. Human Health Risk Assessment
Protocol for Hazardous Waste Combustion Facilities. Office of Solid Waste and Emergency
Response, Washington, DC. EPA-530-R-05-006. September. Available at:
http://www.epa.gov/osw/hazard/tsd/td/combust/risk.htm.
U.S. Environmental Protection Agency (EPA). 2005b. Guidance on Selecting Age Groups for
Monitoring and Assessing Childhood Exposures to Environmental Contaminants. Risk
Assessment Forum. Washington, DC. November. EPA/630/P-03/003F. Available at:
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=146583.
U.S. Environmental Protection Agency (EPA). 2005c. Guidelines for Carcinogen Risk
Assessment. Risk Assessment Forum, Washington, DC. EPA/630/P-03/001F. March.
Available from: http://www.epa.gov/IRIS/cancer032505-final.pdf.
U.S. Environmental Protection Agency (EPA). 2005d. Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens. Risk Assessment Forum: Washington,
D.C. EPA-630/R-03-003F. March. Available at:
http://www.epa.gov/ttn/atw/childrens_supplement_final.pdf.
U.S. Environmental Protection Agency (EPA). 2005e. Analysis of Total Food Intake and
Composition of Individual's Diet Based on the U.S. Department of Agriculture's 1994-96, 1998
Continuing Survey of Food Intakes By Individuals (CSFII) (Final). Office of Research and
Development, National Center for Environmental Assessment, Washington, D.C. EPA/600/R-
05/062F. Available at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=132173.
U.S. Environmental Protection Agency (EPA). 2007a. Prioritized Chronic Dose-Response
Values for Screening Risk Assessments (Table 1). Office of Air Quality Planning and
Standards; June 12, 2007. Available at: http://www.epa.gov/ttn/atw/toxsource/summary.html.
U.S. Environmental Protection Agency (EPA). 2007b. Toxicological Review of 1,1,1-
Trichloroethane (CAS No. 71-55-6) In Support of Summary Information on the Integrated Risk
Information System (IRIS). Office of Research and Development, Washington, DC.
EPA/635/R-03/006. August. Available at: http://www.epa.gov/iris
U.S. Environmental Protection Agency (EPA). 2008a. Child-Specific Exposure Factors
Handbook. Office of Research and Development, Washington, D.C. EPA/600/R-06/096F.
September. Available at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=199243 .
U.S. Environmental Protection Agency (EPA). 2008b. Draft Report on EPA OAQPS Risk and
Technology Review Methodologies: For Review by the EPA Science Advisory Board; Case
Studies - MACT I Petroleum Refining Sources, Portland Cement Manufacturing. Office of Air
Quality Planning and Standards, Office of Air and Radiation, Research Triangle Park, NC. July
14, 2008.
U.S. Environmental Protection Agency (EPA). 2009. Guidance for Implementing the January
2001 Methylmercury Water Quality Criterion. Office of Water, Office of Science and
C-2-97
-------
Technology, Washington, D.C. EPA 823-R-09-002. January. Available at:
http://www.epa.gov/waterscience/criteria/methylmercury/pdf/guidance-final.pdf.
U.S. Food and Drug Administration (FDA). 2009. Draft Risk and Benefit Assessment Report
and Draft Summary of Published Research. Center for Food Safety and Applied Nutrition.
January 15. Available at: http://www.cfsan.fda.gov/seafood1.html
Van den Berg, M., L.S. Birnbaum, M. Denison, M. De vito, W. Farlans, M. Feeley, H. Fiedler, H.
Hakansson, A. Hanberg, L.. Haws, M. Rose, S. Safe, D. Schrenk, C. Tohyama, A. Tritscher, J.
tuomisto, M. Tysklind, N. Walker, and R.E. Peterson. 2006. The 2005 World Health
Organization reevaluation of human and mammalian toxic equivalency factors for dioxins and
dioxin-like compounds. Toxicol Sci. 93(2): 223-41.
Whitehead, R.G., and A.A. Paul. 1981. Infant growth and human milk requirements: A fresh
approach. Lancet 2(8239): 161-163.
C-2-98
-------
ATTACHMENT C-3: Systematic Sensitivity Analysis Variables and
Results
-------
[This page intentionally left blank.]
C-3-i
-------
TABLE OF CONTENTS
1 Systematic Sensitivity Analysis Variables and Results 1
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis 2
Exhibit 1-2. Elasticities and Rankings for the Variables with the Highest Elasticities for
Benzo[a]Pyrene 18
Exhibit 1-3. Elasticities and Rankings for the Variables with the Highest Elasticities for
2,3,7,8-TCDD 11
Exhibit 1-4. Elasticities and Rankings for the Variables with the Highest Elasticities for
Cadmium 13
Exhibit 1-5. Elasticities and Rankings for the Variables with the Highest Elasticities for
Divalent Mercury 15
Exhibit 1-6. Elasticities and Rankings for the Variables with the Highest Elasticities for
Methyl Mercury 18
C-3-ii
-------
1 Systematic Sensitivity Analysis Variables and Results
This attachment provides the tables of the variables included in the systematic sensitivity analysis in
Exhibit 1. The variables are organized into three categories: TRIM.FaTE variables, MIRC farm food chain
variables, and MIRC ingestion and bodyweight variables.
This attachment also provides detailed elasticities and rankings for the variables with the highest
elasticities for Benzo[a]Pyrene (Exhibit 2), 2,3,7,8 - TCDD (Exhibit 3), cadmium (Exhibit 4), divalent
mercury (Exhibit 5), and methyl mercury (Exhibit 6). In each case, the elasticities and rankings are
provided for the local systematic sensitivity analysis (variables perturbed up and down by 5%) and the
range systematic sensitivity analysis (variables perturbed up and down by 50%).
C-3-1
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
TRIM.FaTE Variables
AirTemp
Ambient Air Temperature
AlgaeCarbonCont
Carbon Content of Algae in Pond
AlgaeDensity
Density of Algae in Pond
AlgaeGrowthRate
Growth Rate of Algae in Pond
AlgaeRadius
Radius of Single Algae in Pond
Biomass
Biomass in Pond for all Aquatic Species
ChlorideConc
Water- Chloride Concentration
ChlorophyllConc
Water- Chlorophyll Concentration
EmissionRate
Emission Rate of all PBHAPs
EroRate
Erosion Rate for all Parcels
FishMass
Fish Body Weight for all Aquatic Species
HorizWindSpeed
Horizontal Wind Speed
MixH eight
Mixing Height
Rain
Annual Rainfall
RootSoilAir
Root Soil- Fraction Air
RootSoilOCC
Root Soil- Organic Carbon Fraction
RootSoilpH
Root Soil- pH
RootSoilSand
Root Soil- Fraction Sand
RootSoilVertVel
Root Soil- Average vertical velocity of water (percolation)
RootSoilWat
Root Soil- Fraction Water
RunoffRate
Total Water Runoff Rate
SedDepVel,
SedRusVel
Sediment Deposition Velocity and Resuspension Velocity
SedOCC
Sediment- Organic Carbon Fraction
SedpH
Sediment- pH
SedPorosity
Sediment Porosity
SedSand
Sediment- Fraction Sand
StackH eight
Stack Height
SurfSoilAir
Surface Soil- Fraction Air
SurfSoilOCC
Surface Soil- Organic Carbon Fraction
SurfSoilpH
Surface Soil- pH
SurfSoilSand
Surface Soil- Fraction Sand
SurfSoilVertVel
Surface Soil- Average vertical velocity of water (percolation)
SurfSoilWater
Surface Soil- Fraction Water
SurfWatTemp
Water Temperature
WatFractSand
Water- Fraction Sand
WatOCC
Water- Organic Carbon Fraction
WatpH
Water- pH
WatRetTime
Rentention Time in Pond
WatSuspSed
Water- Suspended Sediment Concentration
C-3-2
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
MIRC Farm Food Chain Variables
100-MAF ef
100 - Moisture adjustment factor, Exposed Fruit
100-MAF ev
100 - Moisture adjustment factor, Exposed Vegetable
100-MAF_pf
100 - Moisture adjustment factor, Protected Fruit
100-MAF_pv
100 - Moisture adjustment factor, Protected Vegetable
100-MAF rv
100 - Moisture adjustment factor, Root Vegetable
Ba beef
PBHAP-specific biotransfer factor, Beef
Ba_dairy
PBHAP-specific biotransfer factor, Dairy
Ba_egg
PBHAP-specific biotransfer factor, Eggs
Ba_pork
PBHAP-specific biotransfer factor, Pork
Ba_poultry
PBHAP-specific biotransfer factor, Poultry
Br_AG_produce_ef
Plant-soil PBHAP bioconcentration factor, Exposed Fruit
Br_AG_produce_ev
Plant-soil PBHAP bioconcentration factor, Exposed Vegetables
Br_AG_prod uce_fo
Plant-soil PBHAP bioconcentration factor, Forage Feed
B r_AG_prod u ce_g r
Plant-soil PBHAP bioconcentration factor, Grain Feed
Br_AG_produce_pf
Plant-soil PBHAP bioconcentration factor, Protected Fruit
Br_AG_produce_pv
Plant-soil PBHAP bioconcentration factor, Protected Vegetable
B r_AG_prod u ce_s i
Plant-soil PBHAP bioconcentration factor, Silage Feed
Br_AG_rootveg
Plant-soil PBHAP bioconcentration factor, Root Vegetables
Bs a b
Soil bioavailability factor for livestock
Bv_AG
Air-to-plant biotransfer factor for aboveground produce for vapor-phase
PBHAP in air
C FishT3
Concentration of PBHAP in whole fish for T3 fish
C FishT4
Concentration of PBHAP in whole fish for T4 fish
C Soil
Concentration of PBHAP in soil from contaminated area
Ca
Average annual total PBHAP concentration in air
Cs root zone feed
Average soil concentration in contaminated area used to grow animal feed
Cs_root_zone_
produce
Average PBHAP concentration in soil at root-zone depth in produce-growing
area
Cs_S_animal_
ingest
PBHAP concentration in surface soil in contaminated area where livestock feed
Drdp
Average annual dry deposition of particle-phase PBHAP
Drwp
Average annual wet deposition of particle-phase PBHAP
FvDc
Fraction of airborne PBHAP in vapor phase
Fw
Fraction of wet deposition that adheres to plant surfaces
Kds
PBHAP-specific soil/water partition coefficient
Kp_ef
Plant surface loss coefficient, Exposed Fruit
Kp_ev
Plant surface loss coefficient, Exposed Vegetable
Kp_fo
Plant surface loss coefficient, Forage
Kp_si
Plant surface loss coefficient, Silage
L1 beef
Loss type 1, Beef
L1_ExpFruit
Loss type 1, Exposed Fruit
L1_ExpVeg
Loss type 1, Exposed Vegetable
L1 Fish
Loss type 1, Fish
L1_pork
Loss type 1, Pork
C-3-3
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
L1_poultry
Loss type 1, Poultry
L1 ProFruit
Loss type 1, Protected Fruit
L1_ProVeg
Loss type 1, Protected Vegetable
L1_RootVeg
Loss type 1, Root Vegetable
L2 beef
Loss type 2, Beef
L2_ExpFruit
Loss type 2, Exposed Fruit
L2 Fish
Loss type 2, Fish
L2_pork
Loss type 2, Pork
L2_poultry
Loss type 2, Poultry
L2_RootVeg
Loss type 2, Root Vegetable
MF d,e
Mammalian metabolism factor
P_a
Density of air
Qp_fo_beef
Quant
ty of forage plant type eaten per animal per day, Beef
Qp_fo_dairy
Quant
ty of forage plant type eaten per animal per day, Dairy
Qp_gr_beef
Quant
ty of grain plant type eaten per animal per day, Beef
Qp_gr_dairy
Quant
ty of grain plant type eaten per animal per day, Dairy
Qp_gr_egg
Quant
ty of grain plant type eaten per animal per day, Eggs
Qp_gr_pork
Quant
ty of grain plant type eaten per animal per day, Pork
Qp_gr_poultry
Quant
ty of grain plant type eaten per animal per day, Poultry
Qp_si_beef
Quant
ty of silage plant type eaten per animal per day, Beef
Qp_si_dairy
Quant
ty of silage plant type eaten per animal per day, Dairy
Qp_si_pork
Quant
ty of silage plant type eaten per animal per day, Pork
Qs beef
Quant
ty of soil eaten by the animal each day, Beef
Qs_dairy
Quant
ty of soil eaten by the animal each day, Dairy
Qs_egg
Quant
ty of soil eaten by the animal each day, Eggs
Qs_pork
Quant
ty of soil eaten by the animal each day, Pork
Qs_poultry
Quant
ty of soil eaten by the animal each day, Poultry
RCF
PBHAP-specific root concentration factor for tubers and root produce
Rp_ef
Interception fraction of the edible portion of plant, Exposed Fruit
Rp_ev
Interception fraction of the edible portion of plant, Exposed Vegetable
Rp_fo
Interception fraction of the edible portion of plant, Forage
Rp_si
Interception fraction of the edible portion of plant, Silage
Tp_ef
Length of plant exposure to deposition per harvest, Exposed Fruit
Tp_ev
Length of plant exposure to deposition per harvest, Exposed Vegetable
Tp_fo
Length of plant exposure to deposition per harvest, Forage
Tp_si
Length of plant exposure to deposition per harvest, Silage
VG AG eff,s
Empirical correction factor, Exposed Fruit
VG AG evf,s
Empirical correction factor, Exposed Vegetable
VG AG foab
Empirical correction factor, Forage
VG AG si
Empirical correction factor, Silage
VG_rootveg f 9
Empirical correction factor, Root Vegetable
Yp_ef
Yield or standing crop biomass, Exposed Fruit
Yp_ev
Yield or standing crop biomass, Exposed Vegetable
Yp_fo
Yield or standing crop biomass, Forage
Yp_si
Yield or standing crop biomass, Silage
C-3-4
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
MIRC Ingestion and Body Weight Variables
BW adit
Body weight, Adult 20-70
BW ch1
Body weight, Child 1-2
BW ch2
Body weight, Child 3-5
BW ch3
Body Weight, Child 6-11
BW ch4
Body Weight, Child 12-19
EF beefa,b
Exposure factor, Beef
EF_dairy a D
Exposure factor, Dairy
EF_egga
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
IR beef ch2
Ingest
on rate, Beef, Child 3-5
IR beef ch3
Ingest
on rate, Beef, Child 6-11
IR beef ch4
Ingest
on rate, Beef, Child 12-19
I R_d a i ry_ad It
Ingest
on rate, Dairy, Adult 20-70
IR_dairy_ch1
Ingest
on rate, Dairy, Child 1-2
IR_dairy_ch2
Ingest
on rate, Dairy, Child 3-5
IR_dairy_ch3
Ingest
on rate, Dairy, Child 6-11
IR_dairy_ch4
Ingest
on rate, Dairy, Child 12-19
IR_egg_adlt
Ingest
on rate, Eggs, Adult 20-70
IR_egg_ch1
Ingest
on rate, Eggs, Child 1-2
IR_egg_ch2
Ingest
on rate, Eggs, Child 3-5
IR_egg_ch3
Ingest
on rate, Eggs, Child 6-11
IR_egg_ch4
Ingest
on rate, Eggs, Child 12-19
IR_ExpFruit_adlt
Ingest
on rate, Exposed Fruit, Adult 20-70
IR_ExpFruit_ch1
Ingest
on rate, Exposed Fruit, Child 1-2
IR_ExpFruit_ch2
Ingest
on rate, Exposed Fruit, Child 3-5
IR_ExpFruit_ch3
Ingest
on rate, Exposed Fruit, Child 6-11
IR_ExpFruit_ch4
Ingest
on rate, Exposed Fruit, Child 12-19
IR_ExpVeg_adlt
Ingest
on rate, Exposed Vegetable, Adult 20-70
IR_ExpVeg_ch1
Ingest
on rate, Exposed Vegetable, Child 1-2
IR_ExpVeg_ch2
Ingest
on rate, Exposed Vegetable, Child 3-5
IR_ExpVeg_ch3
Ingest
on rate, Exposed Vegetable, Child 6-11
IR_ExpVeg_ch4
Ingest
on rate, Exposed Vegetable, Child 12-19
IR Fish adit
Ingest
on rate, Fish, Adult 20-70
IR Fish ch1
Ingest
on rate, Fish, Child 1-2
IR Fish ch2
Ingest
on rate, Fish, Child 3-5
IR Fish ch3
Ingest
on rate, Fish, Child 6-11
IR Fish ch4
Ingest
on rate, Fish, Child 12-19
IR_pork_adlt
Ingest
on rate, Pork, Adult 20-70
IR_pork_ch1
Ingest
on rate, Pork, Child 1-2
IR_pork_ch2
Ingest
on rate, Pork, Child 3-5
IR_pork_ch3
Ingest
on rate, Pork, Child 6-11
IR_pork_ch4
Ingest
on rate, Pork, Child 12-19
IR_poultry_adlt
Ingest
on rate, Poultry, Adult 20-70
IR_poultry_ch1
Ingest
on rate, Poultry, Child 1-2
IR_poultry_ch2
Ingest
on rate, Poultry, Child 3-5
IR_poultry_ch3
Ingest
on rate, Poultry, Child 6-11
IR_poultry_ch4
Ingest
on rate, Poultry, Child 12-19
IR ProFruit adit
Ingest
on rate, Protected Fruit, Adult 20-70
IR ProFruit ch1
Ingest
on rate, Protected Fruit, Child 1-2
IR ProFruit ch2
Ingest
on rate, Protected Fruit, Child 3-5
IR ProFruit ch3
Ingest
on rate, Protected Fruit, Child 6-11
IR ProFruit ch4
Ingest
on rate, Protected Fruit, Child 12-19
IR_ProVeg_adlt
Ingest
on rate, Protected Vegetable, Adult 20-70
IR_ProVeg_ch1
Ingest
on rate, Protected Vegetable, Child 1-2
C-3-6
-------
Exhibit 1-1. Variables Included in the Systematic Sensitivity Analysis
Variable Name
Variable Description
IR_ProVeg_ch2
Ingestion rate, Protected Vegetable, Child 3-5
IR_ProVeg_ch3
Ingestion rate, Protected Vegetable, Child 6-11
IR_ProVeg_ch4
Ingestion rate, Protected Vegetable, Child 12-19
IR_RootVeg_adlt
Ingestion rate, Root Vegetable, Adult 20-70
IR_RootVeg_ch1
Ingestion rate, Root Vegetable, Child 1-2
IR_RootVeg_ch2
Ingestion rate, Root Vegetable, Child 3-5
IR_RootVeg_ch3
Ingestion rate, Root Vegetable, Child 6-11
IR_RootVeg_ch4
Ingestion rate, Root Vegetable, Child 12-19
IR Soil adit
Ingestion rate, Soil, Adult 20-70
IR Soil ch1
Ingestion rate, Soil, Child 1-2
IR Soil ch2
Ingestion rate, Soil, Child 3-5
IR Soil ch3
Ingestion rate, Soil, Child 6-11
IR Soil ch4
Ingestion rate, Soil, Child 12-19
IR water adit
Ingestion rate, Water, Adult 20-70
IR water ch1
Ingestion rate, Water, Child 1-2
IR water ch2
Ingestion rate, Water, Child 3-5
IR water ch3
Ingestion rate, Water, Child 6-11
IR water ch4
Ingestion rate, Water, Child 12-19
a Values cannot be increased by 5
/o because the variable has an upper bound.
b Values cannot be increased by 50% because the variable has an upper bound.
c Values can only be increased by
5% for 2,3,7,8 - TCDD and Benzo[a]Pyrene because the variable has an upper bound.
d Values can only be increased by
5% for Benzo[a]Pyrene because the variable has an upper bound.
e Values can only be increased by 50% for Benzo[a]Pyrene because the variable has an upper bound.
f Values can only be increased by J
¦>% for Benzo[a]Pyrene, 2,3,7,8-TCDD, and methyl mercury because the variable has an upper bound.
g Values can only be increased by
5% for Benzo[a]Pyrene, 2,3,7,8-TCDD, and methyl mercury because the variable has an upper bound.
C-3-7
-------
Exhibit 1-2. Elasticities and Rankings for the Variables with the Highest Elasticities for Benzo[a]Pyrene
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
MixHeight
Mixing Height
TRIM
-1.97
-1.05
-0.95
-0.66
1
1
2
2
EmissionRate
Emission Rate of all Chemicals
TRIM
1.00
1.00
1.00
1.00
3
2
1
1
HorizWindSpeed
Horizontal Wind Speed
TRIM
-1.74
-0.97
-0.88
-0.63
2
3
3
3
Rain
Annual Rainfall
TRIM
0.63
0.63
0.63
0.63
5
4
4
4
Drwp
Average annual wet deposition of
particle-phase chemical
FFC
0.45
0.45
0.45
0.45
6
5
5
5
Fw
Fraction of wet deposition that
adheres to plant surfaces
FFC
0.45
0.45
0.45
0.45
7
6
6
6
MF
Mammalian metabolism factor
FFC
0.41
0.41
0.41
0.41
9
7
7
7
Yp_fo
Yield or standing crop biomass,
Forage
FFC
-0.69
-0.36
-0.33
-0.23
4
8
10
12
Rp_fo
Interception fraction of the edible
portion of plant, Forage
FFC
0.34
0.34
0.34
0.34
10
9
8
8
Ba_dairy
Chemical-specific biotransfer
factor, Dairy
FFC
0.34
0.34
0.34
0.34
12
10
9
9
EF_dairy
Exposure factor, Dairy
Ing./BW
0.34
0.34
N/A
N/A
13
11
FC_dairy
Fraction contaminated, Dairy
Ing./BW
0.34
0.34
N/A
N/A
14
12
Drdp
Average annual dry deposition of
particle-phase chemical
FFC
0.30
0.30
0.30
0.30
16
13
11
10
Qp_fo_dairy
Quantity of forage plant type
eaten per animal per day, Dairy
FFC
0.29
0.29
0.29
0.29
18
14
12
11
F_fo_dairy
Fraction of forage plant type
obtained from contaminated area
used to grow feed, Dairy
Ing./BW
0.29
0.29
N/A
N/A
17
15
Kp_fo
Plant surface loss coefficient,
Forage
FFC
-0.34
-0.25
-0.24
-0.19
11
16
13
15
Yp_ef
Yield or standing crop biomass,
Exposed Fruit
FFC
-0.43
-0.23
-0.20
-0.14
8
17
16
17
100-MAF_ef
100 - Moisture adjustment factor,
Exposed Fruit
FFC
0.22
0.22
0.22
0.22
22
18
14
13
EF_ExpFruit
Exposure factor, Exposed Fruit
Ing./BW
0.22
0.22
N/A
N/A
20
19
C-3-8
-------
Exhibit 1-2. Elasticities and Rankings for the Variables with the Highest Elasticities for Benzo[a]Pyrene
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
FC_ExpFruit
Fraction contaminated, Exposed
Fruit
Ing./BW
0.22
0.22
N/A
N/A
21
20
Rp_ef
Interception fraction of the edible
portion of plant, Exposed Fruit
FFC
0.21
0.21
0.21
0.21
23
21
15
14
Kp_ef
Plant surface loss coefficient,
Exposed Fruit
FFC
-0.27
-0.19
-0.17
-0.13
19
22
17
18
Yp_ev
Yield or standing crop biomass,
Exposed Vegetable
FFC
-0.33
-0.17
-0.16
-0.11
15
23
19
21
100-MAF_ev
100 - Moisture adjustment factor,
Exposed Vegetable
FFC
0.17
0.17
0.17
0.17
26
24
18
16
EF_ExpVeg
Exposure factor, Exposed
Vegetables
Ing./BW
0.17
0.17
N/A
N/A
27
25
FC_ExpVeg
Fraction contaminated, Exposed
Vegetable
Ing./BW
0.17
0.17
N/A
N/A
28
26
Rp_ev
Interception fraction of the edible
portion of plant, Exposed
Vegetable
FFC
0.16
0.16
N/A
N/A
29
27
Kp_ev
Plant surface loss coefficient,
Exposed Vegetable
FFC
-0.21
-0.14
-0.13
-0.10
24
28
20
25
C_Soil
Concentration of chemical in soil
from contaminated area
FFC
0.12
0.12
0.12
0.12
31
29
21
19
EF Soil
Exposure factor, Soil
Ing./BW
0.12
0.12
N/A
N/A
32
30
FC Soil
Fraction contaminated, Soil
Ing./BW
0.12
0.12
N/A
N/A
33
31
IR_dairy_ch1
Ingestion rate, Dairy, Child 1-2
Ing./BW
0.11
0.11
0.11
0.11
34
32
22
20
IR_ExpFruit_adlt
Ingestion rate, Exposed Fruit,
Adult 20-70
Ing./BW
0.10
0.10
0.10
0.10
35
33
23
23
Tp_fo
Length of plant exposure to
deposition per harvest, Forage
FFC
0.17
0.10
0.09
0.06
25
34
26
31
I R_d a i ry_ad It
Ingestion rate, Dairy, Adult 20-70
Ing./BW
0.10
0.10
0.10
0.10
36
35
24
24
L2_ExpFruit
Loss type 2, Exposed Fruit
FFC
-0.10
-0.10
-0.10
-0.10
37
36
25
26
a FFC refers to farm food chain variables and Ing./BW refers to ingestion and body weight variables.
C-3-9
-------
Exhibit 1-3. Elasticities and Rankings for the Variables with the Highest Elasticities for 2,3,7,8 - TCDD
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
MixHeight
Mixing Height
TRIM
-1.99
-1.05
-0.95
-0.67
1
1
2
2
EmissionRate
Emission Rate of all Chemicals
TRIM
1.00
1.00
1.00
1.00
3
2
1
1
HorizWindSpeed
Horizontal Wind Speed
TRIM
-1.80
-0.96
-0.87
-0.61
2
3
3
3
FC Fish
Fraction contaminated, Fish
Ing./BW
0.56
0.56
N/A
N/A
8
4
EF Fish
Exposure factor, Fish
Ing./BW
0.56
0.56
N/A
N/A
7
5
BW adit
Body weight, Adult 20-70
Ing./BW
-0.86
-0.45
-0.41
-0.29
4
6
6
11
C_FishT4
Concentration of chemical in
whole fish forT4 fish
FFC
0.43
0.43
0.43
0.43
9
7
4
5
IR Fish adit
Ingestion rate, Fish, Adult 20-70
Ing./BW
0.43
0.43
0.43
0.43
10
8
5
6
MF
Mammalian metabolism factor
FFC
0.43
0.43
N/A
N/A
11
9
SurfWatTemp
Water Temperature
TRIM
0.32
0.39
0.41
0.53
16
10
7
4
WatOCC
Water- Organic Carbon Fraction
TRIM
-0.62
-0.38
-0.35
-0.25
5
11
9
14
SedDepVel
Sediment Deposition Velocity and
Resuspension Velocity
TRIM
0.28
0.36
0.39
0.30
20
12
8
8
Ba_dairy
Chemical-specific biotransfer
factor, Dairy
FFC
0.32
0.32
0.32
0.32
13
13
10
7
EF_dairy
Exposure factor, Dairy
Ing./BW
0.32
0.32
N/A
N/A
14
14
FC_dairy
Fraction contaminated, Dairy
Ing./BW
0.32
0.32
N/A
N/A
15
15
F_T3
Fraction offish intake that is from
T3
Ing./BW
-0.30
-0.30
-0.30
-0.30
17
16
12
10
F_T4
Fraction offish intake that is from
T4
Ing./BW
0.30
0.30
0.30
0.30
18
17
11
9
Yp_fo
Yield or standing crop biomass,
Forage
FFC
-0.56
-0.30
-0.27
-0.19
6
18
15
19
Rp_fo
Interception fraction of the edible
portion of plant, Forage
FFC
0.28
0.28
0.28
0.28
19
19
13
12
Drdp
Average annual dry deposition of
particle-phase chemical
FFC
0.28
0.28
0.28
0.28
22
20
14
13
WatSuspSed
Water- Suspended Sediment
Concentration
TRIM
-0.37
-0.25
-0.24
-0.19
12
21
16
18
C-3-10
-------
Exhibit 1-3. Elasticities and Rankings for the Variables with the Highest Elasticities for 2,3,7,8 - TCDD
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
Qp_fo_dairy
Quantity of forage plant type
eaten per animal per day, Dairy
FFC
0.23
0.23
0.23
0.23
24
22
17
15
F_fo_dairy
Fraction of forage plant type
obtained from contaminated area
used to grow feed, Dairy
Ing./BW
0.23
0.23
N/A
N/A
23
23
Kp_fo
Plant surface loss coefficient,
Forage
FFC
-0.28
-0.21
-0.20
-0.16
21
24
19
21
I R_d a i ry_ad It
Ingestion rate, Dairy, Adult 20-70
Ing./BW
0.20
0.20
0.20
0.20
25
25
18
16
Rain
Annual Rainfall
TRIM
0.20
0.20
0.20
0.19
26
26
20
17
SurfSoilAir
Surface Soil- Fraction Air
TRIM
0.13
0.14
0.15
0.17
29
27
21
20
C_FishT3
Concentration of chemical in
whole fish forT3 fish
FFC
0.13
0.13
0.13
0.13
28
28
22
22
Cs_S_animal_ing
est
Chemical concentration in
surface soil in contaminated area
where livestock feed
FFC
0.12
0.12
0.12
0.12
31
29
23
23
Bs
Soil bioavailability factor for
livestock
FFC
0.12
0.12
N/A
N/A
30
30
Ba_beef
Chemical-specific biotransfer
factor, Beef
FFC
0.09
0.09
0.09
0.09
33
31
24
24
EF beef
Exposure factor, Beef
Ing./BW
0.09
0.09
N/A
N/A
34
32
FC beef
Fraction contaminated, Beef
Ing./BW
0.09
0.09
N/A
N/A
35
33
Tp_fo
Length of plant exposure to
deposition per harvest, Forage
FFC
0.14
0.08
0.08
0.05
27
34
25
29
Qs_dairy
Quantity of soil eaten by the
animal each day, Dairy
FFC
0.07
0.07
0.07
0.07
38
35
26
25
a FFC refers to farm food chain variables and Ing./BW refers to ingestion and body weight variables.
C-3-11
-------
Exhibit 1 -4. Elasticities and Rankings for the Variables with the Highest Elasticities for Cadmium
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
MixHeight
Mixing Height
TRIM
-1.92
-1.03
-0.93
-0.66
1
1
2
3
HorizWindSpeed
Horizontal Wind Speed
TRIM
-1.92
-1.03
-0.93
-0.66
2
2
3
4
EmissionRate
Emission Rate of all Chemicals
TRIM
1.00
1.00
1.00
1.00
3
3
1
1
Rain
Annual Rainfall
TRIM
0.78
0.75
0.74
0.71
4
4
4
2
Cs_root_zone_pr
oduce
Average chemical concentration
in soil at root-zone depth in
produce-growing area
FFC
0.63
0.63
0.63
0.63
5
5
5
5
Br AG produce
pf
Plant-soil chemical
bioconcentration factor, Protected
Fruit
FFC
0.39
0.39
0.39
0.39
7
6
6
8
IR_ProFruit_ch1
Ingestion rate, Protected Fruit,
Child 1-2
Ing./BW
0.39
0.39
0.39
0.39
10
7
7
9
EF ProFruit
Exposure factor, Protected Fruit
Ing./BW
0.39
0.39
N/A
N/A
8
8
FC_Pro Fruit
Fraction contaminated, Protected
Fruit
Ing./BW
0.39
0.39
N/A
N/A
9
9
100-MAF_pf
100 - Moisture adjustment factor,
Protected Fruit
FFC
0.39
0.39
0.39
0.39
11
10
8
7
BW ch1
Body weight, Child 1-2
Ing./BW
-0.54
-0.28
-0.26
-0.18
6
11
11
13
EroRate
Erosion Rate for all Parcels
TRIM
-0.31
-0.27
-0.26
-0.24
12
12
10
10
SurfWatTemp
Water Temperature
TRIM
0.17
0.27
0.30
0.54
16
13
9
6
IR Fish ch1
Ingestion rate, Fish, Child 1-2
Ing./BW
0.19
0.19
0.19
0.19
15
14
12
12
EF Fish
Exposure factor, Fish
Ing./BW
0.19
0.19
N/A
N/A
13
15
FC Fish
Fraction contaminated, Fish
Ing./BW
0.19
0.19
N/A
N/A
14
16
SurfSoilAir
Surface Soil- Fraction Air
TRIM
0.15
0.17
0.17
0.19
19
17
13
11
L1 ProFruit
Loss type 1, Protected Fruit
FFC
-0.16
-0.16
-0.16
-0.16
18
18
14
14
C_FishT3
Concentration of chemical in
whole fish forT3 fish
FFC
0.15
0.15
0.15
0.15
21
19
15
15
WatRetTime
Rentention Time in Pond
TRIM
0.16
0.15
0.14
0.13
17
20
17
16
SedDepVel
Sediment Deposition Velocity and
Resuspension Velocity
TRIM
0.10
0.13
0.14
0.07
30
21
16
32
C-3-12
-------
Exhibit 1 -4. Elasticities and Rankings for the Variables with the Highest Elasticities for Cadmium
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
100-MAF_ev
100 - Moisture adjustment factor,
Exposed Vegetable
FFC
0.11
0.11
0.11
0.11
23
22
18
18
IR_ExpVeg_ch1
Ingestion rate, Exposed
Vegetable, Child 1-2
Ing./BW
0.11
0.11
0.11
0.11
26
23
19
19
EF_ExpVeg
Exposure factor, Exposed
Vegetables
Ing./BW
0.11
0.11
N/A
N/A
24
24
FC_ExpVeg
Fraction contaminated, Exposed
Vegetable
Ing./BW
0.11
0.11
N/A
N/A
25
25
F_T3
Fraction offish intake that is from
T3
Ing./BW
0.11
0.11
0.11
0.11
28
26
20
20
F_T4
Fraction offish intake that is from
T4
Ing./BW
-0.11
-0.11
-0.11
-0.11
29
27
21
21
SurfSoilOCC
Surface Soil- Organic Carbon
Fraction
TRIM
0.15
0.10
0.10
0.08
20
28
22
25
SurfSoilVertVel
Surface Soil- Average vertical
velocity of water (percolation)
TRIM
-0.11
-0.10
-0.10
-0.09
27
29
23
22
SurfSoilpH
Surface Soil- pH
TRIM
0.13
0.09
0.09
0.07
22
30
25
31
a FFC refers to farm food chain variables and Ing./BW refers to ingestion and body weight variables.
C-3-13
-------
Exhibit 1-5. Elasticities and Rankings for the Variables with the Highest Elasticities for Divalent Mercury
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
EmissionRate
Emission Rate of all Chemicals
TRIM
1.00
1.00
1.00
1.00
5
1
1
1
MixHeight
Mixing Height
TRIM
-1.73
-0.97
-0.89
-0.63
1
2
2
3
HorizWindSpeed
Horizontal Wind Speed
TRIM
-1.73
-0.97
-0.89
-0.63
2
3
3
4
EroRate
Erosion Rate for all Parcels
TRIM
-1.27
-0.77
-0.71
-0.52
3
4
5
7
Rain
Annual Rainfall
TRIM
0.76
0.74
0.74
0.71
6
5
4
2
BW ch1
Body weight, Child 1-2
Ing./BW
-1.17
-0.61
-0.56
-0.39
4
6
8
8
C_Soil
Concentration of chemical in soil
from contaminated area
FFC
0.56
0.56
0.56
0.56
7
7
6
5
EF Soil
Exposure factor, Soil
Ing./BW
0.56
0.56
N/A
N/A
8
8
FC Soil
Fraction contaminated, Soil
Ing./BW
0.56
0.56
N/A
N/A
9
9
IR Soil ch1
Ingestion rate, Soil, Child 1-2
Ing./BW
0.56
0.56
0.56
0.56
10
10
7
6
Cs_root_zone_pr
oduce
Average chemical concentration
in soil at root-zone depth in
produce-growing area
FFC
0.33
0.33
0.33
0.33
11
11
9
9
Br AG produce
pf
Plant-soil chemical
bioconcentration factor, Protected
Fruit
FFC
0.15
0.15
0.15
0.15
12
12
10
10
IR_ProFruit_ch1
Ingestion rate, Protected Fruit,
Child 1-2
Ing./BW
0.15
0.15
0.15
0.15
15
13
11
11
100-MAF_pf
100 - Moisture adjustment factor,
Protected Fruit
FFC
0.15
0.15
0.15
0.15
16
14
12
12
EF ProFruit
Exposure factor, Protected Fruit
Ing./BW
0.15
0.15
N/A
N/A
13
15
FC_Pro Fruit
Fraction contaminated, Protected
Fruit
Ing./BW
0.15
0.15
N/A
N/A
14
16
SurfSoilAir
Surface Soil- Fraction Air
TRIM
0.11
0.13
0.13
0.15
17
17
13
13
Br_AG_rootveg
Plant-soil chemical
bioconcentration factor, Root
Vegetables
FFC
0.11
0.11
0.11
0.11
18
18
14
15
IR_RootVeg_ch1
Ingestion rate, Root Vegetable,
Child 1-2
Ing./BW
0.11
0.11
0.11
0.11
21
19
15
16
C-3-14
-------
Exhibit 1-5. Elasticities and Rankings for the Variables with the Highest Elasticities for Divalent Mercury
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
100-MAF_rv
100 - Moisture adjustment factor,
Root Vegetable
FFC
0.11
0.11
0.11
0.11
23
20
16
17
EF_RootVeg
Exposure factor, Root Vegetables
Ing./BW
0.11
0.11
N/A
N/A
19
21
FC_RootVeg
Fraction contaminated, Root
Vegetable
Ing./BW
0.11
0.11
N/A
N/A
20
22
VG_rootveg
Empirical correction factor, Root
Vegetable
FFC
0.11
0.11
N/A
N/A
22
23
Cs_S_animal_ing
est
Chemical concentration in
surface soil in contaminated area
where livestock feed
FFC
0.08
0.08
0.08
0.08
25
24
17
18
Bs
Soil bioavailability factor for
livestock
FFC
0.08
0.08
N/A
N/A
24
25
L1 ProFruit
Loss type 1, Protected Fruit
FFC
-0.06
-0.06
-0.06
-0.06
26
26
19
19
SedDepVel
Sediment Deposition Velocity and
Resuspension Velocity
TRIM
0.02
0.05
0.06
0.11
48
27
18
14
Ba_egg
Chemical-specific biotransfer
factor, Eggs
FFC
0.05
0.05
0.05
0.05
27
28
20
20
IR_egg_ch1
Ingestion rate, Eggs, Child 1-2
Ing./BW
0.05
0.05
0.05
0.05
30
29
21
21
EF_egg
Exposure factor, Eggs
Ing./BW
0.05
0.05
N/A
N/A
28
30
FC_egg
Fraction contaminated, Eggs
Ing./BW
0.05
0.05
N/A
N/A
29
31
Qs_egg
Quantity of soil eaten by the
animal each day, Eggs
FFC
0.05
0.05
0.05
0.05
31
32
22
22
Ba_poultry
Chemical-specific biotransfer
factor, Poultry
FFC
0.03
0.03
0.03
0.03
32
33
23
23
IR_poultry_ch1
Ingestion rate, Poultry, Child 1-2
Ing./BW
0.03
0.03
0.03
0.03
35
34
24
24
EF_poultry
Exposure factor, Poultry
Ing./BW
0.03
0.03
N/A
N/A
33
35
FC_poultry
Fraction contaminated, Poultry
Ing./BW
0.03
0.03
N/A
N/A
34
36
Qs_poultry
Quantity of soil eaten by the
animal each day, Poultry
FFC
0.03
0.03
0.03
0.03
36
37
25
25
a FFC refers to farm food chain variables and Ing./BW refers to ingestion and body weight variables.
C-3-15
-------
Exhibit 1-6. Elasticities and Rankings for the Variables with the Highest Elasticities for Methyl Mercury
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
SedDepVel
Sediment Deposition Velocity and
Resuspension Velocity
TRIM
0.78
1.56
2.01
3.67
9
1
1
1
BW ch1
Body weight, Child 1-2
Ing./BW
-1.91
-1.00
-0.91
-0.64
1
2
3
4
EmissionRate
Emission Rate of all Chemicals
TRIM
1.00
1.00
1.00
1.00
5
3
2
2
IR Fish ch1
Ingestion rate, Fish, Child 1-2
Ing./BW
0.90
0.90
0.90
0.90
8
4
4
3
FC Fish
Fraction contaminated, Fish
Ing./BW
0.90
0.90
N/A
N/A
7
5
EF Fish
Exposure factor, Fish
Ing./BW
0.90
0.90
N/A
N/A
6
6
MixHeight
Mixing Height
TRIM
-1.02
-0.70
-0.66
-0.51
3
7
5
8
HorizWindSpeed
Horizontal Wind Speed
TRIM
-1.01
-0.70
-0.65
-0.51
4
8
6
9
Rain
Annual Rainfall
TRIM
0.70
0.64
0.63
0.60
10
9
7
6
C_FishT4
Concentration of chemical in
whole fish forT4 fish
FFC
0.55
0.55
0.55
0.55
12
10
8
7
EroRate
Erosion Rate for all Parcels
TRIM
-0.57
-0.45
-0.43
-0.35
11
11
10
11
SedPorosity
Sediment Porosity
TRIM
-0.30
-0.40
-0.43
-0.60
15
12
9
5
C_FishT3
Concentration of chemical in
whole fish forT3 fish
FFC
0.36
0.36
0.36
0.36
13
13
11
10
SurfWatTemp
Water Temperature
TRIM
-0.22
-0.22
-0.22
-0.21
18
14
12
12
F_T3
Fraction offish intake that is from
T3
Ing./BW
-0.19
-0.19
-0.19
-0.19
19
15
13
13
F_T4
Fraction offish intake that is from
T4
Ing./BW
0.19
0.19
0.19
0.19
20
16
14
14
WatSuspSed
Water- Suspended Sediment
Concentration
TRIM
-0.32
-0.18
-0.16
-0.12
14
17
15
16
SurfSoilAir
Surface Soil- Fraction Air
TRIM
0.14
0.14
0.15
0.15
21
18
16
15
WatRetTime
Rentention Time in Pond
TRIM
0.23
0.14
0.13
0.09
17
19
17
17
FishMass
Fish Body Weight for all Aquatic
Species
TRIM
0.10
0.07
0.07
0.06
22
20
19
19
Runoff
Total Water Runoff Rate
TRIM
0.07
0.07
0.07
0.07
23
21
18
18
C_Soil
Concentration of chemical in soil
from contaminated area
FFC
0.05
0.05
0.05
0.05
24
22
20
20
C-3-16
-------
Exhibit 1-6. Elasticities and Rankings for the Variables with the Highest Elasticities for Methyl Mercury
Name of Variable
Changed
Variable Definition
Variable
Category a
Elasticity of Risk for Input
Variable Perturbation
Ranking of Elasticity for
Input Variable
Perturbation
-50%
-5%
5%
50%
-50%
-5%
5%
50%
IR Soil ch1
Ingestion rate, Soil, Child 1-2
Ing./BW
0.05
0.05
0.05
0.05
27
23
21
21
EF Soil
Exposure factor, Soil
Ing./BW
0.05
0.05
N/A
N/A
25
24
FC Soil
Fraction contaminated, Soil
Ing./BW
0.05
0.05
N/A
N/A
26
25
Cs_root_zone_pr
oduce
Average chemical concentration
in soil at root-zone depth in
produce-growing area
FFC
0.04
0.04
0.04
0.04
28
26
22
22
100-MAF_pf
100 - Moisture adjustment factor,
Protected Fruit
FFC
0.03
0.03
0.03
0.03
33
27
25
23
Br AG produce
pf
Plant-soil chemical
bioconcentration factor, Protected
Fruit
FFC
0.03
0.03
0.03
0.03
29
28
23
24
IR_ProFruit_ch1
Ingestion rate, Protected Fruit,
Child 1-2
Ing./BW
0.03
0.03
0.03
0.03
32
29
24
25
a FFC refers to farm food chain variables and Ing./BW refers to ingestion and body weight variables.
C-3-17
-------
Appendix D: Detailed assessment inputs and results
for petroleum refining facilities
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
SUNOCO INC (R&M)/MARCUS HOOK
I—
LU
CL
NEI109
REFINERY
100 GREEN ST PO BOX 426
MARCUS HOOK
Delaware County
PA
Deer Park Refining Limited Partnership (Shell Oil
I—
LU
CL
NEI11119
Products US)
5900 HIGHWAY 225
Deer Park
Harris County
TX
Western Refining Co. LP - North (prev. Chevron
I—
LU
CL
NEI11192
USA Inc.)
6501 TROWBRIDGE DR.
EL PASO
El Paso County
TX
Valero Refining Co. - Port Arthur (prev. Premcor
I—
LU
CL
NEI11200
Refining Group Inc.)
10801 S. GULFWAYDR.
PORT ARTHUR
Jefferson County
TX
I—
LU
CL
NEI11232
HOUSTON REFINING LP
12000 LAWNDALE ST
HOUSTON
Harris County
TX
I—
LU
CL
NEI113
ConocoPhillips Co. (prev. Phillips 66 Co.)
4101 POST RD
TRAINER
Delaware County
PA
I—
LU
CL
NEI11449
BP OIL COMPANY TOLEDO REFINNERY
4001 CEDAR POINT ROAD
OREGON
Lucas County
OH
I—
LU
CL
NEI11450
Sunoco - Toledo
1819 Woodville Road
OREGON
Lucas County
OH
I—
LU
CL
NEI11574
Marathon Petroleum Company LLC
2408 Gambrinus Avenue SW
CANTON
Stark County
OH
Valero Refining Co. (prev. Premcor Refining
I—
LU
CL
NEI11663
Group)
1150 S. METCALF ST.
LIMA
Allen County
OH
BP PRODUCTS NORTH AMERICA INC,
I—
LU
CL
NEI11715
WHITING R
2815 INDIANAPOLIS BLVD.
WHITING
Lake County
IN
1300 S. FORT ST. HES
I—
LU
CL
NEI11885
Marathon Petroleum Company LLC
DEPARTMENT
DETROIT
Wayne County
Ml
I—
LU
CL
NEI12044
Marathon Petroleum Company LLC
1320 LOOP 197 S.
TEXAS CITY
Galveston County
TX
I—
LU
CL
NEI12084
Valero Refining Co. - Corpus Christi West
5900 UP RIVER ROAD
CORPUS CHRISTI
Nueces County
TX
802 US HWY212S, S OF
I—
LU
CL
NEI12458
CENEX HARVEST STATES
LAUREL
LAUREL
Yellowstone County
MT
I—
LU
CL
NEI12459
ConocoPhillips Co. (prev. Conoco Inc.)
401 S 23RD ST
BILLINGS
Yellowstone County
MT
I—
LU
CL
NEI12460
EXXONMOBIL BILLINGS REFINERY
700 EXXONMOBIL RD
BILLINGS
Yellowstone County
MT
I—
LU
CL
NEI12464
MONTANA REFINING
1900 10TH STREET N.E.
GREAT FALLS
Cascade County
MT
Pasadena Refining Systems Inc. (prev. Crown
I—
LU
CL
NEI12480
Central Petroleum Corp.)
111 RED BLUFF ROAD
PASADENA
Harris County
TX
I—
LU
CL
NEI12486
VALERO THREE RIVERS REFINERY
301 LEROY STREET
THREE RIVERS
Live Oak County
TX
I—
LU
CL
NE112711
Valero Refining Co. - Houston
9701 MANCHESTER
HOUSTON
Harris County
TX
Western Refining Co. LP - South (prev. Chevron
I—
LU
CL
NEI12790
USA Inc.)
6500 TROWBRIDGE ST.
EL PASO
El Paso County
TX
Western Refining Co. LP - South (prev. Chevron
I—
LU
CL
NEI12790
USA Inc.)
6501 TROWBRIDGE DR.
EL PASO
El Paso County
TX
I—
LU
CL
NEI12791
Valero Refining Co. - Texas City
1301 LOOP 197 S.
TEXAS CITY
Galveston County
TX
I—
LU
CL
NEI12968
SINCLAIR OIL CORP
902 W25TH ST
TULSA
Tulsa County
OK
I—
LU
CL
NEI12988
ConocoPhillips Co. (prev. Conoco Inc.)
1000 S PINE
PONCA CITY
Kay County
OK
I—
LU
CL
NEI13322
CHEVRON HAWAII REFINERY
91-480 MALAKOLE ST.
KAPOLEI
Honolulu County
HI
TESORO ALASKA COMPANY - KENAI
I—
LU
CL
NEI13371
REFINERY
54741 TESORO ROAD
KENAI
Kenai Peninsula Bori
AK
1 of 6
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
PET NEI18372
Shell Chemical LP
LOCATION ADDRESS IS
NEEDED
MOBILE
Mobile County
AL
PET NEI18394
HUNT REFINING COMPANY
1855 FAIRLAWN RD
TUSCALOOSA
Tuscaloosa County
AL
PET NEI18406
Flint Hills Resources (prev. Williams Alaska Petro
Inc.)
1100 H&H Lane
North Pole
Fairbanks North Star
AK
PET NEI18408
Petro Star Inc. - North Pole
1200 H & H LN.
NORTH POLE
Fairbanks North Star
AK
PET NEI18415
PETRO STAR VALDEZ REFY.
2.5 MILE DAYVILLE RD.
VALDEZ
Valdez-Cordova Cen
AK
PET NEI19587
Chevron USA Inc. - Richmond
841 CHEVRON WAY
RICHMOND
Contra Costa County
CA
PET NEI19834
Shell Oil Products US - Martinez
3485 PACHECO BLVD
MARTINEZ
Contra Costa County
CA
PET NEI19869
ConocoPhillips Co. - Santa Maria (prev. Phillips
66 Co.)
2555 WILLOW ROAD
ARROYO GRANDE
San Luis Obispo Coi
CA
PET NEI19870
ConocoPhillips Co. - Rodeo (prev. Phillips 66
Co.)
1380 SAN PABLO AVE
RODEO
Contra Costa County
CA
PET NEI20103
KERN OIL & REFINING COMPANY
PANAMA LN & WEEDPATCH
HWY
BAKERSFIELD
Kern County
CA
PET NEI20154
SAN JOAQUIN REFINING COMPANY
STANDARD AND SHELL ST
BAKERSFIELD
Kern County
CA
PET NEI20174
Big West Oil LLC (prev. Shell Oil Products US)
6451 ROSEDALE HWY.
BAKERSFIELD
Kern County
CA
PET NEI20467
Chevron USA Inc. - El Segundo
324 W EL SEGUNDO BLVD
EL SEGUNDO
Los Angeles County
CA
PET NEI20616
EDGINGTON OIL COMPANY
2400 E ARTESIA BLVD
LONG BEACH
Los Angeles County
CA
PET NEI20797
VALERO WILMINGTON ASPHALT PLANT
1651 ALAMEDA ST
WILMINGTON
Los Angeles County
CA
PET NEI20966
LUNDAY-THAGARD OIL CO
9301 GARFIELD AVENUE
SOUTH GATE
Los Angeles County
CA
PET NEI21034
ExxonMobil - Torrance
3700 W. 190TH ST.
TORRANCE
Los Angeles County
CA
PET NEI21130
PARAMOUNT PETROLEUM CORP
14708 DOWNEY AV
PARAMOUNT
Los Angeles County
CA
PET NEI21466
Valero Refining Co. - Wilmington (prev. Ultramar
Inc.)
2402 E. ANAHEIM ST.
WILMINGTON
Los Angeles County
CA
PET NEI25450
Valero Refining Co. - Benicia
3400 E 2ND STREET
BENICIA
Solano County
CA
PET NEI25464
Valero Refining - Benicia Asphalt
3001 PARK ROAD
BENICIA
Solano County
CA
PET NEI26101
TENBY INC.
3455 EAST FIFTH STREET
OXNARD
Ventura County
CA
PET NEI26218
Valero Refining Co. (prev. Motiva Enterprises
LLC)
2000 WRANGLE HILL RD
DELAWARE CITY
New Castle County
DE
PET NEI26473
CITGO ASPHALT REFINING COMPANY
FOUNDATION DRIVE
SAVANNAH
Chatham County
GA
PET NEI26489
YOUNG REFINING CORP.
7982 HUEY ROAD
DOUGLASVILLE
Douglas County
GA
PET NEI26533
Tesoro Hawaii Corp.
91-325 KOMOHANA STREET
KAPOLEI
Honolulu County
HI
PET NEI2CA131
BP West Coast Products
1801 E SEPULVEDA BLVD
CARSON
Los Angeles County
CA
PET NEI2CA254
Greka Energy
1660 SINTON RD
SANTA MARIA
Santa Barbara Coun
CA
PET NEI2CA314
Tesoro (prev. Ultramar Inc.)
Avon Refinery
MARTINEZ
Contra Costa County
CA
PET NEI2KS125
COFFEYVILLE RESOURCES REFINING &
MARKETING
400 NORTH LINDEN
COFFEYVILLE
Montgomery County
KS
PET NEI2TX141
Western Refining Co. LP - Marketing Terminal
6501 TROWBRIDGE DR.
EL PASO
El Paso County
TX
PET NEI32353
Countrymark Cooperative Inc.
1200 REFINERY RD
MOUNT VERNON
Posey County
IN
2 of 6
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
PET NEI32762
FRONTIER EL DORADO REFINING COMPANY
1401 S. DOUGLAS ROAD
EL DORADO
Butler County
KS
PET NEI32801
NATIONAL COOPERATIVE REFINERY ASSN
1391 IRON HORSE ROAD
MC PHERSON
McPherson County
KS
PET NEI32864
Marathon Petroleum Company LLC
11631 US ROUTE 23
CATLETTSBURG
Boyd County
KY
PET NEI32997
SOMERSET REFINERY INC
600 MONTICELLO RD
SOMERSET
Pulaski County
KY
PET NEI33007
Calumet Lubricants Co. - Princeton
Calumet Lubricants Co.
10234 HWY157
PRINCETON
Bossier Parish
LA
PET NEI33008
Calumet Shreveport LLC (prev. Calumet
Lubricants Co.)
3333 MIDWAY
SHREVEPORT
Caddo Parish
LA
PET NEI33010
CALCASIEU REFINING CO
4359 W TANK FARM RD
LAKE CHARLES
Calcasieu Parish
LA
PET NEI33030
SHELL CHEMICAL LP/NORCO CHEM PLT
EAST SITE
HWY61 W
NORCO
St. Charles Parish
LA
PET NEI33031
MOTIVA ENTERPRISES LLC/NORCO
REFINERY
15536 River Road
NORCO
St. Charles Parish
LA
PET NEI33039
Calumet Lubricants Co. - Cotton Valley
1756 OLD HWY 7
COTTON VALLEY
Webster Parish
LA
PET NEI34022
FLINT HILLS RESOURCES LP - PINE BEND
JUNCTIONS 52 & 55
INVER GROVE HEIC
Dakota County
MN
PET NEI34050
MARATHON PETROLEUM CO LLC SAINT
PAUL PARK REFINER
300 3RD STREET
SAINT PAUL PARK
Washington County
MN
PET NEI34057
CHEVRON TEXACO PRODUCTS COMPANY,
PASCAGO
250 INDUSTRIAL ROAD
PASCAGOULA
Jackson County
MS
PET NEI34061
HUNT SOUTHLAND REFINING COMPANY
HIGHWAY 11 NORTH
SANDERSVILLE
Jones County
MS
PET NEI34062
Hunt Southland Refining (prev. Southland Oil
Co.)
HIGHWAY 11 NORTH
LUMBERTON
Lamar County
MS
PET NEI34069
ERGON REFINING INC
2611 HAINING ROAD
VICKSBURG
Warren County
MS
PET NEI34862
Sunoco, Inc. (prev. Coastal Eagle Point Oil Co.)
US RT 130 AND 295
WESTVILLE
Gloucester County
NJ
PET NEI34863
CITGO ASPHALT REFINING COMPANY
4 PARADISE RD
PAULSBORO
Gloucester County
NJ
PET NEI34872
Hess Corporation (prev. Amerada-Hess Corp.)
750 CLIFF ROAD
PT. READING
Middlesex County
NJ
PET NEI34873
CHEVRON PRODUCTS COMPANY
1200 STATE ST
PERTH AMBOY
Middlesex County
NJ
PET NEI34898
Navajo Refining Co. - Artesia
501 E Main St
Artesia
Eddy County
NM
PET NEI34907
Giant Refining Co. - Ciniza Refinery
I-40 EXIT 39
JAMESTOWN
McKinley County
NM
PET NEI34912
Giant Industries Inc. - Bloomfield
#50 County Road 4990
Bloomfield
San Juan County
NM
PET NEI363
FRONTIER REFINING INC
2700 EAST 5TH STREET
CHEYENNE
Laramie County
WY
PET NEI371
Little America Refining Co. (Sinclair)
5700 E. HWY. 20/26
CASPER
Natrona County
WY
PET NEI40371
Tesoro - Mandan
900 OLD RED TRAIL N.E.
MANDAN
Morton County
ND
PET NEI404
WYOMING REFINING CO_NEWCASTLE
REFINERY
740 W MAIN STREET
NEWCASTLE
Weston County
WY
PET NEI40531
Wynnewood Refining Co. (prev. Gary-Williams
Energy Corp.)
906 S POWELL
WYNNEWOOD
Garvin County
OK
PET NEI40625
Paramount Petroleum Corp. (prev. Chevron USA)
5501 NW FRONT AVE
PORTLAND
Multnomah County
OR
3 of 6
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
I—
LU
CL
NEI40723
Sunoco Inc. - Philadelphia
3144 PASSYUNKAVE.
Philadelphia
Philadelphia County
PA
I—
LU
CL
NEI40732
UNITED REFINING CO/WARREN PLT
15 BRADLEY ST
WARREN
Warren County
PA
Suncor Energy USA - Denver (prev. Colorado
I—
LU
CL
NEI415
Refining Co.)
5800 BRIGHTON BLVD
COMMERCE CITY
Adams County
CO
Valero Refining Co. (prev. Premcor Refining,
I—
LU
CL
NEI41591
prev. Williams Refining LLC)
543 West Mallory Avenue
Memphis
Shelby County
TN
Total Petrochemicals Inc. (prev. Atofina
I—
LU
CL
NEI41771
Petrochemicals, Inc.)
32ND ST. & HWY. 366
PORT ARTHUR
Jefferson County
TX
I—
LU
CL
NEI41863
Valero Refining Co. - Corpus Christi East
1300 CANTWELL LN.
CORPUS CHRISTI
Nueces County
TX
I—
LU
CL
NEI41864
Flint Hills Resources LP - Corpus Christi West
2825 SUNTIDE RD.
CORPUS CHRISTI
Nueces County
TX
I—
LU
CL
NEI41865
Trigeant LTD
6600 UP RIVER ROAD
CORPUS CHRISTI
Nueces County
TX
I—
LU
CL
NEI42016
Big West Oil Co. (Flying J)
333 W CENTER ST
NORTH SALT LAKE
Davis County
UT
I—
LU
CL
NEI42020
Holly Corp. (prev. Phillips 66 Co.)
393 South 800 West
Woods Cross
Davis County
UT
I—
LU
CL
NEI42025
Silver Eagle Refining
2355 S. 1100 W.
WOODS CROSS
Davis County
UT
I—
LU
CL
NEI42040
Tesoro - Salt Lake City
474 W. 900 N.
SALT LAKE CITY
Salt Lake County
UT
I—
LU
CL
NEI42081
Chevron - Salt Lake City
2351 N1100W
SALT LAKE CITY
Salt Lake County
UT
I—
LU
CL
NEI42309
GIANT YORKTOWN REFINERY
2201 GOODWIN NECK RD
GRAFTON
York County
VA
I—
LU
CL
NEI42370
US OIL & REFINING CO
3001 MARSHALL AVE
TACOMA
Pierce County
WA
I—
LU
CL
NEI42381
TESORO NORTHWEST COMPANY
1020 W MARCH POINT RD
ANACORTES
Skagit County
WA
I—
LU
CL
CN
CO
CO
CN
UJ
Z
Shell - Anacortes
8505 SOUTH TEXAS ROAD
ANACORTES
Skagit County
WA
I—
LU
CL
NEI42413
BP West Coast Products - Cherry Point
4519 GRANDVIEWRD
BLAINE
Whatcom County
WA
I—
LU
CL
NEI42425
ConocoPhillips Co. (prev. Phillips 66 Co.)
3901 UNICKRD
FERNDALE
Whatcom County
WA
I—
LU
CL
NEI42583
MURPHY OIL USA
24TH AVE E AND 26TH ST
SUPERIOR
Douglas County
Wl
I—
LU
CL
NEI43243
SINCLAIR OIL CORP-SINCLAIR REFINERY
BOX 277
SINCLAIR
Carbon County
WY
I—
LU
CL
NEI46556
HOVENSA L.L.C.
1 ESTATE HOPE
CHRISTIANSTED
St. Croix
VI
I—
LU
CL
NEI46752
ERGON - WEST VIRGINIA, INC.
ROUTE 2 SOUTH
NEWELL
Hancock County
WV
I—
LU
CL
NEI46764
American Refining Group Inc.
77 N KENDALL AVE
BRADFORD
McKean County
PA
I—
LU
CL
NEI49781
Marathon Petroleum Company LLC
100 Marathon Ave
Robinson
Crawford County
IL
I—
LU
CL
NEI53702
PDV Midwest Refining LLC (Citgo Petroleum)
135TH ST AND NEW AVE
LEMONT
Will County
IL
INTERSTATE 55 & ARSENAL
I—
LU
CL
NEI53718
EXXONMOBIL OIL CORP
RD
JOLIET
Will County
IL
I—
LU
CL
NEI55835
ConocoPhillips Co. (prev. Phillips 66 Co.)
900 S Central Ave
Roxana
Madison County
IL
I—
LU
CL
NEI6018
SHELL CHEMICAL LP/ST. ROSE REFINERY
11842 RIVER RD
ST. ROSE
St. Charles Parish
LA
I—
LU
CL
NEI6022
ExxonMobil Corp. - Baton Rouge
4045 SCENIC HWY
BATON ROUGE
East Baton Rouge P;
LA
ConocoPhillips Co. - Westlake (prev. Conoco
I—
LU
CL
NEI6062
Inc.)
2200 OLD SPANISH TRAIL
WESTLAKE
Calcasieu Parish
LA
I—
LU
CL
NEI6084
MOTIVA ENTERPRISES,LLC/CONVENT
HWY 70 & HWY 44
CONVENT
St. James Parish
LA
I—
LU
CL
NEI6087
Marathon Petroleum Company LLC
E. BANK OF MS RIVER
GARYVILLE
St. John the Baptist f
LA
Valero Refining - Norco (prev. Orion Refining
I—
LU
CL
NEI6095
Corp)
14902 RIVER RD.
NEW SARPY
St. Charles Parish
LA
4 of 6
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
PET NEI6116
ConocoPhillips Co. - Belle Chasse (prev. Phillips
66 Co.)
15551 HWY23 S
BELLE CHASSE
Plaquemines Parish
LA
PET NEI6123
Chalmette Refining LLC (ExxonMobil)
500 W. ST. BERNARD HWY
CHALMETTE
St. Bernard Parish
LA
PET NEI6127
MURPHY OIL USA, INC./MERAUX REFINERY
2500 E ST. BERNARD HWY
MERAUX
St. Bernard Parish
LA
PET NEI6130
PLACID REFINING CO LLC/PT ALLEN
1940 LA HWY 1, NORTH
PORT ALLEN
West Baton Rouge P
LA
PET NEI6136
Valero Refining Co. - Krotz Springs
HIGHWAY 105 SOUTH
KROTZ SPRINGS
St. Landry Parish
LA
PET NEI6166
Citgo Petroleum Corp. - Lake Charles
2 Ml S
SULPHUR
Calcasieu Parish
LA
PET NEI6375
ConocoPhillips Co. (prev. Phillips 66 Co.)
1400 Park Ave
Linden
Union County
NJ
PET NEI6436
BP - Texas City
2401 5TH AVE. S.
TEXAS CITY
Galveston County
TX
PET NEI6446
Alon USA Energy Inc. (prev. Alon USA LP)
I. 20 AT REFINERY ROAD
BIG SPRING
Howard County
TX
PET NEI6475
Delek Refining Ltd (prev. LaGloria Oil & Gas Co.)
1702 E COMMERCE ST
TYLER
Smith County
TX
PET NEI6519
ConocoPhillips Co. - Sweeny (prev. Phillips 66
Co.)
HWY 35 AND 524 AT OLD
OCEAN
SWEENY
Brazoria County
TX
PET NEI6617
Citgo Refining & Chemical Inc. - Corpus Christi
West
7350 I. 37
CORPUS CHRISTI
Nueces County
TX
PET NEI6963
ConocoPhillips Co. - Borger (prev. Phillips 66
Co.)
STATE HWY. SPUR 119 N.
BORGER
Hutchinson County
TX
PET NEI7130
AGE Refining & Manufacturing
7811 S. PRESA ST.
SAN ANTONIO
Bexar County
TX
PET NEI7134
Flint Hills Resources LP - Corpus Christi East
1700 NUECES BAY
BOULEVARD
CORPUS CHRISTI
Nueces County
TX
PET NEI7233
ExxonMobil Corp. - Beaumont
1795 Burt Street
BEAUMONT
Jefferson County
TX
PET NEI7441
Motiva - Port Arthur
2100 HOUSTON AVE.
PORT ARTHUR
Jefferson County
TX
PET NEI7781
ExxonMobil Corp. - Baytown
2800 DECKER DR
BAYTOWN
Harris County
TX
PET NEI7973
South Hampton Resources Inc. (prev. South
Hampton Refining Co.)
HWY. 418
SILSBEE
Hardin County
TX
PET NEI7988
CITGO CORPUS CHRISTI REFINERY EAST
PLANT
1801 NUECES BAY BLVD.
CORPUS CHRISTI
Nueces County
TX
PET NEI8139
Valero Energy Corp. - McKee (prev. Diamond
Shamrock Refining)
6701 FM 119
SUNRAY
Moore County
TX
PET NEI8612
Gulf Atlantic Operations LLC (prev. Coastal
Mobile Refining Co.)
200 VIADUCT RD.
CHICKASAW
Mobile County
AL
PET NEI876
LION OIL COMPANY
1000 MCHENRY DRIVE
EL DORADO
Union County
AR
PET NEI889
Suncor Energy USA - Commerce City (prev.
Conoco Inc.)
5801 BRIGHTON BLVD
COMMERCE CITY
Adams County
CO
PET NEICA0370
ConocoPhillips Co. - Wilmington (prev. Phillips 66
Co.)
1660 W ANAHEIM ST
WILMINGTON
Los Angeles County
CA
PET NEICA0379
ConocoPhillips Co. - Carson (prev. Phillips 66
Co.)
1520 E SEPULVEDA BLVD
CARSON
Los Angeles County
CA
5 of 6
-------
Table 1 - Facility Identification Information
Facility NEI ID
Facility Name
Address
City
County
State
PET NEICA1057
Tricor Refining (prev. Golden Bear Oil
Specialties)
1134 MANOR ST
OILDALE
Kern County
CA
PET NEICA191C
Shell Oil Products US - Wilmington
2101 E PACIFIC COAST HWY
WILMINGTON
Los Angeles County
CA
PET_NEINJT$89
Valero Refining CO - NJ
VALERO REFINING CO - NJ
PAULSBORO REFINERY 800
BIL
PAULSBORO
Gloucester County
NJ
PET_NEINMT$1J
Navajo Refining Co. - Lovington
5 Ml SE OF LOVINGTION ON
NM 18
LOVINGTON
Lea County
NM
PET_NEIOKT$1'
Valero Refining Company - Oklahoma, Valero
Ardmore Refinery
HWY. 142 & E. CAMERON RD.
ARDMORE
Carter County
OK
PET_NEIPRT$6^
SHELL CHEMICAL YABUCOA INC.
RTE. 901 KM 2.7 CAMINO
NUEVO WARD
YABUCOA
Yabucoa Municipio
PR
PET_NEIWYT$1:
SILVER EAGLE REFINING-EVANSTON
2990 COUNTY RD. 180
EVANSTON
Uinta County
WY
6 of 6
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
PET NEI109
4.1E-06
8.1E-04
1.7E-02
PET NEI11119
6.0E-07
4.1E-04
1.7E-03
PET NEI11192
6.4E-07
9.2E-05
4.2E-03
PET NEI11200
8.8E-07
2.6E-04
4.1E-02
PET NEI11232
4.7E-06
3.0E-03
2.1E-02
PET NEI113
5.3E-07
9.3E-05
2.3E-03
PET NEI11449
9.1E-06
8.0E-04
1.9E-03
PET NEI11450
1.7E-06
1.4E-04
6.5E-03
PET NEI11574
5.2E-06
1.6E-04
2.0E-02
PET NEI11663
4.4E-06
4.3E-04
1.4E-02
PET NEI11715
0.0E+00
0.0E+00
0.0E+00
PET NEI11885
1.8E-07
4.3E-05
2.9E-03
PET NEI12044
6.0E-06
3.9E-04
2.3E-02
PET NEI12084
4.3E-07
7.0E-05
9.9E-04
PET NEI12458
9.8E-07
4.1E-05
2.1E-03
PET NEI12459
1.0E-06
1.3E-05
4.3E-03
PET NEI12460
8.6E-07
4.0E-05
2.9E-03
PET NEI12464
1.0E-06
2.3E-05
5.1E-03
PET NEI12480
6.0E-06
6.0E-04
2.5E-02
PET NEI12486
4.3E-06
1.3E-05
1.0E-02
PET NEI12711
2.84E-05
5.8E-03
1.0E-01
PET NEI12790
3.8E-06
1.7E-04
1.6E-02
PET NEI12791
9.8E-06
1.5E-03
1.7E-02
PET NEI12968
1.9E-06
9.0E-05
1.5E-01
PET NEI12969
1.2E-06
9.3E-05
5.0E-03
PET NEI12988
1.90E-05
5.7E-04
3.7E-02
PET NE113322
1.3E-07
7.7E-06
1.3E-03
PET NE113371
5.2E-06
1.7E-05
1.8E-02
PET NEI18372
3.9E-07
1.0E-04
1.5E-03
PET NEI18394
1.1E-05
1.4E-04
6.6E-02
PET NE118406
8.6E-06
6.5E-05
2.6E-02
PET NEI18408
1.4E-08
5.4E-08
6.1E-05
PET NEI18415
2.4E-08
2.0E-07
1.0E-04
PET NEI19587
4.8E-06
2.0E-03
3.6E-02
PET NEI19834
1.9E-06
1.7E-04
2.6E-02
PET NEI19869
1.4E-07
1.5E-06
6.0E-04
PET NEI19870
5.6E-07
1.0E-04
3.6E-03
PET NEI20103
5.7E-07
1.4E-05
3.6E-03
PET NEI20154
6.5E-09
2.7E-07
1.2E-04
PET NEI20174
1.2E-05
8.3E-04
4.9E-02
PET NEI20467
1.4E-06
2.6E-04
1.1E-02
PET NEI20616
3.4E-07
6.9E-05
1.8E-03
PET NEI20797
3.2E-08
1.2E-05
1.4E-04
PET NEI20966
7.2E-08
4.4E-05
2.6E-04
PET NEI21034
7.6E-07
8.0E-04
2.6E-03
PET NEI21130
2.4E-09
1.2E-06
5.2E-04
PET NEI21466
1.5E-09
9.6E-07
3.2E-04
PET NEI25450
9.9E-07
1.6E-04
2.7E-03
PET NEI25464
4.6E-09
5.1E-07
2.0E-05
PET NEI26101
4.1E-07
3.4E-05
7.2E-04
PET NEI26218
2.9E-07
1.0E-04
1.7E-03
PET NEI26473
9.3E-14
2.7E-11
3.6E-05
PET NEI26489
4.3E-07
1.7E-05
1.8E-03
1 of 3
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
PET NEI26533
7.2E-08
9.5E-06
3.1E-04
PET NEI2CA131003
3.7E-07
3.4E-04
1.2E-03
PET NEI2CA254640
1.2E-07
1.4E-06
4.2E-04
PET NEI2CA314628
2.5E-06
2.9E-04
4.0E-02
PET NEI2KS125003
2.7E-06
4.9E-05
2.9E-02
PET NEI2TX14199
4.3E-08
2.5E-06
1.9E-04
PET NEI32353
2.3E-07
6.0E-06
1.1E-02
PET NEI32762
4.6E-06
1.7E-04
9.2E-02
PET NEI32801
2.1E-06
9.4E-05
1.5E-02
PET NEI32864
1.1E-05
6.6E-04
3.4E-02
PET NEI32997
3.8E-07
2.3E-06
4.4E-03
PET NEI33007
1.3E-08
1.6E-06
2.0E-05
PET NEI33008
9.4E-06
3.5E-04
4.0E-02
PET NEI33010
1.6E-07
1.3E-05
5.5E-04
PET NEI33030
4.8E-06
1.5E-04
2.0E-02
PET NEI33031
1.52E-05
8.9E-04
1.2E-01
PET NEI33039
1.0E-05
2.2E-05
7.9E-03
PET NEI34022
5.4E-07
2.5E-04
1.5E-02
PET NEI34050
1.2E-05
1.1E-03
3.4E-02
PET NEI34057
1.4E-05
1.0E-03
2.2E-01
PET NEI34061
2.9E-07
1.2E-06
4.3E-03
PET NEI34062
4.5E-06
5.4E-06
2.1E-02
PET NEI34069
2.0E-08
1.2E-06
6.3E-05
PET NEI34862
2.1E-06
6.2E-04
8.8E-03
PET NEI34863
3.3E-08
2.1E-05
1.8E-04
PET NEI34872
6.9E-07
8.5E-05
2.6E-03
PET NEI34873
4.6E-06
9.0E-04
2.0E-02
PET NEI34898
2.06E-05
3.4E-04
1.8E-01
PET NEI34907
4.5E-07
9.2E-06
1.7E-03
PET NEI34912
1.8E-07
2.7E-06
7.9E-04
PET NEI363
9.3E-07
3.6E-05
4.4E-03
PET NEI371
4.9E-08
5.6E-07
2.7E-05
PET NEI40371
1.0E-05
1.1E-04
2.7E-02
PET NEI404
4.2E-08
3.7E-07
3.4E-03
PET NEI40531
8.3E-06
4.2E-05
5.1E-02
PET NEI40625
4.3E-07
8.3E-05
1.8E-03
PET NEI40723
2.4E-06
8.1E-04
2.0E-02
PET NEI40732
5.0E-06
1.7E-05
2.7E-02
PET NEI415
1.6E-06
2.4E-04
6.6E-03
PET NEI41591
2.0E-06
2.6E-04
1.1E-02
PET NEI41771
1.3E-05
4.6E-04
4.6E-02
PET NEI41863
1.5E-06
1.6E-04
1.7E-02
PET NEI41864
3.8E-07
6.0E-05
1.1E-03
PET NEI41865
1.9E-08
3.9E-06
8.0E-05
PET NEI42016
2.2E-06
8.1E-05
2.3E-02
PET NEI42020
4.8E-06
1.7E-04
1.9E-02
PET NEI42025
3.0E-06
8.3E-05
1.1E-02
PET NEI42040
1.47E-05
1.3E-04
2.8E-01
PET NEI42081
1.4E-07
4.8E-05
4.3E-04
PET NEI42309
1.501E-05
6.6E-05
6.8E-02
PET NEI42370
5.1E-08
3.3E-05
3.2E-04
PET NEI42381
5.1E-06
4.9E-05
4.1E-02
PET NEI42382
1.9E-07
6.1E-06
7.2E-04
2 of 3
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
PET NEI42413
1.6E-06
1.3E-05
5.2E-03
PET NEI42425
9.8E-07
2.2E-05
4.1E-03
PET NEI42583
9.6E-08
3.3E-06
4.1E-04
PET NEI43243
1.6E-06
6.2E-06
6.8E-03
PET NEI46556
6.0E-06
4.0E-04
3.7E-02
PET NEI46752
1.8E-06
3.2E-05
7.7E-03
PET NEI46764
1.5E-07
2.5E-06
1.0E-03
PET NEI49781
7.4E-07
2.3E-05
3.1E-03
PET NEI53702
2.1E-06
4.8E-04
8.8E-03
PET NEI53718
1.8E-07
6.1E-05
7.4E-04
PET NEI55835
1.7E-08
1.9E-06
7.3E-05
PET NEI6018
1.9E-07
1.2E-05
1.0E-03
PET NEI6022
9.7E-06
1.6E-03
7.1E-02
PET NEI6062
4.6E-06
2.2E-04
2.6E-02
PET NEI6084
1.2E-06
2.3E-05
6.8E-03
PET NEI6087
1.2E-05
5.1E-05
1.9E-01
PET NEI6095
1.3E-05
4.5E-04
2.5E-02
PET NEI6116
7.5E-06
1.2E-04
2.8E-02
PET NEI6123
5.0E-06
1.5E-03
2.0E-02
PET NEI6127
2.5E-06
8.2E-05
1.0E-02
PET NEI6130
5.8E-06
2.5E-04
2.0E-02
PET NEI6136
5.8E-06
1.2E-04
8.6E-03
PET NEI6166
7.3E-06
9.7E-04
2.3E-02
PET NEI6375
1.4E-06
1.7E-03
4.8E-03
PET NEI6436
1.2E-05
1.7E-03
4.6E-02
PET NEI6446
2.4E-06
1.8E-05
9.1E-03
PET NEI6475
1.3E-05
2.2E-04
5.6E-02
PET NEI6519
5.3E-06
4.0E-05
2.3E-02
PET NEI6617
6.8E-08
1.1E-05
5.1E-03
PET NEI6963
2.9E-07
1.9E-05
2.6E-03
PET NEI7130
1.3E-07
1.4E-05
8.7E-04
PET NEI7134
2.8E-06
1.4E-04
1.4E-02
PET NEI7233
5.3E-06
7.9E-04
1.1E-01
PET NEI7441
8.1E-07
8.8E-05
4.3E-03
PET NEI7781
6.0E-06
1.0E-03
5.2E-02
PET NEI7973
4.7E-09
2.0E-07
2.0E-05
PET NEI7988
6.3E-06
4.7E-04
6.1E-02
PET NEI8139
1.5E-06
1.6E-06
4.7E-03
PET NEI8612
6.5E-07
1.6E-04
5.2E-03
PET NEI876
1.45E-05
3.4E-04
3.5E-02
PET NEI889
9.9E-07
9.9E-05
3.8E-03
PET NEICA0370363
2.0E-06
4.1E-04
1.4E-02
PET NEICA0379991
2.6E-07
2.1E-04
2.7E-03
PET NEICA10578
3.1E-07
9.8E-06
1.8E-02
PET NEICA1910268
1.1E-05
1.9E-03
1.4E-02
PET_NEINJT$891
1.2E-06
4.5E-04
6.2E-03
PET_NEINMT$12478
2.4E-07
1.4E-05
2.1E-03
PET_NEIOKT$11009
9.2E-06
6.0E-05
3.8E-02
PET_NEIPRT$64
1.2E-05
1.3E-03
1.1E-01
PET_NEIWYT$12156
1.5E-09
1.3E-07
6.4E-06
1 BOLD/Shaded RED indicates a cancer risk great than 1 in a million or a noncancer HI greater than 1
3 of 3
-------
Table 3 - Maximum Predicted HEM-AERMOD Acute Risks
Facility NEI ID
Maximum Hazard Quotient1
AEGL1
AEGL2
ERPG1
ERPG2
REL
PET NEI109
3E-03
6E-04
3E-03
6E-04
4E-01
PET NEI11119
1E-02
9E-04
1E-03
4E-04
1E-01
PET NEI11192
3E-04
4E-05
2E-04
1E-04
3E-02
PET NEI11200
3E-02
5E-03
7E-03
2E-03
9E-01
PET NEI11232
5E-03
6E-04
5E-03
4E-04
7E-01
PET NEI113
8E-03
7E-04
5E-03
1E-03
7E-01
PET NEI11449
9E-03
4E-03
9E-03
2E-03
2E-01
PET NEI11450
3E-04
4E-05
2E-04
7E-05
3E-02
PET NEI11574
3E-02
1E-03
3E-02
1E-03
2E-01
PET NEI11663
9E-03
6E-04
9E-03
1E-03
1E+00
PET NEI11715
0E+00
0E+00
0E+00
0E+00
0E+00
PET NEI11885
2E-04
2E-05
1E-05
1E-06
0E+00
PET NE112044
5E-02
3E-03
5E-02
3E-03
7E+00
PET NEI12084
1E-01
6E-03
1E-01
6E-03
6E-01
PET NEI12458
3E-01
1E-02
3E-01
1E-02
1E+00
PET NEI12459
1E-03
1E-04
1E-03
2E-04
2E-01
PET NE112460
2E-01
1E-02
2E-01
1E-02
8E-01
PET NE112464
2E-03
3E-04
1E-03
5E-04
2E-01
PET NEI12480
7E-03
5E-04
7E-03
6E-04
9E-01
PET NE112486
4E+00
2E-01
4E+00
2E-01
1E+01
PET NE112711
3E-03
2E-04
1E-03
5E-04
2E-01
PET NEI12790
3E-03
2E-04
3E-03
2E-04
4E-01
PET NEI12791
4E-01
3E-02
4E-01
3E-02
5E-01
PET NEI12968
6E-03
4E-03
6E-03
4E-03
8E-02
PET NEI12969
8E-03
5E-04
8E-03
5E-04
3E-01
PET NEI12988
2E-01
9E-03
2E-01
9E-03
2E+00
PET NE113322
8E-04
5E-05
8E-04
5E-05
1E-01
PET NE113371
5E-03
4E-04
5E-03
8E-04
7E-01
PET NEI18372
5E-03
3E-04
5E-03
6E-04
7E-01
PET NEI18394
5E-03
6E-04
5E-03
1E-03
7E-01
PET NE118406
9E-04
6E-05
9E-04
9E-05
1E-01
PET NEI18408
3E-06
2E-07
3E-06
2E-07
4E-04
PET NEI18415
5E-05
3E-06
5E-05
3E-06
6E-03
PET NEI19587
1E-02
2E-03
1E-02
4E-03
1E+00
PET NEI19834
6E-04
9E-05
3E-04
6E-05
4E-02
PET NEI19869
1E-04
3E-05
1E-04
3E-05
1E-02
PET NEI19870
3E-03
2E-03
3E-03
2E-03
8E-02
PET NEI20103
7E-04
1E-04
7E-04
3E-04
7E-02
PET NEI20154
1E-05
2E-06
8E-06
3E-06
5E-04
PET NEI20174
9E-03
6E-04
9E-03
6E-04
1E+00
PET NEI20467
4E-04
1E-04
6E-04
1E-04
2E-02
PET NEI20616
1E-04
2E-05
1E-04
4E-05
1E-02
PET NEI20797
4E-05
2E-06
4E-05
2E-06
5E-03
PET NEI20966
4E-04
7E-05
4E-04
2E-04
8E-03
PET NEI21034
4E-03
1E-04
4E-03
1E-04
7E-02
PET NEI21130
3E-03
2E-04
3E-03
2E-04
3E-02
PET NEI21466
1E-02
5E-04
1E-02
5E-04
4E-02
PET NEI25450
2E-02
2E-03
7E-03
3E-03
7E-01
PET NEI25464
2E-05
1E-06
2E-05
1E-06
2E-03
PET NEI26101
2E-04
1E-05
2E-04
1E-05
2E-02
PET NEI26218
1E-02
1E-03
2E-03
3E-04
2E-01
PET NEI26473
1E-06
2E-06
1E-06
5E-07
3E-05
PET NEI26489
4E-04
2E-05
4E-04
2E-05
5E-02
1 of 3
-------
Table 3 - Maximum Predicted HEM-AERMOD Acute Risks
Facility NEI ID
Maximum Hazard Quotient1
AEGL1
AEGL2
ERPG1
ERPG2
REL
PET NEI26533
3E-03
2E-04
3E-03
2E-04
4E-01
PET NEI2CA131003
2E-03
1E-04
2E-03
1E-04
1E-01
PET NEI2CA254640
2E-04
1E-05
2E-04
2E-05
2E-02
PET NEI2CA314628
8E-03
1E-03
1E-03
3E-04
2E-01
PET NEI2KS125003
6E+00
3E-01
6E+00
3E-01
2E+01
PET NEI2TX14199
6E-05
4E-06
6E-05
4E-06
8E-03
PET NEI32353
4E-01
2E-02
4E-01
2E-02
1E+00
PET NEI32762
1E-01
5E-03
1E-01
5E-03
6E-01
PET NEI32801
3E-02
1E-03
3E-02
1E-03
4E-01
PET NEI32864
3E-01
2E-02
3E-01
2E-02
5E+01
PET NEI32997
7E-05
4E-06
7E-05
4E-06
9E-03
PET NEI33007
2E-05
3E-06
3E-06
1E-06
5E-04
PET NEI33008
2E-02
4E-03
2E-02
9E-03
5E-01
PET NEI33010
1E-03
1E-04
1E-03
1E-04
2E-01
PET NEI33030
3E-03
2E-04
3E-03
2E-04
4E-01
PET NEI33031
3E-03
2E-04
3E-03
5E-04
2E-01
PET NEI33039
3E-04
5E-05
3E-04
4E-05
4E-02
PET NEI34022
1E-01
7E-03
1E-01
7E-03
1E+00
PET NEI34050
2E-03
3E-04
1E-03
4E-04
2E-01
PET NEI34057
3E-01
2E-02
3E-01
2E-02
6E+00
PET NEI34061
4E-03
2E-04
4E-03
2E-04
4E-02
PET NEI34062
7E-02
5E-03
7E-02
5E-03
9E-01
PET NEI34069
1E-04
7E-06
1E-04
7E-06
1E-02
PET NEI34862
2E-03
3E-04
2E-03
1E-04
3E-01
PET NEI34863
2E-04
3E-05
2E-04
2E-05
3E-02
PET NEI34872
9E-04
8E-05
2E-04
7E-05
1E-02
PET NEI34873
4E-03
3E-04
4E-03
7E-04
6E-01
PET NEI34898
1E-02
4E-04
1E-02
1E-03
2E-01
PET NEI34907
9E-04
6E-05
9E-04
1E-04
1E-01
PET NEI34912
2E-04
3E-05
2E-04
3E-05
3E-02
PET NEI363
1E-03
1E-04
1E-03
1E-04
2E-01
PET NEI371
0E+00
0E+00
0E+00
0E+00
0E+00
PET NEI40371
1E-01
6E-03
1E-01
6E-03
3E+00
PET NEI404
8E-02
3E-03
8E-02
3E-03
3E-01
PET NEI40531
9E-02
6E-03
9E-02
6E-03
1E+00
PET NEI40625
1E-03
7E-05
1E-03
1E-04
1E-01
PET NEI40723
4E-04
5E-05
4E-04
1E-04
5E-02
PET NEI40732
4E-04
3E-05
4E-04
7E-05
6E-02
PET NEI415
5E-03
3E-04
5E-03
3E-04
2E-01
PET NEI41591
1E-01
5E-03
1E-01
5E-03
4E-01
PET NEI41771
2E-02
2E-03
2E-02
2E-03
3E+00
PET NEI41863
4E-02
2E-03
4E-02
2E-03
3E-01
PET NEI41864
4E-03
3E-04
2E-03
1E-04
3E-01
PET NEI41865
2E-03
1E-04
2E-03
1E-04
2E-01
PET NEI42016
9E-01
4E-02
9E-01
4E-02
3E+00
PET NEI42020
2E-03
1E-04
2E-03
3E-04
3E-01
PET NEI42025
2E-03
2E-04
2E-03
3E-04
3E-01
PET NEI42040
2E-02
3E-03
5E-03
2E-03
6E-01
PET NEI42081
2E-03
2E-04
2E-03
5E-04
2E-01
PET NEI42309
2E-02
1E-03
5E-04
1E-04
6E-02
PET NEI42370
2E-04
3E-05
2E-04
8E-05
2E-02
PET NEI42381
1E-02
8E-04
1E-02
1E-03
6E-01
PET NEI42382
6E-04
9E-05
6E-04
1E-04
8E-02
2 of 3
-------
Table 3 - Maximum Predicted HEM-AERMOD Acute Risks
Facility NEI ID
Maximum Hazard Quotient1
AEGL1
AEGL2
ERPG1
ERPG2
REL
PET NEI42413
3E-03
2E-04
3E-03
3E-04
3E-01
PET NEI42425
1E-02
6E-04
1E-02
6E-04
8E-02
PET NEI42583
4E-04
4E-05
4E-04
7E-05
5E-02
PET NEI43243
3E-03
2E-04
3E-03
2E-04
4E-01
PET NEI46556
3E-02
4E-03
2E-02
7E-03
1E+00
PET NEI46752
1E-02
2E-03
1E-02
5E-03
4E-01
PET NEI46764
3E-04
4E-05
2E-04
7E-05
2E-02
PET NEI49781
1E-03
6E-05
1E-03
6E-05
1E-01
PET NEI53702
2E-02
1E-03
2E-02
1E-03
2E+00
PET NEI53718
8E-03
1E-03
8E-03
2E-03
1E+00
PET NEI55835
2E-04
3E-05
2E-04
2E-05
2E-02
PET NEI6018
2E-04
1E-05
2E-04
2E-05
2E-02
PET NEI6022
9E-01
7E-02
9E-01
7E-02
1E+00
PET NEI6062
3E-03
8E-04
3E-03
5E-04
4E-01
PET NEI6084
4E-03
3E-04
4E-04
1E-04
3E-02
PET NEI6087
8E-01
3E-02
8E-01
3E-02
3E+00
PET NEI6095
2E-03
2E-04
2E-03
2E-04
2E-01
PET NEI6116
6E-03
2E-03
6E-03
2E-03
8E-01
PET NEI6123
6E-03
6E-04
6E-03
4E-04
8E-01
PET NEI6127
7E-04
5E-05
7E-04
6E-05
1E-01
PET NEI6130
2E-02
1E-03
2E-02
1E-03
7E-01
PET NEI6136
1E-01
9E-03
1E-01
9E-03
2E+00
PET NEI6166
3E-01
3E-02
3E-01
3E-02
2E+00
PET NEI6375
1E-03
1E-04
8E-04
1E-04
1E-01
PET NEI6436
2E-02
1E-03
2E-02
2E-03
2E+00
PET NEI6446
1E-02
1E-03
1E-02
3E-03
2E+00
PET NEI6475
5E-03
5E-04
5E-03
1E-03
6E-01
PET NEI6519
4E-02
2E-03
4E-02
3E-03
5E+00
PET NEI6617
3E-02
3E-03
3E-02
3E-03
3E-01
PET NEI6963
3E-01
1E-02
3E-01
1E-02
1E+00
PET NEI7130
2E-03
3E-04
1E-03
5E-04
2E-01
PET NEI7134
2E-03
4E-04
2E-03
4E-04
3E-01
PET NEI7233
4E-02
7E-03
4E-02
2E-02
9E-01
PET NEI7441
1E-02
2E-03
4E-03
2E-03
5E+00
PET NEI7781
8E-03
5E-04
8E-03
5E-04
1E+00
PET NEI7973
3E-05
3E-06
3E-05
2E-06
4E-03
PET NEI7988
6E-02
4E-03
6E-02
1E-02
7E+00
PET NEI8139
2E-03
3E-04
2E-03
4E-04
3E-01
PET NEI8612
1E-03
8E-05
1E-03
8E-05
2E-01
PET NEI876
3E-04
2E-03
0E+00
0E+00
0E+00
PET NEI889
1E-02
8E-04
1E-02
8E-04
1E-01
PET NEICA0370363
5E-04
3E-05
5E-04
5E-05
5E-02
PET NEICA0379991
7E-05
1E-05
4E-05
1E-05
2E-03
PET NEICA10578
4E-03
5E-04
8E-04
3E-04
9E-02
PET NEICA1910268
5E-03
7E-04
5E-03
7E-04
8E-02
PET_NEINJT$891
3E-02
1E-03
3E-02
1E-03
3E-01
PET_NEINMT$12478
8E-04
5E-05
8E-04
1E-04
1E-01
PET_NEIOKT$11009
4E-01
2E-02
4E-01
2E-02
6E+00
PET_NEIPRT$64
1E-02
9E-04
1E-02
1E-03
2E+00
PET_NEI WYT$12156
1E-04
7E-06
1E-04
7E-06
1E-02
1 BOLD RED indicates a cancer risk great than 1 in a million or a noncancer risk greater than 1
3 of 3
-------
Table 4 - Maximum Predicted Acute Risks Greater than 1 (Refined Approach)
HEM-3
Refined
Figure No.
Facility NEI ID
Pollutant
Criteria
(Screening)
Results 1
Refined Modeling Notes 2
8-1
PET_NEI12044
Benzene
REL
7
1
NNW of facility
8-2
PET_NEI12486
Glycol Ethers
REL
5
1
W of facility
8-3
PET_NEI12486
Hydrofluoric acid
REL
15
4
W of facility
8-4
PET_NEI12486
Hydrofluoric acid
AEGL-1
4
1
W of facility
8-5
PET_NEI12988
Benzene
REL
2
1
NNW of facility
8-6
PET_NEI2KS125003
Hydrofluoric acid
REL
22
5
E of facility
8-7
PET_NEI2KS125003
Hydrofluoric acid
AEGL-1
6
2
E of facility
8-8
PET_NEI32864
Benzene
REL
45
8
E of facility
8-9
PET_NEI34057
Benzene
REL
6
<1
8-10
PET_NEI34057
Formaldehyde
REL
4
<1
8-11
PET_NEI34057
Methanol
REL
2
<1
8-12
PET_NEI34057
p-Xylene
REL
3
<1
8-13
PET_NEI40371
Glycol Ethers
REL
3
1
E of facility
8-14
PET_NEI41771
Benzene
REL
3
2
W of facility
8-15
PET_NEI42016
Hydrofluoric acid
REL
3
2
ESE of facility
8-16
PET_NEI53702
Benzene
REL
2
<1
8-17
PET_NEI6087
Hydrofluoric acid
REL
3
<1
8-18
PET_NEI6136
Formaldehyde
REL
2
1
WSW of facility
8-19
PET_NEI6166
Benzene
REL
2
1
SE of facility
NA
PET_NEI6436
Benzene
REL
2
2
no refinement; receptor is a census block
8-20
PET_NEI6446
Benzene
REL
2
<1
8-21
PET_NEI6519
Benzene
REL
5
<1
8-22
PET_NEI7441
Glycol Ethers
REL
5
<1
8-23
PET_NEI7988
Benzene
REL
7
1
S of facility
8-24
PET_NEIOKT$11009
Benzene
REL
6
4
E of facility
8-25
PET_NEIPRT$64
Benzene
REL
2
2
W of facility
1 Facilites with a HEM-3 screening acute value greater than 1 were remodeled with a more refined approach
2 Indicates offsite impacts using aerial photographs of facility
1 of 1
-------
Appendix E
Refinement of acute exposure estimates
at petroleum refining facilities
and Portland cement facilities
-------
Appendix El
Refined Acute Assessment for Petroleum Refineries
-------
Initial acute screening risk calculations were performed with the HEM-3 model. HEM-3
estimates acute (1-hour) impacts at both polar and census block receptors. It is assumed for this
short period of time that an exposed individual could be located at any offsite location. The lack
of readily available detailed property boundary information for many of the facilities evaluated
made it difficult to determine whether receptors were on- or offsite. In the absence of such
information, the first ring of polar receptors was placed 100 meters from the plant center for
many facilities. However, these polar rings often transected onsite locations, preventing public
access to exposures at these levels and thereby overestimating exposures. The screening analysis
indicated that 20 facilities had the potential to exceed a 1-hour reference value for one or more
pollutants. To refine the analysis for these 20 facilities, the polar receptors for each facility were
overlaid on an aerial photograph of the facility to determine the offsite receptor with the highest
1-hour exposure. Figures El-1 through El-25 depict the modeled acute hazard quotients for
these facilities. Table El-1 summarizes the results of this refinement by listing the modeled
maximum screening and refined (offsite) hazard quotient values.
2
-------
Table E1-1 - Maximum Modeled Acute Hazard Quotients (Refined Approach)
Figure
No.
Facility NEI ID
Pollutant
Criteria
Screening
HQ
Refined
HQ 1
Refined Modeling
Notes
E1-1
PET NEI12044
Benzene
REL
7
1
NNW of facility
E1-2
PET NEI12486
Glycol Ethers
REL
5
1
W of facility
E1-3
PET NEI12486
Hydrofluoric acid
REL
15
4
W of facility
E1-4
PET NEI12486
Hydrofluoric acid
AEGL-1
4
1
W of facility
E1-5
PET NEI12988
Benzene
REL
2
1
NNW of facility
E1-6
PET NEI2KS125003
Hydrofluoric acid
REL
22
5
E of facility
E1-7
PET NEI2KS125003
Hydrofluoric acid
AEGL-1
6
2
E of facility
E1-8
PET NEI32864
Benzene
REL
45
8
E of facility
E1-9
PET NEI34057
Benzene
REL
6
<1
E1-10
PET NEI34057
Formaldehyde
REL
4
<1
E1-11
PET NEI34057
Methanol
REL
2
<1
E1-12
PET NEI34057
p-Xylene
REL
3
<1
E1-13
PET NEI40371
Glycol Ethers
REL
3
1
E of facility
E1-14
PET NEI41771
Benzene
REL
3
2
W of facility
E1-15
PET NEI42016
Hydrofluoric acid
REL
3
2
ESE of facility
E1-16
PET NEI53702
Benzene
REL
2
<1
E1-17
PET NEI6087
Hydrofluoric acid
REL
3
<1
E1-18
PET NEI6136
Formaldehyde
REL
2
1
WSW of facility
E1-19
PET NEI6166
Benzene
REL
2
1
SE of facility
NA
PET NEI6436
Benzene
REL
2
2
no refinement; receptor
is a census block
E1-20
PET NEI6446
Benzene
REL
2
<1
E1-21
PET NEI6519
Benzene
REL
5
<1
E1-22
PET NEI7441
Glycol Ethers
REL
5
<1
E1-23
PET NEI7988
Benzene
REL
7
1
S of facility
E1-24
PET_NEIOKT$11009
Benzene
REL
6
4
E of facility
E1-25
PET_NEIPRT$64
Benzene
REL
2
2
W of facility
Where facilities had a HEM-3 screening acute HQ greater than 1, HQ values at polar receptors were overlaid on
aerial photographs to determine the maximum offsite value.
3
-------
;ure El-1. NEI12044 Benzene Acute HQ values (based on REL)
1 f; i _ M fTFtf* Wif.-k v . 1 ¦ - ¦ k '
< ILt? : mairlBBr - •
0.59
-o
•W
0.77
4
-------
Figure El-2. NEI12486 Glycol Ethers Acute HQ values (based on REL)
5
-------
Figure El-3. NEI12486 Hydrofluoric Acid Acute HQ values (based on REL)
6
-------
Figure El-4. NEI12486 Hydrofluoric Acid Acute HQ values (based on AEGL-1
7
-------
Figure El-5. NEI12988 Benzene Acute HQ values (based on REL)
8
-------
Figure El-6. NEI2KS125003 Hydrofluoric Acid Acute HQ values (based on REL
9
-------
Figure El-7. NEI2KS125003 Hydrofluoric Acid Acute HQ values (based on AEGL-1)
10
-------
Figure El-8. NEI32864 Benzene Acute HQ values (based on REL
-------
Figure HI-9. NEI34057 Benzene Acute HQ values (based on REL
Im age m 2DOS NQAA ,p -j g -j t al G1 obe
-------
Figure El-10. NEI34057 Fomialdehyde Acute HQ values (based on REL
DlGtTALGLOBE
I m ag e LeJ 200S N Q Aft, p -j g -j t al G1 o b e
-------
Figure El-11. NHI34057 Methanol Acute HQ values (based on REL
DIGITALGLOUE'
• ¦
-------
Figure El-12. NF. 13 405 7 p-Xylene Acute F[Q values (based on REL
DlGtTALGLOBE
* *' 9.11'
-------
Figure El-13. NEI40371 Glycol Ethers Acute HQ values (based on REL
-------
Figure El-14. NEI41771 Benzene Acute HQ values (based on REL
17
-------
-------
igure El-16. NEI53702 Benzene Acute HQ values (based on REL)
Jv, -V 0 '
0L41 •
-------
iP'Tg-iit al Gil]
Figure El-17. NEI6087 Hydrofluoric Acid Acute HQ values (based on REL)
m P ^
0.7 . 0.5 0
^ .>^7 <>•%<> 0« 0.5
j f 0.5 tW <50VW 0-7 °
:0.9 0.7o
V> <>J 0.8 °l c;:f^o0^0.8 0.7 0.4
20
-------
Image Is) 2008 Digi tal Gl obe
Figure El-18. NEI6136 Formaldehyde Acute HQ values (based on REL)
KMj~ * • V" '' '."'JJ—rfiiliB.1!" T- *™ JL45S* r .
21
-------
Figure El-19. NEI6166 Benzene Acute HQ values (based on REL)
wmm,
¦ < ©' 1
o ^
Sl<- >•"
22
-------
Figure El-20. NEI6446 Benzene Acute HQ values (based on REL
23
-------
Figure El-21. NEI6519 Benzene Acute HQ values (based on REL
D1GITALG LOSL
¦
Image l£i 2008 DigitaljGl'bbe
-------
Figure El-22. NEI7441 Glycol Ethers Acute HQ values (based on REL
-------
Figure El-23. NEI7988 Benzene Acute HQ values (based on REL
26
-------
Figure El-24. NEI0KT$11009 Benzene Acute HQ values (based on REL)
% i » |
V
Sg^Waf
'gB • Al
o'-5
1.3
2 5
0" 3.9
o o >
fc66
a46 4siO J-
:
' M ''
£)M&3I HP
'. Ki & 'K- ¦ •' Be;;: ^r
ftiti
,/Ji * ill ¦ ¦ ° fe
2.2.2-4 i7 <
1.5 :?.4?-®f o5"1
O ':5 f,4 o
0.82 1.2 1.4 '1.9 **3.4 49 5.7
o o o o '¦:> gO O
1-74l8 23 3,1
;'-4 "1.7 : :,:®r-?.° 1% 02-8 ' Z;.,
01.5112M 1J ° - ""
#.95
27
-------
Figure El-25. NEIPRT$64 Benzene Acute HQ values (based on REL
DlGITALGLOBE'
(133 fc310.28
-------
Appendix E2
Refined Acute Assessment
For Portland Cement Facilities
-------
Initial acute screening risk calculations were performed with the HEM-3 model. HEM-3
estimates acute (1-hour) impacts at both polar and census block receptors. It is assumed for this
short period of time that an exposed individual could be located at any offsite location. The lack
of readily available detailed property boundary information for many of the facilities evaluated
made it difficult to determine whether receptors were on- or offsite. In the absence of such
information, the first ring of polar receptors was placed 100 meters from the plant center for
many facilities. However, these polar rings often transected onsite locations, preventing public
access to exposures at these levels and thereby overestimating exposures. The screening analysis
indicated that 8 facilities had the potential to exceed a 1-hour AEGL-1 reference value, and 4
facilities had the potential to exceed a 1-hour AEGL-2 reference value for one or more
pollutants. To refine the analysis for these facilities, the polar receptors for each facility were
overlaid on an aerial photograph of the facility to determine the offsite receptor with the highest
1-hour exposure. Figures E2-1 through E2-9 depict the modeled acute hazard quotients for these
facilities. Table E2-1 summarizes the results of this refinement by listing the modeled maximum
screening and refined (offsite) hazard quotient values.
2
-------
Table E2-1 - Maximum Modeled Acute Hazard Quotients (Refined Approach)
Figure
No.
Facility NEI ID
Pollutant
Criteria
Screening
HQ
Refined
HQ 1
Refined Modeling
Notes
E2-1
NEI11181
Hydrochloric acid
AEGL-1
40
6
SW of facility
HQ reduced by same
NEI11181
Hydrochloric acid
AEGL-2
3
<1
factor as AEGL-1 HQ
E2-2
NEI16783
Chlorine
AEGL-1
7
2
W of facility
HQ reduced by same
NEI16783
Chlorine
AEGL-2
2
<1
factor as AEGL-1 HQ
E2-3
NEI16783
Hydrochloric acid
AEGL-1
50
10
W of facility
HQ reduced by same
NEI16783
Hydrochloric acid
AEGL-2
4
<1
factor as AEGL-1 HQ
HQ reduced by same
E2-4
NEI22743
Formaldehyde
AEGL-1
3
<1
factor as AEGL-1 HQ
E2-5
NEI22838
Hydrochloric acid
AEGL-1
20
3
N of facility
HQ reduced by same
NEI22838
Hydrochloric acid
AEGL-2
2
<1
factor as AEGL-1 HQ
E2-6
NEI25375
Hydrochloric acid
AEGL-1
20
3
Multiple Locations
HQ reduced by same
NEI25375
Hydrochloric acid
AEGL-2
2
<1
factor as AEGL-1 HQ
E2-7
NEI338
Formaldehyde
AEGL-1
3
2
E of facility
E2-8
NEI40539
Hydrochloric acid
AEGL-1
9
3
E of facility
E2-9
NEI51527
Hydrochloric acid
AEGL-1
10
2
Multiple Locations
Where facilities had a HEM-3 screening acute HQ greater than 1, HQ values at polar receptors were overlaid on
aerial photographs to determine the maximum offsite value.
3
-------
Figure E2-1. NEI11181 Hydrochloric Acid Acute HQ values (based on AEGL-1)
-------
5
-------
6
-------
Figure E2-4. NEI22743 Formaldehyde Acute FTQ values (based on AEGL-1)
-------
8
-------
Figure E2-6. NEI25375 Hydrochloric Acid Acute HQ values (based on AEGL-1)
20Q8 -i t al Gl obfe
-------
Figure E2-7. NEI338 Formaldehyde Acute HQ values (based on AEGL-1)
lEJlGil iTA Lr GiLOB L
10
-------
Figure E2-8. NEI40539 Hydrochloric Acid Acute HQ values (based on AEGL-1)
11
-------
Figure E2-9. NEI51527 Hydrochloric Acid Acute HQ values (based on AEGL-1)
-------
Appendix F
Statistical analysis of operational parameters to assess air
emissions of chlorinated dibenzodioxins and furans (CDD/F)
from portland cement facilities
-------
Appendix F: Statistical analysis of operational parameters to
assess air emissions of chlorinated dibenzodioxins and furans
(CDD/F) from portland cement facilities
F.1 Introduction
Because very small amounts of CDD/F emitted from any source may become an
important environmental and health stressor, data collection for these substances tends to
be expensive and data scarce. Thus, estimates of risks associated with CDD/F exposure
are often both important and uncertain.
In its 2002 dioxin report, EPA derived a single emission factor of 0.27 ng 2378-
TCDD(teq)/ kg clinker for all non-hazardous waste combustion units for this source
category, based upon stack tests from 13 sites. This factor was developed for all kilns
regardless of type or operational parameters.
This analysis attempts to refine EPA's 2002 effort by searching for statistical correlations
between CDD/F emissions from non-hazardous waste combustors operated by the
Portland cement industry and site-specific operational parameters.
F.2 Methods
We obtained the following operational parameters to use as independent variables:
1. Process parameters1: Temperature at the inlet to the air pollution control device,
flow rate at the stack, kiln type, and estimated kiln clinker capacity.
2. Manufacturing process (type of kiln )2 - wet, dry, dry with preheater, or dry with
both preheater with precalciner.
The hierarchy of data resources for operational data was as follows: 1) stack compliance
tests; 2) PCA survey of HO stack emissions; and 3) 2002 NEI data for HC1. Process data
for HO were considered a plausible surrogate for dioxin because both are emitted from
the same combustion stacks.
We obtained CDD/F emission estimates and emission factors reported to the Toxic
Release Inventory (TRI)3 by 60 non-hazardous waste combustion cement plants from
2002-2006. We then developed site-specific 2378-TCDD, rH-)| emission factors by
dividing the 2378-TCDD,n q, emissions by the estimated clinker capacity data4 for each
facility. For emission data reported as CDD/F congeners, TEQ calculations were based
on the 1998 World Health Organization (WHO) recommendations. The use of separate
1) RTI Memo to EPA (August 31, 2007): Draft: Design Options and Data for Cement Emissions Trading
Market Model: PCA kiln member companies compiled by RTI: Kiln Data 1 l_05_07.xls
2) ENVIRON International Corporations (PCA kiln data from HC1 assessment) and 2002 NEI data
3) Toxic Release Inventory (TRI): http://www.epa.gov/tri
4) RTI Memo to EPA (August 31, 2007): Draft: Design Options and Data for Cement Emissions Trading
Market Model: PCA kiln member companies compiled by RTI: Kiln Data 1 l_05_07.xls
F-1
-------
emission factors for each kiln type helps to account for the variability in production for
each year as well as the variability in the amount of dioxin that is generated by the
combustion process from each type of kiln. This variability is evident when comparing
emission factors between the wet and dry type kilns.
Based upon historic production rates for this source category, EPA assumed that the
maximum allowable production rates were approximate to actual rates. Table F-1
contains a summary of TRI dioxin (TEq) air emissions for this source category.
Table F-1: Dioxiri(jEQ) emission inventory estimates for portland cement for non-hazardous
waste combustion kilns
Source of Data
Air Emissions
(grams/year)
Reporting Universe
2002-2006 TRI Data (Avg)
13.8
60
EPA's Dioxin Report (Calendar
Year - 2000)
17.2
102*
EPA estimate for RTR with
mean emission factor (Avg from
2002-06 TRI Data)
18
97
*-April 1999; Portland Cement NESHAP Final Rule Economic Analysis;
http://epa.gov/ttn/atw/pcem/iv-a-004.pdf
Once these data were assembled we used SAStm software to calculate correlation
coefficients for emissions vs. the two continuous variables (flow and inlet temperature),
and an analysis of variance to test for differences among emission factors from different
kiln types. Correlation analyses were run using a subset of the database that contained a
complete set of temperature and flow measurements. The analysis of variance, for which
only emission factor and kiln type were needed, used the full dataset.
F.3 Results and discussion
F.3.1 Correlation analysis
Results of the correlation analysis using the more limited complete data set are shown in
Table F-2. Reported TCDD/F(Teq) emissions did not correlate significantly with inlet
temperature, stack flow rate, or an interaction between the two.
Table F-2: Results of SAS correlation analysis between TCDD/F(Teq) emissions and
continuous process variables
or portland cement facilities.
Variable
DF
Type I SS
Mean Square
F Value
Pr>F
Temperature
I
0.00018237
0.00018237
1.82
0.1884
Flow
I
0.00010613
0.00010613
1.06
0.3123
Temp*Flow
I
0.00013916
0.00013916
1.39
0.2487
F.3.2 Analysis of variance
Based on the full TRI-derived database (212 observations spanning 5 years), the ANOVA
found statistically significant differences in TCDD/F(Teq) emission factors among the four
F-2
-------
process types (F=17.7, P<0.001). Table F-3 shows the mean emission factors for each
type, and the 95% upper confidence level for each.
Table F-3: Mean and 95% UCL TCDD/F, rK), emission factors for Portland cement
facilities, by kiln type
Kiln type
Mean emission factor
(ng/kg clinker capacity)
95% UCL emission factor
(ng/kg clinker capacity)
Dry
0.110
0.229
Dry with preheater and
precalciner
0.170
0.614
Dry with preheater
0.168
0.377
Wet
0.768
1.877
These emission factor estimates plausibly bracket the single estimate (0.27 ng/kg) from
EPA's 2002 Dioxin Report. Given this plausibility, we characterized CDD/F emissions
by kiln type for the Portland cement risk assessment, and calculated plant-specific risks
separately using the mean and UCL emissions factors.
F-3
-------
APPENDIX G:
Evaluation of Radionuclide Emissions Reported
for Portland Cement Sources
G-i
-------
TABLE OF CONTENTS
G.1 Issue 1
G.2 Approach 2
G.2.1 Estimating Radionuclide Emissions from Portland Cement Sources 2
G.2.2 AERMOD Modeling Approach for Estimating Inhalation Risk at Two
California Facilities 4
G.2.3 CAP88 Approach for Estimating Inhalation Risk at Two California Facilities 4
G.2.4 AERMOD Modeling Approach for Estimating Inhalation Risk at All U.S.
Facilities 5
G.3 Results 6
G.3.1 Emissions 6
G.3.2 Risk 9
G.4 Uncertainties 15
G.5 References 16
G-ii
-------
LIST OF EXHIBITS
Exhibit G-1. Portland Cement Manufacturing Facilities in California Reporting Radionuclide
Emissions in the 2007 NEI Database 1
Exhibit G-2. Details of the Maastricht Facility for Modeling Portland Cement Facilities a 3
Exhibit G-3. Emission Scaling Factors Derived from Data 3
Exhibit G-4. Inhalation Unit Risk Estimation for 210Po and 222Rn 4
Exhibit G-5. Population Scaling Factors Applied to the CAP88 Default Population Files Around
Two California Portland Cement Facilities 5
Exhibit G-6. Estimation of Radionuclide Emissions for the Two California Facilities Using Three
Approaches 6
Exhibit G-7. Estimation of Radionuclide Emissions for All U.S. Portland Cement Facilities,
Based on the Clinker Production Scaling Factor 6
Exhibit G-8. Risk Calculated for Two California Portland Cement Facilities Using AERMOD
Modeling Results and Three Emission Estimation Approaches 9
Exhibit G-9. Risk for Two California Portland Cement Facilities Modeled with CAP88 and Two
Emission Estimation Approaches 10
Exhibit G-10. Estimation of Maximum Incremental Risk from Estimated Radionuclide Emissions
for the Two California Facilities Using Three Approaches and Two Modeling Systems .10
Exhibit G-11. Summary of Maximum Incremental Cancer Risk Modeled in AERMOD and
CAP88, Using Different Radionuclide Emission Estimates 11
Exhibit G-12. Estimation of Maximum Incremental Risk (MIR) for Cancer from Radionuclide
Emissions for All U.S. Portland Cement Facilities, Based on the Clinker Production
Scaling Factor 12
Exhibit G-13. Annual Effective Radiation Dose from Background Sources a 14
Exhibit G-14. Comparison of Maximum Incremental Cancer Risk from Radionuclide Emissions
and Non-Radioactive HAP Emissions from Portland Cement Facilities 15
G-iii
-------
G.1 Issue
Radionuclides, a class of atoms that spontaneously undergo radioactive decay, are regulated as
hazardous air pollutants (HAPs) when they are emitted to the air. Emissions of radionuclides
from industrial facilities are reported in the 2002 National Emissions Inventory (NEI) in mass-
based units of tons (U.S., short) per year. However, the known hazards from radionuclides are
most closely associated with the radioactivity and cancer potency of the material released, not
with the mass of the material released. Radioactive isotopes vary by level of radioactivity
(which can be expressed as picocuries [pCi] per metric ton of the material), cancer potency, and
persistence in the environment. Furthermore, the products of radioactive decay are different
depending on the isotope, and so different isotopes may produce different daughter products
and subsequent radioactive decompositions. Therefore, reporting unspeciated emissions of
radioactive substances from a single facility collectively by mass, rather than individually by
radioactive isotope in units of radioactivity, prevents the accurate estimation of risks posed by
radionuclides emitted from industrial facilities.
The Portland cement manufacturing sector is one source of radionuclide emissions. To
evaluate the potential radionuclide hazards associated with Portland cement manufacturing
facilities, we identified two facilities in California that reported emissions in the 2002 NEI
(EPA 2008a).1 Identifying details of these facilities, including NEI-reported radionuclide
emissions, are presented in Exhibit G-1.2 This exhibit also includes actual clinker production
and particulate matter (PM) emissions because these values are used later in this appendix as
an alternate means of estimating radionuclide emissions.
Exhibit G-1. Portland Cement Manufacturing Facilities in California Reporting
Radionuclide Emissions in the 2007 NEI Database
Facility Name
NTI Site ID
NEI Reported
Radionuclide
Emissions
(ton/yr)
Actual Clinker
Production a
(short ton/yr)
PM Emissions
b
(ton/yr)
Lehigh Southwest Cement Co.
NEICA1505122
8.21 E-03
1.00E+06
2.73E+02
RMC Pacific Materials (CEMEX)
NEI2CA151186
1.42E-08
8.50E+05
3.90E+02
a EPA Dioxin Emission Inventory (2007a).
b CARB (2007).
On a mass basis, emissions reported for these facilities are very low (e.g., the highest emitter,
Lehigh Southwest Cement, reports only 0.00821 ton of radionuclides). Nevertheless, significant
risks may be associated with emissions of this magnitude because, at even minute
concentrations, some radionuclides are associated with large risks. However, which
radionuclides were emitted and how much of each was emitted is uncertain. Also, whether the
facilities are reporting radionuclide emissions in a uniform manner is not certain. Some facilities
may be reporting the total mass of materials that contain radionuclides, even in trace quantities,
and some may be reporting what they believe to be the actual mass of radionuclides emitted.
1 As part of the RTR data review process, EPA reviewed and updated the inventory, so the data analyzed in this
appendix are the most recent data available. Although the most recent version does not include any radionuclide
emissions for the Lehigh Southwest Cement Co. facility, radionuclide emission data from a previous version were
included in this analysis to provide an additional point for analysis.
2 Unless otherwise specified, tons are metric units throughout this appendix.
G-1
-------
This appendix describes an analysis performed on the two facilities presented in Exhibit G-1
and on the Portland cement manufacturing sector. We performed this analysis to:
. Estimate and characterize the actual emissions of radionuclides from Portland
cement manufacturing facilities that report radionuclide emissions and from all U.S.
Portland cement manufacturing facility sources;
. Evaluate the utility of the values reported in NEI for estimating risks from
radionuclides;
. Quantitatively estimate potential incremental inhalation cancer risks posed by
radionuclides emitted by Portland cement manufacturing facilities; and
. Qualitatively evaluate background exposures and risks from inhalation and non-
inhalation exposures to radionuclides.
The analysis has two parts: (1) estimating radionuclide emissions and (2) evaluating resultant
potential incremental cancer risks associated with inhalation exposures to these emissions. We
completed the first part by comparing NEI-reported emissions with modeled emissions, and we
performed the second through a series of modeling exercises. This appendix chronicles the
methods, results, and conclusions of the analysis, and also discusses the assumptions,
uncertainties, and additional data needs.
G.2 Approach
This section presents the approaches used to model radionuclide emissions and to estimate
potential incremental cancer risks from Portland cement manufacturing facilities. We estimated
radionuclide emissions for the two Portland cement sources identified in Exhibit G-1 using the
NEI-reported emissions and scaling factors developed from a "typical" Portland cement facility.
We derived emission factors for a typical facility from the European Commission Radiation
Protection 135 report (Chen et al. 2003), hereafter referred to as the "EU naturally occurring
radioactive material (NORM) report."
The EU NORM report relied upon an analysis by Leenhouts et al. (1996) that examined one
large Portland cement facility in the Netherlands using data from 1990. The heating of clinker in
a cement kiln results in the volatilization of radioactive material, including polonium-210 (210Po)
and radon-222 (222Rn), of which approximately 50 percent and 100 percent, respectively, of the
radionuclide content in the material is assumed to escape (EU NORM report 2003 and
Leenhouts et al. 1996). Uncertainties stemming from the assumption that the facility described
by Leenhouts et al. (hereafter, "the Maastricht facility") is representative of U.S. Portland cement
facilities are discussed in Sections G.2.1 and G.4.
Resultant ground-level radionuclide activity (in pCi/m3) was modeled with HEM3 (Human
Exposure Model-3, EPA 2007b) using the emissions described in Section G.2.1 below
separately for (1) the two California facilities using linearly scalable mass-based emissions of
other pollutants, and (2) all U.S. facilities, using emissions developed with the clinker production
factors.
G.2.1 Estimating Radionuclide Emissions from Portland Cement Sources
To determine whether the emissions reported in NEI are plausible and useful for our purpose
here, we compared NEI-reported emissions to our modeled emissions. The Maastricht facility
was used to develop emission factors that were applied to U.S. Portland cement producers.
Emission factors were developed by assuming that radionuclide emissions from Portland
cement manufacturing facilities were proportional to (1) actual clinker production or (2) PM
G-2
-------
emissions. The parameters of the Maastricht facility used to model U.S. Portland cement
producers are presented Exhibit G-2. Factors derived from the data in Exhibit G-2 for emission
rates of 210Po and 222Rn are presented in Exhibit G-3.
Exhibit G-2. Details of the Maastricht Facility for Modeling Portland Cement Facilities a
Cement production (kton) b
PM Emissions (kton)
210Po (Ci/yr)c
222Rn (Ci/yr)
2,000
8.0
2.108
4.243
a Leenhouts et al. (1996), EU NORM (2003).
b kton = kiloton; Leenhouts et al. report 2,107 kton total cement production, with 980 kton Portland cement, and the
remaining amount as other types of output. They report only 365 kton Portland clinker production. The EU NORM
document rounds to 2,000 kton cement production. Uncertainties related to these data are discussed below.
c Ci/yr = curies per year.
Exhibit G-3. Emission Scaling Factors Derived from Data
from the Maastricht Facility
Method for Estimating Emissions of Radioactivity
Emission Scaling Factor
210Po
222Rn
Based on clinker production (Ci[emitted]/ton[clinker produced])
1.05E-06
2.12E-06
Based on PM emission rate (Ci[emitted]/ton[PM emitted])
2.64E-04
5.30E-04
Other than heating in the kiln, which was identified as the main emission source at the
Maastricht facility, emissions were considered to result from three processes: extraction,
transport, and mixing (Leenhouts et al. 1996). The facility is estimated to release 0.5 to 1
percent by mass of the total input material as dust (Leenhouts et al. 1996). Although the
Maastricht facility provides a well-documented data set and serves as a good case study, the
comparisons between it and U.S. facilities have limitations. First, the data from Maastricht (from
1990) are relatively outdated and may not reflect advances in pollution control technologies,
especially for PM. The emission scaling factor approach assumes that the facilities in California
in 2005 had industrial processes and PM emission control equipment that are identical to those
at the Maastricht facility in 1990 and that radionuclide emission rates are directly proportional to
clinker production volume or PM emissions. Similarly, the scaling factor approach assumes that
radionuclide content in input materials is identical; this may not be true because raw materials
are obtained from different regions of the globe and may contain different quantities of
radiological materials.
The Maastricht facility data were used to estimate the relative contribution of 210Po and 222Rn (94
percent and 6 percent by mass, respectively) to the total radionuclide emissions reported in NEI.
The emissions were then converted to radioactivity units by multiplying by the specific activity
for each radionuclide: 4,493 curies per gram (Ci/g) for 210Po and 153,800 Ci/g for 222Rn (EPA
2001). We assumed that 100 percent of the radionuclide emissions reported in NEI were 210Po
or 222Rn, rather than other radioactive material or non-radioactive material. If the materials
reported as radionuclide emissions in NEI contain a large portion of non-radioactive material,
this assumption would be highly conservative. For the current analysis, this conservative
assumption was applied because the emissions are not speciated, and the proportion of the
NEl-reported emissions that is actually radioactive material is unknown.
Based on knowledge of the source category, some of the 210Po is likely to be physically bound
to solid matter and, therefore, increased PM controls probably result in lower emissions of this
G-3
-------
isotope. Because 222Rn is released in gas phase, however, emissions of this isotope are less
likely to be controlled. Therefore, the Maastricht facility model may overestimate the relative
contribution to overall radionuclide risk from 210Po and underestimate the relative contribution
from 222Rn. In both of these respects, the Maastricht model leads to conservative estimates of
emissions and risk.
G.2.2 AERMOD Modeling Approach for Estimating Inhalation Risk at Two California
Facilities
We applied a linear scaling approach for the California facilities to each emission value
generated by the three approaches detailed in Section G.2.1 above. The linear scaling
approach was developed for the two facilities by modeling emissions of 1.0 ton/yr of a pollutant
with a unit risk estimate (URE) of 1.0 cubic meter per microgram (m3/|jg). (The model results
were then converted into metric units for consistency.) AERMOD3 (EPA 2008b) estimates
ambient air concentrations using Gaussian dispersion equations, so the resulting model output
(in concentration units of jjg/m3) can be multiplied by the actual emissions in ton/yr and the
chemical-specific URE to derive the facility-specific ambient concentrations and risk estimates.
Because AERMOD estimates ambient concentrations resulting from the modeled source(s)
only, background exposures must be calculated independently. Therefore, the highest
estimated concentration at a location where people reside (assumed to be represented by
census block locations in this analysis) is the maximum incremental risk (MIR) estimate.
The MIR estimates were calculated by multiplying the AERMOD-scalable concentration result
by the emission rate (in Ci/yr) and the inhalation URE (in pCi/m3) for each radionuclide. The
inhalation URE was calculated by multiplying the inhalation slope factor (EPA 2001), reported in
risk/pCi, by an estimate of lifetime respiration by an exposed individual of 20 m3/day for 70
years, consistent with the approach EPA used to calculate inhalation UREs for non-radioactive
HAPs (EPA 2002). Therefore, the UREs presented here will lead to more conservative
estimates of incremental risk. The inhalation UREs are presented in Exhibit G-4.
Exhibit G-4. Inhalation Unit Risk Estimation for210Po and 222Rn
Unit
210Po
222Rn
Inhalation slope factor (risk/pCi)a
1.08E-08
1.8E-11
Inhalation URE (pCi/m3)"1 b
5.52E-03
9.20E-06
a EPA 2001.
b Assuming lifetime (70 years) respiration of 20 m3/day.
G.2.3 CAP88 Approach for Estimating Inhalation Risk at Two California Facilities
We used the radioactivity model, Clean Air Assessment Package - 1988 or CAP88 (EPA
2007c), as an alternative means of estimating risk from the two California facilities. The CAP88
model is designed to estimate doses and risks from radionuclide emissions to the air, is
comprised of databases and associated utility programs, and is EPA's preferred model for
assessing radioactivity risk (EPA 2007b).
3 AERMOD = AERMIC Model: AERMIC = AMS/EPA Regulatory Model improvement Committee; AMS = American
Meteorological Society
G-4
-------
Populations exposed to emissions from each facility were determined by using modified
versions of the default CAP88 population files for cities close to each location, rather than
developing site-specific population files for each site.4 Linear population-scaling factors were
applied to each population to develop modified CAP88 population files. These population files
are used by CAP88 to determine which individuals are exposed to radionuclide emissions. We
developed the scaling factors applied to each location, shown in Exhibit G-5, by comparing the
total populations within 70 kilometers (km) of each facility to the default population file in CAP88.
We then applied the scaling factors uniformly to the CAP88 default population files. Therefore,
the results of the modeling may be directionally inaccurate or may not reflect the risks as well as
they might if accurate population files had been developed. The wind direction at the sites is
predominantly west to east, and thus using the Los Angeles population file may lead to a more
conservative measure of risk because the bulk of the population density in Los Angeles is to the
east. The Berkeley population file has the bulk of the population to the south; therefore, using
the Berkeley file may not lead to similarly conservative assumptions. However, the most
important characteristic of the population file when estimating maximum individual risk is the
assumed location of the most exposed individual (i.e., for this analysis, the census block
centroid with the highest modeled concentration). Because the census block locations included
in the population files were not modified, the use of the modified population files based on urban
locations (e.g., Los Angeles) is likely to provide an overestimate of maximum individual risk
when used in sparsely-populated locations.
Exhibit G-5. Population Scaling Factors Applied to the CAP88 Default
Population Files Around Two California Portland Cement Facilities
NTI Site ID
City
CAP88 Default Population File
Scaling Factor
NEICA1505122
Monolith
Los Angeles
0.047
NEI2CA151186
Davenport
Berkeley
0.58
We assumed a mixing height of 643 m, based on the year-long average value of full-day mixing
heights for the Oakland, California meteorological station (EPA 2008c).
CAP88 estimates incremental cancer risks from modeled radionuclide emissions and does not
account for background radioactivity.
G.2.4 AERMOD Modeling Approach for Estimating Inhalation Risk at All U.S. Facilities
To obtain an approximation of total incremental cancer effects associated with radionuclide
emissions from the entire source category, radionuclide emissions were modeled using
AERMOD for all U.S. Portland cement facilities and the results were used to estimate MIR
inhalation cancer risk. Radionuclide emissions for each facility were estimated using the clinker
production scaling factor, detailed in Section G.2.1 above and presented in Exhibit G-3. Where
actual clinker production data were not available for a facility, clinker production was assumed
to equal 95 percent of clinker production capacity, based on the median actual production
relative to production capacity from all facilities having data.
4 CAP88 requires the use of a population file to estimate risk at the MIR location; however, the population data
included with the model cover selected locations only (and none of the locations of the two facilities modeled in this
analysis). Therefore, the existing CAP88 population files were modified to allow us to obtain an estimate of
maximum individual risk for the facilities included in this analysis.
G-5
-------
To facilitate AERMOD modeling of these facilities, stack parameters were compiled that were
assumed to be representative of radionuclide emission release points. We assumed that lead is
released along with radionuclides during the clinker heating, based on knowledge of the
production process. An analysis by ICF of all these facilities revealed that, at 91 percent of the
facilities, the largest release of lead compounds came from the tallest stack. Therefore, the
radionuclide emission source at each of the 91 Portland cement manufacturing facilities in the
United States was assumed to be the tallest stack at each site, usually the cement kiln.
Maximum incremental cancer risks from each facility were estimated by HEM3-AERMOD (EPA
2007a).
G.3 Results
G.3.1 Emissions
Exhibit G-6 presents the results of the two emission factor-based approaches and the NEI-
reported emissions of radionuclides. Using the assumptions described above, the emissions
reported in NEI are several orders of magnitude above those derived using data from the
Maastricht facility, particularly for the Lehigh facility (NEICA1505122). At the Lehigh facility, NEI
emissions are more than eight orders of magnitude higher than those predicted with the PM
emission-scaling factor approach. The emissions calculated with the PM emission scaling
factors are generally an order of magnitude lower than those produced using the clinker
production factors; however, these sets of emission values are much more comparable to one
another than either is to the NEI emissions. Although this difference does not necessarily mean
that the NEI emissions are incorrect, it does compel caution in their use and suggests that they
may not be correlated to actual emissions of radioactive isotopes in a useful way.
Exhibit G-6. Estimation of Radionuclide Emissions for the Two California Facilities
Using Three Approaches
NTI Site ID
Emissions, Based on NEI
Emissions and
Speciation Assumptions
Emissions, Based on
Clinker Production
Scaling Factors
Emissions, Based on PM
Emission Scaling
Factors
210Po
(Ci/yr)
222Rn
(Ci/yr)
210Po
(Ci/yr)
222Rn
(Ci/yr)
210Po
(Ci/yr)
222Rn
(Ci/yr)
NEICA1505122
3.48E+07
7.01 E+07
9.59E-01
1.93E+00
7.20E-02
1.45E-01
NEI2CA151186
6.02E+01
1.21E+02
8.13E-01
1.64E+00
1.03E-01
2.07E-01
The emission values derived for emissions from all 91 U.S. Portland cement manufacturing
facilities using the clinker production-based emission factors are presented in Exhibit G-7,
sorted by 210Po emissions from highest to lowest. The two case study facilities are highlighted
yellow.
Exhibit G-7. Estimation of Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
Emissions (short ton/yr)
Emissions (Ci/yr)
210Po
222Rn
210Po
222Rn
NEI22900
6.69945E-10
3.93935E-11
2.73068E+00
5.49638E+00
NE112018
5.27644E-10
3.10260E-11
2.15066E+00
4.32890E+00
G-6
-------
Exhibit G-7. Estimation of Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
Emissions (short ton/yr)
Emissions (Ci/yr)
210Po
222Rn
210Po
222Rn
NEITX139099J
4.97730E-10
2.92670E-11
2.02874E+00
4.08348E+00
NEI33394
4.21326E-10
2.47744E-11
1.71732E+00
3.45665E+00
NEI34931
4.10064E-10
2.41122E-11
1.67141E+00
3.36426E+00
NEI22838
3.79071E-10
2.22898E-11
1.54508E+00
3.10998E+00
NEI886
3.69776E-10
2.17432E-11
1.50720E+00
3.03372E+00
NEIAL1170004
3.68015E-10
2.16397E-11
1.50002E+00
3.01928E+00
NEI24859
3.67770E-10
2.16253E-11
1.49902E+00
3.01726E+00
NEIFL0860020
3.65493E-10
2.14914E-11
1.48974E+00
2.99859E+00
NEIAZ0250421
3.62856E-10
2.13363E-11
1.47900E+00
2.97695E+00
NEIAL8026
3.53766E-10
2.08018E-11
1.44195E+00
2.90238E+00
NEIKYR0060
3.35341 E-10
1.97184E-11
1.36684E+00
2.75121E+00
NEI20046
3.34850E-10
1.96895E-11
1.36484E+00
2.74718E+00
NEI18621
3.27234E-10
1.92417E-11
1.33380E+00
2.68470E+00
NEI7255
3.14459E-10
1.84905E-11
1.28173E+00
2.57989E+00
NEIT$FNP1408
3.09791E-10
1.82161E-11
1.26270E+00
2.54160E+00
NEIMIB1559
2.95276E-10
1.73625E-11
1.20354E+00
2.42251 E+00
NEIM00990002
2.90875E-10
1.71037E-11
1.18560E+00
2.38640E+00
NEITXT$11924
2.76380E-10
1.62514E-11
1.12652E+00
2.26748E+00
NEI2PRT14359
2.75934E-10
1.62252E-11
1.12470E+00
2.26383E+00
NEI22877
2.66553E-10
1.56736E-11
1.08647E+00
2.18686E+00
NEI52351
2.59183E-10
1.52402E-11
1.05643E+00
2.12639E+00
NEI20130
2.53778E-10
1.49224E-11
1.03440E+00
2.08205E+00
NEIVA2553
2.50410E-10
1.47244E-11
1.02067E+00
2.05442E+00
NEIIA0330060
2.36757E-10
1.39216E-11
9.65016E-01
1.94240E+00
NEICA1505122
2.35353E-10
1.38390E-11
9.59294E-01
1.93089E+00
NEI12238
2.30661 E-10
1.35631 E-11
9.40171E-01
1.89239E+00
NEI34326
2.28474E-10
1.34345E-11
9.31256E-01
1.87445E+00
NEI26277
2.27331 E-10
1.33673E-11
9.26596E-01
1.86507E+00
NEI2PA110039
2.24429E-10
1.31967E-11
9.14768E-01
1.84126E+00
NEIAL1150002
2.21104E-10
1.30012E-11
9.01216E-01
1.81399E+00
NEIPA94-2626
2.16789E-10
1.27474E-11
8.83627E-01
1.77858E+00
NEITXT$11872
2.13243E-10
1.25389E-11
8.69173E-01
1.74949E+00
NEIMIB1743
2.11643E-10
1.24448E-11
8.62651 E-01
1.73636E+00
NEI7376
2.08575E-10
1.22644E-11
8.50147E-01
1.71119E+00
NEISDT$8989
2.07498E-10
1.22011 E-11
8.45756E-01
1.70235E+00
NEI572
2.06035E-10
1.21150E-11
8.39793E-01
1.69035E+00
NEIAL321
2.01696E-10
1.18599E-11
8.22109E-01
1.65476E+00
NEI31319
2.01696E-10
1.18599E-11
8.22109E-01
1.65476E+00
NEITXRBG0259
2.01696E-10
1.18599E-11
8.22109E-01
1.65476E+00
NEI2CA151186
1.99485E-10
1.17299E-11
8.13097E-01
1.63662E+00
NEIUT10303
1.98994E-10
1.17010E-11
8.11094E-01
1.63259E+00
NEI40539
1.93844E-10
1.13982E-11
7.90105E-01
1.59034E+00
G-7
-------
Exhibit G-7. Estimation of Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
Emissions (short ton/yr)
Emissions (Ci/yr)
210Po
222Rn
210Po
222Rn
NEI12739
1.93343E-10
1.13688E-11
7.88063E-01
1.58623E+00
NEI13290
1.91624E-10
1.12677E-11
7.81054E-01
1.57212E+00
NEI32033
1.90641E-10
1.12099E-11
7.77048E-01
1.56406E+00
NEITN0653070
1.89412E-10
1.11377E-11
7.72042E-01
1.55398E+00
NEIIA0330035
1.84090E-10
1.08247E-11
7.50348E-01
1.51031E+00
NEISC0351244
1.78790E-10
1.05130E-11
7.28745E-01
1.46683E+00
NEIFLR001008
1.77847E-10
1.04576E-11
7.24902E-01
1.45910E+00
NEIALT$4449
1.75901E-10
1.03431 E-11
7.16967E-01
1.44313E+00
NEI42038
1.72953E-10
1.01698E-11
7.04951 E-01
1.41894E+00
NEIOK4013107
1.69022E-10
9.93866E-12
6.88930E-01
1.38669E+00
NEIPA58-1290
1.67605E-10
9.85538E-12
6.83157E-01
1.37507E+00
NEI22743
1.67056E-10
9.82309E-12
6.80919E-01
1.37057E+00
NEITN0930008
1.63863E-10
9.63530E-12
6.67901 E-01
1.34437E+00
NEIWA0331133
1.60349E-10
9.42869E-12
6.53579E-01
1.31554E+00
NEIOHT$6526
1.59662E-10
9.38831 E-12
6.50780E-01
1.30990E+00
NEIWV0030006
1.58295E-10
9.30790E-12
6.45206E-01
1.29868E+00
NEIGA1530003
1.57624E-10
9.26846E-12
6.42473E-01
1.29318E+00
NEI46744
1.57100E-10
9.23766E-12
6.40338E-01
1.28888E+00
NEI12976
1.53544E-10
9.02858E-12
6.25844E-01
1.25971E+00
NEI2PRT14367
1.52119E-10
8.94478E-12
6.20036E-01
1.24802E+00
NEI26327
1.48206E-10
8.71467E-12
6.04085E-01
1.21591E+00
NEI51352
1.47894E-10
8.69633E-12
6.02813E-01
1.21336E+00
NEI25375
1.45437E-10
8.55187E-12
5.92800E-01
1.19320E+00
NEINY0394192
1.43963E-10
8.46520E-12
5.86792E-01
1.18111 E+00
NEI51435
1.39787E-10
8.21962E-12
5.69769E-01
1.14684E+00
NEINYT$1163
1.39708E-10
8.21498E-12
5.69447E-01
1.14619E+00
NEINY4192600
1.38277E-10
8.13086E-12
5.63616E-01
1.13446E+00
NEIOK1826
1.36593E-10
8.03182E-12
5.56751 E-01
1.12064E+00
NEI338
1.35350E-10
7.95873E-12
5.51685E-01
1.11044E+00
NEIPAT$1626
1.34657E-10
7.91800E-12
5.48861 E-01
1.10476E+00
NEI51527
1.33645E-10
7.85847E-12
5.44735E-01
1.09645E+00
NEI33699
1.26029E-10
7.41066E-12
5.13693E-01
1.03397E+00
NEITXT$11980
1.14974E-10
6.76060E-12
4.68632E-01
9.43273E-01
NEI446
1.10185E-10
6.47897E-12
4.49110E-01
9.03978E-01
NEINMT$12442
1.06130E-10
6.24055E-12
4.32584E-01
8.70713E-01
NEIWA0331404
1.00960E-10
5.93658E-12
4.11513E-01
8.28301 E-01
NEIME0130002
9.34565E-11
5.49534E-12
3.80927E-01
7.66738E-01
NEI16357
8.05801E-11
4.73820E-12
3.28443E-01
6.61097E-01
NEI33444
7.61581E-11
4.47817E-12
3.10419E-01
6.24817E-01
NEIMT0430001
7.19817E-11
4.23260E-12
2.93396E-01
5.90553E-01
NEITXT$12011
7.00163E-11
4.11703E-12
2.85385E-01
5.74429E-01
NEIMT0310005
6.87879E-11
4.04480E-12
2.80378E-01
5.64351 E-01
G-8
-------
Exhibit G-7. Estimation of Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
Emissions (short ton/yr)
Emissions (Ci/yr)
210Po
222Rn
210Po
222Rn
NEIPA01993-1
6.82837E-11
4.01515E-12
2.78323E-01
5.60214E-01
NEIID0050004
6.38745E-11
3.75589E-12
2.60351 E-01
5.24040E-01
NEIPA23-0797
2.67405E-11
1.57237E-12
1.08994E-01
2.19385E-01
NEI22453
2.50585E-11
1.47346E-12
1.02138E-01
2.05585E-01
NEITX309123F
2.38301 E-11
1.40124E-12
9.71311E-02
1.95507E-01
G.3.2 Risk
Estimated incremental cancer risks by radionuclide using the three approaches described above
for the two California facilities are presented in Exhibit G-8 in table form. Note that the
estimated cancer risks associated with mass emissions as reported in NEI are above or close to
unity, indicating an extremely high probability that cancer could occur in the exposed individual.5
Estimated maximum individual risks calculated using emissions generated with the clinker
production- and PM emission-scaling factors range between approximately 2E-09 and 1E-05.
Exhibit G-8. Risk Calculated for Two California Portland Cement Facilities Using
AERMOD Modeling Results and Three Emission Estimation Approaches
NTI Site ID
Cone.
(Mg/m3)
MIR, Based on NEI
Emissions and
Speciation
Assumptions
MIR, Based on
Clinker Production
Scaling Factors
MIR, Based on PM
Emission Scaling
Factors
210Po
222Rn
210Po
222Rn
210Po
222Rn
NEICA1505122
1.53E-03
2.94E+02
9.86E-01
8.09E-06
2.71 E-08
6.07E-07
2.04E-09
NEI2CA151186
2.53E-03
8.43E-04
2.83E-06
1.14E-05
3.82E-08
1.44E-06
4.82E-09
Maximum incremental risk estimates from CAP88 for each radionuclide are presented in Exhibit
G-9. In general, these estimates tend to be lower than those calculated using the AERMOD
model, except (notably) for the Lehigh facility, where the calculated incremental risks are
approximately an order of magnitude higher. A summary of total risks (taking into account both
radioactive isotopes that are assumed to be emitted) estimated using each approach is
presented in Exhibit G-10 in chart form and Exhibit G-11 in table form.
5 Typically, if estimated cancer risks are calculated to be extremely high (e.g., > 0.01), alternate risk characterization
equations are appropriate. This adjustment has not been made in this case because the estimated risks are being
used primarily for evaluative purposes.
G-9
-------
Exhibit G-9. Risk for Two California Portland Cement Facilities Modeled
with CAP88 and Two Emission Estimation Approaches
NTI Site ID
MIR, Based on NEI Emissions
and Speciation Assumptions
MIR, Based on Clinker
Production Scaling Factors
210Po
222Rn
210Po
222Rn
NEICA1505122
3.95E+01
4.11E+00
1.09E-06
1.13E-07
NEI2CA151186
7.15E-05
8.06E-06
9.66E-07
1.09E-07
Exhibit G-10. Estimation of Maximum Incremental Risk from Estimated Radionuclide
Emissions for the Two California Facilities Using Three Approaches and Two Modeling
Systems
1.00E+03
1.00E+02
1.00E+01
1.00E+00
Certainty
1.00E-01
1.00E-02
o:
1.00E-03
1.00E-04
1.00E-05
1.00E-06
1 in 1 million
1.00E-07
~ LEHIGH SOUTHWEST CEMENT CO.
NEICA1505122
~ RMC PACIFIC MATERIALS
NEI2CA151186
AERMOD, NEI Emis. AERMOD, Clinker Prod. AERMOD, PM Emis. CAP88, NEI Emis. CAP88, Clinker Prod.
Calculation Method
G-10
-------
Exhibit G-11. Summary of Maximum Incremental Cancer Risk Modeled in AERMOD and
CAP88, Using Different Radionuclide Emission Estimates
NTI Site ID
Total MIR, Based on AERMOD Model and:
Total MIR, Based on CAP88
Model and:
NEI Emissions
and Speciation
Assumptions
Clinker
Production
Scaling Factors
PM Emission
Scaling
Factors
NEI Emissions
and Speciation
Assumptions
Clinker
Production
Scaling Factors
NEICA1505122
2.95E+02
8.12E-06
6.09E-07
4.36E+01
1.20E-06
NEI2CA151186
8.46E-04
1.14E-05
1.44E-06
7.96E-05
1.08E-06
Examination of the results presented in Exhibit G-12 indicate that the maximum incremental risk
estimates produced by both models, using non-NEI emission estimates, are near or above one
in a million population. Using NEI mass emission estimates for radionuclides appears to result
in unrealistically high maximum incremental risk estimates regardless of the dispersion model
used. In particular, the NEI emission estimates for the Monolith facility are extremely high,
indicating that the emissions reported in NEI may be incorrect, or that some of the assumptions
used to analyze the emissions are overly conservative.
Both models attributed higher risks from 210Po relative to risks from 222Rn, although the CAP88
model attributed a much higher proportion (approximately 10 percent) of risk to 222Rn compared
to those attributed to 222Rn by AERMOD (0.3 percent). This large differential between the
attribution of risk from each radionuclide by each model reflects large uncertainties introduced
by the assumptions detailed in the modeling approach and also reflects that CAP88 accounts
for daughter products whereas AERMOD does not.
The MIRs from all other emissions from these two California Portland cement facilities are 1E-
07 and 6E-08 for the Monolith and Davenport facilities, respectively. For each facility, the
maximum inhalation individual cancer risk for radionuclides predicted by the AERMOD modeling
of clinker-production scaled emissions are between 1 to 3 orders of magnitude higher than the
inhalation MIRs predicted by AERMOD modeling for all other facility emissions, and the CAP88
predicted risks from clinker production-scaled emissions are between 1 to 2 orders of magnitude
higher. This indicates that radionuclide emissions may be the primary driver for cancer risk from
Portland cement facilities. However, the extremely poor quality of available radionuclide
emissions data prompts caution in the interpretation of these risk values, especially when
comparing to better characterized risks.
Maximum incremental risks were estimated for each of the 91 Portland cement facilities in the
United States using the HEM-3 model as described in Section G.2.4 above. The results of the
HEM-3 modeling are presented in Exhibit G-12, listed from highest to lowest estimated risk.
In general, maximum incremental risks from each facility were approximately 300 times higher
for 210Po than for 222Rn. Of the 91 domestic Portland cement facilities, four are estimated to
have maximum cancer risk higher than 1E-04, or 1 person per 10,000. Approximately 35
percent of the facilities (32) are estimated to have maximum cancer risk higher than 1E-05, or 1
person per 100,000, and all but one facility had maximum cancer risk higher than 1E-06, or 1
person per 1 million. Only for two facilities was the cancer risk greater than 1E-06 due to 222Rn
emissions.
G-11
-------
Exhibit G-12. Estimation of Maximum Incremental Risk (MIR) for
Cancer from Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
MIR
210Po
222Rn
NEI22453
3.48E-04
1.17E-06
NEI22838
3.16E-04
1.06E-06
NEIFL0860020
1.92E-04
6.45E-07
NEI22743
1.56E-04
5.24E-07
NEI338
6.28E-05
2.11E-07
NEI22900
5.08E-05
1.71E-07
NEI7376
4.42E-05
1.49E-07
NEIT$FNP1408
4.39E-05
1.48E-07
NEI2PRT14367
3.48E-05
1.17E-07
NEIAZ0250421
3.14E-05
1.06E-07
NEIID0050004
3.05E-05
1.03E-07
NEI24859
2.60E-05
8.75E-08
NEI33699
2.46E-05
8.29E-08
NEI18621
1.96E-05
6.59E-08
NEINYT$1163
1.86E-05
6.27E-08
NEI32033
1.86E-05
6.25E-08
NEISDT$8989
1.74E-05
5.85E-08
NEINY0394192
1.73E-05
5.82E-08
NEI20046
1.71E-05
5.74E-08
NEI51435
1.54E-05
5.17E-08
NEISC0351244
1.46E-05
4.91 E-08
NEI12739
1.41E-05
4.76E-08
NEIIA0330060
1.40E-05
4.71 E-08
NEI446
1.35E-05
4.55E-08
NEITN0653070
1.26E-05
4.23E-08
NEIOK4013107
1.23E-05
4.13E-08
NEI2PRT14359
1.21E-05
4.06E-08
NEI2CA151186
1.14E-05
3.83E-08
NEITN0930008
1.11E-05
3.74E-08
NEITX139099J
1.08E-05
3.62E-08
NEINY4192600
1.06E-05
3.56E-08
NEI31319
9.75E-06
3.28E-08
NEIPAT$1626
9.30E-06
3.13E-08
NEITXRBG0259
9.17E-06
3.09E-08
NEIMT0430001
8.96E-06
3.02E-08
NEIKYR0060
8.91 E-06
3.00E-08
NEINMT$12442
8.90E-06
3.00E-08
NEI52351
8.89E-06
2.99E-08
NEIAL1150002
8.88E-06
2.99E-08
NEIOK1826
8.37E-06
2.82E-08
NEITXT$11924
8.29E-06
2.79E-08
G-12
-------
Exhibit G-12. Estimation of Maximum Incremental Risk (MIR) for
Cancer from Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
MIR
210Po
222Rn
NEIPA58-1290
8.28E-06
2.79E-08
NEICA1505122
8.09E-06
2.72E-08
NEIOHT$6526
8.03E-06
2.70E-08
NEIFLR001008
7.71 E-06
2.60E-08
NEI7255
7.53E-06
2.53E-08
NEIAL321
7.46E-06
2.51 E-08
NEI13290
7.15E-06
2.41 E-08
NEI886
7.00E-06
2.36E-08
NEITXT$11872
6.42E-06
2.16E-08
NEI12976
6.34E-06
2.14E-08
NEI2PA110039
6.29E-06
2.12E-08
NEIPA94-2626
6.18E-06
2.08E-08
NEI22877
5.80E-06
1.95E-08
NEIAL8026
5.48E-06
1.85E-08
NEI12018
5.26E-06
1.77E-08
NEIWA0331133
5.24E-06
1.76E-08
NEIWV0030006
4.97E-06
1.67E-08
NEIALT$4449
4.96E-06
1.67E-08
NEI51352
4.72E-06
1.59E-08
NEI34931
4.47E-06
1.50E-08
NEI40539
4.27E-06
1.44E-08
NEIGA1530003
4.18E-06
1.41 E-08
NEI12238
3.98E-06
1.34E-08
NEIAL1170004
3.91 E-06
1.32E-08
NEI46744
3.86E-06
1.30E-08
NEIIA0330035
3.70E-06
1.25E-08
NEIMIB1743
3.69E-06
1.24E-08
NEITXT$11980
3.56E-06
1.20E-08
NEI572
3.53E-06
1.19E-08
NEIMIB1559
3.41 E-06
1.15E-08
NEITXT$12011
3.39E-06
1.14E-08
NEIPA23-0797
3.34E-06
1.12E-08
NEITX309123F
3.29E-06
1.11 E-08
NEI51527
3.22E-06
1.09E-08
NEIM00990002
3.22E-06
1.08E-08
NEI33394
3.09E-06
1.04E-08
NEI16357
3.02E-06
1.02E-08
NEI42038
3.01 E-06
1.01E-08
NEIWA0331404
2.96E-06
9.97E-09
NEI33444
2.79E-06
9.39E-09
NEIVA2553
2.76E-06
9.29E-09
G-13
-------
Exhibit G-12. Estimation of Maximum Incremental Risk (MIR) for
Cancer from Radionuclide Emissions for All U.S. Portland Cement
Facilities, Based on the Clinker Production Scaling Factor
NTI Site ID
MIR
210Po
222Rn
NEI20130
2.48E-06
8.36E-09
NEI26327
2.28E-06
7.68E-09
NEI34326
2.00E-06
6.72E-09
NEIME0130002
1.77E-06
5.94E-09
NEI25375
1.53E-06
5.15E-09
NEI26277
1.52E-06
5.12E-09
NEIPA01993-1
1.27E-06
4.29E-09
NEIMT0310005
1.15E-06
3.89E-09
NEIUT10303
3.54E-07
1.19E-09
To provide context for the estimated incremental risks, background radiation risks can be
estimated using reported values for natural radiological background concentrations. Exhibit
G-13 shows the global background radiation dose by exposure route. Using the generic
radiation risk estimate developed by the U.S. Department of Energy (2002) of 0.06 cancer
mortality per sievert (Sv), the total cancer mortality risk from background radiation is calculated
to be 1.4E-04. This estimate of background radiation cancer mortality risk is higher than or
approximately equal to the MIR estimates from radionuclide emissions calculated with clinker
production and PM emission-scaling factors for the two California facilities. Note that the
background risks discussed here are from multiple pathways, including ingestion and dermal
exposure, although inhalation exposure accounts for half of the total background dose.
Exhibit G-13. Annual Effective Radiation Dose from Background Sources a
Exposure Route
Radiation Dose (mSv)b
Estimated MIR
Ingestion
0.3
2E-05
Inhalation
1.2
7.2E-05
Cosmic rays
0.4
2E-05
Terrestrial Gamma rays
0.5
3E-05
Total
2.4
1.4E-04
a United Nations 2000.
b mSv = millisievert
Radionuclide emissions from Portland cement facilities may represent relatively high
incremental cancer risks, as shown by this analysis. Comparing radionuclide risks to high
inhalation risks from emissions of other non-radioactive HAPs from these facilities may be
informative. According to an analysis performed by EPA (2008d), maximum individual lifetime
cancer risk from emissions of non-radioactive HAPs from domestic Portland cement facilities
exceed 1E-6 for only 8 of 91 facilities, compared to 90 facilities potentially exceeding the same
threshold due to radionuclide emissions. Similarly, the analysis showed that only 1 Portland
cement facility exceeded the 1E-5 threshold (to a level of 5E-5, the highest reported for all of the
facilities) due to non-radioactive HAP emissions, while 32 exceeded the same level due to
radionuclide emissions. A summary of the comparison is presented graphically in Exhibit G-14.
G-14
-------
No facilities were shown to exceed the 1E-4 level due to non-radioactive HAP emissions, while
four exceeded it due to radionuclide emissions. Radionuclide emissions may therefore be the
HAP emissions of greatest concern from Portland cement facilities.
Exhibit G-14. Comparison of Maximum Incremental Cancer Risk from Radionuclide
Emissions and Non-Radioactive HAP Emissions from Portland Cement Facilities
100
2
o
73
<1)
<1)
O
X
LU
<1)
.Q
E
90
80
70
60
50
40
30
20
10
¦ Radionuclides ~ Non-radioactive HAPs
1.00E-06
1.00E-05
Maximum Incremental Risk Threshold
1.00E-04
G.4 Uncertainties
Several factors, mostly related to the poor quality of existing data on radionuclide emissions
from Portland cement production facilities, contribute to uncertainties regarding the estimation of
their incremental cancer risks. First, the lack of direct measurements of radionuclides at U.S.
facilities available for this analysis makes evaluation of incremental cancer risks much more
difficult. Additional data of this type would serve as substantial evidence to support (or refute)
the claims and assumptions made in this analysis. Second, this analysis has relied heavily on
emissions from the Maastricht facility measured in 1990. Use of these facility data implicitly
assumes that U.S. and European Portland cement facilities have equivalent input materials,
levels of emission control, and emission profiles. Third, the analysis relies upon many
assumptions and model parameterizations. Although many of the assumptions are
conservative, what their overall effect is on the final risk estimates is not entirely clear. Finally,
the shortcomings in the model formulations for both CAP88 and AERMOD may lead to
inaccuracies in risk estimation. For instance, CAP88 has a static mixing height and may
therefore underestimate exposure during inversion events, because actual doses and resultant
risks may be higher during periods when the mixing height is low (i.e., during an inversion
event). The most important remedy to reduce these uncertainties would be to obtain actual
G-15
-------
measured, speciated radionuclide emission data from U.S. Portland cement manufacturing
facilities.
G.5 References
California Air Resources Board (CARB). 2007. Facility Search Engine, 2005 Criteria & Toxic
plus Risk Data, PM emissions. Available at:
http://www.arb.ca.gov/app/emsinv/facinfo/facinfo.php. (Accessed on April 3, 2008).
Chen, Q.; Degrange, J.P; Gerchikov, M.Y.; Hillis, Z.K.; Lepicard, S.; Meijne, E.I.M.; Smith, K.R.;
and van Weers, A. 2003. Effluent and Dose Control from European Union NORM Industries,
Assessment of Current Situation and Proposal for a Harmonised Community Approach.
Volume 1: Main Report. European Commission.
Leenhouts, H.P.; Stoop, P.; and van Tuinen, S.T. 1996. Non nuclear industries in the
Netherlands and radiological risks, Report No 610053003, National Institute of Public Health
and the Environment.
United Nations. 2000. Report of the United Nations Scientific Committee on the Effects of
Atomic Radiation to the General Assembly. Available at:
http://www.unscear.org/unscear/en/publications/2000_1.html.
U.S. Department of Energy. 2002. Memorandum: Radiation risk estimation from total effective
does equivalents (TEDEs), August 9. Available at:
www. doeal. gov/laso/N EPADocs/l scors_08092002. pdf.
U.S. Environmental Protection Agency (EPA). 1997. Exposure Factors Handbook. Available
at: http://www.epa.gov/ncea/efh/.
U.S. Environmental Protection Agency (EPA). 2001. Health Effects Assessment Summary
Tables (HEAST). Last updated April 16. Available at: http://www.epa.gov/radiation/heast/.
U.S. Environmental Protection Agency (EPA). 2002. A Review of the Reference Dose and
Reference Concentration Processes. Available at:
http://www.epa.gov/ncea/iris/RFD_FINAL%5B1%5D.pdf.
U.S. Environmental Protection Agency (EPA). 2007a. Clinker production estimates, dioxin
emission inventory spreadsheet.
U.S. Environmental Protection Agency (EPA). 2007b. HEM3, HEM-AERMOD application.
Available at: http://www.epa.gov/ttn/fera/data/hem/hem3_users_guide.pdf.
U.S. Environmental Protection Agency (EPA). 2007c. Radiation Risk Assessment Software:
Clean Air Act Assessment Package - 1988 (CAP88), version 3.0. Last updated December 9.
Available at: http://www.epa.gov/rpdweb00/assessment/CAP88/index.html.
U.S. Environmental Protection Agency (EPA). 2008a. 2002 National Emission Inventory (NEI),
CAP and HAP 2002-Based Platform, Version 3. Last updated January 17. Available at:
http://www.epa.gov/ttn/chief/emch/index.html.
G-16
-------
U.S. Environmental Protection Agency (EPA). 2008b. AERMOD. Last updated January 9.
Available at: http://www.epa.g0v/scramOOI/dispersion_prefrec.htm#aermod.
U.S. Environmental Protection Agency (EPA). 2008c. SCRAM Mixing Height Data. Last
updated December 28. Available at: http://www.epa.gov/scram001/mixingheightdata.htm.
U.S. Environmental Protection Agency (EPA). 2008d. Risk and Technology Review - Phase II,
Source Category Information Summary, Preliminary Draft. Portland Cement Manufacturing.
G-17
-------
Appendix H: Detailed assessment inputs and results
for Portland cement manufacturing facilities
-------
Table 1 - Facility Identification Information
Source Category
Facility NEI ID
Facility Name
Address
City
State
Portland Cement
PTC NEI12018
LAFARGE NORTH AMERICA - ALPENA
PLANT
1435 FORD AVE. P.O.
BOX 396
ALPENA
Ml
Portland Cement
PTC NEI12238
LAFARGE NORTH AMERICA INC.
301 EAST FRONT
STREET
BUFFALO
IA
Portland Cement
PTC NEI12739
MONARCH CEMENT COMPANY (THE)
S10-T26S-R18E
HUMBOLDT
KS
Portland Cement
PTC NEI12976
LONE STAR IND INC DBA BUZZI UNICEM
USA-PRYOR
E 5 Ml ON HWY20 THEN
S 2 Ml
PRYOR
OK
Portland Cement
PTC NEI13290
HUNTER PLANT
7781 F.M. 1102
NEW BRAUNFELS
TX
Portland Cement
PTC NEI16357
HEARTLAND CEMENT COMPANY
1765 LIMESTONE LANE
INDEPENDENCE
KS
Portland Cement
PTC NEI18621
ARIZONA PORTLAND CEMENT COMPANY
11115 N. CASA GRANDE
HIGHWAY
RILLITO
AZ
Portland Cement
PTC NEI20046
CALIFORNIA PORTLAND CEMENT CO.
9350 OAK CREEK ROAD
MOJAVE
CA
Portland Cement
PTC NEI20130
NATIONAL CEMENT CO
1 Ml. NOF HWY 138-1 Ml
E/l-5
LEBEC
CA
Portland Cement
PTC NEI22453
RIVERSIDE CEMENT CO UNIT NO.04
1500 RUBIDOUX BLVD
RIVERSIDE
CA
Portland Cement
PTC NEI22743
CALIFORNIA PORTLAND CEMENT CO
695 S RANCHO AV
COLTON
CA
Portland Cement
PTC NEI22838
MITSUBISHI CEMENT 2000
5808 STATE HIGHWAY
18
LUCERNE VALLEY
CA
Portland Cement
PTC NEI22877
TXI RIVERSIDE CEMENT COMPANY
19409 NATIONAL TRAILS
HIGHWAY
ORO GRANDE
CA
Portland Cement
PTC NEI22900
CEMEX - RIVER PLANT
16888 NORTH 'E' STREET
VICTORVILLE
CA
Portland Cement
PTC NEI24859
HANSON PERMANENTE CEMENT
24001 STEVENS CREEK
BLVD
CUPERTINO
CA
Portland Cement
PTC NEI25375
LEHIGH SOUTHWEST CEMENT CO.
15390 WONDERLAND
BLVD
REDDING
CA
Portland Cement
PTC NEI26277
RINKER MATERIALS CORPORATION.
1200 NW137TH AVE
MIAMI
FL
Portland Cement
PTC NEI26327
FLORIDA CRUSHED STONE CO., INC.
10311 CEMENT PLANT
ROAD
BROOKSVILLE
FL
Portland Cement
PTC NEI2CA151186
RMC PACIFIC MATERIALS
HIGHWAY ONE
DAVENPORT
CA
Portland Cement
PTC NEI2PA110039
LEHIGH CEMENT/EVANSVILLE CEMENT
PLT & QUARRY
537 EVANSVILLE RD
FLEETWOOD
PA
Portland Cement
PTC NEI2PRT14359
PUERTO RICAN CEMENT CO. INC.
STATE RD. 123 KM. 8.0
PONCE
PR
Portland Cement
PTC NEI2PRT14367
ESSROC SAN JUAN INC.
PR HAIGHWAY #2 KM 26. /
DORADO
PR
Portland Cement
PTC NEI31319
ESSROC CEMENT CORP.
HIGHWAY 31
SPEED
IN
Portland Cement
PTC NEI32033
LEHIGH CEMENT COMPANY
121 NORTH FIRST
STREET
MITCHELL
IN
Portland Cement
PTC NEI33394
LEHIGH PORTLAND CEMENT
117 MAIN STREET,
SOUTH
UNION BRIDGE
MD
Portland Cement
PTC NEI33444
ESSROC CEMENT
4120 BUCKEYSTOWN
PIKE
BUCKEYSTOWN
MD
1 of 4
-------
Table 1 - Facility Identification Information
Source Category
Facility NEI ID
Facility Name
Address
City
State
Portland Cement
PTC NEI33699
INDEPENDENT CEMENT/ST.
LAWERENCE
1260 SECURITY
ROAD, EXTENDED
HAGERSTOWN
MD
Portland Cement
PTC NEI338
MOUNTAIN CEMENT CO
PO BOX 339
LARAMIE
WY
Portland Cement
PTC NEI34326
LAFARGE NORTH AMERICA INC-
INDEPENDENCE PLANT
2200 N COURTNEY
ROAD
SUGAR CREEK
MO
Portland Cement
PTC NEI34931
LAFARGE BUILDING MATERIALS INC
RT 9W
COEYMANS
NY
Portland Cement
PTC NEI40539
ASH GROVE CEMENT COMPANY
33060 SHIRTTAIL CREEK
RD
DURKEE
OR
Portland Cement
PTC NEI42038
DEVIL'S SLIDE PLANT
6055 E. CROYDON RD.
MORGAN
UT
Portland Cement
PTC NEI446
CEMEX, INC. - LYONS CEMENT PLANT
5134 UTE HWY
LYONS AREA
CO
Portland Cement
PTC NEI46744
CEMEX INC/WAMPUM CEMENT PLT
2001 PORTLAND PARK
WAMPUM
PA
Portland Cement
PTC NEI51352
ILLINOIS CEMENT CO
1601 ROCKWELL RD
LASALLE
IL
Portland Cement
PTC NEI51435
LONE STAR INDUSTRIES INC
PORTLAND AVE
OGLESBY
IL
Portland Cement
PTC NEI51527
DIXON-MARQUETTE CEMENT INC
1914 WHITE OAK LN
DIXON
IL
Portland Cement
PTC NEI52351
LAFARGE MIDWEST INC
2500 PORTLAND RD
GRAND CHAIN
IL
Portland Cement
PTC NEI572
ASH GROVE CEMENT CO
16215 HIGHWAY 50
LOUISVILLE
NE
Portland Cement
PTC NEI7255
ESSROC/NAZARETH LOWER CEMENT
PLT 1
ROUTE 248 AND
EASTON RD
NAZARETH
PA
Portland Cement
PTC NEI7376
NORTH TEXAS CEMENT CO.
2 Ml. N.E. OF
MIDLOTHIAN, TX.
MIDLOTHIAN
TX
Portland Cement
PTC NEI886
HOLCIM (US) INC. PORTLAND PLANT
3500 HWY 120
FLORENCE, 3.8 Ml I
CO
Portland Cement
PTC NEIAL1150002
NATIONAL CEMENT CO OF ALABAMA
LOCATION ADDRESS IS
NEEDED
RAGLAND
AL
Portland Cement
PTC NEIAL1170004
LAFARGE BUILDING MATERIALS
8039 HWY 25
CALERA
AL
Portland Cement
PTC NEIAL321
CEMEX, INC.
1617 ARCOLA ROAD
DEMOPOLIS
AL
Portland Cement
PTC NEIAL8026
HOLCIM INC
3051 HAMILTON BLVD
THEODORE
AL
Portland Cement
PTC_NEIALT$4449
LEHIGH CEMENT COMPANY
8401 SECOND AVENUE
LEEDS, AL
AL
Portland Cement
PTC NEIAZ0250421
PHOENIX CEMENT CO.
3000 W. CEMENT PLANT
RD.
CLARKDALE
AZ
Portland Cement
PTC NEICA1505122
LEHIGH SOUTHWEST CEMENT CO.
13573 TEHACHAPI BLVD.
MONOLITH
CA
Portland Cement
PTC NEIFL0860020
TARMAC AMERICA LLC
11000 NW 121 WAY
MEDLEY
FL
Portland Cement
PTC NEIFLR001008
FLORIDA ROCK INDUSTRIES, INC.
CR 235, 2.5 MILES NE OF
CITY
NEWBERRY
FL
Portland Cement
PTC NEIGA1530003
CEMEX, INC.
2720 HWY 341 SOUTH
CLINCHFIELD
GA
Portland Cement
PTC NEIIA0330035
LEHIGH CEMENT COMPANY - MASON
CITY
700 25TH STREET NW
MASON CITY
IA
Portland Cement
PTC NEIIA0330060
HOLCIM (US) INC. - MASON CITY
1840 N. FEDERAL AVE
MASON CITY
IA
Portland Cement
PTC NEIID0050004
ASH GROVE CEMENT
230 CEMENT ROAD
INKOM
ID
Portland Cement
PTC NEIKYR0060
KOSMOS CEMENT CO
15301 DIXIE HIGHWAY
KOSMOSDALE
KY
2 of 4
-------
Table 1 - Facility Identification Information
Source Category
Facility NEI ID
Facility Name
Address
City
State
Portland Cement
PTC NEIME0130002
DRAGON PRODUCTS CO INC -
THOMASTON
US RT 1
THOMASTON
ME
Portland Cement
PTC NEIMIB1559
CEMEX, INC.
16000 BELLS BAY RD
CHARLEVOIX
Ml
Portland Cement
PTC NEIMIB1743
HOLCIM (US) INC.
15215 DAY RD
DUNDEE
Ml
Portland Cement
PTC NEIM00990002
RC CEMENT COMPANY INC-RIVER
CEMENT CO - SELMA PLAN
1000 RIVER CEMENT
ROAD
FESTUS
MO
Portland Cement
PTC NEIMT0310005
HOLCIM US INC - TRIDENT PLANT
4070 TRIDENT RD
THREE FORKS
MT
Portland Cement
PTC NEIMT0430001
ASH GROVE CEMENT
100 HIGHWAY 518
CLANCY
MT
Portland Cement
PTC_NEINMT$12442
GCC RIO GRANDE, INC. TIJERAS PLANT
11783 STATE HIGHWAY
14 S
TIJERAS
NM
Portland Cement
PTC NEINY0394192
GLENS FALLS LEHIGH CEMENT
COMPANY
120 ALPHA ROAD, OFF
ROUTE 9W
CATSKILL
NY
Portland Cement
PTC NEINY4192600
ST LAWRENCE CEMENT CORP-CATSKILL
QUARRY
RT 9W
CATSKILL
NY
Portland Cement
PTC_NEINYT$1163
GLENS FALLS LEHIGH CEMENT
COMPANY
313 WARREN ST
GLENS FALLS
NY
Portland Cement
PTC_NEIOHT$6526
CEMEX, INC.
3250 LINEBAUGH ROAD
XENIA
OH
Portland Cement
PTC NEIOK1826
HOLCIM US INC
1100 W18TH ST
ADA
OK
Portland Cement
PTC NEIOK4013107
LAFARGE BDLG MATERIALS
2609 N 145TH E AVE
TULSA
OK
Portland Cement
PTC NEIPA01993 1
ARMSTRONG CEMENT &
SUPPLY/WINFIELD
100 CLEARFIELD RD
CABOT
PA
Portland Cement
PTC NEIPA23 0797
LEHIGH CEMENT CO/YORK OPERATIONS
HOKES MILL RD
YORK
PA
Portland Cement
PTC NEIPA58 1290
LAFARGE CORP/WHITEHALL PLT
5160 MAIN ST
WHITEHALL
PA
Portland Cement
PTC NEIPA94 2626
HERCULES CEMENT CO
LP/STOCKERTOWN
501 CENTER ST
STOCKERTOWN
PA
Portland Cement
PTC_NEIPAT$1626
ESSROC/BESSEMER
SECOND ST
BESSEMER
PA
Portland Cement
PTC NEISC0351244
LAFARGE BUILDING MATERIALS
HARLEYVILLE
463 JUDGE ST
HARLEYVILLE
SC
Portland Cement
PTC_NEISDT$8989
GCC DACOTAH
501 N ST ONGE STREET
RAPID CITY
SD
Portland Cement
PTC_NEIT$FNP1408
BALCONES PLANT
AT THE INTERSECTION
OF WALD & SOLMS
ROADS
NEW BRAUNFELS
TX
Portland Cement
PTC NEITN0653070
SIGNAL MOUNTAIN CEMENT CO.
1201 SUCK CREEK ROAD
CHATTANOOGA
TN
Portland Cement
PTC NEITN0930008
CEMEX, INC.
6212 CEMENT PLANT
ROAD
KNOXVILLE
TN
Portland Cement
PTC NEITX139099J
HOLCIM (TEXAS) LP
1800 DOVE LN.
MIDLOTHIAN
TX
3 of 4
-------
Table 1 - Facility Identification Information
Source Category
Facility NEI ID
Facility Name
Address
City
State
Portland Cement
PTC NEITX309123F
LEHIGH PORTLAND CEMENT
100S WICKSON
WACO
TX
Portland Cement
PTC NEITXRBG0259
1604 PLANT
6055 W GREEN
MOUNTAIN ROAD
SAN ANTONIO
TX
Portland Cement
PTC_NEITXT$11872
PORTLAND CEMENT
11551 NACOGDOCHES
ROAD
SAN ANTONIO
TX
Portland Cement
PTC_NEITXT$11924
TEXAS LEHIGH CEMENT CO.
1000 JACK C. HAYS
TRAIL
BUDA
TX
Portland Cement
PTC_NEITXT$11980
MARYNEAL CEMENT PLANT
0.5 Ml. N.W. ON F.M. 608
MARYNEAL
TX
Portland Cement
PTC_NEITXT$12011
CEMEX CEMENT OF TEXAS L.P.
16501 W. MURPHY
ODESSA
TX
Portland Cement
PTC NEIUT10303
LEAMINGTON CEMENT PLANT
HWY132
LEAMINGTON
UT
Portland Cement
PTC NEIVA2553
ROANOKE CEMENT COMPANY
6071 CATAWBA ROAD
TROUTVILLE
VA
Portland Cement
PTC NEIWA0331133
ASH GROVE CEMENT CO, E MARGINAL
3801 E MARGINAL WAY S
SEATTLE
WA
Portland Cement
PTC NEIWA0331404
LAFARGE NORTH AMERICA INC
5400 W MARGINAL WAY
SW
SEATTLE
WA
Portland Cement
PTC NEIWV0030006
CAPITOL CEMENT CORPORATION
1826 SOUTH QUEEN
STREET
MARTINSBURG
WV
4 of 4
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Source Category
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
Portland Cement
PTC NEI12018
6.52E-07
4.95E-05
1.57E-02
Portland Cement
PTC NE112238
1.63E-07
7.84E-05
7.03E-03
Portland Cement
PTC NE112739
4.50E-07
1.04E-05
1.93E-02
Portland Cement
PTC NE112976
8.24E-09
4.13E-07
3.53E-04
Portland Cement
PTC NE113290
1.51E-07
5.03E-05
1.04E-02
Portland Cement
PTC NE116357
7.04E-09
3.17E-07
4.85E-03
Portland Cement
PTC NE118621
7.07E-08
1.06E-05
3.66E-03
Portland Cement
PTC NEI20046
3.65E-08
3.80E-06
2.01 E-03
Portland Cement
PTC NEI20130
2.21 E-08
2.17E-06
3.11E-03
Portland Cement
PTC NEI22453
5.11E-05
8.64E-04
3.89E-01
Portland Cement
PTC NEI22743
3.37E-07
3.27E-05
9.93E-03
Portland Cement
PTC NEI22838
2.16E-06
7.62E-06
7.16E-02
Portland Cement
PTC NEI22877
6.43E-07
4.66E-06
1.44E-01
Portland Cement
PTC NEI22900
1.09E-07
1.35E-05
6.75E-06
Portland Cement
PTC NEI24859
5.56E-08
6.25E-05
1.33E-08
Portland Cement
PTC NEI25375
5.76E-09
2.91 E-06
4.59E-04
Portland Cement
PTC NEI26277
1.85E-08
8.46E-05
8.53E-04
Portland Cement
PTC NEI26327
4.58E-08
1.86E-05
1.53E-03
Portland Cement
PTC NEI2CA151186
5.51 E-08
1.60E-05
2.25E-03
Portland Cement
PTC NEI2PA110039
2.91 E-08
2.04E-05
8.16E-03
Portland Cement
PTC NEI2PRT14359
1.11E-07
3.31 E-05
1.36E-02
Portland Cement
PTC NEI2PRT14367
6.57E-06
4.37E-04
1.36E-01
Portland Cement
PTC NEI31319
1.96E-07
7.47E-05
1.91E-02
Portland Cement
PTC NEI32033
3.69E-07
2.58E-05
2.86E-02
Portland Cement
PTC NEI33394
4.62E-08
6.56E-05
2.69E-03
Portland Cement
PTC NEI33444
6.80E-08
5.15E-05
5.35E-03
Portland Cement
PTC NEI33699
3.20E-08
2.80E-06
1.50E-03
Portland Cement
PTC NEI338
4.51 E-07
4.30E-06
6.46E-02
Portland Cement
PTC NEI34326
1.34E-08
3.17E-05
8.27E-04
Portland Cement
PTC NEI34931
1.52E-07
1.62E-04
7.44E-04
Portland Cement
PTC NEI40539
3.01 E-07
1.10E-06
2.86E-03
Portland Cement
PTC NEI42038
8.12E-09
1.11 E-06
7.46E-05
Portland Cement
PTC NEI446
2.89E-08
6.31 E-06
2.49E-03
Portland Cement
PTC NEI46744
6.98E-08
3.58E-05
8.09E-03
Portland Cement
PTC NEI51352
5.28E-08
8.42E-06
2.14E-03
Portland Cement
PTC NEI51435
2.50E-07
9.34E-06
9.22E-03
Portland Cement
PTC NEI51527
7.06E-07
1.22E-04
1.86E-01
1 of 3
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Source Category
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
Portland Cement
PTC NEI52351
8.43E-08
6.34E-06
2.40E-03
Portland Cement
PTC NEI572
9.72E-09
7.26E-06
1.12E-03
Portland Cement
PTC NEI7255
2.66E-08
2.34E-05
4.22E-03
Portland Cement
PTC NEI7376
9.78E-07
3.01 E-04
1.47E-02
Portland Cement
PTC NEI886
1.20E-07
1.33E-05
4.58E-04
Portland Cement
PTC NEIAL1150002
4.19E-06
7.22E-04
2.40E-03
Portland Cement
PTC NEIAL1170004
8.37E-09
5.14E-06
1.02E-05
Portland Cement
PTC NEIAL321
2.29E-07
7.29E-06
1.88E-01
Portland Cement
PTC NEIAL8026
1.79E-07
9.11E-05
6.25E-04
Portland Cement
PTC_NEIALT$4449
1.06E-08
8.27E-06
1.67E-04
Portland Cement
PTC NEIAZ0250421
1.72E-06
4.90E-05
2.95E-02
Portland Cement
PTC NEICA1505122
1.19E-07
5.25E-06
6.08E-04
Portland Cement
PTC NEIFL0860020
1.95E-06
7.53E-04
6.90E-02
Portland Cement
PTC NEIFLR001008
2.34E-08
2.80E-06
7.64E-04
Portland Cement
PTC NEIGA1530003
8.94E-09
1.31E-06
3.57E-04
Portland Cement
PTC NEIIA0330035
1.59E-07
2.27E-05
4.14E-03
Portland Cement
PTC NEIIA0330060
2.59E-07
2.64E-05
5.66E-04
Portland Cement
PTC NEIID0050004
3.24E-07
3.79E-06
1.71E-03
Portland Cement
PTC NEIKYR0060
1.90E-08
1.96E-05
4.39E-04
Portland Cement
PTC NEIME0130002
6.85E-08
8.11E-06
4.14E-05
Portland Cement
PTC NEIMIB1559
1.72E-07
2.43E-05
7.67E-03
Portland Cement
PTC NEIMIB1743
1.59E-06
1.59E-03
8.94E-02
Portland Cement
PTC NEIM00990002
4.18E-09
3.67E-06
1.61E-03
Portland Cement
PTC NEIMT0310005
1.46E-08
4.20E-07
2.37E-05
Portland Cement
PTC NEIMT0430001
9.54E-08
6.39E-07
4.72E-04
Portland Cement
PTC_NEINMT$12442
2.79E-08
5.39E-06
9.47E-04
Portland Cement
PTC NEINY0394192
1.88E-07
1.12E-05
2.78E-05
Portland Cement
PTC NEINY4192600
2.41 E-07
1.56E-05
3.67E-03
Portland Cement
PTC_NEINYT$1163
3.15E-07
1.38E-05
6.83E-02
Portland Cement
PTC_NEIOHT$6526
2.50E-08
2.01 E-05
7.66E-04
Portland Cement
PTC NEIOK1826
8.96E-08
2.89E-06
1.22E-02
Portland Cement
PTC NEIOK4013107
3.71 E-08
1.19E-05
8.31 E-04
Portland Cement
PTC NEIPA01993-1
1.50E-07
7.97E-05
1.25E-02
Portland Cement
PTC NEIPA23-0797
1.06E-07
4.10E-05
6.46E-03
Portland Cement
PTC NEIPA58-1290
8.81 E-08
5.38E-05
8.79E-03
Portland Cement
PTC NEIPA94-2626
1.57E-07
1.08E-04
1.27E-02
Portland Cement
PTC_NEIPAT$1626
2.99E-07
7.64E-05
3.79E-02
2 of 3
-------
Table 2 - Maximum Predicted HEM-3 Chronic Risks
Source Category
Facility NEI ID
Chronic Risk 1
Cancer MIR
Cancer Incidence
Noncancer Max HI
Portland Cement
PTC NEISC0351244
9.11E-07
5.53E-05
1.95E-02
Portland Cement
PTC_NEISDT$8989
9.70E-08
1.02E-05
9.21 E-03
Portland Cement
PTC_NEIT$FNP1408
1.76E-07
7.96E-05
5.71 E-02
Portland Cement
PTC NEITN0653070
6.22E-08
9.34E-06
1.73E-03
Portland Cement
PTC NEITN0930008
3.87E-08
5.04E-06
7.51 E-04
Portland Cement
PTC NEITX139099J
3.00E-07
5.81 E-04
1.63E-03
Portland Cement
PTC NEITX309123F
3.50E-08
3.09E-06
2.14E-03
Portland Cement
PTC NEITXRBG0259
5.85E-08
6.11E-05
3.77E-03
Portland Cement
PTC_NEITXT$11872
9.89E-08
1.75E-04
8.07E-04
Portland Cement
PTC_NEITXT$11924
1.77E-08
1.19E-05
4.53E-04
Portland Cement
PTC_NEITXT$11980
7.61 E-09
1.46E-07
1.32E-02
Portland Cement
PTC_NEITXT$12011
7.24E-09
8.47E-07
3.17E-05
Portland Cement
PTC NEIUT10303
7.58E-10
3.03E-08
1.97E-06
Portland Cement
PTC NEIVA2553
9.61 E-09
4.10E-06
1.90E-04
Portland Cement
PTC NEIWA0331133
3.01 E-08
1.05E-04
7.21 E-05
Portland Cement
PTC NEIWA0331404
3.17E-08
1.13E-04
1.01 E-04
Portland Cement
PTC NEIWV0030006
5.29E-08
1.49E-05
1.89E-04
1 BOLD indicates a cancer risk great than 1 in a million or a noncancer risk greater than 1
3 of 3
-------
Table 3 - Maximum Predicted Acute Risks (HEM-AERMOD)
Source Category
Facility NEI ID
Maximum Hazard C
Quotient1
REL
AEGL1
AEGL2
ERPG1
ERPG2
Portland Cement
PTC_NEI12018
6.86E-02
5.34E-02
4.37E-03
5.34E-02
4.37E-03
Portland Cement
PTC_NEI12238
8.82E-03
6.86E-03
5.61 E-04
6.86E-03
5.61 E-04
Portland Cement
PTC_NEI12739
1.69E-02
4.63E-03
3.78E-04
4.63E-03
3.78E-04
Portland Cement
PTC_NEI12976
1.72E-03
1.33E-03
1.09E-04
1.33E-03
1.09E-04
Portland Cement
PTC_NEI13290
1.05E-02
8.19E-03
6.70E-04
8.19E-03
6.70E-04
Portland Cement
PTC_NEI16357
5.47E-03
4.25E-03
3.48E-04
4.25E-03
3.48E-04
Portland Cement
PTC_NEI18621
1.67E-01
1.43E-02
1.12E-03
1.43E-02
1.12E-03
Portland Cement
PTC_N El 20046
5.11E-02
3.98E-02
3.25E-03
3.98E-02
3.25E-03
Portland Cement
PTC_NEI20130
5.09E-01
1.19E-02
7.71 E-04
1.19E-02
7.71 E-04
Portland Cement
PTC_NEI22453
3.75E+00
3.14E-01
2.03E-02
3.14E-01
3.16E-02
Portland Cement
PTC_NEI22743
1.14E-01
9.70E-03
6.28E-04
9.70E-03
6.28E-04
Portland Cement
PTC_NEI22838
3.74E-01
1.13E-01
9.26E-03
1.13E-01
9.26E-03
Portland Cement
PTC_NEI22877
1.22E+00
1.05E-01
6.76E-03
1.05E-01
6.76E-03
Portland Cement
PTC_NEI22900
5.31 E-05
0.00E+00
8.03E-11
0.00E+00
3.64E-08
Portland Cement
PTC_NEI24859
2.56E-07
1.96E-09
2.51E-10
1.96E-09
5.95E-10
Portland Cement
PTC_NEI25375
1.50E-02
6.86E-03
5.62E-04
6.86E-03
5.62E-04
Portland Cement
PTC_NEI26277
6.12E-03
4.76E-03
3.89E-04
4.76E-03
3.89E-04
Portland Cement
PTC_NEI26327
3.94E-03
3.06E-03
2.51 E-04
3.06E-03
2.51 E-04
Portland Cement
PTC_NEI2CA151186
2.20E-01
6.73E-03
4.94E-04
6.73E-03
4.94E-04
Portland Cement
PTC_NEI2PA110039
5.83E-03
4.53E-03
3.71 E-04
4.53E-03
3.71 E-04
Portland Cement
PTC_NEI2PRT14359
6.33E-02
1.25E-02
1.02E-03
1.25E-02
1.02E-03
Portland Cement
PTC_NEI2PRT14367
1.02E+00
7.91 E-01
6.47E-02
7.91 E-01
6.47E-02
Portland Cement
PTC_NEI31319
4.01 E-02
6.48E-03
5.30E-04
6.48E-03
5.30E-04
Portland Cement
PTC_NEI32033
3.54E-02
5.72E-03
4.68E-04
5.72E-03
4.68E-04
Portland Cement
PTC_NEI33394
1.06E-02
8.24E-03
6.74E-04
8.24E-03
6.74E-04
Portland Cement
PTC_NEI33444
5.93E-03
2.59E-05
6.27E-06
2.59E-05
5.33E-06
Portland Cement
PTC_NEI33699
9.80E-03
7.62E-03
6.24E-04
7.62E-03
6.24E-04
Portland Cement
PTC_NEI338
1.12E+00
9.58E-02
6.20E-03
9.58E-02
6.20E-03
Portland Cement
PTC_NEI34326
3.59E-02
3.07E-03
1.99E-04
3.07E-03
1.99E-04
Portland Cement
PTC_NEI34931
1.40E-01
0.00E+00
5.30E-05
0.00E+00
4.50E-05
Portland Cement
PTC_NEI40539
3.56E-01
6.94E-04
3.77E-04
6.94E-04
3.20E-04
Portland Cement
PTC_NEI42038
2.01 E-02
0.00E+00
2.12E-05
0.00E+00
1.81 E-05
Portland Cement
PTC_NEI446
7.76E-01
5.18E-03
8.22E-04
5.18E-03
6.98E-04
Portland Cement
PTC_NEI46744
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_NEI51352
6.79E-03
5.28E-03
4.32E-04
5.28E-03
4.32E-04
Portland Cement
PTC_NEI51435
5.68E-02
2.01 E-02
1.65E-03
2.01 E-02
1.65E-03
Portland Cement
PTC_NEI51527
5.39E-02
9.72E-04
7.95E-05
9.72E-04
7.95E-05
Portland Cement
PTC_NEI52351
6.85E-03
5.33E-03
4.36E-04
5.33E-03
4.36E-04
Portland Cement
PTC_NEI572
5.94E-02
5.07E-03
3.28E-04
5.07E-03
3.28E-04
Portland Cement
PTC_NEI7255
3.38E-02
3.62E-03
2.96E-04
3.62E-03
2.96E-04
Portland Cement
PTC_NEI7376
6.28E-02
5.37E-03
3.47E-04
5.37E-03
3.47E-04
Portland Cement
PTC_NEI886
2.47E-02
2.53E-03
2.07E-04
2.53E-03
2.07E-04
Portland Cement
PTC_NEIAL1150002
1.87E-02
1.45E-02
1.19E-03
1.45E-02
1.19E-03
Portland Cement
PTC_NEIAL1170004
5.71 E-04
0.00E+00
6.05E-07
0.00E+00
5.14E-07
Portland Cement
PTC_NEIAL321
3.57E-01
1.20E-02
3.10E-03
1.20E-02
3.10E-03
Portland Cement
PTC_NEIAL8026
3.86E-03
3.12E-04
2.55E-05
3.12E-04
2.55E-05
Portland Cement
PTC_NEIALT$4449
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_NEIAZ0250421
3.19E+01
0.00E+00
9.91 E-06
0.00E+00
8.42E-06
Portland Cement
PTC_NEICA1505122
8.63E-02
2.12E-03
1.73E-04
2.12E-03
1.73E-04
Portland Cement
PTC_N EIFL0860020
4.69E-02
5.94E-03
4.86E-04
5.94E-03
4.86E-04
Portland Cement
PTC_N EIFLR001008
1.08E-02
1.34E-03
1.10E-04
1.34E-03
1.10E-04
Portland Cement
PTC_NEIGA1530003
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_N E11A0330035
6.12E-03
2.28E-03
1.86E-04
2.28E-03
1.86E-04
1 of 2
-------
Table 3 - Maximum Predicted Acute Risks (HEM-AERMOD)
Source Category
Facility NEI ID
Maximum Hazard C
Quotient1
REL
AEGL1
AEGL2
ERPG1
ERPG2
Portland Cement
PTC_NEIIA0330060
2.24E-02
0.00E+00
3.77E-06
0.00E+00
9.39E-06
Portland Cement
PTC_N E11D0050004
3.55E-03
2.76E-03
2.26E-04
2.76E-03
2.26E-04
Portland Cement
PTC_NEIKYR0060
8.72E-03
6.78E-03
5.55E-04
6.78E-03
5.55E-04
Portland Cement
PTC_NEIME0130002
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_NEIMIB1559
5.93E-02
4.62E-02
3.78E-03
4.62E-02
3.78E-03
Portland Cement
PTC_NEIMIB1743
2.41 E+00
6.64E-03
1.99E-03
6.64E-03
1.99E-03
Portland Cement
PTC_N EIM00990002
8.90E-03
6.93E-03
5.67E-04
6.93E-03
5.67E-04
Portland Cement
PTC_NEIMT0310005
1.91E-02
0.00E+00
2.02E-05
0.00E+00
1.72E-05
Portland Cement
PTC_N EIMT04 30001
5.16E-02
0.00E+00
5.46E-05
0.00E+00
4.64E-05
Portland Cement
PTC_NEINMT$12442
6.64E-03
5.16E-03
4.23E-04
5.16E-03
4.23E-04
Portland Cement
PTC_NEINY0394192
2.18E-04
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_NEINY4192600
3.95E-02
0.00E+00
4.18E-05
0.00E+00
3.55E-05
Portland Cement
PTC_NEINYT$1163
1.95E-02
1.67E-03
1.08E-04
1.67E-03
1.08E-04
Portland Cement
PTC_NEIOHT$6526
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
Portland Cement
PTC_NEIOK1826
1.50E-03
2.10E-04
5.44E-05
2.10E-04
5.44E-05
Portland Cement
PTC_NEIOK4013107
1.77E-01
1.51E-02
9.77E-04
1.51E-02
9.77E-04
Portland Cement
PTC_NEIPA01993-1
7.27E-05
5.65E-05
4.62E-06
5.65E-05
2.13E-04
Portland Cement
PTC_NEIPA23-0797
0.00E+00
0.00E+00
0.00E+00
0.00E+00
1.14E-04
Portland Cement
PTC_NEIPA58-1290
1.34E-03
1.04E-03
8.50E-05
1.04E-03
8.50E-05
Portland Cement
PTC_NEIPA94-2626
3.80E-03
7.21 E-04
5.90E-05
7.21 E-04
5.90E-05
Portland Cement
PTC_NEIPAT$1626
1.53E-02
2.26E-03
1.85E-04
2.26E-03
1.85E-04
Portland Cement
PTC_NEISC0351244
7.11E-02
5.34E-02
4.37E-03
5.34E-02
4.37E-03
Portland Cement
PTC_NEISDT$8989
1.88E-06
0.00E+00
1.99E-09
0.00E+00
1.69E-09
Portland Cement
PTC_NEIT$FNP1408
9.47E-02
1.33E-02
3.43E-03
1.33E-02
3.43E-03
Portland Cement
PTC_NEITN0653070
3.04E-01
2.60E-02
1.68E-03
2.60E-02
1.68E-03
Portland Cement
PTC_NEITN0930008
4.07E-02
0.00E+00
4.31 E-05
0.00E+00
3.66E-05
Portland Cement
PTC_NEITX139099J
5.55E-03
2.39E-03
1.96E-04
2.39E-03
1.96E-04
Portland Cement
PTC_NEITX309123F
5.05E-03
3.93E-03
3.22E-04
3.93E-03
3.22E-04
Portland Cement
PTC_NEITXRBG0259
1.42E-03
1.99E-04
5.13E-05
1.99E-04
5.13E-05
Portland Cement
PTC_NEITXT$11872
5.38E-03
1.84E-03
1.51 E-04
1.84E-03
1.51 E-04
Portland Cement
PTC_NEITXT$11924
2.65E-03
2.06E-03
1.69E-04
2.06E-03
1.69E-04
Portland Cement
PTC_NEITXT$11980
1.02E-01
1.43E-02
3.70E-03
1.43E-02
3.70E-03
Portland Cement
PTC_NEITXT$12011
1.40E-02
0.00E+00
1.49E-05
0.00E+00
1.26E-05
Portland Cement
PTC_NEIUT10303
4.07E-03
0.00E+00
4.30E-06
0.00E+00
3.66E-06
Portland Cement
PTC_NEIVA2553
3.88E-03
3.01 E-03
2.47E-04
3.01 E-03
2.47E-04
Portland Cement
PTC_NEIWA0331133
2.97E-02
0.00E+00
3.15E-05
0.00E+00
2.68E-05
Portland Cement
PTC_NEIWA0331404
2.99E-02
0.00E+00
3.16E-05
0.00E+00
2.69E-05
Portland Cement
PT C_N EIWV0030006
0.00E+00
0.00E+00
0.00E+00
0.00E+00
0.00E+00
1 Some maximum acute impacts may be at onsite locations.
Note: BOLD indicates acute risks greater than 1
2 of 2
-------
Table 4 - Maximum Predicted Acute Risks Greater than 1 (Refined Approach)
Facility NEI ID
Pollutant
Criteria
HEM-3
(Screening)
HEM-3/
AERMOD
(Refined)1
Refined Modeling Approach 2
Portland Cement
PTC NEIAZ0250421
Nickel compounds
REL
3.19E+01
To be completedas part of RTR anlaysis
PTC NEI22453
Mercury (elemental
REL
3.75E+00
To be completedas part of RTR anlaysis
PTC NEI22453
Formaldehyde
REL
3.68E+00
To be completedas part of RTR anlaysis
PTC NEIMIB1743
Acrolein
REL
2.41 E+00
To be completedas part of RTR anlaysis
1 Facilites with a HEM-3 screening acute value greater than 1 were remodeled with a more refined approach
2 Indicates modeling technique used to refined estimates
1 of 1
-------
APPENDIX I: Ravena Human Health Risk Assessment
-------
[This page intentionally left blank.]
-------
TABLE OF CONTENTS
1-1 Introduction 1
I-2 Conceptual Model for Potential Exposures/Risks 1
1-2.1 Selection of HAPs for this Analysis 1
I-2.2 Selection of Relevant Exposure Pathways and Approach to Exposure Assessment 2
I-2.3 Overview of Modeling Approach 5
I-2.4 Selection of Facility for Analysis 6
I-3 Fate and Transport Modeling (TRIM.FaTE) 7
1-3.1 Source Characterization 8
I-3.2 Relevant Meteorological Data 8
I-3.3 Extent and Dimensions of Modeled Environment 11
I-3.4 TRIM.FaTE Parcel Design 13
1-3.4.1 Surface Parcel Layout 13
I-3.4.2 Modeled Water Bodies 14
I-3.4.3 Modeled Agricultural Parcels 16
I-3.4.4 Air Parcel Layout 17
I-3.5 Abiotic Environment 19
1-3.5.1 Soil and Watershed Characteristics 19
1-3.5.1.1 Soil Properties 19
1-3.5.1.2 Erosion 19
1-3.5.1.3 Runoff 20
I-3.5.2 Water Body Characteristics 21
1-3.5.2.1 Surface Water and Sediment Properties 21
I-3.5.2.2 Water Transfers 21
I-3.5.2.3 Sediment 22
I-3.6 Terrestrial Plants 22
I-3.7 Aquatic Ecosystem 23
1-3.7.1 Collection of Information on Species Present in Water Bodies 24
I-3.7.2 Creation of Food Webs 24
I-3.7.3 Parameterization of Fish Compartments to be Included in Application 26
I-3.8 Mass Balance Results 28
I-4 Exposure Assessment 29
1-4.1 Approach 29
I-4.2 Exposure Pathways 31
1-4.2.1 Farm Food Chain Media Pathway 33
I-4.2.2 Fish Consumption Pathway 34
I-4.2.3 Breast Milk Pathway 35
I-4.3 Exposure Dose Estimation 35
1-4.3.1 Media Concentrations 35
/-/
-------
1-4.3.2 Exposure Factors 36
1-5 Dose-Response Assessment and Estimation of Human Health Risks 41
I-6 Results and Discussion 41
1-6.1 Ravena Human Health Multipathway Risk Assessment Results 41
I-6.2 2,3,7,8-TCDD 42
1-6.2.1 Estimated Media Concentrations 42
I-6.2.2 Comparison of Modeled Surface Water Concentrations to Measured Values ....46
I-6.2.3 2,3,7,8-TCDD Risk Assessment Results 47
1-6.2.3.1 2,3,7,8-TCDD Estimated Lifetime Cancer Risks 48
I-6.2.3.2 2,3,7,8-TCDD Chronic Non-Cancer Hazard Quotients 51
I-6.2.3.3 2,3,7,8-TCDD Risks and Hazard Quotients Resulting from Dermal
Exposure 53
I-6.2.3.4 Chronic Non-Cancer Hazard Quotients in Nursing Infants 54
I-6.3 Mercury 56
I-6.3.1 Mercury Media Concentrations 56
I-6.3.2 Mercury Risk Assessment Results 63
I-6.3.2.1 Mercury Chronic Non-cancer Hazard Quotients 63
I-6.3.2.2 Mercury Chronic Non-Cancer Hazard Quotients from Dermal Exposure 67
I-6.4 Alternate Modeling Scenario - Incorporation of Fish Harvesting from Ravena Pond 68
I-7 References 71
Attachment 1-1 TRIM.FaTE Inputs for the Ravena Screening Scenario
Attachment I-2 Detailed Ravena Human Health Assessment Exposure, Risk, and
Hazard Quotient Estimates
Hi
-------
LIST OF EXHIBITS
Exhibit 1.2-1. Conceptual Exposure Model for Farmer Scenario 4
Exhibit I.2-2. Fish Consumption Exposure Pathway 5
Exhibit 1.2-3. Location of the Ravena Facility 6
Exhibit 1.3-1. Emissions of Dioxins and Mercury from the Lafarge Facility in Ravena, NY, and Screening
Results 8
Exhibit 1.3-2. Comparison of Historical and Modeled Temperature and Precipitation 9
Exhibit 1.3-3. Frequency Distribution of Wind Direction and Coincidental Wind Speed (2001-2003 Albany
dataset, all time periods) 10
Exhibit 1.3-4. Frequency Distribution of Wind Direction and Speed During Rain Events (2001-2003 Albany
dataset, hours with precipitation only) 11
Exhibit 1.3-5. Frequency Distribution of Calculated Morning/Afternoon Mixing Height (m) 11
Exhibit 1.3-6. Streams, Rivers, and Water Bodies of the Middle Hudson Sub-basin 12
Exhibit 1.3-7. Land Use in Region of the Ravena Facility 13
Exhibit 1.3-8. Overall Modeling Region 14
Exhibit 1.3-9. Modeled and Reported Location of the Ravena Facility 15
Exhibit 1.3-10. Water Bodies Included in the Modeled Region 16
Exhibit 1.3-11. Agricultural Parcels Included in the TRIM.FaTE Scenario 16
Exhibit 1.3-12. Surface Parcel Layout with Water Bodies and Land Use 17
Exhibit 1.3-13. Air Parcel Layout 18
Exhibit 1.3-14. Air and Surface Parcel Layouts (Overlay) 18
Exhibit 1.3-15. Soil Compartment Depths 19
Exhibit 1.3-16. Selected Properties of Soil and Groundwater 19
Exhibit 1.3-17. Selected Surface Water and Sediment Properties 21
Exhibit 1.3-18. Turnover Rates for Ravena Water Bodies 21
Exhibit 1.3-19. Sediment Total Suspended Solids and Burial Rates for Ravena Water Bodies 22
Exhibit 1.3-20. Surface Parcel Layout with Plant Types and Relevant Land Use 23
Exhibit 1.3-21. Aquatic Food Webs for Modeled Water Bodies 27
Exhibit 1.3-22. Distribution of Chemical Mass in Ravena, NY Scenario 29
Exhibit 1.4-1. Ingestion Exposure Scenarios 30
Exhibit 1.4-2. Summary of Ingestion Exposure Pathways and Routes of Uptake 32
Exhibit 1.4-3. Fish Species Assumed to be Consumed in this Assessment 34
Exhibit 1.4-4. Body Weight Estimates Used in This Assessment 37
Exhibit 1.4-5. Age-Specific Ingestion Rates for the FFC Pathway 38
Exhibit 1.4-6. Fish Ingestion Rates for all Scenarios 39
Exhibit 1.4-7. Breast Milk Ingestion Rates for Infants Less Than 1 Year of Age 40
Exhibit 1.5-1. Dose-response Values for PB-HAPs Addressed in this Assessment 41
Exhibit 1.6-2. 2,3,7,8-TCDD Media Concentration Time Series Using 95% UCL Dioxin Emission Rate...43
l-iii
-------
Exhibit 1.6-3. 2,3,7,8-TCDD Air and Surface Soil Concentrations and Dry Particle Deposition Rates
During the 50th Model Year Using 95-Percent UCL Emission Rate 44
Exhibit 1.6-4. 2,3,7,8-TCDD Surface Water Concentrations During the 50th Model Year Using Mean and
95-Percent UCL Emission Rates 45
Exhibit 1.6-5. 2,3,7,8-TCDD Concentration in Fish Species During the 50th Model Year Using the
95-Percent UCL Emission Rate 46
Exhibit 1.6-6. Modeled 2,3,7,8-TCDD Concentrations Compared to Measured Values 47
Exhibit 1.6-7. 2,3,7,8-TCDD Individual Lifetime Cancer Risks 48
Exhibit 1.6-8. Pathway Contributions to 2,3,7,8-TCDD Individual Lifetime Cancer Risks 49
Exhibit 1.6-9. 2,3,7,8-TCDD Individual Lifetime Cancer Risks 50
Exhibit 1.6-10. 2,3,7,8-TCDD Chronic Non-cancer Hazard Quotients (95th Percentile UCL Emission
Factor, RME Ingestion Rates) 52
Exhibit 1.6-11. Pathway Contributions to Divalent Mercury Chronic Non-Cancer Hazard Quotients (95th
Percentile UCL Emission Factor, RME Ingestion Rates) 53
Exhibit 1.6-12. Estimated Lifetime Cancer Risks Associated with Modeled Dermal Exposure to 2,3,7,8-
TCDD 54
Exhibit 1.6-13. Mother and Infant non-cancer Hazard Quotients for 2,3,7,8-TCDD 55
Exhibit 1.6-14. Total Mercury Media Concentration Time Series 57
Exhibit 1.6-15. Mercury Surface Soil Concentrations at 50th Model Year 58
Exhibit 1.6-16. Total Mercury Air and Surface Soil Concentrations and Dry Particle Deposition Rates at
50th Model Year 59
Exhibit 1.6-17. Mercury Surface Water Concentrations During the 50th Model Year 60
Exhibit 1.6-18. Total Mercury Concentration in Fish Species During the 50th Model Year 61
Exhibit 1.6-19. Mercury Speciation Across Different Model Compartments 62
Exhibit 1.6-20. Modeled Mercury Concentrations Compared to Measured Values 63
Exhibit 1.6-21. Mercury Chronic Non-Cancer Hazard Quotients for Ravena 64
Exhibit 1.6-22. Pathway Contributions to Divalent Mercury Chronic Non-Cancer Hazard Quotients 65
Exhibit 1.6-23. Pathway Contributions to Methyl Mercury Chronic Non-Cancer Hazard Quotients 66
Exhibit 1.6-24. Comparison of Hazard Quotients for Ravena Scenario Using Mean and 90th Percentile
Ingestion Rates 67
Exhibit 1.6-25. Divalent Mercury Dermal Hazard Quotients for a Child Aged 1-2 68
Exhibit 1.6-26. Effect of Fish Harvesting on Annually Averaged PB-HAP Concentrations During the 50th
Model Year in Ravena Pond Using 95-Percent UCL Emission Rate 69
Exhibit 1.6-27. Risks and Hazard Quotients in Ravena Pond with and without Fish Harvesting 70
l-iv
-------
1-1 Introduction
Under section 112(f)(2)(A) of the Clean Air Act (CAA), the U.S. Environmental Protection
Agency (EPA) is directed to assess the residual risk from continued emissions of hazardous air
pollutants (HAPs) from source categories regulated under Section 112(d) of the CAA. If existing
maximum achievable control technology (MACT) standards do not provide an "ample margin of
safety" for human health, EPA will promulgate additional emission standards for a source
category. Among other aspects of human health that EPA must consider is the potential for
exposures to HAPs via non-inhalation pathways and the risks associated with such exposures.
In Appendix C to EPA's report to SAB, a screening methodology is described that uses the Total
Risk Integrated Methodology (TRIM), an overall risk assessment modeling system developed by
OAQPS. The results are then used to support residual risk decisions for RTR II categories.
The TRIM-based methodology includes a screening evaluation that determines whether a
source warrants further consideration and then proceeds, as warranted, to more refined, site-
specific assessment involving TRIM. To illustrate the application of the TRIM-based
methodology for refined, site-specific risk assessment, ICF has conducted a case study of
emissions of PB-HAPs from one source in the portland cement source category.
This appendix presents the approach and the results of this case study. The case study
evaluates maximum individual ingestion exposures to mercury and dioxins and estimated
resultant human cancer risks and chronic non-cancer hazards. We targeted a facility that had
geographic characteristics amenable to two basic exposure scenarios (farmer and recreational
angler) chosen to illustrate the application of the methodology. The Ravena LaFarge Portland
Cement Facility in Ravena, New York (NY) was selected for the case study evaluation. The
Ravena facility is close to populated areas, several fishable water bodies, and potential
farmland. Although this facility may not necessarily represent the highest multipathway risk of
all facilities in the source category, it is useful as a demonstration of the intended approach to
be taken when the emissions from a source of interest for RTR de minimis levels and require
refined risk assessment. In turn, this demonstration is expected to be useful for soliciting
feedback on a range of risk assessment-related issues pertaining to EPA's RTR II program.
This document is divided several sections that describe the problem formulation, methodology,
and results for this case study evaluation. Section I-2 describes the conceptual model we
developed for examining potential exposure and risk. Sections I-3, I-4, and I-5 describes the
methods and inputs for the TRIM.FaTE fate and transport modeling, ingestion exposure dose
estimation, and dose-response values and risk characterization calculations conducted for this
assessment. The results of the case study and a limited discussion of results are presented in
Section I-6. References cited in this appendix are listed in Section I-7.
I-2 Conceptual Model for Potential Exposures/Risks
1-2.1 Selection of HAPs for this Analysis
To evaluate non-inhalation exposures and risks for RTR, the EPA has developed a list of 14
persistent, bioaccumulative hazardous air pollutants (PB-HAPs) for which the risks from non-
inhalation exposure pathways may be relevant. OAQPS developed the list based on a two-step
process taking into account the following:
. their presence on three existing EPA lists of persistent, bioaccumulative, and toxic
substances, and
1-1
-------
. a semi-quantitative ranking of toxicity and bioaccumulation potential of the entire list
of HAPs.
The list's development and utility in hazard identification for multipathway risk assessment are
further explained in Chapter 14 and Appendix D of Volume I of EPA's Air Toxics Risk
Assessment (ATRA) Reference Library (EPA 2004a). As described in the RTR Multipathway
Screening TSD (see Appendix G of this report), the first step in evaluating non-inhalation
exposures and risks is to compare HAPs emitted by a facility of interest to the chemicals on this
PB-HAP list. An initial screen is then conducted by comparing the facility-specific total
emissions (in ton per year or TPY) for a given PB-HAP to the de minimis emission rate
calculated using the RTR screening scenario. At each facility, PB-HAPs for which the total
emissions exceed the de minimis emission rate for that chemical (or chemical group) are not
screened out and may be subjected to further analysis.
Facilities in the portland cement manufacturing source category emit a variety of PB-HAPs,
including metals (lead, cadmium, and mercury) and organic compounds (particulate organic
matter and chlorinated dibenzo-p-dioxins and -furans, or "dioxins"). For each facility in this
source category, total emissions for each PB-HAP were compared to de minimis levels to
initially screen for the potential for non-inhalation exposures and risks. The results of this
screening are described in EPA's main report (see Section 3.2 of that report).
Although emissions of every PB-HAP on EPA's list are not reported for every facility in this
source category, more than half of the facilities report mercury emissions. Also, based on data
from individual facilities and knowledge of the portland cement manufacturing process, every
facility is assumed to emit dioxins. Both mercury and dioxins are presumed to be emitted in
relatively large quantities from facilities in this source category. Given the potential for exposure
via non-inhalation pathways to these two PB-HAPs and their relatively high emissions reported
for portland cement facilities, both mercury and dioxins are expected to be chemicals of concern
for the non-inhalation human health risk assessment of this source category for RTR.
Consequently, mercury and dioxin were selected as the chemicals for this case study.
I-2.2 Selection of Relevant Exposure Pathways and Approach to Exposure
Assessment
A multipathway exposure assessment of air toxics typically focuses on two categories of
ingestion pathways: (1) incidental ingestion of contaminated environmental media and (2)
consumption of contaminated food chain constituents. The range of exposure pathways
included in multipathway air toxics assessments is described in Chapters 14 and 15 of EPA's
ATRA Reference Library, Volume 1 (EPA 2004 a,b). For mercury and dioxins, exposures via
the consumption of farm produce, farm animals and animal products, and fish are the primary
concerns.
Mercury compounds that industrial processes emit to the air are typically a mixture of elemental
and divalent mercury species and are not particularly bioaccumulative. However, once
deposited to soil and surface waters, divalent mercury can be converted to methylmercury and
other organic mercury forms that are highly bioaccumulative. Methylation of mercury can occur
in the aquatic environment in particular, where it can enter the aquatic food chain. Elevated
levels of methylmercury have been measured in freshwater and saltwater fish, especially fish
species at higher trophic levels of aquatic food chains, which can accumulate mercury by
consuming small fish (i.e., "biomagnification"). As a result, the consumption offish that contain
methylmercury represents the primary human exposure pathway of concern for mercury.
1-2
-------
People can also be exposed to divalent mercury through FFC exposures and incidental
ingestion of surface soil, although in general these exposures receive less attention and are
typically of less concern relative to the fish-methyl mercury pathway due to high levels of methyl
mercury present in the environment in some locations as a result of historical contamination.
Dioxins do not readily degrade in biotic or abiotic environmental media, and their chemical
characteristics can promote bioaccumulation in aquatic and terrestrial food chains. In addition
to accumulating in fish, dioxins emitted to air can enter the human food chain via the deposition
to soil, surface water, and plant surfaces. Then, these chemicals accumulate in plants
(including some produce) and animals and animal products (e.g., dairy products, eggs, and fish)
that people consume. The consumption of fish and other food thus represents a non-inhalation
exposure pathway of concern for dioxins as well. People can also be exposed to dioxins
through the incidental ingestion of surface soil subject to deposition of dioxins, although this
pathway is generally less significant than pathways involving the food chain.
In a residual risk assessment, a key risk metric of interest with respect to informing policy
decisions is the risk to the individual most exposed (i.e., the "maximum individual risk," or MIR).
For an inhalation risk assessment, the MIR can be approximated (taking into account a range of
assumptions) using modeled long-term average air concentrations associated with a source and
information on where people reside. For an evaluation of non-inhalation exposures, however,
estimating the risk to the "most exposed" individual can be more difficult because chemical
concentrations in environmental media to which people are exposed and individual exposure
patterns associated with ingestion can vary greatly depending on location, timing, and other
factors. For example, people can be exposed to chemicals that accumulate in the FFC by
consuming a variety of fruits and vegetables, each of which may or may not be grown in the
vicinity of the source. The amount of each type of produce consumed can vary widely among
the individuals in a population, as can the fraction of each type of produce that is actually
impacted by emissions from a source.
To simplify the exposure and risk analysis of a multipathway air toxics risk assessment a
scenario approach can be employed. This approach, described in more detail in Chapter 15 of
ATRA (EPA 2004b), entails evaluating a combination of exposure pathways by which an
individual might be exposed to PB-HAPs (i.e., an "exposure scenario"). The scenario approach
provides a systematic method for evaluating the relative importance of exposure pathways (e.g.,
consumption of farm food products vs. consumption of fish) that are of potential concern for
different chemicals and locations. Only scenarios that are plausible for the situation of interest
are typically evaluated, and the assessment usually focuses on those scenarios that are
assumed a priori to lead to the highest individual exposure and risks. Risk metrics such as
incremental lifetime cancer risk and chronic non-cancer hazard quotient are calculated as
appropriate for each scenario, and information regarding the likelihood of a specific exposure
scenario actually occurring can be used to develop estimates of uncertainty for each scenario
and the variations thereof.
For this RTR case study, exposure estimates and risks were calculated for two basic scenarios:
. A farmer scenario, involving an individual living on a farm homestead in the vicinity
of the source who (a) consumes produce grown on and meat and animal products
raised on the farm, and (b) incidentally ingests surface soil at the location of the farm
homestead; and
1-3
-------
. A recreational angler scenario, involving an individual who regularly consumes fish
caught in freshwater lakes in the vicinity of the source of interest.
Variations of these two scenarios were evaluated using different assumptions about location of
the farm homestead or the water body, whether an individual is exposed via both scenarios, the
age of the individual exposed (for non-cancer hazards), the assumed ingestion rate of each food
type, and other factors. In addition, exposure estimates and risks to infants via breastfeeding
were evaluated, with the assumption that the nursing mother was exposed to chemicals via one
or both of the two basic scenarios listed above.
Exhibit 1.2-1 presents the conceptual exposure model for the farmer scenario. The arrows
represent the movement of chemical of concern through the environment and farm food chain.
In this exposure scenario, the hypothetical receptors consume produce, meat, and animal
products, and incidentally ingest soil. Exhibit 1.2-2 presents the conceptual exposure model for
the angler scenario. The hypothetical receptor, a recreational angler, consumes fish from a
contaminated water body.
Exhibit 1.2-1. Conceptual Exposure Model for Farmer Scenario
Forage,
Grain &
Silage
Meat &
Animal
Products
Produci
(Fruits &
Vegetables
Farmer
Portland
Cement
Facility
1-4
-------
Exhibit 1.2-2. Fish Consumption Exposure Pathway
Air
Portland
Cement
Facility
Fish
Aquatic
Food
Web
Soil
Recreational
_____ Angler
1-2.3 Overview of Modeling Approach
Fate and transport modeling of PB-HAPs was completed using the Fate, Transport, and
Ecological Exposure Module (TRIM.FaTE) of EPA's Total Risk Integrated Methodology.
TRIM.FaTE is a fully coupled multimedia model that estimates the flow of pollutants through
time among environmental compartments including air, soil, water, fish, and animals. The
results of TRIM.FaTE modeling are chemical concentrations in abiotic environmental media (air,
soil, surface water, and sediment) and in fish. TRIM.FaTE is essentially a spatially discrete,
multi-compartment box model that partitions chemical mass among phases and between
environmental compartments expressed (in part) using fugacity principles. For detailed
information on TRIM.FaTE, refer to EPA's TRIM website
(http://www.epa.qov/ttn/fera/trim gen.html).
Ingestion exposures were calculated for the two exposure scenarios of interest using Multimedia
Ingestion Risk Calculator (MIRC), and exposure and risk model that uses ingestion exposure
algorithms similar to those found in the Human Health Risk Assessment Protocol (HHRAP)
(EPA 2005). Chemical concentrations in intermediate farm food types (e.g., produce, animal
products) were calculated using biotransfer factors to estimate the food chemical concentration
based on the air and soil concentrations and deposition rates from TRIM.FaTE. Attachment C-2
of the main report provides details of the approach and methods used to calculate ingestion
exposures. Individual lifetime cancer risks for dioxins and chronic non-cancer hazard quotients
for dioxins, methylmercury, and divalent mercury were then calculated using oral cancer slope
factors and ingestion reference doses (RfDs).
1-5
-------
1-2.4 Selection of Facility for Analysis
To narrow the scope of this SAB case study and enable a more in-depth evaluation, we focused
on one portland cement facility. We first identified portland cement facilities that had high
emissions for both mercury and dioxins, assuming that higher emissions of the chemicals would
lead to higher human exposures. Of these facilities, we looked for one with geographic
characteristics amenable to the two basic exposure scenarios (farmer and recreational angler).
Minimum requirements included:
. Close proximity to a freshwater lake of reasonable size,1 and
. Proximity to land used to support a range of agricultural activities (crops and
animals).
The Ravena Lafarge portland cement facility (hereafter referred to as the Ravena facility) in
Ravena, New York (NY) meets these criteria and was selected for evaluation in this case study
(see Exhibit 1.2-3). The Ravena facility is close to populated areas, several fishable water
bodies, and potential farmland. Although this facility may not necessarily represent the highest
multipathway risk of all 91 portland cement facilities, it is useful for demonstrating the methods
of the refined multipathway HHRA (i.e., what to do when the emissions from a source category
exceed the de minimis levels), and this is expected to be useful for soliciting feedback on a
range of risk assessment-related issues pertaining to EPA's RTR II program.
Exhibit 1.2-3. Location of the Ravena Facility
Albany, NY
Nassau Lake
W Farm
Source
E Farm,,
Kinderhook Lake
Water/Wetlands
Pond
Ravena, NY
Alcove Reservoir
Developed
! I Barren/Gravel/Pit
I I Forest
5 I I Grasses
I | Fa rmfPasture
! The goal of the case study was to examine incremental exposure from facility emissions. A very large lake would
dilute the chemical. However, a very small pond would not sustain a fish population large enough to support
regular consumption offish by a local angler.
1-6
-------
The facility is located approximately 12 miles south of Albany, NY, in the southeastern portion of
Albany County (U.S. Census Bureau 2000; population 294,570). The population of Ravena,
NY, the nearest town (located just east of the facility), is 3,369. Nearby counties include
Renesselar, Greene, and Columbia, all in New York. According to the 2002 U.S. Department of
Agriculture (USDA) Census of Agriculture, these four counties support livestock and crops
grown for human and animal consumption (USDA 2002).
For the purpose of the Ravena HHRA, only dioxin and mercury emissions were evaluated. A
scenario layout for the Ravena area was created to use in TRIM.FaTE so that all relevant
ingestion pathways could be modeled. Both divalent mercury and dioxins can accumulate in the
farm food chain, so the scenario layout includes two farm homesteads, on the east and west
sides of the facility. The farm homesteads were located in areas where actual land use is
agricultural.
Methylmercury and dioxins bioaccumulate in fish, so four freshwater water bodies were included
in the Ravena layout to estimate exposure for the angler scenario. The Ravena area
encompasses many other water bodies including the Hudson River, but for the purposes of
TRIM.FaTE modeling, fish populations in only three lakes and one pond were modeled. The
lakes and pond represent a range of sizes and locations that the Ravena facility emissions
could impact. Alcove Reservoir is 7 miles west of the Ravena facility and supplies drinking
water to the city of Albany. Kinderhook Lake (8 miles southeast of the facility) and Nassau Lake
(11 miles northeast) allow recreational fishing. All three of these lakes are large enough to
support large fish populations and were modeled in TRIM.FaTE. A small pond is located 2
miles southwest of the facility. The pond was also modeled, although there is significant
uncertainty whether it is large enough to support a fishable aquatic ecosystem (discussed in
more detail in Section I-6.4). The Ravena facility is within 2 miles of the Hudson River, which
was modeled as a water body in this case study. A fish population was not modeled in the
river.2
For this case study, we modeled dioxin emission rates based on mean and 95th percentile upper
confidence limit emission factors based on the clinker production of the facility. (See Appendix F
to EPA's report to SAB.) The divalent and elemental mercury emissions modeled were those
reported in the 2002 NEI, and transformation of mercury (e.g., divalent mercury into
methylmercury) was included in the model (EPA 2002).
Exposure factors used in this case study are described in Section I-4 of this appendix.
I-3 Fate and Transport Modeling (TRIM.FaTE)
This section describes the TRIM.FaTE modeling conducted for this case study risk assessment.
Most of the material presented here describes the assumptions and data sources used to set
TRIM.FaTE inputs and settings related to meteorological inputs used by the model (Section I-
3.2), the spatial aspects of the modeled region (Section I-3.3), characteristics of abiotic
environmental compartments (Section I-3.5), and plants (Section I-3.6) included in the scenario,
and the aquatic ecosystems set up in each water body of interest (Section I-3.7). In Section I-
3.8, a summary of the distribution of mass among the modeled compartments in the scenario at
the end of the simulation period is presented to provide an overview of the model results (more
2 Incremental concentrations of dioxins and mercury in the river (i.e., those resulting from Ravena facility emissions)
are expected to be significantly lower than incremental concentrations in nearby lakes; the flow of the river will lead
to greater dilution. In addition, the emissions from the Ravena facility are only a small part of the total emissions
that affect the chemical concentrations in the river.
1-7
-------
detailed estimated concentrations are presented with other assessment results later in this
appendix).
1-3.1 Source Characterization
For this case study, TRIM.FaTE was used to model emissions from the Ravena facility of total
dioxins (modeled using 2,3,7,8-TCDD as a surrogate for total emissions calculated as dioxin
TEQ) and mercury (modeled as the appropriate mix of divalent and elemental). Transformation
of mercury to methylmercury was also modeled (thus, three separate mercury species were
modeled in TRIM.FaTE). The modeling scenario duration was 50 years, and emissions of both
mercury and dioxin were assumed to be constant over the course of the simulation.3
TRIM.FaTE was used to estimate chemical concentrations in air, soil, and selected surface
water bodies (and their corresponding benthic sediment layer), as well as components of a
representative aquatic ecosystem in each water body of interest for the risk assessment.
Chemical concentrations were estimated by the model on a bihourly basis for the scenario
duration and used to calculate annual average concentrations.
Estimated mercury and dioxin emissions to air from the Ravena facility are presented in Exhibit
1.3-1. All mercury and dioxin emissions were modeled as coming from the main stack, at a
height of 350 feet to match the reported stack height in NEI. Details regarding emission
estimates are presented in a separate appendix to this report. Despite the fact that divalent and
elemental mercury emissions for the Ravena facility did not exceed the de minimis levels, they
were still modeled for the case study to demonstrate the effect of applying site-specific
parameters The modeled divalent mercury emission rate for Ravena is approximately 70% of
the de minimis level.
Exhibit 1.3-1. Emissions of Dioxins and Mercury from the Lafarge Facility in
Ravena, NY, and Screening Results
PB-HAP
Emissions
(tons per year)
Screening Results
Dioxins/Furans a
95tn percentile upper
confidence limit of estimated
emission factor
3.28E-06
Exceeds de minimis
Levelc
Estimated mean emission
factor
1.34E-06
Exceeds de minimis
Levelc
Mercury - DivalentD
0.05625
Screens outc
Mercury - ElementalD
0.016875
Screens outc
a Emissions estimated based on tons of clinker produced using dioxin emission factors. Details about this
estimation are recorded in Appendix F.
b Emissions reported in 2002 National Emissions Inventory (NEI) (EPA 2002).
c De minimis levels for 2,3,7,8-TCDD TEQ and divalent mercury are 3.18E-08 and 1.64E-01, respectively.
The de minimis level for elemental mercury is larger than and elemental mercury emission rate found in
NEI; calculations of de minimis levels are further described in Appendix C.
I-3.2 Relevant Meteorological Data
TRIM.FaTE uses several meteorological inputs to determine chemical transfers among the air
compartments in a scenario via advective transport (i.e., wind-driven physical movement
through the atmosphere) and from air to underlying soil or water surfaces via deposition
3 Although actual emissions from portland cement facilities may fluctuate with time due to process characteristics,
start-up/shut-down operations, and other factors, modeling the emissions as constant was assumed to be
appropriate for estimating long-term chemical concentrations.
i-8
-------
transfers. These processes determine the long-term spatial patterns of chemical distribution
within the scenario, and modeled concentrations are highly sensitive to the meteorological
inputs used in TRIM.FaTE. Consequently, an initial step in developing the TRIM.FaTE
application for this case study was to collect meteorological data for the expected modeling
region (i.e., the area near Ravena, NY). The long-term trends in these data were then used to
inform the development of the modeling layout that is the basis of this scenario. To maintain
consistency with the development of the TRIM.FaTE application, we present in this section of
the appendix the specifications of the meteorological data and a summary of the long-term
temporal trends.
The meteorological inputs TRIM.FaTE requires include wind speed, wind direction, precipitation,
ambient air temperature, and mixing height. A suitable data set was selected based on how
closely the data are assumed to represent typical long-term conditions near the modeled
source, data availability, and completeness of the data set. For this assessment, hourly surface
meteorological data from the National Oceanic and Atmospheric Administration's National
Climatic Data Center (NCDC) Integrated Surface Hourly (ISH) Database were obtained (NOAA
2001). The ISH Database contains more than 20,000 stations and is quality controlled, and so
was judged to be a reliable source of meteorological data. Using this database, the closest
meteorological station to the Ravena facility is approximately 30 km north at the international
airport in Albany, NY. This surface meteorological station also hosts a radiosonde site, which
results in collocated surface and upper-air meteorological data (the upper-air data set contains
information used to determine mixing height) (NOAA 2007).
Three consecutive years of data (for 2001-2003) were readily available from this data set. To
facilitate the use of these data for a longer application, one of the years (2002) was repeated (to
create a 4-year dataset with the appropriate number of days to account for leap years) over the
duration of the simulation to create a dataset from 2001-2003. Exhibit 1.3-2 shows the 30-year
climate normals for annual total precipitation and annual average daily temperature compared
with the statistics for the 2001-2004 data used in this study and the overall average of the data
set used for modeling. The three years of observed meteorological data used for this study are
warmer and drier than the 1971-2000 NCDC 30-year climate normals for the Albany
meteorological station (NOAA 2003). The parenthetical numbers indicate percent deviation of
the 2001-2003 values from the 30-year normal values.
Exhibit I.3-2. Comparison of Historical and Modeled Temperature and Precipitation
Albany, NY:
Data Used for This Assessment
Statistic
Historical
30-year Normal
(1971-2000)a
2001
2002
2003
Overall b
Annual average of daily
average temperature (°C)
8.6
10
9.9
8.7
9.6
Deviation from normal temperature (°C)
+1.4
(+16%)
+1.3
(+15%)
+0.1
(+1%)
+1.0
(+12%)
Annual precipitation amount
(mm)
980
570
862
919
803
Deviation from normal annual precipitation (mm)
-410
(-42%)
-118
(-18%)
-61
(-6%)
177
(-18%)
a Historical temperature and precipitation data from NOAA's NCDC (2003).
b Overall includes the 2002 year data weighted twice as much as other years because they are repeated to create
a 4-year meteorological time series. This 4-year series was repeated to create the full 50-year meteorological data
set.
1-9
-------
In the natural environment, rainfall amounts that are less than climatologically typical quantities
in the Albany area can have the effect of decreasing modeled wet deposition. Decreased
rainfall may also increase chemical concentrations in water bodies by reducing their volumes
and flush rates, although the water bodies would also collect reduced amounts of chemicals
from their tributaries. The relatively small deviations from normal (warmer and drier) intrinsic to
the data used for modeling were assumed to be acceptable for this assessment.
Hourly average wind speed was reported as calm for about 20 percent of the time for the
repeated 4 year period. When not calm, wind speeds across the repeated 4-year period are
typically less than about 3 meter/second (m/s) 41 percent of the time. The wind direction is
most often from the south (29 percent of the time), with 49 percent of observed wind direction
split fairly evenly from among the north, northwest, and west. These wind direction preferences
indicate that areas south, east, and especially north of the Ravena facility should experience the
greatest dry deposition from facility's emissions. Exhibit 1.3-3 shows the frequency distribution
of wind directions and coincidental wind speeds for the repeated 4-year dataset. In general, the
observed trends in wind direction and speed in the modeling data set are expected to represent
the overall trends for the Ravena area (patterns were similar for all three years and correspond
to general conditions for the mid-Atlantic region).
Exhibit 1.3-3. Frequency Distribution of Wind Direction and Coincidental Wind Speed
(2001-2003 Albany dataset, all time periods)
30%
25%
20%
>
0
1 15%
O"
P
10%
5%
0%
—
R
NE E SE S SW
Wind Direction (from)
W
NW
~ 21 +
¦ 17-21
~ 11-16
¦ 7-10
~ 4-6
~ 1-3
¦ 0-1
Colored stacks
w ithin the bars
represent
w ind speed
(in m/s)
When precipitation occurs, wind speeds less than 3 m/s and wind direction from the south,
north, and west still dominate the wind pattern. However, winds from the north occur slightly
more often than from other directions during precipitation (27 percent of the time). During
precipitation, these wind direction preferences indicate that areas north, east, and especially
south of the Ravena facility should experience the greatest wet deposition from the facility's
emissions (see Exhibit 1.3-4).
Mixing height is used in calculating air concentrations and related processes. In addition, for
time periods when the mixing height is less than the stack height modeled in TRIM.FaTE (i.e.,
350 feet), chemical emissions from the source are transferred to an upper air layer and are not
available for deposition to modeled soil and water surfaces (this situation occurs less than 2
percent of the time for the modeling data set used). About 70 percent of morning mixing height
1-10
-------
values are less than 1,500 meters (m), and the morning frequency distribution decreases
logarithmically with increasing height values (see Exhibit 1.3-5). The afternoon mixing height
values follow a Gaussian distribution with a median value of 1,646 m.
Exhibit 1.3-4. Frequency Distribution of Wind Direction and Speed During Rain Events
(2001-2003 Albany dataset, hours with precipitation only)
30%
25%
20%
o
S 15%
O"
P
10%
5%
0%
NE E SE S SW
Wind Direction (from)
W
NW
~ 21 +
¦ 17-21
~ 11-16
¦ 7-10
~ 4-6
~ 1-3
¦ 0-1
Colored stacks
w ithin the bars
represent
w ind speed
(in m/s)
Exhibit 1.3-5. Frequency Distribution of Calculated Morning/Afternoon Mixing Height (m)
>
o
CT
=500 >=1000 >=1500
<1000 <1500 <2000
>=2000
<2500
>=2500
<3000
>=3000
<3500
>=3500
<4000
>=4000
<4500
>=4500
Height (m)
~ Morning ¦ Afternoon
I-3.3 Extent and Dimensions of Modeled Environment
This section describes the environment for which media concentrations were estimated using
TRIM.FaTE and the geographic characteristics of the modeled environment (e.g., layout of the
modeled domain and geometry of the constituents included).
1-11
-------
The design of the modeling layout was developed based primarily on physical/geographic
characteristics of the watersheds in the area and land-use data for the region. This section
provides a brief overview of the features present in the vicinity of the modeled facility.
As illustrated in Exhibit 1.3-6, the Ravena facility lies within the Middle Hudson Subbasin (HUC-8
Code 02020006).4 Rivers and streams in this subbasin drain to the Hudson River that flows
from north to south through this basin. No major lakes dominate this region, although numerous
reservoirs and lakes are located throughout the subbasin. Based on data from the U.S.
Geological Survey (USGS), land use in the vicinity of the Ravena facility is classified as a
mixture of forested land (with both deciduous and evergreen forests), land in agricultural use
(for pasture and cropland), and commercial and residential uses (see Exhibit 1.3-7). Land use
becomes more urban proceeding northward from the facility toward the city of Albany, NY.
Exhibit 1.3-6. Streams, Rivers, and Water Bodies of the Middle Hudson Sub-basin
Vermont
Middle Hudson Subbasin
my, NY .City. Center
A? i ^
vf?&Whiuftsbn^ Ri ver
Massachusetts
Lefarge PTC Facility in Ravena
Connecticut
40 Kilometei
43°0'0"N
42nCI'0"N
42°0'0"N
75CQ'0"W
74°0'0"W
-£3"0'0"W
New York
a Data obtained from the US Geological Survey (USGS) National Hydrography Dataset (NHD) for the Middle
Hudson Subbasin (USGS 2002b). These data are based on the content of USGS 1:100,000-scale data.
4 This and the following maps describing the modeling spatial layout are in a projected coordinate system, which
means that north is toward the upper-left of the page rather than the top of the page.
1-12
-------
Exhibit 1.3-7. Land Use in Region of the Ravena Facility
Vermont
Middle Hudson Subbasin
J AlbanyY City Center
HudsomRiver
4 /
-J4 ^ /
Lefarge PTC Facility in RavenafNY
\ t r
S. K
40 Kilometers
Massachusetts
Open Water
Iperennial Ice^Snow
1 Low Intensify Residential
Ihligh Intensity Residential
BMc o mm e rc ia I'l n dustr ia lH"r ans p o rtatio r
1 barren Rock/Sand/Clay
1 iQuarries/Strip Mines/O rave I Pits
I Inof^iriimug Forest
^"Evergreen F orest
.Mixed Forest
| | Sh r u bl an d
I . 0 rch a rds/Vin ey a rds/Oth e r
G rass I an ds/H er b ac eo us
| jpasture/H ay
-—J Row C rops
MSmall Grains
~f allow
I Urban/Recreational Grasses
I IWo o dy Wetla n ds
lEmeraent Herbaceous Wetlands
a Land-use data were obtained from the USGS Multi-Resolution Land Characteristics Consortium (MRLC) National
Land Cover Dataset 1992 (USGS 1992). Data were derived from the early-to mid-1990s Landsat Thematic
Mapper scans, have a spatial resolution of 30 m, and contain 21 landcover classifications (e.g., deciduous,
evergreen, and mixed forest; urban/recreational grasses; pasture/hay; row crops; low- and high-intensity residential;
and commercial/industrial/transportation).
I-3.4 TRIM.FaTE Parcel Design
The TRIM.FaTE surface parcel layout is the two-dimensional configuration of soil and water
regions included in the modeled domain; this is overlain by the air parcel layout. These layouts
provide the spatial reference for three-dimensional compartments that hold the modeled
chemical mass.
1-3.4.1 Surface Parcel Layout
The chief goal in designing the surface parcel layout was to accurately capture the watersheds
surrounding the water bodies selected for modeling (i.e., those that contain fish people are
assumed to eat) and the watersheds unique to the tributaries of the Hudson River that are in the
vicinity of the facility. In pursuing this goal, parcel shapes were kept as simple as possible to
reduce complexity in the layout and corresponding run time. As required by TRIM.FaTE, no
parcel is fully contained within any other parcel; all parcels share at least one side or corner with
another parcel.
The overall spatial extent of the modeling scenario is a 770 km2 rectangle that captures several
significant water bodies in the area and their watersheds (see Exhibit 1.3-8). The area
1-13
-------
approaches the metropolitan area of Albany to the north but generally stops short of the
residential areas.
Exhibit 1.3-8. Overall Modeling Region
Albany, NY-City Center
» S U VTsTV-^ I \
Lefarge PTC Facility in Ravena, NY
u l <
-------
this small pond actually is fishable is not clear from the information collected for this
assessment; this uncertainty is discussed in more detail in the results (Section 1-6).
Exhibit 1.3-9. Modeled and Reported Location of the Ravena Facility
TRIM Scenario Source Parcel
for li§fargg?PTC Facility in Ratffcna, NY
"LefargePTC Facility jn Ravena,\NY
(location as reported in NEI)
fc Vrj.^ v rWfr- 'M
1 Kilometers
lOpen Water
I IPerennial loa'Snow
I I Low Intensity Residential
Ihligh Intensity Residential
SsjC o mm e rc ia \i\ n dustr ia l/Tr ans p o rtatio r
rren Rock/Sand/Clay
I iQuarries/Strip Mines/Gravel Pits
I beciduous Forest
Evergreen F orest
Mb
-------
Exhibit 1.3-10. Water Bodies Included in the Modeled Region
Water Body Name
Actual Surface Area
(km2, from NHD a)
Modeled Surface Area of
TRIM.FaTE Parcel (km2)
Nassau Lake
0.654
0.654
Alcove Reservoir
5.511
5.514
Pond (unnamed)
0.020
0.020
Kinderhook Lake
1.341
1.342
Hudson River
22.318
22.329
3 NHD = National Hydrography Dataset
I-3.4.3 Modeled Agricultural Parcels
Agricultural use regions also were included in the modeled domain, to estimate soil
concentrations and other TRIM.FaTE outputs for use in calculating FFC exposures. Two farm
regions were created: one 2.5 km northwest of the facility and the other 5 km south-southeast
(see Exhibit 1.3-11). The locations of these two regions were selected based on land-use
patterns. The northwest location is the closest to the facility with a large area of predominantly
row crops land-use designation. However, the wind pattern in this area, as measured from the
Albany airport, is somewhat evenly split among westerly (i.e., blowing from the west), northerly,
and southerly, and thus this location is generally upwind relative to the facility. A second farm
parcel to the east was also included that is the closest large row crops land-use area that is
approximately downwind from the facility. Each of these farm regions was roughly bisected to
create two parcels (to accommodate modeling of tilled and unfilled surface soil for use in
estimating various farm food media concentrations).
West ./arm Tilled
West Farm Untilled -
I w Source
East Farm Tilled
HlOpen Water
I IPerennial IcefSnow
I Iloiai Intensity Residential
I ll-ligh Intensity Residential
o mm e rc ia VI n dus tr ia VTr ans p o rtatio n
barren Rock/Sand/Clay
I huarries/Strip Mines/O rave I Pits
I beciduous Forest
Evergreen F orest
.Mixed Forest
| jshrubland
j—10 rch a rds A/in eyards/Other
Grasslands/Herbaceous
1 Pastured ay
—Jrow Crops
HSmall Grains
I Fallow
I Urban/Re creation a I Grasses
I IWo o dy Wetla n ds
I [Emergent Herbaceous Wetlands
Exhibit 1.3-11. Agricultural Parcels Included in the TRIM.FaTE Scenario
1-16
-------
Watershed Parcels
The remaining area within the modeled domain was divided according to watershed boundaries,
based on stream and topographic data from NHD, Exhibit 1.3-12 presents the final surface
parcel layout for the TRIM.FaTE scenario. For clarity, the parcel names are omitted; see Exhibit
1.3-11 for parcel names.
Exhibit 1.3-12. Surface Parcel Layout with Water Bodies and Land Use
j JO pen Water
I iF'erennial Ice/Snow
I I Low Intensity Resider
I 'High Intensity Residei
—o mm e rc ia I'l n dustr ia I
rren Rock/Sand/Cli
I iOuarries/Strip Mines A
I beciduous Forest
j Evergreen
.Mixied F ore
'shrubland
I-3.4.4 Air Parcel Layout
Water
Ice/Snow
R esidential
Intensity Residential
ansportation
Rod
-------
TRIM.FaTE Users' Guide for additional discussion of this design (EPA 2003). Overall, 31
parcels, including the source parcel, are included in the air parcel layout.
Exhibit 1.3-14 overlays the air parcel and surface parcel layouts.
Exhibit 1.3-13. Air Parcel Layout
Air4
Air3
Air5
Air9
Air10
Air25
Air21
Air28
Air30
Air29
Kilometers
Exhibit 1.3-14. Air and Surface Parcel Layouts (Overlay)
| Air Parcels
j Surface Parcels
2.5
10 Kilometers
1-18
-------
1-3.5 Abiotic Environment
TRIM.FaTE requires various abiotic environmental properties for each compartment that is
included in the scenario (e.g., the depth of surface soil, soil porosity and water content, erosion
and runoff rates from surface soil to water bodies, suspended sediment concentration, and
others). Where site-specific data were readily available for this assessment they were used.
Regional or national defaults were used in numerous instances, especially for those parameters
that are not expected to influence chemical concentration dramatically. This section lists some
of the important values used for this application. A complete list of TRIM.FaTE inputs for abiotic
compartments is provided in Attachment 1-1 to this appendix.
1-3.5.1 Soil and Watershed Characteristics
1-3.5.1.1 Soil Properties
For this assessment, soils were modeled as three stacked soil layers (surface, root zone, and
vadose zone soil) over ground water. For soils not specifically modeled as land in agricultural
use, the surface soil layer was assumed to be 1 centimeter (cm) deep. Agricultural soils were
assumed to be tilled, and so a depth of 20 cm for the homogeneously mixed "surface soil" layer
was assumed. The tilled soil compartments were used to estimate concentrations in farmed soil
where produce is grown. Depths for surface and subsurface soil layers are presented in Exhibit
1.3-15. Where soils were assumed to be tilled, the thickness of the root zone soil was reduced
accordingly. Depths to and thicknesses of the vadose zone soil and groundwater layers were
identical regardless of whether surface layer was tilled.
Exhibit 1.3-15. Soil Compartment Depths
Untitled Soil (m)
Tilled Soil (m)
Surface soil
0.00-0.01
0.00-0.20
Root soil
0.01 -0.70
0.20-0.70
Vadose soil
0.70-2.10
0.70-2.10
Groundwater
2.10-5.10
2.10-5.10
For most of the basic surface soil properties, values were defined using typical regional or state
values compiled by McKone et al. for use in multipathway modeling (2001). A list of selected
soil properties is shown below in Exhibit 1.3-16.
Exhibit 1.3-16. Selected Properties of Soil and Groundwater
Property
Surface Soil
Root Zone
Soil
Vadose Soil
Groundwater
PH
6.8
6.8
6.8
6.8
Organic carbon content
0.008
0.008
0.003
0.004
Volume fraction, vapor (air content)
0.28
0.25
0.22
--
Volume fraction, liquid (water content)
0.19
0.18
0.17
--
Average downwind vertical velocity of
water infiltrating the soil (m/day)
8.22E-4
8.22E-4
8.22E-4
--
1-3.5.1.2 Erosion
Erosion rates for each surface parcel were estimated using the Universal Soil Loss Equation
(USLE), with a sediment delivery (SD) ratio adjustment. The USLE is intended to predict the
long-term average soil losses from individual field areas (Wischmeier and Smith 1978) and
represents the sheet and rill erosion from a small plot or agricultural field. Application of the
1-19
-------
USLE to an entire watershed requires modification of the equation result to account for
subsequent re-deposition of eroded soil before it reaches the water body. The SD ratio was
developed for this purpose: it estimates the fraction of sediment that reaches a water body
based on the size of the watershed.
The USLE and SD equations use only a few inputs; representative site-specific values were
developed to estimate erosion for this application with readily available data. Rainfall/erosivity
values were used from Albany County for plots west of the Hudson River and Rensselaer
County for regions east of the Hudson River (NRCS 2007). Soils data were obtained from the
Soil Survey Geographic (SSURGO) database for the counties of interest (obtained from the
USDA Natural Resources Conservation Service) to calculate site-specific soil erodibility factors.
Different cover management factors were used for farm parcels and natural forests and grasses
and herbs. For more information on the equations that were used and the derivation of values,
see Attachment 1-1. Calculated erosion rates for each surface soil parcel ranged from 9.5E-5
kilograms per square meter per day (kg/m2/day) to 2.1E-3 kg/m2/day.
The USLE is an empirical model, and therefore modeled conditions must be similar to
conditions for which the model has been calibrated to output useful results. In particular, the
USLE was designed for application to a single slope or field, rather than to an entire watershed.
Using average values across a watershed parcel would likely introduce uncertainties in the
prediction; predictions are improved when individual analyses of the slopes within the watershed
are conducted. We note that the EPA's HHRAP documentation states that using the USLE to
calculate sediment load to a lake from the surrounding watershed can sometimes lead to
overestimates (EPA 2005). The use of area-weighted averages for some of the USLE variables
helps to avoid under- or over-estimating by assuming uniformity across the watershed. The
area-weighted soil erodibility factor (K) and cover management factor (C) are not expected to
contribute significantly to inaccurate soil erosion estimates.
Estimating the length-slope (LS) factor is more challenging than any other factor for the USLE
(Moore and Wilson 1992), especially for complex watersheds. In actual watersheds, the entire
watershed has neither uniform slope length nor uniform slope steepness. Also, due to
nonlinearities in the equation to calculate the LS factor, the assumption of uniformity can result
in underestimates or overestimates of the LS factor. The use of average slope likely would
underpredict the LS factor. An average slope-length of 200 m may be accurate or slightly
greater than average, and thus may slightly overpredict the LS factor by some unknown
amount. Finally, uncertainty is introduced when using the SD ratio to account for the re-
deposition of soil before it reaches the water body. The degree by which the SD ratio
underpredicts or overpredicts actual sediment delivery is unknown. Additional discussion of the
assumptions made in estimating erosion rates for this modeling application and the associated
uncertainties is included in Attachment 1-1.
1-3.5.1.3 Runoff
Runoff from surface parcels into water bodies was calculated by subtracting the annual
evaporation (0.508 m/year, USGS 2004) from the annual precipitation (0.980 m/year, NCDC
2003). This total runoff value includes interflow and ground water recharge; to estimate surface
runoff only, total runoff was reduced by 50 percent per the recommendation included in HHRAP
(EPA 2005). Total runoff rate for all surface parcels except the source parcel was estimated to
equal 4.04E-4 m3/m2/day. The source parcel was not included in runoff because the Ravena
facility is assumed to have different containment configurations than the rest of the area.
1-20
-------
1-3.5.2 Water Body Characteristics
1-3.5.2.1 Surface Water and Sediment Properties
Important surface water and sediment properties for all lakes and the river are shown in Exhibit
1.3-17; sources for these properties are values for all other inputs are listed in Attachment 1-1.
Exhibit 1.3-17. Selected Surface Water and Sediment Properties
Property
Value
Temperature (K)
287
Suspended sediment concentration (kg sediment/ kg water)
0.01
Water column and sediment organic carbon content (kg organic
carbon/ kg solid wet weight)
0.02
Water column and sediment pH
7.3
Chlorophyll concentration (mg/L)
0.0029
Chloride concentration (mg/L)
8.0
Algae density in water column (g/L)
0.0025
I-3.5.2.2 Water Transfers
A water balance was assumed in order to estimate annual flush rates for each modeled water
body. Inputs to each water body included runoff from the surrounding watershed and direct
precipitation to the lake. Outputs from the water body included flushing through the lake outlet
and evaporation from the lake surface.
Long-term average precipitation used to calculate the water balance was obtained from the
Albany airport cooperative observation station. For the water body, this value was added as a
water input, based on surface area of the lake. Runoff from the watershed was calculated by
subtracting annual average evapotranspiration from annual average precipitation and
multiplying the difference by the total watershed area. Evapotranspiration data were obtained
from USGS (2004); a value of 20 inches was assumed to apply across the entire scenario.
Reported runoff values closely matched the value we calculated by this method. For
Kinderhook Lake, the calculated outflow from Nassau Lake was also included as a water input.
Evaporation from the lake surface was subtracted from the water inputs to estimate the
volumetric flow of water leaving the water body. Using surface area and mean depth to
calculate lake volume, a turnover rate in flushes per year was calculated. The values of these
turnover rates are presented in Exhibit 1.3-18.
Exhibit 1.3-18. Turnover Rates for
Ravena Water E
todies
Water Body
Turnover Rate
Kinderhook Lake
3.35
Nassau Lake
4.17
Alcove Reservoir
0.51
Pond
10.30
For water transfer calculations for the river, water velocity is required. The river velocity was
calculated by dividing the average discharge rate of the Hudson (USGS 2008a) by the cross-
sectional area of the Hudson River near Ravena (Oak Ridge National Laboratories 1977). The
estimated river velocity calculated in this way is 0.88 meters per second (m/sec).
1-21
-------
1-3.5.2.3 Sediment
The sediment balance of each watershed/water body system modeled was estimated by
accounting for sediment inputs to the lake based on the erosion calculations and the removal of
sediment from the modeled system via benthic burial and outflow of suspended sediment in the
water column. In this scenario, assumptions about the physical environment were used to
calculate sediment input through erosion and sediment removal through suspended sediment
flushing. All sediment inputs to the watershed are derived from the erosion calculations.
For this modeled system, all sediment that is not flushed out as suspended sediment is
assumed to be buried (i.e., removed from the modeled system by transfer to the consolidated
benthic sediment layer, where it is assumed to no longer interact with the overlying water
column). Suspended sediment depositional velocity is used to calculate total deposition to the
lake bottom, and the difference between deposition and burial is then used to calculate the
sediment that is resuspended. In TRIM.FaTE, resuspension rate is used to internally calculate
burial rather than using the burial rate directly. Resuspension rates were calculated to match
the calculated burial rates.
Based on these calculation methods, burial rates for the three Ravena lakes ranged from
0.0052 kg/(m2-day) to 0.0129 kg/(m2-day), with a value of 0.2066 kg/(m2-day) calculated for the
pond. The burial rate for the pond was set higher than the values for other watebodies in order
to maintain the sediment input/output balance and offset the high erosion rates estimated for the
pond watershed based on the presence of the land use category "Quarries/Strip Mines/Gravel
Pits" (likely associated with the facility) covering about 20 percent of the watershed. In a survey
of 56 lakes across the United States, median burial rates were range from 0.0027 to 0.0137
kg/(m2-day) (USGS 2004), which is comparable to the values calculated for the lake scenarios
(but substantially lower than the burial rate used for the pond). Exhibit 1.3-19 presents
estimated suspended sediment concentrations and calculated burial rates.
Exhibit 1.3-19. Sediment Total Suspended Solids and Burial
Rates for Ravena Water Bodies
Suspend Sediment
Burial Rate
Water Body
Concentration
(kg sediment/
m3 water)
m3
kg
m2 - day
m2 - day
Kinderhook Lake
0.010
5.0E-06
5.2E-03
Nassau Lake
0.010
1.2E-05
1.3E-02
Alcove Reservoir
0.010
6.6E-06
6.9E-03
Pond
0.110a
2.0E-04
2.1E-01
Pond suspended sediment concentration was assumed to be higher than those
for other water bodies because of higher erosion rates and small water body
size.
I-3.6 Terrestrial Plants
Calculations of the areal coverage of each land-use type within each parcel were used to set
each modeling surface parcel's dominant vegetation type (using the National Land Cover
Dataset 1992 (USGS 1992) classifications grouped to match the TRIM.FaTE vegetation types
as described in Section I-3.6). This strategy results in some simplification because most parcels
are at least several square kilometers in area and contain a variety of land-use. However, key
parcels such as the farms are drawn smaller in order to more accurately represent actual land
1-22
-------
use. The TRIM.FaTE vegetation designations are presented in Exhibit 1.3-20, along with the
names of each parcel.
Exhibit 1.3-20. Surface Parcel Layout with Plant Types arid Relevant Land Use
E2
Nai
W1
W2
it Farm Untillei
West Fa n Tilled
E3
E6
: Farm Til
irce
W3
100k Lk.
Untitled
!es.
E4
W5
W4
I Source
| Tilled Farmland
—II Lintllled Farmland
| Pond/Lake
j River
I Deaduous Forest
[ Grasses/Herbs
Hudsbrt River
Modeling plants in TRIM.FaTE requires two additional properties: (1) an "allow exchange"
property that is used in TRIM.FaTE algorithms to determine whether plants are actively growing
(and thus able to exchange chemical mass to and from the ambient air and take up chemical
mass from soil); and (2) a litterfall rate property that dictates when and how fast chemical mass
accumulated by a leaf is transferred to underlying surface soil (to account for chemical transfers
to soil from leaves dropped by deciduous trees and plants). For this assessment, the dates at
which these seasonal events occur were based on the dates of the first and last frosts reported
for Albany, NY (NOAA 1988). The average last day of frost in the spring is April 23, and the first
date of frost in the fall is October 15, assuming a 50-percent probability of a temperature
threshold of 28 °F. Litterfall is assumed to begin on the first day of frost and to end 30 days
after this date, with a litterfall rate of 15 percent of the remaining detritus falling per day.
I-3.7 Aquatic Ecosystem
To estimate risks to human health for the angler scenario, site-specific models of aquatic food
webs were developed in TRIM.FaTE to represent four water bodies in the vicinity of Ravena,
NY: Nassau and Kinderhook Lakes, Alcove Reservoir, and an unnamed small pond near the
facility. Characteristics of the TRIM.FaTE fish compartments used to represent fish in each
water body were based on site-specific fish survey data, supplemented by information from the
open literature.
1-23
-------
The development of each food web consisted of three stages:
1. Collection of local fish survey data for the water bodies from the New York State
Department of Environmental Conservation (NY DEC), including data on the relative
abundance and size/weight distribution of each species, to the extent available;
2. Formulation of simplified food webs for each water body, based on the fish surveys and
other biological and physical data for each water body, with supplemental information on
fish feeding habits, aquatic food webs, and biomass densities for different trophic levels
from the open literature; and
3. Assignment of values for the remaining parameters (e.g., individual body weight,
numeric density per unit area, lipid content) for each biotic compartment for each water
body in TRIM.FaTE from the available data.
Professional judgment was used where available data were incomplete. The process employed
to configure TRIM.FaTE aquatic food webs and set model input properties is summarized here
and discussed in greater detail in Attachment 1-1.
1-3.7.1 Collection of Information on Species Present in Water Bodies
To support the development of the aquatic food webs, ICF contacted fishery biologists at the NY
DEC Region 4 Bureau of Fisheries. The NY DEC conducted surveys offish in Nassau and
Kinderhook Lakes at various times between 1988 and 2006 (NY DEC 2008). Due to the
presence of polychlorinated biphenyls (PCBs) at Nassau and Kinderhook Lakes, there are fish
consumption advisories at these water bodies (NY DOH 2007), and aquatic sampling is
performed to assess current contaminant levels. The New York State Fish and Wildlife
Department published the results offish surveys conducted from 1963 to 1970 for Alcove
Reservoir (NY FWD 1971). This 1971 survey report presented data on average fish weights,
which were used, where applicable, to estimate the average weight per individual fish for each
species in all of the modeled water bodies for this assessment. Because data on fish length or
weight were not available for the other water bodies, average fish weights for each species from
the Alcove report were used as the average fish weights for the same species in the other water
bodies.
No survey or other site-specific data were identified for the small pond. Professional judgment
and published data were used to develop a model food web for the small pond. The food web
for the small pond was developed from an analysis of data presented by Demers et al. (2001)
for two small lakes in Canada. As a conservative position, the small pond was assumed to
sustain a viable fish community from year to year. In each water body, young of the year were
assumed to comprise 15 percent of the total fish biomass on an annual basis biomass.
I-3.7.2 Creation of Food Webs
Food webs for each of the four water bodies were constructed from the information sources
identified above. Several steps were required to construct each food web and to assign
parameter values for all aquatic biotic compartments for TRIM.FaTE:
1. Estimate total standing fish stock (i.e., total fish biomass per unit area) for each water
body based on total biomass estimates reported for similar water bodies in the literature;
2. List for each water body all fish species found in the surveys of the water body;
1-24
-------
3. Identify for each species an average body weight per individual based on the Alcove
Reservoir data;
4. Estimate total biomass caught for each species in the surveys by multiplying the number
of individuals of each species caught over the survey years for the water body by the
average body weight per individual for each species;
5. Estimate the relative total biomass for each species (percentage of total biomass
represented in surveys);
6. Estimate the absolute biomass of each species by multiplying its percent relative
biomass by the estimated total standing fish stock (Step 1);
7. Estimate the numeric density of each fish species (number per unit area) based on
biomass density and average individual weight for each species; and
8. Evaluate the feeding habits of each fish species, as determined from a variety of
sources, relative to the food/prey categories supported by TRIM.FaTE:
. plankton (called algae in TRIM.FaTE; however, it represents both phytoplankton
and zooplankton);
. macrophytes;
. benthic invertebrates (e.g., aquatic insects, crustaceans, mollusks);
. small planktivorous fish (e.g., young of the year, minnows; feed on algae and
zooplankton in the water column);
. larger omnivorous fish that feed on smaller fish in the water column and benthic
invertebrates and/or macrophytes (e.g., sunfish, yellow perch)
. small-to-medium sized benthivores/omnivores that feed primarily on benthic
invertebrates, detritus, and possibly macrophytes (e.g., small carp, white sucker).
Additionally, the lipid content of each species was estimated based on values reported in
national surveys.
The initial estimates of relative abundance for each fish species were based on the fish survey
data. These data are presented for reference in Attachment 1-1. Only the species identified by
fish surveys were assumed to be present in the four modeled water bodies. The body weight of
each individual was assumed to be equal to the average fish weight estimated from the Alcove
surveys. When species were present in the other lakes, but not in the Alcove Reservoir,
professional judgment and readily available data for other locations (e.g., Minnesota fish
surveys) were used to estimate an average individual body weight for the species.
At the small pond, only three species/groups were assumed to be present: largemouth bass,
sunfish (e.g., bluegill or pumpkinseed), and shiners. The mass of the individuals was estimated
based on professional judgment and the Demers et al. (2001) study of two small lakes.
Total relative biomass for each species within a water body was estimated differently for the four
water bodies. At Alcove Reservoir, each species' biomass representation was determined by
taking the observed biomass of the species caught across all survey years and dividing that by
the total fish biomass reported in the Alcove report across all survey years. At both Nassau and
Kinderhook Lakes, the survey data seemed biased towards several species, specifically yellow
and white perch, perhaps due to sampling techniques. We therefore adjusted biomass
representation to reflect a more balanced abundance across different species for these two
1-25
-------
water bodies. For the small pond, the distribution of biomass among the several species was
estimated based on the study published by Demers and co-workers (2001).
I-3.7.3 Parameterization of Fish Compartments to be Included in Application
All fish species were assigned to one of the following five fish compartments established in
TRIM.FaTE:
. Water column carnivore (large predominantly piscivorous species, e.g., walleye and
largemouth bass);
. Water column omnivore (medium-sized fish that feed primarily in the water column,
e.g., sunfish, yellow perch);
. Water column herbivore (more appropriately termed planktivore);
. Benthic carnivore (large carnivorous species, e.g., large bullhead, eel); and
. Benthic omnivore (medium-sized fish that feed primarily on benthic invertebrates).
The compartment to which each species was assigned was determined by its general foraging
habitat (i.e., benthic or water column) and its primary food sources (e.g., invertebrates, smaller
fish, plant material). The total biomass for each of the five fish TRIM.FaTE compartments was
set equal to the sum of the biomass of the species assigned to each compartment.
The diet composition for each of the five fish compartments was calculated as being
proportional to the biomass representation of each species assigned to that compartment. For
example, if largemouth bass comprised 75 percent and smallmouth bass comprised 25 percent
of the biomass of the WCC compartment, then the diet composition of the largemouth bass
multiplied by 0.75 would be added to the diet composition of the smallmouth bass multiplied by
0.25 to estimate the diet composition for the WCC compartment. The four aquatic food webs
developed for the Ravena case study are summarized in Exhibit 1.3-21. Similarly, the lipid
content for each of the five fish compartments in TRIM.FaTE was estimated from the biomass-
weighted lipid content of the individual species assigned to the compartment. Thus, using the
same example, the largemouth bass lipid content, multiplied by 0.75, would be added to the
smallmouth bass lipid content, multiplied by 0.25, to estimate the lipid content of the WCC
compartment.
1-26
-------
Exhibit 1.3-21. Aquatic Food Webs for Modeled Water Bodies
TRIM.FaTE Representative
Compartment Fish Species
Diet Fractions
Algae/
Zooplankton
Macrophytes
Benthic
Invertebrates
Benthic
Omnivores
Water Column
Herbivores
Water Column
Omnivores
Alcove Reservoir
Water Column
Carnivore
Chain pickerel, largemouth bass,
northern pike, walleye
41.1%
25.0%
4.5%
29.5%
Water Column
Omnivore
Bluegill, pumpkinseed, redbreast
sunfish, smallmouth bass, white perch,
yellow perch
7.8%
53.5%
38.8%
Water Column
Herbivore
Black crappie, young of the year
96.3%
3.7%
Benthic
Carnivore
American eel
50.0%
50.0%
Benthic
Omnivore
Bullhead
100%
Kinderhook Lake
Water Column
Carnivore
Largemouth bass, tiger musky, walleye
33.0%
25.7%
7.8%
33.5%
Water Column
Omnivore
Bluegill, pumpkinseed, redbreast
sunfish, rock bass, smallmouth bass,
white perch, white sucker, yellow perch
8.1%
57.9%
34.0%
Water Column
Herbivore
Black crappie, common carp, fantail
darter, golden shiner, young of the year
81.8%
13.5%
4.7%
Benthic
Carnivore
American eel
50.0%
50.0%
Benthic
Omnivore
Bullhead
100%
Nassau Lake
Water Column
Carnivore
Chain pickerel, largemouth bass
25.0%
25.0%
50.0%
Water Column
Omnivore
Bluegill, pumpkinseed, redbreast
sunfish, smallmouth bass, white perch
white sucker, yellow perch
8.7%
61.0%
30.3%
Water Column
Herbivore
Black crappie, common carp, golden
shiner, young of the year
92.4%
2.7%
4.9%
Benthic
Carnivore
American eel
50.0%
50.0%
Benthic
Omnivore
Bullhead
100%
1-27
-------
Exhibit 1.3-21, continued. Aquatic Food Webs for Modeled Water Bodies
Diet Fractions
TRIM.FaTE
Compartment
Representative
Fish Species
Algae/
Zooplankton
Macrophytes
Benthic
Invertebrates
Benthic
Omnivores
Water Column
Herbivores
Water Column
Omnivores
Small Pond
Water Column
Carnivore
Largemouth bass
50.0%
50.0%
Water Column
Herbivore
Golden shiner, young of the year
100%
Benthic
Omnivore
Sunfish
100%
Note: For the purpose of the ecological risk assessment, mallards were included in the four water bodies at the
Ravena site. The mallard diet consists of 67 percent macrophytes and 33 percent benthic invertebrates. See
Appendix J for further discussion.
I-3.8 Mass Balance Results
One summary generated by TRIM.FaTE for each model run is a report of the chemical mass at
the conclusion of the simulation in each compartment type included in the modeled
environment. Exhibit 1.3-22 presents the distribution of modeled chemical mass in the modeled
environment for TRIM.FaTE 2,3,7,8-TCDD and mercury modeling for the Ravena site. The first
section of this table presents the proportion of chemical mass emitted over the 50 year period in
Ravena that was removed from the scenario by transfer to air advection sinks (i.e., the fraction
of mass that was blown out of the modeled environment by wind). The remaining fraction is the
amount emitted and deposited from air to soil, water, and plant surfaces comprising the overall
surface layout via wet and dry deposition processes (including vapor diffusion where
applicable).6 In the second part of this table, the final distribution of the deposited chemical
mass at the ending time step is summarized by media and modeling sink type and within soil
layers.
For dioxins, most of the mass emitted by the modeled source is blown out of the modeled
domain into air sinks, and less than 2 percent of total dioxin emitted is deposited within the
scenario. Of the amount present in the scenario at the end of the simulation (minus emitted
mass in the air advection sinks), 85 percent had degraded; most of the remaining chemical
mass was found in the surface soil. Only trace amounts of the deposited are estimated to
remain present in surface water, and aquatic biota. No abnormal resuspension events are
assumed; taking into account these events would result in higher concentrations and exposure.
Mass distributions between emission factors; the only the total amount in the system changed.
6 Note that the fraction "emitted and deposited" in this table represents the proportion of emitted chemical that is
immediately deposited from air, not the fraction of emitted chemical mass in the soil, water, and plant compartments
at the conclusion of the simulation. Once deposited, chemical mass in TRIM.FaTE can be re-emitted to the air
(e.g., via volatilization of vapor-phase chemical or dust resuspension), transported to another environmental
compartment via advective or other processes, accumulated by biotic compartment types included in the scenario,
metabolized or broken down by abiotic degradation processes, or (in the case of mercury) transformed to another
modeled chemical.
1-28
-------
Five percent of the total emitted chemical mercury mass remained within the bounds of the
Ravena screening scenario; the remaining was found in air sinks.
Of the mass remaining in the scenario, most was found in the soil and sediment, very little had
been lost due to reaction/degradation, and most of the remaining about had been lost in
sediment sinks and removal via the Hudson River outside of the system.
Exhibit 1.3-22. Distribution of Chemical Mass in Ravena, NY Scenario
Compartment
2,3,7,8-
TCDD
(95th)
2,3,7,8-
TCDD
(Mean)
Divalent
Mercury
(Hg2)
Elemental
Mercury
(HgO)
Methyl
Mercury
(MeHg)
Total
Mercury
Distribution of Total Mass Added to Scenario from Modeled Source
Emitted chemical mass
removed from scenario
and transferred to air
sinks
98.1%
98.1%
79.0%
99.9%
a
94.7%
Emitted chemical mass
that is deposited to soil,
water, and plants
1.9%
1.9%
21.0%
0.1%
a
5.3%
Distribution of Mass Remaining in Scenario
Air
0.03%
0.03%
0.002%
0.5%
0.00%
0.01%
Soil
11.2%
11.2%
83.7%
39.3%
88.6%
83.1%
Plants
0.1%
0.1%
0.03%
0.00%
0.00%
0.03%
Surface Water
0.001%
0.001%
0.01%
0.1%
0.01%
0.01%
Sediment
0.1%
0.1%
2.0%
0.8%
0.3%
2.0%
Aquatic Biota D
0.0002%
0.0002%
0.00%
0.00%
0.0003%
0.0000%
Groundwater
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
Reaction Sink
85.6%
85.6%
0.00%
0.00%
0.00%
0.00%
All Other Sinks c
2.9%
2.9%
14.2%
59.4%
11.1%
14.8%
Distribution of Mass in Soil
Surface Soil
99.2%
99.2%
99.4%
2.3%
99.4%
98.7%
Root Zone Soil
0.8%
0.8%
0.6%
97.5%
0.6%
1.3%
Vadose Zone Soil
0.00%
0.00%
0.00%
0.2%
0.00%
0.001%
a No methyl mercury was emitted directly from the point source; it was created via transformation within the TRIM.FaTE
system, therefore the percentage as a fraction of total emissions cannot be calculated
b Compartment includes mallard, but does not include mink.
c Other sinks include soil sinks, sediment sinks for water bodies and the river, and a flush sink for chemical removed by
being carried via the Hudson river.
1-4 Exposure Assessment
1-4.1 Approach
The Ravena facility site-specific HHRA is intended to address non-inhalation (ingestion)
exposures to potential human receptors. Consistent with the scenario assessment approach
described in Section I-2.2, exposures for specific scenarios were estimated using assumed
ingestion activity patterns (i.e., estimating how much of each medium is consumed and the
fraction of the consumed medium that is grown in or obtained from contaminated areas) and
1-29
-------
characteristics of the individual potentially exposed (e.g., age and body weight). For the human
health assessment, three ingestion exposure scenarios were evaluated:
1. Consumption of farm-grown fruits, vegetables, and animal products, and incidental
ingestion of soil;
2. Consumption of self-caught fish from local water bodies; and
3. Ingestion of contaminated breast milk by infant.
Exposure doses (and subsequent risks) associated with dioxins and mercury for each ingestion
pathway were computed separately using the MIRC model so that the pathway(s) of interest for
each PB-HAP could be evaluated separately. Data related to exposure factors and
characteristics of exposed individuals were obtained primarily from EPA's Exposure Factors
Handbook (EPA 1997a, b).
As described in previous sections, exposures were modeled for two hypothetical farm
homesteads (referred to here as West Farm parcel and East Farm parcel) and four potentially
fishable water bodies near the Ravena facility. West Farm parcel is located approximately 2.5
kilometers (km) northwest of the Ravena facility; East Farm parcel is approximately 5 km south-
southeast of the facility. Each parcel is approximately 0.72 km2 in size, and each is roughly
bisected into tilled and unfilled parts. The size and location of the four water bodies are
provided below. The unnamed pond is the closest body of water to the source that could
theoretically support a fish population.
. Alcove Reservoir (5.5 km2) - west of source
. Nassau Lake (0.65 km2) - east of the source
. Kinderhook Lake (1.3 km2) - east of the source
. Small (unnamed) pond (0.020 km2) - south of source
A summary of the sources of contaminated media for each of the three exposure scenarios
evaluated is provided in Exhibit 1.4-1. See Section I-3.3 for site maps and a detailed description
of the spatial layout of the site, including the areas and locations of the farm and watershed
parcels relative to the Ravena facility, as well as land use patterns in the area surrounding the
facility.
Exhibit 1.4-1. Ingestion Exposure Scenarios
Scenario
Source of Ingested Media
Consumption of locally-grown
produce and animal products,
and incidental ingestion of soil
Products and soil from two locations with
agricultural land use:
o East Farm parcel
o West Farm parcel
Consumption of locally-caught
fish by sport anglers
Fish from four water bodies:
o Alcove Reservoir
o Kinderhook Lake
o Nassau Lake
o Small pond to south
Ingestion of contaminated
breast milk by infants
Breast milk; nursing mother would ingest farm
and fish media from most exposed locations
Estimated individual contact rates (i.e., exposure) for each exposure medium were evaluated
using two point estimates - one to represent average or central tendency exposure (CTE) and
another to represent upper-bound or reasonable maximum exposure (RME). The CTE
calculation is used to estimate exposure for individuals with average or typical intake of
1-30
-------
environmental media, taking into account the basic assumptions of the scenario (for example,
for the angler scenario, the exposed individual was assumed to regularly consume self-caught
fish). The RME is used to estimate exposures for individuals exposed via the evaluated
scenarios who are at the high end of the exposure distribution. The intent of the RME is to
estimate a conservative exposure case (i.e., well above the average case) that is still within the
range of possible exposure patterns. For the CTE scenario, consumer-only intakes/ingestion
rates for soil, fish, breast milk, and every FFC medium were set equal to the mean of the
distribution of national or other representative data for that food type. For the RME scenario,
estimates of the 90th percentile of consumer-only intakes/ingestion rates were used. For both
the CTE and RME ingestion exposure scenarios, all food types specific to a certain exposure
scenario (i.e., farmer, angler, or breast-feeding infant) were assumed to be obtained from
locations evaluated in this assessment (i.e., a farm parcel and/or one of the two water bodies, or
from a nursing mother consuming media from one or both of these routes). The approach to
estimating RME ingestion exposure is analogous to EPA's recommended approach for
conducting risk assessments at Superfund sites - that is, we are estimating the "highest
exposure that is reasonably expected to occur at a site," taking into account current and future
(potential) land-use conditions (EPA 1989). The RME is assumed in this instance to
approximate the maximum individual risk (i.e., the upper bound of individual risk based on
conservative assumptions that is unlikely to be exceeded). We have also estimated risk for the
CTE scenario to provide additional information that may be helpful in evaluating the level of
conservatism associated with this MIR estimate.
The conditions defined when conceptualizing and building the scenario were selected so that for
any given individual, a long-term exposure condition would be reasonably likely to be captured,
thereby ensuring that this estimate encompasses the MIR. However, we emphasize again that
because this assessment is designed to estimate the maximum individual risk for the exposure
scenarios evaluated, the results are not intended to represent the actual exposure for a typical
person living in the vicinity of the evaluated source, but rather the estimated exposure for a
person who meets the criteria of the scenarios evaluated (that is, someone who consumes only
produce grown and animals raised on a local farms, and/or someone who regularly consumes
self-caught fish from a local lake). The CTE scenario, therefore, represents an average, "central
tendency" exposure estimate within the relatively strict specifications of the exposure scenarios
developed for this assessment.
The remainder of this section describes the approach for estimating human exposures
associated with the incidental ingestion of soil, ingestion of FFC media, consumption of fish, and
infant consumption of breast milk. A discussion of exposure pathways for potential human
receptors is presented in Section I-4.2. Section I-4.3 describes the approach used to estimate
exposure-related dose for each relevant ingestion source and pathway, and includes a
summary of exposure parameters and assumptions.
I-4.2 Exposure Pathways
A summary of the ingestion exposure pathways evaluated in this assessment is provided in
Exhibit 1.4-2. The quantitative aspects of this non-inhalation evaluation focused primarily on
human exposures via the following ingestion pathways: incidental ingestion of soil; ingestion of
farm-food chain (FFC) media; ingestion offish; and infant ingestion of breast milk. Each
pathway is discussed in the following subsections.
1-31
-------
Exhibit 1.4-2. Summary of Ingestion Exposure Pathways and Routes of Uptake
Ingestion
Exposure Pathway
Intermediate
Environmental Uptake Route
Medium Ingested
Exposure
Pathway - Farm
Animals a
Medium
Process b
Farm-Food Chain
Consumption of
produce
Aboveground produce,
exposed fruits and
vegetables c
NA
Air
Air
Soil
Deposition to leaves/plants
Vapor transfer
Root uptake
Aboveground produce,
protected fruits and
vegetables c
NA
Soil
Root uptake
Belowground produce
NA
Soil
Root uptake
Ingestion of
forage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Beef
Ingestion of
silage
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Ingestion of
forage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Consumption of
farm animals and
Dairy (milk)
Ingestion of
silage
related food
Ingestion of grain
Soil
Root uptake
products
Ingestion of soil
Soil
Ingestion from surface
Pork
Ingestion of
silage
Air
Air
Soil
Direct deposition to plant
Vapor transfer to plant
Root uptake
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Poultry
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Eggs
Ingestion of grain
Soil
Root uptake
Ingestion of soil
Soil
Ingestion from surface
Incidental ingestion
of soil
Surface soil
NA
Surface
soil
Deposition; transfer through
plants; transfer via erosion and
runoff0
Fish
Direct uptake from water and
Consumption offish
Locally-caught fish
(see Exhibit 1.4-3)
NA
Fish
tissue
consumption of other
contaminated media modeled in
TRIM.FaTE d
1-32
-------
Exhibit 1.4-2, continued. Summary of Ingestion Exposure Pathways and Routes of Uptake
Ingestion
Exposure Pathway
Intermediate
Environmental Uptake Route
Medium Ingested
Exposure
Pathway - Farm
Animals a
Medium
Process b
Drinking Water
Ingestion of drinking
water
Surface Water
NA
Surface
Water
Deposition to surface water and
transfer via erosion and runoff
Breast Milk
Consumption of
breast milk
Breast milk
NA
Breast
milk
Ingested by mother and then
partition to breast milk
a NA = not applicable; calculation of intermediate exposure concentrations was required only for the farm animal/animal
product ingestion pathways.
Process by which HAP enters medium ingested by humans.
c For fruits and vegetables, the terms "exposed" and "protected" refer to whether the edible portion of the plant is
exposed to or protected from the atmosphere.
d Modeled in TRIM.FaTE.
Alcove Reservoir is a drinking water reservoir for eastern New York, therefore the exposure
calculations assume that Alcove Reservoir is the only source of drinking water for the exposed
individual. The surface water concentrations of PB-HAPs in Alcove Reservoir modeled by
TRIM.FaTE were used in exposure scenarios.
In addition to ingestion, non-inhalation exposure to PB-HAPs also can occur by way of the
dermal pathway (e.g., through incidental contact with PB-HAP-contaminated soil). However,
dermal absorption of chemicals that are originally airborne is generally a minor pathway of
exposure relative to other exposure pathways such as inhalation exposure or exposure via
ingestion of contaminated crops, soil, or breast milk (EPA 2008, CalEPA 2000). In general, the
assessment followed the protocol for evaluating a reasonable maximum exposure as described
in EPA's Risk Assessment Guidance for Superfund (RAGS), Volume I: Human Health
Evaluation Model, Part E, Supplemental Guidance for Dermal Risk Assessment (EPA 2004c).
1-4.2.1 Farm Food Chain Media Pathway
Data from the 2002 Census of Agriculture (USDA 2002) for Albany, Columbia, Greene, and
Rensselaer Counties of New York were examined to determine the relevant FFC exposure
pathways for this analysis. The census recorded the presence of cattle (for beef and milk), hogs
and pigs, chickens (for eggs and meat), corn, wheat, oats, beans, potatoes, forage, vegetables
and orchards in the four counties included in the Ravena, NY, spatial layout. Based on this
information, the following FFC pathways were evaluated in this assessment:
. Ingestion of homegrown produce (fruits and vegetables),
. Ingestion of homegrown beef,
. Ingestion of milk from homegrown cows,
. Ingestion of homegrown pork, and
. Ingestion of homegrown poultry and eggs.
Exposures to dioxin and mercury (as divalent mercury and methyl mercury) via these FFC
pathways were evaluated.
1-33
-------
1-4.2.2 Fish Consumption Pathway
This site-specific evaluation assessed human exposures via aquatic food chain contamination,
considering both game fish, which are generally top predators in general within aquatic
ecosystems, and bottom-feeding fish that might also be consumed by humans. As described in
Section 1-3.7, data from several fish surveys conducted by the New York State Department of
Environmental Conservation (NY DEC) between 1988 and 2006 were used to estimate the
relative abundance of different fish species in each lake for Kinderhook and Nassau Lakes, and
data for Alcove Reservoir were obtained from NY DEC fish surveys conducted between 1963
and 1970. The food web for the small pond was derived from an analysis of data presented by
Demers et al. (2001) for two small lakes in Ontario. We were not able to confirm what type of
fish population (if any) is present in the pond; the assumptions made for this risk assessment
are intended to err on the conservative side (i.e., by assuming that regular consumption of fish
caught in the pond could occur). The proportion of total fish biomass for each water body
contributed by each species was assigned to one of the following five fish compartments on the
basis of professional judgment and descriptions of their feeding habits available from online
fishing communities and from NYS DEC online documents: benthic omnivore (BO); benthic
carnivore (BC); water column herbivore (WCH); water column omnivore (WCO); or water
column carnivore (WCC). A summary of fish species present in water bodies around the
Ravena facility that are considered in the human health assessment is provided in Exhibit 1.4-3.
See Section l-3.7.3for a discussion of these compartments and the aquatic food web modeled
by the TRIM.FaTE for this simulation. Uncertainties associated with the assumptions regarding
fish populations and consumption rates are discussed in more detail in Section I-6.
Exhibit I.4-3. Fish Species Assumed to be Consumed in this Assessment
Water Body
TRIM.FaTE
Compartment
Type3
Representative Species
Fraction of Total
Angler Consumption
(by mass)
Alcove
Reservoir
WCC
Chain pickerel, largemouth bass, northern pike,
walleye
33%
WCH
Black crappie, young of the year
--
WCO
Bluegill, pumpkinseed, redbreast sunfish,
smallmouth bass, white perch, yellow perch
67%
BC
American eel
--
BO
Bullhead
--
Kinderhook
Lake
WCC
Largemouth bass, tiger musky, walleye
33%
WCH
Black crappie, common carp, fantail darter, golden
shiner, young of the year
--
WCO
Bluegill, pumpkinseed, redbreast sunfish, rock
bass, smallmouth bass, white perch, white sucker,
yellow perch
67%
BC
American eel
--
BO
Bullhead
--
1-34
-------
Exhibit 1.4-3, continued. Fish Species Assumed to be Consumed in this Assessment
Water Body
TRIM.FaTE
Compartment
Type a
Representative Species
Fraction of Total
Angler Consumption
(by mass)
Nassau
Lake
WCC
Chain pickerel, largemouth bass
33%
WCH
Black crappie, common carp, golden shiner,
young of the year
"
WCO
Bluegill, pumpkinseed, redbreast sunfish,
smallmouth bass, white perch, white sucker,
yellow perch
67%
BC
American eel
—
BO
Bullhead
—
Small
(unnamed)
pond
WCC
Largemouth bass
33%
WCH
Golden shiner, young of the year
—
BO
Sunfish
67%
a BO = benthic omnivore; BC = benthic carnivore; WCC = water column carnivore; WCH = water column herbivore;
WCO = water column omnivore
1-4.2.3 Breast Milk Pathway
The U.S. EPA (EPA 1980,1983) and the World Health Organization (WH01985,1989) have
published multiple reports documenting the presence of environmental chemicals and
contaminants in human breast milk. The magnitude of the nursing infant's exposure via
ingestion of contaminated breast milk can be estimated from information on the mother's
exposure, data on the partitioning of the chemical into various compartments of the mother's
body and into breast milk, and information on the infant's consumption of milk and absorption of
the chemical. This pathway is generally of most concern for lipophilic bioaccumulative
chemicals (e.g., dioxins) that can cause developmental effects. The methodology and
algorithms used for the breast milk consumption scenario for this case study are presented
separately in Attachment C-2. Only the results of the analyses are presented in this appendix.
I-4.3 Exposure Dose Estimation
Ingestion exposures for the angler and farmer scenarios for all media were calculated as
average daily doses (ADDs), expressed in milligrams of PB-HAP per kilogram of receptor body
weight per day (mg/kg-day). The equations in MIRC that were used to calculate ADDs for each
of the ingestion pathways are presented in Attachment C-2. Inputs to MIRC used for the
exposure dose estimates (as ADDs) and risk estimates for this assessment included
TRIM.FaTE PB-HAP concentrations, FFC algorithm parameters dictating the chemical quantity
accumulated in produce and animals/animal products, and exposure factors. Each of these
inputs is discussed below.
1-4.3.1 Media Concentrations
For the human health assessment, TRIM.FaTE was used to estimate human exposures to
2,3,7,8-TCDD and mercury via ingestion of locally grown produce and animal products and
ingestion of self-caught fish in several bodies of water in the vicinity of the Ravena facility. As
mentioned above, the water bodies include a farm pond near the facility, the Alcove Reservoir,
Kinderhook Lake, and Nassau Lake. Because it is largely a flow-through system,
concentrations of 2,3,7,8-TCDD and mercury attributable to the Ravena facility were not
estimated for the Hudson River.
1-35
-------
From TRIM.FaTE, the following environmental media concentrations specific to the PB-HAP of
concern were obtained:
. air concentrations (in g/m3);
. air-to-surface deposition rates for both particle and vapor phases (in g/m2-yr);
. ground water concentrations (in g/L);
. fish tissue concentrations for fish in trophic levels three and four (T3 and T4) (in
mg/kg wet weight); and
. concentrations in surface soil and root zone soil (in |jg/g dry weight).
These PB-HAP-specific values were then used to calculate receptor- and exposure scenario-
specific ADDs, using the pathway-specific equations provided in Attachment C-2.
For FFC exposure calculations, concentrations in FFC media were calculated using empirical
biotransfer factors (e.g., soil-to-plant factors, which are the ratios of the concentrations in plants
to concentrations in soil). In general, plant- and animal-specific parameter values, including
chemical-specific transfer factors for FFC media, were obtained from the Hazardous Waste
Companion Database included in HHRAP (EPA 2005). A list of variables and PB-HAP-specific
input parameters, along with the input values used in the evaluation of the FFC pathway, are
provided in Attachment 1-4.
1-4.3.2 Exposure Factors
Specific exposure factors used to estimate ADDs for the evaluated scenarios are summarized in
the following subsections. For this evaluation, exposure characteristics were selected to
calculate average (CTE) and upper-bound (RME) estimates of exposure for the scenarios of
interest. These two estimates were derived by varying only the assumed intake and ingestion
rates for an individual; the values remained the same for both CTE and RME estimates for other
exposure factors (i.e., body weight, exposure frequency, fraction of food from contaminated
sources, and cooking loss). Exposure factors were obtained primarily from EPA's Exposure
Factors Handbook (EPA 1997a, b).
Average ingestion rates used to calculate the CTE estimate were based on mean values
reported for relevant individuals. For the RME estimate, 90th percentile ingestion rates were
used for all food types assumed to be eaten. We realize that such an assumption can lead to a
total food ingestion rate that is extreme; for example, the total amount of food consumed per
day is nearly 6 kg for the farmer exposure scenario if 90th percentile ingestion rates from the
selected data set are assumed for all produce and meat/animal products. One approach to
developing a more realistic estimate of RME (that still evaluates high-end exposures) is to
assume that the exposed individual consumes food at the upper percentile ingestion rate only
for the one or two food types that dictate the total exposure to each chemical based on the
assumptions included in the current assessment, and ingestion rates for other FFC media or
fish are equal to mean reported values. For dioxins, consumption of fish, beef, and dairy
products are the food types that drive the long-term exposures. For mercury, the food types
that dictate exposures vary by chemical species, with beef/dairy consumption driving exposures
to Hg2, ingestion of soil and exposed vegetables and fruits driving exposures to HgO, and fish
consumption driving exposure to MeHg. However, if this approach is taken, the total exposure
(via all ingestion pathways) to each chemical evaluated was found to be roughly the same
regardless of whether the upper percentile ingestion rates are assumed for only the risk-driving
food types or for all food types. Therefore, to simplify the presentation of methods and results,
1-36
-------
we chose to assume ingestion at the 90th percentile for all consumed media (farm products, soil,
and/or fish) included in a scenario for the RME estimate.
Age of Exposed Individual
Exposures (i.e., ADDs) were calculated separately for four children's age groups: 1 to 2 years
(Child Group 1), 3 to 6 years (Child Group 2), 6 to 11 years (Child Group 3), and 12 to 19 years
(Child Group 4)) and for adults. Exposures were also calculated for infants less than 1 year of
age but only for the breast milk ingestion pathway. As described in Attachment C-2, these age
groupings were selected based on the availability of ingestion rate data for adults and children.
Body Weight
Body weights for each age group were estimated from data included in EPA's Exposure Factors
Handbook (EPA 1997a); body weights used were weighted means of the national distribution.
The body weights assumed for this assessment are presented in Exhibit 1.4-4. A single body
weight was used for each age group for all scenarios (i.e., separate RME and CTE estimates of
body weight were not evaluated).
Exhibit 1.4-4. Body Weight Estimates Used in This Assessment
Age of Exposed Individual
Mean Body Weight (kg)
Less than 1 year old
7.8 a
1 to 2 years
12.6 b
3 to 5 years
18.6 c
6 to 11 years
31.8 d
12 to 19 years
64.2 e
Adult
71.4 f
Nursing Mother
66.0 s
a Derived from time-weighted averages of body weights for age groups birth to <1 month, 1 to <3 months, 3 to <6
months, and 6 to <12 months from Table 8-3 of EPA 2008.
b Derived from time-weighted averages of body weights for age groups 1 to <2 years and 2 to <3 years from Table
8-3 of EPA 2008.
c Obtained directly from Table 8-3 of EPA 2008 (age group 3 to <6 years)
d Obtained directly from Table 8-3 of EPA 2008 (age group 6 to <11 years). Value represents a conservative (i.e.,
slightly low) estimate of BW for ages 6 through 11 years.
e Estimated using Table 8-22 of EPA 2008, based on NHANES IV data as presented in Portier et al. (2007). This
estimate was calculated as the average of the 8 single-year age groups from 12 to 13 years through 19 to 20 years.
f Derived from the sample-size weighted average of male and female mean body weights (all races, 18-74 years)
from EPA's 1997 EFH (Tables 7-4 for males and 7-5 for females).
g Used as the maternal body weight only in calculations for the breast milk exposure analysis. This value is from
EPA 2004d.
Intake and Ingestion Rates for Farmer Scenario
Mean and 90th percentile intake and ingestion rate inputs were used for adults and children for
the RME and CTE estimates, respectively. The rates used in this assessment are provided in
Exhibit 1.4-5, Exhibit 1.4-6, and Exhibit 1.4-7.
Values used for intake rates (in gwet weight/kgbody weight-day) in exposure calculations for the farmer
scenario are presented in Exhibit 1.4-5 for exposed fruit, protected fruit, exposed vegetables,
protected vegetables, root vegetables, beef, total dairy, pork, poultry, and eggs. As the units
suggest, the intake rates for produce and animal products are normalized for body weight.
1-37
-------
Exhibit 1.4-5. Age-Specific Ingestion Rates for the FFC Pathway
Product
Age Group
Units a
Child
< 1 yr
Child
1 -2 yrs
Child
3-5 yrs
Child
6-11 yrs
Child
12-19 yrs
Adult
RME - 90th Percentile Consumer Ingestion Rate
Farm Food Item (wet weight)
Beef
NA
4.5
6.7
11.4
3.53
5.39
g/kg-day
Dairy c
NA
148
82
54.7
27.0
34.9
g/kg-day
Eggsc
NA
5.1
2.8
1.80
1.34
1.65
g/kg-day
Exposed Fruit
NA
3.7
5.4
6.98
3.41
5
g/kg-day
Exposed Vegetable
NA
10.7
3.5
3.22
2.35
6.01
g/kg-day
Porkc
NA
4.5
4.4
3.04
2.65
3.08
g/kg-day
Poultry c
NA
7.4
6.8
4.58
3.28
3.47
g/kg-day
Protected Fruitc
NA
53
36
24.1
16.2
15.1 D
g/kg-day
Protected Vegetable
NA
3.9
2.5
2.14
1.85
3.55
g/kg-day
Root Vegetable
NA
7.3
4.3
3.83
2.26
3.11
g/kg-day
Other
Soil (dry weight)
NA
400°
400°
201 e
201 e
201 e
mg/day
Water (volume)
NA
654
834
980
1537
2224
mL/day
Fish (per individual)f
NA
3.24
4.79
6.9
8.95
17
g/day
Fish (per kg BW)
NA
0.26
0.26
0.22
0.14
0.24
g/kg-day
CTE - Mean Consumer Ingestion Rate
Farm Food Item (wet weight)
Beef
NA
1.49
2.21
3.77
1.72
2.63
g/kg-day
Dairy c
NA
67
37
24.8
10.9
17.1
g/kg-day
Eggsc
NA
2.5
1.4
0.86
0.61
0.90
g/kg-day
Exposed Fruit
NA
1.8
2.6
2.52
1.33
2.32
g/kg-day
Exposed Vegetable
NA
3.5
1.7
1.39
1.07
2.17
g/kg-day
Porkc
NA
2.2
2.1
1.49
1.17
1.30
g/kg-day
Poultryc
NA
3.6
3.4
2.13
1.59
1.54
g/kg-day
Protected Fruitc
NA
19
13
8.13
5.44
5.19 D
g/kg-day
Protected Vegetable
NA
2.5
1.3
1.1
0.78
1.3
g/kg-day
Root Vegetable
NA
2.5
1.3
1.32
0.94
1.39
g/kg-day
Other
Soil (dry weight)9
NA
50
50
50
50
50
mg/day
Water (volume)
NA
294
380
447
697
1,098
mL/day
Fish (per individual)f
NA
1.37
2.03
2.71
3.90
6.90
g/day
Fish (per kg BW)
NA
0.11
0.11
0.09
0.06
0.10
g/kg-day
Source: EPA 1997a (Chapter 13), unless otherwise noted in table notes. See Attachment C-2 for additional
information on sources. NA = not applicable.
a Ingestion rates for produce and animal products are normalized to consumer body weight. Ingestion rates for soil
(mg/day) and water (mL/day) are not normalized to body weight.
b This value represents a weighted average for the 20-39 and 40-69 age groups.
c In many cases, ingestion rates for certain child age groups were not available in EPA 1997a. Intakes for these
receptor groups were calculated using the methodology recommended in HHRAP (EPA 2005) (See Attachment C-2).
d This value represents an estimated "upper percentile" for children (EPA 1997a).
e Values represent soil ingestion rates for individuals consuming homegrown food products from Stanek et al. 1997.
f Adult fish ingestion rates are based on data from 1995-1996 and 1998 CSFII as summarized in EPA 2002b; child fish
ingestion rates are based on the same survey data, but estimated by multiplying average two-day consumption rate
for children who consumed fish on one or both days of the survey by the frequency offish consumption (i.e.,
proportion of children that reported consuming fish out of all children sampled).
g Represents CTE from EPA 2008, Table 5-1 for children and recommended mean value for adults from EPA 1997a,
Chapter 4, Table 4-23.
1-38
-------
Therefore, BW values presented in Exhibit 1.4-4 are not used in the farmer's ADD estimates in
this assessment. For estimating adult exposures for the farmer scenario, intake rates provided
in EPA's Exposure Factors Handbook associated with responses to the 1987-88 National Food
Consumption Survey (NFCS; USDA 1980) questionnaire (households who garden, households
who farm, or households who raise animals) were used instead of using the intake data
associated with ages 20 to 39, 40 to 69, and 70 plus years. This is a more conservative
estimate of intake, given that households that farm, garden, or raise animals will typically have a
higher level of intake than those that do not participate in those activities (EPA 1998). Incidental
soil ingestion rates are also presented in this table. For the RME exposure estimate, elevated
soil ingestion rates based on data for individuals who consume only home-grown produce was
used.
Fish Ingestion Rates for Angler Scenario
The daily fish consumption rates (in gfiSh/day) used to estimate exposures for the angler scenario
are presented in Exhibit 4-6. No site-specific fish consumption data were identified for the four
water bodies included in this risk assessment or the Ravena, NY, region. Instead, the 90th
percentile adult fish ingestion rates are based on data from 1995-1996 and 1998 CSFII as
summarized in EPA 2002b; child fish ingestion rates are based on the same survey data, but
estimated by multiplying average two-day consumption rate for children who consumed fish on
one or both days of the survey by the frequency of fish consumption (i.e., proportion of children
that reported consuming fish out of all children sampled). These values are discussed in detail
in Attachment 2 of Appendix C.
The adult mean consumption rate for is 6.9 g/day; the 90th percentile consumption rate offish is
17 g/day.
Exhibit 1.4-6. Fish Ingestion Rates for all Scenarios
Product
Age Group
Units 9
Child < 1
Year
Child 1-2
Years
Child 3-5
Years
Child 6-11
Years
Child 12-19
Years
Adult 20 -
70 years
Ingestion of Fish
Mean
g/kg/day
NAa
1.37
2.03
2.71
3.9
6.9
90th Percentile
g/kg/day
NAa
3.24
4.79
6.9
8.95
17
Source: EPA 2002b
a Infants are assumed to consume only breast milk for one year.
Additionally, Alcove Reservoir was closed to public use as a fishing destination in 1970;
therefore, it is unlikely that the conditions of the angler scenario (i.e., catching and consuming
fish on a regular, long-term basis) would be met. However, exposure estimates were calculated
for this lake as a check. Fish consumption advisories have been published by the State of New
York for Kinderhook and Nassau Lakes (NY DOH 2007). Although the efficacy offish
advisories has been questioned by some investigators, especially for certain ethnic groups that
are more likely to regularly consume self-caught fish,7 it is possible that the specific advisories
for these lakes may reduce the likelihood of regular, long-term consumption from these water
bodies. Finally, the pond included in this risk assessment is probably too small to support a
7 Studies have concluded that fish consumption advisories are often ineffective because anglers are not aware of
advisories (Burger 2000), anglers do not have knowledge of the contaminants, health effects, exposure, or risks
and therefore ignore the advisory (Jardine 2003; Burger et al. 1998; Beehler et al. 2003), and/or question the
credibility of agencies posting the advisories (May and Burger 1996).
1-39
-------
regularly-fished aquatic food web. These uncertainties are discussed in additional detail in the
discussion of results in Section 1-6.
Breast Milk Ingestion Rates for Nursing Infant Scenario
The breast milk ingestion rates (in kg/day) assumed for nursing infants are presented in Exhibit
1.4-7. These values were calculated using the EPA-recommended upper percentile breast milk
ingestion rate for infants nursing up to one year of 980 mL/day (EPA 1997b) and the density of
human milk (1.03 g/mL).
Exhibit I.4-7. Breast Milk Ingestion Rates for
Infants Less Than 1 Year of Age
Percentile
Breast Milk Ingestion Rate
(kg/day)
Mean
0.70864
90tn Percentile
1.0094
The mean ingestion rate was calculated by multiplying the EPA-recommended mean breast milk
ingestion rate for infants nursing up to one year of 688 mL/day (EPA 1997b, Table 14-16) by the
density of human milk (1.03 g/mL) and converting to kg/day, for a final value of 0.70864 kg/day.
The 90th percentile ingestion rate was calculated by multiplying the EPA-recommended upper
percentile breast milk ingestion rate for infants nursing up to one year of 980 mL/day (EPA
1997b, Table 14-16) by the density of human milk (1.03 g/mL) and converting to kg/day, for a
final value of 1.0094 kg/day. As the units suggest, the ingestion rates for breast milk (kg/day)
are not normalized for body weight.
Exposure Frequency
The exposure frequency (EF) represents the number of days per year that an individual
consumes farm food items that are contaminated with PB-HAPs. For the CTE and RME
evaluations, the exposure frequency was set to 365 days per year for all scenarios; a
conservative estimate for constant exposure throughout the year.
Fraction Contaminated
The fraction contaminated (FC) represents the fraction of each food product that is homegrown
(i.e., derived from the environment evaluated in this assessment). Individuals potentially
exposed to PB-HAPs in this evaluation were assumed to derive all potentially contaminated
foodstuffs from the modeled farm and watershed parcels. This means that for the CTE and
RME evaluations, the FC default for all food products was set to 1 (including the value for
nursing mothers used to calculate concentrations in breast milk).
Cooking Loss
Cooking loss (CL) inputs were included to simulate the amount of a food product that is not
ingested due to loss during preparation, cooking, or post-cooking. These inputs are detailed in
Appendix C-2 of the main report.
1-40
-------
1-5 Dose-Response Assessment and Estimation of Human Health
Risks
Estimates of exposure (as ADDs, discussed in Section 1-4) for each PB-HAP and dose-
response data were used to calculate excess lifetime cancer risk and non-cancer hazard for
chronic exposures. Exhibit 1.5-1 provides chemical dose-response data for the PB-HAPs
evaluated in this assessment.
The carcinogenic potency slope factors (CSFs) for ingestion and non-cancer oral reference
doses (RfDs) for chronic exposures for the PB-HAP included in this assessment are provided in
Exhibit 1.5-1. The dose-response values were obtained from tabulated dose-response data that
OAQPS uses for risk assessments of hazardous air pollutants (EPA 2007), and a detailed
discussion of these values is provided in Appendix C. In general, OAQPS chose these values
based on the following hierarchy of sources: EPA's Integrated Risk Information System (IRIS);
the Centers for Disease Control's Agency for Toxic Substances and Disease Registry (ATSDR);
and the California Environmental Protection Agency (CalEPA).
Exhibit 1.5-1. Dose-response Values for PB-HAPs Addressed in this Assessment
Chemical
CAS No.
Cancer Slope Factor
Source
l^kg - day J
Reference Dose
Source
^kg-aay J
Inorganics
Cadmium compounds in food
7440439
not available
1.0E-03
IRIS
Cadmium compounds in water
7440439
not available
5.0E-04
IRIS
Mercury (elemental)
7439976
NA
not available
Mercuric chloride
7487947
not available
3.0E-04
IRIS
Methyl mercury (MeHg)
22967926
not available
1.0E-04
IRIS
Organics
Benzo(a)pyrene
50328
1.0E+01
EPA OAQPS1
not available
2,3,7,8-TCDD
1746016
1.5E+05
EPA ORD
1.0E-09
ATSDR
ATSDR = Agency for Toxic Substances and Disease Registry IRIS = Integrated Risk Information System
EPA OAQPS = EPA's Office of Air Quality Planning and Standards NA = not applicable
EPA ORD = EPA's Office of Research and Development
1 The method to assign oral cancer slope factors for polycyclic organic matter (POM) is the same as was used in the
1999 National Air Toxics Assessment (EPA 1999b). A complete description of the methodology is available at:
http://www.epa.gov/ttn/atw/nata1999/99pdfs/pomapproachjan.pdf.
Sources: Values presented here are consistent with those defined by OAQPS for evaluation of HAPs (EPA 2007).
Sources listed in this table are the original references cited by EPA. For more information and the original
references that provide the derivation of these dose-response values, refer to EPA 2007.
i-6 Results and Discussion
1-6.1 Ravena Human Health Multipathway Risk Assessment Results
The results of the human health risk assessment are presented in this section. Section I-6.2
focuses on the results for 2,3,7,8-TCDD (as a representative of total dioxins), Section I-6.3
1-41
-------
focuses on the results for mercury, and Section 1-6.4 present the results associated with an
alternative modeling scenario in which a fish harvester is introduced into Ravena Pond.
For both dioxin and mercury, the concentrations and human health risks estimated in this
assessment are compared to analogous outputs estimated using the hypothetical multipathway
screening scenario developed for RTR. To accomplish this comparison, the Ravena emission
rates were modeled in the TRIM.FaTE screening scenario layout that is used to derive the RTR
de minimis levels for screening. This comparison helps to illustrate the level of conservatism
associated with the screening scenario and provides additional context for the results estimated
for this site-specific risk assessment. Throughout the multipathway HHRA discussion, the
results of modeling the Ravena emissions in the screening scenario are labeled "Screening
Scenario."
In general, the presentation of results here favors those calculated using reasonable maximum
exposure (RME) ingestion rates that are unlikely to occur but are still within the bounds of what
is possible. Exposures and risks calculated using more typical, central tendency exposure
(CTE) ingestion rates for these scenarios are also presented for comparison in some cases.
Note that most graphs display results plotted on a logarithmic scale.
1-6.2 2,3,7,8-TCDD
The Ravena site-specific TRIM.FaTE scenario was run for 50 years using two emission rates for
2,3,7,8-TCDD and assuming constant emissions for the duration of the simulation. The first
emission rate was calculated using a mean emission factor, and the second rate was calculated
based on the 95-percent upper confidence limit (UCL) of the dioxin emission factor, to provide
an upper bound risk estimate that takes into account the uncertainty regarding the emissions
estimate. A discussion of the derivation of the emission factors is presented in Appendix F.
Media concentrations and risks were estimated for both emission rates; however, results
presented in this section are primarily calculated using the 95-percent UCL emission rate.
1-6.2.1 Estimated Media Concentrations
Exhibit 1.6-2 presents a time series of dioxin concentrations in selected compartments modeled
by TRIM.FaTE using the 95-percent UCL emission rate. Included here are annual average
results for water column carnivores in the Ravena Pond and Alcove Reservoir compartments,
for surface water in the Ravena Pond and the Alcove Reservoir compartments, and surface soil
in the West Farm compartment (the layout of the Ravena TRIM.FaTE scenario is previously
described in Section I-3.4).
1-42
-------
Exhibit 1.6-2. 2,3,7,8-TCDD Media Concentration Time Series Using 95% UCL Dioxin
Emission Rate
1.E-05
o> 1. E-06
E
% 1.E-11
13
5
|*1.E-12
O)
E
« 1.E-13
1.E-14
5 1.E-07
1. E-08
5 >< 1. E-09
W
¦Water Column
Carnivore,
Ravena Pond
¦Water Column
Carnivore,
Alcove
Reservoir
Surface Soil,
West Farm
Unfilled
Surface Soil,
West Farm
Tilled
¦Surface
Water,
Ravena Pond
¦Surface
Water, Alcove
Reservoir
5 10 15 20 25 30 35 40 45 50
Year
As can be seen in Exhibit 1.6-2, concentrations typically increase rapidly over the first fifteen
years and then level off, increasing more slowly over the remainder of the model duration as the
concentration approached a steady-state. For the current analysis, risks were calculated based
on modeled environmental concentrations after fifty years of continuous emissions. This fifty
year period accounts for future impacts from long-term emissions at the assumed rate for the
modeled source.
Annually-averaged air and surface soil concentrations from the 50th model year for each farming
parcel in the Ravena scenario are shown in Exhibit 1.6-3 along with the dry particle deposition
rates to these locations from the final year. Concentrations presented here were modeled using
the 95-percent UCL emission rate, and these concentrations were later used to estimate
concentrations in farm media assumed to be ingested in this scenario. For comparison,
annually-averaged concentrations and deposition rates during the final year of a 50 year run of
the generic RTR screening scenario (run with the same Ravena dioxin emissions rate as the
case study evaluation) are also shown in this chart.
1-43
-------
Exhibit 1.6-3. 2,3,7,8-TCDD Air and Surface Soil Concentrations and Dry Particle
Deposition Rates During the 50th Model Year Using 95-Percent UCL Emission Rate
¦ Air concentration
¦ Soil concentration A Dry particle deposition
1.E-06
1E-11
A
^ 1.E-07
i *
j= -a
o> O)
S "3)
c a
11
C 5
s =
c a>
o c
« o
.!= o
< =
o
OT
£ 1.E-08
^A
1.E-09
1.E-10
A
A
1E-12
1E-13
1E-14 C1
1E-15
East Farm Tilled
East Farm
Unfilled
West Farm
Tilled
West Farm
Unfilled
Screening
Scenario Tilled
Screening
Scenario
Unfilled
In all cases, 2,3,7,8-TCDD surface soil concentrations are higher in the untilled plots than the
tilled plots, with the eastern untilled compartment having the higher concentration of the two
Ravena farm compartments. The chemical mass in the tilled and untilled compartments at a
given location is roughly comparable, but the volume in which the tilled chemical mass is mixed
is larger, resulting in a lower overall concentration. Comparison of the Ravena and screening
scenario soil concentrations for one type of soil compartment (i.e., either tilled or untilled)
illustrates a decrease of about an order of magnitude when site-specific characteristics
(including meteorology and proximity of the farm parcel) are included in the model.
Air concentrations are similar over the Ravena farm compartments. Unlike surface soil
concentrations, dry deposition values tend to be slightly higher in the western tilled soil than in
the eastern tilled soil. This difference likely reflects patterns of wind direction and precipitation
in the data set used for this model scenario.
Annually-averaged surface water concentrations estimated for the 50th year of the simulation for
each water body in the Ravena scenario are shown in Exhibit 1.6-4. The results of modeling the
Ravena dioxin emission rates in the RTR screening scenario are also included for comparison.
1-44
-------
Exhibit 1.6-4. 2,3,7,8-TCDD Surface Water Concentrations During the 50th Model Year
Using Mean and 95-Percent UCL Emission Rates
1.0E-10
¦ Mean Emission Rate
¦ 95% UCL Emission Rate
1.0E-11
at
E
c
I 1.0E-12
c
01
o
c
o
O
1.0E-13
1.0E-14
Alcove Reservoir
Kinderhook Lake
Nassau Lake
Ravena Pond
Screening Scenario
Pond
Water Body
Dioxin concentrations in surface water
were estimated to be highest in the
Ravena Pond, where concentrations were
slightly higher in the screening scenario
pond for a given emission rate. The lowest
concentrations were estimated in Alcove
Reservoir. This result is reasonable given
that Alcove Reservoir is the farthest from
the source of the four water bodies
modeled in the Ravena scenario.
Concentrations of 2,3,7,8-TCDD for each
fish compartment for all water bodies
containing fish in the Ravena and RTR
screening scenarios are presented in
Exhibit 1.6-5. TRIM.FaTE includes
compartments for benthic invertebrates,
benthic omnivores, benthic carnivores,
water column herbivores, water column omnivores, and water column carnivores. The
accompanying text box indicates fish species assumed to be present in the Ravena area water
bodies that are represented by TRIM.FaTE fish compartment types. A complete description of
the development of the aquatic food webs for the Ravena scenario is included in Attachment 1-1.
Fish Species Modeled by the TRIM.FaTE Fish
Compartments
Water Column Herbivore: Black crappie, common
carp, fantail darter, golden shiner, and young of the
year
Benthic Omnivore: Bullhead and sunfish
Water Column Omnivore: Bluegill, pumpkinseed,
redbreast sunfish, rock bass, smallmouth bass,
white perch, white sucker, and yellow perch
Benthic Carnivore: American eel
Water Column Carnivore: Chain pickerel,
largemouth bass, northern pike, tiger musky, and
walleye
1-45
-------
Exhibit 1.6-5. 2,3,7,8-TCDD Concentration in Fish Species During the 50th Model Year
Using the 95-Percent UCL Emission Rate
-•-BO -B-WCH -»-WCO -*-BC -A-WCC
1.E-05 n
1.E-06
C
o
+¦»
2
§ f 1.E-07
o >
r
o «
o >
Q
Q =3,
O ?
h £. 1.E-08
00
r-T
CO
cvT
1.E-09
1.E-10
Alcove Reservoir Kinderhook Lake Nassau Lake Screening Scenario Ravena Pond
Pond
Water Body
Results by location correspond generally to surface water concentration trends. Water column
carnivores and benthic carnivores were estimated to have the highest concentrations of 2,3,7,8-
TCDD in all water bodies except the Ravena Pond. The Ravena Pond does not contain benthic
carnivores, and so water column carnivores and benthic omnivores, the two largest types of fish
in this water body, had the highest concentrations.
Fish that are higher on the food chain (carnivores and omnivores, compared to herbivores and
invertebrates) tend to have higher concentrations, reflecting the biomagnification that occurs for
dioxins in an aquatic food chain. The carnivores and omnivores in the Ravena Pond and the
Screening Scenario Pond have the highest concentrations, consistent with the surface water
concentration trends.
I-6.2.2 Comparison of Modeled Surface Water Concentrations to Measured Values
In Exhibit 1.6-6 estimated fish concentrations for the water bodies modeled in the Ravena
screening scenario are compared to measured values for the Hudson River and nearby bays
(HSF 2007). Note that the concentrations are plotted on a logarithmic scale. For the measured
values, the environmental data were ranked (separately for the river and for bays) to create a
distribution. The modeled concentrations for the Ravena case study corresponding to both the
mean and 95 percent UCL emission factors are presented as two separate series and do not
correspond to the percentiles on the x-axis.
1-46
-------
Fish concentrations measured in the Hudson River and Bay ranged from 0.08 to 600 pg/g wet
weight. By comparison, in the modeled Ravena water bodies, estimated concentrations in fish
ranged from 9E-05 to 6 pg/g wet weight. This outcome seems reasonable because the model
includes only a single source of chemical emissions to the air, while the reported values reflect
all local and regional sources of dioxins, existing background concentrations of dioxins from
historical air sources, and any contributions from non-air sources (likely including historical PCB
contamination introduced to the Hudson River).
Exhibit 1.6-6. Modeled 2,3,7,8-TCDD Concentrations Compared to Measured Values
1000
0.001
0.0001
0.00001
0%
20%
40% 60% 80%
Measurement Percentiles
Measured River
Values (n=325)
Measured Bay
Values (n=134)
~
Modeled TCDD,
Mean Emission
Rate
~
Modeled TCDD,
95% UCL
Emission Rate
Sources for measured
fish concentrations:
Contaminant Assessment
Reduction Program (CARP)
database for bay values and
some measured river values
NOAA for some measured
river values
Notes:
~ = Ravena Pond
¦ = All other water bodies
100%
I-6.2.3 2,3,7,8-TCDD Risk Assessment Results
The annually-averaged concentrations for the 50th year estimated by TRIM.FaTE were used to
calculate lifetime individual cancer risks and chronic non-cancer hazard quotients (HQs) based
on assumptions of ingestion rates for fish, dairy, beef, and other foods. Ingestion rates were
assumed to vary by age, and exposures were calculated for five age groups. Because dioxin
exposure can occur via consumption breast milk by nursing infants, non-cancer hazards in this
subpopulation were also evaluated.
Different combinations and variations of results for the two main exposure scenarios are
presented, including angler-only (ingestion only offish), farmer-only (ingestion only of dairy,
beef, and other), and combined scenarios created for each unique combination of farm parcel
and water body. For the combined scenarios, the scenario location combination with the lowest
risks (the East Farm and the Alcove Reservoir), the highest risks (the West Farm and the
Ravena Pond), and the second highest risks (the West Farm and Nassau Lake) are typically
1-47
-------
presented. The West Farm/Nassau Lake scenario is expected to provide a more realistic high-
end estimate than the West farm/Ravena Pond combination.
1-6.2.3.1 2,3,7,8-TCDD Estimated Lifetime Cancer Risks
Exhibit 1.6-7 below presents the individual lifetime cancer risks for modeled scenarios. The one-
in-a-million and 100-in-a-million cancer risk thresholds are highlighted for reference. The results
were calculated using the TRIM.FaTE results from both the mean and 95-percent UCL emission
rates, and use the 90th percentile RME ingestion rates for all age groups. Only the angler
exposure scenario calculated with fish consumption from Alcove Reservoir yields a lifetime risk
of less than one in a million.
Exhibit 1.6-7. 2,3,7,8-TCDD Individual Lifetime Cancer Risks
1.E-03
~ Mean Emission Rate
~ 95% UCL Emission Rate
1.E-04
,= 1.E-05
rc
3
¦g
">
¦a
- 1.E-06
1.E-07
Alcove
Nassau
Ravena
East Farm
West Farm
East Farm,
West Farm,
West Farm,
Screening
Reservoir
Lake
Pond
Alcove
Nassau
Ravena
Scenario
Reservoir
Lake
Pond
Angler Only
Farmer Only
Combined Scenarios
Note: Presented results assume 90th percentile ingestion rates for all age groups (RME).
the yellow lines mark a risk of 1 in 1 million (1e-6) and of 1 in 10,000 (1e-4).
For the reader's reference,
Exhibit 1.6-8 presents the pathway contribution to dioxin exposures for all combined
farmer/angler scenarios, including results from the screening scenario. In this chart, exposure
scenarios are presented from left to right in order of increasing cancer risk. For the scenario
combination at the Ravena site with the second highest exposures (i.e., consumption of farm
products from the West Farm location and consumption offish caught in Nassau Lake),
consumption of fish is the dominant exposure pathway contributing to overall risk. Conversely,
for the lowest exposure Ravena scenario (i.e., consumption of foodstuffs from East Farm and
Alcove Reservoir), consumption of dairy contributes more to overall risk than exposure to dioxin
via consumption of fish.
1-48
-------
Exhibit 1.6-8. Pathway Contributions to 2,3,7,8-TCDD Individual Lifetime Cancer Risks
100%
80%
>
TO
| 60%
IS
Q.
c
0
1 40%
(9
20%
0%
73%
East Farm
(Risk=2.7E-06)
(Increasing Lifetime Cancer Risk)
119%
66%
9%
B9 WS
34%
54%
East Farm,
Alcove Reservoir
(Risk=3.0E-06)
32%
56%
West Farm,
Nassau Lake
(Risk=6.4E-06)
ingestion Scenario
Screening
Scenario
(Risk=1.1E-04)
West Farm,
Ravena Pond
(Risk=1.7E-04)
~ Soil
¦ Fruits &
Vegetables
~ Egg, Pork,
& Poultry
¦ Beef
~ Dairy
~ Fish
Note: Presented results assume fish harvesting from the Ravena Pond, 95-percent UCL emission rate, and 90th
percentile ingestion rates for all age groups (RME).
Exhibit 1.6-9 presents lifetime cancer risks for each combination of emission rate, exposure
scenario, and ingestion rate (some of these results are included in the two preceding charts).
Estimated individual lifetime cancer risks that exceed one in a million are highlighted in blue.
Only the CTE ingestion rate combined with the mean emission rate for the combined scenarios
using Kinderhook Lake or Alcove Reservoir and either farm parcel produced estimated cancer
risks below the one-in-a-million threshold.
1-49
-------
Exhibit 1.6-9. 2,3,7,8-TCDD Individual Lifetime Cancer Risks
Scenario Type
Water Body
Farm Parcel
95% UCL Emission Rate
Mean Emission Rate
Ratios
RME
Ingestion
Rate
CTE
Ingestion
Rate
RME
Ingestion
Rate
CTE
Ingestion
Rate
Ingestion
RME:CTE
Emissions
95% : Mean
Screening
Screening
Screening
1.1E-04
4.8E-05
4.6E-05
2.0E-05
2.3
2.4
Combined
Farmer and Angler
Ravena Pond
West
1.7E-04
6.8E-05
6.8E-05
2.8E-05
2.4
2.4
East
1.7E-04
6.8E-05
6.8E-05
2.8E-05
2.4
2.4
Nassau Lake
West
6.4E-06
2.8E-06
2.6E-06
1.1E-06
2.3
2.4
East
6.2E-06
2.7E-06
2.5E-06
1.1E-06
2.4
2.4
Kinderhook Lake
West
4.0E-06
1.8E-06
1.6E-06
7.4E-07
2.2
2.4
East
3.8E-06
1.7E-06
1.6E-06
7.1E-07
2.2
2.4
Alcove Reservoir
West
3.2E-06
1.5E-06
1.3E-06
6.0E-07
2.3
2.4
East
3.0E-06
1.4E-06
1.2E-06
5.7E-07
2.4
2.4
Farmer Only
None
West
2.9E-06
1.4E-06
1.2E-06
5.5E-07
2.2
2.4
East
2.7E-06
1.3E-06
1.1E-06
5.2E-07
2.2
2.4
Angler Only
Ravena Pond
None
1.6E-04
6.7E-05
6.7E-05
2.7E-05
2.4
2.4
Nassau Lake
None
3.4E-06
1.4E-06
1.4E-06
5.8E-07
2.4
2.4
Kinderhook Lake
None
1.1E-06
4.5E-07
4.6E-07
1.9E-07
2.4
2.4
Alcove Reservoir
None
2.8E-07
1.2E-07
1.2E-07
4.7E-08
2.4
2.4
Water Ingestion Only
Alcove Reservoir
None
1.3E-13
6.2E-14
5.2E-14
2.5E-14
2.1
2.4
1-50
-------
Cancer risk is between 2- and 2.5-times greater when assuming RME ingestion instead of CTE
ingestion rates. In all cases, using the 95-percent UCL emission rate resulted in cancer risks
approximately 2.4 times greater than using the mean emission rate. Because the emission
rates are run through an identical TRIM.FaTE screening scenario and farm-food chain, it is
reasonable that risks are exactly proportional to the emission rate used. The 95th emission
factor for the modeled facility is 2.4 times greater than the mean emission factor (3.23 g/yr vs.
1.32 g/yr, respectively), and this relationship is maintained throughout the scenario.
In all cases estimated cancer risks are slightly higher when products from the western farm
parcel are consumed rather than the eastern farm parcel. In the special cases where only
consumption of farm food products is considered, risks are approximately 7 percent higher
using the western farm than the eastern farm. When only fish consumption is considered, risks
are more than 400 times greater when fish are consumed from the Ravena Pond than when fish
are consumed from Alcove Reservoir. Risks are approximately 12 times greater when fish are
consumed only from Nassau Lake compared to estimated risks associated with fish
consumption from only Alcove Reservoir.
I-6.2.3.2 2,3,7,8-TCDD Chronic Non-Cancer Hazard Quotients
Chronic non-cancer hazard quotients (HQs) for dioxins were also estimated using modeled
TRIM.FaTE environmental concentrations and exposure estimates. Results are presented for
all age groups in Exhibit 1.6-10 corresponding to 95-percent UCL emission rates, RME ingestion
rates, individual and combined farmer and angler scenarios, both farm locations, and three of
the water bodies. The highest HQs were estimated for children aged 1 to 2, and the lowest HQs
were estimated for adolescents aged 12 to 19, with differences between age groups dictated by
age-specific ingestion rates of farm food products and fish.
2,3,7,8-TCDD HQs for Ravena scenarios not including the pond are generally at least an order
of magnitude lower than 1. When the Ravena emission rates were modeled in the screening
scenario layout the calculated HQs for Child 1-2 and Child 3-5 were both greater than 1. The
application of site-specific parameters reduced these HQs and illustrates the conservative
nature of the screening scenario. If fish from the Ravena Pond are assumed to be consumed,
the HQs were always estimated to be above 1.
The difference in HQs between the CTE and RME (i.e., mean vs. 90th percentile ingestion rates;
results not shown) is, on average, a factor of 2. The spread between the lowest exposure
scenario (Alcove Reservoir with the East Farm) and the second highest exposure scenario
(Nassau Lake and West Farm) is higher when the 90th percentile ingestion rates are used, with
the difference being approximately a factor of 2. When mean ingestion rates are used, there is
a factor of 2 difference in the HQs for these two scenarios. As in the case with cancer risks,
there is a factor of 2.4 difference in HQs when the RME ingestion rates are used instead of the
CTE rates.
Tables of all results for the calculated HQs for 2,3,7,8-TCDD are located in Attachment I-2.
1-51
-------
Exhibit 1.6-10. 2,3,7,8-TCDD Chronic Non-cancer Hazard Quotients
(95th Percentile UCL Emission Factor, RME Ingestion Rates)
I Child 1-2
Child 3-5
I Child 6-11
Child 12-19
I Adult 20-70
10
¦5
3
a
¦g
A
N
RJ
I
0.1
0.01
0.001
0.0001
0.00001
ia
m
¦PI
PI
N
irriri
East Farm
West Farm
Alcove
Reservoir
Nassau Lake
Ravena Pond
East Farm,
Alcove
Reservoir
West Farm,
Nassau Lake
West Farm,
Ravena Pond
Farmer Only
Angler Only
Combined Scenarios
Screening
Scenario
Exhibit 1.6-11 presents the contribution of each ingestion exposure pathway to the overall HQ for
children aged 1-2 and adults. The scenarios examined in the Ravena case study are displayed
in order of increasing HQ, with the location resulting in the lowest HQ displayed at the far left
and the location combination resulting in the highest HQ at the far right.
Modeled fish concentrations in Nassau Lake are approximately an order of magnitude greater
than those in Alcove Reservoir, and, for the West Farm/Nassau Lake scenario, fish
consumption is the risk-driving pathway for all age groups. The influence of the fish
consumption pathway on the HQ increases with age while that of dairy consumption decreases
with age, reflecting the relative magnitude of the ingestion rate for each food type for different
age groups (i.e., children are assumed to consume more dairy products per kg body weight,
while adults are assumed to consume more fish per kg body weight). For the East Farm/Alcove
Reservoir scenario, dairy consumption is the primary risk driver. This pathway has the most
influence on exposures in the youngest age group, with the influence of the dairy pathway
decreasing with age.
1-52
-------
Exhibit 1.6-11. Pathway Contributions to Divalent Mercury Chronic Non-Cancer Hazard
Quotients (95th Percentile UCL Emission Factor, RME Ingestion Rates)
~ Fish ~ Dairy DBeef ~ Egg, Pork, & Poultry ~ Fruits & Vegetables DSoil
Increasing HQ, Child 1-2 Increasing HQ, Adults
100%
9%
9% I
20%
23%
80%
28%
30%
TO
^ 60%
60%
63%
99%
O
c
o
88%
96%
91%
63%
40%
o
ra
71%
u_
61%
59%
20%
33%
31%
11%
0%
East Farm
(HQ=0.05)
East Farm,
AR
(HQ=0.06)
West Farm,
NL
(HQ=0.08)
Child 1-2
West Farm,
RP
(HQ=1.3)
Screening
Scenario
(HQ=1.5)
East Farm
(HQ=0.016)
East Farm,
AR
(HQ=0.018)
West Farm,
NL
(HQ=0.042)
Adult
Screening
Scenario
(HQ=0.74)
West Farm,
RP
(HQ=1.2)
Ingestion Scenario
Ar=Alcove Reservoir; NL=Nassau Lake; RP=Ravena Pond
I-6.2.3.3 2,3,7,8-TCDD Risks and Hazard Quotients Resulting from Dermal Exposure
Non-inhalation exposure to PB-HAPs can occur by way of the dermal pathway through contact
with PB-HAP-contaminated soil and water. However, as discussed in Appendix C of the main
report, dermal absorption of chemicals that are originally airborne is expected to be a relatively
minor pathway of exposure compared to other exposure pathways. The dermal cancer risks
and non-cancer HQs for 2,3,7,8-TCDD were calculated using the methodology discussed in
Appendix C using soil from each of the farm areas and water from Alcove Reservoir (assumed
to be the source of bathing water). The highest combined cancer risks occurred when the
calculations used soil concentrations from the unfilled East Farm compartment. Exhibit 1.6-12
below summarizes these risks assuming both the mean and 95th percentile emission factors.
Lifetime risks using the emission rate in the screening scenario are also shown for comparison.
A table providing risks and HQs for all age groups and farm compartments considered is
provided in Attachment I-2.
The maximum lifetime risk using the site-specific scenario with Alcove Reservoir is 4.9E-9, and
the ingestion risks are at least 200 times the dermal risks. The dermal calculations are highly
conservative and they are compared to the least conservative ingestion scenario (that is, the
mean ingestion rather than the 90th percentile ingestion). Because the dermal lifetime risks
remain over two orders of magnitude lower than the ingestion lifetime risks, dermal exposure is
1-53
-------
not a primary exposure pathway and does not strongly contribute to 2,3,7,8-TCDD lifetime risk
at this site.
Exhibit 1.6-12. Estimated Lifetime Cancer Risks Associated with Modeled Dermal
Exposure to 2,3,7,8-TCDD
Lifetime Risk
Emission
Rate
Farm
Water body
Soil
Water
Soil
arid
Water
Ratio of
Ingestion
to
Dermala
95% UCL
East Farm - Untilled
Alcove Reservoir
2.2E-09
2.7E-09
4.9E-09
>280
Emission
Rate
Screening Scenario
Screening
Scenario
3.9E-08
6.9E-07
7.3E-07
>60
Mean
East Farm - Untilled
Alcove Reservoir
9.0E-10
1.1E-09
2.0E-09
>590
Emission
Rate
Screening Scenario
Screening
Scenario
1.6E-08
2.8E-07
3.0E-07
>150
a Ratio compares total dermal HQ calculated based on RME parameters to total ingestion HQ based on CTE
ingestion rates (i.e., ratio of highest dermal HQ to lowest ingestion HQ).
Ingestion hazard quotients calculated using the mean emission rate and the mean ingestion rate
(i.e., the lowest estimated ingestion hazard quotients) are between approximately 50 and 150
times greater for than the highest estimated dermal hazard quotients (i.e., the HQs calculated
based on the 95-percent UCL emission rate). These results are included in Attachment 2.
I-6.2.3.4 Chronic Non-Cancer Hazard Quotients in Nursing Infants
Dioxin compounds ingested by a lactating mother can partition into breast milk and be passed to
a nursing infant. To evaluate the potential for hazard to a nursing infant at this facility, HQs
were estimated for an infant assumed to ingest breast milk from a woman exposed via
consumption of local farm products and fish during the duration of breast feeding and for nine
years prior to the birth of the child (see Attachment C-2). The infant's average daily ingestion
dose was calculated using the mother's average daily dose, partition factors and other
parameters that estimate the transfer of dioxins from the mother to breast milk, and exposure
factors for the nursing infant (See Attachment C-2, Section 3.4). This dose was then compared
to the same reference dose as the adult (rather than an infant-specific one). Estimated HQs for
the mother and child for the different emission factors, ingestion rates, farms, and water bodies
are shown below in Exhibit 1.6-13. Hazard quotients greater than one are shown in boldface
type.
The infant's hazard quotients tend to be higher than the mother's hazard quotients by a factor of
22 for all cases examined. The infant hazard quotients exceed one when consumption of fish
by the mother from Ravena Pond and the Screening Pond is assumed for all emission factors,
ingestion rates, and farm combinations considered. The hazard quotients shown do not include
the modeled fish harvesting from the Ravena Pond (discussed in Section I-6.4). If fish
harvesting is modeled and exposures are calculated based on those concentrations, the hazard
quotients for the infant and mother decrease by 70-73%, but the infant hazard quotients remain
above one. For the more probable consumptions of fish from Alcove, Kinderhook, or Nassau
Lakes, the infant hazard quotient is lower than one for the highest emission factors and
ingestion rates.
1-54
-------
Exhibit 1.6-13. Mother and Infant non-cancer Hazard Quotients for 2,3,7,8-TCDD
TCDD Emission
Factor
Ingestion
Rate
Farm
Water body
Mother HQ
Infant HQ
Alcove Reservoir
0.02
0.45
East
Kinderhook Lake
0.03
0.58
Nassau Lake
0.04
0.95
90th
Percentile
Ravena Pond
1.18
26.32
Alcove Reservoir
0.02
0.48
West
Kinderhook Lake
0.03
0.61
Nassau Lake
0.04
0.98
Ravena Pond
1.18
26.34
95th Percentile
Screening
Screening
0.77
17.32
UCL
Alcove Reservoir
0.01
0.21
East
Kinderhook Lake
0.01
0.26
Nassau Lake
0.02
0.41
Ravena Pond
0.48
10.77
Mean
Alcove Reservoir
0.01
0.22
West
Kinderhook Lake
0.01
0.27
Nassau Lake
0.02
0.43
Ravena Pond
0.48
10.78
Screening
Screening
0.33
7.47
Alcove Reservoir
0.01
0.19
East
Kinderhook Lake
0.01
0.24
Nassau Lake
0.02
0.39
90th
Percentile
Ravena Pond
0.48
10.77
Alcove Reservoir
0.01
0.20
West
Kinderhook Lake
0.01
0.25
Nassau Lake
0.02
0.40
Ravena Pond
0.48
10.78
Mean
Screening
Screening
0.32
7.08
Alcove Reservoir
0.00
0.09
East
Kinderhook Lake
0.00
0.11
Nassau Lake
0.01
0.17
Ravena Pond
0.20
4.41
Mean
Alcove Reservoir
0.00
0.09
West
Kinderhook Lake
0.01
0.11
Nassau Lake
0.01
0.17
Ravena Pond
0.20
4.41
Screening
Screening
0.14
3.05
1-55
-------
1-6.3 Mercury
Elemental and divalent mercury emitted by the Ravena portland cement facility were modeled
as individual species in TRIM.FaTE. Mercury emissions reported in NEI for the Ravena facility
were not specified with regard to the proportion of elemental and divalent species;
consequently, emissions were modeled as 75% elemental and 25% divalent (the default
mercury speciation for the portland cement source category).
Methyl mercury is created via transformation reactions in the environment, and TRIM.FaTE
includes transformation algorithms to model key transformation reactions. In the charts that
follow, results are presented for individual mercury species and total mercury (i.e., the sum of
mass or concentrations of the three modeled species). TRIM.FaTE mass and concentration
outputs for all modeled mercury species are reported by the model and expressed here in terms
of mass of mercury. For example, methyl mercury concentrations are expressed as "mercury
as methyl mercury" (where the reported value excludes the mass of the carbon and hydrogen
elements present in the compound). This convention is consistent with the way EPA has
defined the oral RfD for methyl mercury.
In most cases, mercury results are plotted on graphs with logarithmic scales.
1-6.3.1 Mercury Media Concentrations
Exhibit 1.6-14 presents a time series of total mercury concentrations for:
Fish species (water column carnivores) in the Ravena Pond and the Alcove
Reservoir compartments,
Surface water in the pond and the Alcove Reservoir, and
Tilled and unfilled surface soil in the western farm compartment.
Similar to the TRIM.FaTE results observed for 2,3,7,8-TCDD, concentrations increase rapidly
over the first fifteen years and then increase less rapidly for the remainder of the model run.
The long-term rate at which modeled mercury concentrations increase appears to be somewhat
higher than dioxin (in other words, in our modeling scenario, mercury concentrations do not
approach steady-state as quickly as dioxins). This is likely due to the fact that dioxin is
assumed to degrade in the environment (with a half-life on the order of years), while mercury
does not degrade. Processes that remove mercury from the modeled system include
volatilization of elemental mercury from soil and water, outflow of dissolved and suspended
sediment-borne mercury from lakes and the river, and other transfers, which collectively appear
to remove mercury at a slower overall rate than processes (including degradation) that affect
dioxin concentrations. Despite this difference, the rate of change of modeled mercury
concentrations is relatively low at the end of the 50-year model run.
1-56
-------
Exhibit 1.6-14. Total Mercury Media Concentration Time Series
1.E-08
Year
1.E+00
u
f 1.E-01
Water Column
Carnivore,
Ravena Pond
0)
IS
§
o
o
t
3
CO
•*-»
>
-a
at
O)
3
o
OT
1.E-02
1.E-03
1.E-04
1.E-05
1.E-06
1.E-07
Surface Soil,
West Farm
Untilled
Surface Soil,
West Farm
Tilled
Water Column
Carnivore,
Alcove
Reservoir
Surface
Water,
Ravena Pond
Surface
Water, Alcove
Reservoir
Elemental, divalent, and methyl mercury surface soil concentrations are presented in Exhibit I.6-
15. Divalent and methyl mercury concentrations (shown in blue and yellow, respectively) are
higher in untilled parcels than in tilled parcels, while elemental mercury concentrations (shown
in blue) are higher in tilled parcels. This trend occurs because elemental mercury is volatile, and
the TRIM.FaTE-estimated volatilization rate is dependent on the estimated vertical
concentration gradient of mercury in the soil. Though roughly the same mass of mercury is
mixed in the tilled and untilled compartments, the soil profile in the tilled parcels is deeper. As a
result, volatilization occurs more slowly and concentrations of elemental mercury are higher
(even though total mercury concentrations are lower).
Divalent mercury concentrations in soil are typically higher than elemental mercury at all
locations because this species deposits much more readily than elemental mercury (as reflected
by the mass distribution summary). Furthermore, the volatilization of elemental mercury
ensures that its residence time in the soil is short compared to divalent mercury. As with
2,3,7,8-TCDD, estimated soil concentrations are higher when the Ravena mercury emissions
are modeled in the RTR screening scenario than when the same emissions are modeled in the
site-specific Ravena layout. This is true for both the tilled and untilled compartments.
1-57
-------
Exhibit 1.6-15. Mercury Surface Soil Concentrations at 50th Model Year
1.E+01
1.E+00
1.E-01
1 -g 1.E-02
2 >
•*-> *-
C T3
o 5>
o = 1.E-03
O —
1.E-04
1.E-05
1.E-06
¦ Divalent Mercury
¦ Elemental Mercury
~ Methyl Mercury
East Farm
Tilled
East Farm
Untilled
West Farm
Tilled
West Farm
Untilled
Screening
Scenario
Tilled
Screening
Scenario
Untilled
Soil Parcel
Total mercury air concentrations (shown in blue), surface soil concentrations (in green), and dry
particle deposition rates (yellow triangles with values corresponding to the right y-axis) are
presented in Exhibit 1.6-16. Soil concentrations were estimated to be higher at the east farm
location while air concentrations and dry deposition rates were slightly higher at the west farm
location. Dry deposition rates are highest for the tilled west farm compartment. These air, soil,
and deposition values were used to calculate chemical concentrations in farm food chain media.
1-58
-------
Exhibit 1.6-16. Total Mercury Air and Surface Soil Concentrations and Dry Particle
Deposition Rates at 50th Model Year
a> *£
o c
c a>
o c
« o
± O
< =
o
OT
¦ Air concentration
¦ Soil concentration
A Dry particle deposition
10
5
j= -a
o) at
¦S "3)
c B
1.1
0.1
£ 0.01
0.001
0.0001
0.00001
K
A
A"
A
A
A
East Farm Tilled East Farm West Farm West Farm Screening Screening
Unfilled Tilled Unfilled Scenario Tilled Scenario
Unfilled
1.E-07
1.E-08
1.E-09 •=
1.E-10 w
o
0)
o
1.E-11 t
1.E-12
1.E-13
Exhibit 1.6-17 shows the surface water concentrations in the Ravena water bodies and in the
Screening Scenario Pond for all mercury species. Concentrations are generally similar across
all three of the large Ravena water bodies. The concentrations in Ravena Pond and the
Screening Scenario Pond are approximately equivalent, with mercury levels far exceeding those
in the other bodies of water. This result is discussed further in Section I-6.4.
1-59
-------
Exhibit 1.6-17. Mercury Surface Water Concentrations During the 50th Model Year
1.E-03
1.E-04
1.E-05
m t
c o> 1.E-06
01 £
o S.
c
o
o
1.E-07
1.E-08
1.E-09
¦ Divalent Mercury
¦ Elemental Mercury
~ Methyl Mercury
Alcove Reservoir Kinderhook Lake Nassau Lake
Ravena Pond
Screening Scenario
Pond
Water Body
Exhibit 1.6-18 illustrates the total mercury concentration in different aquatic species for all water
bodies. Note that the lines in Exhibit 1.6-18 do not indicate trends but rather are included to
assist the reader in making comparisons between concentrations in the water bodies and fish
compartments. Fish in the Ravena Pond and Screening Scenario Pond have the highest total
mercury concentrations, with some results for the pond almost 100 times higher than the larger
lakes and reservoir. Note that the dimensions of the Ravena Pond do not enable it to support
water column omnivores and benthic carnivores (see Section I-6.4).
1-60
-------
Exhibit 1.6-18. Total Mercury Concentration in Fish Species During the 50th Model Year
BO —WCH -• WCO —BC —A—WCC
1.E+00
.E-01
.E-02
O)
O)
.E-03
.E-04
Alcove Reservoir Kinderhook Lake Nassau Lake Screening Scenario Ravena Pond
Pond
Water Body
Exhibit 1.6-19 summarizes the mercury speciation in modeled environmental media. Elemental
mercury (shown in green) is the most predominant form of mercury in the air (roughly reflecting
the emission profile of the modeled source). Most of the mercury in soil and sediment is present
as divalent mercury (shown in blue). Methyl mercury (in yellow) is present in the modeled
aquatic biota compartments, and the fraction of methyl mercury in the aquatic biota
compartments (benthic invertebrates, water column herbivores, and water column carnivores)
increases with higher trophic levels. The water column carnivores have the highest fraction of
methyl mercury while the benthic invertebrates have the lowest.
1-61
-------
Exhibit 1.6-19. Mercury Speciation Across Different Model Compartments
I Divalent Mercury
I Elemental Mercury
Methyl Mercury
100%
¦g 50%
£ 30%
West Farm West Farm West Farm
Tilled U ntilled
Air
Surface Soil
Alcove
Nassau
Alcove
Nassau
Benthic
Reservoir
Reservoir
nvertebrate
Sediment
Surface Water
Water Water
Column Column
Herbivores Carnivores
Aquatic Biota in Nassau Lake
Exhibit 1.6-20 compares divalent, methyl, and total mercury concentrations for fish in all modeled
water bodies in the Ravena scenario to measured values for the Hudson River and surrounding
bays (HRF 2007). Note that the concentrations are plotted on a logarithmic scale. For the
measured values, the environmental data were ranked to show the distribution of values.
The modeled concentrations for the different mercury species are presented as three separate
series and do not correspond to the percentiles on the x-axis. The modeled concentrations from
the pond are shown with triangles and are the only values that overlap the measured
concentrations. All other modeled concentrations are shown with squares, and these
concentrations are lower than any of the measured concentrations. This was the expected
result since the measured values in the Hudson Bay include contributions from many sources,
whereas the Ravena water bodies reflect only the incremental impact of mercury emissions
from the Ravena Portland cement facility.
1-62
-------
Exhibit 1.6-20. Modeled Mercury Concentrations Compared to Measured Values
10000
CARP - Divalent
(n=623)
CARP - Methyl
(n=136)
1000
CARP - Total
(n=136)
Modeled Divalent
Mercury
100
Modeled Methyl
Mercury
O)
Modeled Total
Mercury
O)
O)
Source for measured
u_
fish concentrations:
Contaminant
Assessment Reduction
Program (CARP)
database
Notes:
~ = Ravena Pond
¦ = All other water
bodies
0.01
0%
20% 40% 60% 80%
Measured Percentiles for CARP Fish Tissue Concentrations Distributions
100%
I-6.3.2 Mercury Risk Assessment Results
As in the 2,3,7,8-TCDD analysis, the annually-averaged mercury concentrations from the 50th
year were used to derive chronic non-cancer HQs based on assumptions of ingestion rates for
fish, dairy, beef, and all other foods. Different combinations and variations of the two main
exposure scenarios were investigated, and the results presented here largely assume RME
(reasonable maximum exposure) ingestion rates. For the combined scenarios, the case
associated with the lowest HQ (the West Farm and the Alcove Reservoir) and the case
associated with the highest HQ (the East Farm and the Ravena Pond) are both presented,
along with the second highest HQ case for comparison.
1-6.3.2.1 Mercury Chronic Non-cancer Hazard Quotients
HQs were calculated separately for methyl and divalent mercury. The top graph in Exhibit I.6-
21 presents the estimated divalent mercury HQs using exposure scenarios for farmer ingestion,
angler ingestion, and combinations of these two pathways using RME ingestion rates. Methyl
mercury HQs are shown in the bottom graph in Exhibit 1.6-21. In all cases evaluated, estimated
HQs were less than 1.
1-63
-------
Exhibit 1.6-21. Mercury Chronic Non-Cancer Hazard Quotients for Ravena
Child 3-5
Divalent Mercury
I Child 6-11
Child 1-2
at 0.01
3 0.001
Child 12-19
I Adult 20-70
mm hi
0.0001
0.00001
m
West Farm
East Farm
Alcove
Kinderhook
Ravena Pond
West Farm,
East Farm,
East Farm,
Screening
Reservoir
Lake
Alcove
Kinderhook
Ravena Pond
Scenario
Reservoir
Lake
Farmer Only
Angler Only
Combined Scenarios
Methyl Mercury
¦ Child 1-2
¦ Child 3-5
¦ Child 6-11
E Child 12-19
¦ Adult 20-70
ai 0.01
ra 0.001
0.0001
0.00001
nnii 11
West Farm
East Farm
Alcove
Reservoir
Nassau Lake
Ravena Pond
West Farm,
Alcove
Reservoir
East Farm,
Nassau Lake
East Farm,
Ravena Pond
Farmer Only
Angler Only
Combined Scenarios
Screening
Scenario
Note: Presented results were calculated using the 90th percentile ingestion rates (RME).
1-64
-------
In both figures, hazard quotients are highest with children aged 1-2 and they decrease with age
as children grow older. Estimated hazard quotients for adults are greater than for all child age
groups except the child aged 1-2 group. HQs in the site-specific analysis are approximately an
order of magnitude lower than those for the Screening Scenario analysis. The HQs associated
with fish consumption from the Ravena Pond are an exception to this and are discussed in
further detail in Section I-6.4.
In all scenarios for all chemicals, ingestion of products from the East Farm results in higher HQs
than ingestion of products from the West Farm. The difference in HQs related to consumption
of products from the two farms is greater for divalent mercury than for methyl mercury; this
result makes sense because divalent mercury accumulates more in farm-food chain media than
in fish. Hazard quotients calculated based on consumption of fish from either Kinderhook or
Nassau Lakes are slightly larger than those based on fish consumption from Alcove Reservoir.
The percent contributions to HQ for each exposure pathway were also calculated for divalent
mercury and methyl mercury and are presented in Exhibit 1.6-22 and Exhibit 1.6-23, respectively.
The scenarios are displayed in order of increasing HQ, with the scenario resulting in the lowest
HQ displayed at the far left and the scenario resulting in the highest HQ at the far right.
Exhibit 1.6-22. Pathway Contributions to Divalent Mercury Chronic Non-Cancer Hazard
Quotients
¦ Fruits & Vegetables ~ Egg, Pork, & Poultry ~ Soil DFish BBeef&Dairy
100%
80%
>
ra
&
¦E
-*—«
ra
a.
i*—
o
c
o
60%
~ 40%
o
ra
20%
13%
23%
Increasing HQ, Child 1-2
Increasing HQ, Adults
13%
12%
West Farm
West Farm,
East Farm,
East Farm,
(HQ=0.0034)
AR
NL
RP
(HQ=0.0035)
(HQ=0.0041)
(HQ=0.0231)
Child 1-2
20%
19%
23%
Screening
Scenario
(HQ=0.3440)
Ingestion Scenario
91%
o
20%
12%
West Farm,
East Farm,
East Farm,
Screening
AR
NL
RP
Scenario
(HQ=0.0016)
(HQ=0.0019)
(HQ=0.0194)
(HQ=0.0870)
Adult
1-65
-------
Exhibit 1.6-23. Pathway Contributions to Methyl Mercury Chronic Non-Cancer Hazard
Quotients
¦ Fruits & Vegetables ~ Egg, Pork, & Poultry ~ Soil DFish ¦ Beef & Dairy
100%
80%
Increasing HQ, Child 1-2
Increasing HQ, Adults
>
ra
&
¦E
¦*—<
ra
a.
i*—
o
c
o
o
ra
60%
~ 40%
20%
13%
23%
13%
12%
West Farm
West Farm,
East Farm,
East Farm,
(HQ=0.0034)
AR
NL
RP
(HQ=0.0035)
(HQ=0.0041)
(HQ=0.0231)
Child 1-2
Screening
Scenario
20%
West Farm
(HQ=0.0015)
19%,
23%,
91%,
o
20%
12%
West Farm,
East Farm,
East Farm,
Screening
AR
NL
RP
Scenario
(HQ=0.0016)
(HQ=0.0019)
(HQ=0.0194)
(HQ=0.0870)
Adult
Ingestion Scenario
In general, consumption of soil is associated with the largest contribution to the hazard quotient
for divalent mercury in children aged 1-2, while consumption of fruits and vegetables is the
dominant exposure pathway for divalent mercury in adults. As expected, the ingestion of fish is
the dominant exposure pathway for methyl mercury in all age groups.
CTE (central tendency exposure) and 90th percentile RME ingestion rates were also evaluated
to better understand the differences in hazard quotients when using different ingestion rates.
Exhibit 1.6-24 presents the hazard quotient based on the 90th percentile ingestion rate and
hazard quotient based on the mean ingestion rate for various scenarios for children aged 1-2. A
ratio of hazard quotients using the 90th percentile ingestion rate and mean ingestion rate was
also calculated. In general, when using the 90th percentile ingestion rates for all ages, hazard
quotients are generally 2- to 3-times greater than the hazard quotients associated with the mean
ingestion rates. In many—but not all—cases, the ratio of the HQs decreases slightly as children
age.
1-66
-------
Exhibit 1.6-24. Comparison of Hazard Quotients for Ravena Scenario Using Mean and
90th Percentile Ingestion Rates
Scenario
Farm
Parcel
Water body
90th
Percentile
Ingestion
Rate, HQ
Child 1-2
Mean
Ingestion
Rate, HQ
Child 1-2
Ratio
90th :
Mean HQ
Divalent Mercury
Screening
Screening
Screening Pond
0.344
0.085
4.1
Combined
West
Alcove Reservoir
0.007
0.003
2.7
East
Kinderhook Lake
0.008
0.003
2.6
Ravena Pond
0.023
0.010
2.4
Farm Only
West
-
0.003
0.001
2.7
East
-
0.004
0.002
2.6
Fisherman Only
-
Alcove Reservoir
0.0001
0.00003
2.4
Kinderhook Lake
0.0001
0.00005
2.4
Ravena Pond
0.019
0.008
2.4
Water Ingestion Only
-
-
2.4E-08
1.1E-08
2.2
Methyl Mercury
Screening
Screening
Screening Pond
0.193
0.079
2.5
Combined
West
Alcove Reservoir
0.002
0.001
2.4
East
Nassau Lake
0.003
0.001
2.4
Ravena Pond
0.199
0.084
2.4
Farm Only
West
-
0.0002
0.0001
2.8
East
-
0.0002
0.0001
2.8
Fisherman Only
-
Alcove Reservoir
0.001
0.0003
2.4
Nassau Lake
0.001
0.001
2.4
Ravena Pond
0.198
0.084
2.4
Water Ingestion Only
-
-
1.0E-09
4.6E-10
2.2
Note: Values presented have been rounded
1-6.3.2.2 Mercury Chronic Non-Cancer Hazard Quotients from Dermal Exposure
Dermal exposures were assessed for divalent mercury using the methodology described in
Appendix C. The highest combined HQs occurred when the calculations used soil
concentrations from the unfilled East farm compartment for a child age 1-2. These HQs are
presented in Exhibit 1.6-25 along with HQs derived using the Ravena mercury emission rate in
the Screening Scenario. A table providing hazard quotients for all age groups and farm
compartments considered is provided in Attachment I-2 of this document.
The HQ derived from the site-specific scenario with Alcove Reservoir is 6.7E-3 for children aged
1-2. In this case, the dermal HQs are the same order of magnitude (and somewhat higher than)
the ingestion HQs. Thus, the dermal exposure pathway seems to be of equal importance as the
ingestion pathway for divalent mercury exposures at this site. However, all HQs are well below
1. It should be noted that the dermal calculations are based on highly conservative
assumptions including a high surface area in children that is exposed to soil.
1-67
-------
Exhibit 1.6-25. Divalent Mercury Dermal Hazard Quotients for a Child Aged 1-
2
Hazard Quotient
Farm
Water body
Soil
Water
Soil
and
Ratio of
Ingestion
to
Dermala
Water
East Farm - Untilled
Alcove Reservoir
6.7E-03
3.2E-06
6.7E-03
0.30
Screening
Screening
1.7E-01
1.3E-03
1.7E-01
0.50
a Ratio compares total dermal HQ calculated based on RME parameters to total ingestion HQ based
on CTE ingestion rates (i.e., ratio of highest dermal HQ to lowest ingestion HQ).
I-6.4 Alternate Modeling Scenario - Incorporation of Fish Harvesting from
Ravena Pond
During the development of the conceptual exposure model for the risk assessment of the
Ravena facility, we determined that the possibility of an angler exposure scenario existing for
the Ravena pond was low; however, the Ravena Pond was retained in this case study
evaluation to provide an additional "what-if" analysis. Juxtaposition of the exposure (i.e., fish
ingestion rates) and environmental assumptions associated with the angler-pond scenario
illustrates the implausibility of this scenario. Specifically, it is assumed that the angler fishes at
the pond regularly for a lifetime and consumes his or her catch. Due to the small size of the
pond (20,000 m2 in surface area and 1 m deep), it is unlikely that this water body could sustain
fishable populations at the assumed ingestion rates without regular, substantial restocking of
fish.
The total standing fish biomass in the pond was assumed to be 80 kg on average over the
course of a year. We assumed approximately 5 percent of this total (4 kg) was present as adult
bass, represented in the water column carnivore compartment, and 75 percent of this total (60
kg) was present as benthic omnivores such as catfish or sunfish. The angler is assumed to
consume these two types offish (bass and benthic omnivores) at a ratio of 1:2, with preferential
harvesting of bass. The 90th percentile, reasonable maximum exposure (RME) fish ingestion
rate for an adult angler was assumed to be 17 g/day.
Based on these exposure assumptions and using an exposure duration of 365 days/year, the
amount of bass harvested in one year would need to be about 2 kg (i.e., 17 g/day x 365 days/yr
x 0.33 of total consumption).8 This represents 50% of the standing biomass in the Ravena
Pond. Using the same exposure assumptions, the amount of benthic omnivores harvested
would be nearly 4 kg/yr, which is about 7% of the standing biomass. Note that these
calculations assume that only one single angler is fishing in the pond, and the associated risk
calculations assume that one single consumer eats all of the fish that are caught.
This situation appears not to be ecologically sustainable, and at a minimum is likely to
significantly reduce the chemical concentrations in fish tissues in all fish types. Thus, to obtain
a more realistic estimate of concentrations in fish in the Ravena Pond, we modified the
TRIM.FaTE scenario and incorporated a fish harvesting rate from the pond of 17 g/day to
represent consumption and restocking of the pond within the TRIM.FaTE model. This
harvesting rate corresponds to the 90th percentile fish ingestion rate for humans used to
8 We recognize that these calculations do not take into account other details (for example, the biomass and mass of
fish consumed are directly compared, even though the entire mass of a fish is not consumed), but the intent is a
subjective examination of the underlying assumptions.
1-68
-------
calculate hazard quotients and lifetime cancer risks associated with consumption of
contaminated fish.
Exhibit 1.6-26 compares concentrations of 2,3,7,8-TCDD and total mercury in various aquatic
compartments when TRIM.FaTE was run with and without modeled fish harvesting from the
Ravena Pond compartment. By incorporating fish harvesting within the TRIM.FaTE scenario,
estimated dioxin concentrations in fish consumed by humans decreased by 22% and 38%,and
estimated total mercury concentrations decreased by 5% and 39% in water column herbivores
and carnivores, respectively.
Exhibit 1.6-26. Effect of Fish Harvesting on Annually Averaged PB-HAP Concentrations
During the 50th Model Year in Ravena Pond Using 95-Percent UCL Emission Rate
Compartment
Units
Concentration
Percent Reduction in
With Fish
Harvesting
No Fish
Harvesting
Concentration With
Addition of Harvesting
2,3,7,8-TCDD
Surface Water
mg/L
1.3E-11
1.3E-11
0%
Sediment concentration
ug/g dry wt.
6.3E-08
6.3E-08
0%
Macrophyte
mg/kg wet wt.
1.9E-07
1.9E-07
0%
Benthic Invertebrate
mg/kg wet wt.
6.2E-09
6.2E-09
0%
Benthic Omnivore
mg/kg wet wt.
8.8E-07
8.8E-07
0%
Water Column Herbivore
mg/kg wet wt.
3.7E-06
4.7E-06
22%
Water Column Carnivore
mg/kg wet wt.
3.2E-06
5.2E-06
38%
Mallard
mg/kg wet wt.
3.1E-05
3.1E-05
0%
Mink3
mg/kg wet wt.
4.9E-09
-
-
Total Mercury
Surface Water
mg/L
1.2E-04
1.2E-04
0%
Sediment concentration
ug/g dry wt.
9.1E-01
9.1E-01
0%
Macrophyte
mg/kg wet wt.
5.4E-05
5.4E-05
0%
Benthic Invertebrate
mg/kg wet wt.
5.2E-02
5.2E-02
0%
Benthic Omnivore
mg/kg wet wt.
2.1E-01
2.1E-01
0%
Water Column Herbivore
mg/kg wet wt.
4.3E-02
4.5E-02
5%
Water Column Carnivore
mg/kg wet wt.
1.3E-01
2.1E-01
39%
Mallard
mg/kg wet wt.
3.7E-02
3.7E-02
0%
Mink3
mg/kg wet wt.
6.7E-05
-
-
a The harvester was modeled in the mammalian mink compartment because no human compartment exists in the
TRIM.FaTE modeling system.
Exhibit 1.6-27 presents the corresponding changes in 1) the lifetime cancer risk for 2,3,7,8—
TCDD and 2) the hazard quotient for a child age 1-2 for divalent and methyl mercury when fish
harvesting is included in the model scenario. The comparison reveals that in all cases, the
addition of fish harvesting from the Ravena Pond results in approximately a 27% decrease in
the individual lifetime cancer risk for 2,3,7,8 - TCDD, no matter which ingestion rates or
emission rates are used. This relatively modest change in magnitude is not enough to change
any conclusions with regard to typical risk thresholds (e.g., one-in-a-million or 100-in-a-million).
For mercury, risks are reduced more for the methyl mercury (34%) than divalent mercury (10-
12%), reflecting the fact that the fish pathway is more important for methyl mercury. However,
the hazard quotients are less than one both with and without the modeled fish harvesting.
1-69
-------
Exhibit 1.6-27. Risks and Hazard Quotients in Ravena Pond with and without Fish
Chemical
Ingestion
Rate
Scenario Type
Farm
Parcel
Risk/HQ
Without Fish
Harvesting
from Pond a
Risk/HQ
With Fish
Harvesting
from Pond a
Percent
Reduction in
Risk/HQ with
Addition of
Harvesting a
2,3,7,8-
TCDD, 95th
Emission
Factor
90th
Ingestion
Rates
Combined
West
1.7E-04
1.2E-04
27%
East
1.7E-04
1.2E-04
27%
Fisherman Only
N/A
1.6E-04
1.2E-04
27%
Mean
Ingestion
Rates
Combined
West
6.8E-05
5.0E-05
27%
East
6.8E-05
5.0E-05
27%
Fisherman Only
N/A
6.7E-05
4.9E-05
27%
2,3,7,8 —
TCDD,
Mean
Emission
Factor
90th
Ingestion
Rates
Combined
West
6.8E-05
5.0E-05
27%
East
6.8E-05
5.0E-05
27%
Fisherman Only
N/A
6.7E-05
4.9E-05
27%
Mean
Ingestion
Rates
Combined
West
2.8E-05
2.0E-05
27%
East
2.8E-05
2.0E-05
27%
Fisherman Only
N/A
2.7E-05
2.0E-05
27%
Divalent
Mercury
90th
Ingestion
Rates
Combined
West
2.0E-02
2.2E-02
10% b
East
2.1E-02
2.3E-02
10% b
Fisherman Only
N/A
1.7E-02
1.9E-02
12% b
Mean
Ingestion
Rates
Combined
West
8.4E-03
9.3E-03
10% b
East
8.6E-03
9.6E-03
10% b
Fisherman Only
N/A
7.1E-03
8.0E-03
12% b
Methyl
Mercury
90th
Ingestion
Rates
Combined
West
1.3E-01
2.0E-01
34% b
East
1.3E-01
2.0E-01
34% b
Fisherman Only
N/A
1.3E-01
2.0E-01
34% b
Mean
Ingestion
Rates
Combined
West
5.6E-02
8.4E-02
34% b
East
5.6E-02
8.4E-02
34% b
Fisherman Only
N/A
5.6E-02
8.4E-02
34% b
Represents the lifetime cancer risk for 2,3,7,8 - TCDD and the hazard quotient for a child age 1-2 for divalent
mercury and methyl mercury.
b The change in hazard quotients with and without the harvester was also compared for children of other age
groups and adults, and the percent reduction in risk was found to be broadly consistent across these age groups.
1-70
-------
1-7 References
ATSDR. 1992. Toxicological Profile for Selected PCBs. US Department of Health and Human
Services. Atlanta, GA. TP-92/16.
Beehler, G.P., B.M. McGuiness, and J.E. Vena. 2003. Characterizing Latino Anglers'
Environmental Risk Perceptions, Sport Fish Consumption, and Advisory Awareness. Medical
Anthropology Quarterly 17(1): 99-116.
Burger, J., J. Sanchez, and M. Gochfeld. 1998. Fishing, consumption and risk perception in
fisherfolk along an east Coast Estuary. Environmental Research 77: 25-35.
Burger, J. 2000. Consumption advisories and compliance: The fishing public and the
deamplification of risk. Journal of Environmental Planning and Management 43(4): 471-488.
California Environmental Protection Agency (CalEPA) Office of Environmental Health Hazard
Assessment (OEHHA). 2000. Air Toxics Hot Spots Program Risk Assessment Guidelines; Part
IV, Exposure Assessment and Stochastic Analysis Technical Support Document. Section 6,
Dermal Exposure Assessment. September. Available at:
http://www.oehha.ca.gov/air/hot_spots/pdf/chap6.pdf.
Demers, E., D.J. McQueen, C.W. Ramcharan, A. Perez-Fuentetaja. 2001. "Did piscivores
regulate changes in fish community structure?" Arch. Hydrobiol. Spec. Issues Advanc. Limnol.
56: 49-80.
Grandjean, P., P. Weihe, R.F. White. 1995. Milestone development in infants exposed to
methylmercury from human milk. Neurotoxicology 16(1):27-33.
Hudson River Foundation (HRF). 2007. Contamination Assessment and Reduction Project
(CARP) data archive. Available by request at: http://www.carpweb.org/main.html
Jardine, C.G. 2003. Development of a Public Participation and Communication Protocol for
Establishing Fish Consumption Advisories. Risk Analysis 23(3): 461-471.
Kershaw, T.G., T.W. Clarkson, P.H. Dhahir. 1980. The relationship between blood levels and
dose of methylmercury in man. Arch. Environ. Health 35(1):28-36.
May, H., and J. Burger. 1996. Fishing in a polluted estuary: Fishing behavior, fish
consumption, and potential risk. Risk Analysis 16(4): 459-471.
McKone, T.E., A. Bodnar, and E. Hertwich. 2001. Development and evaluation of state-specific
landscape data sets for multimedia source-to-dose models. University of California at Berkeley.
Supported by U.S. Environmental Protection Agency (Sustainable Technology Division, National
Risk Management Research Laboratory) and Environmental Defense Fund. July. LBNL-43722.
Mierzykowski, SE; Carr, KC. 2004. Contaminant Survey of Sunkhaze Stream and Baker Brook
- Sunkhaze Meadows National Wildlife Refuge. U.S. FWS (Fish and Wldlife Service). Maine
Field Office. Spec. Proj. Rep. FY04-MEFO-2-EC. Old Town, ME. Available at:
http://www.fws.gov/northeast/mainecontaminants/PDF%20files/2004%20Sunkhaze%20Report
%20Final.pdf
1-71
-------
Moore, I. D., and J. P. Wilson. 1992. Length-slope factors for the Revised Universal Soil Loss
Equation: simplified method of estimation. Soil and Water Cons. 47(5): 423-428.
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center
(NCDC). 1988. Freeze/Frost Data. Climatography of the U.S. No. 20, Supplement No. 1.
(Authors: Koss WJ, JR Owenby, PM Steurer, DC Ezell. NCDC, NOAA. Asheville, NC. (p. 107,
Albany WSO listing) January.
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center
(NCDC). 2001. The FCC Integrated Surface Hourly Database, A New Resource of Global
Climate Data. Technical Report 2001-01. Nov 2001. Available at:
http://www1.ncdc.noaa.gOv/pub//data/techrpts/tr200101/tr2001-01.pdf.
National Oceanic and Atmospheric Administration (NOAA), National Climatic Data Center
(NCDC). 2003. United States Climate Normals, 1971-2000. Last updated Sept 8,
2003. Available at: http://www.ncdc.noaa.gov/oa/climate/normals/usnormals.html
National Oceanic and Atmospheric Administration (NOAA). 2007. Radiosonde Database
Access. Forecast Systems Laboratory. Last updated August 2007. Available at:
http://raob.fsl.noaa.gov.
National Resource Conservation Service (NRCS) 2007. Revised Universal Soil Loss Equation,
Version 2 (RUSLE2). Last updated July 11, 2007.
New York State Fish and Wldlife Department (NY FWD). 1971. Fishery Survey of Alcove
Reservoir.
New York State Department of Environmental Conservation (NY DEC). 2008. Printout of fish
survey results from 1988 through 2006 for Kinderhook and Nassau Lakes, forwarded by
Norman R. McBride, NYDEC, to Leiran Biton, ICF International, March 3.
New York State Department of Health (NY DOH). 2007. Chemicals in Sportfish and Game.
Division of Environmental Health Assessment. Prepared 5/18/07. Available at:
http://www.health.state.ny.us/environmental/outdoors/fish/fish.htm.
Oak Ridge National Laboratories. 1977. A Selective Analysis of Power Plant Operation on the
Hudson River with Emphasis on the Bowline Point Generating Station, Volume 1 (Oak Ridge
Tennessee, July 1977) pp. 2.1-2.19. Available at:
http://library.marist.edu/diglib/EnvSci/archives/wtrqlty/oakridge/ornl-
generalconsiderations.html#f212
Portier K., J. Tolson, and S. Roberts. 2007. Body weight distributions for risk assessment. Risk
Analysis 27(1): 11-26.
Smith, A.H. 1987. Infant Exposure Assessment for Breast Milk Dioxins and Furans Derived from
Waste Incineration Emissions. Risk Analysis 7:347-353.
Stanek, E.J., E.J. Calabrese, R. Barnes, P. Pekow. 1997. Soil ingestion in adults - results of a
second pilot study. Toxicol. Environ Safety 36:249-257.
U.S. Census Bureau. 2000. 2000 Census Data. Available at: http://www.census.gov. (Accessed
on April 25, 2008).
1-72
-------
U.S. Department of Agriculture (USDA). 1980. Food and nutrient intakes of individuals in one
day in the United States, Spring 1977. Nationwide Food Consumption Survey 1977-1978. U.S.
Department of Agriculture. Preliminary Report No. 2.
U.S. Department of Agriculture (USDA). 2002. The Census of Agriculture; 2002 State and
County Reports. Available at: http://www.agcensus.usda.gov/index.asp.
US Environmental Protection Agency (EPA). 1980. Acquisition and chemical analysis of
mother's milk for selected toxic substances. Washington, DC. EPA-560/13-80-029. December.
US Environmental Protection Agency (EPA). 1983. Chemicals identified in human breast milk, a
literature search. Washington, DC. EPA-560/5-83-009. October.
U.S. Environmental Protection Agency (EPA). 1989. Risk Assessment Guidance for
Superfund. Volume I. Human Health Evaluation Manual (Part A). Office of Emergency and
Remedial Response (OERR), Washington, D.C. December. OERR 9200 6-303-894. Available
at: http://www.epa.gov/oswer/riskassessment/ragsa/index.htm.
U.S. Environmental Protection Agency (EPA). 1990. Lake Ontario TCDD Bioaccumulation
Study Final Report. Cooperative study including US EPA, New York State Department of
Environmental Conservation, New York State Department of Health, and Occidental Chemical
Corporation. As quoted in The 1994 EPA Dioxin Reassessment - Exposure Document.
Available at: http://www.cqs.com/epa/exposure.
U.S. Environmental Protection Agency (EPA). 1991. Risk Assessment Guidance for Superfund.
Volume 1. Human Health Evaluation Manual, Supplemental Guidance: Standard Default
Exposure Factors. Office of Emergency and Remedial Response, Washington, DC. March 25,
OSWER Directive 9285.6-03.
U.S. Environmental Protection Agency (EPA). 1994. Estimating Exposure to Dioxin-Like
Compounds. External Review Draft. EPA/600/6-88/005Cb. Office of Research and
Development, Washington, DC. Available at: http://nepis.epa.gov/.
U.S. Environmental Protection Agency (EPA). 1997a. Exposure Factors Handbook, Volume I,
General Factors. EPA/600/P-95/002Fa. Office of Research and Development, Washington,
DC. August. Available at: http://www.epa.gov/nceawww1/efh/.
U.S. Environmental Protection Agency (EPA). 1997b. Exposure Factors Handbook, Volume II,
Food Ingestion Factors. EPA-600-P-95-002Fb. Office of Research and Development,
Washington, DC. August. Available at: http://www.epa.gov/nceawww1/efh/.
U.S. Environmental Protection Agency (EPA). 1998. Methodology for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions. EPA-600-R-98-137.
National Center for Environmental Assessment, Cincinnati, OH. Available at:
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=55525.
U.S. Environmental Protection Agency. 2002a. 2002 National Emissions Inventory. Available at:
http://www.epa.gov/ttn/chief/net/2002inventory.html. (Accessed March 20, 2008).
U.S. Environmental Protection Agency (EPA). 2002b. Estimated Per Capita Fish Consumption
in the United States. Office of Water, Office of Science and Technology, Washington, D.C. EPA-
1-73
-------
821- C- 02-003. August. Available at:
http://www.epa.gov/waterscience/fish/files/consumption_report.pdf.
U.S. Environmental Protection Agency (EPA). 2003a. Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) Modeling System, Volume II: Site-based, Regional, and
National Data, Chapter 8. SAB Review Draft. EP-530/D-03-001b. Office of Research and
Development, Athens, GA, and Research Triangle Park, NC, and Office of Solid Waste,
Washington, DC. July. Available at:
http://www.epa.gov/epaoswer/hazwaste/id/hwirwste/risk03.htm.
U.S. Environmental Protection Agency (EPA). 2003b. CSFII Analysis of Food Intake
Distributions. Office of Research and Development, National Center for Environmental
Assessment, Washington, DC. EPA-600-R-03-029. Available at:
http: //cfp u b. e pa. go v/n cea/cf m/record i s p I ay. cf m ?deid=56610.
U.S. Environmental Protection Agency (EPA). 2003. TRIM.FaTE User's Guide, Module 5.
Accessed March 15, 2008. Available at: http://www.epa.gOv/ttn/fera/trim_fate.html#current_user.
U.S. Environmental Protection Agency (EPA). 2004a. Air Toxics Risk Assessment Reference
Library; Volume 1 - Technical Resource Document, Part III, Human Health Risk Assessment:
Multipathway Chapter 14, Overview and Getting Started: Planning and Scoping the
Multipathway Risk Assessment. Office of Air Quality Planning and Standards, Research
Triangle Park, NC. April. Available at:
http://www.epa.gov/ttn/fera/data/risk/vol_1/chapter_14.pdf.
U.S. Environmental Protection Agency (EPA). 2004b. Air Toxics Risk Assessment Reference
Library; Volume 1 - Technical Resource Document, Part III, Human Health Risk Assessment:
Multipathway Chapter 15, Problem Formulation: Multipathway Risk Assessment. Office of Air
Quality Planning and Standards, Research Triangle Park, NC. April. Available at:
http://www.epa.gov/ttn/fera/data/vol_1, chapter_15.pdf.
U.S. Environmental Agency (EPA). 2004c. Risk Assessment Guidance for Superfund Volume
1: Human Health Evaluation Manual (Part E, Supplemental Guidance for Dermal Risk
Assessment). EPA/540/R99/005. Available at:
http://www.epa.gov/oswer/riskassessment/ragse/index.htm.
U.S. Environmental Protection Agency (EPA). 2004d. Estimated Per Capita Water Ingestion and
Body Weight in the United States - An Update. Office of Water, Office of Science and
Technology, Washington, D.C. EPA-822-R-00-001. October. Available at:
http://www.epa.gov/waterscience/criteria/drinking/percapita/2004.pdf
U.S. Environmental Protection Agency (EPA). 2005. Human Health Risk Assessment Protocol
for Hazardous Waste Combustion Facilities. U.S. Environmental Protection Agency, Office of
Solid Waste and Emergency Response, Washington, DC. EPA-530-R-05-006. September.
Available at: http://www.epa.gov/combustion/riskvol.htm.
U.S. Environmental Protection Agency (EPA). 2008a. Child-Specific Exposure Factors
Handbook. Office of Research and Development, Washington, D.C. EPA/600/R-06/096F.
September. Available at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=199243 .
1-74
-------
U.S. Environmental Protection Agency (EPA), Office of Air Quality Planning and Standards
(OAQPS). 2007. Prioritized Chronic Dose-Response Values for Screening Risk Assessments
(Table 1). June 12, 2007. Available at: http://www.epa.gov/ttn/atw/toxsource/summary.html.
U.S. Geological Survey (USGS), U.S. Department of the Interior. 1992. U.S. Geological Survey
Land Cover Institute. Available at: http://landcover.usgs.gov/natllandcover.php. (Accessed
February 19, 2008).
U.S. Geological Survey (USGS). Randall, Allan D. 2004. Mean Annual Runoff, Precipitation,
and Evapotranspiration in the Glaciated Northeastern United States, 1951-80. U.S. Department
of the Interior. Last updated Sept 2, 2004. Available at:
http://ny.water.usgs.gov/pubs/of/of96395/OF96-395.html.
U.S. Geological Survey (USGS), U.S. Department of the Interior. 2008a. Monthly and annual
net discharge, in cubic feet per second, of Hudson River at Green Island, N.Y. 21:39:18. Last
updated: Wednesday, January 16. Available at:
http://ny.water.usgs.gov/projects/dialer_plots/Hudson_R_at_Green_lsland_Freshwater_Dischar
ge.htm
U.S. Geological Survey (USGS), U.S. Department of the Interior. 2008b. U.S. Geological
Survey National Hydrography Dataset. Accessed March 21, 2008. Available at:
http://nhd.usgs.gov/index.html.
West, P., J.M. Fly, R. Marans, F. Larkin and D. Rosenblatt. 1993. The 1991-92 Michigan Sport
Anglers Fish Consumption Study. Final Report to the Michigan Great Lakes Protection Fund,
Michigan Dept. of Natural Resources. University of Michigan, School of Natural Resources, the
Natural Resource Sociology Research Lab. Technical Report #6. May.
Wschmeier, W. H., and D.D. Smith. 1978. Predicting rainfall erosion losses - a guide to
conservation planning. U.S. Department of Agriculture, Agriculture Handbook No. 537.
World Health Organization (WHO). 1985. The quantity and quality of breast milk. Report on the
WHO Collaborative Study on Breast-feeding. Geneva.
World Health Organization (WHO). 1989. Minor and trace elements in breast milk. Report of a
joint WHO/IAEA Collaborative Study. Geneva.
U.S. Geological Survey, 2004. Collection, Analysis, and Age-Dating of Sediment Cores From 56
U.S. Lakes and Reservoirs Sampled by the U.S. Geological Survey, 1992-2001. Scientific
Investigations Report 2004-5184.
1-75
-------
ATTACHMENT 1-1: TRIM.FaTE Inputs for the Ravena Case Study
-------
[This page intentionally left blank.]
-------
TABLE OF CONTENTS
1-1-1 TRIM.FaTE Modeling Inputs 1
1-1-2 Supplemental Information for Exhibit 1-2 - Meteorological and Other Settings 36
1-1-2.1 PCRAMMET 36
1-1-2.2 TRIM.FaTE Processing 37
1-1-3 Supplemental Information for Exhibit 1-6 - Universal Soil Loss Equations 39
1-1-3.1 Universal Soil Loss Equation 39
1-1-3.2 Rainfall/erosivity Factor (R) 39
1-1-3.3 Soil Erodibility Factor (K) 39
1-1-3.4 Length Slope (LS) Factor 40
1-1-3.5 Cover Management Factor (C) 41
1-1-3.6 Supporting Practice Factor (P) 41
1-1-3.7 Total Erosion Losses Per Parcel 41
1-1-3.8 Limitations to This Approach 41
1-1-3.9 Sediment Balance Calculations 42
1-1-3.10 References 44
1-1-4 Supplemental Information for Exhibit 1-12-Aquatic Animals 45
1-1-4.1 Introduction 45
1-1-4.2 Collection of Information on Species Present in Water Bodies 45
1-1-4.3 Creation of Food Webs 50
1-1-4.4 Parameterization of Fish Compartments to be Included in Application 54
1-1-4.5 Fish Harvesting from Ravena Pond 61
1-1-4.6 References 62
1-1-i
-------
LIST OF EXHIBITS
Exhibit 1-1. Non-Chemical-Dependent Air Parameters 2
Exhibit 1-2. Meteorological and Other Settings 3
Exhibit 1-3. Non-Chemical-Dependent Soil Parameters 4
Exhibit 1-4. Erosion and Runoff Fractions 7
Exhibit 1-5. Total Erosion Rates from Surface Soil Volume Elements 8
Exhibit 1-6. Universal Soil Loss Equation Inputs 9
Exhibit 1-7. Surface Soil Terrestrial Plant Types 11
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters 12
Exhibit 1-9. Surface Water Non-Chemical-Dependent Properties 18
Exhibit 1-10. Sediment Non-Chemical-Dependent Parameters 20
Exhibit 1-11. Aquatic Plant Non-Chemical-Dependent Parameters 21
Exhibit 1-12. Aquatic Animal Non-Chemical-Dependent Parameters 22
Exhibit 1-13. Mercury Chemical-Specific Properties 23
Exhibit 1-14. Chemical-Specific Properties for 2,3,7,8-TCDD 24
Exhibit 1-15. Mercury Chemical-Specific Properties for Abiotic Compartments 25
Exhibit 1-16. Chemical-Specific Properties of 2,3,7,8-TCDD for Abiotic Compartments 29
Exhibit 1-17. Mercury Chemical-Specific Properties for Plants 30
Exhibit 1-18. Chemical-Specific Properties of 2,3,7,8-TCDD for Plants 32
Exhibit 1-19. Mercury Chemical-Specific Properties for Aquatic Species 33
Exhibit 1-20. Chemical-Specific Properties of 2,3,7,8-TCDD for Aquatic Species 34
Exhibit 2-1. Completeness of Meteorological Data Types 37
Exhibit 3-1. Soil Erodibility Factor for Watershed Parcels and All Other Parcels 40
Exhibit 3-2. USLE Empirical Intercept Coefficient 42
Exhibit 3-3. Calculated USLE Soil Erosion Rates, Sediment Delivery Ratios, and Adjusted
Erosion Rates for Each Soil Parcel 43
Exhibit 4-1. Fish Survey Data for Alcove Reservoir 46
Exhibit 4-2. Fish Survey Data for Nassau Lake 48
Exhibit 4-3. Fish Survey Data for Kinderhook Lake 49
Exhibit 4-4. Estimated Total Fish Standing Stocks for Water Bodies Near Lafarge Facility 51
Exhibit 4-5. Lipid Content for Fish Species Included in Model Food Webs 52
Exhibit 4-6. Aquatic Species Diets by TRIM.FaTE Model Compartments 53
Exhibit 4-7. Small Pond Parameters: Fish Mass, Abundance, and Model Representation 54
Exhibit 4-8. Alcove Reservoir Parameters: Fish Mass, Abundance, and
Model Representation 55
i-1-ii
-------
Exhibit 4-9. Nassau Lake Parameters: Fish Mass, Abundance, and Model Representation 56
Exhibit 4-10. Kinderhook Lake Parameters: Fish Mass, Abundance, and
Model Representation 57
Exhibit 4-11. Small Pond Model Parameters: Fish Mass, Abundance (Number per Hectare),
and Lipid Content 58
Exhibit 4-12. Alcove Reservoir Model Parameters: Fish Mass, Abundance, and
Lipid Content 58
Exhibit 4-13. Nassau Lake Model Parameters: Fish Mass, Abundance, and Lipid Content 59
Exhibit 4-14. Kinderhook Lake Model Parameters: Fish Mass, Abundance, and
Lipid Content 59
Exhibit 4-15. Small Pond Aquatic Food Web 60
Exhibit 4-16. Alcove Reservoir Aquatic Food Web 60
Exhibit 4-17. Nassau Lake Aquatic Food Web 60
Exhibit 4-18. Kinderhook Lake Aquatic Food Web 61
1-1-rn
-------
[This page intentionally left blank.]
-------
l-1-1TRIM.FaTE Modeling Inputs
This section provides the tables of the detailed modeling inputs for the TRIM.FaTE screening
scenario. Exhibits 1-1 and 1-2 present the air parameters entered into the model. In Section 2
of this attachment, supplemental information is provided for Exhibit 1-2, which is a summary of
meteorological inputs for the TRIM.FaTE analysis.
Exhibits 1-3 through 1-8 present the terrestrial parameters. In Section 3 of this attachment,
supplemental information is provided for Exhibit 1-6, which presents the inputs for the Universal
Soil Loss Equation.
Exhibits 1-9 through 1-12 present the lake parameters, and 1-13 through 1-20 present the
chemical specific parameters. In Section 4 of this attachment, supplemental information is
provided for Exhibit 1-12, which is a summary of non-chemical-dependent parameter inputs for
aquatic animals.
1-1-1
-------
Exhibit 1-1. Non-Chemical-Dependent Air Parameters
Parameter Name
Units
Value Used
Reference
Atmospheric dust load
kg[dust]/m3[air]
6.15E-08
Bidleman 1988
Air density
g/cm3
0.0012
U.S. EPA 1997
Dust density
kg[dust]/m3[dust]
1,400
Bidleman 1988
Fraction organic matter on
particulates
unitless
0.2
Harnerand Bidleman 1998
Height [VE property]
m
varies
Meteorological data used
1-1-2
-------
Exhibit 1-2. Meteorological and Other Settings a
Parameter Name
Units
Value Used
Reference
Meteorological In
puts (all TRIM.FaTE scenario properties, except mixing height)
Air temperature
degrees K
varies daily
MET data assembled from actual 2001-2003 records
Horizontal wind
speed
m/sec
varies daily
MET data assembled from actual 2001-2003 records
Vertical wind
speed
m/sec
0.0
Professional judgment; vertical wind speed not used
by any of the algorithms in the version of the
TRIM.FaTE library used for screening
Wind direction
degrees clockwise
from N (blowing
from)
varies daily
MET data assembled from actual 2001-2003 records
Rainfall rate
m3[rain]/
m2[surface area]-
day
varies daily
MET data assembled from actual 2001-2003 records
Mixing height
(used to set air
VE property
named "top")
m
varies daily
MET data assembled from actual 2001-2003 records
isDay_SteadyStat
e_forAir
unitless
Value not used in current dynamic runs (would need
to be reevaluated if steady-state runs are needed)
isDay_SteadyStat
e_forOther
unitless
Other Settings (all TRIM.FaTE scenario properties)
Start of simulation
date/time
1/1/1990,
midnight
Consistent with met data
End of simulation
date/time
1/1/2020,
midnight
Consistent with met data set; selected to provide a
30-year modeling period
Simulation time
step
hr
1
Selected value
Output time step b
hr
2
Selected value
a For more information, see Section 2 of this attachment.
b Output time step is set in TRIM.FaTE using the scenario properties "simulationStepsPerOutputStep" and
"simulationTimeStep."
1-1-3
-------
Exhibit 1-3. Non-Chemical-Dependent Soil Parameters
Parameter Name
Units
Value Used
Reference
Surface Soil Compartment Type
Air content
volume[air]/
volumefcompartmentl
0.28
McKoneetal 2001 (Table 15)
Average vertical velocity of
water (percolation)
m/day
4.40E-04
Assumed to be 0.2 times average
precipitation for site
Boundary layer thickness
above surface soil
m
0.005
Thibodeaux 1996; McKone et al. 2001
(Table 3)
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2600
Default in McKone et al. 2001
(Table 3)
Thickness -unfilled [VE
propertyla
m
0.01
McKone et al. 2001 (p. 30)
Thickness -tilled [VE
DroDertvla
m
0.20
USEPA 2005
Erosion fraction [Link
property]
unitless
varies b
See Erosion and Runoff Fraction
table.
Fraction of area available
for erosion
m2[area available]/m2[total]
1
Professional judgment; area assumed
rural
Fraction of area available
for runoff
m2[area available]/m2[total]
1
Professional judgment; area assumed
rural
Fraction of area availabe
for vertical diffusion
m2[area available]/m2[total]
1
Professional judgment; area assumed
rural
Fraction Sand
unitless
0.25
Professional judgment
Organic carbon fraction
unitless
0.008
U.S. average in McKone et al. 2001
(Table 16 and A-3)
PH
unitless
6.8
Professional judgment
Runoff fraction [Link
property]
unitless
varies b
See Erosion and Runoff Fraction
table.
Total erosion rate
kg [soil]/m2/day
varies b
See Total Erosion Rates table.
Total runoff rate
m3[water]/m2/day
4.04E-04
Calculated using scenario-specific
precipitation rate and assumptions
associated with water balance.
Water content
volume[water]/
volumefcompartmentl
0.19
McKoneetal 2001 (Table 15)
1-1-4
-------
Exhibit 1-3. Non-Chemical-Dependent Soil Parameters
Parameter Name
Units
Value Used
Reference
Root Zone Soil Compartment Type
Air content
volume[air]/
volumefcomDartmentl
0.25
McKoneetal 2001 (Table 16)
Average vertical velocity of
water (percolation)
m/day
4.40E-04
Assumed as 0.2 times average
precipitation for New England in
McKone et al. 2001
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2,600
McKone et al. 2001 (Table 3)
Fraction Sand
unitless
0.25
Professional judgment
Thickness - unfilled [VE
propertvla
m
0.69
McKone et al. 2001 (Table 16 - Middle
Atlantic value)
Thickness - tilled [VE
propertvla
m
0.6
Adjusted from McKone et al. 2001
(Table 16)
Organic carbon fraction
unitless
0.008
McKone et al. 2001 (Table 16 and A-
3, U.S. Average)
PH
unitless
6.8
Professional judgment
Water content
volume[water]/
volumefcompartmentl
0.21
McKoneetal 2001 (Table 16)
Vadose Zone Soil Compartment Type
Air content
volume[air]/
volumefcompartmentl
0.22
McKoneetal 2001 (Table 17)
Average vertical velocity of
water (percolation)
m/day
4.40E-04
Assumed as 0.2 times average
precipitation for New England in
McKone et al. 2001
Density of soil solids (dry
weight)
kg[soil]/m3[soil]
2,600
Default in McKone et al. 2001 (Table
3)
Fraction Sand
unitless
0.35
Professional judgment
Thickness [VE property]a
m
1.4
McKone et al. 2001 (Table 17)
Organic carbon fraction
unitless
0.003
McKone et al. 2001 (Table 17 and A-
3, U.S. Average)
PH
unitless
6.8
Professional judgment
Water content
volume[water]/
volumefcompartmentl
0.21
McKoneetal 2001 (Table 17-
national average)
1-1-5
-------
Exhibit 1-3. Non-Chemical-Dependent Soil Parameters
Parameter Name
Units
Value Used
Reference
Ground Water Compartment Type
Thickness [VE property]a
m
3
McKone et al. 2001 (Table 3)
Fraction Sand
unitless
0.4
Professional judgment
Organic carbon fraction
unitless
0.004
Professional judgment
PH
unitless
6.8
Professional judgment
Porosity
volume[total pore space]/
volume[compartment]
0.2
Default in McKone et al. 2001
(Table 3)
Solid material density in
aquifer
kg[soil]/m3[soil]
2,600
Default in McKone et al. 2001
(Table 3)
a Set using the volume element properties file.
bSee separate tables for erosion/runoff fractions and total erosion rates.
1-1-6
-------
Exhibit 1-4. Erosion and Runoff Fractions
Originating Compartment
Destination Compartment
Runoff/Erosion Fraction
SurfSoil E1
Hudson River
1
SurfSoil E2
Nassau Lake
1
SurfSoil E3
Kinderhook Lake
1
SurfSoil E4
Hudson River
1
SurfSoil E5
Hudson River
1
SurfSoil E6
Hudson River
1
SurfSoil Tilled East Farm
SurfSoil E5
1
SurfSoil Unfilled East Farm
SurfSoil E5
1
Nassau Lake
Kinderhook Lake
1
SurfSoil W1
Hudson River
1
SurfSoil W2
Hudson River
1
SurfSoil W3
Alcove Reservoir
1
SurfSoil W4
Hudson River
1
SurfSoil W5
Hudson River
1
SurfSoil W6
Hudson River
1
SurfSoil W7
Pond
1
SurfSoil W8
Hudson River
1
SurfSoil Tilled West Farm
SurfSoil W2
1
SurfSoil Unfilled West Farm
SurfSoil W2
1
1-1-7
-------
Exhibit 1-5. Total Erosion Ra
tes from Surface Soil Volume Elements a
Soil Component
Total Erosion Rate (kq/m2-dav)
E1
0.000190
E2
0.000323
E3
0.000220
E4
0.000117
E5
0.000148
E6
0.000153
Efarm tilled
0.001270
Efarm untilled
0.000500
W1
0.000082
W2
0.000310
W3
0.000337
W4
0.000095
W5
0.000231
W6
0.000370
W7
0.005916
W8
0.000410
Wfarm tilled
0.006116
Wfarm untilled
0.000469
a Calculated using the Universal Soil Loss Equation in combination with precipitation rate and other
assumptions.
1-1-8
-------
Exhibit
1-6. Universal Soil Loss Equation Inputs a
Soil
Parcel
Area
Rainfall/E
rosivity
Index
Soil
Erodibility
Index
Slope
Length
Slope
Steepness
Topographical
(LS) Factor
Land Use
km2
100 ft-
ton/acre
ton/ac/(100 ft
ton/acre)
meters
%
unitless
E1
85.224
114
0.3145
200
7.23%
2.206
Grasses
E2
26.719
114
0.3113
200
8.00%
2.539
Deciduous
Forest
E3
44.517
114
0.3102
200
7.86%
2.478
Deciduous
forest
E4
420.32
114
0.3145
200
9.49%
3.248
Grasses
E5
77.709
114
0.3145
200
6.53%
1.924
Grasses
E6
81.925
114
0.3145
200
6.43%
1.886
Grasses
Efarm
tilled
0.358
114
0.3145
200
2.88%
0.484
Tilled
Soil/crops
Efarm
unfilled
0.358
114
0.3145
200
3.98%
0.844
Unfilled corn
50 bu/acre
W1
48.836
105
0.3145
200
4.78%
1.041
Grasses
W2
109.84
105
0.3145
200
9.25%
3.125
Deciduous
W3
86.819
105
0.3222
200
9.45%
3.228
Deciduous
Forest
W4
306.58
105
0.3145
200
7.16%
2.178
Deciduous
Forest
W5
84.025
105
0.3145
200
8.71%
2.864
Deciduous
Forest
W6
15.011
105
0.3145
200
9.42%
3.211
Grasses
W7
2.028
105
0.3145
200
8.65%
2.837
Deciduous
Forest
W8
72.532
105
0.3145
200
10.77%
3.919
Deciduous
Forest
Wfarm
tilled
0.359
105
0.3145
200
7.99%
2.534
Tilled
Soil/crops
Wfarm
unfilled
0.358
105
0.3145
200
4.05%
0.860
Unfilled corn
50 bu/acre
1-1-9
-------
Exhibit 1-6.
Universal Soil Loss Eq
Soil
Parcel
Cover/Mgt
Factor
Supporting
Practices
Factor
Unit Soil Loss
Sediment
Delivery Ratio
b
Calculated
(Adjusted)
Erosion Rate
unitless
unitless
ton/ac/yr
kg/m2/day
kg/m2/day
E1
0.032
1
2.530
0.0016
0.122
0.000190
E2
0.041
1
3.717
0.0023
0.142
0.000323
E3
0.031
1
2.693
0.0017
0.133
0.000220
E4
0.033
1
3.808
0.0023
0.050
0.000117
E5
0.028
1
1.951
0.0012
0.124
0.000148
E6
0.030
1
2.026
0.0012
0.123
0.000153
Efarm
tilled
0.310
1
5.384
0.0033
0.384
0.001270
Efarm
unfilled
0.070
1
2.119
0.0013
0.384
0.000500
W1
0.029
1
1.012
0.0006
0.131
0.000082
W2
0.041
1
4.263
0.0026
0.119
0.000310
W3
0.041
1
4.497
0.0028
0.122
0.000337
W4
0.041
1
2.963
0.0018
0.052
0.000095
W5
0.032
1
3.072
0.0019
0.123
0.000231
W6
0.032
1
3.395
0.0021
0.177
0.000370
W7
0.333
1
31.156
0.0191
0.309
0.005916
W8
0.041
1
5.348
0.0033
0.125
0.000410
Wfarm
tilled
0.310
1
25.940
0.0159
0.384
0.006116
Wfarm
unfilled
0.070
1
1.989
0.0012
0.384
0.000469
uation Inputs
a For more information, see Section 3 of this attachment.
b Calculated using SD = a * (AL)-b; where a is the empirical intercept coefficient (based on the size of the
watershed), AL is the total watershed area receiving deposition (m2), and b is the empirical slope coefficient
(always 0.125).
1-1-10
-------
Exhibit 1-7. Surface Soi
Terrestrial Plant Types
Surface Soil
Volume Element
Surface Soil
Depth (m)
Deciduous
Forest
Grasses/
Herbs
Crops
E1
0.01
X
E2
0.01
X
E3
0.01
X
E4
0.01
X
E5
0.01
X
E6
0.01
X
Efarm tilled
0.20 (tilled)
X
Efarm untilled
0.01
X
W1
0.01
X
W2
0.01
X
W3
0.01
X
W4
0.01
X
W5
0.01
X
W6
0.01
X
W7
0.01
X
W8
0.01
X
Wfarm tilled
0.20 (tilled)
X
Wfarm untilled
0.01
X
1-1-11
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Deciduous a
Value Used
Reference
Leaf Compartment Type
Allow exchange
1=yes, 0=no
seasonalb
-
Average leaf area index
m2[leaf]/
m2[areal
3.4
Harvard Forest, dom. red oak and red
maple, CDIAC website
Calculate wet dep interception
fraction (boolean)
1=yes, 0=no
0
Professional judgment
Correction exponent, octanol to
lipid
unitless
0.76
From roots; Trapp, S. 1995. Model for
uptake of xenobiotics into plants.
Degree stomatal opening
unitless
1
Set to 1 for daytime based on
professional judgment (stomatal
diffusion is turned off at night using a
different property, IsDay)
Density of wet leaf
kg/m3
820
Paterson et al. 1991
Leaf wetting factor
m
3.00E-04
1E-04 to 6E-04 for different crops and
elements, Mullerand Prohl 1993
Length of leaf
m
0.01
Professional judgment
Lipid content
kg/ kg wet
weight
0.00224
European beech, Riederer 1995
Litter fall rate
1/day
seasonalc
-
Stomatal area normalized
effective diffusion path length
1/m
200
Wilmer and Fricker 1996
Vegetation attenuation factor
m2/kg
2.9
Grass/hay, Baes et al. 1984
Water content
unitless
0.8
Paterson et al. 1991
Wet dep interception fraction
unitless
0.2
Calculated based on 5 years of local
met data, 1987-1991
Wet mass of leaf per soil area
kg [fresh leaf]/
m2[area]
0.6
Calculated from leaf area index, leaf
thickness (Simonich & Hites, 1994),
density of wet foliage
Particle on Leaf Compartment Type
Allow exchange
1=yes, 0=no
-
-
Volume particle per area leaf
m3[leaf
particles]/m2[lea
f]
1.00E-09
Based on particle density and size
distribution for atmospheric particles
measured on an adhesive surface,
Coe and Lindberg 1987
1-1-12
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Deciduous a
Value Used
Reference
Root Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
n/a
n/a
Correction exponent, octanol to
lipid
unitless
n/a
n/a
Lipid content of root
kg/kg wet
weight
n/a
n/a
Water content of root
kg/kg wet
weight
n/a
n/a
Wet density of root
kg/m3
n/a
n/a
Wet mass per soil area
kg/m2
n/a
n/a
Stem Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
n/a
n/a
Correction exponent, octanol to
lipid
unitless
n/a
n/a
Density of phloem fluid
kg/m3
n/a
n/a
Density of xylem fluid
kg/cm3
n/a
n/a
Flow rate of transpired water
per leaf area
m3[water]/m2[le
afl
n/a
n/a
Fraction of transpiration flow
rate that is phloem rate
unitless
n/a
n/a
Lipid content of stem
kg/kg wet
weight
n/a
n/a
Water content of stem
unitless
n/a
n/a
Wet density of stem
kg/m3
n/a
n/a
Wet mass per soil area
kg/m2
n/a
n/a
1-1-13
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Grass/Herb a
Value Used
Reference
Leaf Compartment Type
Allow exchange
1=yes, 0=no
seasonalb
-
Average leaf area index
m2[leaf]/
m2[areal
5.0
Mid-range of 4-6 for old fields, R.J.
Luxmoore, ORNL
Calculate wet dep interception
fraction (boolean)
1=yes, 0=no
0
Professional judgment
Correction exponent, octanol to
lipid
unitless
0.76
From roots, Trapp 1995
Degree stomatal opening
unitless
1
Set to 1 for daytime based on
professional judgment (stomatal
diffusion is turned off at night using a
different property, IsDay)
Density of wet leaf
kg/m3
820
Paterson et al. 1991
Leaf wetting factor
m
3.00E-04
1E-04 to 6E-04 for different crops and
elements, Mullerand Prohl 1993
Length of leaf
m
0.05
Professional judgment
Lipid content
kg/ kg wet
weight
0.00224
European beech, Riederer 1995
Litter fall rate
1/day
seasonalc
-
Stomatal area normalized
effective diffusion path length
1/m
200
Wilmer and Fricker 1996
Vegetation attenuation factor
m2/kg
2.9
Grass/hay, Baes et al. 1984
Water content
unitless
0.8
Paterson et al. 1991
Wet dep interception fraction
unitless
0.2
Calculated based on 5 years of local
met data, 1987-1991
Wet mass of leaf per soil area
kg [fresh leaf]/
m2[area]
0.6
Calculated from leaf area index and
Leith 1975
Particle on Leaf Compartment Type
Allow exchange
1=yes, 0=no
seasonalb
-
Volume particle per area leaf
m3[leaf
particles]/m2[lea
f]
1.00E-09
Based on particle density and size
distribution for atmospheric particles
measured on an adhesive surface,
Coe and Lindberg 1987
1-1-14
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Grass/Herb a
Value Used
Reference
Root Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
seasonalb
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995
Lipid content of root
kg/kg wet
weight
0.011
Calculated
Water content of root
kg/kg wet
weight
0.8
Professional judgment
Wet density of root
kg/m3
820
Soybean, Paterson et al. 1991
Wet mass per soil area
kg/m2
1.4
Temperate grassland, Jackson et al.
1996
Stem Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
seasonalb
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995
Density of phloem fluid
kg/m3
1,000
Professional judgment
Density of xylem fluid
kg/cm3
900
Professional judgment
Flow rate of transpired water
per leaf area
m3[water]/m2[le
afl
0.0048
Crank et al. 1981
Fraction of transpiration flow
rate that is phloem rate
unitless
0.05
Paterson et al. 1991
Lipid content of stem
kg/kg wet
weight
0.00224
Leaves of European beech, Riederer
1995
Water content of stem
unitless
0.8
Paterson et al. 1991
Wet density of stem
kg/m3
830
Professional judgment
Wet mass per soil area
kg/m2
0.24
Calculated from leaf and root
biomass density based on
professional judgment
1-1-15
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Agriculture a
Value Used
Reference
Leaf Compartment Type
Allow exchange
1=yes, 0=no
seasonal13
-
Average leaf area index
m2[leaf]/
m2[areal
2.0
Mid-range of 4-6 for old fields, R.J.
Luxmoore, ORNL
Calculate wet dep interception
fraction (boolean)
1=yes, 0=no
0
Professional judgment
Correction exponent, octanol to
lipid
unitless
0.76
From roots, Trapp 1995
Degree stomatal opening
unitless
1
Set to 1 for daytime based on
professional judgment (stomatal
diffusion is turned off at night using a
different property, IsDay)
Density of wet leaf
kg/m3
820
Paterson et al. 1991
Leaf wetting factor
m
3.00E-04
1E-04 to 6E-04 for different crops and
elements, Mullerand Prohl 1993
Length of leaf
m
0.1
Professional judgment
Lipid content
kg/ kg wet
weight
0.00224
European beech, Riederer 1995
Litter fall rate
1/day
seasonalc
-
Stomatal area normalized
effective diffusion path length
1/m
200
Wilmer and Fricker 1996
Vegetation attenuation factor
m2/kg
2.9
Grass/hay, Baes et al. 1984
Water content
unitless
0.8
Paterson et al. 1991
Wet dep interception fraction
unitless
0.2
Calculated based on 5 years of local
met data, 1987-1991
Wet mass of leaf per soil area
kg [fresh leaf]/
m2[area]
0.4
Calculated from leaf area index and
Leith 1975
Particle on Leaf Compartment Type
Allow exchange
1=yes, 0=no
seasonalb
-
Volume particle per area leaf
m3[leaf
particles]/m2[lea
f]
1.00E-09
Based on particle density and size
distribution for atmospheric particles
measured on an adhesive surface,
Coe and Lindberg 1987
1-1-16
-------
Exhibit 1-8. Non-Chemical-Dependent Terrestrial Plant Parameters
Parameter Name
Units
Agriculture a
Value Used
Reference
Root Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
seasonalb
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995
Lipid content of root
kg/kg wet
weight
0.011
Calculated
Water content of root
kg/kg wet
weight
0.8
Professional judgment
Wet density of root
kg/m3
820
Soybean, Paterson et al. 1991
Wet mass per soil area
kg/m2
0.16
Temperate grassland, Jackson et al.
1996
Stem Compartment Type - Nonwoody Only
Allow exchange
1=yes, 0=no
seasonalb
-
Correction exponent, octanol to
lipid
unitless
0.76
Trapp 1995
Density of phloem fluid
kg/m3
1,000
Professional judgment
Density of xylem fluid
kg/cm3
900
Professional judgment
Flow rate of transpired water
per leaf area
m3[water]/m2[le
afl
0.0048
Crank et al. 1981
Fraction of transpiration flow
rate that is phloem rate
unitless
0.05
Paterson et al. 1991
Lipid content of stem
kg/kg wet
weight
0.00224
Leaves of European beech, Riederer
1995
Water content of stem
unitless
0.8
Paterson et al. 1991
Wet density of stem
kg/m3
830
Professional judgment
Wet mass per soil area
kg/m2
0.15
Calculated from leaf and root
biomass density based on
professional judgment
a See separate table for assignment of plant types to surface soil compartments.
b Begins March 9 (set to 1), ends November 7 (set to 0). Nation-wide 80th percentile.
c Begins November 7, ends December 6; rate = 0.15/day during this time (value assumes 99 percent of leaves fall
in 30 days).
1-1-17
-------
Exhibit 1-9. Surface Water Non-Chemical-Dependent Properties
Parameter Name
Units
Value Used
Reference
Constant Across All Water Bodies
Algae carbon content (fraction)
unitless
0.465
APHA 1995
Algae density in water column
g[algae]/L[water]
0.0025
Millard et al. 1996 as cited in ICF
2005
Algae growth rate
1/day
0.7
Hudson et al. 1994 as cited in
Mason et al. 1995b
Algae radius
um
2.5
Mason et al. 1995b
Algae water content (fraction)
unitless
0.9
APHA 1995
Average algae cell density (per
vol cell, not water)
g[algae]/m3[algae]
1,000,000
Mason et al. 1995b, Mason et al.
1996
Boundary layer thickness above
sediment
m
0.02
Cal EPA 1993
Chloride concentration
mg/L
8.0
Kaushal et al. 2005
Chlorophyll concentration
mg/L
0.0029
ICF 2005
Dimensionless viscous
sublayer thickness
unitless
4
Ambrose et al. 1995
Drag coefficient for water body
unitless
0.0011
Ambrose et al. 1995
Fraction Sand
unitless
0.25
Professional judgment
Organic carbon fraction in
suspended sediments
unitless
0.02
Professional judgment
PH
unitless
7.3
Professional judgment
Suspended sediment
deposition velocity
m/day
2
US EPA 1997
Total suspended sediment
concentration
kg [sed i me nt]/m3 [wate r
column]
0.01
US EPA 2005
Water temperature
degrees K
287
US EPA 2005
1-1-18
-------
Exhibit 1-9. Surface Water Non-Chemical-Dependent Properties
Parameter Name
Units
Value Used
Reference
Water Body-specific Inputs
Alcove Reservoir
Flush rate
1/year
0.51
Calculated based on pond
dimensions and flow
calculations.
Depth [VE property]a
m
9.25
Wl DNR 2005 - calculation
based on relationship between
drainage basin and lake area
size.
Nassau Lake
Flush rate
1/year
4.17
Calculated based on pond
dimensions and flow
calculations.
Depth [VE property]a
m
2.90
Wl DNR 2005 - calculation
based on relationship between
drainage basin and lake area
size.
Kinderhook Lake
Flush rate
1/year
3.35
Calculated based on pond
dimensions and flow
calculations.
Depth [VE property]a
m
4.70
Wl DNR 2005 - calculation
based on relationship between
drainage basin and lake area
size.
Pond
Flush rate
1/year
10.30
Calculated based on pond
dimensions and flow
calculations.
Depth [VE property]a
m
2.90
Wl DNR 2005 - calculation
based on relationship between
drainage basin and lake area
size.
Hudson River
Flush rate
1/year
87.04
Calculated based on pond
dimensions and flow
calculations.
Depth [VE property]a
m
6.00
Wl DNR 2005 - calculation
based on relationship between
drainage basin and lake area
size.
Current Velocity
m/s
0.088
Professional judgment
a Set using the volume element properties named "top" and "bottom."
1-1-19
-------
Exhibit
1-10. Sediment Non-Chemical-Dependent Parameters
Parameter Name
Units
Value
Used
Reference
Depth [VE property]a
m
0.05
McKone et al. 2001
(Table 3)
Fraction Sand
unitless
0.25
Professional judgment
Organic carbon fraction
unitless
0.02
McKone et al. 2001 (Table 3)
Porosity of the sediment
zone
volume[total pore space]/
volume[sediment compartment]
0.6
US EPA 1998
Solid material density in
sediment
kg[sediment]/m3[sediment]
2,600
McKone et al. 2001
(Table 3)
a Set using the volume element properties named "top" and "bottom."
1-1-20
-------
Exhibit 1-11. Aquatic Plant Non-Chemica
-Dependent Parameters
Parameter Name
Units
Value Used
Reference
Biomass per water area
kg/m2
0.6
Bonar et al. 1993
Density of macrophytes
kg/L
1
professional judgment
1-1-21
-------
Exhibit 1-12. Aquatic Animal Non-Chemical-Dependent Parameters a
Aquatic Biota
(Consuming Organism)
Fraction Diet
Biomass
(kg/m2)
Body
Weight
Reference
Algae
Bethic
Invertebrate
Macrophyte
Water Column
Herbivore
Water Column
Omnivore
Water Column
Carnivore
Benthic
Omnivore
Benthic
Carnivore
Alcove Reservoir
Water column herbivore
96%
4%
-
-
-
-
-
-
1.30E-03
0.025
Professional judgment
Water column omnivore
8%
53%
-
39%
-
-
-
-
3.34E-03
0.25
Professional judgment
Water column carnivore
-
41%
4%
29%
-
25%
-
9.44E-04
2.0
Professional judgment
Benthic omnivore
-
100%
-
-
-
-
-
-
2.40E-03
2.0
Professional judgment
Benthic carnivore
-
50%
-
-
-
-
50%
-
1.60E-05
2.0
Professional judgment
Benthic invertebrate
-
-
-
-
-
-
-
-
2.00E-02
2.55E-04
Professional judgment
Kinderhook Lake
Water column herbivore
82%
5%
14%
-
-
-
-
-
1.06E-03
0.025
Professional judgment
Water column omnivore
8%
58%
-
34%
-
-
-
-
2.95E-03
0.25
Professional judgment
Water column carnivore
-
33%
-
8%
34%
-
26%
-
3.60E-04
2.0
Professional judgment
Benthic omnivore
-
100%
-
-
-
-
-
-
6.25E-04
2.0
Professional judgment
Benthic carnivore
-
50%
-
-
-
-
50%
-
1.50E-05
2.0
Professional judgment
Benthic invertebrate
-
-
-
-
-
-
-
-
2.00E-02
2.55E-04
Professional judgment
Nassau Lake
Water column herbivore
92%
5%
3%
-
-
-
-
-
9.15E-04
0.025
Professional judgment
Water column omnivore
9%
61%
-
30%
-
-
-
-
2.73E-03
0.25
Professional judgment
Water column carnivore
-
-
-
25%
50%
-
25%
-
8.00E-05
2.0
Professional judgment
Benthic omnivore
-
100%
-
-
-
-
-
-
1.25E-03
2.0
Professional judgment
Benthic carnivore
-
50%
-
-
-
-
50%
-
2.50E-05
2.0
Professional judgment
Benthic invertebrate
-
-
-
-
-
-
-
-
2.00E-02
2.55E-04
Professional judgment
Pond
Fish harvester13
-
-
-
-
-
33%
67%
-
3.57E-03
71.4
Professional judgment
Water column herbivore
100%
-
-
-
-
-
-
-
8.00E-04
0.025
Professional judgment
Water column carnivore
-
50%
-
50%
-
-
-
-
2.00E-04
2.0
Professional judgment
Benthic omnivore
-
100%
-
-
-
-
-
-
3.00E-03
2.0
Professional judgment
Benthic invertebrate
-
-
-
-
-
-
-
-
2.00E-02
2.55E-04
Professional judgment
a For more information, see Section 4 of this attachment.
b Fish harvester is only used in some model runs and is parameterized as one human fisherman. In was modeled as a human-sized mink in
TRIM.FaTE, as a compartment for humans does not exist.
1-1-22
-------
Exhibit 1-13. Mercury Chemical-Specific Properties a
Parameter Name
Units
Value "
Reference
Hg(0)
Hg(2)
MHg
CAS number
unitless
7439-97-6
14302-87-5
2296-92-6
-
Diffusion coefficient in
pure air
m2[air]/day
0.478
0.478
0.456
US EPA 1997
Diffusion coefficient in
pure water
m2[water]/day
5.54E-05
5.54E-05
5.28E-05
US EPA 1997
Henry's Law constant
Pa-m3/mol
719
7.19E-05
0.0477
US EPA 1997
Melting point
degrees K
234
550
443
CARB 1994
Molecular weight
g/mol
201
201
216
US EPA 1997
Octanol-water
partition coefficient
(Kow)
L[water]/kg[octanol]
4.15
3.33
1.7
Mason, et al. 1996
Vapor washout ratio
m3[air]/m3[rain]
1,200
1.6E+06
0
US EPA 1997, based on
Petersen et al. 1995
a All parameters in this table are TRIM.FaTE chemical properties.
bOn this and all following tables, Hg(0) = elemental mercury, Hg(2) = divalent mercury, and MHg = methyl mercury.
1-1-23
-------
Exhibit 1-14
. Chemical-Specific Properties for 2,3,7,8-TCDD
Parameter Name
Units
Value
Reference
CAS number
unitless
1746-01-6
-
Diffusion coefficient in pure
air
m2/day
0.106
US EPA 1999
Diffusion coefficient in pure
water
m2/day
5.68E-05
US EPA 1999
Henry's Law constant
Pa-m3/mol
3.33
Mackay et al. 1992 as cited in U.S.
EPA 2000a
Melting point
degrees K
578.0
Mackay et al. 2000
Molecular weight
g/mol
322.0
Mackay et al. 2000
Octanol-water partition
coefficient (Kow)
L[water]/L[octanol]
6.31 E+06
Mackay et al. 1992a as cited in U.S.
EPA 2000a
1-1-24
-------
Exhibit 1-15. Mercury Chemical-Specific Properties for Abiotic Compartments
Parameter Name
Units
Value
Reference
Hg(0) Hg(2) MHg
Air Compartment Type
Particle dry deposition
velocity
m/day
500
500
500
CalTOX value cited in McKone
et al. 2001
Demethylation rate
1/day
N/A
N/A
0
Professional judgment
Methylation rate
1/day
0
0
0
Professional judgment
Oxidation rate
1/day
0.00385
0
0
Low end of half-life range (6
months to 2 years) in EPA 1997
Reduction rate
1/day
0
0
0
Professional judgment
Washout Ratio
m3[airl/m3[rainl
200,000
200,000
200,000
Professional judgment
Surface Soil Compartment Type
Input characteristic depth
(user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate
value)
Soil-water partition
coefficient
L[ water]/
kgfsoil wet wtl
1,000
58,000
7,000
U.S. EPA 1997
Use input characteristic
depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment
Vapor dry deposition
velocity
m/day
50
2500
0
Hg(0) - from Lindberg et al.
1992 Hg(2) - estimate by
U.S.EPA using the Industrial
Source Complex (ISC) Model -
[See Vol. Ill, App. A of the
Mercury Study Report (USEPA,
1997)1.
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and
Verta 1995 is 3E-2 to 6E-2 /day;
value is average maximum
potential demethylation rate
constant under anaerobic
conditions
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and
Verta 1995 is 2E-4 to 1E-3 /day;
value is average maximum
potential methylation rate
constant under anaerobic
conditions
Oxidation rate
1/day
0
0
0
Value assumed in EPA 1997
Reduction rate
1/day
0
1.25E-05
0
Value used for unfilled surface
soil (2cm), 10% moisture
content, in U.S. EPA 1997;
general range is
(0.0013/day)*moisture content
to (0.0001/day)*moisture
content for forested region
(Lindberg 1996; Carpi and
Lindberg 1997)
1-1-25
-------
Exhibit 1-15. Mercury Chemical-Specific Properties for Abiotic Compartments
Parameter Name
Units
Value
Reference
Hg(0)
Hg(2)
MHg
Root Zone Soil Compartment Type
Input characteristic depth
(user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate
value)
Soil-water partition
coefficient
L[water]/kg[soil wet
wtl
1,000
58,000
7,000
U.S. EPA 1997
Use input characteristic
depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and
Verta 1995 is 3E-2 to 6E-2 /day;
value is average maximum
potential demethylation rate
constant under anaerobic
conditions
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and
Verta 1995 is 2E-4 to 1E-3 /day;
value is average maximum
potential methylation rate
constant under anaerobic
conditions
Oxidation rate
1/day
0
0
0
Value assumed in U.S. EPA
1997
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil
(20cm), 10% moisture content,
in U.S. EPA 1997 (Lindberg
1996; Carpi and Lindberg, 1997)
Vadose Zone Soil Compartment Type
Input characteristic depth
(user supplied)
m
0.08
0.08
0.08
Not used (model set to calculate
value)
Soil-water partition
coefficient
L[water]/kg[soil wet
wtl
1,000
58,000
7,000
U.S. EPA 1997
Use input characteristic
depth (boolean)
0 = no, Else = yes
0
0
0
Professional judgment
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and
Verta 1995 is 3E-2 to 6E-2 /day;
value is average maximum
potential demethylation rate
constant under anaerobic
conditions
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and
Verta 1995 is 2E-4 to 1E-3 /day;
value is average maximum
potential methylation rate
constant under anaerobic
conditions
Oxidation rate
1/day
0
0
0
Value assumed in U.S. EPA
1997
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil
(20cm), 10% moisture content,
in U.S. EPA 1997 (Lindberg
1996; Carpi and Lindberg, 1997)
1-1-26
-------
Exhibit 1-15. Mercury Chemical-Specific Properties for Abiotic Compartments
Parameter Name
Units
Value
Reference
Hg(0)
Hg(2)
MHg
Ground Water Compartment Type
Soil-water partition
coefficient
L[water]/kg[soil wet
wtl
1,000
58,000
7,000
U.S. EPA 1997
Demethylation rate
1/day
N/A
N/A
0.06
Range reported in Porvari and
Verta 1995 is 3E-2 to 6E-2 /day;
value is average maximum
potential demethylation rate
constant under anaerobic
conditions
Methylation rate
1/day
0
0.001
0
Range reported in Porvari and
Verta 1995 is 2E-4 to 1E-3 /day;
value is average maximum
potential methylation rate
constant under anaerobic
conditions
Oxidation rate
1/day
1.00E-08
0
0
Small default nonzero value (0
assumed in U.S. EPA 1997)
Reduction rate
1/day
0
3.25E-06
0
Value used for tilled surface soil
(20cm), 10% moisture content,
in U.S. EPA 1997 (Lindberg
1996; Carpi and Lindberg, 1997)
Surface Water Compartment Type
Algal surface area-specific
uptake rate constant
nmol/[|jm2-day-
nmol]
0
2.04E-10
3.60E-10
Assumes radius = 2.5mm,
Mason et al. 1995b, Mason et
al. 1996; Hg(0) assumed same
as Hg(2)
Dow ("overall Kow")
L[ water]/
kgfoctanoll
0
a
b
Mason et al. 1996
Solids-water partition
coefficient
L[water]/
kg[solids wet wt]
1,000
100,000
100,000
U.S. EPA 1997
Vapor dry deposition
velocity
m/day
N/A
2500
U.S. EPA 1997 (Vol. Ill, App. A)
Demethylation rate
1/day
N/A
N/A
0.013
Average of range of 1E-3 to
2.5E-2/day from Gilmour and
Henry 1991
Methylation rate
1/day
0
0.001
0
Value used in EPA 1997; range
is from 1E-4 to 3E-4/day
(Gilmour and Henry 1991)
Oxidation rate
1/day
0
0
0
Professional judgment
Reduction rate
1/day
0
0.0075
0
Value used in EPA 1997;
reported values range from less
than 5E-3/day for depths greater
than 17m, up to 3.5/day (Xiao et
al. 1995; Vandal et al. 1995;
Mason et al. 1995a; Amyot et al.
1997)
1-1-27
-------
Exhibit 1-15. Mercury Chemical-Specific Properties for Abiotic Compartments
Parameter Name
Units
Value
Reference
Hg(0)
Hg(2)
MHg
Sediment Compartment Type
Solids-water partition
coefficient
L[water]/
kg[solids wet wt]
3,000
50,000
3,000
U.S. EPA 1997
Demethylation rate
1/day
N/A
N/A
0.0501
Average of range of 2E-4 to 1E-
1/day from Gilmour and Henry
1991
Methylation rate
1/day
0
1.00E-04
0
Value used in EPA 1997; range
is from 1E-5 to 1E-
3/day,Gilmour and Henry 1991
Oxidation rate
1/day
0
0
0
Professional judgment
Reduction rate
1/day
0
1.00E-06
0
Inferred value based on
presence of Hg(0) in sediment
porewater (U.S. EPA 1997;
Vandal etal. 1995)
TRIM.FaTE Formula Property, which varies from 0.025 to 1.625 depending on pH and chloride concentration.
'TRIM.FaTE Formula Property, which varies from 0.075 to 1.7 depending on pH and chloride concentration.
1-1-28
-------
Exhibit 1-16. Chemical-Specific Properties of 2,3,7,8-TCDD for Abiotic Compartments
Parameter Name
Units
Value Reference
Air Compartment Type
Deposition Velocity
m/day
500
McKone et al. 2001
Halflife
day
12
Atkinson 1996 as cited in USEPA 2000; vapor phase
reaction with hydroxyl radical
Washout Ratio
m3[air]/m3[rain]
18000
Vulykh et al. 2001
Surface Soil Compartment Type
Input characteristic
depth
m
0.08
Not used (model set to calculate value)
Use input characteristic
depth (boolean)
0 = No, Else = Yes
0
Professional judgment
Halflife
day
3650
Mackay et al. 2000; the degradation rate was cited
by multiple authors, value is for 2,3,7,8-TCDD
Root Zone Soil Compartment Type
Input characteristic
depth
m
0.08
Not used (model set to calculate value)
Use input characteristic
depth
0 = No, Else = Yes
0
Professional judgment
Halflife
day
3650
Mackay et al. 2000; the degradation rate was cited
by multiple authors, value is for 2,3,7,8-TCDD
Vadose Zone Soil Compartment Type
Input characteristic
depth
m
0.08
Not used (model set to calculate value)
Use input characteristic
depth (boolean)
0 = No, Else = Yes
0
Professional judgment
Halflife
day
1008
Average value of the range presented in Mackay et
al. 2000; based on estimated unacclimated aerobic
biodegradation half-life, value is for 2,3,7,8-TCDD
Groundwater Compartment Type
Half-life
day
1008
Average value of the range presented in Mackay et
al. 2000; based on estimated unacclimated aerobic
biodegradation half-life, value is for 2,3,7,8-TCDD
Surface Water Compartment Type
RatioOfConclnAlgaeTo
ConcDissolvedlnWater
(g[chem]/g[algae]) /
(g[chem]/L[water])
1.025
BCF data for green algae for 2,3,7,8-TCDD from
Isense 1978, at 32 days
Half-life
day
2.7
Kim, M., and P. O'Keefe. 1998. as cited in U.S. EPA.
2000.
Sediment Compartment Type
Half-life
day
1095
Estimation based on Adriaens and Grbic-Galic
1992,1993 and Adriaens et al. 1995 as cited in U.S.
EPA 2000.
1-1-29
-------
Exhibit 1-17. Mercury Chemical-Specific Properties for Plants a
Parameter Name
Units
Value
Reference
Hfl(O)
Hq(2)
MHq
Terrestrial Plants
Leaf Compartment Type
Professional judgment
Transfer factor to leaf
particle
1/day
0.002
0.002
0.002
(assumed 1% of transfer
factor from leaf particle to
leaf)
Demethylation rate
1/day
N/A
N/A
0.03
Calculated from Bache et
al. 1973
Methylation rate
1/day
0
0
0
Assumed from Gay 1975,
Bache et al. 1973
Professional judgment;
Oxidation rate
1/day
1.0E+06
0
0
Assumed close to
instantaneous
Reduction rate
1/day
0
0
0
Professional judgment
Particle on Leaf Compartment Type
Transfer factor to leaf
1/day
0.2
0.2
0.2
Professional judgment
Demethylation rate
1/day
N/A
N/A
0
Professional judgment
Methylation rate
1/day
0
0
0
Professional judgment
Oxidation rate
1/day
0
0
0
Professional judgment
Reduction rate
1/day
0
0
0
Professional judgment
Root Compartment Type - Nonwoody Plants Only
b
Alpha for root-root zone
bulk soil
unitless
0.95
0.95
0.95
Selected value
Hg2- geometric mean
Root/root-zone-soil-water
m3[bulk root
n
0.18
1.2
Leonard et al. 1998, John
1972, Hogg et al. 1978;
MHg- assumed, based on
Hogg et al. 1978
partition coefficient
soil]/m3[root]
t-alpha for root-root zone
bulk soil
day
21
21
21
Professional judgment
Demethylation rate
1/day
N/A
N/A
0
Professional judgment
Methylation rate
1/day
0
0
0
Professional judgment
Oxidation rate
1/day
0
0
0
Professional judgment
Reduction rate
1/day
0
0
0
Professional judgment
1-1-30
-------
Exhibit 1-17. Mercury Chemical-Specific Properties for Plants a
Parameter Name
Units
Value
Reference
Hfl(O)
Hfl(2)
MHq
Stem Compartment Type - Nonwoody Plants Only
Transpiration stream
concentration factor
(TSCF)
m3[soil pore
water]/m3[xylem
fluidl
0
0.5
0.2
Calculation from Norway
spruce, Scots pine, Bishop
et al. 1998
Demethylation rate
1/day
N/A
N/A
0.03
Calculated from Bache et
al. 1973
Methylation rate
1/day
0
0
0
Professional judgment
Oxidation rate
1/day
0
0
0
Professional judgment
Reduction rate
1/day
0
0
0
Professional judgment
Aquatic Plants
Macrophyte Compartment Type
Water Column Dissolve
Partitioning Alpha of
Equilibrium
unitless
0.95
0.95
0.95
Selected value
Macrophyte/water
partition coefficient
L[water]/kg[macr
ophyte wet wt]
0.883
q
4.4
Elodea densa, Ribeyre and
Boudou1994
Oxidation rate
1/day
1.00E+09
0
0
Professional judgment
t-alpha
day
18
18
18
Experiment duration from
Ribeyre and Boudou 1994
a TRIM.FaTE currently includes four kinds of terrestrial plants: deciduous forest (not used in screening scenario),
b Roots and stems are not modeled for deciduous or coniferous forest in the current version of TRIM.FaTE.
1-1-31
-------
Exhibit 1-18. Chemical-Specific Properties of 2,3,7,8-TCDD for Plants
Parameter Name
Units
Value
Reference
All Dioxins
Terrestrial Plants
Leaf Compartment Type
Transfer factor to leaf particle
1/day
0.003
Calculated as 1 percent of transfer
factor to leaf; highly uncertain
Half-life
day
70
Arjmand and Sandermann 1985,
as cited in Komoba, et al. 1995;
soybean root cell culture
metabolism test data for DDE
Particle on Leaf Compartment Type
Transfer factor to leaf
1/day
0.3
Professional judgment based on
U.S. EPA 2000c (an estimate for
mercury) and Trapp 1995; highly
uncertain
Half-life
day
4.4
McCrady and Maggard 1993;
photodegradation sorbed to grass
foliage in sunlight; assumed 10
sunlight per day
Root Compartment Type
Half-life
day
70
Arjmand and Sandermann 1985,
as cited in Komoba, et al. 1995;
soybean root cell culture
metabolism test data for DDE
Root Soil Water lnteraction_Alpha
unitless
0.95
Professional judgment
Stem Compartment Type
Half-life
day
70
Arjmand and Sandermann 1985,
as cited in Komoba, et al. 1995;
soybean root cell culture
metabolism test data for DDE
Aquatic Plants
Macrophyte Compartment Type
Half-life
days
70
Arjmand and Sandermann 1985,
as cited in Komoba, et al. 1995;
soybean root cell culture
metabolism test data for DDE
1-1-32
-------
Exhibit 1-19. Mercury Chemical-Specific Properties for Aquatic Species a
Parameter Name
Units
Value
Hq(0) Hq(2) MHq
Reference
Benthic Invertebrate Compartment Type
Alpha of equilibrium for
sediment partitioning
unitless
0.95
0.95
0.95
Selected value
Benthic invertebrate-bulk
sediment partition coefficient
kg[bulk sediment]/
kg[invertebrate wet
wt]
0.0824
0.0824
5.04
Hg(0) - assumed
based on Hg(2)
value; Hg(2) and
MHg - Saouter et
al. 1991
t-alpha for equilibrium for
sediment partitioning
day
14
14
14
Experiment
duration from
Saouter et al. 1991
All Fish Compartment Types
Demethylation rate
1/day
N/A
N/A
0
Professional
judgment
Methylation rate
1/day
0
0
0
Professional
judgment
Oxidation rate
1/day
1.0E+06
0
0
Professional
judgment
Reduction rate
1/day
0
0
0
Professional
judgment
Water-column Carnivore Compartment Type
Assimilation efficiency from
food
unitless
0.04
0.04
0.2
Phillips and
Gregory 1979
Elimination adjustment factor
unitless
3
3
1
Trudel and
Rasmussen 1997
Water-column Herbivore Compartment Type
Assimilation efficiency from
food
unitless
0.04
0.04
0.2
Phillips and
Gregory 1979
Elimination adjustment factor
unitless
3
3
1
Trudel and
Rasmussen 1997
Water-column Omnivore Compartment Type
Assimilation efficiency from
food
unitless
0.04
0.04
0.2
Phillips and
Gregory 1979
Elimination adjustment factor
unitless
3
3
1
Trudel and
Rasmussen 1997
Benthic Omnivore Compartment Type
Assimilation efficiency from
food
unitless
0.04
0.04
0.2
Phillips and
Gregory 1979
Elimination adjustment factor
unitless
3
3
1
Trudel and
Rasmussen 1997
a Screening scenario includes: Benthic Omnivore, Water-column Carnivore, Water-column Herbivore, Water-column
Omnivore.
1-1-33
-------
Exhibit 1-20. Chemical-Specific Properties of 2,3,7,8-TCDD for Aquatic Species
Parameter Name
Units
Value
Reference
Benthic Invertebrate Compartment Type
Clearance constant
unitless
0
Professional judgment
Sediment Partitioning Partition
Coefficient
kg/kg
0.107
TCDD data for sandworm in Rubenstein et al. 1990;
dry weight sediment. PeCDF: multiplied TCDD
partition coefficient for sandworm by congener-
specific bioaccumulation equivalency factor in GLWQI
from U.S. EPA 1999.
Sediment Partitioning Alpha of
Equilibrium
unitless
0.95
Professional judgment
Sediment Partitioning Time to
Reach Alpha of Equilibrium
days
120
TCDD: professional judgment; PeCDF: Rubinstein et
al. 1990; data for TCDF in sandworm.
V_d (ratio of concentration in
benthic invertebrates to
concentration in water)
ml/g
0
Professional judgment
Half-life
day
140
TCDD: estimated based on data for yellow perch in
Keeman et al. 1986b; PeCDF: Sijm et al. 1990 quoted
elimination rate for carp, metabolic rate calculated
assuming 9% metabolites like hepta and hexa
isomers as cited in Muir et al. 1986a
All Fish Compartment Types a
Assimilation efficiency from food
unitless
0.5
TCDD: calculated from data in Kleeman et al. 1986b
trout data as cited in U.S. EPA 1993; PeCDF: used
assimilation efficiency for TCDD in trout
Gamma fish
unitless
N/A b
Thomann 1989
Water Column Carnivore Compartment Type
Chemical Uptake Rate Via Gill
L[water]/
kg [fish wet wt]-
day
104
Muir et al. 1986
Half-life
day
160
TCDD: estimated based on data for rainbow trout in
Kleeman et al. 1986a; PeCDF: Sijm et al. 1990
quoted elimination rate for rainbow trout, metabolic
rate calculated assuming 9% metabolites like hepta
and hexa isomers cited in Muir et al. 1986a
Water Column Herbivore Compartment Type
Assimilation efficiency from
plants
unitless
0.5
TCDD: calculated from data in Kleeman et al. 1986b
trout data as cited in U.S. EPA 1993; PeCDF: used
assimilation efficiency for TCDD in trout
Chemical Uptake Rate Via Gill
L[water]/
kg [fish wet wt]-
day
380
Muir et al. 1986
Half-life
day
140
TCDD: estimated based on data for rainbow trout in
Kleeman et al. 1986a; PeCDF: Sijm et al. 1990
quoted elimination rate for rainbow trout, metabolic
rate calculated assuming 9% metabolites like hepta
and hexa isomers cited in Muir et al. 1986a
1-1-34
-------
Exhibit 1-20. Chemical-Specific Properties of 2,3,7,8-TCDD for Aquatic Species
Parameter Name
Units
Value
Reference
Water Column Omnivore Compartment Type
Assimilation efficiency from
plants
unitless
0.5
TCDD: calculated from data in Kleeman et al. 1986b
trout data as cited in U.S. EPA 1993; PeCDF: used
assimilation efficiency for TCDD in trout
Chemical Uptake Rate Via Gill
L[water]/
kg [fish wet wt]-
day
380
Muir et al. 1986
Half-life
day
140
TCDD: estimated based on data for rainbow trout in
Kleeman et al. 1986a; PeCDF: Sijm et al. 1990
quoted elimination rate for rainbow trout, metabolic
rate calculated assuming 9% metabolites like hepta
and hexa isomers cited in Muir et al. 1986a
Benthic Omnivore Compartment Type
Chemical Uptake Rate Via Gill
L[water]/k
g[fish wet wt]-day
380
Muir et al. 1986
Half-life
day
140
TCDD: estimated based on data for rainbow trout in
Kleeman et al. 1986a; PeCDF: Sijm et al. 1990
quoted elimination rate for rainbow trout, metabolic
rate calculated assuming 9% metabolites like hepta
and hexa isomers cited in Muir et al. 1986a
aScreening scenario includes: Benthic Omnivore, Water-column Carnivore, Water-column Herbivore, Water-column Omnivore.
bN/A = not applicable. This parameter is used in calculating the uptake when measured data are unavailable.
1-1-35
-------
l-1-2Supplemental Information for Exhibit 1-2 - Meteorological and
Other Settings
1-1-2.1 PCRAMMET
The EPA's PCRAMMET meteorological data processor1 is used to combine the surface
meteorological data with the twice-daily mixing height data into the format necessary for many
EPA air quality models. PCRAMMET requires that surface data be in either Solar and
Meteorological Surface Observational Network (SAMSON) format, Hourly United States
Weather Observations (HUSWO) format, or CD-144 format. For this study, the delimited format
downloaded from the ISH database is converted into SAMSON format using the converter
available from RF Lee Consulting.2 This converter does not accurately process the precipitation
information, so hourly precipitation is manually inserted into the SAMSON file that the converter
produces.
PCRAMMET also requires daily morning and afternoon mixing height data, which is not an
explicit field in the upper-air data from NOAA. The EPA's mixing height calculator3 is used to
generate morning and afternoon mixing heights from the FSL-formatted upper-air data. This
calculator uses the Holzworth methodology,4 which requires the morning (07 EST) upper-air
sounding and the daily observed minimum and maximum hourly surface temperatures in order
to perform the calculations.
PCRAMMET converts surface wind directions into vectors (i.e., converts wind directions to
'blowing to' rather than 'blowing from'). PCRAMMET also randomly applies to the wind vector a
variation of -4° to +5° in order to remove the directional bias from the hourly surface reports,
which record wind directions in increments of 10°. Wind directions of 0°, which indicate calm
winds, are set to the wind direction of the previous non-calm hour and then randomly varied by -
4° to +5°.
In PCRAMMET, the cloud layer with the greatest cloud coverage is used to represent the cloud
coverage for that hour. If the ceiling height observation is missing, then PCRAMMET sets the
ceiling height as the height of the lowest cloud layer with cloud coverage that is at least 'broken'
(at least 6/10 cloud coverage).
PCRAMMET interpolates twice-daily mixing heights into hourly values using the maximum
mixing height value from the previous, current, and upcoming day as well as the minimum
height value from the current and upcoming day. Then, two different methodologies are used to
derive the urban and rural mixing height values, respectively.5
For this study, PCRAMMET is set to calculate wet deposition fields. These fields include friction
velocity, Monin-Obukhov length, and roughness length.5
1 http://vwvw.epa. gov/scram001/metobsdata_procaccprogs.htm#pcrammet
2 http://www.rflee.com/RFL_Pages/Meteor.html
3 http://vvww.epa.gOv/scram001/metobsdata_procaccprogs.htm#mixing
4 Holzworth, G., 1967. Mixing Depths, Wind Speeds and Air Pollution Potential for Selected Locations in the United
States. J. Appl. Meteor., 6, 1039-1044.
5 http://www.epa.gov/scram001/userg/relat/pcramtd.pdf
1-1-36
-------
1-1-2.2 TRIM.FaTE Processing
Before the PCRAMMET-processed meteorological files can be processed by the TRIM.FaTE
meteorological processor, some fields must be filtered to replace missing values. The fields of
opaque sky cover, ceiling height, surface temperature, wind speed, wind direction, and mixing
heights cannot have any missing values. The EPA has provided procedures for objectively
replacing missing data in these fields.6 To generally summarize the missing value procedures,
missing opaque sky cover values are replaced by values that depend on the availability and
values of total sky cover and ceiling height; missing ceiling height values are replaced by values
that depend on the availability and values of total sky cover and opaque sky cover; and, missing
surface temperature, wind direction, wind speed, and mixing height values are interpolated from
the values of surrounding times. After these objective measures are used to replace missing
data in these fields, any remaining missing values are subjectively and manually replaced with
values based on observations from surrounding times. Exhibit 2-1 shows the completeness of
the various meteorological datasets and data fields in this study.
Exhibit 2-1. Completeness of Meteorological Data Types a
Year
Data Type
Statistic
Before Missing
Values Objectively
Replaced
After Missing
Values Objectively
Replaced
Hours Not Reporting
0.03%
0.03%
Hours Missing Opaque Sky Cover
2%
0.02%
Surface
Hours Missing Ceiling Height
0.07%
0.07%
Data
Hours Missing Temperature
0.1%
0.02%
2001
Hours Missing Wind Speed
3%
0.7%
Hours Missing Wind Direction
6%
2%
Soundings Not Reporting
5%
5%
Upper-Air
Data
Missing Calculated Morning Mixing
Heights
7%
2%
Missing Calculated Evening Mixing
Heights
3%
0%
Hours Not Reporting
0.06%
0.06%
Hours Missing Opaque Sky Cover
3%
0.07%
Surface
Hours Missing Ceiling Height
0.1%
0.1%
Data
Hours Missing Temperature
0.3%
0.07%
2002
Hours Missing Wind Speed
0.5%
0.09%
Hours Missing Wnd Direction
3%
1%
Soundings Not Reporting
5%
5%
Upper-Air
Data
Missing Calculated Morning Mixing
Heights
8%
5%
Missing Calculated Evening Mixing
Heights
6%
3%
2003
Surface
Hours Not Reporting
0.09%
0.09%
6 See http://vwvw.epa.gov/scram001/surface/missdata.txt.
1-1-37
-------
Exhibit 2-1. Completeness of Meteorological Data Types a
Year
Data Type
Statistic
Before Missing
Values Objectively
Replaced
After Missing
Values Objectively
Replaced
Data
Hours Missing Opaque Sky Cover
4%
0.05%
Hours Missing Ceiling Height
0.2%
0.2%
Hours Missing Temperature
2%
1%
Hours Missing Wind Speed
0.2%
0.07%
Hours Missing Wind Direction
3%
1%
Soundings Not Reporting
5%
5%
Upper-Air
Data
Missing Calculated Morning Mixing
Heights
6%
2%
Missing Calculated Evening Mixing
Heights
5%
0.8%
a The percentage of the 2001-2003 surface and upper-air reports that are completely missing ('Hours Not
Reporting', 'Soundings Not Reporting'), the percentage of non-missing hourly surface reports where specific
surface variables were missing, and the percentage of non-missing upper-air soundings where the morning or
afternoon mixing heights could not be calculated. These percentages of missing data are also shown after the
EPA's objective measures are employed to replace missing values.
Because the Ravena Lefarge Portland Cement scenario is modeled for 1990-2039, the 2001-
2003 meteorological data are duplicated. First, the data from 2002 are duplicated for 2004 (with
a leap day added, which was comprised of the data from 28 February) to create a complete
four-year cycle of data. Then, this four-year cycle of meteorological data are duplicated to fill
the modeling time period.
Finally, this 50-year set of meteorological data are processed by the TRIM.FaTE meteorological
processor. The TRIM.FaTE meteorological processor reverts the PC RAM MET-process wind
directions back into 'blowing from' designation and it converts hourly precipitation amounts
(previously in mm) to a daily precipitation rate in meters (m day"1). Calm wind speeds are set to
0.75 ms"1 so that chemical advection is always occurring. Mixing heights are set to a minimum
of 20 m. Daytime and nighttime hours are identified by inputting the latitude, longitude, and US
time zone of the meteorology station.
1-1-38
-------
l-1-3Supplemental Information for Exhibit 1-6 - Universal Soil Loss
Equations
1-1-3.1 Universal Soil Loss Equation
Sediment delivery for the parcels in this scenario were determined by using the Universal Soil
Loss Equation (USLE). The USLE is the most widely used empirical soil erosion model, which
estimates soil erosion from the product of empirically derived coefficients (Amore et al. 2004).
The values for these coefficients, and the equation itself, have been derived from over 10,000
plot-years of runoff and soil loss data (Pilotti et al. 1977).
The formula for the USLE is the product of five factors, as shown in the equation below:
A = R*K*LS*C*P
where:
A = Total
R = Rainfall/erosivity factor
K = Soil erodibility factor
LS = Combined length-slope factor
C = Cover management factor
P = Supporting practice factor
The USLE is intended to predict the long-term average soil losses from individual field areas
(Wischmeier and Smith, 1978), and represent the sheet and rill erosion from a small plot or
agricultural field. Application of the USLE to an entire watershed requires modification of the
result of the equation to account for subsequent re-deposition of eroded soil before reaching the
water body. The sediment delivery ratio, further described in Section 3.9., was developed for
this purpose, and is an additional factor to determine the amount of sediment that reaches a
water body based on watershed size (Vanoni 1975 in EPA 2005a).
Representative values were determined for each parcel use in the USLE, as outlined below.
1-1-3.2 Rainfall/erosivity Factor (R)
The rainfall/erosivity factor represents the erosive potential of the typical rainfall over a given
period (Wischmeier and Smith, 1978), and due to the typical cyclic nature of rainfall in a given
area can be considered constant for a given location. R values for this scenario were looked up
from county specific data in the Revised Universal Soil Loss Equation 2 (RUSLE2) software
(RUSLE2 was not used to calculate erosion predictions directly due to the intensive site analysis
required for this process, as discussed below). Data was available for both Albany county and
Rensselaer county; therefore, values for parcels located west of the Hudson river were
assumed to have R values the same as Albany county, and values for parcels located east of
the Hudson were assumed to have R values the same as Rensselaer county. These values
were consistent with regional maps of R values.
1-1-3.3 Soil Erodibility Factor (K)
Specific soil types have different natural susceptibilities to erosion, depending on the specific
makeup of their components (Wischmeier and Smith, 1978). To determine the site-specific K
values of the soils around the location in Ravena, NY, soil data was obtained from the Soil
Survey Geographic (SSURGO) database for the counties of interest, in the form of GIS
1-1-39
-------
shapefiles (obtained from Natural Resources Conservation Service with the USDA). The
percentage of each soil type present in a parcel was determined by including all polygons with
their centroid within the parcel. An area weighted average was determined for the three parcels
most driving risk with respect to sediment delivery - the three lake watershed parcels. Given the
very close similarity of these values, especially in comparison to the greater influence of
assumptions related to other USLE variables, the rest of the non-watershed parcels were
assigned the average value of these three parcels. Values for all parcels are given in Exhibit 3-1
below.
Exhibit 3-1. Soil Erodibility Factor for Watershed Parcels and
All Other Parcels
Watershed Parcel
Water Body
Soil Erodibility Factor
(ton/acre/(100 ft-ton/acre))
E2
Nassau Lake
0.3113
E3
Kinderhook Lake
0.3102
W3
Alcove Reservoir
0.3222
All other parcels
N/A
0.3145
1-1-3.4 Length Slope (LS) Factor
The amount of soil eroded from a given field increases as the slope increases, and as the length
of the field increases. For this assessment, the length slope factor was calculated from the
equation provided in Wischmeier and Smith (1978).
This equation for the length-slope factor is:
LS = (^—)m(65.41 sin2 0 + 4.56 sin 0 + 0.65)
72.6
where:
A = the slope length in feet
Q = angle of slope
m = 0.5 if the slope is 5% or more, 0.4 for slopes of 3.5-4.5%, 0.3 for slopes of 1-3%,
and 0.2 on slopes of less than 1 percent.
The field length is measured from the start of erosion to either a well defined channel or
decrease in slope sufficient enough for deposition to occur (Wischmeier and Smith, 1978). This
value would be different for each specific slope within a watershed, and exact evaluation of the
slope length requires detailed analysis of the watershed topography and evaluation of the USLE
for each slope. As an approximation for evaluation purposes, and consistent with the Dioxin
Reassessment (EPA 2004), an average field size was assumed to be 4 hectares. A square field
of 4 hectares translates to a side length of 200 meters. The length of a slope was assumed,
therefore, to be 200 meters in length, or 656 feet.
Average slope for each parcel was determined from GIS data of the topography of the four
counties included. This average slope was used in calculating the LS factor for each parcel.
1-1-40
-------
1-1-3.5 Cover Management Factor (C)
The type of ground cover present on a field plays a major factor in determining the amount of
soil eroded from a slope. Values of the cover management factor can range from less than
0.001 for dense grasses and undisturbed forestland to 1 for bare construction sites. C values
were determined from guidance in Wischmeier and Smith, RUSLE2 values for specific crop
cycles, and GIS data on land use for the area of interest.
For the four farm parcels, representative C values were looked up for specific crop types from
the RUSLE2 software. For unfilled parcels, C values for various grain crops ranged from 0.015
to 0.07, from which a conservative value of 0.07 was assumed (representative of no till corn 50
bushels/acre). For the tilled parcels, an area weighted C value was determined for the top three
vegetable crops reported in the agricultural census. This average covered 85% of all vegetable
crops reported. This method was seen to best represent the C value of farmland in the region,
as the most grown crop (sweet corn) also had the lowest C value, and as such using the C
value for only this crop would likely under-predict sediment delivery. Values for
watermelons/cucumbers were used as a surrogate for pumpkins, and green beans were used
as a surrogate for tomatoes, due to similar plant and growing styles.
For the non-farm parcels, area weighted cover management factors were determined from the
top three reported land uses in that parcel. Values were determined from tables provided by
Wischmeier and Smith (1978). Deciduous forest and Evergreen forest land uses were to have
75% tree cover, with weeds below at 80% ground cover. Mixed forests were assumed to 50%
tree cover and weeds for ground cover, also at 80%. Pasture/hay was assumed to have 80%
ground cover. Finally, quarries/strip mines/gravel pits were assumed to have a C value of 1,
consistent with no ground cover on construction sites.
1-1-3.6 Supporting Practice Factor (P)
Supporting practices include contour tillage, strip-cropping on the contour, and terracing. For
this assessment, no supporting practices were assumed, and therefore a value of 1 was
assigned for all parcels.
1-1-3.7 Total Erosion Losses Per Parcel
By using the above described approach, erosion rates were estimated for each parcel. The
values of these erosion losses were developed on a per-area basis; however, differences in
cover type, soil, and slope in the parcels yielded different per-area erosion rates. The calculated
rates are presented in Section 3.9.
1-1-3.8 Limitations to This Approach
The USLE is an empirical model, and therefore modeled conditions must be similar to
conditions for which the USLE is calibrated. In particular, the USLE is designed for application to
a single slope or field, rather than a whole watershed. Using average values across a watershed
parcel will likely introduce uncertainties in the prediction that would be better predicted by
individual analyses of the slopes within the watershed. It is noted in the HHRAP documentation
that using the universal soil loss equation to calculate sediment load to a lake from the
surrounding watershed can sometimes lead to overestimates (EPA 2005a).
The use of area weighted averages for some of the USLE variables does help to avoid this
problem in not under- or over-estimating by assuming uniformity across the watershed. The
1-1-41
-------
area weighted K (soil erodibility factor) and C (cover management factor) are not expected to
contribute significantly to an inaccurate estimate of soil erosion.
Estimation of the LS factor poses more problems than any other factor of the USLE (Moore &
Wilson, 1992), especially in complex watersheds. In the real watershed, an entire watershed
has neither uniform slope length nor uniform slope steepness. Additionally, due to nonlinearities
in the equation to calculate the LS factor, this assumption can introduce uncertainties into the
degree by which the LS factor is under- or over-estimated. The use of average slope likely will
under-predict the LS factor by some small degree. The use of an average slope-length of 200
meters may be accurate or slightly longer than average, and therefore may slightly over-predict
the LS factor by an unknown amount.
Finally, there is uncertainty in the use of the sediment delivery ratio (SD) to account for the re-
deposition of soil before it reaches the water body. It is not known the degree by which the SD
ratio will under- or over- predict actual sediment delivery.
1-1-3.9 Sediment Balance Calculations
The sediment balance of the watershed is determined by accounting for both the inputs of
sediment from the erosion calculations and the outputs of sediment through removal and burial.
In this scenario, assumptions about the physical environment were used in calculation of the
sediment input through erosion and removal through suspended sediment flushing. All sediment
inputs to the watershed come from the erosion calculations. The sediment delivery ratio
accounts for how much of that is re-deposited within the watershed.
The sediment delivery ratio is calculated using the following equation:
SD = a(A_L)b
where:
SD = sediment delivery ratio
a = empirical intercept coefficient
A_L = total watershed area receiving deposition
b = empirical slope coefficient
The value of the empirical intercept coefficient is determined based on watershed area (see
Exhibit 3-2). The empirical slope coefficient is a unitless constant set to 0.125.
Exhibit 3-2. USLE Empirical Intercept Coefficient
Area of Watershed
(sq. miles)
a
Area < 0.1
2.1
0.1 < Area < 1
1.9
1 < Area < 10
1.4
10 < Area < 100
1.2
Area > 100
0.6
Each parcel's sediment delivery ratio was calculated based on its size, and results are
presented below in Exhibit 3-3. Finally, the adjusted erosion rate was calculated by multiplying
1-1-42
-------
the USLE erosion rate by the sediment delivery ratio. The adjusted erosion rates are the final
erosion values used in the TRIM scenario.
Exhibit 3-3. Calculated USLE Soil Erosion Rates, Sediment Delivery
Ratios, and Adjusted Erosion Rates for Each Soil Parcel
Parcel
Erosion Rate
(USLE)
(kg/m2/day)
Sediment
Delivery Ratio
(unitless)
Adjusted Erosion
Rate (kg/m2/day)
E1
1.6E-03
0.122
1.9E-04
E2
2.3E-03
0.142
3.2E-04
E3
1.7E-03
0.133
2.2E-04
E4
2.3E-03
0.050
1.2E-04
E5
1.2E-03
0.124
1.5E-04
E6
1.2E-03
0.123
1.5E-04
E Farm Tilled
3.3E-03
0.384
1.3E-03
E Farm Untilled
1.3E-03
0.384
5.0E-04
W1
6.2E-04
0.131
8.2E-05
W2
2.6E-03
0.119
3.1E-04
W3
2.8E-03
0.122
3.4E-04
W4
1.8E-03
0.052
9.5E-05
W5
1.9E-03
0.123
2.3E-04
W6
2.1E-03
0.177
3.7E-04
W7
6.9E-03
0.309
2.1E-03
W8
3.3E-03
0.125
4.1E-04
WFarm Tilled
1.6E-02
0.384
6.1E-03
WFarm Untilled
1.2E-03
0.384
4.7E-04
1-1-43
-------
1-1-3.10 References
Amore, Elena. 2004. Scale effect in USLE and WEPP application for soil erosion computation
from three Sicilian basins. Journal of Hydrology.293:100-114.
Moore, I. D., and J. P. Wilson. 1992. Length-slope factors for the Revised Universal Soil Loss
Equation: simplified method of estimation. Soil and Water Cons. 47(5): 423-428.
Pilotti, M., and B. Bacchi. 1997. Distributed evaluation of the contribution of soil erosion to the
sediment yield from a watershed. Earth Surface Processes and Landforms. 22:1239-1251.
U.S. Environmental Protection Agency (EPA). 2004. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (TCDD) and Related Compounds.
Volume 3: Site-Specific Assessment Procedures, NAS Review Draft. U.S. Environmental
Protection Agency, Washington, D.C., EPA/600/P-00/001Cb. Available at:
http://www.epa.gov/ncea/pdfs/dioxin/nas-review/.
U.S. Environmental Protection Agency (EPA). 2005a. Human Health Risk Assessment Protocol
for Hazardous Waste Combustion Facilities (including the Hazardous Waste Companion
Database of chemical-specific parameter values). U.S. Environmental Protection Agency, Office
of Solid Waste and Emergency Response, Washington, DC. EPA-530-R-05-006. September.
Wschmeier, W. H., and D.D. Smith. 1978. Predicting rainfall erosion losses - a guide to
conservation planning. U.S. Department of Agriculture, Agriculture Handbook No. 537.
1-1-44
-------
l-1-4Supplemental Information for Exhibit 1-12 - Aquatic Animals
1-1-4.1 Introduction
To estimate risks to human health and the environment, site-specific models of aquatic food
webs were developed in TRIM.FaTE to represent the four water bodies in the vicinity of Ravena,
NY: Nassau and Kinderhook Lakes, Alcove Reservoir, and an unnamed small pond near the
facility. Characteristics of the fish compartments used to represent each water body were
based on site-specific fish survey data and some additional information from the open literature.
The development of each food web consisted of three stages:
1. Collection of local fish survey data for the water bodies from the New York State
Department of Environmental Conservation (NY DEC), including data on the relative
abundance and size/weight distribution of each species, to the extent available;
2. Formulation of simplified food webs for each water body, based on the fish surveys and
other biological and physical data for each water body, with supplemental information on
fish feeding habits, aquatic food webs, and biomass densities for different trophic levels
from the open literature; and
3. Assignment of values for the remaining parameters (e.g., individual body weight,
numeric density per unit area, lipid content) for each biotic compartment for each water
body in TRIM.FaTE from the available data.
These stages are discussed in greater detail in the sections below. Professional judgment was
used where available data were incomplete.
1-1-4.2 Collection of Information on Species Present in Water Bodies
To support the development of the aquatic food webs, fishery biologists at the NY DEC Region
4 Bureau of Fisheries were contacted. The NY DEC conducted surveys of fish in Nassau and
Kinderhook Lakes at various times between 1988 and 2006. Due to contamination at Nassau
and Kinderhook Lakes, there are fishing restrictions at these water bodies, and aquatic
sampling is performed to assess current contaminant levels. The New York State Fish and
Wildlife Department published the results of fish surveys conducted from 1963 to 1970 for
Alcove Reservoir (NY FWD 1971). No surveys were available for the small pond. Professional
judgment and published data from two small lakes in Canada were used to develop a model
food web for the small pond.
The 1971 Alcove Reservoir fish survey report also presented data on average fish weights,
which were used, where applicable, to estimate the average weight per individual fish for each
species in all water bodies. These data are summarized in Exhibit 4-1. Surveys of the Alcove
Reservoir have not been conducted since 1970 because the reservoir, which serves as a public
1-1-45
-------
Exhibit 4-1. Fish Survey Data for Alcove Reservoir3
Species
Year
Count
Total
Weight (lb)
Average
Weight (lb)
American eel (Anguilla rostrata)
1970
1
4.4
4.40
Black crappie
(Pomoxis nigromaculatus)
1966
6
4
0.67
1965
70
50
0.71
Bluegill (Lepomis macrochirus)
1970
50
20
0.40
1969
36
15
0.42
1968
72
24
0.33
1967
101
42
0.42
1966
81
27
0.33
1965
180
60
0.33
1964
125
55
0.44
1963
825
165
0.20
Bullhead (Ameiurus sp.)
1970
250
150
0.60
1969
40
45
1.13
1968
67
70
1.04
1967
65
65
1.00
1965
840
820
0.98
1964
34
30
0.88
1963
243
146
0.60
Chain pickerel (Esox niger)
1964
5
5
1.00
Largemouth bass
(Micropterus salmoides)
1969
10
12
1.20
1964
49
39
0.80
Northern pike (Esox lucius)
1969
6
8
1.33
1968
6
11
1.83
1967
7
6
0.86
1966
2
3
1.50
1964
3
2
0.67
Pumpkinseed
(,Lepomis gibbosus)
1966
50
17
0.34
1965
170
65
0.38
1964
75
30
0.40
1963
421
84
0.20
Redbreast sunfish (Lepomis sp.)
1964
75
25
0.33
Smallmouth bass
(Micropterus dolomieu)
1970
50
38
0.76
1969
60
64
1.07
1968
98
122
1.24
1967
176
244
1.39
1966
115
130
1.13
1965
30
35
1.17
1964
174
170
0.98
1963
89
100
1.12
Walleye (Stizostedion vitreum)
1970
12
26
2.17
1-1-46
-------
Exhibit 4-1. Fish Survey Data for Alcove Reservoir3
Species
Year
Count
Total
Weight (lb)
Average
Weight (lb)
1969
10
14
1.40
1968
21
33
1.57
1967
167
204
1.22
1966
33
49
1.48
1965
31
42
1.35
1963
7
10
1.43
White perch (Morone americana)
1964
9
8
0.89
White sucker (Catostomus sp.)
1970
3
6
2.00
Yellow perch (Perca flavescens)
1970
10
4
0.40
1968
108
49
0.45
1967
104
48
0.46
1966
30
19
0.63
1965
140
110
0.79
1964
16
14
0.88
1963
324
96
0.30
a NY FWD, 1971.
drinking water supply, has been closed to public fishing. Because data on fish length or weight
were not available for the other water bodies, average fish weights for each species from the
Alcove report were used as the average fish weights for the same species in the other water
bodies. The relevant survey data provided by NY DEC for Nassau and Kinderhook Lakes are
summarized in Exhibit 4-2 and Exhibit 4-3 (NY DEC 2008b).
While these surveys provide some indication of the relative abundance of different fish species
in each water body over the periods of time represented, they do not indicate the absolute
abundance of each species. No estimates were available for total fish standing stock (e.g., total
biomass in the water body or biomass per unit area of the water body). In addition, potential
biases introduced by selection of sampling times and locations and fish capture techniques
have not been evaluated; information on the sampling methods (i.e., gill netting, electro-fishing)
for Nassau and Kinderhook Lakes was not available at the time of this analysis. Personal
communication with the Daniel Zielinski of NY DEC indicated that the fish surveys occurred at or
after dusk, and that the timing of the sampling could have a large effect on the number and type
of fish collected (NY DEC 2008a). Nonetheless, the best available estimates of the relative
abundance of each species in the lakes and reservoir are the relative abundance of each
species as reported in the fish surveys.
The food web for the small pond was developed from an analysis of data presented by Demers
et al. (2001) for two small lakes in Canada. As a conservative position, the small pond is
assumed to sustain a viable fish community from year to year. In each water body, young of the
year were assumed to comprise 15 percent of the total fish biomass on an annual basis
biomass.
1-1-47
-------
Exhibit 4-2. Fish Survey Data for Nassau Lakea
Species
Year
Count
American eel (Anguilla rostrata)
1989
36
2001
3
Black crappie (Pomoxis nigromaculatus)
1989
3
2001
2
Bluegill (Lepomis macrochirus)
1989
100
1997
70
2001
7
Brown bullhead (Ameiurus nebulosus)
1989
2
1997
15
2001
18
Chain pickerel (Esox niger)
1988
4
1989
1
1997
2
Common carp (Cyprinus carpio)
1997
2
2001
1
Golden shiner (Notemigonus crysoleucas)
1989
12
1997
2
Largemouth bass (Micropterus salmoides)
1988
58
1989
16
1997
146
2001
17
Pumpkinseed (Lepomis gibbosus)
1989
100
1997
75
2001
11
Redbreast sunfish (Lepomis auritus)
1997
1
Smallmouth bass (Micropterus dolomieu)
1997
5
2001
3
White perch (Morone americana)
1988
20
1989
10
2001
5
White sucker (Catostomus commersonii)
1989
20
2001
2
Yellow bullhead (Ameiurus nataiis)
2001
3
Yellow perch (Perca fiavescens)
1989
310
1997
321
2001
21
a NY DEC, 2008b.
1-1-48
-------
Exhibit 4-3. Fish Survey Data for Kinderhook Lakea
Species
Year
Count
American eel (Anguilla rostrata)
1993
1
1998
4
2006
10
Black crappie (Pomoxis nigromaculatus)
1998
10
2001
24
2006
5
Bluegill (Lepomis macrochirus)
1998
27
2001
30
2006
12
Brown bullhead (Ameiurus nebulosus)
1998
2
2001
10
2006
1
Common carp (Cyprinus carpio)
1998
100
2001
53
2006
28
Fantail darter (Etheostoma flabellare)
2001
1
Golden shiner (Notemigonus crysoleucas)
1998
1
2001
21
2006
2
Largemouth bass (Micropterus salmoides)
1988
50
1993
20
1998
66
2001
64
2006
49
Pumpkinseed (Lepomis gibbosus)
1998
13
2001
21
2006
38
Redbreast sunfish (Lepomis auritus)
1998
10
2001
1
Rock bass (Ambloplites sp.)
2006
1
Smallmouth bass (Micropterus doiomieu)
1998
7
2001
51
2006
90
Sunfish family (Centrarchidae sp.)
2006
1
Tiger musky (Masquinongy sp.)
2006
1
Walleye (Stizostedion vitreum)
2006
1
1-1-49
-------
Exhibit 4-3. Fish Survey Data for Kinderhook Lakea
Species
Year
Count
White perch (Morone americana)
1988
18
1993
20
1998
19
2001
618
2006
504
White sucker (Catostomus commersonii)
1998
2
2001
12
2006
9
Yellow bullhead (Ameiurus natalis)
2001
1
2006
1
Yellow perch (Perca flavescens)
1998
171
2001
163
2006
97
a NY DEC, 2008b.
1-1-4.3 Creation of Food Webs
Food webs for each of the four water bodies were constructed from the information sources
identified above. Several steps were required to construct each food web and to assign
parameter values for all aquatic biotic compartments for TRIM.FaTE:
1. Estimate total standing fish stock (i.e., total fish biomass per unit area) for each water
body based on total biomass estimates reported for similar water bodies in the literature;
2. List for each water body all fish species found in the surveys of the water body;
3. Identify for each species an average body weight per individual based on the Alcove
Reservoir data;
4. Estimate total biomass caught for each species in the surveys by multiplying the number
of individuals of each species caught over the survey years for the water body by the
average body weight per individual for each species;
5. Estimate the relative total biomass for each species (percentage of total biomass
represented in surveys);
6. Estimate the absolute biomass of each species by multiplying its percent relative
biomass by the estimated total standing fish stock (Step 1);
7. Estimate the numeric density of each fish species (number per unit area) based on
biomass density and average individual weight for each species; and
8. Evaluate the feeding habits of each fish species, as determined from a variety of
sources, relative to the food/prey categories supported by TRIM.FaTE:
. plankton (called algae; however, it represents both phytoplankton and
zooplankton);
1-1-50
-------
. macrophytes;
. benthic invertebrates (e.g., aquatic insects, crustaceans, mollusks);
. small planktivorous fish (e.g., young of the year, minnows; feed on algae and
zooplankton in the water column);
. larger omnivorous fish that feed on smaller fish in the water column and benthic
invertebrates and/or macrophytes (e.g., sunfish, yellow perch)
. small-to-medium sized benthivores/omnivores that feed primarily on benthic
invertebrates, detritus, and possibly macrophytes (e.g., small carp, white sucker).
Additionally, the lipid content of each species was estimated based on values reported in
national surveys.
The total fish standing stock (biomass density of all fish species expressed as kilograms [kg] of
fish [wet weight] per hectare [ha] of lake surface area) was determined from literature sources
for similar water bodies in other locations. For Alcove Reservoir, standing stock estimates from
comparably sized reservoirs from the literature were reviewed. Lynn and Tygart Reservoirs in
West Virginia both have surface areas of approximately 1,700 acres (Yurk and Ney 1989),
slightly larger than Alcove's estimated 1,360 acres. In the late 1980s, Yurk and Ney (1989)
estimated the standing stock in these reservoirs, which were regularly fished, to be 77 and 104
kg/ha, respectively. We note, however, that these water bodies are, on average, 50 to 100
percent deeper than Alcove Reservoir. Because Alcove Reservoir is not fished, we considered
it appropriate to assume a standing stock of 80 kg/ha for Alcove Reservoir despite its more
shallow depth relative to Lynn and Tygart Reservoirs.
The smallest lake discussed by Yurk and Ney (1989) had the smallest standing stock (34
kg/ha), so we assumed that a value in this range would be appropriate for Kinderhook and
Nassau Lakes, which are 134.1 and 64.9 hectares in surface area, respectively. Scaling
standing stock proportionally to lake surface area suggests that the small unnamed pond near
the facility might be too small to sustain game or panfish over the long term. We therefore used
the lowest standing stock reported in Yurk and Ney (1989) as a floor for estimating fish standing
stock for the pond. Exhibit shows the standing stock values selected for all four water bodies.
Exhibit 4-4. Estimated Total Fish Standing Stocks for Water
Bodies Near Lafarge Facility
Water body
Stock (kg/ha)
Stock (g/m2)
Alcove Reservoir
80
8
Nassau Lake
50
5
Kinderhook Lake
50
5
Small "farm" pond
40
4
The initial estimates of relative abundance for each fish species were based on the fish survey
data. Only the species identified by fish surveys were assumed to be present in the four
modeled water bodies. The body weight of each individual was assumed to be equal to the
average fish weight estimated from the Alcove surveys. When species were present in the
other lakes, but not in the Alcove Reservoir, professional judgment and data from other
locations (e.g., Minnesota fish surveys) were used to estimate an average individual body
weight for the species.
1-1-51
-------
At the small pond, only three species/groups were assumed to be present: largemouth bass,
sunfish (e.g., bluegill or pumpkinseed), and shiners. The mass of the individuals was estimated
based on professional judgment and the Demers et al. (2001) study of two small lakes.
Total relative biomass for each species within a water body was estimated differently for Alcove
Reservoir, Nassau and Kinderhook Lakes, and the small pond. At Alcove, each species'
biomass representation was determined by taking the observed biomass of the species caught
across all survey years and dividing that by the total fish biomass reported in the Alcove report
across all survey years. At both Nassau and Kinderhook Lakes, the survey data seemed biased
towards several species, specifically yellow and white perch, perhaps due to sampling
techniques. The biomass representation was therefore adjusted to reflect a more balanced
abundance across different species for these two water bodies. At the small pond, the
distribution of biomass among the several species was estimated based on the Demers et al.
study (2001).
Fish lipid content was estimated from data collected for the 1978 - 1981 National Contaminant
Biomonitoring Program (compiled from Lowe et al. 1985). Because Lowe et al. (1985) did not
report the lipid content of American eels, we used measurements of eel lipid content from a U.S.
Army study of Lake Cochituate in Natick, MA (ICF Consulting 2001). The fish lipid estimates
are presented in Exhibit 4-5.
Exhibit 4-5. Lipid Content for Fish Species Included in Model
Food Webs
Fish species
Lipid content
Source
American eel
16.9%
ICF, 2001
Black crappie
5.7%
Lowe et al., 1985
Bluegill
3.1%
Lowe et al., 1985
Bullhead
2.8%
Lowe et al., 1985
Chain pickerel
1.8%
Mierzykowski and Carr, 2004
Common carp
3.6%
Lowe et al., 1985
Fantail darter
3.5%
Professional judgment
Golden shiner
3.5%
Lowe et al., 1985
Largemouth bass
3.3%
Lowe et al., 1985
Northern pike
2.9%
Lowe et al., 1985
Pumpkinseed
1.5%
Lowe et al., 1985
Redbreast sunfish
5.9%
Lowe et al., 1985
Rock bass
6.2%
Lowe et al., 1985
Smallmouth bass
4.4%
Lowe et al., 1985
Sunfish
5.9%
Lowe et al., 1985
Tiger musky
4.0%
Professional judgment
Walleye
7.9%
Lowe et al., 1985
Young of the year
3.5%
Professional judgment
White perch
17.1%
EPA, 1990b
White sucker
5.1%
Lowe et al., 1985
Yellow perch
4.3%
Lowe et al., 1985
1-1-52
-------
Exhibit 4-6 presents the diets that we assumed for each fish species when determining which
TRIM.FaTE fish compartment in which to include the biomass for that species. These
assumptions are based on abundance of prey species in each water body and on abundance of
species that compete for similar food sources.
Finally, each species was assigned to one of the TRIM.FaTE fish compartment categories as
discussed above.
Exhibit 4-6. Aquatic Species Diets by TRIM.FaTE Model Compartments
Fish Species
Algae/
zooplankton
Macrophytes
Benthic
invertebrates
Benthic
omnivores
Water column
herbivore
Water column
omnivore
American eel
50%
50%
Black crappie
50%
50%
Bluegill (Kinderhook)3
100%
Bluegill (Nassau and Alcove)
50%
50%
Bullhead
100%
Chain pickerel
25%
25%
50%
Common carp
100%
Fantail darter
100%
Golden shiner
100%
Largemouth bass
25%
25%
50%
Largemouth bass (pond)b
50%
50%
Northern Pike
25%
25%
50%
Pumpkinseed
25%
75%
Redbreast sunfish
100%
Rock bass
50%
50%
Smallmouth bass
50%
50%
Tiger musky
50%
50%
White perch
100%
White sucker
25%
75%
Yellow perch
25%
50%
25%
Walleye
50%
25%
25%
Young of the year
100%
a Bluegills in Kinderhook were assumed to feed primarily on invertebrates rather than on water
column herbivores, as they are assumed to do in Nassau and Alcove, because Kinderhook contains
a large population of white perch who also feed exclusively on the relatively sparse herbivorous
population.
Bass in the small pond are assumed to feed in part on benthic invertebrates because most of the
omnivorous fish are assumed to be too large for the bass to swallow.
1-1-53
-------
1-1-4.4 Parameterization of Fish Compartments to be Included in Application
Exhibit 4-7 through Exhibit 4-10 summarize the fish compartments constructed using the
methods discussed in Section 1-1-4.3. All fish species were assigned to one of the following five
fish compartments established in TRIM.FaTE:
. Water column carnivore (WCC - large predominantly piscivorous species, e.g.,
walleye and largemouth bass);
. Water column omnivore (WCO - medium-sized fish that feed primarily in the water
column, e.g., sunfish, yellow perch; see Section 1-1-4.3, Step 8 bullets);
. Water column herbivore (WCH - more appropriately termed planktivore, e.g., black
crappie);
. Benthic carnivore (BC - large carnivorous species, e.g., large bullhead, eel); and
. Benthic omnivore (BO - medium-sized fish that feed primarily on benthic
invertebrates; see Section 1-1-4.3).
The compartment to which each species was assigned was determined by its general foraging
habitat (i.e., benthic or water column) and its primary food sources (e.g., invertebrates, smaller
fish, plant material). The total biomass for each of the five fish TRIM.FaTE compartments was
set equal to the sum of the biomass of the species assigned to each compartment.
Exhibit 4-7. Small Pond Parameters: Fish Mass, Abundance, and Model Representation
Fish Species
Individual
Mass
(g)a
Count
(ha"1)b
Total
Countc
Biomass d
(g ww/m2)
Percentage
Biomass e
Model
Compartmentf
Largemouth bass
1000
2
4
0.2
5.0%
WCC
Sunfish
250
120
240
3
75.0%
BO
Golden shiner
25
80
160
0.2
5.0%
WCH
Young of the year
50
120
240
0.6
15.0%
WCH
Total
322
644
4
100%
a Average individual fish body weights based on data from Alcove Reservoir.
b Abundance or numerical fish density per hectare based on total fish abundance, survey data from Alcove
Reservoir, and individual fish body weights.
c Total abundance or total numerical fish count based on total fish abundance, survey data from Alcove Reservoir,
and individual fish body weights.
d Biomass density per square meter; calculated by multiplying average individual body weight by numeric density per
square meter.
e Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
'TRIM.FaTE model compartment to which this species is assigned.
1-1-54
-------
Exhibit 4-8. Alcove Reservoir Parameters: Fish Mass, Abundance,
and Model Representation
Fish Species
Individual
Mass a
(g)
Count b
(ha"1)
Total
Countc
Biomass d
(g ww/m2)
Percentage
Biomass e
Model
Compartmentf
American eel
2100
0.08
42
0.016
0.2%
BC
Black crappie
350
2.74
1512
0.096
1.2%
WCH
Bluegill
110
58.18
32064
0.64
8.0%
WCO
Bullhead
220
109.1
60120
2.4
30.0%
BO
Chain pickerel
280
0.29
157
0.008
0.1%
WCC
Largemouth bass
410
2.34
1290
0.096
1.2%
WCC
Northern pike
540
1.19
653
0.064
0.8%
WCC
Pumpkinseed
120
33.33
18370
0.4
5.0%
WCO
Redbreast sunfish
120
4.00
2204
0.048
0.6%
WCO
Smallmouth bass
260
61.54
33914
1.6
20.0%
WCO
Walleye
490
15.84
8728
0.776
9.7%
WCC
White perch
900
0.18
98
0.016
0.2%
WCO
Yellow perch
250
25.60
14108
0.64
8.0%
WCO
Young of the year
50
0.24
132
0.0012
15.0%
WCH
Total
315
173597
8
100%
a Average individual fish body weights based on data from Alcove Reservoir.
b Abundance or numerical fish density per hectare based on total fish abundance, survey data from Alcove
Reservoir, and individual fish body weights.
c Total abundance or total numerical fish count based on total fish abundance, survey data from Alcove Reservoir,
and individual fish body weights.
d Biomass density per square meter; calculated by multiplying average individual body weight by numeric density
per square meter.
e Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
f TRIM.FaTE model compartment to which this species was assigned.
1-1-55
-------
Exhibit 4-9. Nassau Lake Parameters: Fish Mass, Abundance, and Model Representation
Fish Species
Individual
Mass a
(g)
Count b
(ha"1)
Total
Countc
Biomass d
(g ww/m2)
Percentage
Biomass e
Model
Compartmentf
American eel
2100
0.12
8
0.025
0.5%
BC
Black crappie
350
2.57
167
0.09
1.8%
WCH
Bluegill
110
34.09
2213
0.375
7.5%
WCO
Bullhead
220
56.82
3688
1.25
25.0%
BO
Chain pickerel
280
0.18
12
0.005
0.1%
WCC
Common carp
300
0.83
54
0.025
0.5%
WCH
Golden shiner
10
50.00
3245
0.05
1.0%
WCH
Largemouth bass
410
1.83
119
0.075
1.5%
WCC
Pumpkinseed
120
26.67
1731
0.32
6.4%
WCO
Redbreast sunfish
120
27.08
1758
0.325
6.5%
WCO
Smallmouth bass
260
40.96
2658
1.065
21.3%
WCO
White perch
900
0.11
7
0.01
0.2%
WCO
White sucker
400
6.25
406
0.25
5.0%
WCO
Yellow perch
250
15.40
999
0.385
7.7%
CO
Young of the year
50
0.15
10
0.00075
15.0%
WCH
Total
263
17069
5
100%
a Average individual fish body weights based on data from Alcove Reservoir
b Abundance or numerical fish density per hectare based on total fish abundance, survey data from Nassau Lake,
and individual fish body weights.
c Total abundance or total numerical fish count based on total fish abundance, survey data from Nassau Lake, and
individual fish body weights.
d Biomass density per square meter; calculated by multiplying average individual body weight by numeric density
per square meter.
e Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
f TRIM.FaTE model compartment to which this species was assigned.
1-1-56
-------
Exhibit 4-10. Kinderhook Lake Parameters: Fish Mass, Abundance, and Model
Representation
Fish Species
Individual
Mass a
(g)
Count b
(ha"1)
Total
Countc
Biomass d
(g ww/m2)
Percentage
Biomass e
Model
Compartmentf
American eel
2000
0.08
10
0.015
0.3%
BC
Black crappie
350
2.86
383
0.1
2.0%
WCH
Bluegill
110
36.36
4876
0.4
8.0%
WCO
Bullhead
220
28.41
3810
0.625
12.5%
BO
Common carp
300
4.75
637
0.1425
2.85%
WCH
Fantail darter
5
50.00
6705
0.025
0.5%
WCH
Golden shiner
10
37.50
5029
0.0375
0.75%
WCH
Largemouth bass
410
2.74
368
0.1125
2.25%
WCC
Pumpkinseed
120
18.75
2514
0.225
4.5%
WCO
Redbreast sunfish
120
20.00
2682
0.24
4.8%
WCO
Smallmouth bass
260
36.54
4900
0.95
19.0%
WCO
White perch
900
4.44
596
0.4
8.0%
WCO
White sucker
400
5.63
754
0.225
4.5%
WCO
Yellow perch
250
20.00
2682
0.5
10.0%
WCO
Rock bass
225
0.22
30
0.005
0.1%
WCO
Walleye
490
4.85
650
0.2375
4.75%
WCC
Tiger musky
500
0.20
27
0.01
0.2%
WCC
Young of the year
50
4.28
573
0.021375
15.0%
WCH
Total
278
37280
5
100%
a Average individual fish body weights based on data from Alcove Reservoir.
b Abundance or numerical fish density per hectare based on total fish abundance, survey data from Kinderhook
Lake, and individual fish body weights.
c Total abundance or total numerical fish count based on total fish abundance, survey data from Kinderhook Lake,
and individual fish body weights.
d Biomass density per square meter; calculated by multiplying average individual body weight by numeric density
per square meter.
e Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
f TRIM.FaTE model compartment to which this species was assigned.
The diet composition for each of the five fish compartments was calculated as being
proportional to the biomass representation of each species assigned to that compartment. For
example, if largemouth bass comprised 75 percent and smallmouth bass comprised 25 percent
of the biomass of the WCC compartment, then the diet composition of the largemouth bass
multiplied by 0.75 would be added to the diet composition of the smallmouth bass multiplied by
0.25 to estimate the diet composition for the WCC compartment.
Similarly, the lipid content for each of the five fish compartments in TRIM.FaTE was estimated
from the biomass-weighted lipid content of the individual species assigned to the compartment.
Thus, using the same example, the largemouth bass lipid content, multiplied by 0.75, would be
added to the smallmouth bass lipid content, multiplied by 0.25, to estimate the lipid content of
the WCC compartment.
1-1-57
-------
Exhibit 4-11 through Exhibit 4-14 present the numeric and biomass densities for each model
compartment as well as the lipid content used to represent the compartment. Exhibit 4-15
through Exhibit 4-18 present the diet composition used for each fish compartment derived as
described above.
Exhibit 4-11. Small Pond Model Parameters: Fish Mass, Abundance (Number per
Hectare), and Lipid Content
Model Compartment3
Individual
Mass b
(g)
Countc
(ha"1)
Total
Countd
Biomass e
(g ww/m2)
Percentage
Biomassf
Lipid
Content
Water column carnivore
1000
2
4
0.2
5.0%
3.3%
Water column herbivore
44
182.9
366
0.8
20.0%
3.5%
Benthic omnivore
250
120
240
3
75.0%
5.9%
Total
305
610
4
100%
aTRIM.FaTE model compartment.
b Average individual fish body weights for each compartment based on biomass-weighted species-specific
individual fish body weights.
c Abundance or numerical fish density per hectare
d Total abundance or total numerical fish count.
e Biomass density per square meter; calculated by multiplying average individual body weight by numeric
density per square meter.
'Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
Exhibit 4-12. Alcove Reservoir Model Parameters: Fish Mass, Abundance,
and Lipid Content
Model Compartmenta
Individual
Mass b
(g)
Countc
(ha"1)
Total
Countd
Biomass e
(g ww/m2)
Percentage
Biomassf
Lipid
Content
Water column carnivore
483
19.53
10760
0.944
11.8%
7.0%
Water column omnivore
214
156.49
86243
3.344
41.8%
3.9%
Water column herbivore
72
179.45
98893
1.296
16.2%
3.7%
Benthic carnivore
2100
0.08
42
0.016
0.2%
16.9%
Benthic omnivore
220
109.09
60120
2.4
30.0%
2.8%
Total
465
256058
8
100%
aTRIM.FaTE model compartment.
b Average individual fish body weights for each compartment based on biomass-weighted species-specific
individual fish body weights.
c Abundance or numerical fish density per hectare
d Total abundance or total numerical fish count.
e Biomass density per square meter; calculated by multiplying average individual body weight by numeric
density per square meter.
'Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
1-1-58
-------
Exhibit 4-13. Nassau Lake Model Parameters: Fish Mass, Abundance,
and Lipid Content
Model Compartment3
Individual
Mass b
(g)
Countc
(ha"1)
Total
Countd
Biomass e
(g ww/m2)
Percentage
Biomassf
Lipid
Content
Water column carnivore
402
1.99
129
0.08
1.6%
3.2%
Water column omnivore
220
124.05
8051
2.73
54.6%
4.2%
Water column herbivore
84
108.73
7057
0.915
18.3%
3.7%
Benthic carnivore
2100
0.12
8
0.025
0.5%
16.9%
Benthic omnivore
220
56.82
3688
1.25
25.0%
2.8%
Total
291.7
18932
5
100%
aTRIM.FaTE model compartment.
b Average individual fish body weights for each compartment based on biomass-weighted species-specific
individual fish body weights.
c Abundance or numerical fish density per hectare
d Total abundance or total numerical fish count.
e Biomass density per square meter; calculated by multiplying average individual body weight by numeric
density per square meter.
'Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
Exhibit 4-14. Kinderhook Lake Model Parameters: Fish Mass, Abundance,
and Lipid Content
Model Compartmenta
Individual
Mass b
(g)
Countc
(ha"1)
Total
Countd
Biomass e
(g ww/m2)
Percentage
Biomassf
Lipid
Content
Water column carnivore
465
7.74
1038
0.36
7.2%
6.4%
Water column omnivore
313
93.97
12602
2.945
58.9%
5.9%
Water column herbivore
110
96.16
12895
1.055
21.1%
3.7%
Benthic carnivore
2000
0.08
10
0.015
0.3%
16.9%
Benthic omnivore
220
28.41
3810
0.625
12.5%
2.8%
Total
226.4
30354
5
100%
aTRIM.FaTE model compartment.
b Average individual fish body weights for each compartment based on biomass-weighted species-specific
individual fish body weights.
c Abundance or numerical fish density per hectare
d Total abundance or total numerical fish count.
e Biomass density per square meter; calculated by multiplying average individual body weight by numeric
density per square meter.
'Percentage biomass; calculated by dividing biomass per unit area by total biomass per unit area.
1-1-59
-------
Exhibit 4-15. Small Pond Aquatic Food Web
TRIM.FaTE Model
Compartment
Algae/
zooplankton
Macrophytes
Benthic
invertebrates
Benthic
omnivores
Water column
herbivore
Water column
omnivore
Water column carnivore
50.0%
50.0%
Water column herbivore
100.0%
Benthic omnivore
100.0%
Exhibit 4-16. Alcove Reservoir Aquatic Food Web
TRIM.FaTE Model
Compartment
Algae/
zooplankton
Macrophytes
Benthic
invertebrates
Benthic
omnivores
Water column
herbivore
Water column
omnivore
Water column carnivore
41.0%
25.0%
4.4%
29.4%
Water column omnivore
7.8%
53.5%
38.8%
Water column herbivore
96.3%
3.7%
Benthic carnivore
50.0%
50.0%
Benthic omnivore
100.0%
Exhibit 4-17. Nassau Lake Aquatic Food Web
TRIM.FaTE Model
Compartment
Algae/
zooplankton
Macrophytes
Benthic
invertebrates
Benthic
omnivores
Water column
herbivore
Water column
omnivore
Water column carnivore
25.0%
25.0%
50.0%
Water column omnivore
8.7%
61.0%
30.3%
Water column herbivore
92.3%
2.7%
4.9%
Benthic carnivore
50.0%
50.0%
Benthic omnivore
100.0%
1-1-60
-------
Exhibit 4-18. Kinderhook Lake Aquatic Food Web
TRIM.FaTE Model
Compartment
Algae/
zooplankton
Macrophytes
Benthic
invertebrates
Benthic
omnivores
Water column
herbivore
Water column
omnivore
Water column carnivore
33.0%
25.7%
7.8%
33.5%
Water column omnivore
8.1%
57.9%
34.0%
Water column herbivore
81.8%
13.5%
4.7%
Benthic carnivore
50.0%
50.0%
Benthic omnivore
100.0%
1-1-4.5 Fish Harvesting from Ravena Pond
During the development of the conceptual exposure model for the risk assessment of the
Ravena facility, we judged that the possibility of an angler exposure scenario existing for the
Ravena pond was low. It is assumed that the angler fishes at the pond regularly for a lifetime
and consumes his or her catch. Due to the small size of the pond, however, it is unlikely that
this water body could sustain fishable populations at the assumed ingestion rates without
regular, substantial restocking offish.
In order to obtain a realistic estimate of concentrations in fish in the Ravena Pond, we modified
the TRIM.FaTE scenario and incorporated a fish harvest rate from the pond of 17 g/day to
represent consumption and restocking of the pond within the TRIM.FaTE model. This harvest
rate corresponds to the 90th percentile fish ingestion rate for an adult angler used to calculate
hazard quotients and lifetime cancer risks from consumption of contaminated fish. For the
purposes of this assessment, we also assumed that the angler consumes two types of fish —
largemouth bass (33% of total consumption) and benthic omnivores (67% of total consumption)
1-1-61
-------
1-1-4.6 References
Demers, E; McQueen, DJ; Ramcharan, CW; Perez-Fuentetaja, A. 2001. Did piscivores rulate
changes in fish community structure? Arch. Hydrobiol. Spec. Issues Advanc. Limnol. 56(1): 49-
80.
ICF Consulting, Inc. 2001. Tier III Deterministic Ecological Risk Assessment. Final Report.
Prepared for U.S. Army Soldier Systems Center (SSC), Kansas Street, Natick, MA 01760.
Lowe, TP; May, TW; Brumbaugh, WG; Kane, DA. 1985. National contaminant biomonitoring
program: Concentrations of seven elements in freshwater fish, 1978-1981. Arch. Environ.
Contam. Toxicol. 14: 363-388.
Mierzykowski, SE; Carr, KC. 2004. Contaminant Survey of Sunkhaze Stream and Baker Brook
- Sunkhaze Meadows National Wildlife Refuge. U.S. FWS (Fish and Wildlife Service). Maine
Field Office. Spec. Proj. Rep. FY04-MEFO-2-EC. Old Town, ME. Available at:
http://www.fws.gov/northeast/mainecontaminants/PDF%20files/2004%20Sunkhaze%20Report
%20Final.pdf
NY DEC (New York State Department of Environmental Conservation). 2008a. Personal
communication between Daniel Zielinski, NY DEC, and Leiran Biton, ICF International, March
12.
NY DEC (New York State Department of Environmental Conservation). 2008b. Printout of fish
survey results from 1988 through 2006 for Kinderhook and Nassau Lakes, forwarded by
Norman R. McBride, NYDEC, to Leiran Biton, ICF International, March 3.
NY FWD (New York State Fish and Wldlife Department). 1971. Fishery Survey of Alcove
Reservoir.
U.S. EPA (Environmental Protection Agency). 1990. Lake Ontario TCDD Bioaccumulation
Study Final Report. Cooperative study including US EPA, New York State Department of
Environmental Conservation, New York State Department of Health, and Occidental Chemical
Corporation. As quoted in The 1994 EPA Dioxin Reassessment - Exposure Document.
Available at: http://www.cqs.com/epa/exposure/
Yurk, JJ; Ney, JJ. 1989. Phosphorous-fish community biomass relationships in southern
Appalachian reservoirs: can lakes be too clean? Lake Reservoir Manag. 5:83-90.
1-1-62
-------
ATTACHMENT 1-2: Detailed Ravena Human Health Assessment
Exposure, Risk, and Hazard Quotient Estimates
-------
[This page intentionally left blank.]
-------
TABLE OF CONTENTS
1-2.1 Detailed Ravena Exposure Concentrations 1
I-2.2 Detailed 2,3,7,8 - TCDD Cancer Risk and Hazard Quotient Results 11
I-2.3 Detailed Mercury Hazard Quotient Results 25
l-2-i
-------
LIST OF EXHIBITS
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-
Specific TRIM.FaTE Scenario, Including Modeled Fish Harvesting from the Ravena Pond 2
Exhibit 1-2. Annually Averaged Concentrations at Year 50 for All Compartments in the TRIM.FaTE
Screening Scenario using Ravena Emission Rates3 8
Exhibit 2-1. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using
UCL Emission Rate and RME Ingestion Rates 12
Exhibit 2-2. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using
UCL Emission Rate and CTE Ingestion Rates 13
Exhibit 2-3. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using
Mean Emission Rate and RME Ingestion Rates 14
Exhibit 2-4. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using
Mean Emission Rate and CTE Ingestion Rates 15
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for 2,3,7,8-TCDD for
all Ravena Scenarios, with Harvester in Pond 16
Exhibit 2-6. Comparison of 2,3,7,8-TCDD Ravena Cancer Risks Using UCL Versus Mean Emission Rates
a 22
Exhibit 2-7. Comparison of 2,3,7,8-TCDD Ravena Cancer Risks Using CTE Versus RME Ingestion Rates
a 23
Exhibit 2-8. 2,3,7,8 TCDD Dermal Hazard Quotients and Lifetime Risks for Water and Soil Contact 24
Exhibit 3-1. Summary Results - Hazard Quotients for Divalent Mercury Scenarios using RME Ingestion
Rates 26
Exhibit 3-2. Summary Results - Hazard Quotients for Divalent Mercury Scenarios using CTE Ingestion
Rates 27
Exhibit 3-3. Summary Results - Hazard Quotients for Methyl Mercury Scenarios using RME Ingestion
Rates 28
Exhibit 3-4. Summary Results - Hazard Quotients for Methyl Mercury Scenarios using CTE Ingestion
Rates 29
Exhibit 3-5. Fractional Pathway of Age-Specific Hazard Quotients for Divalent Mercury for all Ravena
Scenarios, with Harvester in Pond 30
Exhibit 3-6. Fractional Pathway of Age-Specific Hazard Quotients for Methyl Mercury for all Ravena
Scenarios, with Harvester in Pond 33
Exhibit 3-7. Mercury Hazard Quotients in All Water Bodies Compared to Alcove Reservoir3 36
Exhibit 3-8. Comparison of Hazard Quotients Using East Farm Parcels Versus West Farm Parcels in the
Ravena Scenario 37
Exhibit 3-9. Comparison of Hazard Quotients for Ravena Scenario Using CTE and RME Ingestion Rates
38
Exhibit 3-10. Mercury Dermal Hazard Quotients for Water and Soil Contact 39
l-2-ii
-------
1-2.1 Detailed Ravena Exposure Concentrations
Exhibit 1-1 presents annually averaged media concentrations at the 50th year for all
compartments in the Ravena site-specific TRIM.FaTE scenario for 2,3,7,8-TCDD and mercury,
including modeled fish harvesting from the Ravena Pond. The 2,3,7,8 - TCDD emissions
include both a 95th percentile upper confidence limit (UCL) emission and a mean emission. The
reported total mercury emissions from the Ravena facility were separated into elemental and
divalent species based on a speciation factor of 75% elemental and 25% divalent. Methyl
mercury is created via transformation reactions in the environment. Total mercury
concentrations (i.e., the sum of mass or concentrations of these three species) are presented as
well.
Exhibit 1-2 presents annually averaged media concentrations at the 50th year for all
compartments in the TRIM.FaTE screening scenario, using the Ravena emission rates.
1-2-1
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Air
A
r source
i-ig/nr3
4.E-08
9.E-08
1.E-03
4.E-03
9.E-11
6.E-03
Air
A
r 1
i-ig/nr3
1.E-10
3.E-10
2.E-06
1.E-05
1.E-11
2.E-05
Air
A
r 2
i-ig/nr3
2.E-10
4.E-10
3.E-06
2.E-05
3.E-11
3.E-05
Air
A
r 3
i-ig/nr3
5.E-10
1.E-09
1.E-05
6.E-05
4.E-11
7.E-05
Air
A
r 4
|ig/m3
6.E-10
1.E-09
1.E-05
7.E-05
9.E-11
9.E-05
Air
A
r 5
|ig/m3
3.E-10
7.E-10
5.E-06
4.E-05
6.E-11
4.E-05
Air
A
r 6
|ig/m3
1.E-10
4.E-10
2.E-06
2.E-05
4.E-11
2.E-05
Air
A
r 7
|ig/m3
4.E-10
1.E-09
1.E-05
5.E-05
5.E-11
6.E-05
Air
A
r 8
|ig/m3
8.E-10
2.E-09
2.E-05
1.E-04
5.E-11
1.E-04
Air
A
r 9
|ig/m3
1.E-09
3.E-09
3.E-05
1.E-04
1.E-10
2.E-04
Air
A
r 10
|ig/m3
6.E-10
2.E-09
2.E-05
8.E-05
1.E-10
9.E-05
Air
A
r 11
|ig/m3
2.E-09
5.E-09
7.E-05
2.E-04
5.E-11
3.E-04
Air
A
r 12
|ig/m3
3.E-09
8.E-09
1.E-04
4.E-04
6.E-11
5.E-04
Air
A
r 13
|ig/m3
4.E-09
9.E-09
1.E-04
4.E-04
8.E-11
6.E-04
Air
A
r 14
|ig/m3
3.E-09
7.E-09
9.E-05
3.E-04
8.E-11
4.E-04
Air
A
r 15
|ig/m3
3.E-09
6.E-09
9.E-05
3.E-04
6.E-11
4.E-04
Air
A
r 16
|ig/m3
4.E-09
1.E-08
1.E-04
5.E-04
7.E-11
6.E-04
Air
A
r 17
lig/m3
4.E-09
1.E-08
1.E-04
5.E-04
9.E-11
6.E-04
Air
A
r 18
lig/m3
3.E-09
7.E-09
1.E-04
4.E-04
8.E-11
5.E-04
Air
A
r 19
lig/m3
2.E-10
4.E-10
3.E-06
2.E-05
2.E-11
2.E-05
Air
A
r 20
lig/m3
3.E-10
7.E-10
5.E-06
3.E-05
6.E-11
4.E-05
Air
A
r 21
lig/m3
6.E-10
2.E-09
2.E-05
8.E-05
6.E-11
9.E-05
Air
A
r 22
lig/m3
1.E-09
3.E-09
3.E-05
1.E-04
1.E-10
2.E-04
Air
A
r 23
lig/m3
1.E-09
3.E-09
3.E-05
1.E-04
1.E-10
1.E-04
Air
A
r 24
lig/m3
8.E-10
2.E-09
2.E-05
9.E-05
1.E-10
1.E-04
Air
A
r 25
lig/m3
3.E-10
8.E-10
7.E-06
4.E-05
8.E-11
5.E-05
Air
A
r 26
|ig/m3
2.E-10
4.E-10
3.E-06
2.E-05
5.E-11
2.E-05
Air
A
r 27
lig/m3
5.E-10
1.E-09
9.E-06
6.E-05
5.E-11
7.E-05
Air
A
r 28
lig/m3
4.E-10
1.E-09
8.E-06
5.E-05
1.E-10
6.E-05
Air
A
r 29
lig/m3
3.E-10
7.E-10
4.E-06
3.E-05
3.E-11
4.E-05
Air
A
r 30
lig/m3
2.E-10
5.E-10
3.E-06
3.E-05
9.E-11
3.E-05
Air
A
r_Layer2
lig/m3
1.E-12
3.E-12
5.E-08
2.E-07
0.E+00
2.E-07
Surface water
SW AR
mg/L
1.E-14
3.E-14
1.E-07
8.E-08
2.E-09
2.E-07
Water Column Carnivore
SW AR
mg/kg wet weight
8.E-09
2.E-08
7.E-05
2.E-12
6.E-04
6.E-04
Water Column Herbivore
SW AR
mg/kg wet weight
9.E-11
2.E-10
1.E-04
4.E-13
6.E-05
2.E-04
Water Column Omnivore
SW AR
mg/kg wet weight
1.E-09
3.E-09
9.E-05
2.E-12
2.E-04
3.E-04
1-2-2
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Macrophyte
SW AR
mg/kg wet weight
2.E-10
4.E-10
1.E-07
4.E-17
4.E-09
1.E-07
Mallard
SW AR
|ig/g wet weight
3.E-08
7.E-08
2.E-04
7.E-07
4.E-05
3.E-04
Sediment
Sed AR
|ig/g dry weight
2.E-10
4.E-10
6.E-03
9.E-05
1.E-05
6.E-03
Benthic Invertebrate
Sed AR
mg/kg wet weight
2.E-11
4.E-11
3.E-04
5.E-06
4.E-05
4.E-04
Benthic Omnivore
Sed AR
mg/kg wet weight
9.E-10
2.E-09
1.E-04
3.E-12
1.E-04
2.E-04
Benthic Carnivore
Sed AR
mg/kg wet weight
5.E-09
1.E-08
8.E-05
2.E-12
5.E-04
6.E-04
Surface water
SW KL
mg/L
3.E-14
7.E-14
2.E-07
5.E-08
3.E-09
3.E-07
Water Column Carnivore
SW KL
mg/kg wet weight
3.E-08
7.E-08
9.E-05
2.E-12
9.E-04
1.E-03
Water Column Herbivore
SW KL
mg/kg wet weight
3.E-10
7.E-10
2.E-04
2.E-13
7.E-05
3.E-04
Water Column Omnivore
SW KL
mg/kg wet weight
6.E-09
1.E-08
1.E-04
2.E-12
3.E-04
4.E-04
Macrophyte
SW KL
mg/kg wet weight
4.E-10
1.E-09
1.E-07
2.E-17
5.E-09
1.E-07
Mallard
SW KL
|ig/g wet weight
7.E-08
2.E-07
3.E-04
6.E-07
7.E-05
4.E-04
Sediment
Sed KL
|ig/g dry weight
4.E-10
1.E-09
9.E-03
7.E-05
2.E-05
9.E-03
Benthic Invertebrate
Sed KL
mg/kg wet weight
4.E-11
1.E-10
5.E-04
4.E-06
6.E-05
5.E-04
Benthic Omnivore
Sed KL
mg/kg wet weight
2.E-09
4.E-09
2.E-04
5.E-12
2.E-04
3.E-04
Benthic Carnivore
Sed KL
mg/kg wet weight
9.E-09
2.E-08
1.E-04
2.E-12
7.E-04
8.E-04
Surface water
SW NL
mg/L
4.E-14
9.E-14
2.E-07
2.E-08
2.E-09
2.E-07
Water Column Carnivore
SW NL
mg/kg wet weight
1.E-07
3.E-07
5.E-05
4.E-19
1.E-03
1.E-03
Water Column Herbivore
SW NL
mg/kg wet weight
3.E-10
8.E-10
2.E-04
1.E-13
8.E-05
2.E-04
Water Column Omnivore
SW NL
mg/kg wet weight
1.E-08
3.E-08
1.E-04
2.E-12
3.E-04
4.E-04
Macrophyte
SW NL
mg/kg wet weight
6.E-10
1.E-09
9.E-08
2.E-17
5.E-09
1.E-07
Mallard
SW NL
|ig/g wet weight
9.E-08
2.E-07
2.E-04
3.E-07
5.E-05
3.E-04
Sediment
Sed NL
|ig/g dry weight
5.E-10
1.E-09
7.E-03
4.E-05
2.E-05
7.E-03
Benthic Invertebrate
Sed NL
mg/kg wet weight
6.E-11
1.E-10
4.E-04
2.E-06
5.E-05
4.E-04
Benthic Omnivore
Sed NL
mg/kg wet weight
8.E-09
2.E-08
1.E-04
3.E-12
2.E-04
3.E-04
Benthic Carnivore
Sed NL
mg/kg wet weight
4.E-08
1.E-07
9.E-05
1.E-12
7.E-04
8.E-04
Surface water
SW Pond
mg/L
5.E-12
1.E-11
1.E-04
5.E-06
5.E-07
1.E-04
Mink
SW Pond
|ig/g wet weight
2.E-09
5.E-09
3.E-05
4.E-07
3.E-05
7.E-05
Mallard
SW Pond
|ig/g wet weight
1.E-05
3.E-05
3.E-02
9.E-06
6.E-03
4.E-02
Water Column Carnivore
SW Pond
mg/kg wet weight
1.E-06
3.E-06
3.E-02
4.E-11
1.E-01
1.E-01
Water Column Herbivore
SW Pond
mg/kg wet weight
4.E-07
9.E-07
1.E-01
0.E+00
6.E-02
2.E-01
Macrophyte
SW Pond
mg/kg wet weight
8.E-08
2.E-07
5.E-05
7.E-15
1.E-06
5.E-05
Sediment
Sed Pond
|ig/g dry weight
3.E-08
6.E-08
9.E-01
1.E-03
2.E-03
9.E-01
Benthic Invertebrate
Sed Pond
mg/kg wet weight
3.E-09
6.E-09
5.E-02
6.E-05
6.E-03
5.E-02
Benthic Omnivore
Sed Pond
mg/kg wet weight
2.E-06
4.E-06
2.E-02
8.E-11
3.E-02
4.E-02
Surface water
SW River
mg/L
6.E-14
1.E-13
1.E-07
9.E-09
3.E-09
1.E-07
Sediment
Sed River
|ig/g dry weight
8.E-10
2.E-09
6.E-03
2.E-05
1.E-05
6.E-03
1-2-3
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Soil - Surface
SurfSo
1 source
|ig/g dry weight
6.E-07
1.E-06
9.E+00
1.E-03
1.E-01
9.E+00
Soil - Surface
SurfSo
1 W1
|ig/g dry weight
2.E-08
5.E-08
5.E-02
6.E-06
9.E-04
5.E-02
Leaf - Grasses/Herbsa
SurfSo
I W1
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSo
I W1
mg/kg wet weight
1.E-06
3.E-06
5.E-02
9.E-08
3.E-09
5.E-02
Root - Grasses/Herbs
SurfSo
I W1
mg/kg wet weight
1.E-13
3.E-13
1.E-06
0.E+00
1.E-07
1.E-06
Stem - Grasses/Herbs
SurfSo
I W1
mg/kg wet weight
9.E-10
2.E-09
8.E-05
4.E-11
5.E-12
8.E-05
Soil - Surface
SurfSo
I W2
|ig/g dry weight
2.E-08
4.E-08
4.E-02
5.E-06
7.E-04
4.E-02
Leaf - Deciduous Forest3
SurfSo
I W2
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W2
mg/kg wet weight
1.E-06
3.E-06
2.E-02
9.E-08
3.E-10
2.E-02
Soil - Surface
SurfSo
I W3
|ig/g dry weight
3.E-09
8.E-09
8.E-03
1.E-06
1.E-04
8.E-03
Leaf - Deciduous Forest3
SurfSo
I W3
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W3
mg/kg wet weight
1.E-07
3.E-07
5.E-03
8.E-09
9.E-11
5.E-03
Soil - Surface
SurfSo
I W4
|ig/g dry weight
6.E-09
1.E-08
1.E-02
2.E-06
2.E-04
1.E-02
Leaf - Deciduous Forest3
SurfSo
I W4
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W4
mg/kg wet weight
2.E-07
6.E-07
1.E-02
2.E-08
2.E-10
1.E-02
Soil - Surface
SurfSo
I W5
|ig/g dry weight
9.E-09
2.E-08
2.E-02
3.E-06
4.E-04
2.E-02
Leaf - Deciduous Forest3
SurfSo
I W5
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W5
mg/kg wet weight
4.E-07
1.E-06
2.E-02
3.E-08
3.E-10
2.E-02
Soil - Surface
SurfSo
I W6
|ig/g dry weight
2.E-08
5.E-08
1.E-01
1.E-05
2.E-03
1.E-01
Leaf - Deciduous Forest3
SurfSo
I W6
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W6
mg/kg wet weight
1.E-06
2.E-06
8.E-02
7.E-08
9.E-10
8.E-02
Soil - Surface
SurfSo
I W7
|ig/g dry weight
2.E-08
4.E-08
5.E-02
6.E-06
8.E-04
5.E-02
Leaf - Deciduous Forest3
SurfSo
I W7
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W7
mg/kg wet weight
1.E-06
3.E-06
2.E-01
7.E-08
8.E-10
2.E-01
Soil - Surface
SurfSo
I W8
|ig/g dry weight
1.E-08
3.E-08
4.E-02
5.E-06
7.E-04
4.E-02
Leaf - Deciduous Forest3
SurfSo
I W8
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I W8
mg/kg wet weight
5.E-07
1.E-06
2.E-02
4.E-08
4.E-10
2.E-02
Soil - Surface
SurfSo
I E1
|ig/g dry weight
7.E-09
2.E-08
2.E-02
2.E-06
3.E-04
2.E-02
Leaf - Deciduous Forest3
SurfSo
I E1
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I E1
mg/kg wet weight
5.E-07
1.E-06
2.E-02
3.E-08
5.E-10
2.E-02
Soil - Surface
SurfSo
I E2
|ig/g dry weight
3.E-09
8.E-09
8.E-03
1.E-06
1.E-04
8.E-03
Leaf - Deciduous Forest3
SurfSo
I E2
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I E2
mg/kg wet weight
3.E-07
7.E-07
2.E-02
2.E-08
3.E-10
2.E-02
Soil - Surface
SurfSo
I E3
|ig/g dry weight
4.E-09
9.E-09
1.E-02
1.E-06
2.E-04
1.E-02
Leaf - Deciduous Forest3
SurfSo
I E3
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I E3
mg/kg wet weight
2.E-07
5.E-07
1.E-02
2.E-08
4.E-10
1.E-02
Soil - Surface
SurfSo
I E4
|ig/g dry weight
4.E-09
1.E-08
1.E-02
1.E-06
2.E-04
1.E-02
1-2-4
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Leaf - Deciduous Forest3
SurfSo
I E4
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Deciduous Forest
SurfSo
I E4
mg/kg wet weight
2.E-07
5.E-07
9.E-03
2.E-08
4.E-10
9.E-03
Soil - Surface
SurfSo
I E5
|ig/g dry weight
1.E-08
3.E-08
3.E-02
4.E-06
5.E-04
3.E-02
Leaf - Grasses/Herbs a
SurfSo
I E5
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSo
I E5
mg/kg wet weight
5.E-07
1.E-06
2.E-02
4.E-08
2.E-09
2.E-02
Root - Grasses/Herbs
SurfSo
I E5
mg/kg wet weight
6.E-14
2.E-13
8.E-07
0.E+00
9.E-08
9.E-07
Stem - Grasses/Herbs
SurfSo
I E5
mg/kg wet weight
4.E-10
9.E-10
4.E-05
2.E-11
4.E-12
4.E-05
Soil - Surface
SurfSo
I E6
|ig/g dry weight
9.E-09
2.E-08
3.E-02
3.E-06
5.E-04
3.E-02
Leaf - Grasses/Herbs a
SurfSo
I E6
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSo
I E6
mg/kg wet weight
5.E-07
1.E-06
3.E-02
4.E-08
2.E-09
3.E-02
Root - Grasses/Herbs
SurfSo
I E6
mg/kg wet weight
6.E-14
1.E-13
7.E-07
0.E+00
8.E-08
8.E-07
Stem - Grasses/Herbs
SurfSo
I E6
mg/kg wet weight
4.E-10
9.E-10
4.E-05
2.E-11
3.E-12
4.E-05
Soil - Surface
SurfSo
I WFT
|ig/g dry weight
8.E-10
2.E-09
4.E-03
1.E-04
7.E-05
5.E-03
Leaf - Agriculture - Generala
SurfSo
I WFT
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Agriculture - General
SurfSo
I WFT
mg/kg wet weight
1.E-06
3.E-06
2.E-01
5.E-08
2.E-10
2.E-01
Root - Agriculture - General
SurfSo
I WFT
mg/kg wet weight
2.E-15
4.E-15
4.E-08
0.E+00
4.E-09
4.E-08
Stem - Agriculture - General
SurfSo
I WFT
mg/kg wet weight
7.E-10
2.E-09
3.E-04
4.E-11
3.E-13
3.E-04
Soil - Surface
SurfSo
I WFU
|ig/g dry weight
1.E-08
2.E-08
5.E-02
6.E-06
9.E-04
5.E-02
Leaf - Agriculture - Generala
SurfSo
I WFU
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Agriculture - General
SurfSo
I WFU
mg/kg wet weight
5.E-07
1.E-06
6.E-02
2.E-08
1.E-09
6.E-02
Root - Agriculture - General
SurfSo
I WFU
mg/kg wet weight
6.E-14
2.E-13
1.E-06
0.E+00
2.E-07
2.E-06
Stem - Agriculture - General
SurfSo
I WFU
mg/kg wet weight
3.E-10
9.E-10
9.E-05
4.E-11
3.E-12
9.E-05
Soil - Surface
SurfSo
I EFT
|ig/g dry weight
8.E-10
2.E-09
5.E-03
1.E-04
8.E-05
5.E-03
Leaf - Agriculture - Generala
SurfSo
I EFT
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Agriculture - General
SurfSo
I EFT
mg/kg wet weight
1.E-06
3.E-06
2.E-01
4.E-08
5.E-10
2.E-01
Root - Agriculture - General
SurfSo
I EFT
mg/kg wet weight
2.E-15
4.E-15
4.E-08
0.E+00
4.E-09
4.E-08
Stem - Agriculture - General
SurfSo
I EFT
mg/kg wet weight
5.E-10
1.E-09
3.E-04
2.E-11
6.E-13
3.E-04
Soil - Surface
SurfSo
I EFU
|ig/g dry weight
1.E-08
3.E-08
7.E-02
8.E-06
1.E-03
7.E-02
Leaf - Agriculture - Generala
SurfSo
I EFU
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Agriculture - General
SurfSo
I EFU
mg/kg wet weight
1.E-06
3.E-06
2.E-01
4.E-08
2.E-09
2.E-01
Root - Agriculture - General
SurfSo
I EFU
mg/kg wet weight
9.E-14
2.E-13
2.E-06
0.E+00
2.E-07
2.E-06
Stem - Agriculture - General
SurfSo
I EFU
mg/kg wet weight
5.E-10
1.E-09
3.E-04
2.E-11
5.E-12
3.E-04
Soil - Root Zone
RootSoil source
|ig/g dry weight
6.E-11
1.E-10
7.E-04
7.E-04
1.E-05
1.E-03
Soil - Root Zone
RootSoil W1
|ig/g dry weight
2.E-12
5.E-12
4.E-06
5.E-06
7.E-08
9.E-06
Soil - Root Zone
RootSoil W2
|ig/g dry weight
2.E-12
4.E-12
4.E-06
4.E-06
6.E-08
8.E-06
Soil - Root Zone
RootSoil W3
|ig/g dry weight
3.E-13
8.E-13
7.E-07
8.E-07
1.E-08
1.E-06
Soil - Root Zone
RootSoil W4
|ig/g dry weight
6.E-13
1.E-12
1.E-06
1.E-06
2.E-08
2.E-06
1-2-5
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
So
1 - Root Zone
RootSoil W5
|ig/g dry weight
9.E-13
2.E-12
2.E-06
2.E-06
3.E-08
4.E-06
So
1 - Root Zone
RootSoil W6
|ig/g dry weight
2.E-12
6.E-12
1.E-05
1.E-05
2.E-07
2.E-05
So
1 - Root Zone
RootSoil W7
|ig/g dry weight
2.E-12
4.E-12
5.E-06
6.E-06
9.E-08
1.E-05
So
1 - Root Zone
RootSoil W8
|ig/g dry weight
1.E-12
3.E-12
3.E-06
4.E-06
6.E-08
7.E-06
So
1 - Root Zone
RootSoil E1
|ig/g dry weight
8.E-13
2.E-12
2.E-06
2.E-06
3.E-08
3.E-06
So
1 - Root Zone
RootSoil E2
|ig/g dry weight
4.E-13
9.E-13
7.E-07
8.E-07
1.E-08
1.E-06
So
1 - Root Zone
RootSoil E3
|ig/g dry weight
4.E-13
1.E-12
8.E-07
1.E-06
1.E-08
2.E-06
So
1 - Root Zone
RootSoil E4
|ig/g dry weight
4.E-13
1.E-12
9.E-07
1.E-06
2.E-08
2.E-06
So
1 - Root Zone
RootSoil E5
|ig/g dry weight
1.E-12
3.E-12
3.E-06
3.E-06
4.E-08
5.E-06
So
1 - Root Zone
RootSoil E6
|ig/g dry weight
1.E-12
2.E-12
2.E-06
3.E-06
4.E-08
5.E-06
So
1 - Root Zone
RootSoil WFT
|ig/g dry weight
3.E-14
7.E-14
1.E-07
4.E-05
2.E-09
4.E-05
So
1 - Root Zone
RootSoil WFU
|ig/g dry weight
1.E-12
3.E-12
4.E-06
5.E-06
7.E-08
9.E-06
So
1 - Root Zone
RootSoil EFT
|ig/g dry weight
3.E-14
7.E-14
1.E-07
5.E-05
2.E-09
5.E-05
So
1 - Root Zone
RootSoil EFU
|ig/g dry weight
2.E-12
4.E-12
6.E-06
7.E-06
1.E-07
1.E-05
So
1 - Vadose Zone
VadoseSoil source
g/g dry we
ght
1.E-24
3.E-24
3.E-17
5.E-13
4.E-19
5.E-13
So
1 - Vadose Zone
VadoseSoil W1
g/g dry we
ght
4.E-26
1.E-25
2.E-19
3.E-15
3.E-21
3.E-15
So
1 - Vadose Zone
VadoseSoil W2
g/g dry we
ght
3.E-26
8.E-26
1.E-19
3.E-15
2.E-21
3.E-15
So
1 - Vadose Zone
VadoseSoil W3
g/g dry we
ght
7.E-27
2.E-26
3.E-20
5.E-16
4.E-22
5.E-16
So
1 - Vadose Zone
VadoseSoil W4
g/g dry we
ght
1.E-26
3.E-26
4.E-20
9.E-16
7.E-22
9.E-16
So
1 - Vadose Zone
VadoseSoil W5
g/g dry we
ght
2.E-26
5.E-26
7.E-20
1.E-15
1.E-21
1.E-15
So
1 - Vadose Zone
VadoseSoil W6
g/g dry we
ght
5.E-26
1.E-25
4.E-19
7.E-15
6.E-21
7.E-15
So
1 - Vadose Zone
VadoseSoil W7
g/g dry we
ght
4.E-26
9.E-26
2.E-19
5.E-15
4.E-21
5.E-15
So
1 - Vadose Zone
VadoseSoil W8
g/g dry we
ght
2.E-26
6.E-26
1.E-19
3.E-15
2.E-21
3.E-15
So
1 - Vadose Zone
VadoseSoil E1
g/g dry we
ght
2.E-26
4.E-26
6.E-20
1.E-15
1.E-21
1.E-15
So
1 - Vadose Zone
VadoseSoil E2
g/g dry we
ght
8.E-27
2.E-26
3.E-20
5.E-16
4.E-22
5.E-16
So
1 - Vadose Zone
VadoseSoil E3
g/g dry we
ght
8.E-27
2.E-26
3.E-20
7.E-16
5.E-22
7.E-16
So
1 - Vadose Zone
VadoseSoil E4
g/g dry we
ght
9.E-27
2.E-26
3.E-20
7.E-16
6.E-22
7.E-16
So
1 - Vadose Zone
VadoseSoil E5
g/g dry we
ght
2.E-26
5.E-26
9.E-20
2.E-15
2.E-21
2.E-15
So
1 - Vadose Zone
VadoseSoil E6
g/g dry we
ght
2.E-26
5.E-26
8.E-20
2.E-15
1.E-21
2.E-15
So
1 - Vadose Zone
VadoseSoil WFT
g/g dry we
ght
4.E-27
1.E-26
3.E-20
2.E-13
6.E-22
2.E-13
So
1 - Vadose Zone
VadoseSoil WFU
g/g dry we
ght
2.E-26
6.E-26
2.E-19
3.E-15
3.E-21
3.E-15
So
1 - Vadose Zone
VadoseSoil EFT
g/g dry we
ght
4.E-27
1.E-26
3.E-20
2.E-13
6.E-22
2.E-13
So
1 - Vadose Zone
VadoseSoil EFU
g/g dry we
ght
3.E-26
8.E-26
2.E-19
5.E-15
4.E-21
5.E-15
Groundwater
GW source
g/L
1.E-32
3.E-32
6.E-24
2.E-19
1.E-25
2.E-19
Groundwater
GW W1
g/L
4.E-34
1.E-33
4.E-26
1.E-21
7.E-28
1.E-21
Groundwater
GW W2
g/L
3.E-34
8.E-34
4.E-26
1.E-21
7.E-28
1.E-21
Groundwater
GW W3
g/L
7.E-35
2.E-34
8.E-27
2.E-22
1.E-28
2.E-22
1-2-6
-------
Exhibit 1-1. Annually Averaged Concentrations at Year 50 for All Compartments in the Ravena Site-Specific TRIM.FaTE
Scenario, Including Modeled Fish Harvesting from the Ravena Pond
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Groundwater
GW W4
g/L
1.E-34
3.E-34
1.E-26
3.E-22
2.E-28
3.E-22
Groundwater
GW W5
g/L
2.E-34
5.E-34
2.E-26
5.E-22
3.E-28
5.E-22
Groundwater
GW W6
g/L
5.E-34
1.E-33
1.E-25
3.E-21
2.E-27
3.E-21
Groundwater
GW W7
g/L
4.E-34
9.E-34
8.E-26
2.E-21
1.E-27
2.E-21
Groundwater
GW W8
g/L
2.E-34
6.E-34
4.E-26
9.E-22
7.E-28
9.E-22
Groundwater
GW E1
g/L
2.E-34
4.E-34
2.E-26
4.E-22
3.E-28
4.E-22
Groundwater
GW E2
g/L
8.E-35
2.E-34
8.E-27
2.E-22
1.E-28
2.E-22
Groundwater
GW E3
g/L
8.E-35
2.E-34
1.E-26
2.E-22
2.E-28
2.E-22
Groundwater
GW E4
g/L
9.E-35
2.E-34
1.E-26
3.E-22
2.E-28
3.E-22
Groundwater
GW E5
g/L
2.E-34
5.E-34
3.E-26
7.E-22
5.E-28
7.E-22
Groundwater
GW E6
g/L
2.E-34
5.E-34
3.E-26
6.E-22
4.E-28
6.E-22
Groundwater
GW WFT
g/L
4.E-35
1.E-34
2.E-24
5.E-20
3.E-26
5.E-20
Groundwater
GW WFU
g/L
2.E-34
6.E-34
5.E-26
1.E-21
8.E-28
1.E-21
Groundwater
GW EFT
g/L
4.E-35
1.E-34
2.E-24
5.E-20
3.E-26
5.E-20
Groundwater
GW EFU
g/L
3.E-34
8.E-34
7.E-26
2.E-21
1.E-27
2.E-21
a Annually averaged leaf concentrations are unavailable because of the seasonally changing leaf compartments.
1-2-7
-------
Exhibit 1-2. Annually Averaged Concentrations at Year 50 for All Compartments in the TRIM.FaTE Screening Scenario
using Ravena Emission Rates3
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Air
Air source
|ig/m3
8.E-08
2.E-07
3.E-03
1.E-02
2.E-09
1 .E-02
Air
Air N1
|ig/m3
2.E-08
6.E-08
9.E-04
3.E-03
2.E-11
4.E-03
Air
Air N2
|ig/m3
2.E-08
4.E-08
6.E-04
2.E-03
2.E-11
3.E-03
Air
Air N3
|ig/m3
1.E-08
3.E-08
4.E-04
1.E-03
2.E-11
2.E-03
Air
Air N4
|ig/m3
5.E-09
1.E-08
1.E-04
6.E-04
7.E-12
7.E-04
Air
Air N5
|ig/m3
3.E-09
6.E-09
6.E-05
3.E-04
4.E-12
4.E-04
Air
Air S1
|ig/m3
2.E-08
6.E-08
9.E-04
3.E-03
5.E-09
4.E-03
Air
Air S2
|ig/m3
2.E-08
4.E-08
6.E-04
2.E-03
6.E-09
3.E-03
Air
Air S3
|ig/m3
1.E-08
3.E-08
4.E-04
1.E-03
6.E-09
2.E-03
Air
Air S4
|ig/m3
5.E-09
1.E-08
1.E-04
6.E-04
1.E-09
7.E-04
Air
Air S5
|ig/m3
3.E-09
6.E-09
6.E-05
3.E-04
7.E-10
4.E-04
Surface water
SW_pond
mg/L
3.E-12
7.E-12
6.E-05
6.E-06
5.E-07
6.E-05
Water Column Carnivore
SW_pond
mg/kg wet weight
1.E-06
3.E-06
4.E-03
2.E-21
8.E-02
9.E-02
Water Column Herbivore
SW_pond
mg/kg wet weight
1.E-08
3.E-08
2.E-02
0.E+00
1 .E-02
3.E-02
Water Column Omnivore
SW_pond
mg/kg wet weight
8.E-08
2.E-07
7.E-03
5.E-13
2.E-02
2.E-02
Macrophyte
SW_pond
mg/kg wet weight
2.E-08
4.E-08
1.E-05
8.E-16
3.E-07
1.E-05
Sediment
Sed_pond
|ig/g dry weight
3.E-08
8.E-08
9.E-01
6.E-03
2.E-03
9.E-01
Benthic Invertebrate
Sed_pond
mg/kg wet weight
3.E-09
8.E-09
5.E-02
3.E-04
7.E-03
6.E-02
Benthic Omnivore
Sed_pond
mg/kg wet weight
5.E-08
1.E-07
1.E-02
2.E-12
2.E-02
3.E-02
Benthic Carnivore
Sed_pond
mg/kg wet weight
4.E-07
9.E-07
1.E-02
8.E-13
5.E-02
7.E-02
Soil - Surface
SurfSoil source
|ig/g dry weight
1.E-06
3.E-06
3.E+01
5.E-03
5.E-01
3.E+01
Soil - Surface
SurfSoil N1
|ig/g dry weight
2.E-07
6.E-07
2.E+00
4.E-04
3.E-02
2.E+00
Leaf - Grasses/Herbs D
SurfSoil N1
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSoil N1
mg/kg wet weight
1.E-04
3.E-04
3.E+01
3.E-06
2.E-07
3.E+01
Root - Grasses/Herbs
SurfSoil N1
mg/kg wet weight
2.E-12
6.E-12
1.E-04
0.E+00
1.E-05
1.E-04
Stem - Grasses/Herbs
SurfSoil N1
mg/kg wet weight
3.E-08
7.E-08
3.E-02
2.E-09
4.E-10
3.E-02
Soil - Surface
SurfSoil N6
|ig/g dry weight
1.E-08
3.E-08
3.E-01
1.E-02
5.E-03
3.E-01
Soil - Surface
SurfSoil N7
|ig/g dry weight
2.E-07
4.E-07
1.E+00
2.E-04
2.E-02
1.E+00
Leaf - Grasses/HerbsD
SurfSoil N7
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSoil N7
mg/kg wet weight
8.E-05
2.E-04
2.E+01
2.E-06
2.E-07
2.E+01
Root - Grasses/Herbs
SurfSoil N7
mg/kg wet weight
2.E-12
4.E-12
8.E-05
0.E+00
9.E-06
9.E-05
1-2-8
-------
Exhibit 1-2. Annually Averaged Concentrations at Year 50 for All Compartments in the TRIM.FaTE Screening Scenario
using Ravena Emission Rates3
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Stem - Grasses/Herbs
SurfSoil N7
mg/kg wet weight
2.E-08
5.E-08
2.E-02
1.E-09
3.E-10
2.E-02
Soil - Surface
SurfSoil N3
ng/g dry weight
1.E-07
3.E-07
9.E-01
2.E-04
2.E-02
1.E+00
Leaf - Grasses/Herbs D
SurfSoil N3
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSoil N3
mg/kg wet weight
5.E-05
1 .E-04
1.E+01
1.E-06
1.E-07
1.E+01
Root - Grasses/Herbs
SurfSoil N3
mg/kg wet weight
1.E-12
3.E-12
6.E-05
0.E+00
6.E-06
6.E-05
Stem - Grasses/Herbs
SurfSoil N3
mg/kg wet weight
1.E-08
3.E-08
1.E-02
8.E-10
2.E-10
1 .E-02
Soil - Surface
SurfSoil N4
ng/g dry weight
7.E-08
2.E-07
6.E-01
1 .E-04
1 .E-02
6.E-01
Leaf - Coniferous Forest
SurfSoil N4
mg/kg wet weight
4.E-08
1.E-07
6.E-02
1.E-08
2.E-12
6.E-02
Leaf Particle - Coniferous
Forest
SurfSoil N4
mg/kg wet weight
4.E-05
9.E-05
1.E+01
2.E-06
3.E-10
1.E+01
Soil - Surface
SurfSoil N5
ng/g dry weight
3.E-08
9.E-08
2.E-01
4.E-05
4.E-03
2.E-01
Leaf - Coniferous Forest
SurfSoil N5
mg/kg wet weight
2.E-08
6.E-08
3.E-02
5.E-09
9.E-13
3.E-02
Leaf Particle - Coniferous
Forest
SurfSoil N5
mg/kg wet weight
2.E-05
5.E-05
4.E+00
9.E-07
1.E-10
4.E+00
Soil - Surface
SurfSoil S1
ng/g dry weight
2.E-07
6.E-07
2.E+00
4.E-04
3.E-02
2.E+00
Leaf - Grasses/Herbs
SurfSoil S1
mg/kg wet weight
-
-
-
-
-
-
Leaf Particle - Grasses/Herbs
SurfSoil S1
mg/kg wet weight
1.E-04
3.E-04
3.E+01
3.E-06
3.E-07
3.E+01
Root - Grasses/Herbs
SurfSoil S1
mg/kg wet weight
2.E-12
6.E-12
1.E-04
0.E+00
1.E-05
1 .E-04
Stem - Grasses/Herbs
SurfSoil S1
mg/kg wet weight
3.E-08
7.E-08
3.E-02
2.E-09
4.E-10
3.E-02
Soil - Surface
SurfSoil S4
ng/g dry weight
1.E-07
2.E-07
1.E+00
2.E-04
2.E-02
1.E+00
Leaf - Coniferous Forest
SurfSoil S4
mg/kg wet weight
4.E-08
1.E-07
6.E-02
1.E-08
3.E-10
6.E-02
Leaf Particle - Coniferous
Forest
SurfSoil S4
mg/kg wet weight
4.E-05
9.E-05
1.E+01
2.E-06
5.E-08
1.E+01
Soil - Surface
SurfSoil S5
ng/g dry weight
4.E-08
1.E-07
3.E-01
6.E-05
5.E-03
3.E-01
Leaf - Coniferous Forest
SurfSoil S5
mg/kg wet weight
2.E-08
6.E-08
3.E-02
5.E-09
2.E-10
3.E-02
Leaf Particle - Coniferous
Forest
SurfSoil S5
mg/kg wet weight
2.E-05
5.E-05
4.E+00
9.E-07
3.E-08
4.E+00
Soil - Root Zone
RootSoil source
ng/g dry weight
2.E-10
4.E-10
3.E-03
3.E-03
6.E-05
7.E-03
Soil - Root Zone
RootSoil N1
ng/g dry weight
4.E-11
9.E-11
3.E-04
3.E-04
6.E-06
6.E-04
Soil - Root Zone
RootSoil N6
ng/g dry weight
5.E-13
1.E-12
1.E-05
2.E-03
2.E-07
2.E-03
Soil - Root Zone
RootSoil N7
ng/g dry weight
3.E-11
6.E-11
2.E-04
2.E-04
4.E-06
4.E-04
Soil - Root Zone
RootSoil N3
ng/g dry weight
2.E-11
5.E-11
2.E-04
1.E-04
3.E-06
3.E-04
1-2-9
-------
Exhibit 1-2. Annually Averaged Concentrations at Year 50 for All Compartments in the TRIM.FaTE Screening Scenario
using Ravena Emission Rates3
Compartment
Volume Element
Units
Mean
TCDD
UCL
TCDD
Divalent
Mercury
Elemental
Mercury
Methyl
Mercury
Total
Mercury
Soil - Root Zone
RootSoil N4
|ig/g dry weight
1.E-11
3.E-11
9.E-05
8.E-05
2.E-06
2.E-04
Soil - Root Zone
RootSoil N5
|ig/g dry weight
5.E-12
1.E-11
4.E-05
3.E-05
6.E-07
7.E-05
Soil - Root Zone
RootSoil S1
|ig/g dry weight
4.E-11
9.E-11
3.E-04
3.E-04
6.E-06
6.E-04
Soil - Root Zone
RootSoil S4
|ig/g dry weight
1.E-11
4.E-11
1.E-04
1.E-04
2.E-06
2.E-04
Soil - Root Zone
RootSoil S5
|ig/g dry weight
6.E-12
2.E-11
5.E-05
4.E-05
8.E-07
9.E-05
Soil - Vadose Zone
VadoseSoil source
g/g dry weight
2.E-24
5.E-24
8.E-17
4.E-13
1.E-18
4.E-13
Soil - Vadose Zone
VadoseSoil N1
g/g dry weight
5.E-25
1 .E-24
1.E-17
6.E-14
2.E-19
6.E-14
Soil - Vadose Zone
VadoseSoil N6
g/g dry weight
5.E-26
1 .E-25
2.E-18
2.E-12
4.E-20
2.E-12
Soil - Vadose Zone
VadoseSoil N7
g/g dry weight
4.E-25
9.E-25
7.E-18
4.E-14
1.E-19
4.E-14
Soil - Vadose Zone
VadoseSoil N3
g/g dry weight
3.E-25
6.E-25
5.E-18
3.E-14
9.E-20
3.E-14
Soil - Vadose Zone
VadoseSoil N4
g/g dry weight
1.E-25
4.E-25
3.E-18
1.E-14
4.E-20
1.E-14
Soil - Vadose Zone
VadoseSoil N5
g/g dry weight
7.E-26
2.E-25
1.E-18
6.E-15
2.E-20
6.E-15
Soil - Vadose Zone
VadoseSoil S1
g/g dry weight
5.E-25
1 .E-24
1.E-17
6.E-14
2.E-19
6.E-14
Soil - Vadose Zone
VadoseSoil S4
g/g dry weight
2.E-25
5.E-25
3.E-18
2.E-14
6.E-20
2.E-14
Soil - Vadose Zone
VadoseSoil S5
g/g dry weight
8.E-26
2.E-25
1.E-18
7.E-15
2.E-20
7.E-15
Groundwater
GW source
g/L
4.E-32
1.E-31
1.E-23
3.E-19
2.E-25
3.E-19
Groundwater
GW N1
g/L
9.E-33
2.E-32
2. E-24
4.E-20
3.E-26
4.E-20
Groundwater
GW N6
g/L
1.E-33
2.E-33
4.E-23
1.E-18
7.E-25
1.E-18
Groundwater
GW N7
g/L
7.E-33
2.E-32
1 .E-24
3.E-20
2.E-26
3.E-20
Groundwater
GW N3
g/L
5.E-33
1.E-32
9.E-25
2.E-20
1 .E-26
2.E-20
Groundwater
GW N4
g/L
3.E-33
7.E-33
4.E-25
9.E-21
7.E-27
9.E-21
Groundwater
GW N5
g/L
1.E-33
3.E-33
2.E-25
4.E-21
3.E-27
4.E-21
Groundwater
GW S1
g/L
9.E-33
2.E-32
2. E-24
4.E-20
3. E-26
4.E-20
Groundwater
GW S4
g/L
4.E-33
9.E-33
5.E-25
1.E-20
8.E-27
1.E-20
Groundwater
GW S5
g/L
2.E-33
4.E-33
2.E-25
4.E-21
3.E-27
4.E-21
a For more information of the TRIM.FaTE screening scenario, refer to Appendix C.
b Annually averaged leaf concentrations are unavailable because of the seasonally changing leaf compartments.
1-2-10
-------
1-2.2 Detailed 2,3,7,8 - TCDD Cancer Risk and Hazard Quotient
Results
This section provides tables showing detailed cancer risk and hazard quotient modeling
estimates for 2,3,7,8 - TCDD for all the different ingestion scenarios (combinations of the
selected soil compartment, water body compartment, ingestion rate, and emission rate)
considered. Exhibit 2-1 through Exhibit 2-4 provide the estimated hazard quotients and
individual lifetime cancer risk estimates using the combinations of 95th percentile (upper
confidence limit, or UCL) and mean emission rates as well as 90th percentile (reasonable
maximum exposure, or RME) and mean (central tendency exposure, or CTE) ingestion rates.
Exhibit 2-5 gives detailed individual lifetime cancer risk and age-specific hazard quotient
estimates broken down by different ingestion pathways. Exhibit 2-6 and Exhibit 2-7 provide
comparisons and percent changes in individual lifetime cancer risk when using either UCL or
mean emission rates and using either RME or CTE ingestion rates, respectively. Finally, Exhibit
2-8 provides dermal hazard quotients and risk estimates due to exposure to water in Alcove
Reservoir for all emission rates and age groups.
1-2-11
-------
Exhibit 2-1. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using UCL Emission Rate and
RME Ingestion Rates
Note: HQs greater than 1 and risks greater than 1E-06 are in boldface type.
Scenario Type
Water body
Farm Parcel
Harvesting from
Ravena Pond
C
Child
(1-2)
hronic Non
Child
(3-5)
-Cancer Haz
Child
(6-11)
.ard Quotier
Child
(12-19)
it
Adult
(20-70)
Individual
Lifetime
Cancer
Risk
Screening
Screening
Screening
n/a
1.456
1.086
0.892
0.480
0.739
1.1E-04
Combined
Ravena Pond
West
Harvesting
0.971
0.950
0.800
0.508
0.863
1.2E-04
No Harvesting
1.314
1.293
1.089
0.694
1.180
1.7E-04
East
Harvesting
0.967
0.948
0.798
0.507
0.862
1.2E-04
No Harvesting
1.310
1.291
1.087
0.693
1.179
1.7E-04
Nassau Lake
West
n/a
0.084
0.062
0.051
0.027
0.042
6.4E-06
East
n/a
0.080
0.060
0.049
0.026
0.041
6.2E-06
Kinderhook Lake
West
n/a
0.066
0.044
0.036
0.017
0.025
4.0E-06
East
n/a
0.062
0.042
0.034
0.017
0.024
3.8E-06
Alcove
Reservoir
West
n/a
0.060
0.038
0.030
0.014
0.019
3.2E-06
East
n/a
0.056
0.035
0.029
0.013
0.018
3.0E-06
Farmer Only
-
West
n/a
0.058
0.035
0.029
0.013
0.017
2.9E-06
-
East
n/a
0.053
0.033
0.027
0.012
0.016
2.7E-06
Angler Only
Pond
-
Harvesting
0.914
0.915
0.771
0.495
0.846
1.2E-04
-
No Harvesting
1.256
0.681
1.258
1.060
1.163
1.6E-04
Nassau Lake
-
n/a
0.026
0.027
0.022
0.014
0.025
3.4E-06
Kinderhook Lake
-
n/a
0.009
0.009
0.007
0.005
0.008
1.1E-06
Alcove
Reservoir
-
n/a
0.002
0.002
0.002
0.001
0.002
2.8E-07
Water Ingestion Only
-
-
n/a
0.000
0.000
0.000
0.000
0.000
1.3E-13
1-2-12
-------
Exhibit 2-2. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using UCL Emission Rate and
CTE Ingestion Rates
Note: HQs greater than 1 and risks greater than 1E-06 are in boldface type.
Scenario Type
Water body
Farm Parcel
Harvesting from
Ravena Pond
C
Child
(1-2)
hronic Non
Child
(3-5)
-Cancer Haz
Child
(6-11)
.ard Quotier
Child
(12-19)
it
Adult
(20-70)
Individual
Lifetime
Cancer
Risk
Screening
Screening
Screening
n/a
0.632
0.463
0.362
0.206
0.323
4.8E-05
Combined
Ravena Pond
West
Harvesting
0.412
0.403
0.315
0.221
0.352
5.0E-05
No Harvesting
0.557
0.549
0.428
0.302
0.480
6.8E-05
East
Harvesting
0.410
0.402
0.314
0.221
0.351
5.0E-05
No Harvesting
0.555
0.548
0.428
0.302
0.480
6.8E-05
Nassau Lake
West
n/a
0.037
0.027
0.021
0.012
0.018
2.8E-06
East
n/a
0.035
0.026
0.020
0.011
0.018
2.7E-06
Kinderhook Lake
West
n/a
0.029
0.019
0.015
0.007
0.012
1.8E-06
East
n/a
0.027
0.018
0.014
0.007
0.011
1.7E-06
Alcove
Reservoir
West
n/a
0.027
0.016
0.013
0.006
0.009
1.5E-06
East
n/a
0.025
0.015
0.012
0.006
0.009
1.4E-06
Farmer Only
-
West
n/a
0.026
0.015
0.012
0.005
0.008
1.4E-06
-
East
n/a
0.024
0.014
0.011
0.005
0.008
1.3E-06
Angler Only
Pond
-
Harvesting
0.386
0.388
0.303
0.216
0.343
4.9E-05
-
No Harvesting
0.531
0.297
0.533
0.416
0.472
6.7E-05
Nassau Lake
-
n/a
0.011
0.011
0.009
0.006
0.010
1.4E-06
Kinderhook Lake
-
n/a
0.004
0.004
0.003
0.002
0.003
4.5E-07
Alcove
Reservoir
-
n/a
0.001
0.001
0.001
0.001
0.001
1.2E-07
Water Ingestion Only
-
-
n/a
0.000
0.000
0.000
0.000
0.000
6.2E-14
1-2-13
-------
Exhibit 2-3. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using Mean Emission Rate and
RME Ingestion Rates
Note: HQs greater than 1 and risks greater than 1E-06 are highlighted in boldface type.
Scenario Type
Water body
Farm Parcel
Harvesting from
Ravena Pond
C
Child
(1-2)
hronic Non
Child
(3-5)
-Cancer Haz
Child
(6-11)
.ard Quotier
Child
(12-19)
it
Adult
(20-70)
Individual
Lifetime
Cancer
Risk
Screening
Screening
Screening
n/a
0.596
0.444
0.365
0.196
0.302
4.6E-05
Combined
Ravena Pond
West
Harvesting
0.398
0.389
0.327
0.208
0.353
5.0E-05
No Harvesting
0.538
0.529
0.445
0.284
0.483
6.8E-05
East
Harvesting
0.396
0.388
0.327
0.208
0.353
5.0E-05
No Harvesting
0.536
0.528
0.445
0.284
0.483
6.8E-05
Nassau Lake
West
n/a
0.034
0.025
0.021
0.011
0.017
2.6E-06
East
n/a
0.033
0.024
0.020
0.011
0.017
2.5E-06
Kinderhook Lake
West
n/a
0.027
0.018
0.015
0.007
0.010
1.6E-06
East
n/a
0.025
0.017
0.014
0.007
0.010
1.6E-06
Alcove
Reservoir
West
n/a
0.024
0.015
0.012
0.006
0.008
1.3E-06
East
n/a
0.023
0.014
0.012
0.005
0.007
1.2E-06
Farmer Only
-
West
n/a
0.024
0.014
0.012
0.005
0.007
1.2E-06
-
East
n/a
0.022
0.013
0.011
0.005
0.007
1.1E-06
Angler Only
Pond
-
Harvesting
0.374
0.374
0.316
0.203
0.346
4.9E-05
-
No Harvesting
0.514
0.279
0.515
0.434
0.476
6.7E-05
Nassau Lake
-
n/a
0.011
0.011
0.009
0.006
0.010
1.4E-06
Kinderhook Lake
-
n/a
0.003
0.004
0.003
0.002
0.003
4.6E-07
Alcove
Reservoir
-
n/a
0.001
0.001
0.001
0.000
0.001
1.2E-07
Water Ingestion Only
-
-
n/a
0.000
0.000
0.000
0.000
0.000
5.2E-14
1-2-14
-------
Exhibit 2-4. Estimated Hazard Quotients and Individual Lifetime Cancer Risks for 2,3,7,8-TCDD Using Mean Emission Rate and
CTE Ingestion Rates
Note: HQs greater than 1 and risks greater than 1E-06 are highlighted in boldface type.
Scenario Type
Water body
Farm Parcel
Harvesting from
Ravena Pond
C
Child
(1-2)
hronic Non
Child
(3-5)
-Cancer Haz
Child
(6-11)
.ard Quotier
Child
(12-19)
it
Adult
(20-70)
Individual
Lifetime
Cancer
Risk
Screening
Screening
Screening
n/a
0.258
0.189
0.148
0.084
0.132
2.0E-05
Combined
Ravena Pond
West
Harvesting
0.169
0.165
0.129
0.091
0.144
2.0E-05
No Harvesting
0.228
0.224
0.175
0.124
0.197
2.8E-05
East
Harvesting
0.168
0.165
0.129
0.090
0.144
2.0E-05
No Harvesting
0.227
0.224
0.175
0.124
0.196
2.8E-05
Nassau Lake
West
n/a
0.015
0.011
0.008
0.005
0.008
1.1E-06
East
n/a
0.014
0.010
0.008
0.005
0.007
1.1E-06
Kinderhook Lake
West
n/a
0.012
0.008
0.006
0.003
0.005
7.4E-07
East
n/a
0.011
0.007
0.006
0.003
0.005
7.1E-07
Alcove
Reservoir
West
n/a
0.011
0.007
0.005
0.002
0.004
6.0E-07
East
n/a
0.010
0.006
0.005
0.002
0.004
5.7E-07
Farmer Only
-
West
n/a
0.011
0.006
0.005
0.002
0.003
5.5E-07
-
East
n/a
0.010
0.006
0.005
0.002
0.003
5.2E-07
Angler Only
Pond
-
Harvesting
0.158
0.159
0.124
0.088
0.141
2.0E-05
-
No Harvesting
0.217
0.121
0.218
0.170
0.193
2.7E-05
Nassau Lake
-
n/a
0.005
0.005
0.004
0.003
0.004
5.8E-07
Kinderhook Lake
-
n/a
0.001
0.001
0.001
0.001
0.001
1.9E-07
Alcove
Reservoir
-
n/a
0.000
0.000
0.000
0.000
0.000
4.7E-08
Water Ingestion Only
-
-
n/a
0.000
0.000
0.000
0.000
0.000
2.5E-14
1-2-15
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
UCL Emission Rate, CTE Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
0.7%
0.9%
1.1%
1.4%
1.6%
1.4%
Egg, Pork, & Poultry
2.0%
3.0%
2.7%
4.5%
3.2%
3.2%
Beef & Dairy
93.6%
90.0%
90.2%
85.0%
85.8%
87.1%
Fish
3.7%
6.0%
6.0%
9.1%
9.4%
8.3%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
0.7%
1.0%
1.1%
1.5%
1.7%
1.5%
Egg, Pork, & Poultry
1.4%
2.2%
2.0%
3.3%
2.4%
2.3%
Beef & Dairy
94.3%
91.2%
91.2%
86.5%
87.0%
88.3%
Fish
3.4%
5.7%
5.7%
8.6%
8.9%
7.8%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE ngestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
0.6%
0.8%
0.9%
1.1%
1.2%
1.1%
Egg, Pork, & Poultry
1.8%
2.6%
2.3%
3.6%
2.5%
2.5%
Beef & Dairy
84.3%
76.4%
76.6%
67.0%
67.2%
70.0%
Fish
13.2%
20.2%
20.2%
28.3%
29.0%
26.4%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
0.7%
0.8%
1.0%
1.2%
1.3%
1.2%
Egg, Pork, & Poultry
1.3%
1.9%
1.7%
2.6%
1.9%
1.9%
Beef & Dairy
85.7%
78.2%
78.2%
69.0%
69.0%
71.8%
Fish
12.3%
19.1%
19.1%
27.1%
27.7%
25.1%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
0.5%
0.5%
0.6%
0.7%
0.8%
0.7%
Egg, Pork, & Poultry
1.4%
1.8%
1.6%
2.2%
1.6%
1.6%
Beef & Dairy
66.1%
53.7%
53.9%
42.0%
41.8%
45.1%
Fish
32.0%
44.0%
43.9%
55.0%
55.8%
52.6%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
0.5%
0.6%
0.7%
0.8%
0.9%
0.8%
Egg, Pork, & Poultry
1.0%
1.3%
1.2%
1.7%
1.2%
1.2%
Beef & Dairy
68.0%
55.8%
55.8%
44.0%
43.7%
47.0%
Fish
30.4%
42.2%
42.3%
53.6%
54.3%
51.0%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-16
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
UCL Emission Rate, CTE Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.6%
3.4%
3.4%
2.2%
2.1%
2.4%
Fish
94.2%
96.4%
96.4%
97.7%
97.8%
97.5%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
6.1%
3.7%
3.7%
2.3%
2.3%
2.6%
Fish
93.8%
96.2%
96.2%
97.5%
97.6%
97.3%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, CTE Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
0.8%
0.9%
1.2%
1.4%
1.8%
1.5%
Egg, Pork, & Poultry
1.8%
2.7%
2.3%
4.3%
3.6%
3.3%
Beef & Dairy
93.3%
90.1%
90.1%
85.4%
83.5%
85.7%
Fish
3.9%
6.2%
6.4%
8.9%
11.0%
9.4%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
0.9%
0.9%
1.3%
1.5%
1.9%
1.7%
Egg, Pork, & Poultry
1.3%
2.0%
1.7%
3.2%
2.7%
2.4%
Beef & Dairy
94.1%
91.2%
91.0%
86.9%
84.9%
87.0%
Fish
3.6%
5.8%
6.0%
8.4%
10.4%
8.8%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
0.8%
0.8%
1.0%
1.1%
1.3%
1.2%
Egg, Pork, & Poultry
1.6%
2.3%
1.9%
3.4%
2.7%
2.6%
Beef & Dairy
83.7%
76.2%
75.9%
67.7%
63.1%
67.2%
Fish
13.8%
20.6%
21.1%
27.7%
32.8%
28.9%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
0.8%
0.8%
1.1%
1.2%
1.5%
1.3%
Egg, Pork, & Poultry
1.2%
1.7%
1.4%
2.5%
2.0%
1.9%
Beef & Dairy
85.0%
77.9%
77.3%
69.7%
65.0%
69.1%
Fish
12.9%
19.5%
20.1%
26.5%
31.5%
27.6%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-17
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
UCL Emission Rate, RME Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
0.6%
0.5%
0.7%
0.7%
0.8%
0.8%
Egg, Pork, & Poultry
1.3%
1.6%
1.3%
2.2%
1.6%
1.6%
Beef & Dairy
64.9%
53.2%
52.6%
42.8%
37.4%
41.8%
Fish
33.1%
44.6%
45.4%
54.3%
60.2%
55.8%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
0.6%
0.6%
0.8%
0.8%
0.9%
0.8%
Egg, Pork, & Poultry
0.9%
1.2%
1.0%
1.6%
1.2%
1.2%
Beef & Dairy
66.9%
55.3%
54.4%
44.8%
39.1%
43.7%
Fish
31.5%
42.9%
43.8%
52.8%
58.7%
54.2%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.4%
3.3%
3.2%
2.2%
1.8%
2.1%
Fish
94.5%
96.5%
96.6%
97.6%
98.1%
97.8%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
UCL Emission Rate, RME Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.8%
3.6%
3.5%
2.4%
1.9%
2.3%
Fish
94.1%
96.3%
96.4%
97.5%
98.0%
97.6%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
0.7%
0.9%
1.1%
1.4%
1.6%
1.4%
Egg, Pork, & Poultry
2.0%
3.0%
2.7%
4.5%
3.2%
3.2%
Beef & Dairy
93.6%
90.0%
90.2%
85.0%
85.8%
87.1%
Fish
3.7%
6.0%
6.0%
9.1%
9.4%
8.3%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
0.7%
1.0%
1.1%
1.5%
1.7%
1.5%
Egg, Pork, & Poultry
1.4%
2.2%
2.0%
3.3%
2.4%
2.3%
Beef & Dairy
94.3%
91.2%
91.2%
86.5%
87.0%
88.3%
Fish
3.4%
5.7%
5.7%
8.6%
8.9%
7.8%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-18
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
Mean Emission Rate, CTE Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
0.6%
0.8%
0.9%
1.1%
1.2%
1.1%
Egg, Pork, & Poultry
1.8%
2.6%
2.3%
3.6%
2.5%
2.5%
Beef & Dairy
84.3%
76.4%
76.6%
67.0%
67.2%
70.0%
Fish
13.2%
20.2%
20.2%
28.3%
29.0%
26.4%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
0.7%
0.8%
1.0%
1.2%
1.3%
1.2%
Egg, Pork, & Poultry
1.3%
1.9%
1.7%
2.6%
1.9%
1.9%
Beef & Dairy
85.7%
78.2%
78.2%
69.0%
69.0%
71.8%
Fish
12.3%
19.1%
19.1%
27.1%
27.7%
25.1%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
0.5%
0.5%
0.6%
0.7%
0.8%
0.7%
Egg, Pork, & Poultry
1.4%
1.8%
1.6%
2.2%
1.6%
1.6%
Beef & Dairy
66.1%
53.7%
53.9%
42.0%
41.8%
45.1%
Fish
32.0%
44.0%
43.9%
55.0%
55.8%
52.6%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
0.5%
0.6%
0.7%
0.8%
0.9%
0.8%
Egg, Pork, & Poultry
1.0%
1.3%
1.2%
1.7%
1.2%
1.2%
Beef & Dairy
68.0%
55.8%
55.8%
44.0%
43.7%
47.0%
Fish
30.4%
42.2%
42.3%
53.6%
54.3%
51.0%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.6%
3.4%
3.4%
2.2%
2.1%
2.4%
Fish
94.2%
96.4%
96.4%
97.7%
97.8%
97.5%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, CTE Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
6.1%
3.7%
3.7%
2.3%
2.3%
2.6%
Fish
93.8%
96.2%
96.2%
97.5%
97.6%
97.3%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-19
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
Mean Emission Rate, RME Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
0.8%
0.9%
1.2%
1.4%
1.8%
1.5%
Egg, Pork, & Poultry
1.8%
2.7%
2.3%
4.3%
3.6%
3.3%
Beef & Dairy
93.3%
90.1%
90.1%
85.4%
83.5%
85.7%
Fish
3.9%
6.2%
6.4%
8.9%
11.0%
9.4%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
0.9%
0.9%
1.3%
1.5%
1.9%
1.7%
Egg, Pork, & Poultry
1.3%
2.0%
1.7%
3.2%
2.7%
2.4%
Beef & Dairy
94.1%
91.2%
91.0%
86.9%
84.9%
87.0%
Fish
3.6%
5.8%
6.0%
8.4%
10.4%
8.8%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
0.8%
0.8%
1.0%
1.1%
1.3%
1.2%
Egg, Pork, & Poultry
1.6%
2.3%
1.9%
3.4%
2.7%
2.6%
Beef & Dairy
83.7%
76.2%
75.9%
67.7%
63.1%
67.2%
Fish
13.8%
20.6%
21.1%
27.7%
32.8%
28.9%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
0.8%
0.8%
1.1%
1.2%
1.5%
1.3%
Egg, Pork, & Poultry
1.2%
1.7%
1.4%
2.5%
2.0%
1.9%
Beef & Dairy
85.0%
77.9%
77.3%
69.7%
65.0%
69.1%
Fish
12.9%
19.5%
20.1%
26.5%
31.5%
27.6%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
0.6%
0.5%
0.7%
0.7%
0.8%
0.8%
Egg, Pork, & Poultry
1.3%
1.6%
1.3%
2.2%
1.6%
1.6%
Beef & Dairy
64.9%
53.2%
52.6%
42.8%
37.4%
41.8%
Fish
33.1%
44.6%
45.4%
54.3%
60.2%
55.8%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
0.6%
0.6%
0.8%
0.8%
0.9%
0.8%
Egg, Pork, & Poultry
0.9%
1.2%
1.0%
1.6%
1.2%
1.2%
Beef & Dairy
66.9%
55.3%
54.4%
44.8%
39.1%
43.7%
Fish
31.5%
42.9%
43.8%
52.8%
58.7%
54.2%
Soil
0.1%
0.1%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-20
-------
Exhibit 2-5. Fractional Pathway of Cancer Risks and Age-Specific Hazard Quotients for
2,3,7,8-TCDD for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-
11
Child 12-
19
Adult 20-
70
Cancer
Mean Emission Rate, RME Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.4%
3.3%
3.2%
2.2%
1.8%
2.1%
Fish
94.5%
96.5%
96.6%
97.6%
98.1%
97.8%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Mean Emission Rate, RME Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
Egg, Pork, & Poultry
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
Beef & Dairy
5.8%
3.6%
3.5%
2.4%
1.9%
2.3%
Fish
94.1%
96.3%
96.4%
97.5%
98.0%
97.6%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-21
-------
Exhibit 2-6. Comparison of 2,3,7,8-TCDD Ravena Cancer Risks Using UCL Versus Mean
Emission Rates3
Scenario Type
Farm Parcel
Water body
Mean
Emission
Factor
UCL
Emission
Factor
Ratio of
UCL: Mean
EF
Screening
Screening
Screening
4.6E-05
1.1E-04
2.4
Combined
West
Alcove
1.3E-06
3.2E-06
2.4
Kinderhook
1.6E-06
4.0E-06
2.4
Nassau
2.6E-06
6.4E-06
2.4
Pond
5.0E-05
1.2E-04
2.4
East
Alcove
1.2E-06
3.0E-06
2.4
Kinderhook
1.6E-06
3.8E-06
2.4
Nassau
2.5E-06
6.2E-06
2.4
Pond
5.0E-05
1.2E-04
2.4
Farm Only
West
-
1.2E-06
2.9E-06
2.4
East
-
1.1E-06
2.7E-06
2.4
Fisherman Only
-
Alcove
1.2E-07
2.8E-07
2.4
Kinderhook
4.6E-07
1.1E-06
2.4
Nassau
1.4E-06
3.4E-06
2.4
Pond
4.9E-05
1.2E-04
2.4
Water Ingestion
Only
-
-
5.2E-14
1.3E-13
2.4
a Selected scenarios use RME percentile ingestion rates and contain a fish harvester in Ravena Pond
1-2-22
-------
Exhibit 2-7. Comparison of 2,3,7,8-TCDD Ravena Cancer Risks Using CTE Versus RME
Ingestion Rates3
Emission
Rate
Scenario Type
Farm
Parcel
Water body
CTE
Ingestion
Rates
RME
Ingestion
Rates
Ratio of
RME : CTE
Screening
Screening
Screening
4.8E-05
1.1E-04
2.3
Alcove
1.5E-06
3.2E-06
2.2
West
Kinderhook
1.8E-06
4.0E-06
2.2
Nassau
2.8E-06
6.4E-06
2.3
Combined
Pond
5.0E-05
1.2E-04
2.4
Alcove
1.4E-06
3.0E-06
2.2
East
Kinderhook
1.7E-06
3.8E-06
2.2
Nassau
2.7E-06
6.2E-06
2.3
UCL
Pond
5.0E-05
1.2E-04
2.4
Farm Only
West
-
1.4E-06
2.9E-06
2.2
East
-
1.3E-06
2.7E-06
2.2
Alcove
1.2E-07
2.8E-07
2.4
Fisherman Only
Kinderhook
4.5E-07
1.1E-06
2.4
Nassau
1.4E-06
3.4E-06
2.4
Pond
4.9E-05
1.2E-04
2.4
Water Ingestion
Only
-
-
6.2E-14
1.3E-13
2.1
Screening
Screening
Screening
2.0E-05
4.6E-05
2.3
Alcove
6.0E-07
1.3E-06
2.2
West
Kinderhook
7.4E-07
1.6E-06
2.2
Nassau
1.1E-06
2.6E-06
2.3
Combined
Pond
2.0E-05
5.0E-05
2.4
Alcove
5.7E-07
1.2E-06
2.2
East
Kinderhook
7.1E-07
1.6E-06
2.2
Nassau
1.1E-06
2.5E-06
2.3
Mean
Pond
2.0E-05
5.0E-05
2.4
Farm Only
West
-
5.5E-07
1.2E-06
2.2
East
-
5.2E-07
1.1E-06
2.2
Alcove
4.7E-08
1.2E-07
2.4
Fisherman Only
Kinderhook
1.9E-07
4.6E-07
2.4
Nassau
5.8E-07
1.4E-06
2.4
Pond
2.0E-05
4.9E-05
2.4
Water Ingestion
Only
-
-
2.5E-14
5.2E-14
2.1
a Selected scenarios use UCL emission rates and contain a fish harvester in Ravena Pond
1-2-23
-------
Exhibit 2-8. 2,3,7,8 TCDD Dermal Hazard Quotients and Lifetime Risks for Water and Soil Contact
Age-
UCL Emission Factor
Mean Emission Factor
Specific
Hazard
Quotient
or
Lifetime
Risk
Exposure
Media
East Farm
-Tilled,
Alcove
Reservoir
East
Farm -
Untitled,
Alcove
Reservoir
West
Farm -
Tilled,
Alcove
Reservoir
West
Farm -
Unfilled,
Alcove
Reservoir
Screen
East
Farm -
Tilled,
Alcove
Reservoir
East
Farm -
Untitled,
Alcove
Reservoir
West
Farm -
Tilled,
Alcove
Reservoir
West
Farm -
Untitled,
Alcove
Reservoir
Screen
Child < 1
HQ
Soil
2.58E-06
4.42E-05
2.56E-06
3.25E-05
7.82E-04
1.06E-06
1.81 E-05
1.05E-06
1.33E-05
3.20E-04
Water
3.92E-05
3.92E-05
3.92E-05
3.92E-05
9.93E-03
1.60E-05
1.60E-05
1.60E-05
1.60E-05
4.06E-03
Soil and
Water
4.18E-05
8.34E-05
4.17E-05
7.16E-05
1.07E-02
1.71 E-05
3.41 E-05
1.71 E-05
2.93E-05
4.38E-03
Child 1-2
HQ
Soil
2.18E-06
3.73E-05
2.16E-06
2.74E-05
6.60E-04
8.90E-07
1.52E-05
8.82E-07
1.12E-05
2.70E-04
Water
3.48E-05
3.48E-05
3.48E-05
3.48E-05
8.81 E-03
1.42E-05
1.42E-05
1.42E-05
1.42E-05
3.60E-03
Soil and
Water
3.69E-05
7.20E-05
3.69E-05
6.21 E-05
9.47E-03
1.51 E-05
2.95E-05
1.51 E-05
2.54E-05
3.87E-03
Child 3-5
HQ
Soil
2.06E-06
3.52E-05
2.04E-06
2.59E-05
6.23E-04
8.41 E-07
1.44E-05
8.34E-07
1.06E-05
2.55E-04
Water
3.14E-05
3.14E-05
3.14E-05
3.14E-05
7.96E-03
1.28E-05
1.28E-05
1.28E-05
1.28E-05
3.26E-03
Soil and
Water
3.35E-05
6.66E-05
3.34E-05
5.73E-05
8.58E-03
1.37E-05
2.73E-05
1.37E-05
2.34E-05
3.51 E-03
Child 6-
11 HQ
Soil
1.61E-06
2.76E-05
1.60E-06
2.03E-05
4.89E-04
6.60E-07
1.13E-05
6.54E-07
8.31 E-06
2.00E-04
Water
2.61 E-05
2.61 E-05
2.61 E-05
2.61 E-05
6.62E-03
1.07E-05
1.07E-05
1.07E-05
1.07E-05
2.71 E-03
Soil and
Water
2.77E-05
5.37E-05
2.77E-05
4.64E-05
7.11 E-03
1.13E-05
2.20E-05
1.13E-05
1.90E-05
2.91 E-03
Child 12-
19 HQ
Soil
1.23E-06
2.10E-05
1.22E-06
1.54E-05
3.72E-04
5.02E-07
8.59E-06
4.97E-07
6.31 E-06
1.52E-04
Water
2.05E-05
2.05E-05
2.05E-05
2.05E-05
5.20E-03
8.40E-06
8.40E-06
8.40E-06
8.40E-06
2.13E-03
Soil and
Water
2.18E-05
4.15E-05
2.17E-05
3.60E-05
5.58E-03
8.90E-06
1.70E-05
8.90E-06
1.47E-05
2.28E-03
Adult HQ
Soil
5.49E-07
9.40E-06
5.44E-07
6.91 E-06
1.66E-04
2.25E-07
3.85E-06
2.23E-07
2.83E-06
6.81 E-05
Water
1.49E-05
1.49E-05
1.49E-05
1.49E-05
3.77E-03
6.09E-06
6.09E-06
6.09E-06
6.09E-06
1.54E-03
Soil and
Water
1.54E-05
2.43E-05
1.54E-05
2.18E-05
3.94E-03
6.31 E-06
9.94E-06
6.31 E-06
8.91 E-06
1.61 E-03
Lifetime
Risk
Soil
1.29E-10
2.20E-09
1.28E-10
1.62E-09
3.90E-08
5.26E-11
9.02E-10
5.22E-11
6.62E-10
1.60E-08
Water
2.72E-09
2.72E-09
2.72E-09
2.72E-09
6.89E-07
1.11E-09
1.11E-09
1.11 E-09
1.11 E-09
2.82E-07
Soil and
Water
2.85E-09
4.92E-09
2.84E-09
4.34E-09
7.28E-07
1.16E-09
2.01 E-09
1.16E-09
1.77E-09
2.98E-07
1-2-24
-------
1-2.3 Detailed Mercury Hazard Quotient Results
This section provides tables showing detailed hazard quotient modeling estimates for divalent
mercury and methyl mercury for all the different ingestion scenarios (combinations of selected
soil compartment, water body compartment, and ingestion rate) considered. Exhibit 3-1 and
Exhibit 3-2 provide hazard quotient estimates using the 90th percentile (reasonable maximum
exposure, or RME) and mean (central tendency exposure, or CTE) ingestion rates for divalent
mercury, respectively. Exhibit 3-3 and Exhibit 3-4 provide hazard quotient estimates using the
RME and CTE ingestion rates for methyl mercury, respectively. Exhibit 3-5 and Exhibit 3-6
gives detailed age-specific hazard quotient estimates broken down by different ingestion
pathways for divalent and methyl mercury, respectively. Exhibit 3-7 highlights the differences in
hazard quotient estimates when using the Alcove Reservoir compared to all other water bodies.
Exhibit 3-8 compares the hazard quotient estimates generated using the east farm versus the
west farm TRIM.FaTE compartments, and Exhibit 3-9 provide comparisons and percent
changes in hazard quotient that arise from using either RME or CTE ingestion rates. Finally,
Exhibit 3-10 provides dermal hazard quotients due to exposure to water in Alcove reservoir for
all age groups.
Attachment 1-2
1-2-25
SAB Review Draft - 05/22/09
-------
Exhibit 3-1. Summary Results - Hazard Quotients for Diva
ent Mercury Scenarios using RM
E Ingestion Rates
Ingestion
Rates
Scenario Type
Water body
Farm
Parcel
Harvester in
Ravena Pond?
HQ
Child
(1-2)
HQ
Child
(3-5)
HQ
Child
(6-11)
HQ
Child
(12-19)
HQ
Adult
(20-70)
Screening
Screening
Screening
Harvester
0.344
0.233
0.120
0.072
0.087
West
Harvester
0.020
0.019
0.016
0.010
0.017
Ravena Pond
No Harvester
0.022
0.021
0.018
0.011
0.019
East
Harvester
0.021
0.020
0.016
0.010
0.017
No Harvester
0.023
0.022
0.018
0.012
0.019
Combined
Nassau Lake
West
Harvester
0.004
0.002
0.002
0.001
0.002
East
Harvester
0.004
0.003
0.002
0.001
0.002
Kinderhook Lake
West
Harvester
0.004
0.002
0.002
0.001
0.002
RME
East
Harvester
0.004
0.003
0.002
0.001
0.002
Ingestion
Alcove Reservoir
West
Harvester
0.004
0.002
0.002
0.001
0.002
Rate
East
Harvester
0.004
0.003
0.002
0.001
0.002
Farm Only
-
West
Harvester
0.003
0.002
0.002
0.001
0.002
-
East
Harvester
0.004
0.003
0.002
0.001
0.002
Pond
-
Harvester
0.017
0.017
0.014
0.009
0.016
-
No Harvester
0.019
0.010
0.019
0.016
0.018
Fisherman Only
Nassau Lake
-
Harvester
0.000
0.000
0.000
0.000
0.000
Kinderhook Lake
-
Harvester
0.000
0.000
0.000
0.000
0.000
Alcove Reservoir
-
Harvester
0.000
0.000
0.000
0.000
0.000
Water Ingestion Only
-
-
Harvester
0.000
0.000
0.000
0.000
0.000
1-2-26
-------
Exhibit 3-2. Summary Results - Hazard Quotients for Diva
Ingestion
Rates
Scenario Type
Water body
Farm
Parcel
Harvester in
Ravena Pond?
HQ
Child
(1-2)
HQ
Child
(3-5)
HQ
Child
(6-11)
HQ
Child
(12-19)
HQ
Adult
(20-70)
Screening
Screening
Screening
Harvester
0.085
0.057
0.041
0.026
0.033
West
Harvester
0.008
0.008
0.006
0.004
0.007
Ravena Pond
No Harvester
0.009
0.009
0.007
0.005
0.008
East
Harvester
0.009
0.008
0.006
0.005
0.007
No Harvester
0.010
0.009
0.007
0.005
0.008
Combined
Nassau Lake
West
Harvester
0.001
0.001
0.001
0.000
0.001
East
Harvester
0.002
0.001
0.001
0.001
0.001
Kinderhook Lake
West
Harvester
0.001
0.001
0.001
0.000
0.001
CTE
East
Harvester
0.002
0.001
0.001
0.001
0.001
Ingestion
Alcove Reservoir
West
Harvester
0.001
0.001
0.001
0.000
0.001
Rate
East
Harvester
0.002
0.001
0.001
0.001
0.001
Farm Only
-
West
Harvester
0.001
0.001
0.001
0.000
0.001
-
East
Harvester
0.002
0.001
0.001
0.001
0.001
Pond
-
Harvester
0.007
0.007
0.006
0.004
0.006
-
No Harvester
0.008
0.004
0.008
0.006
0.007
Fisherman Only
Nassau Lake
-
Harvester
0.000
0.000
0.000
0.000
0.000
Kinderhook Lake
-
Harvester
0.000
0.000
0.000
0.000
0.000
Alcove Reservoir
-
Harvester
0.000
0.000
0.000
0.000
0.000
Water Ingestion Only
-
-
Harvester
0.000
0.000
0.000
0.000
0.000
ent Mercury Scenarios using CTE Ingestion Rates
1-2-27
-------
Exhibit 3-3. Summary Results - Hazard Quotients for Met
hyl Mercury Scenarios using RM
E Ingestion Rates
Ingestion
Rates
Scenario Type
Water body
Farm
Parcel
Harvester in
Ravena Pond?
HQ
Child
(1-2)
HQ
Child
(3-5)
HQ
Child
(6-11)
HQ
Child
(12-19)
HQ
Adult
(20-70)
Screening
Screening
Screening
Harvester
0.193
0.188
0.155
0.099
0.167
West
Harvester
0.132
0.132
0.111
0.071
0.122
Ravena Pond
No Harvester
0.199
0.199
0.168
0.108
0.184
East
Harvester
0.132
0.132
0.111
0.071
0.122
No Harvester
0.199
0.199
0.168
0.108
0.184
Combined
Nassau Lake
West
Harvester
0.002
0.002
0.001
0.001
0.001
East
Harvester
0.002
0.002
0.001
0.001
0.001
Kinderhook Lake
West
Harvester
0.001
0.001
0.001
0.001
0.001
RME
East
Harvester
0.001
0.001
0.001
0.001
0.001
Ingestion
Alcove Reservoir
West
Harvester
0.001
0.001
0.001
0.000
0.001
Rate
East
Harvester
0.001
0.001
0.001
0.001
0.001
Farm Only
-
West
Harvester
0.000
0.000
0.000
0.000
0.000
-
East
Harvester
0.000
0.000
0.000
0.000
0.000
Pond
-
Harvester
0.132
0.132
0.111
0.071
0.122
-
No Harvester
0.198
0.108
0.199
0.167
0.184
Fisherman Only
Nassau Lake
-
Harvester
0.001
0.001
0.001
0.001
0.001
Kinderhook Lake
-
Harvester
0.001
0.001
0.001
0.001
0.001
Alcove Reservoir
-
Harvester
0.001
0.001
0.001
0.000
0.001
Water Ingestion Only
-
-
Harvester
0.000
0.000
0.000
0.000
0.000
1-2-28
-------
Exhibit 3-4. Summary Results - Hazard Quotients for Met
Ingestion
Rates
Scenario Type
Water body
Farm
Parcel
Harvester in
Ravena Pond?
HQ
Child
(1-2)
HQ
Child
(3-5)
HQ
Child
(6-11)
HQ
Child
(12-19)
HQ
Adult
(20-70)
Screening
Screening
Screening
Harvester
0.079
0.078
0.060
0.043
0.068
West
Harvester
0.056
0.056
0.044
0.031
0.049
Ravena Pond
No Harvester
0.084
0.084
0.066
0.047
0.075
East
Harvester
0.056
0.056
0.044
0.031
0.049
No Harvester
0.084
0.084
0.066
0.047
0.075
Combined
Nassau Lake
West
Harvester
0.001
0.001
0.001
0.000
0.001
East
Harvester
0.001
0.001
0.001
0.000
0.001
Kinderhook Lake
West
Harvester
0.001
0.001
0.000
0.000
0.000
CTE
East
Harvester
0.001
0.001
0.000
0.000
0.001
Ingestion
Alcove Reservoir
West
Harvester
0.000
0.000
0.000
0.000
0.000
Rate
East
Harvester
0.000
0.000
0.000
0.000
0.000
Farm Only
-
West
Harvester
0.000
0.000
0.000
0.000
0.000
-
East
Harvester
0.000
0.000
0.000
0.000
0.000
Pond
-
Harvester
0.056
0.056
0.044
0.031
0.049
-
No Harvester
0.084
0.047
0.084
0.066
0.075
Fisherman Only
Nassau Lake
-
Harvester
0.001
0.001
0.000
0.000
0.001
Kinderhook Lake
-
Harvester
0.001
0.001
0.000
0.000
0.000
Alcove Reservoir
-
Harvester
0.000
0.000
0.000
0.000
0.000
Water Ingestion Only
-
-
Harvester
0.000
0.000
0.000
0.000
0.000
lyl Mercury Scenarios using CTE Ingestion Rates
1-2-29
-------
Exhibit 3-5. Fractional Pathway of Age-Specific Hazard Quotients for Divalent Mercury
for all F
lavena Scenarios, with H
arvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult
CTE Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
57.7%
55.8%
60.7%
59.8%
64.5%
Egg, Pork, & Poultry
34.4%
34.8%
28.4%
31.4%
26.8%
Beef & Dairy
2.0%
2.8%
5.1%
3.5%
3.9%
Fish
1.9%
2.7%
2.8%
3.0%
3.4%
Soil
4.0%
3.9%
3.0%
2.2%
1.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
62.0%
60.1%
64.6%
64.0%
68.2%
Egg, Pork, & Poultry
29.5%
30.0%
24.2%
26.9%
22.7%
Beef & Dairy
1.8%
2.6%
4.6%
3.2%
3.5%
Fish
2.3%
3.2%
3.3%
3.6%
4.0%
Soil
4.4%
4.2%
3.2%
2.4%
1.5%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
57.1%
55.1%
59.8%
58.9%
63.4%
Egg, Pork, & Poultry
34.0%
34.3%
28.0%
31.0%
26.3%
Beef & Dairy
2.0%
2.7%
5.0%
3.5%
3.8%
Fish
2.9%
4.1%
4.2%
4.5%
5.1%
Soil
4.0%
3.8%
2.9%
2.2%
1.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
61.3%
59.1%
63.5%
62.8%
66.8%
Egg, Pork, & Poultry
29.1%
29.5%
23.8%
26.4%
22.2%
Beef & Dairy
1.8%
2.5%
4.6%
3.1%
3.4%
Fish
3.4%
4.9%
4.9%
5.3%
6.0%
Soil
4.3%
4.1%
3.1%
2.4%
1.5%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
57.6%
55.7%
60.5%
59.6%
64.3%
Egg, Pork, & Poultry
34.3%
34.7%
28.4%
31.3%
26.7%
Beef & Dairy
2.0%
2.8%
5.1%
3.5%
3.8%
Fish
2.1%
3.0%
3.1%
3.3%
3.8%
Soil
4.0%
3.8%
3.0%
2.2%
1.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
61.9%
59.9%
64.4%
63.7%
68.0%
Egg, Pork, & Poultry
29.4%
29.9%
24.1%
26.8%
22.6%
Beef & Dairy
1.8%
2.5%
4.6%
3.2%
3.5%
Fish
2.5%
3.6%
3.6%
3.9%
4.4%
Soil
4.4%
4.2%
3.2%
2.4%
1.5%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-30
-------
Exhibit 3-5. Fractional Pathway of Age-Specific Hazard Quotients for Divalent Mercury
for all F
lavena Scenarios, with H
arvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult
CTE Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
10.5%
7.5%
8.0%
7.4%
7.1%
Egg, Pork, & Poultry
6.2%
4.7%
3.8%
3.9%
3.0%
Beef & Dairy
0.4%
0.4%
0.7%
0.4%
0.4%
Fish
82.2%
86.9%
87.2%
88.0%
89.3%
Soil
0.7%
0.5%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
9.7%
6.9%
7.3%
6.8%
6.5%
Egg, Pork, & Poultry
4.6%
3.5%
2.8%
2.8%
2.2%
Beef & Dairy
0.3%
0.3%
0.5%
0.3%
0.3%
Fish
84.7%
88.8%
89.0%
89.8%
90.9%
Soil
0.7%
0.5%
0.4%
0.3%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
57.5%
55.0%
63.7%
63.0%
67.8%
Egg, Pork, & Poultry
26.6%
27.3%
23.3%
27.4%
22.9%
Beef & Dairy
1.9%
3.0%
5.7%
3.0%
3.3%
Fish
1.7%
2.5%
2.7%
2.8%
3.5%
Soil
12.2%
12.1%
4.6%
3.7%
2.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
60.8%
58.3%
67.1%
66.8%
71.1%
Egg, Pork, & Poultry
22.4%
23.1%
19.6%
23.2%
19.2%
Beef & Dairy
1.8%
2.7%
5.2%
2.7%
3.0%
Fish
2.0%
2.9%
3.2%
3.3%
4.1%
Soil
13.0%
12.9%
4.9%
3.9%
2.6%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
57.0%
54.3%
62.8%
62.1%
66.6%
Egg, Pork, & Poultry
26.4%
27.0%
23.0%
27.0%
22.5%
Beef & Dairy
1.9%
3.0%
5.6%
3.0%
3.3%
Fish
2.6%
3.8%
4.1%
4.2%
5.3%
Soil
12.1%
12.0%
4.5%
3.6%
2.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
60.2%
57.4%
66.0%
65.6%
69.6%
Egg, Pork, & Poultry
22.2%
22.8%
19.3%
22.8%
18.8%
Beef & Dairy
1.7%
2.7%
5.1%
2.7%
2.9%
Fish
3.0%
4.4%
4.8%
5.0%
6.2%
Soil
12.9%
12.7%
4.8%
3.9%
2.5%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-31
-------
Exhibit 3-5. Fractional Pathway of Age-Specific Hazard Quotients for Divalent Mercury
for all F
lavena Scenarios, with H
arvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult
RME Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
57.4%
54.9%
63.5%
62.9%
67.5%
Egg, Pork, & Poultry
26.6%
27.3%
23.2%
27.3%
22.8%
Beef & Dairy
1.9%
3.0%
5.7%
3.0%
3.3%
Fish
1.9%
2.8%
3.0%
3.1%
3.9%
Soil
12.2%
12.1%
4.6%
3.7%
2.4%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
60.7%
58.1%
66.9%
66.5%
70.8%
Egg, Pork, & Poultry
22.4%
23.0%
19.6%
23.1%
19.2%
Beef & Dairy
1.8%
2.7%
5.1%
2.7%
3.0%
Fish
2.2%
3.3%
3.5%
3.7%
4.6%
Soil
13.0%
12.9%
4.9%
3.9%
2.6%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
11.4%
7.9%
8.5%
8.2%
7.2%
Egg, Pork, & Poultry
5.3%
3.9%
3.1%
3.6%
2.4%
Beef & Dairy
0.4%
0.4%
0.8%
0.4%
0.4%
Fish
80.5%
85.9%
86.9%
87.4%
89.7%
Soil
2.4%
1.7%
0.6%
0.5%
0.3%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
10.6%
7.3%
7.8%
7.5%
6.6%
Egg, Pork, & Poultry
3.9%
2.9%
2.3%
2.6%
1.8%
Beef & Dairy
0.3%
0.3%
0.6%
0.3%
0.3%
Fish
83.0%
87.8%
88.7%
89.1%
91.1%
Soil
2.3%
1.6%
0.6%
0.4%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-32
-------
Exhibit 3-6. Fractional Pathway of Age-Specific Hazard Quotients for Methyl Mercury
for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult 20-70
CTE Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
12.5%
8.8%
8.1%
7.7%
6.5%
Egg, Pork, & Poultry
1.0%
0.7%
0.6%
0.6%
0.5%
Beef & Dairy
2.8%
3.1%
5.6%
3.7%
3.6%
Fish
0.829217
0.868195
0.853748
0.877842
0.892817
Soil
0.8%
0.5%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
11.6%
8.2%
7.5%
7.1%
6.0%
Egg, Pork, & Poultry
0.7%
0.5%
0.4%
0.4%
0.3%
Beef & Dairy
2.1%
2.2%
4.1%
2.7%
2.6%
Fish
84.9%
88.6%
87.6%
89.5%
90.9%
Soil
0.7%
0.5%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
8.7%
6.0%
5.5%
5.2%
4.4%
Egg, Pork, & Poultry
0.7%
0.5%
0.4%
0.4%
0.3%
Beef & Dairy
1.9%
2.1%
3.8%
2.5%
2.4%
Fish
88.2%
91.0%
90.0%
91.7%
92.8%
Soil
0.5%
0.4%
0.3%
0.2%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
8.0%
5.5%
5.1%
4.8%
4.0%
Egg, Pork, & Poultry
0.5%
0.4%
0.3%
0.3%
0.2%
Beef & Dairy
1.4%
1.5%
2.8%
1.8%
1.8%
Fish
89.6%
92.2%
91.6%
92.9%
93.9%
Soil
0.5%
0.3%
0.3%
0.2%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
7.5%
5.2%
4.8%
4.5%
3.8%
Egg, Pork, & Poultry
0.6%
0.4%
0.3%
0.4%
0.3%
Beef & Dairy
1.7%
1.8%
3.3%
2.2%
2.1%
Fish
89.7%
92.2%
91.3%
92.8%
93.7%
Soil
0.5%
0.3%
0.2%
0.2%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
6.9%
4.8%
4.4%
4.1%
3.5%
Egg, Pork, & Poultry
0.4%
0.3%
0.2%
0.3%
0.2%
Beef & Dairy
1.2%
1.3%
2.4%
1.6%
1.5%
Fish
91.0%
93.3%
92.7%
93.9%
94.7%
Soil
0.4%
0.3%
0.2%
0.2%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-33
-------
Exhibit 3-6. Fractional Pathway of Age-Specific Hazard Quotients for Methyl Mercury
for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult 20-70
CTE Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.1%
0.1%
0.1%
0.1%
0.0%
Egg, Pork, & Poultry
0.0%
0.0%
0.0%
0.0%
0.0%
Beef & Dairy
0.0%
0.0%
0.0%
0.0%
0.0%
Fish
99.9%
99.9%
99.9%
99.9%
99.9%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
CTE Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.1%
0.1%
0.1%
0.0%
0.0%
Egg, Pork, & Poultry
0.0%
0.0%
0.0%
0.0%
0.0%
Beef & Dairy
0.0%
0.0%
0.0%
0.0%
0.0%
Fish
99.9%
99.9%
99.9%
99.9%
99.9%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Alcove Reservoir, East Farm
Fruits & Vegetables
13.0%
9.1%
8.4%
8.9%
7.1%
Egg, Pork, & Poultry
0.8%
0.6%
0.5%
0.6%
0.4%
Beef & Dairy
3.1%
3.6%
6.3%
3.3%
3.0%
Fish
80.6%
84.9%
84.2%
86.7%
89.3%
Soil
2.5%
1.8%
0.6%
0.5%
0.3%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Alcove Reservoir, West Farm
Fruits & Vegetables
12.1%
8.4%
7.8%
8.3%
6.5%
Egg, Pork, & Poultry
0.6%
0.4%
0.3%
0.4%
0.3%
Beef & Dairy
2.2%
2.6%
4.7%
2.4%
2.2%
Fish
82.7%
86.8%
86.6%
88.5%
90.8%
Soil
2.3%
1.7%
0.6%
0.5%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Kinderhook Lake, East Farm
Fruits & Vegetables
9.1%
6.3%
5.8%
6.1%
4.8%
Egg, Pork, & Poultry
0.6%
0.4%
0.3%
0.4%
0.3%
Beef & Dairy
2.1%
2.5%
4.4%
2.3%
2.0%
Fish
86.5%
89.6%
89.1%
90.9%
92.7%
Soil
1.8%
1.2%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Kinderhook Lake, West Farm
Fruits & Vegetables
8.3%
5.7%
5.3%
5.6%
4.4%
Egg, Pork, & Poultry
0.4%
0.3%
0.2%
0.3%
0.2%
Beef & Dairy
1.6%
1.8%
3.2%
1.6%
1.5%
Fish
88.1%
91.0%
90.8%
92.2%
93.8%
Soil
1.6%
1.1%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-34
-------
Exhibit 3-6. Fractional Pathway of Age-Specific Hazard Quotients for Methyl Mercury
for all Ravena Scenarios, with Harvester in Pond
Pathway
Child 1-2
Child 3-5
Child 6-11
Child 12-19
Adult 20-70
RME Ingestion Rate, Nassau Lake, East Farm
Fruits & Vegetables
7.9%
5.4%
5.0%
5.3%
4.1%
Egg, Pork, & Poultry
0.5%
0.4%
0.3%
0.3%
0.2%
Beef & Dairy
1.9%
2.1%
3.8%
2.0%
1.7%
Fish
88.2%
91.0%
90.5%
92.1%
93.7%
Soil
1.5%
1.1%
0.4%
0.3%
0.2%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Nassau Lake, West Farm
Fruits & Vegetables
7.3%
5.0%
4.6%
4.8%
3.8%
Egg, Pork, & Poultry
0.4%
0.3%
0.2%
0.2%
0.2%
Beef & Dairy
1.4%
1.6%
2.8%
1.4%
1.3%
Fish
89.6%
92.2%
92.1%
93.2%
94.6%
Soil
1.4%
1.0%
0.3%
0.3%
0.1%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Ravena Pond, East Farm
Fruits & Vegetables
0.1%
0.1%
0.1%
0.1%
0.0%
Egg, Pork, & Poultry
0.0%
0.0%
0.0%
0.0%
0.0%
Beef & Dairy
0.0%
0.0%
0.0%
0.0%
0.0%
Fish
99.9%
99.9%
99.9%
99.9%
99.9%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
RME Ingestion Rate, Ravena Pond, West Farm
Fruits & Vegetables
0.1%
0.1%
0.1%
0.1%
0.0%
Egg, Pork, & Poultry
0.0%
0.0%
0.0%
0.0%
0.0%
Beef & Dairy
0.0%
0.0%
0.0%
0.0%
0.0%
Fish
99.9%
99.9%
99.9%
99.9%
99.9%
Soil
0.0%
0.0%
0.0%
0.0%
0.0%
Water
0.0%
0.0%
0.0%
0.0%
0.0%
1-2-35
-------
Exhibit 3-7. Mercury Hazard Quotients in All Water Bodies Compared to Alcove
Reservoir3
Chemical
Farm
Parcel
Water Body
HQ
Child
1-2
Child 1-2
Ratio of
Water Body
: Alcove
Reservoir
HQ
Adult
20-70
Adult 20-70
Ratio of
Water Body
: Alcove
Reservoir
Alcove Reservoir
0.004
-
0.002
-
West
Kinderhook Lake
0.004
1.0
0.002
1.0
Nassau Lake
0.004
1.0
0.002
1.0
Ravena Pond
0.020
5.7
0.017
10.8
Alcove Reservoir
0.004
-
0.002
-
Divalent
East
Kinderhook Lake
0.004
1.0
0.002
1.0
Mercury
Nassau Lake
0.004
1.0
0.002
1.0
Ravena Pond
0.021
5.0
0.017
9.4
Alcove Reservoir
0.000
-
0.000
-
None
Kinderhook Lake
0.000
1.5
0.000
1.5
Nassau Lake
0.000
0.7
0.000
1.1
Ravena Pond
0.017
215.0
0.016
237.9
Alcove Reservoir
0.001
-
0.001
-
West
Kinderhook Lake
0.001
1.4
0.001
1.5
Nassau Lake
0.002
1.2
0.001
1.7
Ravena Pond
0.132
81.4
0.122
148.3
Alcove Reservoir
0.001
-
0.001
-
Methyl
East
Kinderhook Lake
0.001
1.4
0.001
1.5
Mercury
Nassau Lake
0.002
1.1
0.001
1.7
Ravena Pond
0.132
80.1
0.122
145.9
Alcove Reservoir
0.001
-
0.001
-
None
Kinderhook Lake
0.001
1.5
0.001
1.5
Nassau Lake
0.001
1.2
0.001
1.8
Ravena Pond
0.132
90.7
0.122
163.3
a Results include a harvester in the Ravena Pond. Results were also consistent when using the RME ingestion
rate and the CTE ingestion rates.
1-2-36
-------
Exhibit 3-8. Comparison of Hazard Quotients Using East Farm Parcels Versus West
Farm Parcels in the Ravena Scenario
Chemical
Ingestion Rate
Scenario Type
Water body
HQ Child
(1-2)
Alcove Reservoir
18%
Combined
Kinderhook Lake
17%
RME
Nassau Lake
17%
Ravena Pond
3%
Divalent
Farm Only
-
18%
Mercury
Alcove Reservoir
20%
Combined
Kinderhook Lake
19%
CTE
Nassau Lake
20%
Ravena Pond
3%
Farm Only
-
20%
Alcove Reservoir
3%
Combined
Kinderhook Lake
2%
RME
Nassau Lake
2%
Ravena Pond
0%
Methyl Mercury
Farm Only
-
15%
Alcove Reservoir
2%
Combined
Kinderhook Lake
2%
CTE
Nassau Lake
1%
Ravena Pond
0%
Farm Only
-
16%
a The change in hazard quotients using each farm parcel was also compared for children of other age groups and
adults, and the percent reduction in risk was found to be fairly consistent across these age groups. Results include
a harvester in the Ravena Pond.
1-2-37
-------
Exhibit 3-9. Comparison of Hazard Quotients for Ravena Scenario Using CTE and RME
Ingestion Rates
Chemical
Scenario
Farm
Parcel
Water body
RME
Ingestion
Rate, HQ
Child 1-2
CTE
Ingestion
Rate, HQ
Child 1-2
Ratio
RME : CTE
HQ
Divalent
Mercury
Screening
Screening
Screening Pond
0.344
0.085
4.1
Combined
West
Alcove Reservoir
0.004
0.001
2.7
Kinderhook Lake
0.004
0.001
2.7
Nassau Lake
0.004
0.001
2.7
Ravena Pond
0.022
0.009
2.4
East
Alcove Reservoir
0.004
0.002
2.6
Kinderhook Lake
0.004
0.002
2.6
Nassau Lake
0.004
0.002
2.6
Ravena Pond
0.023
0.010
2.4
Farm Only
West
-
0.003
0.001
2.7
East
-
0.004
0.002
2.7
Fisherman
Only
-
Alcove Reservoir
0.000
0.000
2.4
Kinderhook Lake
0.000
0.000
2.4
Nassau Lake
0.000
0.000
2.4
Ravena Pond
0.019
0.008
2.4
Water
Ingestion Only
0.000
0.000
0.0
Methyl
Mercury
Screening
Screening
Screening Pond
0.193
0.079
2.5
Combined
West
Alcove Reservoir
0.001
0.000
2.4
Kinderhook Lake
0.001
0.001
2.4
Nassau Lake
0.002
0.001
2.4
Ravena Pond
0.199
0.084
2.4
East
Alcove Reservoir
0.001
0.000
2.4
Kinderhook Lake
0.001
0.001
2.4
Nassau Lake
0.002
0.001
2.4
Ravena Pond
0.199
0.084
2.4
Farm Only
West
-
0.000
0.000
2.8
East
-
0.000
0.000
2.8
Fisherman
Only
-
Alcove Reservoir
0.001
0.000
2.4
Kinderhook Lake
0.001
0.001
2.4
Nassau Lake
0.001
0.001
2.4
Ravena Pond
0.198
0.084
2.4
Water
Ingestion Only
.
.
0.000
0.000
2.2
1-2-38
-------
Exhibit 3-10. Mercury Dermal Hazard Quotients for Water and Soil Contact
Age-
Specific
Hazard
Quotient
Exposure
Media
East Farm
- Tilled,
Alcove
Reservoir
East Farm -
Untitled,
Alcove
Reservoir
West Farm
- Tilled,
Alcove
Reservoir
West Farm -
Untitled,
Alcove
Reservoir
Screen
Child < 1
HQ
Soil
4.44E-04
6.66E-03
4.02E-04
4.77E-03
1.70E-01
Water
3.19E-06
3.19E-06
3.19E-06
3.19E-06
1.30E-03
Soil and Water
4.47E-04
6.67E-03
4.05E-04
4.77E-03
1.71E-01
Child 1-2
HQ
Soil
3.74E-04
5.62E-03
3.39E-04
4.03E-03
1.43E-01
Water
2.83E-06
2.83E-06
2.83E-06
2.83E-06
1.15E-03
Soil and Water
3.77E-04
5.62E-03
3.42E-04
4.03E-03
1.44E-01
Child 3-5
HQ
Soil
3.54E-04
5.31 E-03
3.20E-04
3.80E-03
1.35E-01
Water
2.55E-06
2.55E-06
2.55E-06
2.55E-06
1.04E-03
Soil and Water
3.56E-04
5.31 E-03
3.23E-04
3.80E-03
1.36E-01
Child 6-11
Soil
2.78E-04
4.17E-03
2.51 E-04
2.98E-03
1.06E-01
Water
2.12E-06
2.12E-06
2.12E-06
2.12E-06
8.64E-04
Soil and Water
2.80E-04
4.17E-03
2.53E-04
2.99E-03
1.07E-01
Child 12-19
HQ
Soil
2.11E-04
3.17E-03
1.91 E-04
2.27E-03
8.07E-02
Water
1.67E-06
1.67E-06
1.67E-06
1.67E-06
6.79E-04
Soil and Water
2.13E-04
3.17E-03
1.93E-04
2.27E-03
8.14E-02
Adult HQ
Soil
9.45E-05
1.42E-03
8.55E-05
1.02E-03
3.61 E-02
Water
9.22E-07
9.22E-07
9.22E-07
9.22E-07
3.75E-04
Soil and Water
9.54E-05
1.42E-03
8.65E-05
1.02E-03
3.65E-02
1-2-39
-------
APPENDIX J: Ecological Risk Assessment Case Study - Lafarge Ravena Portland
Cement Facility
-------
TABLE OF CONTENTS
J-1 Introduction 1
J-1.1 Scope of this ERA 1
J-1.2 Modeling of the Ravena Facility 2
J-1.3 Organization of this Appendix 2
J-2 Problem Formulation 2
J-2.1 Selection of Ecological HAPs of Concern 3
J-2.1.1 Mercury and Dioxins 3
J-2.1.2 Hydrogen Chloride 4
J-2.2 Site Selection 5
J-2.2.1 Site Selection for Mercury and Dioxins 5
J-2.2.2 Site Selection for Hydrogen Chloride 5
J-2.3 Selection of Assessment Endpoints 5
J-2.3.1 For Mercury and Dioxins at Ravena Facility 6
J-2.3.2 For Hydrogen Chloride 7
J-2.4 Modeling Fate and Transport 7
J-2.4.1 For Mercury and Dioxins at Ravena Facility 7
J-2.4.2 For Hydrogen Chloride 7
J-3 Methods 8
J-3.1 HAP Emissions Data 8
J-3.2 Mercury and Dioxins 8
J-3.2.1 TRIM.FaTE Aquatic Ecosystem Modeling 8
J-3.2.2 Exposure Assessment 9
J-3.2.3 Ecological Effects Assessment 10
J-3.2.4 Risk Characterization 12
J-3.3 Hydrogen Chloride 12
J-3.3.1 Facility Ranking 12
J-3.3.2 Refined Facility Ranking 14
J-3.3.3 Exposure Assessment - Site-specific Data 15
J-3.3.4 Terrestrial Environments 15
J-3.3.5 Ecological Effects Assessment 16
J-3.3.6 Ecological Risk Characterization 16
J-4 Results 17
J-4.1 Results for Mercury and Dioxins 17
J-4.1.1 Exposure Assessment 17
J-4.1.2 Ecological Effects Assessment 36
J-4.1.3 Risk Characterization 45
J-4.1.4 Uncertainties in Ravena ERA Related to Mercury and Dioxin 46
J-4.2 Results for HCI 48
J-4.2.1 Results for Facility-Ranking Analysis 48
J-4.2.2 Indirect Ecological Effects Assessment 52
J-4.2.3 Indirect Ecological Risk Characterization 55
J-5 References 61
Attachment J-1 - Ecological Risk Assessment Case Study Supporting Documents
j-i
-------
LIST OF EXHIBITS
Exhibit 4-1. Distribution of Length of Fish Consumed by Common Mergansers in Michigan
(Alexander 1977) 20
Exhibit 4-2. Annual Mean Adult Body Weights and Food Ingestion Rtes Assumed for Wildlife
Species 24
Exhibit 4-3. Fraction Diet Assumptions for Wildlife Feeding from Alcove Reservoir, Nassau
Lake, and Kinderhook Lake 24
Exhibit 4-4. Fraction Diet Assumptions for Wildlife Feeding from Ravena Pond as Modeled in
TRIM.FaTE 25
Exhibit 4-5. Estimated Average Daily Ingestion Rate of Each Food Type in the Diets of Wildlife
Species from Alcove Reservoir, Nassau Lake, and Kinderhook Lake (g/day) ....25
Exhibit 4-6. Estimated Average Daily Ingestion Rate of Each Food Type in the Diets of Wildlife
Species from Ravena Pond (g/day) 26
Exhibit 4-7. Concentrations (pg/g) of Methyl Mercury in Compartments of the TRIM.FaTE
Aquatic Food Web at Year 50 - Based on Mean Measured Annual Hg Emission
Rate 26
Exhibit 4-8. Concentrations (pg/g) of Divalent Mercury in Compartments of the TRIM.FaTE
Aquatic Food Web at Year 50 - Based on Mean Measured Annual Hg Emission
Rate 27
Exhibit 4-9. Concentrations (pg/g) of 2,3,7,8-TCDD in Compartments of the TRIM.FaTE Aquatic
Food Web at Year 50 with Mean Emission Rate 28
Exhibit 4-10. Concentrations (pg/g) of 2,3,7,8-TCDD in Compartments of the TRIM.FaTE
Aquatic Food Web at Year 50 with 95-percent UCL Emission Rate 28
Exhibit 4-11. Biomass of Fish Harvested by a Single Angler Fishing in Ravena Pond Relative to
Standing Biomass of Fish in Each Compartment 29
Exhibit 4-12. Concentrations (pg/g) of Mercury and 2,3,7,8-TCDD in the Ravena Pond Aquatic
Compartments at Year 50 Without Fish Harvesting by Humans or Wildlife 30
Exhibit 4-13. Concentrations (pg/g) of Mercury and 2,3,7,8-TCDD in the Ravena Pond Aquatic
Compartments at Year 50 With 17 Grams Fish Harvested per Day by One Angler
from Two Fish Compartments 30
Exhibit 4-14. 2,3,7,8-TCDD Concentrations in Aquatic Foodweb Compartments With and
Without Angler Harvesting of 17 Grams of Fish Daily in Ravena Pond 31
Exhibit 4-15. Concentrations of Divalent and Methyl Mercury in Aquatic Foodweb
Compartments, With and Without Angler Harvesting of 17.0 grams of Fish Daily
in Ravena Pond 32
Exhibit 4-16. Speciated Mercury Concentrations for Surface Water, Sediment, and Biota in
Nassau Lake (ppm [SW: mg/L; sediment: pg/g dry weight; algae, Bl, fish: pg/g
wet weight]) 33
Exhibit 4-17. Tree Swallow Intake of MeHg (pg/g-day) 33
Exhibit 4-18. Common Merganser Intake of MeHg (pg/g-day) 34
Exhibit 4-19. Bald Eagle Intake of MeHg (pg/g-day) 34
Exhibit 4-20. Mink Intake of MeHg (pg/g-day) 34
Exhibit 4-21. Wildlife Intakes of Hg+2 (pg/g-day) at Ravena Pond 35
Exhibit 4-22. Tree Swallow Intake of 2,3,7,8-TCDD (pg/g-day) 35
Exhibit 4-23. Common Merganser Intake of 2,3,7,8-TCDD (pg/g-day) 35
Exhibit 4-24. Bald Eagle Intake of 2,3,7,8-TCDD (pg/g-day) 36
Exhibit 4-25. Mink Intake of 2,3,7,8-TCDD (pg/g-day) 36
Exhibit 4-26. Summary of Wildlife TRVs (pg[chemical]/kg[BW]-day) 44
Exhibit 4-27. Hazard Quotients for Wildlife Exposure to Methyl Mercury 45
Exhibit 4-28. Hazard Quotients for Wildlife Exposure to Divalent Mercury 45
j-ii
-------
Exhibit 4-29. Hazard Quotients for Wildlife Exposure to 2,3,7,8-TCDD 45
Exhibit 4-30. Final Hazard Scores for Top Thirteen Portland Cement Facilities Emitting HCI... 51
Exhibit 4-31. Measurements of Water pH for Alcove Reservoir in Albany County, NY 56
Exhibit 4-32. Measurements of Water pH for Kinderhook Lake in Albany County, NY 56
Exhibit 4-33. Measurements of Soil pH and Effective CEC for Sensitive Terrestrial
Environments Near Portland Cement Facilities Emitting HCI 58
Exhibit 4-34. Measurements of Surface Water pH for Sensitive Aquatic Environments Near
Portland Cement Facilities Emitting HCI 61
j-iii
-------
J-1 Introduction
Section 112 of the Clean Air Act (CAA) establishes a two-stage regulatory process for
addressing emissions of hazardous air pollutants (HAPs) from stationary sources. In the first
stage, section 112(d)(2) requires the U.S. Environmental Protection Agency (EPA) to develop
technology-based standards for categories of industrial sources (e.g., Portland Cement
manufacturing, pulp and paper mills). EPA has largely completed the initial Maximum
Achievable Control Technology (MACT) standards as required under this provision. Under
section 112(d)(6), EPA must review each of these technology-based standards at least every
eight years and revise a standard, as necessary, "taking into account developments in
practices, processes and control technologies." In the second stage, EPA is required under
section 112(f)(2) to assess the health and environmental risks that remain after sources come
into compliance with MACT. If additional risk reductions are necessary to protect public health
with an ample margin of safety or to prevent adverse environmental effects, EPA must develop
standards to address these remaining risks. This second stage of the regulatory process is
known as the residual risk stage, and EPA is implementing this stage in accordance with its
Risk and Technology Review (RTR) Assessment Plan (EPA 2006).
This appendix presents an ecological risk assessment (ERA) in support of the RTR analysis for
the Portland Cement manufacturing industry. In particular, ERAs are presented for releases of
hydrogen chloride (HCI), mercury (Hg), and dioxins (specifically, 2,3,7,8-tetrachlorodibenzo-p-
dioxin or 2,3,7,8-TCDD) from Portland Cement facilities in the United States. Potential
ecological effects from HCI emissions include both direct damage to the foliage of plants and
indirect adverse effects from the acidification of soils and surface waters. Ecological effects of
concern for Hg and dioxins are similar to those for humans, because these chemicals tend to
bioaccumulate in food chains and because they are highly toxic to mammals and other
vertebrate wildlife.
J-1.1 Scope of this ERA
For HCI, this Appendix evaluates possible indirect ecological effects that might result if pH were
reduced in soils and surface waters in the vicinity of one or more Portland Cement facilities. It
describes a screening-level ranking assessment to identify which facilities are expected to pose
the highest risk, if any, of indirect effects of HCI deposition on nearby sensitive ecosystems.
Available data are examined for top-ranked facilities to determine if there is any evidence to
date of such effects. A separate document evaluates possible direct effects of HCI in air to the
foliage of plants in the vicinity of these facilities (Appendix K).
For Hg and dioxin, this Appendix presents a site-specific, refined ERA for the Portland Cement
manufacturing facility with the highest releases of these chemicals, the Ravena Lefarge
Portland Cement Facility near Ravena, New York. Although the selection of chemicals and the
Ravena facility initially was based on a screen of Portland Cement facilities in relation to de
minimis emission rates derived to protect human health, the same chemicals and site also are
appropriate for evaluating potential ecological risks from persistent and bioaccumulative (PB)
hazardous air pollutants (HAPs) emitted by this source category. The Ravena ERA evaluates
whether post-MACT emissions from the facility could result in concentrations of Hg and dioxins
in nearby aquatic food webs sufficient to cause adverse effects in local populations of birds and
mammals that feed from these food webs.
For future RTR risk assessments, EPA is developing a systematic tiered ecological screening
methodology to identify chemicals and facilities of most ecological concern.
j-1
-------
J-1.2 Modeling of the Ravena Facility
The Ravena ERA and human health risk assessment (HHRA) (Appendix I) for Hg and dioxin
were designed to use common data sources, methods, and assumptions when possible to
ensure consistency and efficiency. The Ravena ERA and HHRA use the same emissions data
and environmental fate and transport modeling for Hg and dioxin. Aquatic food webs that could
be used to assess both human and wildlife exposures were developed for surface waters in the
vicinity of the Ravena facility. Those aspects of the Ravena ERA problem formulation and
methods that are shared with the Ravena HHRA are summarized briefly in this Appendix and
presented in detail in the HHRA documentation (Appendix I).
J-1.3 Organization of this Appendix
The problem formulation for and methods used in the ERA are described in Section J-2 and
Section J-3, respectively, of this Appendix. Section J-4 presents the results of the ERA, and
Section J-5 identifies the references cited.
J-2 Problem Formulation
"Problem formulation" in an ecological risk assessment (ERA) defines one or more hypotheses
regarding the potential ecological effects to be evaluated and establishes the scope and
methods for analyzing risks (EPA 1998). Problem formulation for evaluation of residual risks
requires several decisions, including identification of chemicals of potential concern (COPC)
emitted by source category facilities. For source categories with a large number of facilities
across the United States, screening or ranking facilities based on emissions and other factors,
exposure pathways, and surrogate measures of risk can narrow the focus of an assessment to
the facilities most likely to pose risks. These considerations hold for both human health risk
assessments (HHRAs) and ERAs, even though problem formulation is a phrase first used in the
context of ERAs (EPA 1998). This initial phase of a risk assessment includes identification of
appropriate assessment endpoints for the COPC and the measures or models that can be used
to evaluate or predict adverse changes in those endpoints. Problem formulation generally
concludes with development of an analysis plan for the risk assessment.
Available resources can be put to the best use when an HHRA and an ERA are planned
together, with fate and transport analyses serving double-duty where possible. Problem
formulation activities for the HHRA and the ERA for the Ravena facility were conducted jointly.
The HHRA for Portland Cement is described in an earlier Appendix (Appendix I). Part of the
ERA for the Ravena facility was planned side-by-side with the HHRA and uses the same fate
and transport modeling setup as used for the HHRA. Therefore, many aspects of ERA problem
formulation, including selection of the Ravena facility and its spatial modeling layout, are
described in detail in the documentation of the Ravena HHRA (Appendix I). The documentation
of the Ravena ERA in this Appendix refers to relevant sections of the HHRA Appendix, but
focuses on aspects of the problem formulation (e.g., ecological endpoints) and analyses that are
unique to the ERA.
Problem formulation for the ERA is described below in four parts. Selection of hazardous air
pollutants (HAPs) of potential ecological concern is described in Section J-2.1. Site selection for
a refined ecological risk assessment for those HAPs is described in Section J-2.2. Selection of
"assessment endpoints" is documented in Section J-2.3. Remaining aspects of problem
formulation are described in Section J-2.4. Because the starting information, assessment
endpoints, and approach to exposure and effects analysis, as well as to risk characterization,
J-2
-------
differ for the ecological chemicals of concern, Sections J-2.1 through J-2.4 are each divided into
two subsections based on COPC.
J-2.1 Selection of Ecological HAPs of Concern
The Portland Cement source category encompasses 91 facilities in the United States identified
from the National Emissions Inventory (NEI), as described in the initial emissions screening
analysis presented in Section 3.2 of EPA's report to SAB. An initial screen of persistent and
bioaccumulative (PB-HAPs) was conducted for those facilities by comparing the facility-specific
total emissions for a given PB-HAP to de minimis emission quantities derived for human health
risk endpoints. For persistent and bioaccumulative (PB) HAPs, the de minimis emission
quantities were estimated using a conservatively constructed screening scenario within
TRIM.FaTE to estimate chemical fate and transport, including transfer through both terrestrial
and aquatic food chains. At each facility, PB-HAPs for which the total emissions exceed the de
minimis emissions quantity for that chemical (or chemical group) were selected to further
analysis. The de minimis screening analysis is documented further in Appendix C.
Although emissions of every PB-HAP on EPA's list are not reported for every facility in this
source category, over half of the facilities report emissions of mercury (Hg). In addition, based
on measurements at individual facilities and knowledge of the Portland cement manufacturing
process, it was assumed that every facility emits dioxins. Both Hg and dioxins are presumed to
be emitted in relatively large quantities from at least some facilities in this source category.
Given the potential for human exposure via non-inhalation pathways to these two PB-HAPs and
the relatively high emissions of these chemicals reported for Portland cement facilities, both Hg
and dioxins were expected to be chemicals of concern for the non-inhalation human health risk
assessment (see Appendix I).
Overall, for the ERA we considered two categories of HAPs separately: (1) those that are
sufficiently persistent and bioaccumulative to reach levels in aquatic food chains that are toxic to
piscivorous wildlife at environmental concentrations (e.g., in air, soil, and water) unlikely to
cause direct toxicity to any other group of organisms and (2) those that might cause direct
adverse ecological effects at lower air concentrations than are of concern for human health.
From the first category of chemicals, which roughly corresponds to the PB-HAPs of concern for
human health, Hg and dioxins were selected as described in Section J-2.1.1 below. From the
second category of chemicals, which includes hydrogen fluoride (toxic to plants), we selected
HCI as described in Section J-2.1.2. Data from NEI indicated that hydrogen fluoride (HF) is
emitted from only 3 of the 91 facilities; HF was therefore not included in the ERA for the Ravena
facility.
J-2.1.1 Mercury and Dioxins
PB-HAPs might pose threats to ecological receptors at lower environmental concentrations than
those that pose human health risks because several wildlife species feed almost exclusively on
aquatic prey, while human diets generally are more diversified and, therefore, humans generally
consume less fish per unit body weight than piscivorous wildlife. Based on the prevalence of
HAP emission from Portland Cement facilities and the toxicity of those HAPs to wildlife, we
concluded that dioxins and Hg are the PB-HAPs from this source category most likely to pose a
risk to wildlife predators of aquatic organisms.
The same characteristics of dioxins and Hg that resulted in their selection for the HHRA indicate
their potential for adverse effects in piscivorous wildlife. Dioxins and Hg in both its methylated
(MeHg) and divalent (Hg+2) forms, are toxic to non-human mammals and other classes of
J-3
-------
vertebrates, including birds. Wildlife that feed from aquatic food chains, particularly those that
consume primarily larger fish, tend to be the components of ecosystems that are most highly
exposed to bioaccumulative chemicals.
For both the HHRA and the ERA, ICF used 2,3,7,8-TCDD, the most toxic dioxin congener, to
represent total dioxins. Available data also suggest that 2,3,7,8-TCDD is the most
bioaccumulative of the dioxins in part because it is the least well metabolized by vertebrates and
invertebrates alike and in part because it is more readily taken up than other dioxins and furans.
For Hg, in addition to total Hg, ICF considered three species of Hg - divalent (Hg+2), elemental
(HgO), and methyl mercury (MeHg) - for purposes of modeling fate and transport. The
proportion of total Hg present as MeHg in top predatory fish generally is high, more than 90
percent in most studies (EPA 2001, 2009).
J-2.1.2 Hydrogen Chloride
Hydrogen chloride is the chemical released in the highest quantities each year by many
Portland Cement facilities. Hydrogen chloride also is one of the few HAPs that might produce
direct toxic effects on vegetation or other direct adverse ecological effects near facilities at
concentrations lower than a reference concentration (RfC) for the protection of human health.
At sufficiently high or prolonged air exposures, HCI can directly impact the structure and
function of plant leaves at several levels. The derivation of short-term (1 hr) and long-term (e.g.,
days to weeks) air concentration benchmarks for the protection of plant communities from direct
effects of HCI on leaves is documented in a separate Appendix (Appendix K). Comparison of
the long-term benchmark to the RfC for HCI for humans indicates that air concentrations
protective of humans also will protect plant communities from adverse effects due to chronic
exposure to airborne HCI (Appendix K). The 1-hr HCI air concentration benchmark for foliar
damage, on the other hand, is lower than the most conservative reference exposure
concentration for the protection of human health. See Appendix K for the assessment of risk of
foliar damage from direct exposure of plants to airborne HCI.
Local emissions of HCI also, however, might cause adverse ecosystem effects indirectly
through a gradual decrease in the pH of receiving ecosystems. Lower pH in soils and surface
waters can increase the bioavailability of inorganic contaminants (e.g., aluminum, mercury,
selenium) and cause several types of deleterious effects at multiple levels of biological
organization (Brezonik et al. 1991; Sparling 1995). A significant issue in assessing these effects
is that they are mediated in large part by changes in pH in the receiving surface water and in
soils. Whether pH will change in rain, soils, and surface waters near a facility in response to
local emissions of HCI to air depends on several characteristics of the environment, particularly
current regional levels of acid deposition from all sources, including oxides of nitrogen (NOx)
and sulfur (SOx), and the acid buffering capacity of the receiving ecosystem. In addition, for
soils, the type, depth, and slope also may influence the extent to which pH changes in response
to acid deposition. Whether adverse effects are likely to occur in response to a given change in
pH (range or central tendency) also depends on many environmental factors, including the pH in
the absence of local HCI emissions, the occurrence of other possibly toxic chemicals in soils
and surface waters that may become more bioavailable at lower pH, and the sensitivity of local
organisms to acid conditions (e.g., acid-tolerant plants).
Rainfall in remote areas (away from anthropogenic sources of air pollution) generally has a
slightly acidic pH of approximately 5.6, because carbon dioxide and water in the air react
together to form carbonic acid, a weak acid. Anthropogenic contributions, particularly NOx and
J-4
-------
S0X, to lowering pH (increasing acidity) of rainfall varies regionally, with highest acidities in the
northeastern United States.
In aquatic ecosystems, different plant and animal species have different tolerances for pH of the
water. Loss of sensitive species due to acidification can, depending on location, change aquatic
community structure in ways deleterious to wildlife and to human welfare. In addition, changes
in surface water pH can affect the toxicity of other pollutants to aquatic organisms in the system,
particularly ionic inorganic chemicals and weak organic acids or bases.
In terrestrial ecosystems, soil pH affects the solubility and bioavailability of inorganic nutrient
and pollutant chemicals to plants and can affect microbial community processes. Strongly acid
soils can result in sufficient aluminum+3 activity to be toxic to plants.
This appendix documents a screening-level ERA to evaluate the likelihood that any Portland
Cement facilities might cause any of the adverse indirect effects listed above through releases
of HCI.
J-2.2 Site Selection
In this section, site selection for a refined ecological analysis is first described for Hg and dioxin
(Section J-2.2.1) and then for HCI (Section J-2.2.2).
J-2.2.1 Site Selection for Mercury and Dioxins
The approach used to select the Ravena facility for the HHRA and ERA is documented in
Appendix I. Briefly, we first identified Portland cement facilities that had high emissions for both
Hg and dioxins, assuming that higher emissions of the chemicals would lead to higher human
exposures. For these facilities, we looked for one that had suitable geographic characteristics
for the two basic human health exposure scenarios (i.e., fisher and farmer) and for an ERA. The
Ravena facility was considered appropriate for the ERA because of its proximity to pond, lake,
reservoir, forest, and field habitats for wildlife. With those habitats, several piscivorous wildlife
species can reasonably be expected to be present.
J-2.2.2 Site Selection for Hydrogen Chloride
As a consequence of the influence of environmental characteristics on pH change and
ecological effects of pH change in response to localized HCI emissions, ICF concluded that it
could not estimate a de minimis emission rate for HCI applicable to all facilities. A different
approach to identifying facilities of concern from among the 91 facilities under consideration was
needed than that used for PB-HAPs.
Using the emission data for HCI from Portland facilities across the United States compiled for
use in the HHRA, ICF conducted a proximity and vulnerability screening assessment for
ecologically sensitive environments, as described in Section J-3.3.
J-2.3 Selection of Assessment Endpoints
Assessment endpoints are "explicit expressions of the actual environmental value that is to be
protected, operationally defined by an ecological entity and its attributes" (EPA 1988). Different
assessment endpoints are appropriate for Hg and dioxins (Section J-2.3.1) and HCI (Section J-
2.3.2).
J-5
-------
J-2.3.1
For Mercury and Dioxins at Ravena Facility
The available literature indicates that piscivorous wildlife tend to be early warning indicators in
ecosystems known to be contaminated with Hg and polychlorinated biphenyls (PCBs). Adverse
effects, particularly reproductive and developmental impairments, associated with these
chemical groups have been observed in piscivorous wildlife where other adverse ecological
effects are not evident (Colborn 1991; Giesy et al. 1994; Gilbertson et al. 1991; Keith 1996; Rice
and O'Keefe 1995). In addition, measurements of these chemicals in animal tissues tend to be
higher for piscivorous wildlife than for other wildlife species (Sheffy and St. Amant 1982; Wren
et al. 1986, Wolfe et al. 2006). The persistence, toxicity, and bioaccumulative potential for
2,3,7,8-TCDD are similar to those of many PCBs. The focus of the ERA for dioxins and Hg,
therefore, was narrowed to a risk assessment primarily for piscivorous wildlife in the area
surrounding the Ravena facility.
From an ecological perspective, adverse effects on piscivorous avian and mammalian
populations are likely to be the most sensitive endpoint for TCDD and Hg, largely because they
will be the most highly exposed organisms. Terrestrial food "chains" (e.g., plants —~ herbivores
[such as voles, mice, and rabbits] —~ predators [such as hawks, owls, and canines]) tend to be
"shorter" than aquatic food chains (e.g., algae or detritus —~ zooplankton or benthic
invertebrates —~ small fish —~ larger fish —~ piscivorous wildlife). Moreover, the wildlife species
that have been documented with the highest tissue concentrations of, and in some cases
substantial observed impacts from, bioaccumulative chemicals from the environment tend to be
the piscivores (e.g., mink, otter, osprey, gulls, terns, cormorants, mergansers, bald eagles)
(Colborn 1991; Environment Canada 1991; Eisler 1987; Giesy et al. 1994; Gilbertson et al.
1991; Sheffy and St. Amant 1982; Rice and O'Keefe 1995; Wolfe et al. 2006). Thus, adverse
effects from bioaccumulative chemicals can occur in piscivores at lower environmental
concentrations than are likely to cause adverse effects on other ecological receptors.
Three piscivorous wildlife species were selected for the Ravena ERA based on their likely
presence in the area, their dietary habits, and their overall body size, which affects metabolic
rates and might affect effective dose relative to body weight. In addition, tree swallows were
included to represent consumers of benthic invertebrates (insects) from potentially
contaminiated aquatic environments around the Ravena facility.
• Tree swallows (Tachycineta bicolor) are aerial insectivores. These relatively small
passerine birds (20 grams) have an energetically intensive method of foraging
(catching insects on the wing), a relatively high metabolic rate and food ingestion rate
relative to their body weight, and females can consume up to 100% of their body
weight daily when forming eggs (clutches of 4 to 6 eggs common). For this ERA, it is
assumed that 100% of the tree swallow diet consists of insects emerging from aquatic
environments, and 100% of those insects lived in the benthos as nymphs. In short,
tree swallows consume a diet that is equivalent to 100% benthic invertebrates.
• Common mergansers (Mergus merganser) are included in the ERA primarily for two
reasons: (1) they can and do capture and consume larger (e.g., up to 25 to 30 cm),
higher-trophic-level fish in general than do other birds that might be present near
Ravena (e.g., belted kingfishers are limited to fish of 10 cm or less; great blue herons
forage in shallow areas with smaller fish than those available to mergansers), and (2)
their diet consists entirely offish, unlike some other semi-piscivorous birds that might
consume both terrestrial and aquatic organisms (e.g., great blue herons).
• Bald eagles (Haliaeetus leucocephalus) are large birds (approximately 4.5 kg) with a
relatively low metabolic rate compared to smaller birds. Therefore, a toxicity
J-6
-------
reference dose in mg/kg-day scaled to bald eagles from smaller experimental species
on the basis of metabolic rate would be lower than for smaller bird species. Bald
eagles are long-lived and can scavenge relatively large fish (and other dead animals)
from surface waters as well as catch and carry large live prey under some
circumstances. Finally, although no longer classified as endangered, they continue to
be rare in much of their historic range.
• American minks (Neovison vison) are piscivorous and are likely to occur in the area
around the Ravena facility. EPA has quantified exposure factors for mink in itsl993
Wildlife Exposure Factors Handbook, although some more recent studies also are
available. Mink reproductive success is sensitive to environmental contaminants,
most notably PCBs and Hg. Other potentially exposed mammalian species in the
area of the Ravena facility (e.g., raccoons) may take some or all of their diet from
aquatic ecosystems, but are more omnivorous and variable in their diet than mink,
and so are likely to be less exposed than mink.
J-2.3.2 For Hydrogen Chloride
The types of ecological effects that might be expected from indirect effects of HCI deposition
have been described in Section J-2.1.2. Because those effects cannot yet readily be predicted
on the basis of modeling HCI fate and transport and soil and water chemistry at a local level,
evaluation of assessment endpoints will require observed evidence that adverse ecological
effects consistent with increased acidification are occurring near a facility. Attribution to the
facility will only be possible if there is a clear gradient of effects with increasing distance from
the facility.
J-2.4 Modeling Fate and Transport
J-2.4.1 For Mercury and Dioxins at Ravena Facility
For both the ERA and the HHRA, TRIM.FaTE was used to simulate air dispersion, deposition,
and transport of Hg and TCDD emissions from the Ravena facility and to predict concentrations
of Hg and dioxin in fish for four water bodies in the vicinity of the facility. For Hg, three forms
(HgO, Hg+2, and MeHg) were modeled with transformations among forms simulated as
appropriate in various environmental media. For the HHRA, TRIM.FaTE also was used to
calculate chemical concentrations in additional exposure media (e.g., locally grown produce and
animal products). The water bodies include a small pond (Ravena Pond) near the facility
(located south of the facility), the Alcove Reservoir (located west of the source), and Kinderhook
Lake and Nassau Lake (both located east of the source). See Appendix I for site maps and a
detailed description of the spatial layout of the site, including the areas and locations of the farm
and watershed parcels relative to the Ravena facility, as well as land-use patterns in the area
surrounding the facility.
The ERA evaluates risks to piscivorous wildlife species that obtain prey from the four water
bodies and an insectivorous bird that is assumed to obtain its prey from aquatic environments
around the Ravena facility. Concentrations of TCDD and Hg attributable to the Ravena facility
were not estimated for the nearby Hudson River because it is largely a flow-through system.
J-2.4.2 For Hydrogen Chloride
EC/R modeled fate and transport in air of HCI from Portland Cement facilities across the United
States. See Appendix K for a description of that overall approach.
J-7
-------
J-3 Methods
This section describes the method used for the Ravena ERA. Section J-3.1 identifies the
source of HAP emissions data for the ERA. Section J-3.2 presents the ERA methods used for
mercury (Hg) and dioxins (TCDD). The methods for hydrogen chloride (HCI) are presented in
Section J-3.3.
J-3.1 HAP Emissions Data
Mercury and HCI emissions data for the Ravena, NY, facility were obtained or derived from the
inventory compiled for the Portland cement source category for this case study (based on the
2002 National Emissions Inventory [NEl]). As discussed in the documentation of the initial
emissions screening analysis (Appendix C), the Hg data were examined to confirm that default
speciation was applied to emissions reported as "total Hg" or similar designations. Items
entered in the NEI as "mercury" or "mercury and compounds" were divided into estimated
divalent and elemental Hg emissions. This speciation was achieved using speciation factors by
source category provided by EPA. For Portland Cement, 25 percent of the emissions were
assumed to be divalent mercury (Hg+2) and 75 percent were assumed to be elemental mercury
(HgO).
The NEI does not include dioxin/furan emissions, so a separate analysis was conducted to
estimate the dioxins/furans emissions for Portland Cement facilities (Appendix F). Clinker
production data (in tons per year) were obtained for each facility. Emission factors then were
applied to the clinker production data to calculate a mean and 95th percent upper confidence
limit (UCL) emission rate for 2,3,7,8-TCDD equivalents (TEQs). Both the mean and 95th percent
UCL emission estimates were used in the Ravena ERA.
J-3.2 Mercury and Dioxins
J-3.2.1 TRIM.FaTE Aquatic Ecosystem Modeling
Part of the site-specific HHRA assesses human exposures via aquatic food chain
contamination, considering both bottom-feeding fish that might be consumed by humans and
game fish, in general the top predators in aquatic ecosystems. The same aquatic food webs
developed in TRIM.FaTE for the HHRA are appropriate for use in estimating dose to the wildlife
species chosen as assessment endpoints.
Aquatic food webs in TRIM.FaTE were developed to predict bioaccumulation of chemicals in a
small "farm" pond near the Ravena facility (hereafter called Ravena Pond), Alcove Reservoir,
Kinderhook Lake, and Nassau Lake. There are nine groups of aquatic organisms included in
TRIM.FaTE for one or more of these water bodies:
1. Plankton includes both algae and zooplankton modeled as a "phase" of the water
column;
2. Macrophytes which can accumulate and "sequester" some chemicals (modeled as a
separate compartment);
3. Benthic invertebrates such as mollusks, Crustacea, and aquatic insect nymphs that
consume periphyton and detritus (modeled as a compartment in chemical equilibrium
with bottom sediments);
4. Benthivorous fish which are bottom-feeding fish (e.g., bullhead catfish) that
consume primarily benthic invertebrates;
J-8
-------
5. Bottom-feeding carnivores (i.e., eels) that consume both benthic invertebrates and
fish;
6. Water column planktivores, such as young-of-the-year for many species and other
small fish (e.g., shiners) that consume primarily planktonic organisms;
7. Water column omnivores, which are larger fish that consume invertebrates and
smaller fish from both the benthic and pelagic environments (e.g., "panfish" like
bluegill, yellow perch, and young age classes of the game species);
8. Water column piscivores, which are larger size game fish species that primarily
consume smaller fish in pelagic and/or benthic environments (e.g., walleye,
largemouth bass); and
9. Mallard ducks, which consume aquatic insects, invertebrates, and vegetation (ducks
included as prey for bald eagles as discussed in Section J.4.1.1.1).
The parameterization of these compartments (with the exception of mallards) is described in
Appendix I, Attachment 1-1. Briefly, for Kinderhook and Nassau Lakes, data from several fish
surveys conducted by the New York State Department of Environmental Conservation (NYS
DEC) between 1988 and 2006 were used to estimate the relative abundance of different fish
species in each lake. Data on fish species presence and fish weights in the Alcove Reservoir
were obtained from the NYS DEC fish surveys conducted between 1963 and 1970, after which
surveys ceased and the Reservoir was closed to public uses. The same fish weight data were
used for the other water bodies. The proportion of total fish biomass for each water body
contributed by each species was assigned to one of five fish compartments (numbers 4 through
8 above) on the basis of descriptions of their feeding habits available from online fishing
communities and from NYS DEC online documents. The food web for the small "farm" pond
was derived from an analysis of data presented by Demers et al. (2001) for two small lakes in
Ontario.1
J-3.2.2 Exposure Assessment
Assumptions about the composition of each species' diet were developed based on published
field observations and methods outlined in EPA's Wildlife Exposure Factors Handbook (EPA
1993 WEFH). These assumptions are expressed as percentages of the total diet obtained from
the TRIM.FaTE food web compartments described earlier.
A weight-of-evidence approach was used to develop appropriate ingestion rates for each food
type. Measurements of fish ingestion by captive animals, if available, were compared to
estimates of ingestion rates of free-living animals based on measured or allometric predictions
of average metabolic rates for free-living animals, a central tendency estimate of the gross
energy content of the food type, and the assimilation efficiency of the food type. Where
ingestion estimates differed from measurements, the methods used in primary studies were
reviewed and an ingestion value was selected to best represent an annual average daily
ingestion rate for free-living adult animals.
Mercury and dioxin intakes (exposure doses) for the five potential wildlife receptors were
calculated from concentrations of Hg and TCDD in the TRIM.FaTE aquatic food web
1 To provide the most accurate predictions possible with TRIM.FaTE, it is important to account for all of the plant and
fish biomass that might accumulate chemicals in each water body. The latter requires assigning measured fish
biomass densities for all fish species to the smaller number of fish compartments that are used in TRIM.FaTE to
represent different "trophic" groups of fish (e.g., species/sizes of fish that feed on the same general type/size of
foods in the same general environments - benthic and pelagic). For Ravena Pond, the fish harvest rate by humans
and wildlife would reduce concentrations in fish, as discussed in Section J-4.1.4.
J-9
-------
compartments (i.e., wildlife food types) described above. For each wildlife species, the
chemical concentration in each food type was multiplied by the average daily ingestion rate for
that food type. The resulting daily chemical intake rates for each food type were summed
across food types, and the total chemical intake was normalized to adult body weight to
estimate exposure dose (i.e., micrograms chemical ingested per gram of body weight per day,
|jg/g-day, equivalent to milligrams chemical per kilogram body weight per day, mg/kg-day).
J-3.2.3 Ecological Effects Assessment
Several sources were reviewed for wildlife toxicity reference values (TRVs) for Hg (expressed
either as total Hg or MeHg) and 2,3,7,8-TCDD; however, only one source proved to be useful
for the Ravena ERA:
• EPA's 1995 Great Lakes Water Quality Initiative (GLWQI) Criteria Documents for the
Protection of Wildlife: DDT; Mercury; 2,3,7,8-TCDD; PCBs (adequate documentation
of TRVs expressed as chemical dose);
• EPA's 2005 TRIM Ecotox Database (inadequate documentation - did not describe
quality of full database for each chemical; original toxicity values not included;
dosimetric scaling between experimental and wildlife species performed using body
weights that were not reported)
(http://www.epa.aov/ttn/fera/data/trim/ecotoxdatabaseDoc-Nov152005);
• EPA's 2005 Ecological Soil Screening Levels (http://www.epa.qov/ecotox/ecossl/) (no
values for Hg or TCDD);
• EPA's Region 9 Biological Technical Assistance Group (BTAG) Recommended
Toxicity Reference Values for Mammals/ for Birds (last revised 11/21/2002) (no
values for TCDD; Hg values based on EPA 1995 above);
• California Office of Environmental Health Hazard Assessment Ecotox Database,
developed in collaboration with the University of California at Davis
(http://www.oehha.org/cal ecotox/) (no values for TCDD or MeHg);
• The Risk Assessment Information System (RAIS) sponsored by the US Department
of Energy (DOE) Office of Environmental Management, Oak Ridge Operations
(ORO) Office, Ecological Benchmarks (http://rais.ornl.qov/homepaae/
benchmark.shtml), provides access to a large number of ecological benchmarks
(expressed as concentrations in soil, water, sediments, or biota), developed by
numerous state and federal agencies (no values expressed as doses or chemical
intake rates for wildlife); and
• US DOD's Wildlife Toxicity Assessment Program (http://chppm-
www.apqea.army.mil/erawq/tox/) (did not include Hg or TCDD).
We concluded that only the EPA 1995 GLWQI documents for Hg and 2,3,7,8-TCDD, with
modifications specified in EPA's 1997 Mercury Study Report to Congress (MeRTC), were
adequate for establishing and documenting TRVs, expressed as doses (chemical intake in
mg/kg-day) for wildlife in the Ravena ERA. The GLWQI documents represent the only source
that documented the available toxicity data at the time, why a study was selected as the critical
study upon which to base a reference dose, and which uncertainty factors (UFs) were needed
and what their values should be given specific limitations of the available database and the
critical study.
Concerns have been expressed that EPA's 1995 wildlife criteria may be overly conservative.
These concerns, however, center on the food chain model for the Great Lakes (e.g., Wolfe and
Norman 1998) from which the criteria expressed as water concentrations were back-calculated.
J-10
-------
Food chains modeled on data form the Great Lakes might predict more bioaccumulation than is
likely in smaller ecosystems with shorter food chains overall. The original critical toxicity studies
and uncertainty factors applied to those studies have been used widely by other state and
federal agencies. For the Ravena ERA, ICF is not using the Great Lakes food chain model;
instead it is using food webs developed specifically for the water bodies near the Ravena
facility.
Given the date of publication of the GLWQI wildlife criteria, ICF conducted a literature search
using keywords in online biobliographic databases for more recent information on the toxicity of
Hg and TCDD to wildlife. We found that a large proportion of recent wildlife ecotoxicity studies
have focused on Hg (few on TCDD) and on correlating wildlife tissue concentrations (including
chemical concentrations in mammalian fur and in bird feathers and eggs, as well as chemical
concentrations in dead or sacrificed animal liver, kidney, brain, muscle, and fat) with adverse
reproductive outcomes in field situations (e.g., Barr 1998, de Sorbo and Evers 2005, Evers et al.
2004, Evers and Reaman 1998, Heath and Frederick 2005, Hoffman et al. 1996, Mierle et al.
2000, Thompson 1996, Wolfe and Norman 1998, Wolfe et al. 1998). These studies are
intended to determine the utility of various wildlife species as monitors of environmental
pollution or to generate exposure-response relationships at the population level in the field. The
measures of exposure, tissue-specific chemical concentrations, in general were only weakly
associated with measures of chemical concentrations in potential prey species, limiting the utility
of the studies for relating chemical intake (dose) to effect levels.
ICF considered the option of using toxicity reference values for mammals and birds expressed
as tissue concentrations to compare to TRIM.FaTE-estimated tissue concentrations in wildlife.
We decided against that approach for several reasons.
• There are as yet no consensus TRVs based on wildlife tissue concentrations at a
federal level for Hg or 2,3,7,8-TCDD.
• One would need to determine which tissue concentrations would be most appropriate
for establishing a TRV; some effects data are related to kidney and liver
concentrations, which tend to be high, while other effects data are related to target
tissue concentrations (e.g., Hg in brain tissues), while still other effects data are
related to blood concentrations, which can be collected from wildlife without their
sacrifice. For birds, chemical concentrations in eggs often are related with egg shell
thinning, breakage, or hatching success.
• For a risk assessment that starts with emissions of chemicals to air, it would be
necessary to predict not only the uptake of chemicals from the environment, but also
the distribution of chemical to different organs in the bodies of birds and mammals.
That would require the addition of PBPK models for Hg and 2,3,7,8-TCDD in birds
and mammals.
Finally, for Hg and 2,3,7,8-TCDD, ICF used EPA's approach of using an inter-species
uncertainty factor of only 3 to account for unknown toxicokinetic and toxicodynamic differences
among different species of bird and mammal. The most often used alternative is to scale dose
between species on the basis of relative metabolic rate (body weight raised to approximately the
0.75 power). For bioaccumulative chemicals that accumulate over time in specific tissues,
however, metabolic rate may have only a minor influence on accumulated chemical residues in
tissues.
j-11
-------
J-3.2.4 Risk Characterization
A two-stage approach is used to characterize ecological risks from Hg and dioxin emissions
from the Ravena facility. In the first phase, hazard quotients (HQs) (exposure doses divided by
TRVs) are calculated by chemical for each of the five wildlife species and four water bodies.
HQs exceeding 1.0 indicate a potential for individual animals to be adversely affected by their
exposure.
For those scenarios where HQ values exceed 1.0, a preliminary evaluation of the potential for
population-level effects is conducted. Specifically, the maximum number of individuals of a
species for which the HQ exceeds 1.0 is estimated based on available data on population
densities or territory size as reported in available literature. If the maximum number of
individuals with an HQ greater than 1.0 is one or two, as could be the case for the small Ravena
Pond, then the threat of population-level effects would be considered to be negligible. However,
if the estimated HQ exceeds 1.0 for multiple individuals, we could not exclude the possibility of
population-level risks without further analysis..
J-3.3 Hydrogen Chloride
ICF conducted a proximity and vulnerability screening assessment for possible indirect effects
of HCI deposition on ecologically sensitive environments. Portland Cement facilities were
ranked according to emission rates, the pH of regional rainfall, surface water alkalinity, and
proximity to sensitive environments, as described below.
J-3.3.1 Facility Ranking
ICF conducted an initial ranking of all Portland Cement Facilities emitting HCI according to three
indicators of ecological risk: (1) "background" acid deposition (regional pH of rainfall) in the area
surrounding each facility, (2) surface water alkalinity (an indicator of acid buffering capacity or
resistance to changing pH), and (3) annual HCI emissions reported by each facility. The
background acid deposition and the surface water alkalinity were used as indicators of
ecosystem susceptibility to additional acid deposition. Annual HCI emissions were used as an
indicator of potential additional acid deposition due to Portland Cement facilities.
Regional pH of rainfall (i.e., background acid deposition) is one indicator of ecosystem
"susceptibility". The pH of rainfall shows regional patterns across the United States that have
resulted from multiple point and non-point sources of chemical precursors of acid rain that
change slowly over time. Areas subject to rainfall of relatively low pH already may be under
stress from acid deposition. At a minimum, the buffering capacity of ecosystems in areas of
highly acidic rainfall is likely to have been lowered from "natural" levels for the area.
ATTACHMENT J-1 Exhibit 1 provides a map of the pH of rainfall across the United States as
measured by the National Atmospheric Deposition Program, National Trends Network.
A second indicator of ecosystem susceptibility is its ability to buffer acid deposition as indicated
by measurements of surface water alkalinity in an area. Water alkalinity, which can be
expressed as mg of calcium carbonate and magnesium bicarbonate per liter (mg/L or ppm by
J-12
-------
weight) or simply milliequivalents of carbonate per liter (meq/L), is an indicator of the water's
ability to absorb hydrogen ions (H+) without changing pH. The carbonate and hydrogen ions
react to produce carbonic acid and then water and carbon dioxide. The higher the alkalinity, the
more acid must be added for a noticeable reduction in pH to occur. The alkalinity of surface
waters is due in large part to geological characteristics of the area, including the type of parent
materials that weather to soils and sediments. Thus, alkalinity can be considered an indicator of
the ecosystem sensitivity to acid deposition.
ATTACHMENT J-1 Exhibit 2 provides a copy of EPA's map of surface water alkalinity across
the United States. We use this map as an indicator of both surface water and soil alkalinity as
influenced by parent geological materials. Other factors can influence alkalinity as well (e.g.,
soil type and grain); however, for a national-scale screen, we consider this map adequate to
identify areas of likely low acid buffering capacity in surface waters and in soils.
To quantify a facility-specific indicator of exposure, ICF used NEI data on emission rates (in tons
per year) of HCI for all Portland Cement facilities.
The Portland Cement facilities were ranked according to the product of scores assigned to the
susceptibility and emission factors as described below. Each factor was scored on a scale of 1
to 5.
• Potential ecosystem susceptibility - background acid deposition. Background acid
deposition was assigned a score based on the measured pH of rainfall (log scale
maintained) using the map in
•
• ATTACHMENT J-1 Exhibit 1. Considering the range of pH of rainfall across the
United States, we used a scoring range of 1 to 5, with a score of 1 representing a pH
of < 4.5, a score of 2 representing a pH of > 4.5 and < 4.7, a score of 3 representing a
pH of > 4.7 and < 4.9, a score of 4 representing a pH of > 4.9 and <5.1, and a score
of 5 representing a pH of > 5.1.
• Potential ecosystem susceptibility - acid buffering capacity. EPA maps of surface
water total alkalinity (meq/L) are based on five alkalinity categories. We assigned
scores based on those categories: a score of 1 for areas with total alkalinity less than
50 meq/L; 2 for alkalinity of 50 to 100 meq/L; 3 for alkalinity of 100 to 200; 4 for
alkalinity of 200 to 400; and 5 for total alkalinity of surface waters greater than 400
meq/L.
• Facility emissions rate. Facility HCI emissions were ranked from 1 to 5 based on the
estimated HCI emission rates, with facilities in the top 20th percentile receiving a score
of 1 and those in the lowest 20th percentile receiving a score of 5.
Equation J-1 shows the calculation of a preliminary hazard ranking score for each facility.
Based on the individual factor scoring system above, the lower the hazard score, the higher the
possible ecological risks of adverse indirect effects of HCI deposition.
J-13
-------
Equation J-1:
Preliminary Hazard Score facility n = Rainfall pH facuityn x Surface Water Alkalinity facmy n
X Emissions facility n
J-3.3.2 Refined Facility Ranking
The next step to refine the ecological risk ranking for the facilities was to identify nearby
sensitive environments on maps of the area surrounding each facility and to score facilities on
the basis of distance to the sensitive environment. This step not only identified ecosystems of
potential concern, but it also allowed quantification of a proximity factor, another indicator of the
potential for exposure because of dispersal of contaminants in air with increasing distance from
an emission source.
Sensitive environments, as defined in EPA's Hazard Ranking System for potential Superfund
sites, may include areas such as wildlife refuges, national parks, waterfowl staging areas, water
bodies, and wetlands larger than 5 acres (EPA 1992). Because this step requires substantially
more effort than quantifying and scoring the susceptibility and emission factors, we intended to
conduct this step only for the ten facilities with the lowest preliminary hazard score (i.e., highest
preliminary risk ranking). Because several facilities received the same score, we used a
criterion of a preliminary hazard score of 20 or less to identify facilities for the sensitive
environment proximity assessment (total of thirteen facilities).
The distances between the thirteen facilities with a hazard score of 20 or less and sensitive
environments were estimated using GIS maps with data layers for sensitive environments
provided by Environmental Systems Research Institute, Inc. (ESRI) (ESRI 2006). Sensitive
environments included in the data layers are water bodies (e.g., canals, glaciers, lakes,
reservoirs, streams, swamps, and marshes); national, state, and local forests; and national,
state, and local parks. For this analysis, we first identified the nearest sensitive environment of
any type. Finding that about half of the environments identified in this way were extremely large
bodies of water (e.g., the Mississippi River, Lake Michigan) for which localized emissions of HCI
from Portland Cement facilities are not likely to affect pH, we reviewed those environments
again to find smaller bodies of water on which localized acid deposition might have an effect.
For consistency, we identified the nearest "small" water bodies for all thirteen facilities.
Using a measuring tool in MapWindowfor GIS maps, ICF measured the shortest distance
between each facility and the shore of the closest water body (excluding the Great Lakes and
large rivers). For three facilities, the distance to a nearby terrestrial sensitive environment (state
and national forests and a state park) also was measured. Finally, we assigned a proximity
score to each facility. The proximity score was equal to the square root of the distance between
the facility and the sensitive environment. Although we have not seen a precedent for this, we
selected the square root function to quantify this indicator on the basis of a simple conceptual
model of primarily horizontal chemical dispersion in all compass point directions with increasing
distance from a source. The score was rounded to one significant digit, and stopped at a top
score of 5. For example, the proximity factor for a separation of 4 km would be assigned a
score of 2; a separation of 10 km (square root = 3.2) would correspond to a score of 3; and a
separation of 23 km (square root of 4.8) would receive a score of 5, as would any separation
greater than 25 km.
ICF calculated a final composite hazard score for each of the thirteen facilities by multiplying the
susceptibility, emissions, and proximity scores (see Equation J-2). The final scores were ranked
to determine the facilities most likely to pose ecological risks, if any.
J-14
-------
Equation J-2:
Final Hazard Score facmyn = Rainfall pH facmyn x Surface Water Alkalinity faCiiity n x Emissions faCiiityn
x Proximity facimn
J-3.3.3 Exposure Assessment - Site-specific Data
For the HCI exposure assessment, the indirect effects of concern are mediated through
changes in soil or surface water pH. Predictions of the rate of HCI deposition required to
produce pH changes would require substantial site-specific information and model development.
ICF therefore considered existing measurements of soil and surface water pH values as an
indicator of the possibility of indirect adverse ecological effects resulting from acidification.
Note, however, that such measurements serve only as another screening tier. They do not
indicate the relative contribution of a Portland Cement facility and "background" regional acid
deposition to the measured pH values. An analysis of relative source contributions would be
warranted only if the screening criteria suggested non-negligible effects consistent with
acidification in the vicinity of a Portland Cement facility.
Originally, ICF intended to focus on only one or two facilities with the lowest composite hazard
score (highest likelihood of ecological effects) to search for local measurements of pH in soils or
surface waters to compare with pH ecotoxicity benchmarks or for reports of acidification or
adverse ecological effects in the vicinity of the facility. The types of nearby sensitive
environments, however, vary substantially for the thirteen facilities with the lowest composite
score (highest hazard). We therefore looked for localized data for all thirteen locations.
J-3.3.4 Terrestriai En vironments
For analysis of terrestrial environments, we defined "areas of interest" as the nearest boundary
of the sensitive environment to the respective Portland Cement facility. Each area of interest
was less than 60 acres. Measurements of soil pH and other parameters for the upper soil
layers in ecologically valued and protected areas (e.g., state parks) close to three Portland
Cement facilities were obtained from the U.S. Department of Agriculture's Web Soil Survey
(USDA 2008). The Web Soil Survey provides data from the National Cooperative Soil Survey,
which is a partnership of federal, regional, state, local and private agencies that provides
information about soils across the United States. Most data were collected over the past 40
years, and approximately 75 percent of the data are less than twenty years old. Soil data are
available for more than 95 percent of the counties in the United States. We found, however,
that the range of pH values reported for a single soil layer and soil type was sufficiently high (at
least 1 or 1.5 pH unit, and often more) that it is unlikely to be predictive of plant community
responses. This limitation is not surprising given that pH measurements are sensitive to
humidity, temperature, and other parameters that can vary seasonally and daily. We therefore
also considered the cation-exchange capacity measured for the surface soil layers in the area of
interest as a more precise indicator of local soil conditions. The higher the cation-exchange
capacity of a soil, the higher the soil buffering capacity, and the more resistant the soil is to
changes in pH with acid deposition.
A final line of evidence for the possibility of indirect ecological effects of HCI emissions from a
Portland Cement facility would be reports of adverse ecological effects in the vicinity of a facility
that are consistent with effects of soil or surface water acidification.2 For example, evidence of
2 The existing facilities have been emitting HCI against a background rate of acid deposition for many years.
Confirmation of ecological risks by reported observations of existing ecological impacts might, therefore, be
J-15
-------
a gradient of aluminum toxicity to plants with distance from a facility would be consistent with
excessive acidification of soils possibly due to HCI emitted by the facility. We searched state
departments of natural resources and other Internet sites for possible reports of adverse
ecological effects that might be related to acid deposition. In the absence of site-specific data,
we examined aerial photographs of the area surrounding a facility to determine if there was any
evidence visible in the photographs of adverse effects on vegetation in the vicinity of a facility.
For nearby surface waters, we searched for pH measurements in EPA's STORET Database
(EPA 2008). STORET is an operational data repository that is updated continuously with water
data for all states, territories, and jurisdictions in the United States. STORET contains water
quality data in addition to biological and physical data. When water quality data were not
available in STORET, ICF sought alternative data sources that are described in Section J-4.2.
J-3.3.5 Ecological Effects Assessment
To evaluate potential aquatic ecological effects associated with measured surface water pH, we
relied on EPA's documentation of its criteria for pH for freshwaters of the United States (EPA
1986).
To evaluate potential terrestrial ecological effects associated with measured soil pH values, we
considered recommendations for soil pH for maximizing plant growth, including consideration of
the moderately acid-tolerant native and agricultural plant species. Use of this type of
benchmark for soil pH assumes that the concentrations of other potentially toxic heavy metals
(e.g., aluminum) are not above the range of "background" levels that characterize most of the
United States.
To evaluate potential terrestrial community effects of measured cation-exchange capacity, we
used the U.S. Department of Agriculture's (USDA) classification of cation-exchange capacity.
J-3.3.6 Ecological Risk Characterization
Where local measurements of surface water or soil pH are available, they are compared to
EPA's ambient water quality criteria (AWQC) for pH for the protection of aquatic life (Section J-
4.2.1.1) or the soil pH benchmark (Section J-4.2.1.2). Where the local surface water or soil pH
measurements are below the ecotoxicity benchmarks or where there are observations of
adverse ecological effects in the vicinity of a facility, further investigation might be warranted.
We consider surface waters with measured pH values below the AWQC criterion of 6.5 to be at
risk of reduced biodiversity due to the loss of acid-sensitive species. We consider soils with pH
levels below 5.5 to possibly be at risk of reduced plant biodiversity (i.e., species restricted to the
more acid-tolerant groups), which might affect plant community structure. As described in
Section J-4.2.1.2, we consider soils with pH values above 6.0 at negligible risk of indirect
adverse effects from existing acid deposition. In between pH 5.5 and 6.0, additional lines of
evidence are needed before one would conclude that some level of adverse effect is possible.
As noted in Section J-3.3.3 above, pH values below (more acidic than) the pH benchmarks or
observations of adverse effects consistent with acidification in the vicinity of a Portland Cement
facility do not indicate whether or to what extent the effects are attributable to releases of HCI
from the facility compared to regional acid deposition. Attribution of effects to a facility would
possible. In particular, any adverse effects that decrease with increasing distance from the facility and that are
consistent with effects associated with acidification might be due to HCI emissions from the facility.
J-16
-------
require additional lines of evidence, such as a gradient of decreasing adverse effects on plants
with increasing distance from the facility. Exposure-response evidence of this type would not
necessarily identify HCI alone as the chemical causing affects (it could be HCI in combination
with some heavy metals released in lower quantities), but it would strongly suggest the facility
as the source. A relative-source-attribution analysis is warranted only if the screening data
indicate that adverse ecological effects might occur or be occurring in the vicinity of a facility.
J-4 Results
Detailed results are presented separately for mercury (Hg) and dioxins (2,3,7,8-TCDD) (Section
J-4.1) and hydrogen chloride (HCI) (Section J-4.2).
J-4.1 Results for Mercury and Dioxins
This section includes the results of the exposure assessment (Section J-4.1.1), the dose-
response analysis for Hg and 2,3,7,8-TCDD (Section J-4.1.2), and risk characterization (Section
J-4.1.3). Key data and model limitations and uncertainties of the ecological risk assessment
(ERA) for Hg and TCDD are discussed in Section J-4.1.4.
J-4.1.1 Exposure Assessment
An exposure assessment was performed for "individuals" of each of the four wildlife species of
concern to calculate daily doses of Hg and TCDD from ingestion of prey obtained from surface
water bodies near the Ravena facility. Exposure doses were calculated for each wildlife species
and each water body modeled for the Ravena facility assuming that an individual animal
obtained all of its food from the single water body. This required a characterization of each
species' mean body weight, an assumed diet for the TRIM.FaTE modeled food web
compartments, and food ingestion rates.
TRIM.FaTE estimates of concentrations of Hg and 2,3,7,8-TCDD in each prey type were
multiplied by daily prey ingestion rates to estimate the daily intake for each chemical for each
wildlife species and water body (i.e., Ravena Pond near the facility, Alcove Reservoir,
Kinderhook Lake, and Nassau Lake). The chemical intakes (doses) were normalized to body
weight (i.e., mg chemical ingested per kg animal body weight per day). Exposure estimates for
individuals of the same wildlife species varied by location because of differences in several
factors across the water bodies (described in Appendix I), including:
• water and sediment chemical concentrations estimated by TRIM.FaTE;
• the food webs constructed for each water body (food chains were shorter and fewer
fish compartments were included in the small pond relative to the three large water
bodies based on our experience in parameterizing aquatic food webs for use in
TRIM.FaTE for case studies in Maine, New York, and the RTR screening scenario);
• total biomass of fish assumed for each water body relative to the volume of water in
the system; and
• the distribution of fish biomass across the fish compartments as estimated from local
fish surveys of all water bodies except the small pond.
Construction of the aquatic food webs was described briefly in Section J-3.2.1 and in detail in
Appendix I. The exposure assumptions used for each of the avian and mammalian wildlife
species are described below.
J-17
-------
J-4.1.1.1 Exposure Assumptions
For each wildlife species, we identified values for three exposure factors: (a) adult body weight,
(b) the percentage of total diet obtained from the food types included in the TRIM.FaTE model,
and (c) an estimated daily ingestion rate for each consumed food type. For (a) adult body
weights, we used mean values reported in the literature for populations closest to New York.
For (b), we used reviews of the dietary habits of wildlife species recently prepared by ICF for
EPA's Office of Water, but not yet published. For these data, we cite original sources here
rather than the secondary Office of Water documents. We emphasize that prey selection and
dietary habits of different wildlife species vary with location, time of year, habitat, relative
abundance of different prey species, breeding status of both predators and prey, and other
factors. Nonetheless, some attributes of diet composition are fairly common for some of the
more specialized predators, including swallows and piscivorous wildlife, as discussed for each
species below.
One of the most important considerations in modeling bioaccumulation of chemicals through
food chains is the size of fish consumed, which loosely corresponds to trophic level depending
on the species of fish and their feeding habits. There generally are limits to the size of fish that
can be captured and swallowed by avian wildlife that swallow their prey whole (e.g., merganser,
swallows). For wildlife that can consume larger fish by tearing pieces off while standing on land
(e.g., mink, eagle), the distribution offish sizes consumed depends on fish availability and
population age/size class structure, the size offish in habitats fished by the wildlife, and in the
agility of the fish in escaping capture compared with the abilities of the wildlife species. We
therefore evaluated available data on the size of fish consumed. For species that consume both
aquatic and terrestrial prey (e.g., mink, bald eagle) in many locations, we conservatively
assumed 100 percent consumption of aquatic prey.
For (c) we intended the food ingestion rates to represent an annual average ingestion rate for a
free-living animal rather than a breeding-season-only ingestion rate, even though the bird
species are likely to migrate away from the site during the winter (particularly the swallow). We
selected the measurement or estimate of food ingestion rates that we judged most likely to
represent a free-living metabolic rate (FMR) averaged across all seasons.
The exposure assumptions for the four wildlife species evaluated for the Ravena facility site-
specific ecological risk assessment are described below.
Tree Swallow
Body weight. The mean weight of 82 birds of both sexes captured at the Powdermill Nature
Center in Pennsylvania during spring migration was 20.1 ± 1.58 grams (g), with a range of 15.6
to 25.4 g (Dunning 1984,1993, citing unpublished data by the PNC). The mean body weight for
12 birds was 21.6 ± 1.9 g (Williams 1988). The first mean value (i.e., 20.1 g) is used to
represent tree swallows throughout the year.
Diet composition. As aerial insectivores, tree swallows consume virtually 100 percent small
flying insects, including adult midges, mosquitoes, mayflies, and other groups with aquatic larval
forms (Quinney and Ankney 1985). For this ERA, we assumed that 100 percent of the insects
consumed by swallows had been aquatic nymphs in the water body under consideration.
Food ingestion rate. Using doubly labeled water to study free-living (field) metabolic rates
(FMRs) in tree swallows in New Brunswick Canada, Williams (1988) found that incubating
J-18
-------
females required 118.9 ± 9.3 kiloJoules per day (kJ/d) (mean ± SD; n=9; average body weight
[BW] 22.6 g), or 1.3 kcal/g BW-d. Females feeding young exhibited higher energy requirements
and lower body weights: 128.3 ± 21.3 kJ/d for females with three young (n=5; average adult
female body weight 18.8 ± 2.0 g) and 136.4 ± 15.6 kJ/d for females with five young (n=11;
average adult female body weight 19.4 ±1.2 g). Those daily energy expenditures equal 1.6 and
1.7 kcal/g BW-d, respectively. Williams (1988) noted that the FMR for aerial-feeding
insectivorous passerines, such as swallows, is higher than the FMR for ground- or tree-feeding
insectivorous passerine birds, such as sparrows, of similar size. We estimate from their data
and discussion that the FMR for 20 g swallows is perhaps as much as 33 percent higher than
for non-aerial feeding passerines of similar size.
To estimate an FMR more in keeping with a year-round food ingestion rate for chronic
exposures, we used Nagy et al.'s (1999) allometric equation for passerine birds [FMR (kJ/d) =
10.4 * BW(g)0 64 = 79.8 kJ/d for tree swallows weighing 20 g]. Note that measured FMRs (using
doubly labeled water) for barn swallows (95.8 kJ/d for 20.4 g bird) and house martins (79.8
kJ/day for 19 g bird) cited by Nagy et al. (1999) are similar to the allometric estimate for tree
swallows. An FMR of 79.8 kJ/d equals 19.1 kcal/d, or 0.96 kcal/g BW-d.
The FMR estimated using the allometric equation for passerine birds from Nagy et al. (1999) is
about 26 percent less than the mean value of 1.3 kcal/g BW-d measured for incubating females
(males take over incubation for short periods to allow the females to feed) and 44 percent less
than the mean value of 1.7 kcal/g BW-d measured for females feeding five nestlings (Williams
1988). Those observations are consistent with Williams estimate that swallows require
approximately a third more calories per day to forage for food than do ground- or tree-foraging
insectivorous passerines. The weather during the field study in New Brunswick was cool and
moist, possibly requiring more energy for thermoregulation that would be required in New York.
Nonetheless, to be conservative, we use the measured FMRs instead of the allometric-model
estimate of FMR. Wejudge that the FMR during incubation is likely to be somewhat lower than
an annual average and that the FMR when feeding a clutch of 5 young is probably substantially
higher than an annual average energy requirement (not considering migration). We therefore
use an FMR of 1.4 kcal/g BW-d as the energetic requirement for tree swallows in this ERA.
To estimate an insect ingestion rate on a wet-weight basis, ICF used the procedure
recommended in EPA's (1993) Wildlife Exposure Factors Handbook. Our estimate of a wet-
weight insect ingestion rate for tree swallows of 1.33 g/g BW-day is based on the following
assumptions:
• tree swallow body weight - 20.1 g/bird (Dunning 1984, 1993);
• energetic requirement is 1.4 kcal/g BW-day (see discussion above);
• gross energy (GE) content of insects - 22.09 kJ/g dry weight (5.28 kcal/g dry weight)
(Bell 1990);
• water content of insects - 72.5 percent (midpoint of range of 70 to 75 percent) (Bell
1990); and
• enerqy assimilation efficiency (AE) for birds consuminq insects - 72 percent (USEPA
1993a, Table 4-3).
Common Merganser
Body weight. Adult male common mergansers typically are heavier than adult females;
however, not all investigators report weights separately for the sexes. In addition, Feltham
J-19
-------
(1995) noted that although female mergansers were smaller than males, and tended to have
higher FMRs, those differences were not significant.
There is some seasonal variation in body weights. Anderson and Timken (1972) found that as
winter temperatures in South Dakota, Minnesota, and Oklahoma became colder, the average
body weights of mergansers increased. The body weight assumed for the Ravena ERA is 1.27
kg, which is the mean body weight of 124 adults and juveniles of both sexes measured in winter
in Michigan (Salyer and Lagler 1940).
Diet composition. Common mergansers typically forage in the shallower parts of large water
bodies (e.g., lower reaches or mouths of rivers), moving to the middle reaches as the slower
moving waters freeze over in winter (Salyer and Lagler 1940). They typically locate their prey
by swimming on the surface and half-submerging their heads to look underwater (White 1937).
They then pursue and capture their prey during short (10 to 20 second) dives (Salyer and Lagler
1940). In very shallow water, they sometimes feed by probing under rocks and sticks while
partially submerged (Salyer and Lagler 1940).
The diet of common mergansers varies with local abundance of prey (Timken and Anderson
1969, White 1937). Several studies comparing fish availability with the composition of common
merganser diets suggest that the birds consume the most abundant of the suitably sized
available prey (White 1957, Huntington and Roberts 1959, Latta and Sharkey 1966, Sjoberg
1988, and McCaw et al. 1996 as cited in Mallory and Metz 1999).
For the exposure assessment, ICF used a review of the available literature to develop
assumptions concerning the diet composition of common mergansers for this ERA. We used
data, summarized in Exhibit 4-1, on the length distribution offish reported caught in Michigan by
Alexander (1977), with some consideration of studies from other locations (e.g., White 1936,
1967 and Huntington and Roberts 1959) and experimental choice studies (Latta and Sharkey
1966).
Exhibit 4-1. Distribution of Length of Fish Consumed by Common Mergansers in Michigan
(Alexander 1977)
Measure
Length of Fish (inches)
1
2
3
4
5
6
7
8
9
10-13
Number of Fish
Consumed
77
65
50
45
27
16
23
19
21
6
Percentage
(n = 349)
22%
19%
14%
13%
8%
5%
7%
5%
6%
<2%
Latta and Sharkey (1966) reported that the largest captive merganser (1.7 kg) could consume a
trout with a girth of 15.8 cm, while the smallest merganser (0.94 kg) could swallow trout with
girth of up to only 12.5 cm. Offering six mergansers a total of 25 trout, 5 in each of five trout-
size categories between 9.9 and 21 cm in length (that is 20 percent of the total trout in each size
category), Latta and Sharkey (1966) found that the larger prey were captured less often than
expected on the basis of their relative abundance: 28 percent of all trout consumed (N = 130)
were from 9.9 to 11 cm (approximately 4 inches); 28 percent were from 12 to 13 cm
(approximately 5 inches); 24 percent were from 15 to 16 cm (approximately 6 inches); 15
percent were from 17.5 to 18.3 cm (7 inches), and only 5 percent were from 20 to 21 cm
(approximately 8 inches), although 20 percent of the trout offered were in that size range.
J-20
-------
White (1936,1937) found that common mergansers in Nova Scotia consumed tomcod and
rainbow smelt that averaged 21 (8.3 inches) and 18 cm (7.1 inches), respectively, although
some tomcod up to 27 cm (10.6 inches) were consumed. Huntington and Roberts (1959) found
that 67 percent of 344 fish consumed by common mergansers in New Mexico were less than 10
cm (4 inches) in length; 84 percent were less than 20 cm (8 inches) in length, 94 percent were
less than 24 cm (10 inches) in length. Less than 1 percent were greater than 30.5 cm (12
inches). The majority of the fish consumed (86 percent) were gizzard shad.
Given the fish compartments modeled in TRIM.FaTE, we assumed that medium-sized
benthivorous fish and water column planktivorous fish (e.g., shiners 4 inches or less) were each
35 percent of the mergansers' total diet (on a wet-weight basis). Medium-sized "panfish" (water
column omnivores 5 to 10 inches) were assumed to comprise 25 percent of the diet. Finally, we
allowed 5 percent of the diet to be water column piscivorous fish (e.g., largemouth bass) greater
than 10 inches to account for the larger fish consumed by common mergansers.
Food ingestion rate. Based on analyses of stomach contents and observed feeding rates,
Salyer and Lagler (1940) estimated that American mergansers consume fish at a rate of
between one third and one half of their body weight daily during winter in Michigan (0.33 to 0.50
g/g BW-day). Alexander (1977) also estimated a food ingestion rate of 0.33 g/g BW-day for
mergansers consuming fish in Michigan. Gooders and Boyer (1986) estimated that mergansers
consume an average of 445 g/d, or more than 0.33 g/g BW-day. Feltham (1995) used the
doubly labeled water technique to demonstrate that males and females of M.m. merganser
released on Scottish Rivers required 522 g and 480 g of food, equivalent to 0.32 and 0.40 g/g-
day, respectively. Latta and Sharkey (1966) found that 8 captive common mergansers
consumed between 0.183 and 0.257 g/g-day (mean ± SD of 0.208 ± 0.035 g/g-day). Based on
these studies, we assumed a fish consumption rate of 0.33 g/g-day for the Ravena ERA.
Bald Eagle
Body weight. As for most raptors, female bald eagles typically weigh more than males. Snyder
and Wiley (1976) reported a mean weight of 5.35 kg for 37 female and 4.13 kg for 35 male bald
eagles. The adult body weight assumed for the Ravena ERA was 4.5 kg, which the average
body weight for males and females combined used by Stalmaster and Gessaman (1984) and
Craig et al. (1988) in their studies of bald eagle FMR.
Diet composition. Our assumption for the composition of the diet of bald eagles in the Ravena
area is based on a review primary data sources (e.g., Dunstan and Harper 1975, Bowerman
1993, Grubb and Hensel 1978, Kozie and Anderson 1991, Todd et al. 1982) summarized in
USEPA's draft Trophic Level and Exposure Analyses for Selected Piscivorous Birds and
Mammals, Volumes 2 and 3 (USEPA 2005b). Overall, we assumed the diet to consist of 80
percent fish and 20 percent ducks. Further assumptions about the composition of the fish diet
were based on length offish documented by Bowerman (1993) for eagles in Michigan and
Watson et al. (1991) for eagles in the Columbia River estuary. As reported in these studies, fish
species consumed tend to include slow-moving benthivores, particularly suckers and catfish, as
well as gizzard shad and carp, which generally are herbivores/detritivores. The high end of the
proportion of piscivorous fish included in the bald eagle diet among the studies reviewed is 30
percent. Often, gizzard shad or other slow-swimming, lower trophic level species predominate.
For the Ravena water bodies, we did not include the latter two species. The pelagic species
caught by eagles can include salmonids, pike, and bass. For the Ravena ERA, we assumed
the following breakdown of fish for the diet in addition to 20 percent ducks: 28 percent
benthivores, 28 percent water column omnivores, and 24 percent water column carnivores
J-21
-------
Food ingestion rate. Food ingestion rates were estimated separately for free-living adult bald
eagles with diets consisting of 100 percent fish or 100 percent birds (i.e., mallards). Because
birds have more metabolizable energy per unit wet weight than do fish, the eagle's caloric
needs may be met with a smaller ingestion rate of birds than of fish (USEPA 2005b). The
separate ingestion rates for fish and birds were then used to calculate a fresh-weight ingestion
rate for bald eagles for a combined diet of 80 percent fish and 20 percent mallard on a wet-
weight basis.
The ingestion rate for a diet consisting entirely of fish is based on similar results obtained by two
separate research teams. Stalmaster and Gessaman (1984) observed captures of pre-weighed
salmon provided at artificial feeding stations in Washington State. Although the eagles may
have fed elsewhere on occasion, Stalmaster and Gessaman (1984) believed that the feeding
stations provided most of the eagles' intake. They estimated the adult (including both sexes)
ingestion rate to be 0.12 g/g-day. Craig et al. (1988) obtained the same estimate for bald eagles
in Connecticut. Both research teams assumed the eagles weighed approximately 4.5 kg.
No feeding studies were available for bald eagles consuming waterfowl. Therefore, we
estimated a mallard ingestion rate using the procedure and assumptions recommended in
EPA's (1993a) Wildlife Exposure Factors Handbook. The assumptions included a GE content
of 1.2 kcal/g fish and 2.0 kcal/g bird wet weight (EPA 1993a, Table 4-1) and a general energy
AE (EPA 1993a, Table 4-3) of 79 percent for birds consuming fish and 78 percent for birds
eating other birds. We then could estimate the average metabolizable energy (ME) for fish and
birds in Equation J- below.
Equation J-3:
ME = GE x AE
MEnsh = 1.2 kcal/g fish wet wt x 0.79 = 0.95 kcai ME/g fish wet wt
MEmaiiard =2.0 kcai/g bird wet wtx 0.78= 1.56 kcai ME/g bird wet wt
Next, the average ME was calculated for a diet of 80 percent fish and 20 percent mallard:
MEaverage = (0.8 x 0.95 kcal/g) + (0.2 x 1.56 kcal/g)
= 1.07 kcai ME/g combined diet wet wt
A total food ingestion rate (FIR) then could be estimated from the MEaVerageand FMR estimated
with data from Stalmaster and Gessaman (1984), normalized to body weight (4.5 kg), using
EquationJ-4 below.
Equation J-4:
FIRtotal = FMR / MEaverage
FIRtotai = (0.114 kcal/g BW-day) / (1.07 kcai ME/g food wet wt)
FIRtotai = 0.1065 g/g-BW-day wet weight
Assuming a body weight of 4.5 kg, an adult eagle is estimated to consume a total of 480 g/d
fresh food (FlRtotai = 1 -065 g/g-day x 4.5 kg x 1000 g/kg = 480 g/d). Still assuming that the
combined diet includes 80 percent fish and 20 percent mallard, the daily ingestion rates for each
food type are calculated as follows:
J-22
-------
FIRfish = 480 g/d x 0.8
= 384 g/d
FIRmallard = 480 g/d X 0.2
= 96 g/day
Mink
Body weights. Mink body size varies greatly throughout the species' range (adult males
reaching no more than 1.4 kg in the east but up to 2.3 kg in the west of the United States
according to Harding 1934 as cited in Linscombe et al. 1982). Males weigh markedly more than
females (in some populations, almost twice as much). Mitchell (1961) reported mean body
weights for wild mink captured in summer in Montana as 0.55 kg for females (n = 25) and 1.04
kg for males (n = 5). We averaged these mean male and female body weights to estimate a
mink body weight of 0.8 kg for use in the Ravena ERA.
Diet composition. We reviewed the diet composition of mink reported on a wet weight basis by
Alexander (1977) for southern Michigan year-round and by Sealander (1943) for the same area,
but in the winter only. The other studies of mink diets summarized by EPA (1993b, 2005b) were
based on measurements of remains in scats, which provide a poor indication of the proportion
of diet on a wet-weight basis. In addition, we considered southern Michigan to be an adequate
surrogate location for New York. Based on the two Michigan mink studies and the assumption
of 100 percent aquatic prey, we specified the following diet for the Ravena ERA:
• 24 percent benthic invertebrates (e.g., crayfish),
• 25 percent medium-sized, benthivorous fish (e.g., small bullheads),
• 1 percent benthic carnivores (e.g., eels - captured in proportion to their relative
biomass density in the Ravena water bodies),
• 25 percent small water column planktivorous fish, and
• 25 percent medium-sized "panfish" (water column omnivores).
Mink generally are not fast enough to capture the larger game fish. In reality, mink generally
consume some of their diet from terrestrial sources; the diet specified above will introduce a
conservative bias that could be reexamined later. In addition, mink are likely to include
amphibians in their diet.
Food ingestion rates. Studies of captive mink indicate that mink eat at least 12 to 16 percent of
their body weight in food daily. Assuming a body weight 1.4 kg (e.g., for a male mink) and
Cowan et al.'s (1957) food-ingestion model derived from measures of prey consumed by captive
mink, Arnold and Fritzell (1987) estimated that mink require 180 g/day fresh prey. Normalized
to body weight, that food ingestion estimate is equivalent to 0.13 g/g-day. Bleavins and Aulerich
(1981) measured food ingestion rates of farm-raised mink provided a diet of whole chicken (20
percent), commercial mink cereal (17 percent), ocean fish scraps (13 percent), beef parts,
cooked eggs, and powdered milk. The moisture content of the feed overall was 66.2 percent.
On this diet, the food consumption rate of female mink was 0.16 ± 0.0075 SE g/g-day, and that
of male mink was 0.12 ± 0.0048 SE g/g-day (Bleavins and Aulerich 1981). Farrell and Wood
(1968) documented how food requirements of mink depended on their activity level. Mink
maintained in small cages required 20 kcal/100 g of body weight compared with 26 kcal/100 g
body weight when the same animals were housed in larger ranch-type cages (Farrell and Wood
1968).
J-23
-------
Nagy's (1987) allometric equation for estimating an FMR for non-herbivorous mammals predicts
that a mink weighing 0.8 kg would require 196 kcal/d. Assuming that mink prey entirely on fish,
that fish provide 1.2 kcal/g GE wet weight (EPA 1993a, Table 4-1), and that mink consuming
fish exhibit an AE of 91 percent (EPA 1993a, Table 4-3), the fish ingestion rate of a male mink
weighing 1.04 kg would be 0.22 g/g-day and of a female mink weighing 0.55 kg would be 0.24
g/g-day. For an average adult of 0.8 kg, the fish ingestion rate would be 0.23 g/g-day.
Assuming that mink prey entirely on benthic invertebrates, that crayfish provide the same GE as
shrimp (1.1 kcal/g wet weight; EPA 1993, Table 4-1), and that the AE is similar to mammals
consuming insects (87 percent; EPA 1993, Table 4-3), the invertebrate ingestion rate would be
very similar rounded to two significant digits, or 0.24 g/g-day. With a diet of 75 percent fish and
25 percent benthic invertebrates, we set a food ingestion rate for mink in the Ravena ERA of
0.23 g/g-day.
Assuming that wild mink are likely to be more active than captive mink, the higher food ingestion
rates estimated from Nagy's (1987) allometric equation appear to be more appropriate for
wildlife exposure analyses than the food ingestion rates measured for captive animals.
Summary of Exposure Assumptions
Exhibit 4-2 through Exhibit 4-6 summarize the exposure assumptions developed above for the
four wildlife species included in the Ravena ERA. These assumptions were used to estimate
daily intake rates of Hg and TCDD by each species.
Average adult body weight and food ingestion rates are summarized in Exhibit 4-2. Because
the averages are for adults of both sexes, exposure estimates calculated with these
assumptions may be over- or under-estimated by gender for sexually dimorphic species. For
example, adult male mink are markedly larger than females, and adult female bald eagles are
larger than males. Regional variation in body weight among different populations also may
mean that the body weights assumed for the Ravena ERA are over- or under-estimated.
Exhibit 4-2. Annual Mean Adult Body Weights and Food Ingestion Rtes Assumed for
Wildlife Species
Species
Mean Adult Body Weight (kg)
Food Ingestion Rate (g/day)
Tree Swallow
0.0201
26.9
Common Merganser
1.27
419.1
Bald Eagle
4.5
179.6
Mink
0.8
478.3
Exhibit 4-3 summarizes the percentages of each species' diet assumed for each of the nine
TRIM.FaTE aquatic food web compartments in Alcove Reservoir, Nassau Lake, and Kinderhook
Lake. As discussed above, these assumptions are based on published reports of feeding
behavior of free-living animals, and do not account for regional and seasonal variations (e.g.,
due to local variations in prey abundance).
Exhibit 4-3. Fraction Diet Assumptions for Wildlife Feeding from Alcove
Reservoir, Nassau Lake, and Kinderhook Lake
Food Type
Tree
Swallow
Common
Merganser
Bald Eagle
Mink
Algae
0%
0%
0%
0%
J-24
-------
Macrophytes
0%
0%
0%
0%
Benthic Invertebrates
100%
0%
0%
24%
Benthivorous Fish
0%
35%
28%
25%
Bottom-feeding Carnivores
0%
0%
0%
1%
Water Column Planktivores
0%
35%
0%
25%
Water Column Omnivores
0%
25%
28%
25%
Water Column Carnivores
0%
5%
24%
0%
Mallard Duck
0%
0%
20%
0%
All Food Types
100%
100%
100%
100%
Because Ravena Pond is smaller than the other water bodies, its food web was shortened to be
consistent with available literature. Specifically, bottom-feeding carnivores and water column
omnivores were not included in the Ravena Pond food web. Because these food types are not
available to wildlife species that feed at Ravena Pond, the percentages of available food types
in each species' diet were scaled upward so that the percentages would sum to 100 percent.
Exhibit 4-4 displays the diet compositions for species for feeding at Ravena Pond.
Exhibit 4-4. Fraction Diet Assumptions for Wildlife Feeding from Ravena Pond
as Modeled in TRIM.FaTE
Food Type
Tree
Swallow
Common
Merganser
Bald Eagle
Mink
Algae
0%
0%
0%
0%
Macrophytes
0%
0%
0%
0%
Benthic Invertebrates
100%
0%
0%
32%
Benthivorous Fish
0%
47%
43%
34%
Bottom-feeding Carnivores
-
-
-
-
Water Column Planktivores
0%
47%
0%
34%
Water Column Omnivores
-
-
-
-
Water Column Carnivores
0%
6%
37%
0%
Mallard Duck
0%
0%
20%
0%
All Food Types
100%
100%
100%
100%
The diet composition percentages in Exhibit 4-3 and Exhibit 4-4 were multiplied by total daily
food ingestion rates in Exhibit 4-2, to estimate daily ingestion rates for each of the nine food
types. These ingestion rate estimates are presented in Exhibit 4-5 for Alcove Reservoir,
Kinderhook Lake, and Nassau Lake, and in Exhibit 4-6 for Ravena Pond.
Exhibit 4-5. Estimated Average Daily Ingestion Rate of Each Food Type in
the Diets of Wildlife Species from Alcove Reservoir, Nassau Lake, and
Kinderhook Lake (g/day)
Food Type
Tree
Swallow
Common
Merganser
Bald Eagle
Mink
Algae
-
-
-
-
Macrophytes
-
-
-
-
Benthic Invertebrates
26.9
-
-
43.1
Benthivorous Fish
-
146.7
133.9
44.9
J-25
-------
Bottom-feeding Carnivores
--
--
--
1.8
Water Column Planktivores
--
146.7
--
44.9
Water Column Omnivores
--
104.8
133.9
44.9
Water Column Carnivores
--
21.0
114.8
--
Mallard Duck
--
--
95.7
--
All Food Types
26.9
419.1
478.3
179.6
Exhibit 4-6. Estimated Average Daily Ingestion Rate of Each Food Type in
the Diets of Wildlife Species from Ravena Pond (g/day
Food Type
Tree
Swallow
Common
Merganser
Bald Eagle
Mink
Algae
-
-
-
-
Macrophytes
-
-
-
-
Benthic Invertebrates
26.9
-
-
58.2
Benthivorous Fish
-
197.0
206.0
60.7
Bottom-feeding Carnivores
-
-
-
-
Water Column Planktivores
-
197.0
-
60.7
Water Column Omnivores
-
-
-
-
Water Column Carnivores
-
25.1
176.6
-
Mallard Duck
-
-
95.7
-
All Food Types
26.9
419.1
478.3
179.6
J-4.1.1.2 Exposure Concentrations
Exhibit 4-7 through Exhibit 4-10 present estimated exposure concentrations for Hg and TCDD in
each of the nine biotic and two abiotic compartments of the TRIM.FaTE aquatic food web in
year 50 of the Ravena TRIM.FaTE simulation. For Hg, results are presented separately for
methyl mercury (MeHg) and divalent mercury (Hg+2) because it is the methylated form that
bioaccumulates (i.e., is not readily eliminated from animals). The MeHg fish tissue
concentrations associated with the mean emissions rate of Hg from the Ravena facility are
presented in Exhibit 4-7, and the analogous data for Hg+2 are presented in Exhibit 4-8.
Exhibit 4-7. Concentrations (pg/g) of Methyl Mercury in Compartments of the
TRIM.FaTE Aquatic Food Web at Year 50 - Based on Mean Measured Annual Hg
Emission Ratea
Compartment
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Surface Water
4.8E-07
2.0E-09
2.5E-09
2.6E-09
Sediment
1.9E-03
1.3E-05
1.5E-05
1.9E-05
Algae
1.7E-02
7.0E-05
8.6E-05
9.1E-05
Macrophytes
9.6E-07
4.0E-09
4.9E-09
5.2E-09
Benthic Invertebrates
5.9E-03
4.0E-05
4.8E-05
6.1E-05
Benthivorous Fish
3.0E-02
1.4E-04
2.2E-04
1.8E-04
Bottom-feeding Carnivores
N/A
4.9E-04
7.3E-04
6.6E-04
Water Column Planktivores
5.8E-02
6.3E-05
8.1E-05
7.4E-05
J-26
-------
Water Column Omnivores
N/A
1.9E-04
2.9E-04
2.9E-04
Water Column Carnivores
1.7E-01
5.6E-04
1.1E-03
8.7E-04
Mallard Duck
6.5E-03
4.4E-05
5.3E-05
6.8E-05
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the
Ravena Pond food web.
a Concentrations in surface water are in mg [MeHg]/L [water], which is equivalent to pg [MeHg]/g [water].
Concentrations in bulk sediment are in pg [MeHg]/g [sediment] dry weight. Concentrations in biota are pg
[MeHg]/g [biotal] wet weight.
Exhibit 4-8. Concentrations (|jg/g) of Divalent Mercury in Compartments of the
TRIM.FaTE Aquatic Food Web at Year 50 - Based on Mean Measured Annual Hg
Emission Ratea
Compartment
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Surface Water
1.1E-04
1.4E-07
1.6E-07
2.1E-07
Sediment
9.1E-01
6.1E-03
7.2E-03
9.3E-03
Algae
4.9E-01
5.9E-04
6.9E-04
9.0E-04
Macrophytes
5.3E-05
1.3E-07
9.1E-08
1.4E-07
Benthic Invertebrates
4.6E-02
3.2E-04
3.7E-04
4.8E-04
Benthivorous Fish
1.6E-02
1.0E-04
1.3E-04
1.6E-04
Bottom-feeding Carnivores
N/A
8.0E-05
9.4E-05
1.2E-04
Water Column Planktivores
1.5E-01
1.5E-04
1.7E-04
1.9E-04
Water Column Omnivores
N/A
9.0E-05
1.1E-04
1.4E-04
Water Column Carnivores
3.5E-02
6.7E-05
4.5E-05
9.2E-05
Mallard Duck
3.1E-02
2.1E-04
2.5E-04
3.2E-04
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the
Ravena Pond food web.
a Concentrations in surface water are in mg [Hg2+]/L [water], which is equivalent to pg [Hg2+]/g [water].
Concentrations in bulk sediment are in pg [Hg2+]/g [sediment] dry weight. Concentrations in biota are pg
[Hg2+]/g [biotal] wet weight.
MeHg and Hg+2 exposure concentrations (Exhibit 4-7 and Exhibit 4-8, respectively) are
presented in micrograms of MeHg per gram of the food type in wet weight (|jg/g ww). For
example, the concentration of MeHg estimated for mallard ducks feeding in Ravena Pond of
0.0065 |jg/g wet weight represents the average whole-body concentration of MeHg in an
individual duck.
As described in Section J-3.1, dioxin emissions monitoring data were not available for the
Ravena facility. Therefore, mean and 95-percent UCL emission rates were using emissions
factors based on clinker capacity and process type. Whole-body fish tissue and abiotic media
concentrations estimated for the mean and 95-percent UCL emissions estimates from year 50
of the Ravena TRIM.FaTE simulation are presented in Exhibit 4-9 and Exhibit 4-10,
respectively.
J-27
-------
Exhibit 4-9. Concentrations (|jg/g) of 2,3,7,8-TCDD in Compartments of the
TRIM.FaTE Aquatic Food Web at Year 50 with Mean Emission Ratea
Compartment
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Surface Water
5.3E-12
1.1E-14
3.8E-14
2.8E-14
Sediment
2.6E-08
1.6E-10
5.4E-10
4.0E-10
Algae
3.4E-09
7.1E-12
2.4E-11
1.8E-11
Macrophytes
7.7E-08
1.7E-10
5.6E-10
4.1E-10
Benthic Invertebrates
2.5E-09
1.6E-11
5.8E-11
4.5E-11
Benthivorous Fish
1.9E-06
8.9E-10
7.8E-09
1.5E-09
Bottom-feeding Carnivores
N/A
4.6E-09
3.9E-08
8.6E-09
Water Column Planktivores
3.6E-07
8.8E-11
3.4E-10
3.0E-10
Water Column Omnivores
N/A
1.3E-09
1.2E-08
5.7E-09
Water Column Carnivores
2.1E-06
7.7E-09
1.0E-07
2.9E-08
Mallard Duck
1.3E-05
2.8E-08
9.4E-08
7.0E-08
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
a Concentrations in surface water are in mg [2,3,7,8-TCDD]/L [water], which is equivalent to pg/g.
Concentrations in bulk sediment are in pg [2,3,7,8-TCDD]/g [sediment] dry weight. Concentrations in biota are
pg[2,3,7,8-TCDD]/g [biota] wet weight.
Exhibit 4-10. Concentrations (pg/g) of 2,3,7,8-TCDD in Compartments of the
TRIM.FaTE Aquatic Food Web at Year 50 with 95-percent UCL Emission Rate3
Compartment
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Surface Water
1.3E-11
2.7E-14
9.2E-14
6.8E-14
Sediment
6.3E-08
3.9E-10
1.3E-09
9.7E-10
Algae
8.3E-09
1.7E-11
5.9E-11
4.3E-11
Macrophytes
1.9E-07
4.1E-10
1.4E-09
1.0E-09
Benthic Invertebrates
6.2E-09
3.8E-11
1.4E-10
1.1E-10
Benthivorous Fish
4.7E-06
2.2E-09
1.9E-08
3.7E-09
Bottom-feeding Carnivores
N/A
1.1E-08
9.6E-08
2.1E-08
Water Column Planktivores
8.8E-07
2.2E-10
8.4E-10
7.3E-10
Water Column Omnivores
N/A
3.2E-09
3.0E-08
1.4E-08
Water Column Carnivores
5.2E-06
1.9E-08
2.5E-07
7.2E-08
Mallard Duck
3.1E-05
6.8E-08
2.3E-07
1.7E-07
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
a Concentrations in surface water are in mg [2,3,7,8-TCDD]/L [which = mg/kg or pg/g]. Concentrations in bulk
sediment are in pg [2,3,7,8-TCDD]/g [sediment] dry weight. Concentrations in biota are pg [2,3,7,8-TCDD]/g
[biota] wet weight.
J-28
-------
2,3,7,8-TCDD exposure concentrations based on the 95-percent UCL emission rate are
generally less than an order of magnitude greater than exposure concentrations based on the
mean emission rate.
Chemical concentrations in the fish compartments are highest in Ravena pond which has the
highest ratio of catchment area to water volume of the four water bodies evaluated. Previous
case studies using TRIM.FaTE had indicated that harvest rate offish from a water body
influenced chemical concentrations in the fish compartments by removing chemical from the
system. The influence of a given harvest rate on chemical concentrations in fish decreased with
increasing total fish biomass in the water body. We therefore decided to investigate the
consequences of simulating a single angler harvesting fish from Ravena Pond (as assumed in
the Ravena HHRA, Appendix I) on the chemical concentrations in fish compartments.
A single mammalian TRIM.FaTE compartment was added to the terrestrial parcel adjacent to
Ravena Pond. The body weight of the angler, who was assumed to live in the house near
Ravena Pond, was set to 71.4 kg, the average weight of an adult human as used in the Ravena
HHRA (Appendix I). We assumed that the angler would harvest fish at a rate of 17 g/day (90th
percentile fish ingestion rate, see Appendix C, Attachment 2) from Ravena Pond, and of the fish
harvested 67 percent were fish from the benthic omnivore (BO) fish compartment and 33
percent were fish from the water column carnivore (WCC) compartment to reflect the
assumptions used in the Ravena HHRA (Appendix I). We assumed that the small forage fish
(water column herbivore) were too small to be keepers. Note that the actual fish harvest rate
expressed as total biomass offish removed from Ravena Pond that would correspond to the
human fish ingestion rate listed here is likely to be 2 to 3 times higher because humans do not
consume an entire fish. The fillet generally constitutes a third of the wet weight of a fish, and
edible muscle with skin generally is no more than half of the wet weight of a fish. Although
wildlife also may feed on fish from Ravena Pond, we did not include consumption of fish by
wildlife.
Exhibit 4-11 compares the annual angler fish harvest to the standing biomass for each fish
compartment in Ravena Pond, which has a surface area of 20,000 m2. Note that as the angler
removes fish (with chemical) from the pond, TRIM.FaTE maintains the same biomass for the
fish compartment. That model feature is consistent with recruitment of younger, less
contaminated fish, into the pond adult fish population at a rate equal to the removal rate of the
adults.
Exhibit 4-11. Biomass of Fish Harvested by a Single Angler Fishing in
Ravena Pond Relative to Standing Biomass of Fish in Each Compartmenta
Compartment-specific Properties
Water
Column
Carnivore
Benthic
Omnivore
Total Fish
Biomass
Fish Biomass Density (kg ww/m2)
0.00020
0.0030
0.0040
Total Fish Biomass in Pond (kg)
4.01
60.2
80.2
Fish Biomass (kg) Harvested Annually by
Single Angler at Mean Fish Ingestion Rate (17
g/day)
4.14
2.07
6.21
a Total surface area of Ravena Pond is approximately 20,000 m2.
Exhibit 4-11 illustrates several points. First, an angler harvesting fish at a rate of 17.0 g/day
would need to catch fish at other water bodies in addition to Ravena Pond. The water column
J-29
-------
carnivore harvest rate of 4.14 kg/year is not possible with a standing stock of only 4.01 kg total
in Ravena Pond. The benthic omnivore harvest rate of 2.07 kg/year from a standing biomass of
60.2 kg is more reasonable and likely to be sustainable.
Exhibit 4-12 summarizes the chemical concentrations in each of the aquatic compartments for
Ravena Pond as presented in the previous exhibits. Exhibit 4-13 summarizes the chemical
concentrations in the two fish compartments for which concentration changed with the addition
of the single angler harvester.
Exhibit 4-12. Concentrations (|jg/g) of Mercury and 2,3,7,8-TCDD in the Ravena Pond
Aquatic Compartments at Year 50 Without Fish Harvesting by Humans or Wildlife 3
Compartment
Ravena Pond
MeHg
Hg+2
Mean TCDD
95th UCL
TCDD
Surface Water
4.8E-07
1.1E-04
5.3E-12
1.3E-11
Sediment
1.9E-03
9.1E-01
2.6E-08
6.3E-08
Algae
1.7E-02
4.9E-01
3.4E-09
8.3E-09
Macrophytes
9.6E-07
5.3E-05
7.7E-08
1.9E-07
Benthic Invertebrates
5.9E-03
4.6E-02
2.5E-09
6.2E-09
Benthivorous Fish
3.0E-02
1.6E-02
1.9E-06
4.7E-06
Bottom-feeding Carnivores
N/A
N/A
N/A
N/A
Water Column Planktivores
5.8E-02
1.5E-01
3.6E-07
8.8E-07
Water Column Omnivores
N/A
N/A
N/A
N/A
Water Column Carnivores
1.7E-01
3.5E-02
2.1E-06
5.2E-06
Mallard Duck
6.5E-03
3.1E-02
1.3E-05
3.1E-05
N/A= Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
a Concentrations in surface water are in mg/L. Concentrations in sediment are in mg/kg [sediment] dry wt.
Concentrations in biota are mg/kg wet weight.
Exhibit 4-13. Concentrations (|jg/g) of Mercury and 2,3,7,8-TCDD in the Ravena Pond
Aquatic Compartments at Year 50 With 17 Grams Fish Harvested per Day by One
Angler from Two Fish Compartments 3
Compartment
Ravena Pond
MeHg
Hg+2
Mean TCDD
95th Percentile
TCDD
Benthivorous Fish
3.0E-02
1.6E-02
1.9E-06
4.7E-06
Water Column Carnivores
1.7E-01
3.5E-02
2.1E-06
5.2E-06
a Concentrations in biota are pg [2,3,7,8-TCDD]/g [biota] wet weight. Concentrations for all other aquatic
compartments are the same as in Exhibit 4-12; they were unaffected by harvesting 17 grams of fish daily (33
percent from water column carnivore and 67 percent from benthic omnivore (benthivorous) fish compartments).
With the addition of an angler harvesting 17.0 g/day or 6.21 kg/year offish from Ravena Pond,
chemical concentrations in the fish compartments which the angler harvests decrease
substantially. In other fish compartments, chemical concentrations do not change with the
addition of the angler. Exhibit 4-14 below presents the 2,3,7,8-TCDD concentrations with and
without the angler in Ravena Pond. When the angler is present, fish concentrations in both the
water column carnivore and benthic omnivore fish compartments are reduced by 21 percent for
the benthic omnivore, and 38 percent for the water column carnivore. The proportional
reduction in chemical concentration in the water column carnivore represents an unrealistically
high harvest rate. The concentrations in the water column herbivore and the benthic
J-30
-------
invertebrates included for comparison are the same with and without the angler harvesting fish.
The concentrations in these compartments remain unchanged because there are no changes to
the rate at which their predators consume biomass (and chemical) from those compartments.
Exhibit 4-14. 2,3,7,8-TCDD Concentrations in Aquatic Foodweb Compartments With and
Without Angler Harvesting of 17 Grams of Fish Daily in Ravena Pond
1.E-05
¦ Without Harvester
~ With Harvester
1.E-06
c P
S s
o I
o S
1.E-07
1.E-08
1.E-09
Benthic Invertebrate
Benthic Omnivore
Water Column Herbivore Water Column Carnivore
Using the 95-percent UCL dioxin emission rate.
Exhibit 4-15 presents the mercury concentrations with and without an angler harvesting fish
from Ravena Pond. With harvesting, fish concentrations are reduced substantially for the water
column carnivore for both divalent and methyl mercury (20 percent and 42 percent,
respectively). Fish concentrations are reduced to a lesser amount with the angler present for
the benthic omnivore fish compartment (2 percent Hg+2 reduction, 7 percent MeHg reduction).
J-31
-------
Exhibit 4-15. Concentrations of Divalent and Methyl Mercury in Aquatic Foodweb
Compartments, With and Without Angler Harvesting of 17.0 grams of Fish Daily in
Ravena Pond
¦ Methyl Mercury, Without Harvester
~ Methyl Mercury, With Harvester
¦ Divalent Mercury, Without Harvester
~ Divalent Mercury, With Harvester
0.1
I s
m ¦*->
o
c 5
,9 u>
u>
3 E
o ^
l_
o
0.01
0.001
Benthic Invertebrate
Benthic Omnivore
Water Column Herbivore
Water Column Carnivore
Exhibit 4-15 also illustrates that the proportion of total mercury present as MeHg is higher for the
water column carnivore than for the water column herbivore (higher for the higher fish trophic
level), which is consistent with reports from the literature (EPA 2005). While humans harvest
primarily the larger fish, wildlife that swallow their prey whole (e.g., mergansers) generally
harvest smaller fish, and so may be exposed to less MeHg than indicated by measurements or
estimates of total mercury.
Exhibit 4-16 shows the TRIM.FaTE predicted speciation of mercury in the aquatic ecosystem
compartments for Nassau Lake. Two of the fish compartments, the benthic carnivore and water
column omnivore, were not modeled in Ravena Pond; therefore, Nassau Lake is used to
illustrate Hg speciation among aquatic compartments. The majority of the total Hg in
algae/zooplankton, macrophytes, and benthic invertebrates (89, 95, and 88 percent,
respectively) is in the inorganic, Hg+2, form. At the next benthic trophic level (benthivorous
fish), TRIM.FaTE estimated 37 percent Hg+2 and 63 percent MeHg. At the next trophic level in
the water column (water column planktivores), however, TRIM.FaTE estimated 67 percent Hg+2
and 33 percent MeHg. Given that MeHg comprises 11 percent of the total Hg in the diets of
both the benthivorous fish (100% benthic invertebrates) and the water column planktivorous fish
(97% plankton), this difference in Hg speciation at the next higher trophic level was not
expected. Possible reasons are still under examination. As expected for top predators, most of
the total Hg in the bottom-feeding carnivorous fish, 89 percent, was estimated to be MeHg,
while 96 percent was estimated to be MeHg in the water column carnivores.
J-32
-------
Exhibit 4-16. Speciated Mercury Concentrations for Surface Water, Sediment, and
Biota in Nassau Lake (ppm [SW: mg/L; sediment: pg/g dry weight; algae, Bl, fish: pg/g
wet weight])
Total Hg
Cone.
Divalent Hg
Elemental Hg
Methyl Hg
Foodweb Compartment
Cone.
% Total
Hg
Cone.
% Total
Hg
Cone.
% Total
Hg
Surface Water
1.9E-07
1.6E-07
86%
2.4E-08
13%
2.5E-09
1%
Sediment
7.2E-03
7.2E-03
99%
3.9E-05
1%
1.5E-05
0%
Algae/zooplankton
7.8E-01
6.9E-01
89%
0.0E+00
0%
8.6E-02
11%
Macrophytes
9.6E-08
9.1E-08
95%
1.6E-17
0%
4.9E-09
5%
Benthic Invertebrates
4.2E-04
3.7E-04
88%
2.0E-06
0%
4.8E-05
11%
Benthivorous Fish
3.4E-04
1.3E-04
37%
2.9E-12
0%
2.2E-04
63%
Bottom-feeding Carnivores
8.2E-04
9.4E-05
11%
1.0E-12
0%
7.3E-04
89%
Water Column Planktivores
2.5E-04
1.7E-04
67%
1.2E-13
0%
8.1E-05
33%
Water Column Omnivores
4.0E-04
1.1E-04
28%
2.3E-12
0%
2.9E-04
72%
Water Column Carnivores
1.2E-03
4.5E-05
4%
3.8E-19
0%
1.1E-03
96%
Mallards
3.0E-04
2.5E-04
82%
3.1E-07
0%
5.3E-05
17%
J-4.1.1.3 Exposure Doses
MeHg and 2,3,7,8-TCDD exposure doses for the four wildlife species included in the Ravena
ERA were estimated by first multiplying the estimated chemical concentrations in each food type
(Exhibit 4-7 through Exhibit 4-10) by the daily ingestion rates of each food type (Exhibit 4-5 and
Exhibit 4-6) to yield average daily intake rates for each chemical for each surface water body.
Then, the intake rates were divided by body weights (Exhibit 4-2) to calculate the body-weight
normalized chemical intake rate or dose (|jg/g-day). Intakes of MeHg and 2,3,7,8-TCDD are
calculated for each wildlife species, food type, and water body. Because of the different health
endpoints for MeHg and Hg+2, the exposures are estimated separately for each. For Ravena
Pond, we used the most conservative scenario to calculate the fish compartment concentrations
- no fish harvesting by anglers or wildlife.
MeHg intake rates are presented in Exhibit 4-17 through Exhibit 4-20. Each exhibit includes the
MeHg intake rates for one wildlife species by water body and by prey type. Total MeHg intake
rates for each water body are shown in the bottom row of each exhibit. These total MeHg intake
rates are compared to reference toxicity values to calculate hazard indices in Section J-4.1.2.
Exhibit 4-17. Tree Swallow Intake of MeHg (pg/g-day)
Food Type
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthic Invertebrates
1.6E-04
1.1E-06
1.3E-06
1.6E-06
Total
1.6E-04
1.1E-06
1.3E-06
1.6E-06
J-33
-------
Exhibit 4-18. Common Merganser Intake of MeHg (jjg/g-day)
Food Type
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthivorous Fish
5.9E-03
2.0E-05
3.2E-05
2.6E-05
Water Column Planktivores
1.1E-02
9.3E-06
1.2E-05
1.1E-05
Water Column Omnivores
N/A
2.0E-05
3.0E-05
3.0E-05
Water Column Carnivores
4.3E-03
1.2E-05
2.3E-05
1.8E-05
Total
2.2E-02
6.1E-05
9.7E-05
8.6E-05
N/A = Not applicable. Water column omnivores are not included in the Ravena Pond food web.
Exhibit 4-19. Bald Eagle Intake of MeHg (jjg/g-day)
Food Type
Water Body
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthivorous Fish
6.1E-03
1.9E-05
2.9E-05
2.4E-05
Water Column Omnivores
N/A
2.5E-05
3.9E-05
3.9E-05
Water Column Carnivores
3.0E-02
6.5E-05
1.3E-04
1.0E-04
Mallard Duck
6.2E-04
4.3E-06
5.0E-06
6.5E-06
Total
3.7E-02
1.1E-04
2.0E-04
1.7E-04
N/A = Not applicable. Water column omnivores are not included in the Ravena Pond food web.
Exhibit 4-20. Mink Intake of MeHg (jjg/g-day)
Food Type
Water Body
Ravena Pond
Alcove Res
Nassau Lake
Kinderhook
Lake
Benthic Invertebrates
3.4E-04
1.7E-06
2.1E-06
2.6E-06
Benthivorous Fish
1.8E-03
6.2E-06
9.7E-06
8.1E-06
Bottom-feeding Carnivores
N/A
8.8E-07
1.3E-06
1.2E-06
Water Column Planktivores
3.5E-03
2.9E-06
3.7E-06
3.3E-06
Water Column Omnivores
N/A
8.4E-06
1.3E-05
1.3E-05
Total
5.6E-03
2.0E-05
3.0E-05
2.8E-05
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
Given the relative magnitude of MeHg intakes from Ravena Pond compared with the three other
water bodies (more than two orders of magnitude higher), we estimated intakes of Hg+2
(divalent mercury) for wildlife at Ravena Pond only as shown in Exhibit 4-21.
J-34
-------
Exhibit 4-21. Wildlife Intakes of Hg+2 (jjg/g-day) at Ravena Pond
Food Type
Species
Swallow
Merganser
Bald Eagle
Mink
Benthic Invertebrates
1.2E-03
-
-
2.7E-03
Benthivorous Fish
-
3.1E-03
3.2E-03
9.5E-04
Bottom-feeding Carnivores
-
-
-
-
Water Column Planktivores
-
2.9E-02
-
9.0E-03
Water Column Omnivores
-
-
-
-
Water Column Carnivores
-
8.9E-04
6.2E-03
-
Mallard Duck
-
-
2.9E-03
-
Total
1.2E-03
3.3E-02
1.2E-02
1.3E-02
-- = No ingestion of this compartment.
As discussed in the previous section, mean and 95-percent UCL emission rates for 2,3,7,8-
TCDD were used to estimate exposure concentrations. Because the exposure concentrations
based on the 95-percent UCL emission rates were generally within an order of magnitude of the
exposure concentrations based on mean emissions rate, exposure doses for 2,3,7,8-TCDD
were calculated using only the 95-percent UCL emission rates (Exhibit 4-10). The resulting
2,3,7,8-TCDD exposure doses are presented in Exhibit 4-22 through Exhibit 4-25.
Exhibit 4-22. Tree Swallow Intake of 2,3,7,8-TCDD (|jg/g-day)a
Water Body
Food Type
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthic Invertebrates
1.7E-10
1.0E-12
3.8E-12
2.9E-12
Total
1.7E-10
1.0E-12
3.8E-12
2.9E-12
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
Exhibit 4-23. Common Merganser Intake of 2,3,7,8-TCDD (|jg/g-day)a
Water Body
Food Type
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthivorous Fish
9.3E-07
3.2E-10
2.8E-09
5.5E-10
Water Column Planktivores
1.7E-07
3.2E-11
1.2E-10
1.1E-10
Water Column Omnivores
N/A
3.4E-10
3.1E-09
1.5E-09
Water Column Carnivores
1.3E-07
4.0E-10
5.2E-09
1.5E-09
Total
1.2E-06
1.1E-09
1.1E-08
3.6E-09
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
N/A = Not applicable. Water column omnivores are not included in the Ravena Pond food web.
J-35
-------
Exhibit 4-24. Bald Eagle Intake of 2,3,7,8-TCDD (jjg/g-day)3
Water Body
Food Type
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthivorous Fish
9.8E-07
2.9E-10
2.6E-09
5.0E-10
Water Column Omnivores
N/A
4.3E-10
4.0E-09
1.9E-09
Water Column Carnivores
9.1E-07
2.2E-09
2.9E-08
8.3E-09
Mallard Duck
3.0E-06
6.5E-09
2.2E-08
1.6E-08
Total
4.9E-06
9.4E-09
5.7E-08
2.7E-08
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
na = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
Exhibit 4-25. Mink Intake of 2,3,7,8-TCDD (|jg/g-day)a
Water Body
Food Type
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Benthic Invertebrates
3.6E-10
1.7E-12
6.2E-12
4.7E-12
Benthivorous Fish
2.9E-07
9.7E-11
8.6E-10
1.7E-10
Bottom-feeding Carnivores
N/A
2.0E-11
1.7E-10
3.8E-11
Water Column Planktivores
5.3E-08
9.7E-12
3.8E-11
3.3E-11
Water Column Omnivores
N/A
1.4E-10
1.3E-09
6.2E-10
Total
3.4E-07
2.7E-10
2.4E-09
8.6E-10
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
N/A = Not applicable. Bottom-feeding carnivores and water column omnivores are not included in the Ravena
Pond food web.
J-4.1.2 Ecological Effects Assessment
As described in Section J-2, protection of local populations of three species of piscivorous and
one species of insectivorous wildlife were selected as the ecological assessment endpoints. To
evaluate the risks associated with exposure to 2,3,7,8-TCDD and Hg in fish or insects, a
benchmark dose below which population-level effects are considered unlikely is needed for
each combination of chemical and receptor species. This section describes the derivation of
these benchmarks, termed toxicity reference values (TRVs), for the local wildlife of concern near
the Ravena facility.
A TRV for a mammalian or avian species is calculated from a "critical" study, reporting the
highest dose at which no adverse effects on reproduction, development, or survival are
observed. This test dose (TD) or point of departure (POD) might be reduced by one or more
uncertainty factors (UF) that reflect the limitations of the database from which the critical study
was selected. The resultant POD/UF value can be referred to as a toxicity benchmark or TRV.
During risk characterization, TRVs are compared with the estimated exposure (dose) to assess
a hazard quotient (HQ) for adverse effects.
ICF reviewed TRVs that have been developed for avian and mammalian wildlife over the past
several decades as described in Section J-3.2.3. The best documented of those values were
published by EPA in its 1995 Great Lakes Water Quality Initiative Criteria Documents for the
J-36
-------
Protection of Wildlife: DDT, Mercury, 2,3,7,8-TCDD, PCBs (EPA 1995a). EPA performed some
additional review of the mercury criteria for wildlife in its 1997 Mercury Study Report to
Congress (EPA 1997 Volume 5). We therefore began with those values and conducted a
literature search for more recent studies that might indicate the need for revision of one or more
of those values to lower doses. The TRVs are expressed in units of milligrams[chemical]/
kilogram[fresh body weight (BW)]-day (mg/kg-day), micrograms[chemical] /kilogram[BW]-day
(|jg/kg-day), or |jg/g-day (equivalent to mg/kg-day). Where use of dietary concentrations (e.g.,
ppm Hg in the diet expressed as mg[Hg]/kg diet) might be confused with dose to an animal, BW
is specified in the units of dose.
J-4.1.2.1 Calculation of Wildlife TRVs for 2,3,7,8-TCDD
In the Great Lakes Water Quality Initiative Criteria Documents for the Protection of Wildlife, EPA
(1995a) calculated wildlife toxicity values for the effects of 2,3,7,8-TCDD on avian and
mammalian species. EPA conducted a computer-based and manual search for published
studies on the effects of 2,3,7,8-TCDD available in the literature through approximately 1994. As
a result of this search, 26 adequately documented reports of dose-response data were identified
and summarized in EPA's 1995 GLWQI wildlife criteria report. The derivations of the avian and
mammalian TRVs for 2,3,7,8-TCDD are described below.
Avian 2,3,7,8-TCDD TRV
As of 1995, EPA had identified only one comprehensive avian dose-response study on the
effects of 2,3,7,8-TCDD that was adequate for the calculation of an avian TRV. Three
publications by Nosek et al. (1992a,b, 1993) outline the effects of 10 weekly 2,3,7,8-TCDD i.p.
injections on ring-necked pheasants at levels equivalent to 0.0014 |jg[TCDD]/kg[BW]-day, 0.014
|jg/kg-day, and 0.14 |jg/kg-day. Pheasants in the 0.14 |jg[TCDD]/kg[BW]-day group showed a
significant decrease in egg production and increase in the mortality of embryos from fertilized
eggs. These effects were not seen in the other two dose groups. Based on these results, EPA
(1995a) concluded that the lowest-observed-adverse-effect level (LOAEL) for fertility and
embryo mortality in pheasants was 0.14 |jg[TCDD]/kg[BW]-day, and the no-observed-adverse-
effect level (NOAEL) was 0.014 |jg[TCDD]/kg[BW]-day. EPA selected this study for use in
developing wildlife criteria for the GLWQI because it showed meaningful endpoints for long-term
(70 days) i.p. administration of 2,3,7,8-TCDD.
For wildlife TRVs in general, EPA considers three uncertainty factors that might need to be
applied to a dose from a critical study. Interspecies uncertainty factors (UFA) are used to
develop TRVs for species other than the test species to account for toxicokinetic and
toxicodynamic differences between the species. Because gallinaceous birds are thought to be
among the most sensitive avian species, EPA set the UFA for a TRV for belted kingfisher,
herring gull, and bald eagle equal to 1. The Nosek et al. (1992a,b, 1993) studies are
subchronic; in order to extrapolate the results to chronic effects, a sub-chronic-to-chronic
uncertainty factor (UFS) was set equal to 10. This factor accounts for the rate of steady-state
accumulation and whole-body elimination of 2,3,7,8-TCDD. Lastly, the NOAEL was identified
by the investigators, indicating that the LOAEL-to-NOAEL uncertainty factor (UFL) can be set
equal to 1.
The NOAEL of 0.014 |jg[TCDD]/kg[BW]-day for adverse reproductive effects in pheasants
divided by a compound UF of 10 results in an avian TRV of 0.0014 |jg/kg[BW]-day for 2,3,7,8-
TCDD.
J-37
-------
EPA (1995a) conducted a brief sensitivity analysis to illustrate the significance of the
assumptions made on the value calculated for the TRV. The first assumption evaluated was
that gallinaceous birds are the most sensitive of the avian species. If the UFA for all
representative species were set equal to 3, rather than 1, the avian TRV would equal 0.47
picograms/kg[BW]-day (i.e., 0.00047 |jg/kg-day).
No additional data have been identified to indicate the need for a lower avian TRV for 2,3,7,8-
TCDD.
Mammalian 2,3,7,8-TCDD TRV
As of 1994, chronic or subchronic studies of the effects of 2,3,7,8-TCDD on mammalian wildlife
species were not available. EPA therefore reviewed studies of TCDD toxicity to laboratory
mammals. The Agency identified five adequate chronic and subchronic studies of the effects of
dietary exposure to 2,3,7,8-TCDD, as described in the GLWQI wildlife criteria document (EPA
1995a). After considering the relevance and adequacy of each study, EPA selected the three-
generation rat study using three dietary doses of 2,3,7,8-TCDD conducted by Murray et al.
(1979) as the TD (POD) for three reasons: it covered a wide range of reproductive effects; both
a NOAEL and a LOAEL were identified by the investigators; and the test species was exposed
to the 2,3,7,8-TCDD over three generations.
Sprague-Dawley rats of the fO, f1, and f2 generations were exposed to dietary doses of 0.001,
0.01, or 0.1 |jg[TCDD]/kg[BW]-day for 90 days prior to and throughout gestation (Murray et al.
1979). In the fO generation of the 0.1 |jg[TCDD]/kg[BW]-day group, the fertility, litter size, and
neonatal survival of pups was significantly lower while the incidence of stillbirths was
significantly higher than the control group. In the 0.01 |jg/kg-day group, no effect on fertility was
evident in the fO generation, but the f1 and f2 generations exhibited significantly lower fertility,
litter sizes, and postnatal body weights and a significantly higher incidence of still-births. At the
lowest dose, there was no significant difference between the fertility of experimental and control
animals. Murray et al. (1979) concluded that the LOAEL and NOAEL in this study of
reproductive endpoints in Sprague-Dawley rats were 0.01 and 0.001 |jg[TCDD]/kg[BW]-day,
respectively.
The UFs considered in the mammalian analysis are the same factors as considered in the avian
analysis. EPA set the UFL to 1 because the critical study identified a NOAEL. EPA set the UFS
to 1 because the study covered three generations. Given the limited number of mammalian
species for which chronic data were available, and considering the high sensitivity of mink to
PCBs and other chemicals, EPA determined that the UFA to extrapolate from rats to mink
should be 10.
The NOAEL of 0.001 |jg/kg-day for adverse effects on reproductive endpoints in rats divided by
a composite UF of 10 results in a mammalian TRV for 2,3,7,8-TCDD of 0.0001 |jg/kg[BW]-day.
EPA conducted a brief analysis to assess how sensitive the calculated TRV was to assumptions
included in its derivation. Although Murray et al. (1979) concluded that the 0.001 |jg[TCDD]/
kg[BW]-day dose was a NOEAL for rats, others have reinterpreted the same data and
concluded that 0.001 |jg/kg-day is actually the LOAEL. Using a LOAEL of 0.001 |jg/kg-day and
a UFL of 3, the resulting TRV (mammalian) would be 0.00033 |jg[TCDD]/kg[BW]-day, a less
conservative value. Another assumption EPA evaluated is whether mink are the most sensitive
mammalian species. Bowman et al. (1989a,b) had identified NOAEL and LOAEL values for
survival to weaning of young Rhesus monkeys of 0.00012 and 0.0059 |jg[TCDD]/kg[BW]-day,
respectively. If this NOAEL were used as the POD instead, setting the UFA values for mink and
J-38
-------
otter to 1, the resulting mammalian TRV for 2,3,7,8-TCDD would be 0.000012 |jg/kg-day, a
lower value. In general, however, EPA does not use toxicity results for primates, which often
are very sensitive to chemical exposures, as a POD for North American mammalian wildlife
species.
No additional data have been identified to indicate the need for a lower mammalian TRV for
2,3,7,8-TCDD.
J-4.1.2.2 Calculation of Wildlife TRVs for Mercury
Wildlife that consume aquatic prey may ingest mercury in its divalent state (Hg+2) and as
methyl mercury (MeHg). As illustrated in Exhibit 4-16 above, although TRIM.FaTE predicts that
most mercury in the top predators, or "game" fish, is in the form of MeHg, that might not hold
true for the smaller forage fish that can comprise a large proportion of the fish diet consumed by
wildlife that swallow their prey whole (e.g., kingfishers, mergansers, cormorants). For example,
for Nassau Lake near Ravena, TRIM.FaTE predicted that a majority of the Hg in small water
column planktivorous fish is present as Hg+2 (Exhibit 4-16). TRIM.FaTE predicts that most (88
percent) of the total Hg present in benthic invertebrates, which are consumed after emergence
by swallows, is present as Hg+2.
For humans, EPA has derived separate reference doses for Hg+2 and MeHg. For wildlife, we
also consider these two forms of mercury separately.
Avian Methyl Mercury TRV
EPA summarized subchronic and chronic toxicity test results for birds exposed to Hg in its
GLWQI wildlife criteria document for mercury (EPA 1995a). The most robust data identifying
both LOAEL and NOAEL values from the data examined were the mallard studies by Heinz
(1974, 1975,1976b, and 1979). These studies covered three generations, quantified several
different measures of reproductive success, and provided dose-response information even
though a NOAEL was not identified.
Heinz (1974, 1975, 1976a, 1976b, 1979) assessed the effects of dietary MeHg in mallards in
two sets of experiments. In the first set, Heinz (1974, 1975,1976a) exposed adult mallards to
commercial feed treated with MeHg dicyandiamide at concentrations of 0, 0.5, or 3.0 ppm from
18 months of age through two consecutive breeding seasons. Egg production stopped earlier in
the 3 ppm group compared with the 0.5 ppm and control groups (Heinz 1974). The number of
normal hatchlings and survival of hatchlings through one week were significantly reduced in the
3.0 ppm group but not in the 0.5 ppm group, compared with the control group. During the
second breeding season, most measures of reproduction for hens exposed to 3.0 ppm had
improved from the first breeding season and matched control levels, with the exception of
normal hatchlings surviving through one week, which remained significantly lower (Heinz
1976a). The LOAEL and NOAEL determined from these studies for the reproductive
performance of adult mallards exposed to MeHg in their diet is 3.0 ppm Hg and 0.5 ppm Hg,
respectively.
The second series of experiments considered the effects of dietary MeHg on reproduction and
behavior in three consecutive generations of mallards. The second season offspring from adult
mallards exposed to MeHg dicyandiamide at 0.5 ppm dietary Hg were themselves exposed to
0.5 ppm dietary Hg from 9 days of age through their third reproductive season (Heinz 1976b).
The offspring of these birds then were exposed to 0.5 ppm dietary Hg beginning at 9 days of
age (Heinz 1979). Both a statistically significant increase in eggs laid outside of the nest box
J-39
-------
and decrease in the number of one-week-old ducklings produced were observed in the second
generation exposed to dietary concentrations of 0.5 ppm Hg (Heinz 1976b). These trends were
observed in the third generation, but were not significant (Heinz 1979); however, these data
combined with the results from the second generation were significantly different from controls
on both measures (Heinz 1979). These results suggest that MeHg at 0.5 ppm Hg in the diet
may be associated with reproductive effects in multigenerational exposure; therefore, a LOAEL
of 0.5 ppm Hg for MeHg was inferred. Multiplying the LOAEL by the average food ingestion rate
for treated mallards in the second and third generation (i.e., 0.156 kg/kg-day) results in a
LOAEL for MeHg of 0.078 mg[Hg]/kg[BW]-day (or |jg[Hg]/g[BW]-day), the value used as the
POD (EPA 1995a, 1997).
EPA evaluated the three standard UFs, although the Agency revised its estimate of the species-
to-species UFa in its Report to Congress (EPA 1997).
• To extrapolate the results from the mallard to other species of birds in other orders or
families, for the GLWQI, EPA concluded that a UFAgreater than 1 would be required
(EPA 1995a). Of the avian species for which data were presented in the GLWQI
criteria document (EPA 1995a), the mallard and pheasant appear to be the most
sensitive. The pheasant study used an exposure duration of only 12 weeks, and the
LOAEL was determined to be 0.093 mg[Hg]/kg[BW]-day (Fimreite 1971). With such a
short duration study, EPA concluded that the pheasant might be even more sensitive
than the mallard. EPA therefore assigned an intermediate value of 3 for the UFA to
extrapolate to other species of birds (EPA 1995a).
• In its Report to Congress, however, EPA (1997) decided to set the UFA to a value of
1.0 instead of 3. The decision was based on a review of the literature that indicated
piscivorous birds are better able to detoxify MeHg than non-piscivorous birds (Dietz et
al. 1990), apparently including mallards which consume benthic invertebrates in the
former category.
• A UFs greater than 1 was not necessary because Heinz's studies covered three
generations.
• For the GLWQI, EPA set the UFL to 2 because the LOAEL appeared to EPA to be
very near the threshold for effects of Hg on mallards (EPA 1995a). For the Report to
Congress, EPA set the UFL to 3, citing the GLWQI methodology (EPA 1995b).
The resultant GLWQI TRV for mallards, obtained by dividing the NOAEL of 78 |jg[Hg]/kg[BW]-
day by UFL of 2 would be 39 |jg[Hg]/kg[BW]-day. For the GLWQI, the TRV for other species of
birds, for which the UFA is 3 is 13 |jg[Hg]/kg[BW]-day. For its Mercury Report to Congress, the
NOAEL of 78 |jg/kg-day divided by a total UF of 3 established an avian TRV of 26 |jg/kg-day for
all avian species (EPA 1997).
Of the two EPA avian TRVs for MeHg, we prefer the GLWQI TRV for two reasons. First, we are
not confident that mallards, which consume benthic invertebrates that might have relatively low
MeHg in relation to Hg+2 content, have the same higher ability as piscivorous birds to detoxify
MeHg. In addition, birds larger than mallards (e.g., merganser, bald eagle) might have, on
average, longer lives lives than mallards, which might result in higher tissue concentrations of
Hg, particularly in older birds, for the same daily exposure dose per unit body weight. We agree
that the identified LOAEL represents a low level of adverse effects, and that a UFL of 2 to
estimate a NOAEL is likely to be adequate. We therefore use the value of 0.13 |jg[Hg]/kg[BW]-
day as the avian TRV for the Ravena ERA.
J-40
-------
ICF conducted a literature review to identify more recent publications with exposure-response
information for the effects of MeHg on avian wildlife. Recent studies of common loons in Maine
indicate that MeHg concentrations in fish of 0.05 |jg[Hg]/g[diet wet weight] or less pose "low
risks" and of 0.05 to 0.15 |jg[Hg]/g[diet] pose "moderate risks" of adverse effects on
reproduction (Evers et al. 2004). Evers et al. (2004) provided some evidence that fish
concentrations of 0.15 |jg[total Hg]/g[fish wet weight (ww)] were roughly associated with loon
blood total Hg concentrations of 3 |jg/g, which was a LOAEL associated with lowered
reproductive success in the field, although the criteria for a "LOAEL" were unclear. For adult
female loons weighing 4.7 kg, and a fish ingestion rate of 15 percent of the adult body weight
estimated from Nagy's (1987) allometric equation for non-passerine birds, a value which is
slightly less than the fish ingestion rate of 20 percent of body weight measured for growing 35-
day-old loon chicks by Fournier et al. (2002) using doubly labeled water, 0.15 |jg[Hg]/g[fish ww]
would correspond to an exposure dose of 0.11 |jg[Hg]/g[BW]-day. Using a food ingestion rate
of 20 percent of body weight daily, the exposure dose would be 0.14 |jg[Hg]/g[BW]-day (Evers
et al. 2004). Using a UFL of 3 to extrapolate from a possible LOAEL to a NOAEL, a UFA of 3 for
inter-species variation in sensitivity, and UFS of 1 (field exposures are of the duration of
interest), a final TRV for birds based on field data from Evers et al. (2004) would be 0.012 to
0.016 |jg[total Hg]/g[BW]-day. Therefore, the MeHg TRV of 0.013 |jg[Hg]/g[BW]-day
established by EPA in 1995 is consistent with the more recent data.
Mammalian Methyl Mercury TRV
From its review of available subchronic and chronic toxicity studies on the effect of MeHg on
mammalian species for its GLWQI, EPA selected a NOAEL for MeHg of 0.16 mg [Hg]/kg[BW]-
day as the POD (EPA 1995a). The NOAEL is from a 93-day study of MeHg chloride
administered in the diet to mink (Wobeser et al. 1976b).
Wobester et al. (1976b) exposed adult female mink to dietary concentrations of MeHg chloride
of 1.1, 1.8, 4.8, 8.3, and 15.0 ppm Hg for up to 93 days. Clinical signs of Hg intoxication
(anorexia and ataxia) were observed in all mink exposed to concentrations of 1.8 ppm Hg and
greater. All five of the mink exposed to 1.8 ppm Hg developed ataxia: two of the mink died, and
the remaining three were killed following onset of symptoms for examination. The investigators
determined that dietary Hg concentration was directly related to the time of the onset of toxic
effects and death. Pathological alterations in the nervous system were observed at the 1.1 ppm
Hg in the diet, but additional clinical symptoms were absent; therefore, EPA initially concluded
that this dietary concentration would not have clear implications for population-level effects on
mink (EPA 1995a). Using the captive mink body weight of 1.0 kg and food ingestion rate of 0.15
kg/day, the dietary concentration of 1.1 ppm Hg was converted to a NOAEL of 0.16
mg[Hg]/kg[BW]-day (or jjg[Hg]/g[BW]-day).
After obtaining the doctoral thesis of Wobeser (Wobeser 1973), EPA concluded that the effects
observed at the 1.1 ppm concentration in the diet, lesions of the central nervous system and
axonal degeneration, were sufficiently adverse to consider that exposure to be a LOAEL; EPA
used data from the first part of the study to identify a NOAEL of 0.33 ppm (EPA 1997). In
addition, EPA recalculated the doses using data on the weights of female mink and kits used in
the experiments, for a NOAEL for MeHg for mink of 55 |jg[Hg]/kg[BW]-day.
In order to extrapolate the results from this study to a chronic TRV for mink, EPA evaluated two
UFs: a subchronic-to-chronic factor and an interspecies uncertainty factor. A third UF, to
estimate a NOAEL from a LOAEL, was not needed.
J-41
-------
• Wobester et al. (1976b) had concluded that the pathological alterations observed at
the 1.1 ppm dietary concentration after 93 days would have resulted in distinct clinical
signs of toxicity had the exposure period been longer. In a prior study, Wobester et
al. (1976a) determined that the NOAEL for MeHg for adult mink was 0.05
mg[Hg]/kg[BW]-day over a 145-day dietary exposure period. This NOAEL is
approximately a factor of 3 less than the 93-day NOAEL of 0.16 mg[Hg]/kg[BW]-day
for MeHg discussed above. Considering a mink's lifetime of 6 or 7 years, 145 days
represents a relatively short subchronic exposure. Therefore, for the GLWQI, EPA set
the UFS to 10 (EPA1995a). When using a lower dose for the NOAEL, however, EPA
set the UFS to 3 for the Report to Congress (EPA 1997).
• The UFa was set to 1 for mink because mink was the test species (EPA 1995a,
1997).
The resultant MeHg TRV for mink, obtained by dividing the NOAEL for mink by the product of
the UFs discussed above was 0.016 and 0.018 mg[Hg]/kg[BW]-day for the GLWQI (EPA 1995a)
and the Report to Congress (EPA 1997), respectively. Because we are concerned that the 93-
day exposure is short compared to the lifetime of a mink, and because MeHg is only slowly
eliminated and therefore tends to increase in concentration in older animals (EPA 1997), we
prefer to retain the more conservative UFS to 10 from the GLWQI to use with the more
conservative NOAEL of 55 |jg[Hg]/kg[BW]-day to estimate a TRV for mink of 1.8 |jg[Hg]/kg[BW]-
day.
ICF conducted a literature review to identify more recent publications with exposure-response
information for the effects of MeHg on mammalian wildlife. While many recent studies focus on
the relationship between environmental Hg contamination and mammalian wildlife total Hg
tissue concentrations in North America (e.g., Halbrook et al. 1994, Mierle et al. 2000, Thompson
1996, Yates et al. 2004, Wolfe et al. 1998), we did not identify any new data linking exposure
doses to adverse effects at lower levels than the mink study used by EPA in 1995 and 1997.
Avian and Mammalian TRVs for Divalent Mercury
We did not identify any TRVs developed for Hg+2 for avian or mammalian wildlife; concern and
research have largely focused on MeHg or, in some cases, total Hg assuming most of it is
methylated.
Mammalian TRV for Divalent Mercury
To determine whether we should expend the effort to derive a TRV for Hg+2, we first compared
the human reference dose (RfD) for Hg+2 (i.e., 3 |jg/kg-day) to the human RfD for MeHg, (i.e., 1
|jg/kg-day). The two RfDs were derived from different health endpoints using substantially
different (UFs) as described briefly below.
RfD for Mercuric Chloride. The human RfD for chronic oral exposure to mercuric chloride
(essentially Hg+2) is 3E-4 mg[Hg+2]/kg[BW]-day based on autoimmune effects (i.e.,
"...formation of mercuric-mercury-induced autoimmune glomerulonephritis") (EPA IRIS). Dose
conversions for three studies of the Brown Norway Rat (Druet et al. 1978; Bernaudin et al. 1981;
Andres 1984) were used to derive the RfD. The conversions were a factor of 0.739 to convert
the weight of HgCI2 to Hg+2, a factor of 1 for the different routes of exposure (i.e., an
assumption of 100 percent absorption efficiency by both the subcutaneous (s.c.) and oral routes
of exposure), and a factor to estimate an average daily exposure from the days per week
injections were administered. The three identified LOAELs, as converted, were 0.226, 0.317,
J-42
-------
and 0.633 mg[Hg+2]/kg[BW]-day. A composite uncertainty factor (UF) of 1000 was applied to
the LOAEL determine the RfD. The UF included a factor of 10 for conversion from LOAEL to
NOAEL, a factor of 10 for use of subchronic studies, and a factor of 10 for both animal-to-
human extrapolation (interspecies variation) and sensitive human populations (intraspecies
variation).
RfD for Methyl Mercury. The human RfD for chronic oral exposure to MeHg is 1E-4 mg
[MeHg]/kg[BW]-day based on developmental neuropsychological impairment (EPA IRIS). (We
note that the molecular weights of MeHg and Hg are similar.) Human epidemiological studies
were used to derive the RfD (Grandjean et al. 1997). Surrogate data on maternal daily dietary
intake were used for the observed developmental effects in children exposed in utero. Maternal
daily dietary intake rates were calculated primarily from concentrations in cord blood. Almost all
of the Hg in cord blood was MeHg. A benchmark-dose (BMD), not NOAEL/LOAEL, approach
was used to identify a POD within the observed range of response in the critical study of
neurological impairment in children. The lower 95 percent confidence limit of the BMD in
maternal blood corresponding to a 5 percent response above the control (BMDL05) ranged from
46 ppb to 79 ppb. This blood concentration corresponded to a range of maternal daily intakes
of 0.858 to 1.472 pg [MeHg]/kg[BW]-day. A composite uncertainty factor of 10 was applied to
the estimated maternal daily intake. The UF included a factor of 3 to account for
pharmacokinetic variability within humans and uncertainty in estimating an ingested mercury
dose from cord-blood mercury concentration and a factor of 3 for pharmacodynamic variability
among humans.
A comparison of the derivation of the human RfDs for Hg+2 and MeHg indicates that although
similar in magnitude, they are based on substantially different studies, health endpoints, and
types of uncertainty. The chronic oral RfD for Hg+2 includes a composite UF of 1000, whereas
the UF is only 10 in the RfD derivation for MeHg. The Hg+2 RfD is based on a LOAEL (as
opposed to a NOAEL), animal (as opposed to human) data, and subchronic exposures. We
conclude that comparing the human RfD for Hg+2 to that for MeHg does not provide an
adequate basis by which to compare the chronic toxicity of Hg+2 to MeHg in wildlife.
Comparison of Chronic Organic and Inorganic Hg Toxicity in Rats. We identified one animal
study that provides sufficiently similar experiments by which to compare the relative chronic
toxicity of organic and inorganic mercury. Fitzhugh et al. (1950) compared the toxicity of dietary
organic (phenyl mercuric acetate) and inorganic (mercuric acetate) Hg to rats exposed for up to
two years. For both experiments, groups of rats were exposed at 0 (control), 0.1, 0.5, 2.5,10,
40, or 160 ppm Hg in the diet. The organic form reduced growth in males at 10 ppm Hg in
males (40 ppm reduced growth in both males and females); survival was reduced only in the
160 ppm group. The inorganic form reduced growth in males exposed at 160 ppm, but no other
adverse effects were observed. The LOAEL for reduced growth in males, therefore, was 160
ppm Hg in the diet as inorganic mercury and 10 ppm Hg as organic mercury, suggesting that
the organic form was more toxic than the inorganic form for chronic exposures.
Fitzhugh et al. (1950) calculated the doses associated with 10, 40, and 160 ppm diets to be
0.15, 0.6, and 2.4 mg[Hg]/rat-day, but did not report body weights for the different groups, did
not distinguish males from females, and did not consider the reduced body weight of rats in
groups that exhibited reduced body weight. Assuming a body weight of 0.175 kg, and using
data presented by the investigators where possible, EPA (1995a) calculated the doses as 0.56,
2.2, and 14 mg[Hg]/kg-day for these studies (EPA 1995a). Thus, the LOAEL and NOAEL for
growth in rats (male) for organic Hg are 2.2 and 0.56 pg/g-day, respectively; and a LOAEL for
growth in male rats and a NOAEL for reproduction and development for inorganic Hg is14 pg/g-
j-43
-------
day for this study. Thus, the LOAEL for inorganic Hg is between 4 and 6 times higher than the
LOAEL for the same effect (reduced growth in males) for organic mercury.
Limitations of this study include its age, that the sensitive endpoints of reproduction and
neurodevelopment (for organic Hg) and nephrotoxicity (for Hg+2) were not assessed, and that
the chemical form of organic mercury may or may not be absorbed and distributed to the body
as is MeHg. In addition, data on the final tissue concentrations of Hg were not reported;
whether final tissue concentrations of total Hg were similar or lower for the Hg+2 group than for
the MeHg groups is not known. Given the higher clearance rate of Hg+2 than MeHg, it is
possible that the lower LOAEL for the organic Hg than for Hg+2 results in part from its
accumulation in tissues over time.
Derivation of Mammalian TRV for Hg+2 for Autoimmune Giemeruionephritis. Using the rat
Hg+2 toxicity studies cited in the derivation of the human RfD, we identify a POD for mammalian
wildlife as the geometric mean of the three LOAELs identified for autoimmune
glomerulonephritis in rats of 0.3 mg[Hg+2]/kg[BW]-day or 300 |jg[Hg+2]/kg[BW]-day. We
propose a UFL of 10 to estimate a NOAEL for a population-level effect from a LOAEL for a
sublethal individual effect that might affect reproductive success or survival. In addition, we
applied UFAof 3 is applied to account for toxicodynamic differences among mammals.
Toxicokinetic differences among mammalian wildlife species for Hg+2 should be estimated on
the basis of metabolic rate (body weight to the 3/4 power) relative to the metabolic rate of the
Brown Norway rat, because Hg+2 is readily eliminated by animals (in contrast to MeHg). To
extrapolate from a rat weighing 0.175 kg to a 1 kg mink, the POD is multiplied by a factor of 1.55
(i.e., 0.175 °25 /1 -°25) providing a final mink TRV of 16 jjg[Hg+2]/kg[BW]-day. The mink TRV
derived as described above for Hg+2 is about 9 times higher than the mink TRV for MeHg of 16
jjg[Hg+2]/kg[BW]-day.
Avian TRV for Divalent Mercury
We did not identify any chronic toxicity values for dietary Hg+2 exposure for birds or appropriate
data by which to estimate the chronic toxicity of Hg+2 by comparison with MeHg, although acute
toxicity tests indicate similar acute toxicities (EPA 1995a, Table 2-5). Given that it is unlikely
that chronic exposure to Hg+2 is as toxic as chronic exposure to MeHg in birds owing in part to
the more rapid elimination of Hg+2 than MeHg, we judge that the TRV for Hg+2 can be at least
2 times the MeHg TRV for birds smaller than mallards and at least 5 times the MeHg TRV for
birds larger than mallards, such as bald eagles and common mergansers.
Summary
A summary of the wildlife TRVs used in this risk assessment is provided in Exhibit 4-26.
Exhibit 4-26. Summary of Wildlife TRVs (jjg[chemical]/kg[BW]-day)
Chemical
Avian Values
Mink Values
POD
(Mg/kg-day)
UFTo,
TRV
(Mg/kg-day)
POD
(Mg/kg-day)
UFto,
TRV
(Mg/kg-day)
2,3,7,8-TCDD
14 E-03
10
1.4 E-03
1.0 E-03
10
0.10 E-3
Methyl Mercury
78
6
13
55
30
1.8
Divalent Mercury
N/A
N/A
Smaller birds: 26
Larger birds: 65
300
30/1.55 =
19
16
N/A = Not applicable.
J-44
-------
J-4.1.3 Risk Characterization
In Section J.3.2.4, a two-stage approach was proposed to characterize ecological risk of MeHg
and dioxin exposure. In the first step, HQs are calculated as an indicator of potential adverse
effects at the level of the individual organism. For those species or locations where HQ values
exceeded 1.0, a second stage analysis is conducted to provide a preliminary evaluation of
potential population-level effects.
For the first stage of the risk characterization, hazard quotients were calculated by dividing the
total exposure doses in Exhibit 4-17 through Exhibit 4-25 by the applicable avian or mammalian
TRVs in Exhibit 4-26. The resulting HQs for MeHg, Hg+2, and TCDD are presented in Exhibit
4-27, Exhibit 4-28, and Exhibit 4-29, respectively.
Exhibit 4-27. Hazard Quotients for Wildlife Exposure to Methyl Mercury3
Water Body
Wildlife Species
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
0.605
0.004
0.005
0.006
Common Merganser
1.304
0.004
0.006
0.005
Bald Eagle
0.634
0.002
0.003
0.003
Mink
3.919
0.014
0.021
0.020
a Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
Exhibit 4-28. Hazard Quotients for Wildlife Exposure to Divalent Mercurya,b
Water Body
Wildlife Species
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
2.37
<1
<1
<1
Common Merganser
0.40
<1
<1
<1
Bald Eagle
0.04
<1
<1
<1
Mink
0.98
<1
<1
<1
a Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
bThe HQs for Hg+2 are likely to be less than 1.0 at water-bodies other Ravena Pond given that exposure
doses are more than two orders of magnitude lower for wildlife consuming prey from those water bodies.
Exhibit 4-29. Hazard Quotients for Wildlife Exposure to 2,3,7,8-TCDD a,b
Water Body
Wildlife Species
Ravena Pond
Alcove
Reservoir
Nassau Lake
Kinderhook
Lake
Tree Swallow
0.01
0.00004
0.0001
0.0001
Common Merganser
0.70
0.001
0.01
0.002
Bald Eagle
0.77
0.001
0.01
0.004
Mink
4.27
0.003
0.03
0.01
a Exposure doses are based on the estimated 95-percent UCL dioxin emission rates.
b Hazard quotients highlighted in blue and bold indicate exceed the hazard quotient threshold of 1.
J-45
-------
All HQs for all species are below 1.0 for all chemicals with a few exceptions. For MeHg, an HQ
of 1.3 was calculated for the common merganser in Ravena Pond. For Hg+2, an HQ of 2.4 was
calculated for the tree swallow in Ravena Pond. Finally, for 2,3,7,8-TCDD, an HQ of 4.3 was
calculated for the mink in Ravena Pond. No hazard quotients were found to be greater than at
water bodies other than Ravena Pond.
Ravena Pond has a surface area of only 0.02 km2 and a shoreline of approximately 0.8 km. At
that size, at most a few pairs of swallows, one pair of mergansers, and one female mink might
forage there each season. De Graaf et al. (1981) reported tree swallows breeding at a density
of about 0.0007 pairs/ha or 0.7 pair/km2. Typical breeding densities for common mergansers
are 0.07 to 0.11 breeding pairs/ km2, or about 1 pair per 10 km2 of habitat containing many
suitable bodies of water (Cadman et al. 1987, Erskine 1987, and Ross 1987 as cited in Mallory
and Metz 1999). The density of female mink in the vicinity of a Michigan river was reported to
be 0.006 per hectare or 0.6 per km2 (Marshall 1936). Along a Montana river, Mitchell (1961)
reported densities of mink of between 0.03 and 0.085 individuals per hectare of area near shore
or 3 to 8.5 individuals per km2. Estimating the mink density per unit river shoreline instead of
per unit area, Marshall (1936) reported the mink density along the Michigan river to be 0.6
mink/km.
Populations of piscivorous and insectivorous wildlife are not expected to be adversely affected
by MeHg, Hg+2, and dioxin attributable to emissions from the facility. Adverse effects on the
reproductive success of a single pair or female of a non-endangered avian species should not
result in any population-level effects. These results indicate that emissions of neither Hg or
2,3,7,8-TCDD from the Ravena facility are expected to result in adverse toxic effects on the
selected wildlife species. As discussed in Section J-2, the wildlife species chosen for the
Ravena ERA are expected to be the most highly exposed species likely to be found near the
Ravena facility.
J-4.1.4 Uncertainties in Ravena ERA Related to Mercury and Dioxin
This section identifies uncertainties and limitations of the data and approaches used for the
Ravena ERA for MeHg, Hg+2, and 2,3,7,8-TCDD. Where possible, we qualitatively identify the
likely direction in which these limitations may affect the relevant results.
• As discussed in problem formulation (Section J-2), some aspects of the analysis
scope were determined based on screening analyses or other decisions associated
with the HHRA. For example, the contaminants of concern for both the HHRA and
Ravena ERA were selected based on a de minimis emissions screen that identified
mercury and dioxins as the persistent and bioaccumulative chemicals of highest
potential concern from a human health perspective. Although human health criteria
were used to select the HAPs, we determined, based on release quantities, chemical
characteristics, and toxicity to ecological receptors, that these pollutants, as well as
HCI, are the pollutants most likely to pose ecological risks. However, it is possible
that a systematic ecological screening analysis may have identified additional HAPs
to include in the ERA.
• The Ravena ERA does not address or include background concentrations of total Hg,
MeHg, or 2,3,7,8-TCDD. It also does not consider the impact of other environmental
pollutants and sources (e.g., PCB contamination in the nearby Hudson River) on the
baseline condition of the wildlife receptors.
• The assessment endpoints for the ERA include piscivores and insectivores that were
chosen because they are likely to be the most highly exposed receptors given their
J-46
-------
feeding habits and the high bioaccumulation potential for MeHg and 2,3,7,8-TCDD.
However, it is possible that other locally present species that were not considered
(e.g., because their local presence has not been documented, such as river otter, or
because relevant data to estimate values for their body weights, diets, or food
ingestion rates is not readily available) might be more highly exposed.
The Ravena ERA used fate and transport modeling results performed for the HHRA,
although the aquatic food webs were constructed with both the HHRA and ERA in
mind. Details of the TRIM.FaTE fate and transport modeling methodology and its
potential limitations are discussed in Appendix I.
The TRIM.FaTE aquatic food web models for the four water bodies are based on
information obtained from NYS DEC and other sources. Like any model, the aquatic
foodweb models are fairly gross simplifications that may not accurately estimate
exposure concentrations.
Although the TRIM.FaTE aquatic foodweb models were conceived using field data on
the composition and relative abundance offish species in local aquatic communities,
the original model does not consider the influence of harvesting fish on removal of
chemicals from the aquatic ecosystem. For the two lakes and reservoir, it is unlikely
that this limitation is of significance. For Ravena Pond, however, the mean and 90
percentile human harvest rates alone are not sustainable or not possible, and smaller
harvest rates plus predation by any wildlife species are likely to remove sufficient
chemical from the system to significantly lower the chemical concentration in the
benthic omnivores and water column carnivores.
Mercury speciation as predicted by TRIM.FaTE indicates a higher proportion of Hg+2
than MeHg in water column planktivores. If this is an overestimate of Hg+2 and an
underestimate of MeHg in that compartment, the exposures to MeHg may be
somewhat underestimated for mink and mergansers. This bias is too small to affect
our conclusions, however.
Emission rates for mercury are based on data from the 2002 NEI (taking into account
any recent changes made as a part of the SAB analyses). These data are not
necessarily consistent with current emissions rates. The NEI does not include TCDD
emissions data for the Ravena facility. Therefore, emissions factors were used to
estimate mean and 95-percent UCL 2,3,7,8-TCDD emission rates. The 95-percent
UCL emission rates were used in the ERA. Ecological exposures and risks estimated
based on the 95-percent UCL emission rates may be overly conservative.
Exposure factors for which values were estimated or assumed for the wildlife species
include body weight, diet composition, and feeding rate. These data were obtained
from the literature. Published values, particularly for diet composition, can vary
substantially depending on location, time of year, sex, or other factors. In choosing
assumptions using the available literature, we considered several factors, including
the sample size, distance or latitude difference from the Ravena facility, and study
methods (e.g., using captive vs. free-living animals) with the general goal of using the
most robust and representative data. Where temporal variations were evident,
assumptions were based on average annual values if possible. Adult body weights
were averages for adult males and females. Because some species display
significant sexual dimorphism, the body weights and, to a lesser extent, the food
ingestion rates are under- or over-estimates for each sex.
ICF used diet composition information from the literature to make assumptions about
the percentage of each species' diet obtained from each of the nine food types
included in the TRIM.FaTE aquatic food web model. In making these assumptions,
ICF judged the closest match between the reported prey species/types and the biotic
J-47
-------
compartments included in the model based on the diet and size of the prey
species/types. This limitation may cause over- or under-estimation of actual
ecological exposure levels. In some cases, we made conservative assumptions to
facilitate modeling exposures, such as assuming that mink and bald eagles take 100
percent of their prey from aquatic ecosystems, assuming year-round residency for the
migratory birds (see next bullet), assuming all prey come from a single water body
near the Ravena facility, and using a high-end estimate of consumption of top
predatory fish by bald eagles.
• Because chronic exposures are of concern, with both 2,3,7,8-TCDD and MeHg
accumulating in tissues over time, ingestion rates for free-living adult animals are
used and no attempt has been made to define ingestion rates for nestling birds fed by
their parents, by adult female birds laying eggs, or by pregnant or lactating mink.
Note that while mink are resident year-round, the mergansers and bald eagles
migrate further southward to follow the open (not ice-covered) water, while the
swallows migrate to South America for the winter. Thus, the bird species would not
be exposed year-round to dioxins and methyl mercury that originated with the Ravena
facility. Given the global nature of contamination of aquatic ecosystems with dioxins
and Hg, however, it is likely that the birds will be exposed year-round to these
chemicals.
The only issue that might affect our conclusions of negligible ecological risks from Hg and
TCDD emissions from the Ravena facility is omission of existing background concentrations of
these chemicals, particularly, Hg.
J-4.2 Results for HCI
This section discusses the facility-ranking analysis (Section J-4.2.1) and the indirect ecological
effects assessment for HCI (Section J-4.2.2).
J-4.2.1 Results for Facility-Ranking Analysis
The preliminary facility ranking according to ecological hazards was based on scores for three
factors (Section J-4.2.1.1). For the facilities with the lowest scores (i.e., highest potential for
ecological hazards), proximity to specially valued ecosystems was assessed (Section J-4.2.1.2)
to determine the facilities for which to examine readily available lines of evidence for indirect
ecological effects that might result from ongoing emissions of HCI from the facility.
J-4.2.1.1 Preliminary Facility Ranking
ICF conducted an initial ranking of all Portland Cement facilities that emit HCI based on three
indicators of ecological risk: (1) background acid deposition (regional pH of rainfall;
ATTACHMENT J-1 Exhibit 1), (2) surface water alkalinity (
J-4 8
-------
ATTACHMENT J-1 Exhibit 2) as an indicator of both surface water and soil alkalinity of a
geographic area, and (3) annual HCI emissions reported for the facility. Please refer to Section
J-3.3.1 for a detailed discussion of these indicators. We assigned background acid deposition
(an indicator of potential ecosystem susceptibility), surface water alkalinity (an indicator of
inherent ecosystem susceptibility [sensitivity] to acid deposition), and annual HCI emissions (an
indicator of potential exposure) scores from 1 to 5, with 1 indicating greatest potential for
ecological effects and 5 indicating lowest potential for ecological effects. These three scores
were multiplied for each facility to generate preliminary facility-specific ecological hazard scores.
Ecosystem susceptibility scores (both rainfall pH and surface water alkalinity), facility emission
scores, and preliminary ecological hazard scores are presented in
ATTACHMENT J-1 Exhibit 3. Background acid deposition and annual HCI emission scores for
the facilities were relatively evenly distributed from 1 through 5, allowing a reasonable ranking of
facilities based on the product of those two scores. The vast majority of facilities, however,
were located in areas of high surface water alkalinity (i.e., > 400 meq/L) indicating a high
buffering capacity. Few of the facilities that reported HCI emissions were located in areas
identified by EPA as having lower surface water alkalinity, as shown in
ATTACHMENT J-1 Exhibit 4. Only three facilities were located in areas with alkalinity less than
50 meq/L, and two of those are in Puerto Rico and not included in this analysis of the
conterminous United States. Thus, the surface water alkalinity score, which we used as an
indicator of both aquatic and local terrestiral ecosystems' abilities to resist changes in pH with
acid deposition, was a poor discriminator among Portland Cement facilities.
Four Portland Cement facilities shared the lowest preliminary ecological hazard score of ten,
indicating that these facilities may have the greatest potential to cause adverse ecological
effects. The four highest hazard (lowest score) facilities are located in Albany County, New
York; Hernando County, Florida; Carroll County, Maryland; and Dorchester County, South
Carolina. Each of these facilities had background acid deposition scores of 1 (with regional pH
of rainfall in the range of 4.5 to 4.7), alkalinity scores of 5 (with surface water alkalinity
measurements exceeding 400 meq/L), and facility emission scores of 2 (with total emissions
between approximately 50 and 180 tons per year).
Nine additional Portland Cement facilities had a preliminary hazard score of 20 or less. A facility
in Santa Cruz County, California, had the only alkalinity score of 1 (less than 50 meq/L). The
ecosystems in the vicinity of this facility are likely to have very limited acid buffering capacity,
J-4 9
-------
and soils and surface waters might show reduced pH with added HCI deposition from the
facility. This facility, however, is in a region of low background acid deposition (score of 5) and
is in the second lowest quintile of annual HCI emissions (emissions score of 4).
J-4.2.1.2 Refined Facility Ranking
To further refine the facility ranking to focus assessment of evidence of ecological effects,
facilities were selected with preliminary ecological risk scores of 20 or less (the top thirteen
facilities) to conduct a proximity analysis. Nearby ecologically valued areas were identified for
the top thirteen facilities. Ecologically valued environments were described in Section J-3 and
included reservoirs, rivers, lakes, and parks and preserves. Very large water bodies (e.g., the
Great Lakes and major rivers), which are not expected to show changes in pH from localized
HCI emissions, were excluded from the analysis. A proximity score was assigned to each
facility based on the square root of the distance between the facility and the valued
environment. The proximity score was rounded to one significant digit and stopped at a top
score of 5 (any separation greater than 25 km was assigned a proximity factor of 5).
A final hazard score was determined by calculating the product of the background rainfall pH
score, the alkalinity score, the facility emission score, and the proximity score for each of the top
thirteen facilities. Final hazard scores for each facility are reported in Exhibit 4-30.
J-50
-------
Exhibit 4-30. Final Hazard Scores for Top Thirteen Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Rainfall pH
Score
Alkalinity
Score
Emissions
Score
Preliminary
Hazard
Score
Closest
Sensitive
Environment
Distance
to Sensitive
Environment
(km)
Proximity
Score
Final Hazard
Score
PTC_NEI34931
Albany County,
NY
2
5
1
10
Alcove
Reservoir
11.30
3
30
PTC_NEI26327
Hernando
County, FL
2
5
1
10
Withlacoochee
State Forest
11.15
3
30
PTC_NEIMIB1559
Charlevoix
County, Ml
3
5
1
15
Lake Charlevoix
4.42
2
30
PTC_NEI51435
La Salle
County, IL
3
5
1
15
Illinois River
5.71
2
30
PTC_NEI12018
Alpena County,
Ml
3
5
1
15
Elbow Lake
11.37
3
45
PTC_NEIPAT$1626
Lawrence
County, PA
1
5
3
15
Evans Lake
9.76
3
45
PTC_NEI33394
Carroll County,
MD
2
5
1
10
Patuxent River
State Park
24.83
5
50
PTC_NEISC0351244
Dorchester
County, SC
2
5
1
10
Lake Moultrie
28.75
5
50
PTC_N EIM00990002
Jefferson
County, MO
3
5
1
15
Moredock Lake
17.49
4
60
PTC_NEI51352
La Salle
County, IL
4
5
1
20
Illinois River
6.41
3
60
PTC_NEI2CA151186
Santa Cruz
County, CA
5
1
4
20
Big Basin
Redwoods
State Park
10.97
3
60
PTC_NEI31319
Clark County,
IN
2
5
2
20
Quick Creek
Reservoir
38.52
5
100
PTC_NEI7255
Northhampton
County, PA
2
5
2
20
Beltzville Lake
30.03
5
100
Appendix J
J-51
SAB Review Draft - 05/22/09
-------
Four Portland Cement facilities shared the lowest final hazard score of 30 (with proximity scores
of 2 or 3), indicating that these facilities may have the greatest potential to cause harm to local
valued ecosystems due to indirect effects of HCI deposition to soils or surface waters. These
four facilities are located in Albany, New York (the Ravena facility); Hernando County, Florida;
Charlevoix County, Michigan; and La Salle County, Illinois.
Calculation of the final hazard score suggests a different ranking of the thirteen facilities than
the preliminary hazard scores indicated. The facilities in New York and Florida remained in the
top four, but the inclusion of the proximity score removed facilities in Maryland and South
Carolina from the top four and replaced them with facilities in Michigan and Illinois.
J-4.2.2 Indirect Ecological Effects Assessment
The localized indirect ecological effects of HCI released from Portland Cement facilities, if any,
would be mediated through changes in the pH of surface waters or the top layers of soil (plant
root zone). We therefore attempted to identify pH benchmarks associated with ecological
effects for surface waters (Section J-4.2.2.1) and soils (Section J-4.2.2.2).
J-4.2.2.1 Benchmarks for Surface Waters
Several water characteristics are related to the potential for acid loading to cause adverse
effects: pH, alkalinity, and water hardness. We discuss each below to clarify how they might be
used in assessing risks to aquatic communities.
PH
For freshwaters, EPA (1976, still current as of 2008) has recommended that pH be no lower
than 6.5. "pH" equals the negative of the log (base 10) of hydrogen ion (H+) activity in the water.
The materials in natural waters that most influence pH include carbon dioxide (C02), carbonic
acid (H2C03), bicarbonate ion (HC03 ), and carbonate ions (C03 ). The pH of surface waters
affects the toxicity of many chemical compounds to aquatic life by changing the degree of
dissociation for weak acids or bases; the undissociated compounds generally are more
bioavailable than are the hydrophilic dissociated ions. In general, one cannot identify a
"threshold" pH for adverse effects on aquatic life because of the influence of pH on the toxicity
of other chemicals that may or may not be present. Typically, water is not directly lethal to fish
at pH values as low as 5; however, several common water pollutants are more toxic at lower
pH.
The European Inland Fisheries Advisory Commission (EIFAC 1969) recommended that pH in
the range of 5.0 to 6.0 was unlikely to cause adverse effects to freshwater fish unless the
concentration of free C02 in water was higher than 20 mg/L (or if excess iron salts were
available). The Commission observed that fish in waters with pH in a range from 6.0 to 6.5
were unlikely to be harmed unless the concentration of free C02 was higher than 100 mg/L,
which it often can be. The Commission concluded that pH values between 6.5 and 9.0 should
not harm fish, although the toxicity of other chemical contaminants might be enhanced within
this range (EIFAC 1969, EPA 1976).
In establishing its lower criterion for pH in freshwaters of 6.5, EPA (1976) summarized additional
studies of freshwater organisms exposed to different pH levels. In a 13-month (1-generation)
exposure study using fathead minnows, Mount (1973) found fish deformities at pH values of 4.5
and 5.2 and reduced egg production and hatchability at a pH of 6.6 compared with the control
fish at pH 7.5. Bell (1971) examined the responses of two species of caddisfly, four species of
J-52
-------
stonefly, two species of dragonfly, and one species of mayfly nymphs to water at different pH
values. He found 50 percent mortality in mayfly nymphs in water at a pH of 5.4 for 30 days. He
also found the "50-percent-emergence" effect level to be as high as pH 6.6 for some species. A
pH of 6.5 appears, therefore, to be a reasonable screening value for adverse effects on
freshwater aquatic animals. ICF did not attempt to identify additional literature on the effects of
pH on aquatic organisms.
Alkalinity
Another water quality parameter relevant to interpreting acid loading to surface waters is
alkalinity as described earlier. Water alkalinity is the sum of the components in water that tend
to elevate water pH above a value of 4.5. Such materials include carbonates, bicarbonates,
phosphates, and hydroxides (EPA 1986). Some of the materials that contribute to alkalinity
(e.g., carbonates) reduce the toxicity of metals in surface waters by complexing with the metals
so that they are no longer bioavailable. Alkalinity is measured by titration with a standardized
acid to a pH value of 4.5, and it generally is expressed as mg/L equivalents of calcium
carbonate (CaC03). Photosynthesis by aquatic plants generates dissolved carbon dioxide,
which can acidify waters with limited buffering capacity (low alkalinity and high acidity).
The National Academy of Sciences' National Academy of Engineering (NAS/NAE 1974)
recommended that natural alkalinity not be reduced more than 25 percent, and for areas with
naturally low alkalinity (e.g., below 20 mg/L as calcium carbonate), alkalinity should not be
reduced further. In 1976, EPA established a lower bound criterion for alkalinity in freshwaters of
20 mg/L as calcium carbonate (EPA 1976); however, that criterion was replaced with a narrative
statement in 1986 (EPA 1986).
Water hardness
Water hardness also is expressed in units of mg/L as calcium carbonate. However, water
hardness reflects polyvalent metallic ions dissolved in water, primarily calcium and magnesium
in fresh waters, but also iron, strontium, and manganese. Based on human water uses,
classification of water hardness into soft (0 to 75 mg/L CaC03), moderately hard (75 to 150
mg/L), hard (150 to 300 mg/L), and very hard (300 mg/L or higher) (Sawyer 1960; EPA 1986).
Limestone is a natural source of water hardness.
Freshwater hardness can be divided into the carbonate and non-carbonate fractions. The
carbonate fraction is chemically equivalent to the bicarbonates present, and so carbonate water
hardness is considered equal to alkalinity. In general, the toxicity of metals to aquatic
organisms is reduced at higher levels of carbonate hardness/alkalinity (EPA 1986).
J-4.2.2.2 pH Benchmark for Soils
The pH of soils across the United States varies both regionally and locally due to a wide variety
of contributing factors. EPA does not have a soil criterion for pH of which we are aware, nor
have we identified a "threshold" for "adverse effects" established by any other agency. Because
of long-term adaptation of plants and soil communities to more acidic conditions in some
regions of the country, in some types of habitats, and with some soil types, there is no single
soil pH value that could serve as an ecotoxicity benchmark in all areas or all regions.
Nonetheless, we examined literature on soil pH associated with agricultural and horticultural
practices to identify a pH benchmark to assist in screening Portland Cement facilities for the
potential to cause indirect ecological effects of HCI deposition.
J-53
-------
There are many different methods of measuring soil pH, some of which provide different results.
Compared with shaking a soil sample with water to measure pH, the calcium chloride (CaCI2)
extraction method tends to result in pH estimates of 0.5 to 0.8 pH units lower.3 The CaCI2
method does not provide an actual soil solution pH, but rather a result that depends on soil
solution pH and hydrogen ions that are readily available through cation exchange. (Cation
exchange capacity is discussed in greater detail at the end of this section.) Most measurements
of pH include a temperature correction to a standard temperature of 25 °C.
Soil pH can change seasonally, daily, and hourly depending on temperature and moisture
content of the soil. Soil pH generally is reported as a range of pH values for a specified soil
depth (USDA 1998).
For agricultural areas, intensive growing of crops can cause a reduction in soil pH; however,
farmers can amend the soil to bring its pH back to more neutral levels. Several factors affect
soil pH in agricultural lands in particular:
• Addition of organic matter to soil to improve soil aeration and nutrients can result in
acidification as the organic materials decompose. Liming can restore pH.
• Addition of ammonium fertilizers to soils results in the production of nitrates, which
can hydrolyze in soils to nitric and nitrous acids.
• Harvesting crops can remove some of the alkaline elements (cations) originally in the
soil, reducing the soils buffering capacity. (See below for a discussion of soil
buffering capacity.)
Phosphorus, one key nutrient for plants, generally is most soluble at soil pH of 6.5 (6.0 to 7.0).
Some nutrients are more soluble at lower pH values and some at higher pH values. A pH range
of 6.0 to 7.0 generally is considered most favorable for growth for most species of plants
because it provides the highest availability of plant nutrients overall. Soil micro-organisms that
contribute to the availability of nitrogen, sulfur, and phosphorus in soils perform well in a pH
range of 6.6 to 7.3. Soils with a pH less than 5.5 generally have a low availability not only of
phosphorus, but also magnesium, calcium , and molybdenum (USDA 1998). Extremely acidic
soils (pH 4.0 to 5.0) often result in sufficiently high concentrations of soluble aluminum (and
sometimes iron and manganese) to be toxic to many species of plants (SUNY ESF;
http://www.esf.edu/pubprog/brochure/soilph/soilph.htm).
Acid-tolerant plants, such as rhododendrons, blueberries, azaleas, and certain pines and other
coniferous trees, can grow in soils of pH 4.0 to 5.0, depending on the species (and the source of
information), although pH of 5.0 to 5.5 may result in more vigorous growth for these species
because of increased nutrient availability.
The ability of a soil to resist changes in pH depends on its buffering capacity. In general, soils
with high clay and organic matter content and high cation exchange capacities tend to have
higher buffering capacities. Cation exchange capacity (CEC) is the ability of a soil to hold,
retain, and exchange cations (i.e., positively charged ions) such as calcium, magnesium,
potassium, sodium, ammonium, aluminum, and hydrogen (Daniels and Haering 2006). Soils
are generally characterized by a negative surface charge. Negative charges attract cations and
prevent their leaching. The higher the CEC, the more cations it can retain. A soil's CEC is
calculated by adding the charge equivalents of potassium, ammonium, calcium, magnesium,
aluminum, sodium, and hydrogen that are extracted from the exchangeable fraction of the soil.
3 http://www.bettersoils.com.au/module2/2 3.htm
J-54
-------
Low CEC values are in the range of 1 to 10 ten milliequivalents per 100 grams (meq/100g) and
high CEC values are in the range of 11 to 50 meq/100g. Soils with low CEC are often
characterized by high sand content and low clay content, low organic matter content, and low
soil pH. A low CEC indicates that the soil is not resistant to changes in pH or other chemical
changes and that the soil is more prone to cation leaching. Soils with high CEC often have low
sand and higher silt content, and moderate to high organic matter content. A high CEC
indicates that the soil is resistant to changes in pH and is less prone to cation leaching. Thus,
soils with a high CEC have a greater buffering capacity than do soils with a low CEC (Daniels
and Haering 2006).
Based on this information, we conclude that soils with pH lower than 4.0 are likely to produce
adverse effects in most species of plants, including those adapted to acid soil conditions. At a
pH less than 4.5, many plants would exhibit reduced growth owing to reduced availability of key
nutrients, but necrosis and death are possible where metal ions are mobilized. In areas for
which acid-tolerant plants are not native, pH values less than 5.5 are likely to cause adverse
effects on plant growth and survival. Soils with CEC values less than 11 meq/100g have low
buffering capacity and are less resistant to changes in soil pH than soils with CEC values
greater that 11 meq/100g.
J-4.2.3 Indirect Ecological Risk Characterization
We identified local measurements of surface water pH and local measurements of soil pH for
the top four facilities according to the final hazard ranking.
J-4.2.3.1 Catskill State Park Ecological Risk Characterization
Surface Waters
The Ravena facility (Facility ID NEI34931), located in Albany County, New York, is close to
Alcove Reservoir and Kinderhook Lake. EPA's STORET database did not have local
measurements of water pH for Alcove Reservoir or Kinderhook Lake when this analysis was
conducted. However, data for these water bodies were available from other sources. The City
of Albany Department of Water provided water pH data for Alcove Reservoir (NYS FWD).
Annual averages for 2007 and 2008 are presented in Exhibit 4-31. pH measurements were
taken at the surface, 5 feet, 20 feet, 34 feet, 48 feet, and at the bottom (65 feet) in 2007 and
2008. Annual pH averages of measurements taken at all depths ranged from 7.0 to 7.8. The
lowest annual average of water pH, 7.0, is above the EPA pH benchmark of 6.5. Therefore,
adverse ecological effects on fish and other aquatic wildlife associated with low water pH are
not anticipated for the Alcove Reservoir near the Ravena Facility.
J-55
-------
Exhibit 4-31. Measurements of Water pH for Alcove Reservoir in Albany
County, NY
Depth in Alcove Reservoir (ft)
Surface
5
20
34
48
65 (Bottom)
2007
Sample Size
9
7
9
9
9
9
Average (SD)
7.7 (0.3)
7.8 (0.2)
7.7 (0.2)
7.2 (0.4)
7.1 (0.4)
7.0 (0.4)
Exceeds pH
Benchmark
(6.5)
No
No
No
No
No
No
2008
Sample Size
4
4
4
4
4
4
Average (SD)
7.5 (0.2)
7.7 (0.2)
7.7 (0.2)
7.5 (0.2)
7.4 (0.2)
7.2 (0.2)
Exceeds pH
Benchmark
(6.5)
No
No
No
No
No
No
Data collected from Kinderhook Lake were provided by a private citizen to whom ICF was
referred by the New York Department of Environmental Conservation (NYS DEC). Data are
presented in Exhibit 4-32. From 2001 to 2008, Kinderhook Late was treated with alum to bind
phosphate and to reduce blue-green algae growth in the summer. In 2001, only the surface
water was treated and only surface water pH values were collected. Since 2001,
measurements were taken at the surface and at 20 feet. Since alum is acidic, application to the
deep regions may have lowered the pH by several tenths, but the pH returned to the pre-
treatment values presented in Exhibit 4-32 within two days. Measurements taken before 2004
were obtained using a pH meter that tested up to pH 10. Measurements taken after 2004 were
obtained using a color test with a limit of pH 8.2. It is not anticipated that this test
underestimated pH levels significantly because surface water values obtained with the original
meter did not register values above 8.3. Surface water (1 foot) and deep water (20 feet) pH
values are all above the EPA pH benchmark for surface water of 6.5. Therefore, adverse
effects on aquatic communities from the ongoing HCI deposition near the Ravena facility are not
anticipated.
Exhibit 4-32. Measurements of Water pH for Kinderhook Lake in Albany
County, NY
Year
Sample
Size
Surface Water pH
Annual Averaqe
(SD)
Deep Water pH
Annual Averaqe
(SD)
PH
Benchmark
Exceeds pH
Benchmark
2001
2
8.1 (0.1)
NA (NA)
No
2002
3
8.0 (0.4)
7.3 (0.2)
No
2003
3
8.1 (0.1)
7.1 (0.1)
No
2004
3
8.2 (0)
7.5 (0.3)
6.5
No
2005
1
8.2 (NA)
7.2 (NA)
No
2006
1
8.2 (NA)
7.2 (NA)
No
2007
1
8.2 (NA)
7.2 (NA)
No
J-56
-------
Soils
ICF used EPA's surface water alkalinity map (
ATTACHMENT J-1 Exhibit 2) to determine if the Ravena facility is located in an area of low
surface water alkalinity (as an indicator of soil alkalinity). As illustrated in
ATTACHMENT J-1 Exhibit 5, the Ravena facility modeling domain appears to lie almost entirely
in an area with high surface water buffering capacity.
ICF also used the USDA's Web Soil Survey to obtain local measurements of soil pH for Catskill
State Park, an ecologically valued terrestrial environment, which is approximately 30 km from
the Ravena facility (
ATTACHMENT J-1 Exhibit 6). An area of interest (AOI) of 11.2 acres was defined at the
nearest boundary of the park to the Ravena facility. Soil data for this AOI are presented in
Exhibit 4-33.
J-57
-------
Exhibit 4-33. Measurements of Soil pH and Effective CEC for Sensitive Terrestrial Environments Near Portland Cement
Facilities Emitting HCI
Facility ID
Facility
Location
Nearest
Terrestrial
Sensitive
Environment
Soil Type
(% of AOI)
Soil Depth
(inches)
Soil pH
Measurement
Effective CEC
(meq/1 OOg)
pH and ECEC
Benchmarks for
Soila b
Relationships to
pH and CEC
Benchmarks
Lewbeach and
Willowemoc channery
silt loams, moderately
steep, very bouldery
(= 60%)
Oto 6
3.6 to 5.5
0.0 to 2.9
High, Low
PTC_NEI3
4931
Albany
County, NY
Catskill State
Park
6 to 21
3.6 to 5.5
0.0 to 4.0
pH:
< 4.0 = high risk
4.0 to 5.5 =
moderate risk
High, Low
Vly Halcott complex,
strongly sloping, very
Oto 2
3.6 to 6.0
0.1 to 82
> 5.5 = low risk
CEC:
1 to 10 meq/1 OOg
High, High
rocky
(=40%)
2 to 28
3.6 to 5.5
0.1 to 15
low buffering
High, High
capacity
11 to 50 meq/1 OOg
Arrendo Fine Sand
Oto 8
4.5-6.0
0.2 to 1.6
Medium, Low
PTC_NEI2
Hernando
Withlacoochee
(= 30%)
8 to 62
4.5-6.0
0.0 to 3.2
= high buffering
capacity
Medium, Low
6327
County, FL
State Forest
Candler Fine Sand
0 to 4
4.5 to 5.5
0.1 to 1.8
Medium, Low
(= 70%)
4 to 48
4.5 to 5.5
0.0 to 1.6
Medium, Low
a pH <4.0 = high risk adverse effects in most species of plants, including those adapted to acid soil conditions; pH 4.0 to 5.5 = medium risk to plant species; ph > 5.5 =
low risk to all plant species
b CEC 1 to 10 meq/1 OOg = low buffering capacity and low resistance to changes in soil pH; CEC 11 to 50 meq/1 OOg = high buffering capacity and resistant to changes
in soil pH
J-58
-------
This AOI was characterized by two types of soil. Lewbeach and Willowmoc soils comprised
approximately 60 percent of the AOI and were characterized by USDA (2008) as "silt loams"
and "moderately steep." The top soil (0 to 6 inches) and deeper soil (6 to 21 inches) pH ranged
from 3.6 to 5.5. Effective cation-exchange capacity (ECEC) for the top soil layer ranged from
0.0 to 2.9 meq/100g, and for the deeper soil ranged from 0.0 to 4.0. Of note is the large range
in soil pH that characterizes the same soil layer and type, limiting the value of the soil pH
benchmarks indicated in Section J-4.2.2.2.
The lower boundary of the pH range for the Lewbeach and Willowmoc soils is below the
benchmark of 4.0, indicating a possibly high risk of adverse effect on many plant species,
including acid tolerant plants. The CEC values for the Lewbeach and Willowmoc soils also are
low, suggesting that significant parts of the Catskill State Park's soil has a low acid buffering
capacity and is not resistant to changes in soil pH. Note that the Catskill State Park is in an
area of the EPA surface water alkalinity map associated with alkalinity measurements of less
than 100 mg/L.
ATTACHMENT J-1 Exhibit 5).
The other soil type, Vly-Halcott soil, comprised approximately 40 percent of the AOI and was
described by USDA (2008) as "complex, strongly sloping" and "very rocky." The top two inches
of soil had soil pH values in the range of 3.6 to 6.0, indicating a low pH below the soil
benchmark of 4.0. However, this soil had CEC values ranging up to 82 meq/100g, suggesting
that at least patches of this soil have high acid buffering capacity. Deeper soil (2 to 28 inches)
of the same type had similarly low pH values (3.6 to 5.5) and relatively high CEC values (up to
15 meq/100g).
Given the possibly high sensitivity of the Catskill State Park area to further acid deposition, GIS
was used to evaluate the overlap between the HCI air concentrations estimated around the
Ravena facility and nearby parks.
ATTACHMENT J-1 Exhibit 6 illustrates that deposition of HCI emitted by the facility is unlikely to
reach the Catskill State Park.
J-59
-------
ATTACHMENT J-1 Exhibit 7 shows that deposition of HCI emitted by the facility might reach
into the John Boyd Thacher and Hudson River Islands State Parks, but only in areas with
surface water alkalinity greater than 400 mg/L (as CaC03) and therefore soils with a relatively
high acid buffering capacity. HCI emissions from the Ravena facility, therefore, are not
expected to produce indirect adverse ecological effects associated with acidification of either
surface waters or valued terrestrial ecosystems near the facility. This conclusion is supported
by aerial photography in the vicinity of the facility which indicates no discernable adverse effects
(e.g., die-back, chlorosis) on coniferous vegetation and no indication of adverse effects (e.g.,
excess fallen trunks) in the deciduous portions of the forests.4
J-4.2.3.2 Withlacoochee State Forest Ecological Risk Characterization
The Florida facility in Hernando County (NEI126327) is closest to Withlacochee State Forest, an
ecologically valued terrestrial environment located approximately 11 km from the Portland
Cement facility. We defined an AOI of 57.4 acres at the boundary of the forest closest to the
Hernando County facility using USDA's Web Soil Survey. Soils in this AOI were characterized
as either Arrendo fine sand, which comprised approximately 30 percent of the AOI, or Candler
fine sand, which comprised approximately 70 percent of the AOI (USDA 2008). EPA's surface
water alkalinity map suggests that the acid buffering capacity of surface waters in the area
surrounding the facility is not unusually low (i.e., greater than 400 mg/L).
The range of soil pH values listed for Candler top soil (0 to 4 inches) and deeper soils (4 to 48
inches) were the same: 4.5 to 5.5. CEC was slightly higher in the top soil (0.1 to 1.8 meq/100g)
than in deeper soils (0.0 to 1.6 meq/100g). The soil pH range is consistent with an acid-tolerant
plant community, but is unlikely to support acid-intolerant species. A low CEC suggests that the
soil does not have a high acid buffering capacity, suggesting a low tolerance for further
acidification. Soil pH in the other soil type, Arrendo fine sand, also did not differ for surface soil
(0 to 8 inches) and deeper soil (8 to 62 inches): pH range of 4.5 to 6.0. The lower boundary of
this range indicates some risk to acid-intolerant plant species; however, the vegetation of the
area may have evolved to be acid-tolerant given the sandy nature of both types of soils.
Effective CEC in Arrendo top soil ranged from 0.2 to 1.6 meq/100g and in deeper soil ranged
from 0.0 to 3.2 meq/100g. The range of soil pH measurements in Arrendo soil suggests
moderate risk to acid-intolerant plant species, and low CEC suggests that the soil does not have
significant buffering capacity, and might not be resistant to changes in soil pH. The sandy soils
in this area, however, may have contributed to the development of acid-tolerant plant
communities.
Inspection of aerial photographs of this facility was inconclusive owing to patchy distribution of
land uses near the facility and areas of grass and shrubs predominating in some directions and
with trees only apparent in other directions from the facility. Without additional data for the
environment in the vicinity of this facility, we cannot draw any conclusions regarding the
potential for indirect ecological effects of HCI emitted from the facility. HCI emissions rates and
background acid deposition rates are lower than for Ravena, however.
J-4.2.3.3 Lake Charlevoix Ecological Risk Characterization
The nearest sensitive environment to the Michigan facility (NEIMIB1559) is Lake Charlevoix,
which is approximately 4 km from the Portland Cement facility. EPA's STORET database had
local measurements of surface water pH for Lake Charlevoix that ranged from 7.7 to 8.3 (Exhibit
4-34). This range is above the EPA pH benchmark for surface water of 6.5. Therefore, adverse
4 EnviroMapper for Envirofacts located at: http://www.epa.gov/enviro/emef/.
J-60
-------
ecological effects on fish and other aquatic wildlife associated with low surface water pH are not
anticipated.
Exhibit 4-34. Measurements of Surface Water pH for Sensitive Aquatic
Environments Near Portland Cement Facilities Emitting HCI
Facility ID
Facility
Location
Nearest
Aquatic
Sensitive
Environment
Surface Water
PH
Measurement
PH
Benchmark
Exceeds pH
Benchmark
PTC_NEIMIB1559
Charlevoix
County, Ml
Lake
Charlevoix
7.7 to 8.3
6.5
No
PTC_NEI51435
La Salle
County, IL
Depue Lake
7.5 to 9.0
6.5
No
J-4.2.3.4 Lake Depue Ecological Risk Characterization
The Illinois River is the closest sensitive environment to the facility in La Salle County, Illinois
(NEl 151435). However, EPA's STORET database did not have local surface water data for the
Illinois River when this analysis was conducted. Local surface water measurements were,
however, available for Depue Lake, which is approximately 20 km from the Portland Cement
facility (see Exhibit 4-34). The reported surface water pH ranged from 7.5 to 9.0. This range is
above EPA's 6.5 pH benchmark for surface water, thus adverse ecological effects associated
with low surface water pH are not anticipated.
J-4.2.3.5 Summary of Risk Characterization for Indirect Effects of HCI
ICF used four factors to identify the most likely of the 91 Portland Cement facilities to pose risks
of indirect ecological effects associated with HCI deposition surrounding a facility. Four facilities
tied for the low hazard score (highest potential ecological risks). For all four facilities, pH values
in all nearby bodies of water were above the EPA pH criterion for freshwater of 6.5. For the
terrestrial environments, several lines of evidence indicate that the Ravena facility is not likely to
cause indirect adverse ecological effects associated with soil acidification in nearby valued or
unprotected terrestrial environments. Data from the Florida facility are inconclusive, but are
consistent with sustainable acid-tolerant plant communities which often occur in sandy soils.
J-5 References
Alexander, G. R. 1977. Food of vertebrate predators on trout waters in north central lower
Michigan. Michigan Academician 10: 181-195.
Anderson, B. W., and Timken, R.L. 1972. Sex and age ratios and weights of common
mergansers. J. Wildi. Manage. 36: 1127-1133.
Andres, P. 1984. IgA-IgG disease in the intestine of Brown Norway rats ingesting mercuric
chloride. Clin. Immunol. Immunopathol. 30: 488-494. (As cited in EPA Mercuric Chloride RfD
documentation in IRIS, last revised 05/01/1995).
Arnold, T. W., and Fritzell, E.K. 1987. Food habits of prairie mink during the waterfowl breeding
season. Can. J. Zool. 65: 2322-2324.
J-61
-------
Barr, J.F. 1996. Aspects of common loon (Gavia immei) feeding biology on its breeding
ground. Hydrobiologica 321: 119-114.
Bell, H.L. 1971. Effect of low pH on the survival and emergence of aquatic insects. Water Res.
5:313. (As cited in USEPA 1976.)
Bell, G.P. 1990. Birds and mammals on an insect diet: a primer on diet composition analysis
in relation to ecological energetics. Stud. Avian Biol. 13: 416-422.
Bernaudin, J.F., Druet, E., Druet, P., and Masse, R. 1981. Inhalation or ingestion of organic or
inorganic mercurials produces auto-immune disease in rats. Clin. Immunol. Immunopathol. 20:
129-135. (As cited in EPA Mercuric Chloride RfD documentation in IRIS, last revised
05/01/1995).
Bleavins, M. R. and R.J. Aulerich. 1981. Feed consumption and food passage in mink (Mustela
vison) and European ferrets (Mustela putorius furo). Lab. Animal Sci. 31: 268-269.
Bowerman, W. W. 1993. Regulation of bald eagle (Haliaeetus leucocephalus) productivity in
the great lakes basin: an ecological and toxilogical approach [Ph.D. dissertation]. Department of
Fisheries and Wildlife Institute for Environmental Toxicology, and Ecology and Evolutionary
Biology Program. Michigan State University.
Bowman R.E., S.L. Schantz, M.L. Gross, S. Ferguson. 1989a. Behavioral effects in monkeys
exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin transmitted maternally during gestation and for
four months of nursing. Chemosphere 18:235-242.
Bowman, R.E., S.L. Schantz, N.C.A. Weerasinghe, M.L. Gross, D.A. Barsotti. 1989b. Chronic
dietary intake of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) at 5 and 25 parts per trillion in the
monkey: TCDD kinetics and dose-effect estimate of reproductive toxicity. Chemosphere 18:243-
252.
Brezonik, P.L., S.O. King, and C.E. Mach. 1991. The influence of water chemistry on trace
metal bioavailability and toxicity to aquatic organisms. In: M.C. Newman and A.W. Mcintosh
(eds.) Advances in Trace Substances Research: Metal Ecotoxicology, Concepts and
Applications. Chelsea, Ml: Lewis Publishers, Inc. pp. 1-31.
Cadman, M. P., P. F. J. Eagles and F. M. Helleinen. 1987. Atlas of breeding birds of Ontario.
Univ. of Waterloo, Waterloo. (As cited in Mallory and Metz 1999.)
Colborn, T.I. 1991. Epidemiology of Great Lakes bald eagles. J. Environ. Health Toxicol. 4:
395-453.
Cowan, I. M., A.J. Wood, W.D. Kitts. 1957. Feed requirements of deer, beaver, bear, and mink
for growth and maintenance. Trans. North Am. Wildl. Conf. 22: 179-188.
Craig, R. J., E. S. Mitchell, J.E. Mitchell. 1988. Time and energy budgets of bald eagles
wintering along the Connecticut River. J. Field Ornithol. 59: 22-32.
Daniels, W.L. and Haering, K.C. 2006. Concepts of Basic Soil Science. In The Mid-Atlantic
Nutrient Management Handbook. February. Available at:
http://www.mawaterqualitv.org/Publications/manmh.htm.
J-62
-------
Dietz, R., Nielsen, C.O., Hansen, M.M., and Hansen, C.T. 1990. Organic mercury in Greenland
birds and mammals. Sci. Tot. Environ. 95: 41-51. (As cited in EPA 1997.)
De Graff, R.M., Witman, G.M., Lanier, J.W., Hill, B.J., and Keniston, J.M. 1981. Forest Habitat
Birds of the Northeast. USDA Forest Service. Northeast Forest Experiment Station and
Eastern Region. 598 pp. (As cited in EPA 2005c.)
Demers, E., D.J. McQueen, C.W. Ramcharan, A. Perez-Fuentetaja. 2001. Did piscivores
regulate changes in fish community structure? Arch. Hydrobiol. Spec. Issues Advanc. Limnol.
56: 49-80.
DeSorbo, C.R., and Evers, D.C. 2005. Evaluating exposure of Maine's Bald Eagle population
to Mercury: Assessing Impacts on productivity and Spatial Exposure Patterns. Report BRI
2005-08. Gorham, ME: BioDiversity Research Institute; 27 pp.
Druet, P., Druet, E., Potdevin, F., and Sapin, C. 1978. Immune type glomerulonephritis
induced by HgCI2 in the Brown Norway rat. Ann. Immunol. 129C: 777-792. (As cited in EPA
Mercuric Chloride RfD documentation in IRIS, last revised 05/01/1995).
Dunning, J.B., Jr. 1984. Body Weights of 686 Species of North American Birds. Monograph
No. 1. Western Bird Banding Assoc., Tucson, Arizona.
Dunning, J. B. Jr., ed. 1993. CRC Handbook of Avian Body Masses. Boca Raton, FL: CRC
Press.
Dunstan, T. C. and J.F. Harper. 1975. Food habits of bald eagles in north-central Minnesota.
J. Wildl. Manage. 39: 140-143.
EIFAC. 1969. Water quality criteria for European freshwater fish - extreme pH values and
inland fisheries. Prepared by European Inland Fisheries Advisory Commission, Working Party
on Water Quality Criteria for European Freshwater Fish. Water Res. 3: 593. (As cited in EPA
1976).
Eisler, R. 1987. Mercury Hazards to Fish, Wildlife, and Invertebrates: A Synoptic Review.
Contaminant Hazard Reviews Report No. 10. Biological Report 85(1.10). Laurel, MD: U.S.
Department of the Interior, Fish and Wildlife Service. April.
Environment Canada. 1991. Toxic Chemicals in the Great Lakes and Associated Effects.
Department of Fisheries and Oceans, Health and Welfare Canada. Cat. No. En 37-94/1990E;
ISBN 0-662-18316-9.
Erskine, A. J. 1987. Waterfowl breeding population surveys, Atlantic provinces. Can. Wildl.
Serv. Occas. Pap. Ser. no. 60. (As cited in Mallory and Metz 1999.)
ESRI. 2006. ESRI Data and Maps. Environmental Systems Research Institute, Inc. Redlands,
CA.
Evers, D.C., and Reaman, P.S. 1998. A comparison of mercury exposure between artificial
impoundments and natural lakes measured in common loon and their prey. Unpubl. Report. (As
cited in Evers et al. 2004.)
Evers, D.C., Lane, O.P., Savoy, L., and Goodale, W. 2004. Assessing the impacts of
methylmercury on piscivorous wildlife using a wildlife criterion value based on the Common
J-63
-------
Loon, 1998-2003. Report BRI 2004-2005 submitted to the Maine Depatment of Environmental
Protection. Gorham, ME: BioDiversity Research Institute.
Farrell, D. J. and A.J. Wood. 1968. The nutrition of the female mink (Mustela vison). II. The
energy requirement for maintenance. Can. J. Zool. 46: 47-52.
Feltham, M. J. 1995. Consumption of Atlantic salmon smolts and parr by goosanders:
estimates from doubly-labeled water measurements of captive birds released on two Scottish
rivers. J. Fish. Biol. 46: 273-281.
Fimreite, N. 1971. Effects of Methyl Mercury on Ring-necked Pheasants. Canadian Wildlife
Service Occasional Paper Number 9. Department of the Environment; 39 pp.
Fitzhugh, O.G., Nelson, A.A., Laug, E.P., and Kunze, F.M. 1950. Chronic oral toxicities of
mercuri-phenyl and mercuric salts. Indust. Hyg. Occup. Med. 2:433-442.
Fournier, F., Karasov, W.H., Kenow, K.P., Meyer, M.W., and Hines, R.K. 2002. The oral
bioavailability and toxicokinetics of methylmercury in common loon (Gavia immei) chicks.
Comp. Biochem. Physiol. Part A 133: 703-714.
Giesy, J.P., J.P. Ludwig, and D.E. Tillit. 1994. Embryolethality and deformities in colonial, fish-
eating, water birds of the Great Lakes region: assessing causality. Environ. Sci. Technol. 28:
128a-135a.
Gilbertson, M., T.J. Kubiak, P.J. Ludwig, et al. 1991. Great Lakes embryo mortality, edema,
and deformities syndrome (GLEMEDS) in colonial fish-eating birds: similarity to chick edema
disease. J. Toxicol. Environ. Health 33: 455-520.
Gooders, J. and T. Boyer. 1986. Ducks of North America and the Northern Hemisphere. New
York, NY: Facts on File.
Grubb, T. G. and R.J. Hensel. 1978. Food habits of nesting bald eagles on Kodiak Island,
Alaska. Murrelet 59: 70-72.
Halbrook, R.S., Jenkins, J.H., Bush, P.B., and Seabolt, N.D. 1994. Sublethal concentrations of
mercury in river otters: monitoring environmental contamination. Arch. Environ. Contam.
Toxicol. 27: 306-310.
Hall, E. R. 1981. The Mammals of North America. 2nd ed. New York, NY: John Wiley and
Sons; 1,181 pp.
Harding, A. R. 1934. Mink Trapping. A. R. Harding, Columbus, OH; 171 pp.
Heath, J.A., and Frederick, P.C. 2005. Relationships among mercury concentrations,
hormones, and nesting effort of white ibises (Eudocimus albus) in the Florida Everglades. Auk
122: 255-267.
Heinz, G.H. 1974. Effects of low dietary levels of methyl mercury on mallard reproduction. Bull.
Environ. Contam. Toxicol. 11: 386-392.
Heinz, G.H. 1975. Effects of methylmercury on approach and avoidance behavior of mallard
ducklings. Bull. Environ. Contam. Toxicol. 13: 554-564.
J-64
-------
Heinz, G.H. 1976a. Methylmercury: second-year feeding effects on mallard reproduction and
duckling behavior. J. Wildl. Manage. 40: 82-90.
Heinz, G.H. 1976b. Methylmercury: Second-generation reproductive and behavioral effects on
mallard ducks. J. Wildl. Manage. 40: 710-715.
Heinz, G.H. 1979. Methylmercury: reproductive and behavioral effects on three generations of
mallard ducks. J. Wildl. Manage. 43: 394-401.
Hoffman, D.J., Rice, C.P., and Kubiak, T.J. PCBs and dioxins in birds. 1996. In: Beyer, W.N.,
Heinz, G.H., and Redmon-Norwood, A.W. (eds.) Environmental Contaminants in Wildlife:
Interpreting Tissue Concentrations. Society of Environmental Toxicology and Chemistry
(SETAC) Special Publications Series. Boca Raton, FL: CRC Press/Lewis Publishers; pp. 165-
207.
Huntington, E. H. and A.A. Roberts. 1959. Food Habits of the Merganser in New Mexico. New
Mexico Dep. Game Fish. Bull. No. 9.
Keith, J.O. 1996. Residue analyses: how they were used to assess the hazards of
contaminants to wildlife. In: W.N. Beyer, G. Heinz, and A.W. Redmon-Norwood (eds.)
Environmental Contaminants in Wildlife: Interpreting Tissue Concentrations. A Special
Publication of SETAC. Boca Raton, FL: Lewis Publishers; pp. 1-47.
Kozie, K. D. and R.K. Anderson. 1991. Productivity, diet, and environmental contaminants in
bald eagles nesting near the Wisconsin shoreline of Lake Superior. Arch. Environ. Contam.
Toxicol. 20: 41-48.
Lagler, K. F. and B. T. Ostenson. 1942. Early spring food of the otter in Michigan. J. Wildl.
Manage. 6: 244-254.
Latta, W. C. and R.F. Sharkey. 1966. Feeding behavior of the American merganser in captivity.
J. Wildl. Manage. 30: 17-23.
Lauhachinda, V. 1978. Life history of the river otter in Alabama with emphasis on food habits
[Ph.D. dissertation]. Auburn, AL: University of Alabama.
Liers, E. E. 1951. Notes on the river otter (Lutra canadensis). J. Mammal. 32: 1-9.
Linscombe, G., N. Kinler, R.J. Aulerich. 1982. Mink. In: Chapman, J. A.; Feldhammer, G. A.,
(eds.), Wild Mammals of North America. Baltimore, MD: Johns Hopkins University Press; pp.
329-643.
Mallory, M. and K. Metz. 1999. Common merganser (Mergus merganser). In: Poole, A.; Gill,
F., eds., The Birds of North America, No. 442. Philadelphia, PA: The Birds of North America,
Inc. Retrieved from the BNA Online: http://bna.birds.cornell.edu/bna/species/442.
Marshall, W.H. 1936. Notes on live-trapping mink. J. Mammal. 30: 416-423.
McCaw, J. H., Ill, P.J. Zwank, R.L. Steiner. 1996. Abundance, distribution, and behavior of
common mergansers wintering on a reservoir in southern New Mexico. J. Field Ornithol. 67:
669-679. (As cited in Mallory and Metz 1999.)
J-65
-------
Melquist, W. E. and M.G. Hornocker. 1983. Ecology of river otters in west central Idaho. In:
Kirkpatrick, R. L., ed. Wildlife Monographs: Vol. 83. Bethesda, MD: The Wildlife Society; 60 pp.
Mierle, G., Addison, E.M., MacDonald, K.S., and Joachim, D.G. 2000. Mercury levels in tissues
of otters from Ontario, Canada: variation with age, sex, and location. Enviorn. Toxicol. Chem.
19: 3044-3051.
Mitchell, J. L. 1961. Mink movements and populations on a Montana river. J. Wildl. Manage.
25: 48-54.
Mount, D.I. 1973. Chronic effect of low pH on fathead minnow survival, growth and
reproduction. Water Res. 7: 987. (As cited in USEPA 1976.)
Murray, F.J., F.A. Smith, K.D. Nitschke, C.G. Huniston, R.J. Kociba and B.A. Schwetz. 1979.
Three generation reproduction study of rats given 2,3,7,8-tetrachlorodobenzo-p-dioxin( TCDD)
in the diet. Toxicol. Appl. Pharmacol. 50:241-252.
Nagy, K. A. 1987. Field metabolic rate and food requirement scaling in mammals and birds.
Ecol. Mono. 57: 111-128.
Nagy, K. A., I. A.Girard, T.K. Brown. 1999. Energetics of free-ranging mammals, reptiles and
birds. Annual Review of Nutrition 19: 47-277.
NAS/NAE. 1974. Water Quality Criteria - 1972. National Academy of Sciences, National
Academy of Engineering. Washinton, DC: U.S. Government Printing Office.
New York State Fish and Wildlife Department (NYS FWD). 1971. Fishery Survey of Alcove
Reservoir.
New York State Department of Environmental Conservation (NYS DEC). 2008a. Personal
communication between Daniel Zielinski, NYS DEC, and Leiran Biton, ICF International, March
12.
NYS DEC. 2008b. Printout of fish survey results from 1988 through 2006 for Kinderhook and
Nassau Lakes, forwarded by Norman R. McBride, NYDEC, to Leiran Biton, ICF International,
March 3.
Nosek, J.A., J.R. Sullivan, S.S. Hurley, J.R. Olson, S.R. Craven, R.E. Peterson. 1992a.
Metabolism and disposition of 2,3,7,8-tetrachlorodibenzo-p-dioxinin ring-necked pheasant hens,
chicks, and eggs. J. Toxicol. Environ. Health 35:153-164.
Nosek, J.A., J.R. Sullivan, S.S. Hurley, S.R. Craven, R.E. Peterson. 1992b. Toxicity and
reproductive effects of 2,3,7,8-tetrachlorodibenzo-p-dioxitno xicity in ring-necked pheasant
hens. J. Toxicol. Environ. Health 35157-198.
Nosek, J.A., J.R. Sullivan, T.E. Amundson, S.R. Craven, L.M. Miller, A.G. Fitzpatrick, M.E.
Cook, R.E. Peterson. 1993. Embryotoxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the ring-
necked pheasants. Environ. Toxicol. Chem. 12:1215-1222.
Palmer, E. L. and H. S. Fowler. 1975. Fieldbook of Natural History. New York, NY: McGraw-
Hill Book Co.
Quinney, T.E. and C.D. Ankney. 1985. Prey size selection by Tree Swallows. Auk 102: 245-250.
J-66
-------
Ross, R. K. 1987. Interim report on waterfowl breeding pair surveys in northern Ontario, 1980-
1983. Can. Wildl. Serv. Prog. Rep. no. 168. (As cited in Mallory and Metz 1999.)
Rice, C.P., and P. O'Keefe. 1995. Sources, pathways, and effects of PCBs, dioxins, and
dibenzofurans. In: D.J. Hoffman, B.A. Rattner, G.A. Burton, Jr., and J. Cairns, Jr. (eds.)
Handbook of Ecotoxicology. Boca Raton, FL: Lewis Publishers of CRC Press, Inc.; pp. 424-468.
Salyer, J. C. and K.F. Lagler. 1940. The food and habits of the American merganser during
winter in Michigan, considered in relation to fish management. J. Wildl. Manage. 4: 186-219.
Saywer, C.N. 1960. Chemistry for Sanitary Engineers. New York, NY: McGraw-Hill. (As cited
in EPA 1986.)
Sheffy, T.B., and J.R. St. Amant. 1982. Mercury burdens in furbearers in Wisconsin. J. Wildl.
Manage. 46: 1117-1120.
Sjoberg, K. 1988. Food selection, food-seeking patterns and hunting success of captive
goosanders Mergus merganser and red-breasted mergansers M. serrator in relation to the
behaviour of their prey. Ibis 130: 79-93. (As cited in Mallory and Metz, 1999.)
Snyder, N. F. and J.W. Wiley. 1976. Sexual size dimorphism in hawks and owls of North
America. Orni. Monogr. 20.
Sparling, D.W. 1995. Acidic deposition: a review of biological effects. In: D.J. Hoffman, B.A.
Rattner, G.A. Burton, Jr., and J. Cairns, Jr. (eds.) Handbook of Ecotoxicology. Boca Raton, FL:
Lewis Publishers of CRC Press, Inc.; pp. 301-329.
Stalmaster, M. V. and J.A. Gessaman. 1984. Ecological energetics and foraging behavior of
overwintering bald eagles. Ecol. Monogr. 54: 407-428.
Stanne. S.P., R.G. Panetta, B.E. Froist. 1996. The Hudson, an Illustrated Guide to the Living
River. Rutgers University Press. New Brunswick, NJ.
Thompson, D.R. 1996. Mercury in birds and terrestrial mammals. In: Beyer, W.N., Heinz,
G.H., and Redmon-Norwood, A.W. (eds.) Environmental Contaminants in Wildlife: Interpreting
Tissue Concentrations. Society of Environmental Toxicology and Chemistry (SETAC) Special
Publications Series. Boca Raton, FL: CRC Press/Lewis Publishers; pp. 341-356.
Timken, R. L. and B.W. Anderson. 1969. Food habits of common mergansers in the
northcenteral USA (Mergus merganser) fish. J. Wildl. Manage. 33: 87-91.
Todd, C.S., L.S. Young, R.B. Owen Jr., F.J. Gramlich. 1982. Food habits of bald eagles in
Maine. J. Wildl. Manage. 46: 636-645.
Toweill, D. E. and J.E. Tabor. 1982. River otter. In: Chapman, J. A.; Feldhammer, G. A., eds.
Wild mammals of North America. Baltimore, MD: Johns Hopkins University Press; pp. 688-703.
U.S. Department of Agriculture (USDA). 1998. Soil Quality Information Sheet. Soil Quality
Indicators: pH. Natural Resources Conservation Service. Available at:
http://soils.usda.qov/sqi/publications/files/indicate.pdf.
U.S. Department of Agriculture (USDA). 2008. Web Soil Survey. Soil Survey Staff, Natural
Resources Conservation Service. Available at: http://websoilsurvev.nrcs.usda.gov/.
J-67
-------
U.S. Environmental Protection Agency (EPA). 1976. Quality Criteria for Water. Office of Water
and Hazardous Materials, Water Planning and Standards, Washington, DC. NTIS PB-263-943.
July. ("Red Book") Available at: http://www.epa.qov/waterscience/criteria/redbook.pdf.
U.S. Environmental Protection Agency (EPA). 1986. Quality Criteria for Water. Office of Water
Regulations and Standards Washington, DC. EPA 440/5-86-001. May. ("Gold Book")
Available at: http://www.epa.aov/waterscience/criteria/goldbook.pdf.
U.S. Environmental Protection Agency (EPA). 1992. Hazard Ranking System Guidance Manual,
Appendix A, Sensitive Environments. Office of Solid Waste and Emergency Response,
Washington, DC. EPA/540/R/92/026.
U.S. Environmental Protection Agency (EPA). 1993a,b. Wildlife Exposure Factors Handbook,
Volumes I and II. Office of Research and Development, Washington, DC. EPA/600/R-93/187a
and b.
U.S. Environmental Protection Agency (EPA). 1995a. Great Lakes Water Quality Initiative
Criteria Documents for the Protection of Wildlife: DDT; Mercury; 2,3,7,8-TCDD; PCBs. Office of
Water, Office of Science and Technology, Washington, DC. EPA/820/B-95/008. March.
U.S. Environmental Protection Agency (EPA). 1995b. Final water quality guidance for the Great
Lakes system: Final Rule. Fed. Regist. 60(56): 15366-15425. March 23.
U.S. Environmental Protection Agency (EPA). 1997. Mercury Study Report to Congress.
Volume VI: An Ecological Assessment for Antrhopogenic Mercury Emissions in the United
States. Office of Air Quality Planning and Standards and Office of Research and Development.
EPA-452/R-97-008. December.
U.S. Environmental Protection Agency (EPA). USEPA. Guidelines for Ecological Risk
Assessment. Risk Assessment Forum, Washington, DC. EPA/630/R-95/002F. April.
U.S. Environmental Protection Agency (EPA). Water Quality Criterion for the Protection of
Human Health: Methylmercury. Office of Water, Washington, DC. January. EPA 823-R-01-001.
Available at: http://www.epa.aov/waterscience/criteria/methvlmercurv/pdf/mercurv-criterion.pdf.
U.S. Environmental Protection Agency (EPA). 2005a. National Emissions Inventories for the
U.S., Version 2. Available at: http://www.epa.gov/ttn/chief/net/.
U.S. Environmental Protection Agency (EPA). 2005b. Trophic Level and Exposure Analyses for
Selected Piscivorous Birds and Mammals - Volume I: Analyses of Species in the Great Lakes
Basin; Volume II: Analyses of Species in the Conterminous United States; and Volume III:
Appendices. Draft Final. Office of Water, Office of Science and Technology, Washington, DC.
U.S. Environmental Protection Agency (EPA). 2005c. Evaluation of TRIM.FaTE Volume II:
Model Performance Focusing on Mercury Test Case. Office of Air Quality Planning and
Standards, Emissions Standards and Air Quality Strategies and Standards Division, Research
Triangle Park, North Carolina. EPA-453/R-05-002. July.
U.S. Environmental Protection Agency (EPA). 2006. Risk and Technology Review (RTR)
Assessment Plan, Draft for EPA Science Advisory Board Review. Office of Air and Radiation,
Research Triangle Park, November 20. Available at: http://www.epa.gov/ttn/atw/rrisk/rtrpg.html.
J-68
-------
U.S. Environmental Protection Agency (EPA). 2008. STORET Database. Available at:
http://www.epa.gov/storet/index.html.
U.S. Environmental Protection Agency (EPA). 2009. Guidance for Implementing the January
2001 Methylmercury Water Quality Criterion. Office of Water, Office of Science and Technology.
Washington, D.C. January. EPA 823-R-09-002. Available at:
http://www.epa.aov/waterscience/criteria/methvlmercurv/guidance-final.html
White, H. C. 1936. The food of kingfishers and mergansers on the Margaree River, Nova
Scotia. J. Biol. Board Can. 2: 299-309.
White, H. C. 1937. Local feeding of kingfishers and mergansers. J. Biol. Board Can. 3:
323-338.
White, H. C. 1957. Food and natural history of mergansers on salmon waters in the Maritime
provinces of Canada. Bull. Fish. Res. Board Canada no. 116.
Williams, J.B. 1988. Field metabolism of tree swallows during the breeding season. Auk'\05:
706-714.
Wobeser, G.A. 1973. Ph.D. Dissertation. Aquatic Mercury Pollution: Studies of its occurrence
and pathologic effect on fish and mink. University of Saskatchewan (Canada). Disseration
Number 73-24, 819. University Microfilms, Ann Arbor, Ml.
Wobeser, G., N.D. Nielsen, B. Schiefer. 1976a. Mercury and mink I: the use of mercury
contaminated fish as a food for ranch mink. Can. J. Comp. Med. 40:30-33.
Wobeser, G., N.D. Nielsen, B. Schiefer. 1976b. Mercury and mink II: experimental methyl
mercury intoxication. Can. J. Comp. Med. 40:34-45.
Wolfe, M., and Norman, D. 1998. Effects of waterborne mercury on terrestrial wildlife at Clear
Lake: evaluation and testing of a predictive model. Environ. Toxicol. Chem. 17: 214-227.
Wolfe, M.F., Schwarzbach, S., and Sulaiman, R.A. 1998. Effects of mercury on wildlife: a
comprehensive review. Environ. Toxicol. Chem. 17: 146-160.
Wolfe, M.F., T. Atkeson, W. Bowerman, J. Burger, D.C. Evers, M.W. Murray, and E. Zillioux.
2006. Wildlife indicators. In: R. Harris, D.P. Krabbenhoft, R. Mason, M.W. Murray, R. Reash,
and T. Saltman (eds.) Ecosystem Responses to Mercury Contamination: Indicators of Change.
Pensacola, FL: Society of Environmental Toxicology and Chemistry (SETAC) Press; pp. 123-
190.
Wren, C.D., P.M. Stoikes, and K.L. Fischer. 1986. Mercury levels in Ontario mink and otter
relative to food levels and environmental acidification. Can. J. Zool. 64: 2854-2859.
Yates, D., Evers, D.C., and Savoy, L. 2004. Developing a Mercury Exposure Profile for Mink
and River Otter in Maine. Report BRI 2004-09 submitted to Maine Department of
Environmental Protection and Maine INIsand Fisheries and Wildlife. Gorham, ME: BioDiversity
Research Institute; 24 pp.
J-69
-------
ATTACHMENT J-1: Ecological Risk Assessment Case Study Supporting
Documents
-------
[This page is intentionally left blank.]
-------
TABLE OF CONTENTS
Exhibit 1. Background Acid Precipitation Monitoring Data from USGS 1
Exhibit 2. Total Alkalinity of Surface Waters of the Conterminous United States 2
Exhibit 3. Ecosystem Background Deposition, Alkalinity, Facility Emission, and Preliminary
Ecological Hazard Scores for Portland Cement Facilities Emitting HCI 3
Exhibit 4. Location of Portland Cement Facilities Relative to Surface Water Alkalinity in the
Conterminous United States 8
Exhibit 5. Location of Ravena Facility Modeling Domain Relative to Surface Water Alkalinity... 9
Exhibit 6. HCI Concentrations in Air Estimated from HCI Emissions from Ravena Facility
Relative to Nearest Large Parks and Preserves 10
Exhibit 7. HCI Concentrations in Air Estimated from HCI Emissions from the Ravena Facility
Relative to Nearby Smaller Valued Parks and Other Areas 11
j-1-i
-------
Exhibit 1. Background Acid Precipitation Monitoring Data from USGS
Hydrogen ion concentration as pH from measurements
made at the Central Analytical Laboratory, 2006
*
' k" 5.2 -
• * "' fKjk
¦- 5.2 * 54 S 3 « - * r&L- I 48
6.3 56
»/ h v - * J
- . "~t" , *" ^
• 5.6 * " 51 )4S ^ jc" 4.6 46—
5.5 "7 I 5.0 J s
"s W--74H.5 i "
a® Sff m r -f.b . * * -if V» 7 ^ 1
55 _ V ',, 48-4.7 46-
v V •*' ( * 46 ut*srW§ ;
» - -• w- v 1 -m
a j
ltT " 5.7 * 46 4.5 *5 __.>4 6 *
„(3-0 , a 4.7 . '* - '~4 4 .~ »
A . *$?-¦ ,V
j j j. a. I ' J. -
* 54 r.. L $!(>,„ 5 8
54 *
53 5.4 5.1 * * ->/• ""
.Jj-is.-jrj ty-*« UfrpN
»¦ " • " .' • .» .« ' ¦)«)
Si
~$
S3
55 5.0 > •
•
5.3
*
*
5.1
«V )*&
®7
r 4 * ^
5.Q
i _
•
yi Y- •.- , £fr; „^"S
52
¦ fY^.J ^ u «%¦
40 4... fl* «>..
^—4.9 LST*7 fl-64 6,^_-^—--4S *
•JLJ-
47\ V 5v£"5i3i
53 V 55 _. *5 I V Mf.1.5,2
¦~.» 4el $£•>$. 1
53 I fc 48 k / «l J * *jf
fitpa m* pioturpdL! \ {_ JS 4.7* . \ 4J?-4,9
WW U ,. v M ( "i;-A^v-'N «.«
«w* a,a '¦r^m ' \Mr «-w
PR2R M W V *,r' : u \ «'te
vmi *» 52/ \ \ 4.t.ts
\ 51 *L3»4.4
w i «*,?
Nsttefral fttmw ptaw fteretifcn Pwgrartff^tsin## Trentf9 Netwerfc
ttt^rM^.muiuo, tdu
j-1-1
-------
James M. Omernik1, Glenn E.Griffith2,
Jeffrey T. Irish \ and Colleen B. Johnson
in small-scale maps. This map was compiled using
picture of surface water alkalinity.
precise i
Omernik,
I C.F.. Powers. 1983. Total alkalinity i
Annals of the Association
Exhibit 2. Total Alkalinity of Surface Waters of the Conterminous United States
Total Alkalinity of Surface Waters
Albers Equal Area Projection
Total
Alkalinity
(c*eq/l)
¦ 50-100
~ 100-200
~ 200-400
~ >400
1 Environmental Research Laboratory
U.S. Environmental Protection Agency
Corvallis, Oregon 97333
Available from http://qeodata.epa.qov/WAF/Total%20Alkalinity%20of%20Surface%20Waters%20of%20the%20US.xml
J-1-2
-------
Exhibit 3. Ecosystem Background Deposition, Alkalinity,
Facility Emission, and Preliminary Ecological Hazard Scores for Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Acid
Deposition
(pH of
Rainfall)
Background
Exposure
Score
Surface
Water
Alkalinity
(meq/L)
Alkalinity
Score
Emissions
(TPY)
Emissions
Score
Preliminary
Hazard Score
PTC_NEI34931
Albany County,
NY
4.5-4.6
2
>400
5
71.08
1
10
PTC_NEI26327
Hernando
County, FL
4.6-4.7
2
>400
5
48.48
1
10
PTC_NEI33394
Carroll County,
MD
4.5-4.6
2
>400
5
180.00
1
10
PTC_NEISC0351244
Dorchester
County, SC
4.6-4.7
2
>400
5
97.53
1
10
PTC_NEIMIB1559
Charlevoix
County, Ml
4.7-4.8
3
>400
5
323.47
1
15
PTC_NEI51435
La Salle County,
IL
4.8-4.9
3
>400
5
39.70
1
15
PTC_NEI12018
Alpena County,
Ml
4.7-4.8
3
>400
5
474.56
1
15
PTC_NEIPAT$1626
Lawrence
County, PA
4.4-4.5
1
>400
5
10.90
3
15
PT C_N EIM00990002
Jefferson County,
MO
4.8-4.9
3
>400
5
72.15
1
15
PTC_NEI51352
La Salle County,
IL
5.0-5.1
4
>400
5
32.69
1
20
PTC_NEI31319
Clark County, IN
4.6-4.7
2
>400
5
18.86
2
20
PTC_NEI7255
Northampton
County, PA
4.5-4.6
2
>400
5
16.85
2
20
J-1-3
-------
Exhibit 3. Ecosystem Background Deposition, Alkalinity,
Facility Emission, and Preliminary Ecological Hazard Scores for Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Acid
Deposition
(pH of
Rainfall)
Background
Exposure
Score
Surface
Water
Alkalinity
(meq/L)
Alkalinity
Score
Emissions
(TPY)
Emissions
Score
Preliminary
Hazard Score
PTC_NEI2CA151186
Santa Cruz
County, CA
>5.3
5
<50
1
4.79
4
20
PTC_NEI12238
Scott County, IA
5.2-5.3
5
>400
5
76.60
1
25
PTC_NEI26277
Miami-Dade
County, FL
5.1-5.2
5
>400
5
63.17
1
25
PTC_NEI22838
San Bernardino
County, CA
5.1-5.2
5
>400
5
38.89
1
25
PTC_NEIPA01993-1
Butler County,
PA
4.4-4.5
1
>400
5
0.43
5
25
PTC_NEI52351
Massac County,
IL
4.7-4.8
3
>400
5
32.49
2
30
PTC_NEIAL1150002
St. Clair County,
AL
4.7-4.8
3
>400
5
28.11
2
30
PTC_NEI32033
Lawrence
County, IN
4.6-4.7
2
>400
5
13.57
3
30
PTC_NEI2PA110039
Berks County,
PA
4.5-4.6
2
>400
5
11.74
3
30
PTC_NEIVA2553
Botetourt County,
VA
4.5-4.6
2
>400
5
8.54
3
30
PTC_NEIKYR0060
Jefferson County,
KY
4.5-4.6
2
>400
5
8.20
3
30
PTC_NEIAL8026
Mobile County,
AL
4.6-4.7
2
200-400
4
4.04
4
32
PTC_NEITX139099J
Ellis County, TX
4.9-5.0
4
>400
5
18.53
2
40
PT C_N EIPA94-2626
Northampton
County, PA
4.5-4.6
2
>400
5
4.27
4
40
J-1-4
-------
Exhibit 3. Ecosystem Background Deposition, Alkalinity,
Facility Emission, and Preliminary Ecological Hazard Scores for Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Acid
Deposition
(pH of
Rainfall)
Background
Exposure
Score
Surface
Water
Alkalinity
(meq/L)
Alkalinity
Score
Emissions
(TPY)
Emissions
Score
Preliminary
Hazard Score
PTC_NEI33699
Washington
County, MD
4.5-4.6
2
>400
5
3.80
4
40
PTC_N EIF LR001008
Alachua County,
FL
4.8-4.9
3
>400
5
8.60
3
45
PTC_NEI51527
Lee County, IL
4.8-4.9
3
>400
5
8.02
3
45
PTC_NEI22877
San Bernardino
County, CA
5.1-5.2
5
>400
5
32.14
2
50
PTC_NEI16357
Montgomery
County, KS
>5.3
5
>400
5
31.01
2
50
PTC_N E11A0330035
Cerro Gordo
County, IA
5.1-5.2
5
>400
5
22.49
2
50
PTC_NEI25375
Shasta County,
CA
>5.3
5
>400
5
20.00
2
50
PTC_NEI20046
Kern County, CA
>5.3
5
>400
5
19.53
2
50
PTC_NEI12739
Allen County, KS
>5.3
5
>400
5
19.50
2
50
PTC_NEIPA58-1290
Lehigh County,
PA
4.5-4.6
2
>400
5
2.15
5
50
PT C_N EITN0653070
Hamilton County,
TN
4.6-4.7
2
>400
5
1.98
5
50
PTC_NEITX309123F
McLennan
County, TX
5.0-5.1
4
>400
5
7.68
3
60
PTC_NEIMIB1743
Monroe County,
Ml
4.6-4.7
3
>400
5
3.83
4
60
PTC_NEIAL321
Marengo County,
AL
4.8-4.9
3
>400
5
2.52
4
60
J-1-5
-------
Exhibit 3. Ecosystem Background Deposition, Alkalinity,
Facility Emission, and Preliminary Ecological Hazard Scores for Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Acid
Deposition
(pH of
Rainfall)
Background
Exposure
Score
Surface
Water
Alkalinity
(meq/L)
Alkalinity
Score
Emissions
(TPY)
Emissions
Score
Preliminary
Hazard Score
PTC_NEI18621
Pima County, AZ
5.2-5.3
5
>400
5
16.55
3
75
PTC_NEI13290
Comal County,
TX
5.1-5.2
5
>400
5
14.22
3
75
PTC_NEI34326
Jackson County,
MO
5.2-5.3
5
>400
5
14.16
3
75
PTC_NEITXT$11924
Hays County, TX
5.1-5.2
5
>400
5
7.40
3
75
PTC_NEI886
Fremont County,
CO
5.0-5.1
4
>400
5
6.43
4
80
PTC_NEI7376
Ellis County, TX
4.9-5.0
4
>400
5
3.25
4
80
PTC_NEITXT$11872
Bexar County,
TX
5.1-5.2
5
>400
5
6.31
4
100
PTC_NEI12976
Mayes County,
OK
5.2-5.3
5
>400
5
3.56
4
100
PTC_N EIF L0860020
Miami-Dade
County, FL
5.1-5.2
5
>400
5
3.07
4
100
PT C_N EICA1505122
Kern County, CA
>5.3
5
>400
5
2.97
4
100
PTC_NEI338
Albany County,
WA
>5.3
5
>400
5
2.23
5
125
PTC_NEI40539
Baker County,
OR
>5.3
5
>400
5
2.00
5
125
PTC_NEI20130
Kern County, CA
>5.3
5
>400
5
1.96
5
125
PTC_NEINMT$12442
Bernalillo County,
NM
5.1-5.2
5
>400
5
1.95
5
125
PTC_N E11D0050004
Bannock County,
ID
>5.3
5
>400
5
1.78
5
125
J-1-6
-------
Exhibit 3. Ecosystem Background Deposition, Alkalinity,
Facility Emission, and Preliminary Ecological Hazard Scores for Portland Cement Facilities Emitting HCI
Facility
Facility
Location
Acid
Deposition
(pH of
Rainfall)
Background
Exposure
Score
Surface
Water
Alkalinity
(meq/L)
Alkalinity
Score
Emissions
(TPY)
Emissions
Score
Preliminary
Hazard Score
PTC_NEI446
Boulder County,
CO
5.1-5.2
5
>400
5
1.51
5
125
PTC_NEITXT $11980
Nolan County,
TX
>5.3
5
>400
5
0.80
5
125
PTC_NEI22743
San Bernardino
County, CA
5.1-5.2
5
>400
5
0.33
5
125
PTC_NEI22453
Riverside
County, CA
5.1-5.2
5
>400
5
0.17
5
125
J-1-7
-------
Exhibit 4. Location of Portland Cement Facilities Relative to Surface Water Alkalinity in the Conterminous United States
O The 91 PTC Facilities
Total Alkalinity
J-1-8
-------
Exhibit 5. Location of Ravena Facility
Modeling Domain Relative to Surface Water Alkalinity
Total Alkalinity
(eq/l)
| <50
| 50-100
| 100-200
| 200-400
>400
20 Kilometers
J-1-9
-------
Exhibit 6. HCI Concentrations in Air Estimated from
HCI Emissions from Ravena Facility Relative to Nearest Large Parks and Preserves
AdirdrBdaok- F4;
7
Concentrations (ug/cu-m)
Q0.002 -0.003
Pjo.004 -0.005
P]0.006 -0.007
[JO.OOS -0.009
[]0.01 -0.012
[]0.013 -0.014
[]0.015 -0.016
0.017 -0.018
0.019 -0.02
[3>.021 -0.022
[Ho.023 -0.025
0.026 -0.027
_|0.028 -0.029
|0.03 -0.031
0.032 -0.033
|0.034 -0.036
0.037 -0.04
§0.041 -0.044
0.045 -0.049
|0.05 -0.053
§0.054 -0.06
§0.061 -0.066
§0.067 -0.075
y0.076 -0.088
(0.089 -0.11
|0.111 -0.14
§0.141 -0.184
§0.185 -0.236
Bo.237 -0.548
Cockapon^it^t^te Forest
CockaponsVt^^ite, Forest
Cock^ponsefrSpfte forest
4
50 Kilometers
Cats kill State
ft
J-1-10
-------
Exhibit 7. HCI Concentrations in Air Estimated from HCI Emissions from the Ravena
Facility Relative to Nearby Smaller Valued Parks and Other Areas
John Boyd Thacher State Park
Grafton Lakes State Park
Taconic
ail State Partt
% fllf
Mount Greylock State Pari*
A
Cherry Plain State Patf
Hudso
River
State Park
Rogers Island Game MGMT Area
01 ana Historic Site
Lake Taghkanic State Park
Cats kill State Park
10
20 Kilometers
Natl, State, and Local Forests and P arks
HCI Concentrations
(ug/m3)
-0.008 -0.001
I | 0.002-0.003
~j 0.004 -0.005
] 0.006 -0.007
0.008 -0.009
~ 0.01 -0.012
] 0.013-0.014
| 10.015-0.016
| 10.017-0.018
~ | 0.019-0.02
j 0.021 -0.022
H 0^023-0.025
0 026 - 0 027
0.028 -0.029
H 0.03-0.031
| 0.032 -0.033
if 0.034-0.038
Hi 0 0:39 "° 042
3 0.043 -0.046
0 047 "0 ¦051
B 0.052 -0.057
fU 0.058 -0.064
H 0.065 -0.073
¦ 0.074-0.088
0.089-0.116
0.117-0.16
¦I 0.161 -0.223
I 0.224-0.548
J-1-11
-------
Appendix K
Development of a threshold concentration for foliar damage caused by
ambient hydrogen chloride concentrations
K-l
-------
Table of Contents
K.l Introduction 3
K.2 Summary of Studies 3
K.2.1 Phytotoxicity of Hydrogen Chloride Gas with a Short-Term Exposure 3
K.2.2 Foliar and Microscopic Observations of Bean Leaves Exposed to Hydrogen
Chloride Gas 6
K.2.3 Reversible Fine Structural Alterations of Pinto Bean Chloroplasts Following
Treatment with Hydrogen Chloride Gas 8
K.2.4 Peroxidase Activity in Plant Leaves Exposed to Gaseous HC1 or Ozone 9
K.2.5 Photosynthesis and Respiratory Consequences of Hydrogen Chloride Gas
Exposures of Phaseolus Vulgaris L. and Spinacea Oleracea L 9
K.2.6 Histological Effects of Aqueous Acids and Gaseous Hydrogen Chloride on Bean
Leaves 11
K.2.7 The Phytotoxicity of Designated Pollutants on Plant Species 12
K.3 Development of Gaseous HC1 Ecological Exposure Thresholds 12
K.3.1 Recommended Methods for Developing Ecological Exposure Thresholds 14
K.3.2 Development of Screening-Level HC1 Ecological Thresholds 16
K.3.3 Comparison of Modeled HC1 Air Concentration Estimates to Screening Level
Ecological Thresholds 18
K.4 References 19
K-2
-------
K.1 Introduction
Exposure thresholds have been developed for many chemicals for exposure of plants and animals
to media such as water, soil and sediments. Screening level ecological exposure thresholds are
frequently compared to measured or modeled media concentrations to determine whether a full
ecological risk assessment is required. Unfortunately, few exposure thresholds have been
developed for direct air exposures, and risk assessments done under the air toxics program do not
routinely assess these exposures. As a result, in order to develop a case study for direct plant
exposure to a hazardous air pollutant we were required to develop an exposure threshold for use
in this assessment. A literature search was conducted through several university libraries in the
attempt to locate information that could be used in developing screening-level hydrogen chloride
(HCl) ecological exposure thresholds for foliar damage. Over 50 scientific databases were
accessed in the literature search, yielding the studies described below.
In the late 1970's and early 1980's, a series of studies were conducted by the Statewide Air
Pollution Research Center of the University of California at Riverside to determine the impact of
gaseous hydrogen chloride (HCl) on plants. These experiments, designed to examine the effects
of massive, nearly instantaneous, releases of HCl on vegetation, and the subsequent journal
articles, were supported by grants from the U.S. Air Force Office of Scientific Research to assess
the potential damage to plants, as compared to controls, from short-term exposures to high
concentrations of gaseous HCl in the exhaust from some types of solid-fuel rockets.
In addition to the studies at the University of California at Riverside, the Air Force summarized
supported studies done at the University of California at Irvine in the document The
Phytotoxicity of Designated Pollutants on Plant Species (USAF, 1983). The designated
pollutants are HCl and aluminum oxide emitted from the solid rocket fuel used in the rockets that
launch the space shuttle.
These exposure conditions could also be characteristic of mass releases from spills or equipment
failures. EPA is interested in determining how the results of these studies could be used to
develop gaseous HCl ecological exposure thresholds to compare to short-term average (1 to 24
hour) and long-term average (annual) air concentrations of gaseous HCl from routine industrial
releases to estimate the potential for them to cause foliar damage. The studies are summarized
below.
K.2 Summary of Studies
K.2.1 Phytotoxicity of Hydrogen Chloride Gas with a Short-Term Exposure
In the introduction to this journal article, Lerman, et. al. (1976) provided a synopsis of research
on damage from gaseous HCl plant exposures from the early 1900's on. This information is
presented in table form in Table K-l.
The purpose of this study was to determine the concentration of HCl required to induce
morphological injury to eight types of ornamental plants: aster, calendula, cornflower, cosmos,
American marigold, French marigold, nasturtium, and zinnia. Several ages of plants were
K-3
-------
exposed for 20 minutes to gaseous HC1 concentrations ranging from 1 to 35 mg/m3.1 The degree
of injury to the plants was evaluated 24 and 48 hours after exposure by external examination
using an arbitrary 1 to 10 scale. The evaluation was based on number of injured leaves per plant,
estimated percentage of foliar surface affected, and the overall appearance of the plant.
Table K-l. Synopsis of Early Research on Damage from Gaseous HC1 Plant Exposures
Exposure
Results
Reference
Single exposure, 2
day duration at 5-20
ppm (8-30 mg/m3)
Seedlings of viburnum and larch killed in less than 2
days
Haselhoff
and Lindau,
1903
Single exposure, 1
hour duration at
1,000 ppm (1,500
mg/m3)
Bleached lesions on leaves of fir, birch and oak
80 exposures,
duration 1 hour/day
at 2,000 ppm (3,000
mg/m3)
Necrosis on margins of maple, birch and pear tree
leaves
Single exposure, 2
hour duration at 5
ppm (8 mg/m3)
28 day old tomato plants developed interveinal
bronzing followed by necrosis within 72 hours
Shriner and
Lacasse,
1969
Single exposure, 4
hour duration at 3-
43 ppm (5-65
mg/m3)
2-5 year old seedlings, 12 types of coniferous and
broadleaf trees:
- Most sensitive Liriodendron tulipfera (tulip tree)
visible injury at the lowest concentration of 3 ppm (5
mg/m3)
- Most sensitive conifer Pinus strobes (white pine)
visible injury at 8 ppm (12 mg/m3)
- No injury to Thuja occidentalis (white cedar) at 43
ppm (65 mg/m3)
Means and
Lacosse,
1969
Single exposure, 5
min. duration at 95,
300, and 2,071 ppm
(140, 460, and
3,150 mg/m3)
Mature, flowering marigold plants:
95 ppm - little or no visible injury
300 ppm - temporary wilting and leaf spots
2,071 ppm - severe wilting, marginal and interveinal
leaf necrosis, stem collapse, and death
Lind and
London,
1971
208 hours of
exposure within 2
weeks at 1.6 mg/m3
(1 ppm)
Slight necrosis and chlorosis on Spinacia oleracea L.
(spinach)leaves
Mausch
et.al., 1973
Single exposure, 29
hour duration, 0.5
mg/m3 (0.3 ppm)
Carrots:
- Exposed 45 days after germination showed 32.2 to
49.7% decrease in crop yield
- Exposed 96 days after germination showed only
5.3% crop yield decrease
Hulensberg,
1974
1 mg HCl/m3 = ppm x 1.52
K-4
-------
Exposure
Results
Reference
- Winter grape only leaf discoloration, radishes no
damage
- Tomato, cucumber and bush bean plants showed leaf
damage and increase in leaf chlorides
- Reduction in yield severe in tomato and slight in
cucumber plants
Source: Lerman et. al., 1976
Table K-2 presents a qualitative summary of injury symptoms as a function of HC1
concentration. Table K-3 presents a comparison of the relative sensitivity of the eight plant
species. The relative sensitivity is expressed as the concentration of gaseous HC1 required to
cause 10% relative injury calculated using first and second order, polynomial type regression
equations.
Table K-2. Injury symptoms on Eight Plant Species 23 Hours After Exposure to HC1 Gas
HC1
Aster
Calendula
Centaurea
Cosmos
Concentration
21-35 mg/m3
Temporary
Temporary
Extensive
Extensive
wilting,
wilting, lower
necrosis, rolling,
necrosis, extensive
extensive
surface,
speckling,
rolling, flower
interveinal
discoloration,
temporary
discoloration,
bronzing on
necrosis. Younger
wilting,
tipburn of sepals.
lower leaf
the leaf the more
discoloration.
surface, necrosis
distal the damage.
of young tissue.
10-20 mg/m3
Interveinal
Bronzing of lower
Discoloration
Tipburn, tip
bronzing on
leaf surface,
along the leaf
rolling.
lower surface,
interveinal
margin, rolling.
trace of
necrosis, marginal
necrosis.
discoloration.
1.5-9 mg/m3
Trace of
Trace of lower
Tipburn.
necrotic spots on
surface bronzing.
young leaves.
HC1
Marigold, Fr.
Marigold, Am.
Nasturtium
Zinnia
Concentration
Dwarf
Sen. Dirksen
21-35 mg/m3
Severe necrosis
Severe necrosis,
Interveinal
Bronzing on basal
of almost all
extensive rolling,
bleached lesions,
leaf portions,
leaves, rolling.
tipburn of sepals
on younger
extensive necrosis
on flowers.
leaves, in
and rolling on rest
addition,
of leaf.
marginal
Occasional petal
bleaching and
necrotic spots.
K-5
-------
rolling.
10-20 mg/m3
Discoloration,
Interveinal
Discoloration,
Speckling,
necrosis of mid-
discoloration of
necrotic
interveinal
aged leaves,
mid-aged leaves,
speckling,
bronzing.
some rolling.
some rolling.
rolling.
1.5-9 mg/m3
Traces of
Traces of necrosis
Traces of
Traces of lower
necrosis or
or discoloration.
discoloration.
surface bronzing.
discoloration.
Source: Lerman et. a/., 1976
Table K-3. Relative Sensitivity of Eight Species of Ornamental Plants to HC1 Gas
Species
-2
Concentration of HC1 Gas (mg/m ) in a 20 Minute
Exposure Required to Cause 10% Relative Injury
Cosmos
6.5
Marigold (French)
8.8
Marigold (American)
9.5
Zinnia
15.3
Nasturtium
15.7
Calendula
16.1
Centaurea
18.3
Aster
29.9
Source: Lerman et. al., 1976
K.2.2 Foliar and Microscopic Observations of Bean Leaves Exposed to Hydrogen
Chloride Gas
In this Endress et.al. (1978) experiment, pinto beans 8 days and 12 days from seeding were
exposed for 20 minutes to gaseous HC1 concentrations ranging from 6 to 54.2 mg/m3. The plants
were evaluated immediately after exposure, and at 30 minutes and 1, 2, 3, and 24 hours after
exposure. The leaves were evaluated for visible effects and for cellular level changes using
microscopy.
The first visible symptom of injury to primary leaves was glazing of the lower leaf surface
followed by injury to the upper leaf surface. Interveinal necrosis and/or rolling of the leaf
occurred as the HC1 concentration increased. These symptoms were similar to those observed
for other species. Table K-4 presents the extent of necrotic lesions observed from the multiple
exposure concentrations.
In sectioned leaf tissue, the glazing appeared to result from collapse of epidermal cells that seem
to result from deformation of both the inner and outer cell walls. A frequently observed
K-6
-------
symptom related to cell collapse was plasmolysis2 of the protoplast. The cytoplasm left its
normal position by retracting from the cell wall. In cases of more severe HC1 stress, irreversible
plasmolysis occurred as well as cell wall collapse. Mesophyll tissue was usually affected to a
lesser extent than the epidermis, with the most common symptom of injury being plasmolysis.
Other changes noted included the formation of vesicles or small vacuoles and the formation of
crystals in mesophyll cells. Table K-5 presents the microscopic cellular injury observed and the
concentrations of gaseous HC1 necessary to invoke the injuries.
Table K-4. Extent of Necrotic Lesions on Primary Leaves of Pinto Bean 24 Hours After
Exposure for 20 Minutes to Varying Concentrations of I
[CI Gas
Treatment Concenration
mg HCl/m3
Average Leaf Area
cm2
Range of Average
Necrotic Area* cm
Range of Average
% Necrotic Area*
0
16.21+4.31**
0
0
6.0
16.63 + 5.31
0
0
11.3
18.00 + 4.34
0
0
17.9
19.49 + 5.95
0.40 + 0.52**
1.6 + 2.1**
(0.1-1.0)
(0.3-4.0)
25
17.99 + 5.90
0.1
0.1
32
16.67 + 6.51
0.57 + 1.00
4.9 + 10.9
(0.4-4.30)
(0.1-45.8)
41.3
13.83 + 6.71
1.20 + 1.66
11.0+ 16.8
(0.08-7.80)
(0.4-78.9)
54.2
11.46 + 6.52
5.69 + 4.25
55.6 + 34.2
(0.10-16.80)
(0.6-96.7)
Calculations of average necrotic area and average percent necrotic area excluded leaves devoid
of necrotic lesions.
**+ Standard deviation. No attempt was made to determine statistical significance between
treatments because of large variabilities within treatments.
Source: Endress et.al., 1978
Table K-5. Microscopic Cellular Injury Symptoms Observed in Sectioned Primary Leaves
of Pinto Bean Following Exposure to Several Concentrations of Gaseous HC1*
Symptom and Location
Immediate Post-
24 Hour Post-
Fumigation
Fumigation
mg HCl/m3
mg HCl/m3
Plasmolysis in
adazial epidermis
>6.0
>6.0
palisade parenchyma
>17.9
>17.9
spongy parenchyma
>17.9
>17.9
abaxial epidermis
>17.9
11.3, 17.9
2 Plasmolysis is the contraction of cells within plants due to the loss of water through osmosis. It is the cell
membrane peeling off of the cell wall and the vacuole collapsing, plasmolysis occurs when a plant cell's membrane
shrinks away from its cell wall. This phenomenon occurs when water is drawn out of the cell and into the
extracellular (outside cell) fluid. The movement of water occurs across the membrane moving from an area of high
water concentration to an area of lower water concentration outside the cell. It is unlikely to occur in nature, except
in severe conditions. http://www.bio-medicine.org/biology-definition/Plasmolvsis/
K-7
-------
Symptom and Location
Immediate Post-
Fumigation
mg HCl/m3
24 Hour Post-
Fumigation
mg HCl/m3
Vacuolar particulates in
adazial epidermis
>6.0
>6.0
palisade parenchyma
6.0-21.1
54.2
spongy parenchyma
6.0, 11.3
54.2
abaxial epidermis
>11.3
41.3, 54.2
Vesiculation in
adazial epidermis
>6.0
>6.0
palisade parenchyma
>6.0
>6.0
spongy parenchyma
>6.0
>6.0
ab axial epidermis
>11.3
>6.0
Chloroplast crystals in
palisade parenchyma
>11.3
54.2
spongy parenchyma
>11.3
54.2
Collapse of
palisade parenchyma
>17.9
>17.9
spongy parenchyma
>17.9
>17.9
Glazing of
adazial epidermis
>21.9
>17.9
ab axial epidermis
>17.9
>6.0
*When primary leaf tissue was sampled immediately after the 20 minute exposure to HC1,
particulates were present in the vacuoles of abaxial epidermal cells from leaves treated with 11.3
or greater mg HCl/m3, but only leaves exposed to 41.3 or 54.2 mg HCl/m3 had abaxial epidermal
cells with vacuolar particulates when the tissue was sampled 24 hours after the HC1 treatment.
Source: Endress et.al., 1978
K.2.3 Reversible Fine Structural Alterations of Pinto Bean Chloroplasts Following
Treatment with Hydrogen Chloride Gas
In this 1979 study, Endress et. al. examined the development of injury to cells following
treatment with HC1 gas and looked for reversible chloroplast alterations. Pinto bean plants that
were 8, 12, and 16 days from seeding were exposed for 20 minutes to 6 to 54.2 mg/m3
concentrations of gaseous HC1 and examined at multiple intervals up to 24 hours. Tissue
samples from the two primary leaves were prepared for electron microscopy.
Chloroplast structure was distinctly modified in all tissue samples. But not in all cells of each
sample. A distinctive feature of the chloroplast appearance was the presence of crystalline
structures. Crystals were not observed in cells treated with 6 mg/m3 nor 54.2 mg/m3 gaseous
HC1. One percent of chloroplasts contained crystals in cells treated with 11.3 mg/m3 with the
frequency of crystals increasing rapidly above that concentration to all chloroplasts containing
crystals after treatment at 41.3 mg/m3 HC1. Recovery of chloroplasts was found when samples
were observed at 0.5, 1, 2, 3 and 4 hours after treatment with 21.1 mg/m3 HC1. The frequency of
crystals declined from 65 percent of chloroplasts containing crystals immediately after treatment,
to 15 percent at 0.5 hours after treatment, and no remaining crystals after 4 hours. Other authors
have hypothesized that crystal formation reflects a generalized stress response. In the sample
K-8
-------
cells characterized by severe plasmolysis, dehydration of chloroplasts and associated crystal
formation irreparable cell membrane damage occurred. It is not clear whether the normal repair
mechanism was inhibited by HC1 exposure or if the magnitude of membrane injury was greater
than the capacity of the repair process.
The authors compared their results from 20 minute exposures to work reported by Masuch et. al.
(1973) on chronic exposures. They reported that chronic exposures of spinach to HC1 gas (0.13
and 0.25 mg/m3 for 43 hours within 5 days and 1.6 mg/m3 for 208 hours within 14 days)
increased the average frequency of changes in the chloroplasts, although they did not report the
presence of crystals.
Altered mitochondrial appearance was one of the most consistent indicators of HC1 exposure in
this study. In tissues sampled at longer periods after exposure, the mitochondria retained their
altered morphology. The authors were able to find qualitative structural differences between 8,
12, and 16 day old plants exposed to similar HC1 concentrations with the greatest number in the
12 day old plants.
K.2.4 Peroxidase Activity in Plant Leaves Exposed to Gaseous HCi or Ozone
This Endress et.al. study (1980) was designed to determine if peroxidase activity was elevated in
bean and tomato leaf tissues that did not show macroscopic injury after exposure to HCI and
ozone gases. Perioxidase is an enzyme found in almost all higher plants and animals that is
associated with cellular growth and development. Pinto bean plants 12 days from seeding and
tomato plants 88 days from seeding were exposed for 20 minutes to multiple concentrations of
gaseous HCI and ozone. The HCI concentrations used were 0, 4.08, and 12.52 mg/m3. The
treated plants were sampled immediately for peroxidase and at 24 and 48 hours for both
peroxidase and macroscopic injury. Preparations for determining peroxidase levels included
preparation of enzyme assay solutions and for polyacrylamide slab gel electrophoresis.
Of the four independent experiments (a) bean and HCI, (b) tomato and HCI, (c) bean and ozone,
and (d) tomato and ozone, only the tomato plants exposed to ozone showed a concentration
related significantly different level of perioxodase activity than the controls. Leaves were also
scored at 24 and 48 hours after exposure for macroscopic injury symptoms. The visible injury
was statistically related to the pollutant treatment in all case except HCI and tomato. Greater
than 10 percent of bean leaves showed injury at an HCI exposure level of 4.08 mg/m3, while
approximately 75 percent showed injury and 20 percent showed necrosis at 12.52 mg/m3.
Several previous studies of plants exposed to stressors including air pollutants have shown that
increased total peroxidase activity or altered isozyme patterns are frequently induced. Others
indicate that crop yields may be reduced by exposure to air pollutants, even though no
discernable macroscopic injury symptoms were present. Total peroxidase activity appears
unsuitable as a biomarker of latent injury.
K.2.5 Photosynthesis and Respiratory Consequences of Hydrogen Chloride Gas
Exposures of Phaseolus Vulgaris L. and Spinacea Oleracea L.
The purpose of the Endress, et. al. 1982 study was to measure the photosynthetic and respiratory
activities of plant leaf tissue following exposures of pinto bean plants at 8 to 16 days from
seeding for 20 minutes to gaseous HCI concentrations of 3.3 and 45.4 mg/m3. Locally
K-9
-------
purchased spinach leaves were treated in the same manner. Visible foliar damage was estimated
at 24 hours after exposure. Immediate sampling was completed on isolated chloroplasts in an
assay solution to determine both light and dark oxygen evolution. Additionally, leaf discs were
used to estimate photosynthesis and respiration rates.
Chloropyll levels sampled immediately following exposure were slightly higher in exposed
plants than in controls. However, of the 8, 10, 12, 14, and 16 day old plants only the 14 day old
sample showed a statistically significant increase. When plants were sampled 24 hours after
treatment, chlorophyll levels were slightly less in the exposed plants than in the controls.
Chloroplasts were sensitive to HC1 gas exposure, but as the time between exposure and sampling
increased many chloroplasts gradually recovered their normal appearance while a few others
became totally disrupted.
Exposure to HC1 that resulted in <15% necrotic leaf injury appeared to stimulate both
photosynthesis and respiration. These rates decreased linearly with increased injury severity.
Leaf discs sampled 24 hours after treatment generally exhibited greater rates of photosynthesis
and respiration than those sampled immediately following exposure. Table K-6 presents the
rates observed. No significant difference was found between control and exposed plants in
oxygen evolution or consumption among the spinach plants except for the variety Bloomsdale.
Significantly higher respiration rates at both sample times were exhibited by variety Bloomsdale.
Treated pinto bean chloroplasts exposed to HC1 concentrations ranging from 9.5 to 21.8 mg/m3
evolved less oxygen than controls. Significant reductions in oxygen evolution occurred
following treatment with 14.9 or 18.5 mg HCl/m3. Chloroplasts of spinach also exhibited
reduced oxygen evolution. The 24 hour samples showed no recovery by from the initial
depressed rates of oxygen evolution. Table K-7 presents the rates of oxygen evolution.
Foliar injury and photosynthetic rates for discs that were dipped in various concentrations of
dilute liquid hydrochloric acid was comparable to with that from the gaseous HC1 treatment.
Oxygen evolution from isolated spinach chloroplasts was examined with regard to pH. Increased
acidification of the reaction solution caused a linear inhibition of the oxygen evolving capability
regardless of the acid used: HC1, H2SO4, or HNO3. Unlike the leaf disc experiments, no recovery
was observed in the 24 hour sample.
Table K-6. Comparison of Photosynthesis and Respiration Rates Exhibited by S. oleracea
S. oleracea (Spinach)
% of Control
Variety
0 hour
24 hour
Photosynthetic Rate
Melody
109
105
Bloomsdale
84
112
Avon
87
109
Respiration Rate
Melody
115
93
Bloomsdale
123*
142*
Avon
101
99
Leaf discs were taken from plants exposed either to 22.1+5.01
K-10
-------
S. oleracea (Spinach)
Variety
% of Control
0 hour 24 hour
Photosynthetic Rate
mg anhydrous HC1 /m or carbon filtered air (control) for 20
minutes. Data are mean values of four experiments.
Significance: * PO.Ol determined by Student's t test.
Source: Endress, et. al. 1982
Table K-7. Oxygen Evolution of Chloroplasts Isolated from P. vulgasis and S. oleracea
Species
mg HC1
/m3
jil O2 / hr-mg Chlorophyll
% of Control
Control
HC1
P. vulgaris
9.5
180.5
170.9
94.7
(pinto bean)
14.9
199.1
83.1
41.7*
18.5
110.4
75.8
68.7*
20.7
64.5
58.5
90.7
21.8
81.1
63.3
78.4
S. oleracea
27.7
163.6
106.5
65.1***
(spinach var.
29.2
143.3
87.2
60.9**
Melody)
31.7
212.8
157.0
73.8*
S. oleracea
28.9
223.2
211.6
94.8
(spinach var.
29.0
199.5
174.0
87.2
Boomsdale)
30.4
261.3
209.5
80.2
Data presented are average of a minimum of five samples per treatment with
carbon filtered air or anhydrous HC1 at concentrations indicated. Significance: *
P<0.05, **P<0.01, ***P<0.001.
Source: Endress, et. al. 1982
Photosynthetic C02 fixation in HC1 treated pinto beans was followed by examining the activity
of RubPCase. After exposure to HC1 gas, experiments showed an initial sharp decrease in
RubPCase activity followed by a continued but more gradual decrease. Low concentrations of
HC1 stimulated RubPCase and with minimal or no necrotic injury observed, but with either
increasing severity of injury or HC1 concentration, RubPCase activity decreased. Sampling 24
hours after treatment showed enzyme activity was not as depressed and recovery in samples
exposed to 20 mg HCl/m3 or lower.
K.2.6 Histological Effects of Aqueous Acids and Gaseous Hydrogen Chloride on
Bean Leaves
This study (Swiecki et. al., 1982) was conducted to look for the possible mechanism of action for
gaseous HC1 phytotoxicity. From previous work, it was hypothesized that gaseous HC1
condensed as aqueous acid on leaf surfaces due to HCl's high water solubility and the high
humidity at the leaf boundary layer. This experiment compared injury symptoms following
treatment with aqueous HC1 or HC1 gas and assessed whether the injury was attributable to H+,
C1-, or a combination of the two. Using 20 minute exposures to 12 day old pinto bean plants,
aqueous acids' and chloride salts' effects were compared to the effect of gaseous HC1 exposures
of 14.5 to 19 mg/m3.
K-ll
-------
Injury was observed in all leaves when observed 1 hour after exposure to dilute aqueous HC1
concentrations and other dilute acids. For treatments exhibiting injury, the 24 hour
observations differed from the 1 hour observations only by minor increases in visible injury.
Equivalent concentrations of aqueous HC1, H2SO4, and HNO3 produced essentially identical
injury symptoms. Equivalent levels of injury were produced in leaves exposed to 0.06N (pH =
1.45) aqueous HC1 or 15-30 mg gaseous HCl/m3 for 20 minutes.
Injury from 12 day old pinto bean leaf aqueous HC1 exposure was similar to injury the author
had found in 8 day old leaves exposed to gaseous HC1. The dependency of injury susceptibility
on tissue age was a generally agreed upon hypothesis and using microscopic features, this study
found the only difference in age appears to be in the numbers of affected cells and not in the type
of injury. At the levels tested, the authors found that effect of the chloride anion was
inconsequential relative to the hydrogen ion concentration. Further, they found the generalized
injury response to acid and particularly aqueous acid corresponds closely to injury caused by
sulfate acid precipitation.
K.2.7 The Phytotoxicity of Designated Pollutants on Plant Species
Additional Air Force funded research on plant exposures to HC1 and aluminum oxide was
conducted at the University of California at Irvine (Granett, 1984). In this research aluminum
oxide particulate matter was found to be nontoxic. The exposure of plants to HC1 was conducted
using acidic mist solutions These HC1 solutions ranged from 0 to 5,000 ppm by volume, with
measured pH ranging from 4.91 to 0.75. The researchers either sprayed the mist on the plant or
soaked the plant in the solution, rinsed, and then checked for effects. Since the effects were
compared to doses in terms of the pH of the solutions rather than air concentrations, we decided
not to include these results in our development of ecological exposure thresholds.
K.3 Development of Gaseous HCI Ecological Exposure Thresholds
The series of studies described provide increasingly detailed information about gaseous HCI
injury to plants from visual observations, photosynthetic and oxygen evolution rates, and
electron microscopy of localized cellular damage following exposure. The hypothesis of the
process for the damage remains consistent - gaseous HCI condenses on the leaf surface
producing an aqueous acid solution that promotes cellular injury. Degree of injury is
proportional to exposure to gaseous HCI and this injury is a response to exposure to an acid
rather than being specific only to HCI. Unfortunately, the data was developed to determine the
impact of exposure to gaseous HCI from short-term high concentration exposures. While this
data can be extrapolated for use in developing acute ecological exposure thresholds, more
uncertainty is involved in extrapolating the data to develop chronic ecological exposure
thresholds. As a result, one might look to the literature on acid rain injury to increase the data
available for use in developing screening-level chronic ecological exposure thresholds. The
results of the studies, emphasizing the more conservative results are summarized in Table K-8.
Table K-8. Summary of Studies from Literature Review of Gaseous HCI and Foliar
Damage
Study Summary
K-12
-------
Study
Summary
Phytotoxicity of Hydrogen
Chloride Gas with a Short-
Term Exposure
(Lerman et. al., 1976)
After 20 minute exposure, 10% damage to most sensitive 1 of 8 plant
types at 6.5 mg/m3. Changes in 7 of 8 types at lowest concentration -
1 mg/m3. (Plant types: aster, calendula, centaurea, cosmos, dwarf
marigold, marigold, nasturtium, and zinnia.)
Foliar and Microscopic
Observations of Bean Leaves
Exposed to Hydrogen Chloride
Gas (Endress et.al., 1978)
After 20 minute exposure to pinto bean plants, damage found at
lowest exposure concentration of 6 mg/m3. At 25 mg/m3 10% showed
necrotic lesions.
Reversible Fine Structural
Alterations of Pinto Bean
Chloroplasts Following
Treatment with Hydrogen
Chloride Gas
(Endress et.al., 1979)
After 20 minute exposure to pinto bean plants, no observed adverse
effects at lowest exposure concentration of 6 mg/m3. 1% damaged
after 20 minutes exposure to 11.3 mg/m3.
Peroxidase Activity in Plant
Leaves Exposed to Gaseous
HCl or Ozone (Endress et.al.,
1980)
After 20 minute exposure to bean and tomato plants, increase in
peroxidase activity found at lowest exposure concentration of 4
mg/m3. 25%) necrotic or injured bean leaves at 4 mg/m3. 10%> tomato
necrotic or injured at 12.5 mg/m3.
Photosynthesis and
Respiratory Consequences of
Hydrogen Chloride Gas
Exposures of Phaseolus
Vulgaris L. and Spinacea
Oleracea L.
(Endress et.al., 1982)
After 20 minutes exposure to gaseous HCl concentrations of 3.3 and
45.4 mg/m3, measured photosynthetic and respiratory activities of
spinach and pinto bean plants at 8 to 16 days from seeding. Exposure
to HCl that resulted in <15% necrotic leaf injury appeared to stimulate
both photosynthesis and respiration. These rates then decreased
linearly with increased injury severity. Bean plants exposed to HCl
concentrations ranging from 9.5 to 21.8 mg/m3 evolved less oxygen
than controls with significant reductions following treatment with 14.9
or 18.5 mg HCl/m3. Similar reduction for spinach. No recovery in 24
hour samples.
Histological Effects of Aqueous
Acids and Gaseous Hydrogen
Chloride on Bean Leaves
(Swiecki et. al., 1982)
After 20 minute exposure to pinto bean plants, damage found at
lowest exposure concentration of 15 mg/m3.
Cited in (Lerman et.al., 1976)
Haselhoff and Lindau, 1903:
Single exposure, 2 day duration at 5-20 ppm (8-30 mg/m3). Seedlings
of viburnum and larch killed in less than 2 days
Cited in (Lerman et.al., 1976)
Shriner and Lacasse, 1969:
Single exposure, 2 hour duration at 5 ppm (8 mg/m3)28 day old
tomato plants developed interveinal bronzing followed by necrosis
within 72 hours
Cited in (Lerman et.al., 1976)
Means and Lacosse, 1969:
Single exposure, 4 hour duration at 3-43 ppm (5-65 mg/m3) to 12
types of 2-5 year old coniferous and broadleaf seedlings: most
sensitive broadleaf (tulip tree) visible injury at 3 ppm (5 mg/m3); most
sensitive conifer (white pine) visible injury at 8 ppm (12 mg/m3) No
injury to least sensitive Thuja occidentalis (white cedar) at 43 ppm (65
mg/m3).
K-13
-------
Study
Summary
Cited in (Lerman et.al., 1976)
Hulensberg, 1974:
Single exposure, 29 hour duration, 0.5 mg/m3 (0.3 ppm)
Carrots exposed 45 days after germination showed 32.2 to 49.7%
decrease in crop yield; exposed 96 days showed only 5.3% crop yield
decrease. Winter grape only leaf discoloration, radishes no damage.
Tomato,cucumber and bush bean plants showed leaf damage and
increase in leaf chlorides. Reduction in yield severe in tomato and
slight in cucumber plants.
Cited in (Lerman et.al., 1976)
Masuch et. al., 1973:
Exposure of spinach to HC1 gas for 43 hours within 5 days at 0.13 to
0.25 mg/m3 and for 208 hours within 14 days at 1.6 mg/m3 resulted in
changes in chloroplasts. Authors attributed differences between
Masuch's study and theirs could be attributed to differences between
acute and chronic exposures, differences in species characteristics, or
both.
K.3.1 Recommended Methods for Developing Ecological Exposure Thresholds
The EPA Guidelines for Ecological Risk Assessment (EPA, 1998) discusses the use of a stressor-
response analysis in characterization of ecological effects. Point estimates are frequently
adequate for simple assessments or comparative studies of risk with a median effect level
frequently used because the level of uncertainty is minimized at the midpoint of the regression
curve. The guidance points out that a 50% effect level for an endpoint such as survival may not
be appropriately protective for an assessment endpoint. Median effect levels can be used for
preliminary assessment or comparative purposes especially when used in combination with
uncertainty factors. Selection of a different effect level (10%, 20%, etc.) can be arbitrary unless
there is some clearly defined benchmark for the assessment endpoint, making it preferable to
carry several levels of effect or the entire dose response curve forward to risk estimation.
At the conservative end of the spectrum, EPA's Region 5 Superfund Office (EPA, 2007a)
recommends the use of No-Observed-Adverse-Effect-Levels (NOAELs) for screening level
ecological effects evaluations.3 They suggest the NOAEL be from scientific studies that
exposed the plants or animals to the chemical for a long time (chronic). Short- or medium-time
exposure studies are less desirable because it may take a long time of exposure to a chemical in
order for there to be an adverse effect. However, they point out that time should be measured
relative to the life span of the plant or animal being studied. For a plant or animal with a short
life span, it may only be necessary to have a relatively short study. They further guide the risk
assessor to be consistently conservative in selecting literature values, to describe the limitations
of using the data for the assessment, and to discuss the uncertainty before moving onto the risk
calculation. Region 5 also has available a setoff ecological screening levels that include
exposures to chemicals through air (EPA, 2003). Unfortunately, they have not developed values
for HC1 for any exposure medium.
3 IRIS defines the No-Observed-Adverse-Effect Level (NOAEL) as the highest exposure level at which there are no
biologically significant increases in the frequency or severity of adverse effect between the exposed population and
its appropriate control; some effects may be produced at this level, but they are not considered adverse or precursors
of adverse effects.
K-14
-------
In the Department of Energy document providing benchmarks for soil to plant chemical
exposures, Toxicological Benchmarks for Screening Contaminants of Potential Concern for
Effects on Terrestrial Plants: 1997 Revision (Efroymson, et.al., 1997), growth and yield are
selected as the two significant parameters because 1) they are the most common class of
response parameters reported from phytotoxicity studies and 2) they are ecologically significant
responses both in terms of the plant population and the ability of the vegetation to support higher
trophic levels.
They recommend 20% reduction in growth or yield as the threshold for significant effects to be
consistent with other screening benchmarks and with current regulatory practice. They justify
the 20% level because most regulatory criteria are based on concentrations in toxicity tests that
cause effects that are statistically significantly different than the controls. On average, those
concentrations correspond to greater than a 20% difference effects. Additionally, in programs
such as Superfund, regulatory actions may be based on comparisons of biological parameters
measured on a contaminated site to those from reference sites. Differences between those
parameters must be greater than 20% to be reliably detected in such studies. Therefore, the 20%
effects level is treated as a conservative approximation of the threshold for regulatory concern.
Using the method for deriving soil benchmarks based on the National Oceanic and Atmospheric
Administration's (NOAA) method for deriving the Effects Range Low (ER-L). This method has
been recommended as a sediment screening benchmark by EPA Region 9. The ER-L is the 10th
percentile of the distribution of various toxic effects thresholds for various organisms in
sediments. Justifications include that the phytotoxicity of a chemical in soil is a random
variable, the toxicity of the soil at a given site is drawn from the same distribution, and the
assessor should be 90% certain of plants growing in that soil. Analogously, site-specific
atmospheric conditions, including the concentrations of other pollutants, would affect the
phytotoxicity of a chemical in air in the same manner.
The 10th percentile phytotoxicity benchmarks are derived by rank ordering the Lowest-Observed-
Effects-Level (LOEL)4 values and selecting a number that approximates the 10th percentile. If
ten or fewer values are available for a chemical, the one with the lowest LOEC is used. Though
the derivation of a benchmark through this method implies a significant impact on approximately
10%) of the species, the authors defend their level of conservatism because: 1) the benchmarks
are for the community level and a loss of 10% of the community species is likely acceptable and
2) the benchmarks derived by this method have proved to be conservative in practice.
Finally, the authors attempt to assign levels of confidence to the benchmarks:
1.Low Confidence - Benchmarks based on 10 or fewer literature values.
2.Moderate Confidence - Benchmarks based on 10 to 20 literature values.
3.High Confidence - Benchmarks based on over 20 literature values.
4
IRIS defines the Lowest-Observed-Effect Level (LOEL or LEL) as the lowest dose or exposure level at which a
statistically or biologically significant effect is observed in the exposed population compared with an appropriate
unexposed control group.
K-15
-------
Based on professional judgment, the authors confidence in a benchmark were lowered a level if
the range of plant species is narrow or if the 10th percentile is the lowest value tested and caused
a greater than 30% reduction in the measured growth parameters.
K.3.2 Development of Screening-Level HCI Ecological Thresholds
As recommended in EPA Guidelines for Ecological Risk Assessment, we attempted to develop
and carry several levels of effect forward to risk estimation in terms of multiple screening-level
ecological exposure thresholds. However, with fewer than 10 studies to use in the development
of a threshold, a more conservative approach of basing the threshold on the lowest LOEL was
selected (Efroymson, et.al., 1997).
A first step in developing thresholds is to adjust from the exposure duration of the experiments to
the desired acute and chronic exposure durations. Because of the time periods available from the
modeling exercise, the acute or short-term exposure duration we are interested in is one-hour.
The chronic or long-term exposure duration is one-year. To extrapolate from the many
experiments using 20 minute exposure durations to the desired one-hour duration, and from the
few longer term studies to an one-year duration, we will follow the recommendations of EPA's
Office of Research and Development (EPA, 2007b) for human health exposure duration
adjustments as described below.
Haber's Law (i.e., C x t = k, where C = concentration, t = time, and k = a constant) traditionally
has been used to relate exposure concentration and duration to a toxic effect (Rinehart and Hatch,
1964). Specifically, the equation implies that exposure concentration or duration may be
adjusted to attain a cumulative exposure constant (k) which relates to a toxic response of specific
magnitude. Work by ten Berge et al. (1986), affirmed that chemical-specific relationships
between exposure concentration and exposure time may be exponential rather than linear; i.e.,
the expression now becomes Cn x t = k, where n represents a chemical-specific exponent.
Upon examining the concentration and time relationship of the lethal response to approximately
20 chemicals, ten Berge et al. (1986) reported that the empirically derived value of n varied from
0.8 to 3.5. The magnitude of the exponent (n) provides insight into the relationship between
exposure concentration and exposure duration such that if n = 1, the toxic response to the
chemical is dependent solely upon total dose (i.e., a linear relationship, or Haber's Law).
Generally, if n < 1, the exposure duration is the determinant of the toxic response and if n > 1,
the exposure concentration is the primary determinant of the toxic response.
Ten Berge developed an exponent value of one for HCI. Thus for HCI a linear relationship exists
and 20 minute exposure concentrations can be extrapolated to one-hour exposure concentrations
by multiplying by 1/3 (20 minutes / 60 minutes). Because of the lack of data from long term
exposures, we will use an uncertainty factor of 10 to extrapolate from acute to chronic exposure
thresholds. This is a value EPA frequently uses in developing human health dose response
values when long term studies are not available. Table K-9 presents the impacts at the lowest
levels noted for potential use in establishing the screening-level thresholds. While potentially
overestimating impacts to more resistant species, the level is more likely to be protective of all
species, including those not studied.
Table K-9. Results of Gaseous HCI Studies for Use in Development of Screening-Level
Thresholds
K-16
-------
Selected Significant Impacts from 20 Minute Exposures
1.5 m «/m '
Lowest concentration in study
Changes in 7 of 8 plant types
6.5 in«/m3
10% leaves damaged in 1 plant
type
(Lerman et.al., 1976)
4 mg/m3
Lowest concentration in study
Increase in perioxidase activity
25% necrotic or injured bean
leaves
11. 5 mg/m3
25% necrotic or injured tomato
leaves
(Endress et.al., 1980)
9.5 to 21.8 mg/m3
Lowest concentration in
study Evolved less oxygen
than controls
14.9 or 18.5 mg HCl/m3
Significant reductions in
oxygen evolution
(Endress et.al., 1982)
Selected Significant Impacts from 2 to 4 Hour Exposures
3 nig/in3
Lowest concentration in study
Visible injury to most sensitive
broadleaf of 12 tree species
after 4 hour exposure
(Means andLacasse, 1969; as
cited in Lerman et.al., 1976)
8 mg/m3
Interveinal bronzing followed
by necrosis after 2 hour
exposure
(Shriner and Lacasse, 1969; as
cited in Lerman et.al., 1976)
8-30 mg/m3
Death to viburnum and birch
seedlings after 2 hour
exposure
(Haselhoff and Lindau, 1903;
as cited in Lerman et.al.,
1976)
Selected Significant Impacts from Longer Exposure Durations
0.13 to 0.25 mg/m3
Changes in spinach chloroplasts
after 43 hours exposure within 5
days
(Masuch et.al., 1973; as cited in
Endress et.al., 1979)
0.5 mg/m3
32.2 to 49. 7% crop yield
decrease 45 day old carrots
exposed for 29 hours
(Hulensberg, 1974; as cited in
Lerman et.al., 1976)
1.6 mg/m3
Necrosis and chlorosis in
spinach after 208 hours
exposure within 2 weeks
(Masuch et.al., 1973; as cited
in Lerman et.al., 1976)
K-17
-------
Two studies were selected to be used in establishing the LOEL and LOAEL for establishing
screening-level phytotoxicological thresholds for HC1. The Lerman, et.al. study (1976) was
determined to be most appropriate for estimating an LOEL. The 20 minute plant exposure to the
lowest concentration in the study, 1.5 mg/m3, resulted in changes in 7 of the 8 plant types in the
study. Adjusted to an one-hour exposure duration, the LOEL is 0.5 mg/m3 or 500 |ig/m3. The
Endress et. al. 1980 study was determined to be most appropriate for estimating an LOAEL. The
20 minute plant exposure to the lowest concentration in the study, 4 mg/m3, resulted in 25%
necrotic or injured pinto bean leaves. Adjusted to an one-hour exposure duration, the LOAEL is
1 mg/m3 or 1,000 |ig/m3.
Because of the lack of ample data to statistically establish a ecotoxicological threshold, and the
critical exposure concentrations were the lowest in the studies, we recommend using the more
conservative LOEL value of 0.5 mg/m3 for the short-term screening-level phytotoxicological
threshold for HC1. Applying the factor of 10 extrapolation for using a short-term study to
establish a chronic dose response value, our recommended long-term screening-level
phytotoxicological threshold for HC1 is 0.05 mg/m3 or 50 |ig/m3.
Screening-level phytotoxicological thresholds for HC1 based on Lerman, et.al. (1976):
Critical Effect Point of Departure Acute Threshold Chronic UF Chronic
Threshold for
Leaf chanses LOEL CAdiV 0.5 me/m3 5xl0_1 me/m3 10 5xl0"2me/m3
K.3.3 Comparison of Modeled HCI Air Concentration Estimates to Screening
Level Ecological Thresholds
The highest one-hour average HCI air concentration of HCI modeled was compared to the
ecological threshold developed. This maximum one-hour air concentration estimate is 2 mg/m3.
To calculate the hazard quotient (HQ) for foliar damage, the air concentration estimate is divided
by the screening level ecological threshold. In this case, a short term screening level ecological
HQ of 4 is calculated for potential foliar damage (2 mg/m3 / 0.5 mg/m3 = 4).
An HQ > 1 indicates that there is the potential for foliar damage to plants from the estimated air
concentration. However, due to the lack of data, this relates to potential individual leaf damage
rather than the 20% reduction in growth or yield recommended as the threshold for significant
effects. Thus, the HQ resulting from the screening level ecological threshold is a conservative
value and a value so near to 1 cannot be construed to mean that significant ecological damage
would be anticipated. Rather, because in addition to the use of a conservative threshold due to
lack of data coupled with the highest one-hour average air concentration modeled, it is more
likely that HCI would not cause significant effects to exposed plants.
The comparison of long term average modeled air concentrations is discussed in another section.
Because the long term screening level ecological threshold is greater than the reference
concentration (RfC) used to assess noncancer adverse health effects, the RfC is protective of
both human health and of potential foliar damage and estimating a long term HQ for ecological
effects is not necessary.
K-18
-------
K.4 References
Efroymson, R.A., M.E. Will, G.W. Suter II, and A.C. Wooten, 1997. Toxicological Benchmarks
for Screening Contaminants of Potential Concern for Effects on Terrestrial Plants: 1997
Revision. Prepared for the U.S. Department of Energy by Oak Ridge National Lab, Oak Ridge,
TN. ES/ER/TM-85/R3.
Endress, A.G., T. J. Swiecki, and O.C. Taylor, 1978. Foliar and Microscopic Observations of
Bean Leaves Exposed to Hydrogen Chloride Gas. Environmental and Experimental Botany,
Vol. 18, pp. 139-149.
Endress, A.G., J.T. Kitasako, and O.C. Taylor, 1979. Reversible Fine Structural Alterations of
Pinto Bean Chloroplasts Following Treatment with Hydrogen Chloride Gas. Botanical Gazette,
Vol. 140, No. 1, pp. 11-19.
Endress, A.G., S. J. Suarez, and O.C. Taylor, 1980. Peroxidase Activity in Plant Leaves Exposed
to Gaseous HCl or Ozone. Environmental Pollution, (Series A) 22, pp.47-58.
Endress, A.G., S. J. Suarez, and O.C. Taylor, 1982. Photosynthesis and Respiratory
Consequences of Hydrogen Chloride Gas Exposures of Phaseolus Vulgaris L. and Spinacea
Oleracea L. Environmental Pollution (Series A) 29, pp. 13-26.
EPA, 1998. Guidelines for Ecological Risk Assessment. Risk Assessment Forum. EPA/630/R-
95/002F. Federal Register 63(93) 26846.
EPA, 2007a. Region 5 Superfund, Screening Level Ecological Effects Evaluation. Online at:
www.epa.gov/region5superfund/ecology/erasteps/erastepl.html
EPA, 2007b. Standing Operating Procedures for the Development of Provisional Advisory
Levels (PALs) for Chemicals. Prepared for U.S. EPA National Homeland Security Research
Center by Oak Ridge National Lab, Oak Ridge, TN.
Lerman, S., O.C. Taylor, and E.F. Darley, 1976. Phytotoxicity of Hydrogen Chloride Gas with a
Short-Term Exposure. Atmospheric Environment, Vol.10, pp. 873-878.
Rhinehart W.E. and T. Hatch, 1964. Concentration-time product (CT) as an expression of dose
in sublethal exposures to phosgene. Industrial Hygiene Journal. 25:545-553.
Swiecki, T.J., G. Anton, and O.C. Taylor, 1982. Histological Effects of Aqueous Acids and
Gaseous Hydrogen Chloride on Bean Leaves. American Journal of Botany, Vol. 69, pp. 141-
149.
Ten Berge, W.F., A. Zwart, and L.M. Applebaum, 1986. Concentration-time mortality response
relationship of irritant and systematically acting vapours and gases. Journal of Hazardous
Materials. 13(3):301-309.
K-19
-------
Appendix L: Statistical comparison of monitored and
modeled ambient benzene concentrations near
petroleum refineries in Texas City, TX
Roy Smith
Jim Hirtz
EPA Office of A ir Quality Planning and Standards
-------
L-l. Introduction
The risk assessment performed by the EPA for the petroleum refinery source category to
support its proposed rulemaking regarding residual risks has been criticized by many,
primarily those who suggest that the emissions estimates used as the basis for the risk
assessment are too low by a factor of 10 to 100. EPA has countered this criticism by utilizing
an extensive review process, including multiple elements of public and expert review, to
develop the inventory of emissions and source information used in its risk assessments.
This analysis compares ambient monitoring data for benzene from two monitoring sites near
two petroleum refineries in Texas City, TX to dispersion modeling results for those facilities.
The monitors were selected for this exercise because of their proximity to large refineries and
relatively complete datasets of hourly benzene measurements. We did this to assess the
general magnitude of uncertainty, and the possibility of bias, in our facility-specific
emissions estimates for benzene, recognizing that benzene exposures tend to drive total
cancer risk estimates for refinery emissions and also that benzene emissions originate from
many common sources (primarily mobile) besides refineries.
This case study illustrates both the utility and limitations of conducting such an assessment,
and provides a general indication of whether our benzene emissions estimates for the two
facilities in question are reasonable representations of actual benzene emissions during the
monitoring year. It attempts to answer the question, "Are our benzene emission estimates
truly low by a factor of 10 to 100 (at least for these 2 facilities), or are they close enough to
be useful in residual risk decision-making?" We attempt to answer this last part keeping in
mind the 2 order of magnitude range of MIR values embodied in the residual risk decision
framework.
L-2 Methods
L-2.1 Monitoring sites
Benzene monitoring data were obtained from the Texas Commission on Environmental
Quality (TCEQ) for two continuous flame ionization detection (FID) monitors located in
Texas City, TX. The latitude, longitude, and dates for the monitors are provided in Table L-
1. The FID monitors are the most common gas chromatograph monitors with reliable
detection limits for volatile organic compounds (VOCs) such as benzene, toluene, and xylene
(BTX). These two benzene monitors are each located within 300 meters of major industrial
sources that emit benzene, including three large refineries (BP Refining, Marathon, and
Valero Refining) and one chemical manufacturing facility (Sterling Chemical). Figure L-l
shows an aerial view of the study area.
Table L-l: Texas Citv. TX benzene monitors
Monitoring Site
BP-31 st
Marathon-Ashland
Latitude
29.381361
29.377
Longitude Census Tract Monitoring Period
-94.940806 I 48167721900 r June 1,2003 to September 30,2007
-94.9104 j 48167722400 October 1,2004 to September 30,2007
L-l
-------
Figure L-l. Aerial photo of Texas City, TX.
Monitor
chemical
Monitor
8 © 2008 Europa Technologies
Image Houslon-Galvestori Araafcourcll
C5k0Eflb$Gl
L-2
-------
L-2.2 Monitoring data
The BP monitor at 31st Street began collecting continuous monitoring data for benzene on
June 1, 2003. The monitoring data were averaged to 1-hour values, with analytical
results from 06/01/2003 to 09/30/2007, coupled with hourly measurements of wind speed
and wind direction. The Marathon monitor at 11th Street began operation on October 1,
2004. The hourly monitoring data, also including wind speed and direction, included the
period from 10/01/2004 to 09/30/2007. Table L-2 contains a summary of the annual
average concentrations for these sites.
Table L-2: Monitored benzene concentrations - annual averaee
Monitor
Year
Avg. Cone (ug/m"1)
# Hours
BP-31 st
2004
5 41
7716
BP-3'1 st
2005
8,74
7454'
BP-31 st
2006
: 5.67
7634
BP-31 st " ""
All Hours
5.74
33205
Marathon-Ashlarid
2005
i 7.47
5377 "
Marathon-Ashland
2006
6.81 r
5817
Marathon-Ashland
All Hours
7 17 r
16860
The raw hourly ambient data were evaluated and adjusted so that non-detected ( ND)
values were replaced with V2 the minimum detection limits (MDLs). In addition,
measurements that lacked matching hourly wind directions were omitted, in order to
support a statistical analysis of directional source contributions at the monitor. These
adjustments had little effect on the annual averages. For example, the adjustments caused
the annual average benzene concentration at the Marathon monitor for 2006 to decrease
from 6.81 to 6.72 /ig/m3, or about 1.3%.
L-2.3 Modeling data
Modeling results developed for the petroleum refinery source category at the census
tracts where the monitors are located indicate that benzene is responsible for over 90% of
the estimated cancer risk associated with these three petroleum refineries. The monitors
are located within 200 meters of residential areas and are relatively close to the locations
where the MIR for the BP refinery and the Marathon refinery were identified based on
the RTR modeling. In addition, the RTR modeling results for each monitor indicated that
the modeled benzene concentrations at the BP monitoring site were overwhelmingly
(greater than 99%) influenced by emissions from the BP facility, and that modeled
benzene concentrations at the Marathon monitoring site were predominantly (greater than
85%) influenced by emissions from the Marathon facility, with the remaining influence
coming largely from Valero.
L-2.4 Model to monitor comparison
To compare modeled refineries emissions data from the RTR database (referred to as
"modeling data") to the ambient monitor concentrations (referred to as "monitoring
data"), we had to prepare the emissions data from the three refineries and one chemical
L-3
-------
plant in the vicinity of the two monitors. We assumed constant hourly emissions and
developed estimates of hourly ambient concentrations using dispersion modeling that was
largely based on the RTR modeling, with adjustments to emission rates from each
individual emission point made by multiplying them by the ratio of the total TRI
emissions from that facility in the monitor year to the TRI emissions from that facility in
the base RTR modeling year. We focused the comparison on 2004 data at the BP
monitor and 2006 data at the Marathon monitor. We chose 2004 for the BP monitor
because the initial RTR modeling run for the BP refinery was based on emissions data
from the year 2004, affording us the opportunity to perform a direct comparison. For the
Marathon monitor, we chose not to use data from 2005 because a large explosion and fire
occurred at one of the refineries in 2005, altering annual benzene emission levels for
2006 in an unknowable way. Although the original RTR modeling run for the Marathon
refinery was based on emissions data from 2002, we revised that modeling run for this
analysis by scaling benzene emissions using TRI information for 2006 {i.e., adjusting by
the ratio of 2006 benzene emissions to 2002 benzene emissions) and using
meteorological data from 2006.
Meteorological data were available from the Texas City Ball Park (just north of the
refineries) and the Galveston airport (about 14 km SSE of the refineries). Both stations
exhibited southerly winds. However, given that the emission sources were south of the
BP and Marathon monitors, the Galveston winds were considered more representative of
the area because of the more open exposure of the Galveston instrument tower, especially
from the south. Galveston was less affected by obstacles around the tower. Also, when
comparing 2004 and 2006, the Galveston winds are more consistent in direction, a
general southeast direction, while the ball park shifts from predominantly south in 2004
to south and southeast for 2006. The variations in wind roses for the Ball Park site
between 2004 and 2006, given the consistency of the patterns at Galveston for those two
years, may be indicative of an exposure problem for the Ball Park site resulting in very
localized influences. For these reasons the modeling was conducted using data
(including high-altitude data) from the Galveston airport.
We conducted one AERMOD run to develop hour-by-hour estimated benzene
concentrations at the BP monitor site for the year 2004 using 2004 emissions data for the
BP refinery, 2002 emissions data from Marathon, Valero, and Sterling (adjusted to 2004
using TRI activity indices), and meteorological data from the Galveston airport for 2004.
We conducted a second AERMOD run to simulate hourly benzene concentrations at the
Marathon monitor site for the year 2006 using emissions data from Marathon, Valero,
and Sterling from 2002 (scaled to 2006 using TRI activity data), and hourly
meteorological data from the Galveston airport for 2006. The BP refinery emissions
were omitted from the Marathon monitor site comparison due to the 2005 explosion,
which disrupted activities at the BP refinery in 2006. All modeling options were
identical to those used in the RTR baseline petroleum refinery assessment modeling.
In addition to preparing the emissions data, we also adjusted the monitoring data to focus
our analysis as specifically as possible on the benzene contribution from petroleum
refineries. To help characterize the impact of benzene emissions from the petroleum
L-4
-------
refineries on each of the monitors, we first estimated the contribution at each monitor that
could be attributed to unmodeled sources such as mobile and area sources. To do this, we
extracted estimates of the ambient benzene concentration contributions for all other
sources besides the major industrial sources (this included area sources, mobile sources,
and long-range transport) from the 2002 National Air Toxics Assessment (NATA)1 at
each of the census tracts within 20 km of the monitors. We created an isopleth map of
these contributions to develop estimates at each of the monitors (Figure L-2), giving us
an annual background benzene concentration estimate of 1.0 /ig/m3 at the BP monitor and
1.4 /ig/m3 at the Marathon monitor. These background estimates were subtracted from
the individual monitor data, thereby limiting the comparison between modeled estimates
and measurements to contributions from the refineries. We recognize that the
contribution from background sources can vary on an hourly basis, and that this
simplistic approach cannot be valid for any time scale less than annual. Thus, a great
deal of variation in monitored data may be caused by changes in background
contribution.
Since our initial modeling determined that the BP monitor is overwhelmingly influenced
by benzene emissions from the BP facility (greater than 99%), we use the results of our
comparisons to derive inferences about the BP emissions inventory. Since our initial
modeling determined that the Marathon monitor was predominantly influenced by
benzene emissions from the Marathon facility (greater than 85%), we use the results of
our comparisons to derive inferences about the Marathon emissions inventory. We
recognize that this can lead to greater uncertainties regarding the interpretation at the
Marathon monitor relative to the BP monitor.
L-2.5 Statistical analyses
We used SAS software to perform an analysis of variance comparing mean modeled and
monitored benzene concentrations among 16 equal wind direction sectors of 22.5 degrees
each, numbered clockwise with sector #1 centered on zero degrees. (H0: No difference
exists in benzene concentration with wind sector; Ha: Benzene concentration varies with
wind sector.)
We used Excel software regression analysis to assess the effects of wind sector by
plotting average monitored and modeled concentrations and hourly monitor data by wind
sector. We also examined if a regression existed between wind speed and the ratio of
hourly monitored to modeled benzene concentrations, a measure of model error (H0:
Model error does not vary with wind speed; Ha: Model error varies with wind speed.)
Separate regressions were run for each wind sector, and for the combination of wind
sectors coming from the sources to the monitors. Finally, we developed frequency
distributions of monitor-to-model ratios to examine possible short-term, high-emission
events.
1 1999 NATA Tables - Pollutant Specific Database: http://www.epa.gov/ttn/atw/natal999/tables.Mml
L-5
-------
Figure L-2. Background benzene concentration isopleths (//g/m3), Texas City, I X,
BP Monitor
Marathon Monitor
L-6
-------
L-3. Results and discussion
L-3.1 Comparison of annual average concentrations
A simple comparison of the annual average modeled concentrations at each monitor with
their measured values (minus the estimated contribution of background sources) shows
that, for the BP facility, the modeled estimate (4.0 /ig/m3) is lower than the measured
value (4.5 /ig/m3) by only about 11%. This suggests that our annual benzene emission
estimates for this facility are close to the actual values, potentially being underestimates
by this same amount, 11%. This difference may also be due to other uncertainties, in our
estimate of background source contributions, for example.
A similar comparison for the Marathon facility shows a greater difference, the modeled
estimate of 2.1 /ig/m3 being lower than the corresponding monitored value of 5.5 /ig/m3
by about 72%. This suggests that our annual benzene emission estimates for this facility
may be underestimated by 72%, i.e. low by a factor of 2.6.
Annual means for modeled estimates, monitor data, and the difference between them (A),
are shown in Tables L-3 and L-4. Data were stratified by wind sector in the expectation
that the monitors would be most strongly influenced by the BP and Marathon refinery
emissions when the respective sources were directly upwind and less strongly influenced
at other times. All three quantities varied significantly with wind sector at both monitors
(P<0.001). Shaded cells in Tables L-3 and L-4 indicate that the monitor was downwind
of the source. An analysis of variance comparing the means of hourly modeled and
monitored concentrations (with results shown in the "P<" column) indicate that 28 out of
32 pairs of modeled and monitored annual means were significantly different at the
P- 0.05 level or less). The same results are shown graphically on Figures L-3 and L-4,
below.
Table L-3. BP Monitor, 2004. Comparison of monitored and modeled means of hourly
benzene concentrations, by wind sector (north 1).
Wind Socioi
Mollis
A
Moniloi (j/ni.-S
Modol (j/lii.'-S
I'
1
485
1.75679397
1.8424446
0.0856506
0.001
2
523
1.05785342
1.233489
0.1756355
0.001
3
349
1.2798939
1.506188
0.2262941
0.001
4
359
1.01127838
1.1558281
0.1445497
0.001
5
586
1.15972321
2.1760263
1.0163031
0.001
6
757
2.44873468
4.2204341
1.7716994
0.001
/
992
3.1 !>ii994i>2
9.22691; i!i
6.06 /923 /
0.001
< t
O
1091
4. /(>/22i>0!>
10./390!>41
1 !..!)0()2i)22
0.001
9
!.»9
O.I !>(>!>34(>2
4.3366263
4.4931609
0./I.4
10
189
1.19336501
1.5591725
0.3658075
0.023
11
136
0.17954026
0.7641936
0.5846534
0.205
12
132
0.60588579
1.0385461
0.4326603
0.001
13
160
0.8036979
0.9544022
0.1507042
0.005
14
233
0.72722058
0.8481373
0.1209167
0.001
L-7
-------
Wind Sector
hours
A
Monitor jUQlmd,
Model ^g/mS
P<
15
197
0.93016269
1.0969735
0.1668108
0.001
16
274
1.22653114
1.3211776
0.0946465
0.001
Table L-4. Marathon monitor, 2006. Comparison of monitored and modeled means of
hourly benzene concentrations, by wind sector (north=l).
Wind Sector
hours
A
Monitor jUQlmd,
Model ^g/mS
P<
1
549
1.93530075
1.9589727
0.02367193
0.001
2
387
0.50857501
0.5664341
0.0578591
0.001
3
213
1.00833775
1.0939155
0.08557775
0.001
4
220
0.84957914
5.1024727
4.25289359
0.578
5
389
5.93874823
6.6997172
0.760969
0.001
6
525
8.12713267
13.2994095
5.17227686
0.001
7
852
6.09892079
11.4159108
5.31699001
0.001
8
973
2.2703778
4.2261274
1.95574964
0.001
9
570
3.1886853
5.125814
1.93712874
0.001
10
166
4.48408789
4.9027831
0.41869524
0.001
11
117
3.25752179
3.5228034
0.26528162
0.001
12
74
1.67502297
1.8754324
0.20040946
0.001
13
124
2.10250782
2.1906774
0.0881696
0.094
14
156
0.34009731
0.410141
0.07004372
0.001
15
138
1.61167471
1.7061449
0.09447022
0.023
16
199
0.6831391
0.7269548
0.04381568
0.001
The effects of both refineries can be clearly seen as elevated benzene levels in both the
measured and modeled concentrations. Although the differences between modeled and
monitored results were statistically significant, the magnitude of the differences may not
be important from a policy perspective, namely the residual risk decision framework
which allows for some amount of uncertainty in the assessments by its design.
For the BP monitor, the absolute difference between average measured and modeled
concentrations for 13 of 16 wind sectors was less than 2 «g/m3. Wind sectors 7 and 8,
where the BP monitor was downwind of the BP facility and concentrations were highest,
had the largest model-to-monitor variation. While modeling of sector 7 resulted in an
underestimate relative to monitored data, sector 8 showed an overestimate. In contrast to
the results for all wind sectors combined, for wind sectors 7 through 9 the BP monitor
concentrations averaged 0.81 «g/m3 (9.1 %) less than the modeled estimates; one
explanation for this finding may be that the 2004 emissions inventory for the BP facility
was overstated.
For the Marathon monitor, mean differences between modeled and measured
concentrations were less than 2 «g/m3 for eight of the 16 wind sectors. For wind sectors
5 to 7, where the Marathon monitor was downwind of the Marathon facility and
concentrations were highest, modeled estimates appear to underestimate monitored
concentrations by an average of 6.7 «g/m3 (a factor of 2.6, agreeing closely with the
results for all wind sectors combined). One explanation for these results may be that the
L-8
-------
2006 emissions inventory for the Marathon facility may have understated actual
emissions by 2.6-fold.
L-9
-------
Figure L-3.
BP Monitor, 2004:
Mean Modeled and Monitored Benzene Concentrations
by Wind Sector
| - Modeled (ug/m3) Monitored (ug/m3) |
1
13
9
Figure L-4.
Marathon Monitor, 2006:
Mean Modeled and Monitored Benzene Concentrations
by Wind Sector
| - Modeled (ug/m3) Monitored (ug/m3) |
1
13
9
L-10
-------
L-3.2 Comparison of hourly concentrations
Even if the inventories for both facilities and the dispersion modeling were perfectly
accurate, we could not expect the hourly monitor data to match the hourly modeled
estimates. This is because the emissions inventory is composed only of annual emission
data, and contains no information on hourly variations in emissions. Furthermore, the
monitored data were adjusted only for average background, not hourly background.
Thus, the modeled estimates capture only variation in meteorological effects, whereas the
monitored data capture variation in meteorology, emission rate, and hourly background
variation for which we could not compensate.
For this reason, Figures 1.-5 and L-6 compare hourly monitor data with annual average
modeled estimates for each wind sector. Figure 1.-5 (for the BP site) shows
approximately the same number of points on either side of the 1:1 correlation line,
suggesting little overall bias. However, the 1:1 line does not appear to fit the data well,
suggesting that the model tends on average to underestimate lower monitor
concentrations (e.g., 1 /^g/m3) and overestimate higher ones (e.g., 5 fig/m3). This may be
a result of the simplistic background adjustment of the monitor data or contributions from
other nearby refineries. Short-term variations in background may have caused many of
the high monitor readings at the lower right of Figure 1.-5.
The same trend appears in Figure L-6 (for the Marathon site). However, this figure also
shows that most monitor data points are below the 1:1 correlation line, suggesting an
overall low bias for the model consistent with Figure L-4 above. This low bias appears
most prominently where the model predicted low concentrations, e.g., less than 1 fig/m3,
but the monitor measured levels above 10 /ig/m3, presumably when the monitor was not
downwind of the Marathon refinery. As with Figure L-5, the average background
adjustment method or contributions from nearby refineries may have contributed to this
effect, but Figure L-4 suggests that the model was also biased low.
L-3.3 Regression analysis of hourly concentrations and wind speed
Regression analysis of hourly monitor-to-model ratios (Figures L-7 and L-8, below)
shows that the relationship between measured and modeled concentrations at both
monitors was significantly correlated with wind speed. As wind speed increased at each
location, the tendency for the model to underestimate the measured concentration
increased, reaching more than tenfold at the highest wind speeds. Both the regression
slope and R2 values increased when source was directly upwind of the monitor, vs. the
regressions for winds from all sectors, showing that this effect was somewhat stronger
when the monitor was directly affected by the source.
L-3.4 Hourly monitor-to-model ratios and short-term events
Hourly monitor data were divided by corresponding hourly modeled estimates to develop
hourly monitor-to-model ratios. Figures L-9 and L-10 show frequency distributions of
these ratios for each monitor, at times when the monitor was downwind of the source.
Both distributions are approximately log-normal, but with somewhat exaggerated tails on
the low side. Figure L-9 shows that the mode of the distribution for the BP site occurs at
a 1:1 ratio (mean = 1.2, median = 0.73), further supporting the suggestion that the
modeled estimates were reasonably unbiased. The mode of the distribution for the
Marathon site (Figure L-10) occurred at a monitor-to-model ratio of 2.5 (mean = 2.7,
L-l 1
-------
median = 1.9), consistent with a 2.6-fold underestimate by the model suggested by the
ANOVA results above.
Possible short-term, high-emission events at either facility, representing periods when the
emission rate substantially exceeded the annual average, should appear on Figures L-9
and L-10 as the highest monitor-to-model ratios. In screening for potential acute risks,
OAQPS uses a default assumption that the maximum hourly emission rate may exceed
the annual average rate by tenfold, so that the location with the greatest modeled 1-hour
concentration (based on average emissions but hourly meteorology data) may experience
a concentration ten times higher than modeled. Therefore, the number of monitor-to-
model ratios above ten is of interest. For the BP site, 0.8% of the ratios (16 of 1984)
exceeded ten, and the highest was 20. For the Marathon site, 2.7% of the ratios (35 of
1277) exceeded ten, and the highest was 15. The possible contribution to this result by
short-term variations in both background and contributions from other nearby refineries
is not known, but could be important.
Narrowing this comparison to the highest concentrations at the monitor, application of
the OAQPS default acute screening method (i.e., using the peak-to-mean emission factor
of 10, in combination with hourly meteorology data) at the BP facility results in an
estimate of a peak hourly benzene concentration at the monitor of 2140 /ig/m3, whereas
the maximum measured hourly concentration at the monitor was only 130 /ig/m3, more
than 16 times lower than our screening estimate. This suggests that our screening
approach for this facility is very conservative, and that peak emission rates for this
facility may not vary much from their average values. Application of the default
screening method at the Marathon facility results in an estimate of a peak hourly benzene
concentration at the monitor of 960 /ig/m3, whereas the maximum measured hourly
concentration at the monitor was only 275 /ig/m3, lower by a factor of about 3.5. This
suggests that our screening approach is also conservative for the Marathon facility, but
less so than for the BP facility. If, however, we had adjusted our annual emission
estimates for the Marathon facility to remove bias (i.e., based on the analysis shown
above), the conservatism of our screening methodology goes back up, with our estimate
of peak hourly benzene concentration being 7 times greater than the peak value actually
measured at the monitor.
This result suggests that the tenfold default assumption captured a very high percentage
(though not all) of short-term emissions events at these facilities, but was more than
sufficiently conservative in screening the highest hourly concentrations.
L-3.5 Summary
Modeled concentrations averaged about 11% less than measured concentrations at the BP
facility, but about 72% less at the Marathon facility. When this comparison was stratified
by wind sector (16 sectors of 22.5 degrees each), 28 of 32 model-to-monitor pairs were
significantly different (at the P- 0.05 level or less). For 26 of the 28 significant results the
monitored concentration exceeded the modeled estimate. Limiting the comparison to
periods when the monitors were downwind of their respective sources, monitor
concentrations averaged 0.81 fig/m3 (9.1 %) less than the modeled estimates at the BP
site, but 6.7 fig/m3 (260%) more than modeled estimates at the Marathon site. Given that
both monitors were modeled with similar input data and uncertainties, except for the
emissions data and the calendar year, these results suggest that the inventory for the BP
refinery may be reasonably accurate and possibly slightly overestimated, but the
L-12
-------
inventory for the Marathon site may be somewhat underestimated. However, these
discrepancies are within the range of those expected for such model-to-monitor
comparisons, and may be the result of model error and inaccuracies in other model
inputs.
Comparing hourly monitor data with annual average modeled estimates by wind sector,
the model tended on average to underestimate lower monitor concentrations (e.g., 1
/ig/m3) and overestimate higher ones (e.g., 5 /ig/m3) at both sites. For the BP site (Figure
L-5) the modeled results appeared unbiased, but for the Marathon site (Figure L-6) the
modeled estimated appeared to be biased low, consistent with the comparison of means
for all wind directions.
As wind speed increased at each location, the tendency for the model to underestimate
the measured concentration increased, reaching more than tenfold at the highest wind
speeds. This effect was somewhat stronger when the monitor was directly affected by the
source. This result suggests that the model performs better at lower wind speeds, and is
biased low at higher wind speeds.
Frequency distributions of monitor-to-model ratios (Figures L-9 and L-10) provide
further support for the suggestion that the modeled estimates were reasonably unbiased at
the BP site, but biased low for the Marathon site.
Only 0.8% of monitor-to-model ratios at the BP site exceeded ten, but 2.7% exceeded 10
at the Marathon site. The maximum ratios were 20 and 15, respectively. However, none
of these ratios exceeded 10 when monitor concentrations were highest, suggesting that
the OAQPS approach for screening short-term emissions and exposures was very
conservative at both facilities, and that refinery emissions do not vary dramatically in
time. This conservatism would be further increased if the Marathon emissions inventory
did indeed prove to be underestimated, and was corrected.
L-3.6 Uncertainty
1. Inventory data. The RTR inventory may contain errors in amounts, locations, or
release parameters that would affect the dispersion modeling results. In particular,
activity at the BP refinery was disrupted during 2006, but nevertheless emitted an
unknown amount of benzene that may have influenced monitored concentrations at the
Marathon monitor site. Also, benzene emissions from shipping activities, roadways, and
more distant industrial sources may have influenced both monitors, although the
background adjustment was applied with the intent of removing this effect from the
analysis.
The inventory is limited to annual emission rates, and lacks any information on short-
term variations. Furthermore, it does not include emissions from upset conditions or
emergency releases that the monitors may have measured. Finally, the inventory data for
one of the two facilities had to be adjusted because an explosion altered its emissions.
2. Background adjustment. We estimated the average contribution of unmodeled sources
and subtracted it from the monitor data to improve the comparison with modeled
estimates. A model-to-monitor comparison done as part of NATA
(http://www.epa.gov/nata/mtom pre.html) found that the median model-to-monitor ratio
at 87 sites was 0.93, that 89% of ratios were within a factor of 2 and 59% were within
L-13
-------
30%. Although the comparison results for benzene were better than for any other HAP,
this amount of uncertainty must still be considered substantial, and it is likely that the
background adjustment contributes to it.
In addition, adjustment for short-term background variations was not possible, adding
further uncertainty to short-term model-to-monitor comparisons.
3. Monitor data. As with any study involving sampling and analysis, the monitor data are
subject to both sampling and measurement error. The monitors were also affected by
unmodeled (background) benzene sources, including vehicles and industrial facilities
outside the modeled domain. Some adjustment of the monitor data was also necessary:
hours that lacked wind data were dropped, and non-detect results were entered as one-
half the method detection limit.
4. Meteorological data. While the meteorological data station at the Galveston airport
was considered to be more representative for this application of AERMOD given the
focus on southerly winds, the assessment for a routine application of AERMOD for the
same facilities including receptors for all directions around the emission sources may lead
to a different conclusion. We recognize that selecting the most appropriate
meteorological data for coastal locations is challenging.
Meteorological data are also subject to sampling and measurement errors, and they may
fail to accurately represent conditions for every hour monitored.
5. Other modeling inputs, (a) We relied on information supplied by the American
Petroleum Institute for stack and other release parameters, but we did not independently
verify these data, (b) For BP Chemical, BP Refinery, and Marathon Refinery, the TRI
indicates that a large fraction of benzene releases are fugitive emissions, which may drive
much of the concentration picked up at the two nearby monitoring sites, (c) We believe
that the ways we used to adjust emissions across years were reasonable, but they are still
uncertain, (d) The modeling did not consider topography or building downwash. Each of
these sources of uncertainty will be reflected in the results, but the aggregate error and
direction of potential bias, if any, are unknown.
Of these uncertainties, the most important are probably those associated with the
inventory and the adjustment for background. In particular, the lack of short-term
variability in both these databases effectively limits the input variation available to the
model, and prevents it from fully reproducing the monitor results. This effect limits our
ability to draw conclusions.
Nevertheless, several trends seem clear. It appears that the modeling effort represented
one facility reasonably accurately but underestimated the other by more than twofold.
There is no way to know which (if either) facility is representative of the whole sector.
The model tended to overestimate low monitored concentrations and underestimate high
ones, perhaps not surprising since the model captured only some of the sources of
variability in the monitor data. The model's tendency to underestimate high monitor
levels increased with wind speed, and the increase was more pronounced when the source
was directly upwind. Despite this tendency, however, we found that the OAQPS acute
exposure screen (which assumes ten times the annual emission rate, worst-case
meteorology, and a receptor at the monitor) was protective for these facilities by a
substantial margin.
L-14
-------
L.4 An Alternative Viewpoint
One EPA staff reviewer of this document, Fred Talcott, disagreed with some of the
methods used in this model-to-monitor analysis and some of the conclusions reached by
the authors of this Appendix. Since we did not have time to reach a consensus view on
this issue within EPA or conduct some of the additional analyses suggested by Fred, in
Attachment L-l we present his comments and suggestions as an alternative viewpoint for
the SAB panel to consider as they develop their comments on this analysis and its
interpretations.
L-15
-------
Figure L-5.
Monitored Hourly Benzene Concentrations (ug/m3) vs. Annual Average Modeled
Concentrations for each wind direction, BP Monitor Site, 2004
100 i
0.0001 0.001 0.01 0.1 1 10 100 1000
Monitored Benzene Concentration
L-16
-------
Figure L-6.
Monitored Hourly Benzene Concentrations (ug/m3) vs. Annual Average Modeled
Concentrations for each wind direction, Marathon Monitor Site, 2006
Measured Benzene Concentration
L-17
-------
Figure L-7.
10000
1000
100
Wind Speed vs. Ratio of Hourly Monitored and Modeled Benzene Concentrations,
BP Monitor Site, 2004
All wind directions • Wind from source
10
0.1
0.01
0.001
0.0001
y = 0.0659e
R = 0.26S1
¦ y = 0.647e
R2 = 0.092
0.00001
10
12
14
16
Wind Speed (m/s)
L-18
-------
Figure L-8.
10000
1000
Wind Speed vs. Ratio of Hourly Monitored to Modeled Benzene Concentrations,
Marathon Monitor Site, 2006
All wind directions • Wind from source
100
10
0.1
0.01
R = 0.1122
y = 0.4702e
R = 0.1837
y = 0.8326e
0.001
10
12
14
16
18
Wind Speed (m/s)
L-19
-------
Figure L-9.
Frequency Distribution of Ratio of Hourly Monitored Benzene Concentration to Mean Modeled
Concentration, BP Monitor Site, 2004, Monitor Downwind of Source
Monitor to Model Ratio
L-20
-------
Figure L-10.
Frequency Distribution of Ratio of Hourly Monitored Benzene Concentration to Mean Modeled
Concentration, Marathon Monitor Site, 2006, Monitor Downwind of Source
135
n
28
0 0 0
114
105
83
70
44
48
30
23
K
a
27
I
62
111
103
¦
80
53
47
50
28
7
I
& & & s$ J3 # # ,0s
^ f ^ ^ 1? /" J? # # ^ ^ <£< ,0'f J* N#
Monitor to Model Ratio
L-21
-------
Attachment L-1: Alternative View: A Summary of One EPA Reviewer Comments
on the Model- to-Monitor Comparison
The following is a summary of comments from Fred Talcott, EPA's Office of Policy, Economics, and Innovation
(OPEI) on the current draft of the Model- to-Monitor comparison conducted in conjunction with the Petroleum
Refinery case study. He felt that the current draft does not adequately address these comments, and they are
provided for SAB reviewers as a plausible critique of and alternative to the current draft of Appendix L.
Conclusion: The approach makes sense and much of it is well-presented. Nevertheless, I urge you to qualify
your conclusion that the inventory for the Marathon may be an underestimate, and re-run the analysis
considering the Specific Comments, provided below. The difference in the modeled concentration may be
explained primarily or in part by an underestimate in emissions, but the analysis provided does not make a case
for that conclusion.
Specific Comments:
The focus of these comments is on the "Marathon" monitor (the model compares quite well for the BP monitor).
First: Missing sources.
The analysis uses 2002 emissions for the Marathon, Valero, and Sterling facilities, scaled to assumed
2006 emissions using TRI activity data. Benzene concentrations based on these emissions are compared to
2006 measurements at the "Marathon" monitor. Emissions from the BP refinery were included for the "BP"
monitor comparison for 2004, but not for the "Marathon" comparison for 2006. The reason is the explosion and
fire at the BP refinery in 2005; emissions from that site would be hard to estimate for the 2006 comparison year.
While it is likely that fugitives would have been emitted from the BP site from clean-up and any residual storage
and production activities in 2006, the analysis assumes 2006 emissions from BP are zero.
Further, the analysis does not estimate benzene emissions from shipping and barge traffic in the channel
surrounding the Marathon facility on two sides, nor from shipping activities a little further afield in the channel
between Galveston and Texas City on the mainland. This omission affects modeling at the "Marathon" site
much more than at the "BP" site, which is about 5 km further inland.
Omitting these sources of benzene (emissions from the BP facility and nearby shipping and barge
activities) probably contributes to the underestimate of benzene concentrations at the "Marathon" monitor site.
Second: Meteorological data.
The analysis considered, but rejected, the use of the data from the "Ball Park" met station, and opted to
use met data from the Galveston Airport.
The Ball Park met station is located in an urban area, 1 to 2 km from the "BP" and "Marathon" monitors,
and 2 to 4 km from the modeled facilities. The Galveston Airport met station is located across the bay on the
field at the airport, about 14 km from the refineries. The Galveston Airport met station is located in an area
where there are few building obstructions, about 1 km from a bay on one side and about 2 km from the Gulf of
Mexico on the other.
Although both met stations have roughly similar distributions of wind directions (though Ball Park is more
from the SE, and Galveston from the south), there is more variation in the wind roses for 2004 and 2006 for the
Ball Park station than for the Galveston station. While this variation is given as a reason for preferring the
Galveston station over the Ball Park station, this logic does not make sense. The meteorology is more variable
at the Ball Park met station that is almost on top of the monitors and the sources OAQPS is modeling. The
modeling should reflect the actual variability, and not the steadier, and less representative, conditions 14 km
away at the Galveston station.
OAQPS has provided this reviewer a document titled "Galveston, TX and Texas City Ball Park, TX
meteorological towers." What is not pointed out in the document, but is quite apparent in comparing the wind
roses, is how much greater the wind speed is at the Galveston station in comparison to the Ball Park station.
Reading off the wind roses (Figure 3 for the Ball Park station and Figure 5 for the Galveston station), about 54%
of measurements are below 7 knots for the Ball Park station, but only 19% are below 7 knots for the Galveston
station. Fewer than 6% of the measurements for the Ball Park stations are greater than 11 knots, but about
32% of the measurements at the Galveston station are above 11 knots, with some above 17 knots. That is, it's
a great deal "blowier" on the field on a narrow island facing the Gulf of Mexico than it is inland in a developed
area. This is no surprise.
I .-22
-------
The effect of using an artificially high wind speed is to push plumes from both stacks and fugitive
emissions further down-wind, and artificially to dilute the estimated concentrations at points close to the sources,
such as the two monitoring sites.
Use of the Galveston Airport met station probably contributes a sizable amount to the underestimate of
benzene concentrations at both the "Marathon" and the "BP" monitor sites.
I suggest that you (a) include the information in the document titled "Galveston, TX and Texas City Ball
Park, TX meteorological towers" as part of appendix L, and (b) re-run the analysis, using the met data from the
Ball Park met station.
Third: Adjusting for "background" benzene levels is highly uncertain.
It is interesting and reasonable to try to adjust the monitored levels downward to account for the
contributions from sources other than the industrial facilities you are modeling. The study took the 2002 NATA
results for area and mobile sources at Census tracts in the Texas City area, and created an isopleth map
(Figure 2), interpolating for values at the two monitoring sites. These estimated annual concentrations were
then subtracted from the monitored values for each of the 8,700 hours for the year-long comparisons.
It is not easy to guess whether this might result in an upward or a downward bias in the resulting model to
monitor comparison, but you do need to address this as a possible important source of uncertainty. Looking at
the model to monitor comparisons for the 1996,1999, and 2002 NATA, the study finds substantial uncertainties
for benzene and for the other HAPs that were analyzed. These are not, regrettably, broken down by type of site
(those heavily influenced by industrial emissions versus those with mostly area and mobile sources), but
differences of a factor of over 2.0 are common in the comparisons. Thus, the adjustment factor should be
viewed as having a substantial uncertainty, and this uncertainty then translates into uncertainty in the Texas City
comparisons for the modeled facilities versus the background-adjusted monitors. A part of that is also the
compounding factor of subtracting the estimated annual background from each of the hourly observations.
Thus, this background adjustment inserts a non-trivial uncertainty into the analysis process.
Fourth: Other sources of uncertainty need to be summarized and discussed.
Information provided to this reviewer said that you relied on the API effort to get stack and other release
parameters, as well as any updated stack parameters provided through the FR process. I don't know what we
can say about the completeness and the QC behind these release parameters. For BP Chemical, BP Refinery,
and Marathon, TRI indicates that a large fraction of benzene releases are from fugitives. Fugitives would seem
to drive much of the concentration picked up at the two nearby monitoring sites. I suggest that OAQPS take a
careful look at the release parameters for the facilities and make a judgment about how accurate or how
approximate they may be.
The study uses reasonable, but still uncertain, ways of adjusting across years. It is not clear, but it
appears that for some years you scaled emissions by the ratio of TRI emissions, and used the TRI "activity
ratios" for others. Acknowledge that there is uncertainty in these ratios, and that adds to the uncertainty in the
whole.
Surface roughness can be an important determinant of concentrations, especially from fugitives and in the
near field. You should document the values that you used (different for different wind directions?) and the
rationale. Then, make some qualitative statement about how any uncertainty in surface roughness may
translate into uncertainties in modeled concentrations at the monitoring sites.
With regard to stability class, please indicate which met site was the source for upper-atmospheric met
data. What was the distance from Texas City? Qualitatively discuss how the different location may affect the
validity of the hourly stability class used in the modeling, and the resulting uncertainty in the modeling results.
Make a clear, concise, and coherent statement of the presence and importance of each of these sources
of uncertainty, and say something semi-quantitative or qualitative about their combined impact on the modeling
results.
Fifth: Put these results into the context of wider model to monitor comparisons.
OAQPS has conducted and made available model to monitor comparisons for a few dozen FIAPs for the
1996, 1999, and 2002 NATA analyses. For 1996, the median model-to-monitor ratio for benzene was 0.92; it
was 1.47 for 2002. The modeled benzene concentration was within a factor of 2 of the monitored value in 89%
of cases in 1996, and 69% in 2002. Modeling results were generally not as good for the other FIAPs. If it is true
that 31 % of NATA predictions for benzene were off by two-fold in the most recent version of NATA, that helps to
put the 2.6-fold difference found for Marathon in a different light.
I .-23
-------
Many factors contribute to the model's predictions being different from the monitors -- emission rates,
location errors and approximations (especially for area and mobile sources), meteorological data, release
parameters, surface roughness, etc.
What are we to make of the fact that the computations found for modeling at the "Marathon" site were 2.6-
fold lower than the monitored values?
I make the case, above, that missing sources and use of the wrong met data may explain a significant part
of the under-prediction. But I think that we have an obligation to acknowledge that there is sizable uncertainty in
this kind of modeling, and that the 2.6-fold difference may well be within the noise of this noisy enterprise.
Specific recommendations:
1. Modify the Summary section to say that the differences between model and monitor values at the
"Marathon" site may be explained principally by the omitted sources and the choice of the met station. Indicate
that discrepancies such as these may be within the range of what can be expected in modeling of this type.
2. Include a discussion of the use of the Ball Park versus the Galveston Airport met stations.
3. If at all possible, re-run the analyses using the Ball Park met data. Use the full two years of met data from
the Ball Park station, since there seems to be sizable variation from year-to-year. If computational resources
are an issue, OAQPS might take a random selection of hours, or perhaps six-hour or daily segments from the
2004 and the 2006 Ball Park years, as a short-cut.
4. Add a summary section about all the sources of uncertainties in this modeling. See the first four points,
above.
5. Include a section of ASPEN/NATA model to monitor comparisons. I suggest showing the benzene
comparisons for each of the 1996,1999, and 2002 comparisons, and perhaps just the 2002 data for the other
HAPs. Can you answer the question: What fraction of model to monitor comparisons are within a factor of 1.5;
2; 2.5; 3; 10 (separately for high and for low)? FYI, for the 35 HAPs in Table 1 of the "Comparison of 2002
Model-Predicted Concentrations to Monitored Data," more than half of the HAPs missed the monitored value by
more than a factor of 2-fold in at least 50% of the observations, i.e., more than a two-fold error is more than just
common-place. Greater than a two-fold error is the norm.
I .-24
-------
Appendix M: Sensitivity analysis of uncertainty in risk
estimates resulting from estimating exposures at
Census block centroids near petroleum refineries
-------
^eos%
\ UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
S RESEARCH TRIANGLE PARK, NC 27711
\pr 0^
OFFICE OF
AIR QUALITY PLANNING
AND STANDARDS
MEMORANDUM
DATE:
SUBJECT:
FROM:
TO:
December 23, 2008
Uncertainties Associated with the Use of Census Block Centroids as
Receptors for Chronic Exposure
Mark Morris —
Sector Based AssessmenrGroup (C539-02)
David Guinnup
Sector Based Assessment Group (C539-02)
The HEM-AERMOD model estimates ambient concentrations at the geographic
centroids of census blocks (using the 2000 Census), and at other receptor locations that
can be specified by the user. In cases where the census block centroids are found to be
located on facility property (as determined from aerial imagery), receptors are moved to
the nearest off-site location. The model accounts for the effects of multiple facilities
when estimating concentration impacts at each block centroid, and assesses chronic
exposure and risk only for census blocks with at least one resident (i.e., locations where
people may reasonably be assumed to reside rather than receptor points at the fenceline of
a facility). Chronic ambient concentrations are calculated as the annual average of all
estimated short-term (one-hour) concentrations at each block centroid. Possible future
residential use of currently uninhabited areas is not considered. Census blocks, the finest
resolution available in the census data, are typically comprised of approximately 40
people or about ten households.
The use of census block centroids as receptors for chronic exposure instead of
actual residence locations introduces uncertainty into the risk assessment because
residences within a census block may not be located near the centroid. This is minimized
in highly populated areas because census blocks are typically small in such areas. In less
populated areas, census blocks are typically large, and residences may be nearer to or
farther from the source than the centroid, resulting in higher or lower actual exposures at
those locations than at the centroid. However, this would not seem to introduce bias
Internet Address (URL) • http://www.epa.gov
Recycled/Recyclable • Printed with Vegetable Oil Based Inks on Recycled Paper (Minimum 25% Postconsumer)
-------
because residences seem equally likely to be located on either side of the centroid relative
to the source.
To test for possible systematic bias associated with the use of census block
ccntroids as receptors for chronic exposure, we analyzed a sample of 21 petroleum
refineries. We overlaid census block boundaries and cancer risks at census block
centroids on aerial photographs of refineries with estimated cancer MIR values greater
than or equal to 10 in a million. In some cases, we also overlaid cancer risk contours to
allow estimates of cancer risk at residences that may not be reflected by the census block
or polar receptors. These photographs are given in Figures 1 through 21. The cancer risk
contours were created with geographic information system software using "natural
neighbor" interpolation, which finds (for every point in the modeling domain) the closest
polar receptor risk values and applies weights to them based on proportionate areas in
order to interpolate a value. For several facilities, the density of polar receptors was not
adequate to create cancer risk contours, and estimates of cancer risk at residences were
made using the nearest census block and polar receptors. Because census block centroids
sometimes fall within facility boundaries, I IEM-AERMOD assigns zero risk to census
block centroids that are within a user-specified distance (default is 30 m) of any emission
source. This does not ensure that HEM-AERMOD will not assign a cancer risk to a
census block centroid within facility boundaries, but it is the method that we use in the
absence of specific facility boundary information. As can be seen, several of the figures
show census block centroids within facility boundaries.
) '
Table 1 gives the estimated facility-specific cancer MIR values at the census
block centroid and at the nearest residence for 21 petroleum refineries with cancer MIR
values greater than or equal to 10 in a million. In eleven cases, the census blocks were
small, with a typical distance from the centroid to the block boundary less than 100 m. In
these cases, we estimate that the MIR values at the census block centroid and nearest
residence are identical. There were two cases where census blocks were relatively large,
but for which the residences were located near the centroid. In these cases, we also
estimate that the MIR values at the census block centroid and nearest residence are
identical. In the remaining eight cases, the census blocks were relatively large, and the
MIR values at the centroid were higher than the values estimated at the nearest residence,
with the overestimates ranging from 40 to 2000 percent. In seven of these cases, the
census blocks overlap both facility property and adjacent residential areas. In such
situations, MIR estimates at the centroid are biased high because most of the area
between the centroid and the boundary of the block nearest the facility is not residential.
In summary, in this analysis of facility-specific MIR values, the centroid-
generated values overestimated the residence-generated values by 40 to 2000 percent in
less than half the cases, were equivalent in over half the cases, and there were no cases ^
where the value at a residence exceeded that at the centroid of the census block
containing the residence. The MIR estimate for the source category as a whole was the
same using either methodology. While it is possible that exposures at a residence in a
large census block could be higher than at the centroid of the block, this analysis supports
the use of the centroid as a reasonable representation of the MIR for the nearest receptor,
-------
and it provides strong evidence that the use of the centroid is not creating a low bias in
the overall risk results, indicating, in fact, the tendency for this approach to overestimate
MIR values for the highest risk sources, and thus the MIR for the source category as a
whole.
-------
Table 1. Comparison of Risks at Census Block Centroid Versus Nearest Residence
Facility NEI ID
Maximum Indiv
idual Cancer Risk (in a
Million)
Census Block
Percent
Overestimate
Census Block
Nearest Residence
NEI876
10
10
0
NEI6022
10
10
0
NEI6087
10
1
1000
NEI6436
10
10
0
NEI6475
10
10
0
NEI12711
30
30
0
NEI12791
10
10
0
NEI12988
20
20
0
NEI20174
10
7
40
NED2864
10
10
0
NED 3 031
20
10
100
NED 3 03 9
10
3
300
NED4050
10
10
0
NED4057
10
10
0
NED4898
20
20
0
NEI40371
10
5
100
NEI41771
10
10
0
NEI42040
10
3
300
NEI42309
20
1
2000
NEI CA1910268
10
10
0
NEIPRT$64
10
5
100
-------
Figure 1. Cancer Risk for NEI876
Meters
-------
Figure 2. Cancer Risk for NEI6022
NEI6022 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Centroids
MIR
*> r*
750 f,000
—Haanai Meters j
U 125 250 500
-------
Figure 3. Cancer Risk for NEI6087
NE16087 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Ceniroids
0.00067 - 0.027
0.038 - 0.19
Meters
Residences
-------
Figure 4. Cancer Risk for NEI6436
NEI6436 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Centroids
-------
Figure 5. Cancer Risk for NEI6475
NEI6475 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Centroids
-------
Meters
NEI12711 Cancer Risk
(in a Million)
| | Census Block Boundaries
Census Block Centroids
¦ MIR
-------
Figure 7. Cancer Risk for NEI12791
NEI12791 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Centroids
Meters
-------
Figure 8. Cancer Risk for NEI12988
NEI12988 Cancer Risk
(in a Million)
~ Census Slock Boundaries
Census Block Central ds
Meters
-------
Figure 9. Cancer Risk for NEI20174
Residences
NEI20174 Cancer Risk
(in a Million)
~ Census Block Boundaries
Census Block Centroids
¦ MIR
000025 -0.22
0 23 - 1.9
¦|2- 5.4
j I 5j5- 11
112- 18
19-28
29 - 39
¦I <40 - 55
-------
Figure 10. Cancer Risk for NEI32864
NE132864 Cancer Risk \ /
(in <1 Million) -
I Census Block Boundaries
1 1
Census Block Ceritroids
¦ MIR
| 10.0023 - 33
| 13.4- 4.9
|H| 5 - 8.2
| 8.3 - 15
.16-23
23-55
56- 110
~ 120 - 220
¦ 230-430
H 443 - 870
-------
Figure 11. Cancer Risk for NEI33031
Meters
f.
HH
• Jm
tL ":£~ J6A
sfe: it ¦ ¦ > .C*/
&."*W v
•***// ' 47/, \
» w* 'jT^/l,*% >
N ii
NEI33031 Cancer Risk
{in a Million)
Census Block Boundaries
Census Block Centroids
MIR
O Polar Receptor
5.V
!^V
¦4WM[W
L ' " .i£ . ¦ Wk -
9i
SBBfeggg
qi tal ul rjbe
-------
Figure 12. Cancer Risk for NEI33039
Meters
Residences *
NEI33039 Cancer Risk
(in a Million)
~ Census Block Boundaries
Census Block Centroids
¦ MIR
] 0.0013 -0.053
J 0.054 -0.31
m| 0.32 - 0.67
0.88 -1.8
11.9 -3.3
3.4 -5.3
J 5.4-7.8
^¦7.9-13
-------
Figure 13. Cancer Risk for NEI34050
NEI34050 Cancer Risk
(in a Million)
Census Block Boundaries
Census Block Centra ids
-------
Figure 14. Cancer Risk for NEI34057
digital Globe'
NEI34057 Cancer Risk
(in a Million*
~ Census Block Boundaries
Census Block Centroids
¦ MIR
0.0061 - 1.3
1.4- 1.8
V
\ V£Y® ;*
1,000
¦¦Meters
^ijr ij ffiQQ S^Gn^Pi tal £1 o be
-------
Figure 15. Cancer Risk for NEI34898
Residences
NEI34898 Cancer Risk
(in a rvillioni
ensus Block Boundaries
Census Block Centre ids
.0013- 0.15
Meters!
-------
Figure 16. Cancer Risk for NEI40371
Residences
NEMO371 Cancer Risk
(in a MNIion)
~ Census Block Boundaries
Census Block Centroids
¦ MIR
0.0013- 03
| 0.31 - 0.4
^^0.41 - 0.69
^0.7-1.5
11.6- 4
4.1 - 11
~~1 12 - 31
-------
Figure 17. Cancer Risk for NEI41771
Meters
NEW!771 Cancer Risk
(in a Million)
~ Census Block Boundaries
Census Block Centroids
¦ MIR
-------
Figure 18. Cancer Risk for NEI42040
Residences
450 600
Meters
MEI42040 Cancer Risk
(ill a billion)
I ICensus Block Boundaries
Census Block Centroids
¦ MIR
I lO.Q - 0.12
¦ 0.13- 0.53
¦ 0.54- 1.4
I 11.5 - 2J8
¦ 2.9 - 4.7
I 14.8 - 7.1
H11 - 15
-------
Figure 19. Cancer Risk for NEI 42309
IIEI42309 Cancer Risk
(in a billion)
~ Census Block Boundaries
Cere us Block Centroids
¦ MR
~000041- 0.11
¦0.12-0.15
¦0.16-026
~027 - 0j57
~058 -15
~ 16-4
~4.1 - 11
¦ 12-33
Meters
Residences
Residences
-------
Meters
NEICA1910268 Cancel Risk
(in a Million)
~ Census Block Boundaries
Census Block Centroids
¦ MIR
no.00012- 0.4
~ 0.41 - 2
~ 2.1 - 4.4
~ 4.5-7.9
~ 8-13
~ 14- 19
~ 20- 27
~ 23- 37
~ 38- 54
~ 55- 100
-------
Figure 21. Cancer Risk for NEIPRTS64
NEI PRT$64 Cancer Risk
{in a Million)
f | Census Block Boundaries
Census Bloci; Centnoids
¦ MIR
~ 0.00035-1.1
~ 1.2-1.6
1^1.7-2.7
| 2.8-4.9
| | 5-9.5
I 196-19
~ 20 -38
39-78
~'79-160
70-320
Residences
-------
Appendix N: Analysis of the effect of considering long-term
mobility of receptor populations on estimates of lifetime
cancer risk
N-l
-------
N.1 Methods
This appendix describes probabilistic calculations performed on the estimated 70-year
inhalation cancer risk output from the HEM-SCREEN model for the petroleum refining1
and Portland cement source categories. The goal of these calculations was to adjust the
HEM-SCREEN estimates to account for the receptor population's periodic relocation to
new residences, either within or outside the exposure area. These calculations do not
account for the effects of short-term behavior patterns (e.g., daily commuting or time
spent outdoors) on exposure, nor do they consider sources of variation and uncertainty
other than relocation and emigration.
N.1.1 Residence Time Input
In previous air toxics assessments EPA used a residence time frequency distribution from
the EPA Exposure Factors Handbook (1996), based on an analysis by Johnson and Capel
(1992). This analysis has two shortcomings that currently limit its usefulness for residual
risk assessments. First, the underlying data were from 1987. Second, the approach was
based on one-year move rates that appear to have underestimated residence time for long-
term residents, as reported by the US Census Bureau in its Surveys of Income and
Program Participation (SIPP, 1996 and 2001).
At EPA's request, Ted Johnson (one of the original authors) updated the Johnson and
Capel (1992) analysis to reflect the more recent SIPP data and a newer, more complete
modeling approach. Johnson's model randomly selected subjects from the US Census
Bureau's American Community Survey database and estimated (1) time already spent in
the residence, (2) future time to be spent in the residence, and (3) future length of life.
These estimates were combined to predict the total time, past and future, that the subject
would occupy the current residence. Johnson then compared the modeling results with
SIPP residence time data and adjusted the results to compensate for "residential inertia"
(i.e., a tendency in the SIPP data for long-term residents to have lower-than-expected
move rates). EPA is in the process of updating the Exposure Factors Handbook, and
expects to replace its current residence time recommendations (Table N-l, below) with
Johnson's new estimates (Table N-2). However, the entire Handbook must undergo
scientific review, and we are not certain when that process will be complete.
Table N-l. Residence time estimates (in years) from Johnson and Capel (1992).
From To Probability
0
1.5
0.05
1.5
2.5
0.05
2.5
3.5
0.15
3.5
9
0.25
9
16
0.25
1 For the petroleum refineries source category, the modeling exercise was conducted using the NPRM draft
baseline assessment. Thus, the "before" results may differ somewhat from the final version of the
assessment presented in the main report.
N-2
-------
From To Probability
16
26
0.15
26
33
0.05
33
41
0.03
41
47
0.01
47
51
0.005
51
55
0.003
55
59
0.001
59
85
0.001
Table N-2. Revised residence time estimates (in years) described above.
From To Probability
0
1
0.05
1
2
0.05
2
5
0.15
5
12.6
0.25
12.6
27.2
0.25
27.2
45.6
0.15
45.6
56.3
0.05
56.3
74.9
0.04
74.9
81
0.005
81
91
0.004
91
100
0.001
N.1.2 Emigration Input
The second distribution describes the likelihood that each relocation will remove the
individual from the exposure area. This distribution is based on a regression analysis of
5-year population migration data (US Census Bureau, 2003) from seven states: Maine,
Connecticut, Virginia, Ohio, Louisiana, Nebraska, and Montana. These states were non-
randomly selected to provide a range of areas with different sizes and population
characteristics that included large/small, rural/urban, east/west, dense/sparse population,
and counties varying widely in size. Finally, two states (CT and ME) also included
Census data broken down by townships (169 for CT and 523 for ME), supporting an
extension of the regression into areas smaller than counties. Land areas were calculated
using population density data from the late 1990's (Wright, 2003).
The regression indicates a highly significant (P<0.00005, R2 = 0.75) inverse relationship
between the fraction of moves from outside a jurisdiction (e.g., a state, county, or
township) and the area of that jurisdiction (Figure N-l). This regression confirms
common sense - as the target area becomes smaller, it becomes less likely that a random
movers will "hit" it. The modeling domain for the source category was considered in the
aggregate (rather than separately for each facility) to allow for the possibility that a
person who moves away from one facility could relocate near another facility in the same
source category. For the sake of simplicity in this analysis, the total population size was
assumed to be constant, and the rate of emigration from the area was therefore assumed
equal to the rate of immigration. That is, each person coming in replaced a person who
N-3
-------
left. This is unlikely to be true in the real world, but including a population growth
variable would require site-specific information for each facility.
Figure N-l. Association between immigration rate and total area.
Mean fraction of the moves from OUTSIDE a jurisdiction:
Based on 2000 Census data from counties in ME, CT, VA, OH, LA
NE, and MT, and townships (~) in CT and ME
0.8
0.7
0.6
0.5
0.4
0.3
0.2
FT = 0.7496
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
total (land and water) area - square miles
The HEM-SCREEN assessment included 153 refineries and 91 Portland cement
facilities, all for which we had input data at the time this analysis was developed. Each
facility had a circular modeled domain with a 20-km radius, for a total modeled area of
about 120,000 and 70,000 square miles, respectively. These areas correspond to those at
the far right of Figure N-l, suggesting an emigration rate of about 20%. Therefore, the
calculations for all receptors assumed that 20% of those who changed residences left the
exposure area and 80% relocated to another residence within it. The estimates of total
modeled area did not consider overlap of neighboring facilities, which could cause an
underestimate of the emigration rate. On the other hand, some facilities in each source
category could not be modeled, which could overestimate emigration. No allowance was
made for individuals returning to the exposure area after once leaving it.
There are several important sources of uncertainty associated with this approach:
1. The regression data were selected non-randomly in order in maximize the
diversity of the included populations (e.g., with respect to density, land area, etc.)
These data may not be fully representative of the populations affected by the
emissions from these source categories.
N-4
-------
2. The population size was assumed to be constant. To the extent that this is not true
for the receptor populations affected by these source categories, this important
simplifying assumption will underestimate the number of people exposed where
populations are growing and overestimate it where populations are contracting.
The uncertainty associated with constant population size is probably minor
compared to other uncertainties.
3. The Census data represented a comparison of respondents' residence with their
residence five years earlier. People may have moved more than once during this
period, potentially biasing the immigration rate estimates either low or high. The
Census data themselves are a subset of the population who filled out longer data
forms, who may not fully represent the entire population.
4. This analysis has not specifically considered several demographic variables that
are known to strongly influence move rates, e.g., age, income, marital status, and
owner/renter status. These factors may vary substantially among Census tracts,
meaning that move rates may also vary substantially. Applying central tendency
move rates to the entire modeled domain means represents an important source of
uncertainty.
5. Individuals who once emigrated from the modeled area were assumed never to
return to it. Because the size of the unmodeled area is so much larger than that of
the modeled area, this assumption probably did not have a strong effect on the
results.
6. The total modeled areas of these assessments represent the upper limit of the
regression, and the estimated emigration rate therefore is subject to greater error
than an area in the regression's center (e.g., 1000 to 10,000 square miles). The
regression itself is subject to statistical error, meaning the true relocation rates
should be viewed as falling within a range of approximately 10-30%, rather than
fixed at exactly 20%. Using a different relocation rate within this range might
have produced significantly different results.
N.1.3 Lifetime Inhalation Cancer Risk Input
The initial lifetime (70-year) inhalation cancer risk estimates from the HEM-SCREEN
model included risk estimates for approximately 53 million and 89 million individuals
who live within 20 km of one or more of the modeled Portland cement and petroleum
refining facilities, respectively. However, in order to focus on the most-exposed
subpopulation of potential regulatory interest, mobility adjustments were calculated only
for individuals whose 70-year risk estimate was 1 in a million or greater.
2. Probabilistic Calculations. Calculations were performed using Crystal Ball™ and
Microsoft Excel™ software. Each probabilistic simulation included 100,000 trials based
on the Monte Carlo sampling method, using the same random number seed to ensure
repeatability.
N-5
-------
The probabilistic calculations for all individuals began by randomly selecting one
individual from the HEM-SCREEN distribution of lifetime risks. This individual was
assigned a residence time (selected randomly from the residence time distribution), and
the risk associated with the exposure in that residence was calculated as follows:
Equation 1
RiskR = x rt
70y
Where: RiskR = Estimated cancer risk from years in residence R
Risk/ = Estimated total lifetime cancer risk (from HEM-SCREEN)
RT= Residence time (y)
This individual was also assigned a random binary emigration value that determined if
he/she moved to another home within the exposure area at the end of the residence time,
or left the assessment either by emigration or death. If the binary emigration value for
the individual was 1, the person was deemed not to have emigrated and was randomly
assigned another residence within the exposure area. If the emigration value was zero,
the person was deemed to have emigrated. Individuals were tracked through seven
residences until they either emigrated or reached 70 years of total exposure (i.e., "died").
The procedure was limited to seven residences to optimize calculation times, because test
runs showed that virtually all individuals either emigrated or died before reaching an
eighth residence.
Individuals who emigrated were assumed not to return to the exposure area in a
subsequent move, which could potentially underestimate the lifetime exposure of some
individuals. Total lifetime risk was the aggregate of risks associated with all residences
occupied, as follows:
Equation 2
n
RiskT = RiskH
R=1
Where: RiskT = Estimated total cancer risk associated with multiple residences
RiskR = Estimated cancer risk from years in residence R
n = Total number of residences occupied by the individual, up to 7
The simulation results were extrapolated into the full population. Because individuals
who "died" or moved away were replaced by new individuals, the size of the true
receptor population was greater than was considered by the 70-year analysis. The size of
this full receptor population was determined by the ratio of the average aggregate
residence time to the total assumed 70-y lifetime.
N-6
-------
N.2 Results
Outputs of the probabilistic residence time adjustment are shown below in Table N-3.
Figures N-3 and N-4 compare estimated 70-y cancer risk distribution with the residence-
time adjusted risks for petroleum refineries and portland cement facilities, respectively.
Table N-3. Comparison of populations exceeding three lifetime inhalation cancer risk
benchmarks, with and without adjustment for long-term mobility, for two source
categories.
Cancer Risk
Portland Cement
Petroleum Refineries
Unadjusted
Adjusted
Unadjusted
Adjusted
> le-4
0
0
0
0
> le-5
125
43
4,378
2,556
> le-6
5,066
2,955
430,800
292,003
Figure N-3.
Petroleum Refineries Source Category:
Effect of Adjusting for Long-Term Mobility
On Estimated Lifetime Cancer Risks >= 1 in 1 million
70-year risk estimate Adjusted risk estimate |
0.9
0.8
0.7
-Q
re
¦Q
O
0.6
o 0.5
>
3
E
3
o
0.4
0.3
0.2
0.1
0
¦9
¦8
¦7
¦6
¦5
¦4
Log10 of Estimated Individual Risk
N-7
-------
Figure N-4.
Portland Cement Source Category:
Effect of Adjusting for Long-Term Mobility
On Estimated Lifetime Cancer Risks >= 1 in 1 Million
70-year risk estimate —"—Adjusted risk estimate
0.9
0.8
0.7
-Q
re
¦Q
O
0.6
o 0.5
>
3
E
3
o
0.4
0.3
0.2
0.1
0
-9
¦8
¦7
¦6
¦5
¦4
Log10 of Estimated Individual Risk
N.3 References
Johnson T, and Capel JA. 1992. Monte Carlo approach to simulating residential
occupancy periods and its application to the general U.S. population. Research Triangle
Park, NC: U.S. Environmental Protection Agency, Office of Air Quality and Standards.
US Census Bureau, 2003. State-to-State Migration Flows: 1995 to 2000 (CENSR-8);
Migration for the Population 5 Years and Over for the United States, Regions, States,
Counties, New England Minor Civil Divisions, Metropolitan Areas, and Puerto Rico:
2000 (PHC-T-22). http://www.census.gov/population/www/cen2000/phc-t22.html
US Census Bureau, 2007. Survey of Income and Program Participation. Available
online at http ://www. sipp. census. gov/sipp/
US Census Bureau, 2007. American Community Survey. Available online at
http://www.census.gov/acs/www/
US EPA, 1997. Exposure Factors Handbook. National Center for Environmental
Assessment, Office of Research and Development. PB98-124225. Available online at
http://www.epa.gov/ncea/efh/ Residence time data in Table 15-167.
Wright, J. W. 2003. The New York Times Almanac. Penguin Books, New York, NY.
ISBN: 0142003670.
N-8
-------
Appendix O: Potential importance of hazardous air pollutants
lacking dose-response values
-------
0.1 Introduction
In their comments following the SAB consultation on the first RTR risk assessment plan in
December 2006, the panel expressed concern about that EPA's quantitative assessment methods
generally omit risks from HAPs that lack peer-reviewed dose-response assessments. The panel
requested a sensitivity analysis to test how important the effects of unassessed HAPs might be to
the total risk. In response to this comment, we conducted a simple "what-if' analysis based on
median and upper-bound estimates of toxic potency for these substances. We included in this
analysis the Portland cement and petroleum refinery1 source categories individually, and also all
US sources combined. Calculations were done separately for cancer and effects other than
cancer.
This study is intended as a rough range-finding exercise to examine the potential magnitude of
risks posed by HAPs that lack dose-response assessments, and to prioritize HAPs for toxicity-
testing and dose-response assessment. The results are not intended to propose dose-response
ranges for unassessed compounds in refined risk assessments done in support of regulatory
decisionmaking.
0.2 Methods
The analysis was based on toxicity-weighting of the 2002 NEI, a process that provides an
estimate of relative potential cancer risk and noncancer respiratory hazard posed by each HAP.
Health risks associated with exposure to environmental chemicals are a function of (1) the
amount of chemical released, (2) the toxicity of the chemical, (3) the dispersion of the chemical
in the environment (as influenced by release conditions and meteorology), and (4) receptor
exposure (as influenced by receptor location and behavior). Toxicity-weighting represents a
partial analysis of health risks, using information covering only areas (1) and (2). Toxicity-
weighting is useful as a screening tool because the data are readily available, the analysis can be
conducted quickly, and the inputs account for a large part of the variation in risks obtained from
a complete assessment. However, toxicity weighting is useful only for relative estimates of risk,
and the omission of information in areas (3) and (4) means that toxicity-weighted results are
more uncertain than results from a complete assessment. Toxicity-weighting is most
appropriately used as a screening and prioritization tool.
We weighted the pollutant emissions as follows: (1) for noncancer respiratory effects, the
emitted amount for each chemical was divided by its RfC or similar chronic no-effect exposure
level; (2) for cancer, the emitted amount of each chemical was multiplied by its inhalation URE
for cancer.
For HAPs that lacked an RfC or URE, we selected as surrogates the following range of values
selected from the universe of chronic RfCs and UREs in the OAQPS table of prioritized chronic
dose-response values for inhalation exposure (http://www.epa.gov/ttn/atw/toxsource/tablel .pdf):
1 For the petroleum refineries source category, the analysis was conducted using the NPRM draft baseline inventory,
which differs slightly from the final version of the inventory described in the main report.
0-1
-------
IVivcnlilc of
loxicilv
Rf(-
(nm ivT)
i r i:
( 1 /JU 111')
5
2.28
1.0e-6
25
0.2
6.0e-6
50
0.0098
6.8e-5
75
0.00065
6.1e-4
95
0.000023
4.8e-2
All HAPs lacking an RfC were assigned this range of surrogate RfCs. Only HAPs lacking a URE
but having an EPA or IARC WOE equivalent to "possible carcinogen" or greater were assigned
the range of surrogate UREs. Toxicity-weighted emissions (TWEs) for cancer and noncancer
effects were kept separate. TWE's were normalized by dividing each score by the maximum
TWE from all chemicals that had a dose-response value.
We did not attempt to reduce these toxicity ranges (e.g., by grouping HAPs by chemical class or
structure-activity characteristics) because there is no universally accepted grouping system.
Developing and defending such a system would require a major effort that would be beyond the
scope of a range-finding exercise.
0.3 Results and Discussion
Results of the analysis are shown in Figures 0-1 to 0-6. TWEs appear as points for chemicals
that have dose-response values and ranges for those that do not. TWE ranges for both
carcinogens and noncarginogens spanned about five orders of magnitude (as did the surrogate
RfC and URE ranges in the table above). Chemicals on each figure are shown in order of
decreasing TWE, with the median TWE value used for sorting ranges. The graphs, with one
exception, were limited to the 40 chemicals with the highest TWEs.
For petroleum refineries, Figure 0-1 shows that four unassessed noncarcinogens (2,2,4-
trimethylpentane, POMs, biphenyl, and carbonyl sulfide) are emitted in amounts that could
produce a relative TWE of 0.1 or higher if they had 75th percentile toxicity or worse. Figure 0-2
shows only one unassessed carcinogen, quinoline, that could produce a relative TWE of 0.1 or
higher if it had 95th percentile carcinogenic potency.
For Portland cement facilities, Figure 0-3 shows five chemicals (carbonyl sulfide, POM, 1,3-
propane sultone, chromium III, and bromoform) that could produce a relative TWE of 0.1 or
higher at 95th percentile toxicity. Of these, only carbonyl sulfide would have a TWE of 0.1 or
higher at 75th percentile toxicity. Figure 0-4 shows that no unassessed carcinogens would be
likely to contribute a TWE greater than 0.1, even at 95th percentile potency.
Considering HAPs emitted from all sources nationally, Figures 0-5 shows shows five chemicals
(2,2,4-trimethylpentane, carbonyl sulfide, POM, and propionaldehyde) that could produce a
relative TWE of 0.1 or higher at 95th percentile toxicity. Of these, only 2,2,4-trimethylpentane
would have a TWE of 0.1 or higher at 75th percentile toxicity. Figures 0-6a and 0-6b show one
unassessed carcinogen, ethyl acrylate, with the potential for a TWE greater than 0.1, if it had 95th
percentile potency.
2 Low RfCs connote high toxicity, so the RfC decreases as toxicity increases. UREs are directly proportional to
carcinogenic potency, so the URE increases as potency increases.
0-2
-------
This toxicity-weighting analysis, while obviously simplistic, is nevertheless useful for
determining whether particular assessments have overlooked any potentially important
unassessed chemicals, and for informing decisions prioritizing pollutants for toxicity testing and
dose-response assessment. Obvious candidates for study or dose-response assessment that
emerge from the analysis include 2,2,4-trimethylpentane, carbonyl sulfide, POM, biphenyl,
propionaldehyde, and ethyl acrylate. Similar analyses can be conducted easily on other source
categories, and with other inventory years, to identify new candidates.
In addition to the limitations discussed in the introduction above, it's important to reiterate that
TWE scoring of carcinogens was limited to substances that lacked a URE but had a WOE
determination of "possible carcinogen" or worse. This assumes, in effect, that all chemicals that
lack a WOE, or that have a WOE of "no data," are not carcinogens. This is unlikely to be true,
and for this reason this analysis may underestimate the potential TWE contributions of
unassessed carcinogens.
0-3
-------
1.0E+03 -i
Figure 0-1. Petroleum Refinineries: Noncancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs lacking RfCs compared with TWEs HAPs with RfCs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E+02
1.0E+01
1.0E+00
1.0E-01
ra
E 1.0E-02
1.0E-03
1.0E-04
1.0E-05
I.0E-06
^ ^ vcf
c!^
& & & & r& v v v«
V //rf
ft- ^ ^ ^ o*
oO^'
&
0-4
&
-------
1.0E+00
1.0E-01
1.0E-02
1.0E-03
"D
n 1.0E-04
Figure 0-2. Petroleum Refineries: Cancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs with WOE of "possible" or higher but lacking UREs, compared with TWEs HAPs with UREs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E-05
1.0E-06
1.0E-07
^ ^ c/ / # / / /
J? ^ ^ . cS^ v
J.0 .0 . ?> J.0 .N0 .0 . >$>
> ^ ^ .# J# .
^ J* ^ v
•5^ >& >&> sS> s& -
* J" x/ J" # J'
° .*« <5S /. sS>
/o- <£> ^ <0- *£¦ yfp ^ ,s
7// ^ <•*/>
oP >&> c£> >& r& ^ & r& >&>
^ r?> v ^ . «***?>
0-5
-------
1.0E+01
1.0E+00
1.0E-01
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
Figure 0-3. Portland Cement Facilities: Noncancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs lacking RfCs compared with TWEs HAPs with RfCs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
CP*8
' ^ vO° o° kP XS° -cK
* ///A
V
^ /v/ #
J? x
-------
1.0E+00
1.0E-01
1.0E-02
1.0E-03
Figure 0-4. Portland Cement Facilities: Cancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs with WOE of "possible" or higher but lacking UREs, compared with TWEs HAPs with UREs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E-05
1.0E-06
1.0E-04
.V ,^
/•
cF
/" Y/S////S'// '*'// ,-rr-
~
,
-------
1.0E+01
Figure 0-5. All NEI sources: Noncancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs lacking RfCs compared with TWEs HAPs with RfCs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E+00
1.0E-01 -
Uj 1.0E-02
"D
n 1.0E-03
ra
E
i.
o
z 1.0E-04
1.0E-05
1.0E-06
1.0E-07
0-8
-------
1.0E+00
Figure 0-6a. All NEI Sources: Cancer Tox-Weighted Emissions for HAPs 1-40
TWE ranges for HAPs with WOE of "possible" or higher but lacking UREs, compared with TWEs HAPs with UREs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E-01
LU
> 1.0E-02
"D
0)
N
re
£
O 1.0E-03
1.0E-04
1.0E-05
^ -e? cP -rsv o° ^ & r&s o° cP cP .<0' .0° .<0' .<0' S~ _B° <<3< ^ .Jf* ,\^ ,„<° j# ^ /O' A-
ssT & ,\<^ ^ .-I? ^
" ,1> ^ -r^ .J? ^ ^
\« ^ Kl?
- „!¦ #
tx-
c/
A
_# <$¦
o
-------
1.0E+00
1.0E-01
1.0E-02
1.0E-03
1.0E-04
1.0E-05
1.0E-06
1.0E-07
Figure 0-6b. All NEI Sources: Cancer Tox-Weighted Emissions for HAPs 41-80
TWE ranges for HAPs with WOE of "possible" or higher but lacking UREs, compared with TWEs HAPs with UREs
(Ranges are 5th, 25th, 50th, 75th, and 95th percentile TWEs)
1.0E-08
/ <& \>s\ .($> <-^ rP .?& X<> .?&
& J'.J'
3 0<.C
•V * ysj s*y ** s'fjrs
r „pn
-------
Appendix P: Comparison of RTR Emissions Inventory Data and
Refineries Emissions Model (REM) Data
1
-------
P.1 Background
Throughout the development of the Risk and Technology Review (RTR) program, one
potentially significant area of uncertainty has been the quality of emissions data from individual
sources. The general approach has been to model these sources based on data contained in the
National Emissions Inventory (NEI), provide the results of this modeling to the public in an
Advanced Notice of Proposed Rulemaking, soliciting further data from individual sources or
state/local air pollution agencies. While this approach has proved somewhat successful,
questions often remain as to the variable quality of much of the data. Given the requirement to
examine the potential risks from all hazardous air pollutants listed in the Clean Air Act,
inconsistencies often remain across both pollutants and individual sources within a category.
EPA's Science Advisory Board (SAB) and EPA's Office of Inspector General (OIG) have
commented on the emissions uncertainties associated with the RTR rulemaking. In addition to
expressing concern over emissions uncertainties, both groups have suggested that EPA conduct
sensitivity analyses regarding the potential uncertainties in emissions data. Independent of these
reviews, EPA has considered anecdotal data on petroleum refineries emissions and has expressed
concern that refinery emissions and risk estimates may be understated in the NEI. 1
For these reasons, we have modeled risk from petroleum refineries using a different set of
emissions data, generated using the Refineries Emissions Model (REM). Our aim is to compare
two different, but reasonable, sets of emissions data to examine the potential scope of uncertainty
in the emissions data and the implications of these differences for estimated cancer inhalation
risks. This appendix documents this alternative approach used to assess the baseline emissions
and risks from the petroleum refineries MACT I source category. This analysis is based on an
emission factor approach, using emission factors along with facility-specific production and
throughput data to estimate emissions. In addition to its relevance for this particular source
category, this analysis may serve as an example for the RTR program more broadly. Other
source categories may have even less certain emissions data in the NEI, perhaps making them
candidates for this type of analysis.
In the present analysis, we employed a HAP emissions model developed specifically for
petroleum refineries known as the Refinery Emissions Model, or REM (RTI, 2002; Lucas,
2007b). This model was used to generate an "REM" emissions database, including emissions
estimates for each refinery in the source category. These emission estimates are compared, by
individual pollutant, with those generated using the RTR method (see section 2.2.1 of RTR
Methodologies Report). We then used these emissions data to develop alternative risk estimates
that are compared to the risk estimated using the RTR emissions data.
P.2 Methods
Emissions and excess cancer risk associated with Refinery MACT 1 emission sources have been
estimated from data reported in the NEI; these emission and associated risk estimates are
described in the main body of this report. The detailed emission factor analysis described in this
1 See EPA Docket No. EPA-HQ-OAR-2003-0146, "Potential Low Bias of Reported VOC Emissions from the
Petroleum Refining Industry."
P-2
-------
appendix is being used to provide an alternative baseline HAP emission estimate for all Refinery
MACT 1 emission sources. The modeling approach and assumptions used to estimate the
emissions and the source release characteristics have been described elsewhere (RTI, 2002). In
essence, the emission factors used in this model are based on MACT compliance, but for the
most part do not take into account the impact of any state/local regulations, or any "overcontrol"
on the part of facilities beyond MACT requirements. For example, because cooling towers do
not currently have a MACT standard, we assume they are uncontrolled, whereas it is likely that
some portion of cooling towers is controlled for state regulations or other reasons. We also
know that some portion of external floating roof storage vessels have some controls.
Although this analysis uses process-specific production capacities, it provides emission estimates
by source type for the facility. For example, the analysis models emissions from classes of
storage vessels (e.g., crude oil tanks, gasoline tanks), not the emissions from individual storage
vessels. It also accumulates and assigns the emissions to one large area source representing the
tank farm rather than attempting to estimate the number and characteristics of individual storage
vessels. Since the analysis by RTI (2002), some enhancements to the emission estimates and
source characteristic assumptions were made, partly as a result of the "22 Refinery Study"
(Lucas, 2007b). These enhancements are described in the Addendum of this appendix.
The REM uses facility-specific data on the types of processes and their capacities to estimate
emissions for each refinery in the United States (US) and its territories. The original model
includes algorithms for estimating emissions from various petroleum refinery MACT 1 sources,
i.e., storage vessels, equipment leaks, wastewater treatment systems, cooling towers, flares,
product loading, as well as from various MACT 2 sources, i.e., process heaters, boilers, catalytic
cracking units, catalytic reforming units, and sulfur recovery plants (RTI, 2002). While the
overall framework of the model is the same, some revisions to the model have been made, as
mentioned above. For this analysis, emissions output were only estimated for the MACT 1
sources. The product loading estimate assumes all light and middle distillates are loaded in
tanker trucks. Marine vessel loading operations, when co-located at a refinery (and therefore
subject to Refinery MACT 1), are typically controlled; these emissions were included in the
emissions estimates for flares. Similarly, nearly all miscellaneous vents at a refinery are
controlled and the emissions from these vents are also included in the flare emission estimates.
Table K in the Addendum is a complete table of facility-specific assumptions used for the REM
analysis. Table L shows all of the emissions results by facility. For comparison, Table M
provides the RTR emissions by facility.
After emissions estimates have been developed, a dispersion/risk analysis was undertaken.
Chronic inhalation exposure concentrations and associated health risk from each facility of
interest were estimated using the Human Exposure Model in combination with the American
Meteorological Society/EPA Regulatory Model dispersion modeling system (HEM-AERMOD,
sometimes called HEM3). More details on the HEM modeling system and the approach used to
estimate health risks is outlined in Section 2.2.2 of the RTR Methodologies Report. The REM
analysis consisted of modeling 1512 refineries nationwide. The RTR (NEI-based) risk modeling
2 While we have emissions data for 153 facilities, 151 facilities included in the REM dataset were modeled in this analysis. One
facility that was not modeled is thought to be a duplicate and the other has minor emissions and does not have a
corresponding RTR facility with an NEI_ID.
P-3
-------
included 156 refineries. However, there are some adjacent refineries that have come under
single ownership. At times, the permits for these facilities are merged and the State reports the
emissions as a single refinery. In other cases, the permits are kept separate and the State reports
emissions for these facilities separately (although under the definition of facility within the CAA,
the plants are contiguous and under common ownership/control, so they should be a single
facility for the purposes of the CAA). Thus, while there is a small discrepancy in the number of
refineries modeled, the two analyses effectively cover identical refining operations.
The risk associated with each facility's estimated emissions was evaluated using the same
dispersion models, exposure assumptions, and unit risk factors that were used to estimate risk
based on the RTR data. It is important to note, however, that unlike the RTR database for which
it is possible to report source-specific locations and release characteristics (18-42% of emission
points include unique data, depending on the parameter), these details are not included in REM.
Instead we made assumptions about location and other specifications that are described in the
Addendum. For example, the REM risk analysis is based on all emissions being released at or
near the centroid of the facility and uses default emission source release parameters. As such,
differences in the risk results between RTR and REM may be a function of emissions magnitude,
but they may also be caused by differences in release characteristics (e.g., individual storage
vessels vs. tank farms), and/or emission source locations.
P.3 Comparison of Emission Estimates - REM vs. RTR
The total nationwide HAP emissions estimate at baseline projected by the REM emissions
estimates is about 17,800 tons/yr; the total nationwide HAP emissions estimate in the RTR
dataset for refineries is about 6,820 tons/yr. Thus, the REM analysis projects approximately 2.6
times higher emissions than the RTR data. As indicated in Table 1, benzene emissions are
estimated to be about 1,990 tons/yr nationwide in REM, whereas the RTR dataset includes a total
of 693 tons/yr of benzene emissions. REM includes 135 tons/yr of 1,3-butadiene, which is about
8.3 times higher than the 16.2 tons/yr reported in RTR. For naphthalene, REM estimates
emissions of about 113 tons/yr, and the RTR data set contains about 77.0 tons/yr. Thus, while
the REM data indicate higher emissions of benzene and 1,3-butadiene than the NEI by a factor of
8.3, naphthalene emissions are only about 47 percent higher.
Table 1 shows the HAP emissions estimates for those pollutants included both in the RTR and
REM datasets. Overall emissions are higher in the REM dataset for 17 of the 19 HAPs that
appear in both estimates. However, there are 37 pollutants that are shown as emissions from at
least one facility in RTR that are not included in REM. In addition, REM assumes that most
pollutants would be expected from essentially all refineries; only six pollutants are reported to be
emitted from more than 100 RTR facilities, and nine pollutants are reported by 50 or more
facilities. The pollutants reported by the most facilities in RTR are benzene, toluene, ethyl
benzene, xylenes, hexane, naphthalene, cumene, 1,3 butadiene, and methanol.
As shown in Table 1, overall REM emissions are higher by a factor of 2.6, but this factor varies
significantly among pollutants. Part of this difference stems from the fact that the REM analysis
applies the emission factors at all petroleum refineries, but RTR reports emissions of these
pollutants only for a subset of facilities. For some pollutants, the fraction of sources where
emissions are reported in the RTR database represents a majority of facilities (e.g., for benzene,
P-4
-------
144 of 156 refineries report emissions), but for others this percentage represents significantly less
than half of refineries (see Tables 1 and 2).
As we have noted, REM only covers a subset of HAPs, though we believe this includes most of
the major hydrocarbons thought to be common to virtually all petroleum refineries. However, the
RTR database includes emission estimates for 37 pollutants not covered by REM, reported to be
emitted from anywhere from 1 to 34 facilities nationwide. Table 2 lists these pollutants, the
amount or RTR-reported emissions, and the number of refineries that reported these emissions.
Several of these (e.g., vinyl chloride) are considered to be highly toxic. It is not clear whether
these are erroneously reported, a function of specific products of a given refinery, or whether
they represent systematic under-reporting for the other refineries in the source category.
At a facility level, there is great variability in the magnitude of difference in emissions. About
two-thirds of the facilities have emissions estimates from REM and RTR within the same order
of magnitude. However, many REM emission estimates are over one order of magnitude higher,
and some are over 1000 times higher. It is unclear from this analysis what factors are driving
these differences (e.g., lack of reporting of certain pollutants, difference in quantity of certain
pollutants, or incorrect assumptions about emissions in REM).
Table
: Comparison of HAP Emission Estimates Bet
ween RTR and REM Datasets
Polliiliinl
RTR 1.inissions (tp\)
# l-'iicililics \\/ RTR l-'niissituis
RIM I'.inissioiis U|)\)
# l-'iicililics «/ REM l-'.niissituis
1,3-Butadiene
16.2
71
135
153
2,2,4-
T rime thy lpentane
137
48
1170
153
Benzene
693
144
1990
153
Biphenyl
3.28
21
11.0
153
Cresols
8.64
27 or 28
112
153
Cumene
52.1
1
162
153
Ethyl Benzene
244
129
506
153
Formaldehyde
7.74
28
23.0
153
Hexane
1180
127
4770
153
Methanol
549
61
10.9
152
Methyl Isobutyl
Ketone
92.0
5
925
152
Methyl Tert-
Butyl Ether
347
45
2220
153
Naphthalene
77.0
104
113
153
Phenol
17.1
42
88.3
153
POM 71002a
16.0
44 to 61
5.15
151
POM 72002b
5.28
23 to 58
7.18
151
Styrene
5.46
25
372
153
Toluene
1650
135
3010
153
Xylenes (Mixture
of o, m, and p
Isomers)
1570
128 to 156
2200
153
TOTAL
6670
156
17800
153
A POM 71002 is a modeling category that contains the following pollutant descriptions from RTR and/or REM: chrysene, polycyclic organic
matter, PAH total, benz[a]Anthracene, 16-PAH, and PNA/PAH.
B POM 72002 is a modeling category that contains the following pollutant descriptions from RTR and/or REM: anthracene, fluorine,
phenanthrene, pyrene, benzo[g,h,i]Perylene, fluoranthene, acenaphthene, and perylene.
More information on the POM modeling categories can be found at http://www.epa.gov/ttn/atw/natal999/99pdfs/pomapproachian.pdf.
P-5
-------
Table 2. Pollutants in the RTR but not included in the REM database ranked in
decreasing order by # facilities reporting.
IIAP ( iileiion
1". miss ions (lp\)
# l-'iicililics Rcpnrlinii '
Tetrachloroethylene (Perchloroethylene)
15.3
34
Hydrogen Fluoride (Hydrofluoric Acid)
51.3
32
Diethanolamine
36.7
22
Carbon Disulfide
3.52
15
Carbonyl Sulfide
2.01
15
Acetaldehyde
0.195
14
Ethylene Dichloride (1,2-Dichloroethane)
0.609
11
Hydrochloric Acid (Hydrogen Chloride [Gas Only])
10.7
10
Ethylene Dibromide (Dibromoethane)
0.695
8
Ethylene Glycol
21.8
8
Glycol Ethers
3.16
4 to 8
POM 76002b
0.0000482
2 to 7
POM 75002°
0.000625
2 to 6
Carbon Tetrachloride
1.66
5
Methyl Chloroform (1,1,1-Trichloroethane)
0.979
5
Trichloroethylene
0.567
5
Dioxins/Furans
0.000105
3 to 5
Chlorobenzene
0.144
4
Methylene Chloride (Dichloromethane)
0.202
4
Bis(2-Ethylhexyl)Phthalate (Dehp)
0.0013
3
Vinyl Acetate
0.0825
3
Dibenzofuran
0.0254
2
Methyl Chloride (Chloromethane)
0.012
2
p-Dioxane
0.013
2
1,1,2,2-Tetrachloroethane
0.0052
1
1,2,4-Trichlorobenzene
0.0003
1
1,4-Dichlorobenzene
0.000245
1
Acetophenone
0.0840
1
Acrylonitrile
0.0015
1
Aniline
0.026
1
Ethyl Chloride
0
1
Methyl Bromide (Bromomethane)
0
1
Pentachlorophenol
0.002
1
p-Phenylenediamine
0.031
1
Propylene Oxide
0
1
Quinoline
0.037
1
Vinyl Chloride
0.137
1
TOTAL
150
A A range of numbers may be presented because of potential facility overlap resulting from the aggregation of multiple RTR pollutants into a
single HAP category. We did not go back and determine the actual number of overlapping facilities within those categories.
bPOM 76002 contains individual POM species for which the UREs are between: 5e-5
-------
P.4 Comparison of Risk Estimates
In general, we see a modest increase in risk estimates for REM compared to RTR modeling.
Table 3 indicates that the highest maximum individual risk (MIR) for an individual facility (i.e.,
the source category MIR) for the REM and RTR analyses. The highest MIR using REM data is
30 in one million (3 x 10"5) using the high-end benzene potency and 20 in 1 million using the
low-end benzene potency.3 The source category MIR for the REM analysis was driven by
benzene, naphthalene, and POM. The highest MIR (the source category MIR) is also 30 in 1
million (3 x 10"5) based on RTR data, but it occurs at a different facility. The source category
MIR for the RTR analysis was driven by naphthalene and POM. Because benzene is not a driver
at this facility, the MIR using RTR data is also 30 in 1 million using the low-end benzene
potency estimate.
Additionally, the distribution of individual facility MIRs for the entire source category is shifted
upward using REM data as compared to RTR, assuming the high-end cancer potency value.
Using the REM emissions estimates 135 facilities have an MIR greater than 1 in 1 million and 45
facilities have a MIR greater than 10 in 1 million. Using the RTR emissions data, 77 facilities
had MIRs greater than 1 in 1 million and 5 facilities had MIRs greater than 10 in 1 million. We
do not know what the distribution of REM or RTR facility MIR estimates would be using the
equally probable lower estimate of benzene potency.
The estimate for cancer incidence using the REM emissions estimates is three to four times
higher than the incidence estimate using the RTR emissions estimates. Using the low-end cancer
potency value, the REM incidence is 0.1 excess cases per year and the RTR incidence is 0.03
excess cases per year. Using the high-end benzene cancer potency value, the REM incidence is
0.2 excess cancer cases per year and the RTR incidence is 0.05 excess cancer cases per year.
These results are also displayed in Table 3 below. Looking across facilities, about two-thirds of
the facilities, as analyzed, were within the same order of magnitude, most of the rest were one
order of magnitude different, and a handful of outliers were two or more orders of magnitude
different. Table J in the Addendum shows the full set of cancer incidence estimates.
The EPA IRIS assessment for benzene provides a range of plausible unit risk estimates. This
comparative analysis used the highest value in that range, 7.8E-06 per ug/m3, and provides a
conservative estimate of potential benzene cancer risks. The low end of the range is 2.2E-06 per
ug/m3. We applied this low-end value to estimate the potential range in cancer incidence, shown
in Table 3, but did not use it in any other aspect of the REM analyses. In the RTR analysis, we
were able to report the source category MIR because benzene was not a driver at that facility.
The distribution of facility MIRs in both REM and RTR is based on the high-end benzene cancer
potency value. Therefore, the distribution of facility MIRs from the REM and RTR analyses
could be lower (and not necessarily proportionately so) when the lower estimate for benzene is
applied.
3 The EPA IRIS assessment for benzene provides a range of plausible unit risk estimates between 2.2E-06 per ug/m3 and 7.8E-06
per ug/m3. While we originally did this analysis using the high-end of that range, we have since tried to add low-end
calculations where possible without completely remodeling.
P-7
-------
Table 3. Summary of Risk Estimates Projected from the Rr
"R and REM Analyses
Pa ram el or
KIM
KIK
Number of facilities modeled
151
156
Annual HAP emissions (tons/yr)
17,800
6,820
Highest Maximum Individual Lifetime Cancer Risk (MIR, in
20 to 30
30
(naphthalene,
POM)
1 million) from any one Refinery
(benzene,
naphthalene,
POM)
No. Facilities with MTR > 100 in 1 million
0
0
No. Facilities with MTR > 10 in 1 million
41
5
No. Facilities with MTR > 1 in 1 million
135
77
Estimated Cancer Incidence (excess cancer cases per year)
0.1 to 0.2
0.03 to 0.05
Contribution of HAP to Cancer Incidence^
benzene
63%
48%
naphthalene
17%
21%
1,3-butadiene
11%
5%
pomb
6%
15%
A These percentage contributions are based on the high-end benzene cancer potency value. They likely will be
different assuming the low-end benzene cancer potency value.
B POM refers to groups 71002 and 72002 in the REM dataset because no other groups are represented in REM.
P. 4.1 Facility Risks
Looking across facilities, the relative ranking of facility-specific MIRs varied between the RTR
and REM approaches. Table 4 shows the 20 highest facility MIRs using REM data, the
corresponding RTR MIR estimates, and the magnitude of difference in the emissions estimates.
Table 5 similarly shows the 20 highest facility-specific MIRs based on the RTR data and the
corresponding MIRs using REM data.
Only two facilities are ranked among the top 20 facilities in both analyses. Interestingly, all but
one MIR estimates based on RTR data (Table 5) are higher than the corresponding REM MIR
estimates at those same facilities; however, these differences are less than 10-fold and almost
half (9) are roughly the same, i.e., have ratios of 1. Similarly, the highest MIRs using REM data
are almost all higher than corresponding RTR MIRs (Table 4), but there is more variability in the
magnitude of difference. About half of the MIRs for these facilities are less than 10-fold higher
than the corresponding RTR MIR estimates. Also, two of these facilities have a three-order
magnitude of difference. A full comparison of MIR estimates is included as Table I in the
Addendum.
As mentioned previously, this section is based on cancer MIR values assuming the high-end
benzene cancer potency value. The comparisons would likely be different assuming the low-end
benzene potency value because of the difference in benzene emissions estimates between REM
and RTR datasets; however, without specifically calculating those values, we cannot say how
different they would be.
P-8
-------
Table 4. 20 Highest Maximum Individual Risk at REM facilities vs.
iTR estimates
Facility II)
UKM Cancer MIU
(in 1 million)
UTU ( ancer MIU
(in 1 million)
Ualio
(UKM/UTU)
PET NEI34872
30
1
40
PET NEI109
30
4
7
PET NEI46556
30
6
5
PET NEI40732
30
5
6
PET NEI20467
20
1
20
PET NEICA1910268
20
10
2
PET NEI6022
20
10
2
PET NEI7781
20
6
3
PET NEI11450
20
2
10
PET NEI11192
20
1
30
PET NEI20154
20
0.007
3000
PET NEI18406
20
9
2
PET NEI6130
20
6
3
PET NEI11574
20
5
4
PET NEI42309
20
20
1
PET NEI13371
20
5
4
PET NEICA0370363
20
2
10
PET NEI6519
10
5
2
PET NEB 3 03 9
10
10
1
PET NEI876
10
20
0.7
A Numbers in this table are rounded to one significant digit. Facilities were determined by sorting first by
descending REM Cancer MIR then by descending REM Cancer Incidence, and the list was capped at 20.
Table 5. 20 Highest Maximum Individual Risk at RTR facilities vs. REM estimates A
Kacililv II)
UTU ( ancer MIU
(in 1 million)
UIM ( ancer MIU
(in 1 million)
Ualio
(UTUrUI.M)
PET NEI12711
30
9
3
PET NEI34898
20
7
3
PET NEI12988
20
10
2
PET NED 3 031
20
10
2
PET NEI42309
20
20
0.8
PET NEI42040
20
10
1
PET NEI876
20
10
1
PET NEI34057
10
2
9
PET NEI41771
10
9
1
PET NEI6475
10
5
3
PET NEI6095
10
1
9
PET NEI6087
10
8
1
PET NEI6436
10
10
1
PET NEIPRT$64
10
2
5
P-9
-------
U I K (-jinccr MIR
KIM ( sincor MIK
Usilio
I'sicililv II)
(in 1 in ill ion)
(in 1 million)
(UTUrklM)
PET NEI34050
10
5
2
PET NEI20174
10
5
2
PET NEI18394
10
8
1
PET NEICA1910268
10
20
0.5
PET NED2864
10
10
1
PET NEI40371
10
4
2
A Numbers in this table are rounded to one significant digit. Facilities were determined by sorting first by
descending REM Cancer MIR then by descending REM Cancer Incidence, and the list was capped at 20.
P.4.2 Pollutant Risks
For the highest REM facility MIR, benzene, naphthalene, and POM were the risk drivers,
assuming the high-end benzene potency value. Naphthalene and POM were the risk drivers for
the RTR MIR. Benzene, naphthalene, 1,3-butadiene, and POM were the risk drivers for the
REM cancer incidence. These were also drivers for the RTR cancer incidence, but benzene and
1,3-butadiene contribute more overall using REM data. We did not assess how these HAP
contributions to cancer incidence using REM or RTR would change assuming the lower estimate
of benzene cancer potency.
Whereas we determined the emissions of benzene are about three times more in REM than RTR,
they make up about 15% more of the relative cancer incidence risk. 1,3-butadiene emissions are
about eight times greater in REM than RTR, and their relative contribution to overall cancer
incidence is about double using REM than using RTR data. The relative influence of
naphthalene on cancer incidence is roughly the same, and the influence of POM is greater using
RTR than using REM data. RTR contains two more toxic groups of POM that are not included
in the REM data. In addition to total quantity of these pollutants emitted, the number of facilities
reporting these pollutants (only two-thirds of RTR facilities report naphthalene and one-half
report 1,3-butadiene emissions) along with the relative contributions of pollutants that are not
included in REM may also influence these contributions. The relative contributions of individual
pollutants to the overall REM and RTR cancer incidence would likely change when calculated
using the low-end benzene potency value.
P.5 Limitations and Uncertainty
While this analysis provides a general comparison of the standard inventory approach to
gathering emissions data to the emission factor approach, using REM in this case, it is not
without significant uncertainties. Some of the major differences are described in detail, and
Table 5 includes a list of specific differences in the two approaches.
P. 5.1 Emissions Estimates
Both RTR and REM emissions data are modeled estimates, based on few, if any, actual site-
specific measured data. RTR emissions estimates typically do not include record of calculation
method and are based on the 2002 NEI with some information updated through 2005. They are
rarely measured and there may be some similarities between the method used for REM and the
P-10
-------
methods used at some facilities and/or states to compile RTR data. REM-based emissions
estimates are calculated using emissions factors and are generally a function of production and
process charge capacities based on the Energy Information Administration's (EIA) Petroleum
Supply Annual 2004 (EIA, 2005). As with any generic approach, the REM analysis cannot
account for differences due to site-specific modifications at individual facilities, and actual
emissions may be greater or less than estimated for the purposes of this analysis. The
assumptions made about particular emission point (e.g., storage tanks) specifications are
explained in the Addendum to this Appendix. Also, as all these data are limited to annual
emission rates; this analysis does not attempt to estimate short-term releases or health risks
associated with such releases.
For REM, we have attempted to use emission factors that are consistent with the requirements
associated with the existing MACT regulations. For those emission points not controlled by the
existing MACT standards (i.e., cooling towers), no controls are assumed. If facilities control
emissions beyond the level of MACT, whether to meet state/local regulations, to provide a
"buffer" below those allowed under MACT, or for any other reasons (e.g., occupational exposure
reduction), those controls are not reflected in this analysis other than some state control
considered for equipment leaks. The fact that additional control beyond what is allowed (either
uncontrolled or to meet MACT), is not considered in REM may account for differences between
RTR and REM emission estimates. The extent of the difference they account for is unknown
because we do not have facility-specific control data.
P.5.2 Pollutant Coverage
The REM covers 19 organic pollutants and pollutant categories that represent the majority of
HAP emissions by mass. This does not represent the full range of possible pollutants emitted
from at least some facilities in this source category; the RTR database reports emissions of 37
additional HAP categories for MACT I petroleum refineries. While much less important in
terms of gross emissions, several of these pollutants are relatively potent in terms of their
potential health effects. For example, RTR indicates that some facilities emit from MACT I
processes tetrachloroethylene (i.e., perchloroethylene), some of the more toxic POM species,
dioxins/furans, and vinyl chloride, some of which have relatively high cancer potency values,
and the REM analysis does not address these pollutants.
We are uncertain to the extent to which these missing pollutants should be considered for more
facilities within this source category, and we have not evaluated the impact they would have on
overall MIR and cancer incidence if we did include them more broadly.
P. 5.3 Facility Risk Modeling
Whereas RTR sometimes contains detailed emission point specifications (18-42% of the time,
depending on the parameter, for petroleum refineries), REM estimates emissions more broadly,
using default stack parameters and not accounting for specific number of emission points or their
locations within the facility. Therefore, we have had to make assumptions about the size and
location of these sources within the facility, as described elsewhere in this appendix (for
example, see Addendum). Placing area sources in the center of facilities tends to dampen the
extreme risk estimates from those sources, assuming that risk is independent of where these
sources are actually located. Therefore, this may result in an understatement of high MIRs.
P-ll
-------
While these differences influence the risk results, we are uncertain of the magnitude of their
influence without doing more targeted and detailed analysis of this question.
Table 5. Differences between the REM-based and RTR-based emissions and risk estimates
RTR
RUM
1 Emissions estimates methodology
Methods unreported. For some
emission points, such as equipment
leaks and cooling towers, refineries
may estimate using monitoring data
and equipment leak correlation
equations. There is no national
requirement to produce emissions
estimates using a standard protocol
or identifying what emissions
points must be reported.
Emissions factors from AP-42 or RTI
(2002).
Pollutant coverage
There are no national requirements
for what pollutants must be
reported. For petroleum refineries,
RTR happens to contain 56 HAP
reported at between one and 144
facilities of 156. Additionally,
there are no standards for speciating
data. For example, sometimes
VOCs are reported but not
speciated by HAP and they are not
included in RTR.
REM was designed to include 19
pollutants that were thought to cover
the common pollutants from all
refineries. As such, emissions of these
pollutants are estimated for each
facility.
Level of control assumed
If controls are on, emissions
estimates account for them, but
RTR does not have facility-specific
control information.
REM generally assumes facilities are
controlled at the MACT level. For
cooling towers, which are not
currently controlled by MACT are
assumed to be uncontrolled. Estimates
for equipment leaks account for
control requirements from states and
consent decrees.
Modeling parameters
Depends on what information is
provided in RTR. While about
40% of emission points include
facility-specific stack height, only
about 20% include facility-specific
temperature, diameter, flow rate,
and velocity. If facility-specific
data are not known, national-,
source classification code (SCC)-,
or standard industrial classification
(SlC)-defaults are applied.
Assume emission points are located in
the center of the facility. Apply tiered
size categories based on refinery crude
capacity. REM assigns stack
parameters based on the generalized
SCC.
P-12
-------
P.6 Summary
Emissions estimation and risk modeling are complex processes and given the uncertainties
discussed above (e.g., differences in modeling of area sources), it is challenging to draw firm
conclusions as to the reasons for these findings. The following are the salient points we believe
one can take from this analysis:
1. Across all petroleum refineries and HAPs, total HAP emissions estimated with REM are
2.6 times higher than those in the RTR database. At the individual facility level, the
differences can span an order of magnitude or more.
2. On an aggregate level, the MIR results of the REM analysis are similar to the RTR
results. The source category MIR for both the REM and RTR analyses was 30 in a
million (though not at the same facilities) using the high-end benzene cancer potency
value. Using the low-end benzene value, the MIR in the REM analysis dropped to 20 in
1 million while the MIR in the RTR analysis remained 30 in 1 million because the source
category MIR for the RTR analysis was driven by POM and naphthalene, and not
benzene. The source category MIR for the REM analysis was driven by benzene,
naphthalene, and POM.
3. Assuming the high-end benzene potency, we found a shift toward higher facility MIR
estimates. 135 facilities in the REM analysis have MIR estimates greater than 1 in 1
million and 41 facilities have MIR estimates greater than 10 in 1 million, whereas in the
RTR analysis, 77 facilities have risks greater than 1 in 1 million and five facilities have
MIR estimates greater than 10 in 1 million. We do not know what the distribution of
facility MIR estimates for REM or RTR is using the equally-probable low-end estimate
of benzene potency.
4. The top 20 facilities with the highest MIRs based on RTR data have REM-based MIR
estimates within the same order of magnitude. For the top 20 REM-based MIR estimates,
there was somewhat more variability in the magnitude of differences compared to RTR-
based MIR estimates; 14 of these facilities showed differences in estimates of less than an
order of magnitude, but the remainder of differences were at least a factor of 10 (and as
high as 3,000-fold). Using the low-end benzene estimate may alter these differences,
depending on the relative amounts of benzene estimated at each facility.
5. The facilities with the highest MIRs (using the high-end benzene cancer potency value)
in either approach are generally different facilities. This suggests a more pronounced
difference in the influence of the emissions estimation approach at the facility level than
in aggregate. Additionally, the facilities with the highest MIRs in either case, with two
exceptions, are not among the facilities with the most dramatic differences in emissions.
These order of magnitude changes for facilities did not shift any individual facilities to
have MIRs greater than or equal to 100 in 1 million, but we cannot judge how alternative
emissions estimation approaches might affect other source categories. We did not
evaluate this issue using the low-end cancer potency value.
6. Depending on which benzene cancer potency estimate is used, the estimate for cancer
incidence using the REM emissions estimates is three to four times higher than the
P-13
-------
incidence estimate using the RTR emissions estimates (using the high-end benzene
potency estimate, REM incidence is 0.2 cases per year and RTR incidence is 0.05 cases
per year; using the low-end benzene potency estimate, REM incidence is 0.1 cases per
year and RTR incidence is 0.03 cases per year).
7. Petroleum refineries are highly regulated facilities for which emissions are thought to be
relatively well understood compared to many other source categories. The relative
similarity in MIRs may be unique in this case. It is difficult to generalize the results of
this analysis to other source categories.
P-14
-------
P.7 References
EIA. 2005. Petroleum Supply Annual 2004. Prepared by the Energy Information
Administration, Washington, DC. Available at:
http://www.eia.doe.gOv/oil_gas/petroleum/data_publications/petroleum_supply_annual/p
savolume 1 /psavolume 1 hi storical. html
Lucas, B. 2007a. Memorandum from B. Lucas, EPA/SPPD, to Project Docket File (EPA
Docket No. EPA-HQ-OAR-2003-0146). Average Refinery Stream Composition. August
6, 2007. Docket Item No. EPA-HQ-OAR-2003-0146-0003.
Lucas, B. 2007b. Memorandum from B. Lucas, EPA/SPPD, to Project Docket File (EPA
Docket No. EPA-HQ-OAR-2003 -0146). Collection of Detailed Benzene Emissions Data
from 22 Petroleum Refineries. August 20, 2007. Docket Item No. EPA-HQ-OAR-2003 -
0146-0015.
Lucas, B. 2008. Memorandum from B. Lucas, EPA/SPPD, to Project Docket File (EPA Docket
No. EPA-HQ-OAR-2003-0146). Storage Vessels: Revised Control Options and Impact
Estimates. October 30, 2008. Docket Item No. EPA-HQ-OAR-2003-0146-0144.
RTI. 2002. Petroleum Refinery Source Characterization and Emission Model for Residual Risk
Assessment. Prepared for U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards, Research Triangle Park, NC. EPA Contract No. 68-D6-0014.
July 2, 2002.
TANKS 4.09d Software, www.epa.gov/ttn/chief/software/tanks/index.html.
U.S. EPA (Environmental Protection Agency). 1995. Compilation of Air Pollutant Emission
Factors. Section 5.1. AP-42. Office of Air Quality Planning and Standards, Research
Triangle Park, NC.
U.S. EPA (Environmental Protection Agency). 1998. Locating and Estimating Air Emissions
from Sources of Benzene. EPA-454/R-98-011. Office of Air Quality Planning and
Standards, Research Triangle Park, NC.
U.S. EPA Office of Air Quality Planning and Standards, Office of Air and Radiation. 2009.
Draft Final Baseline Residual Risk Assessment forMACTI Petroleum Refining Sources.
January 15, 2009.
P-15
-------
Addendum
Details ofModelins Approach
The basis for this emission factors modeling approach can be found in Petroleum Refinery
Source Characterization and Emission Model for Residual Risk Assessment (RTI, 2002), with
more recent adjustments as described here. REM emissions estimates are generally a function of
production and process charge capacities.
Revised Emission Factors for Equipment Leaks
For equipment leaks, based on the 22 Benzene Study (Lucas, 2007b), revised emission factors
were developed to account for different stringencies of leak detection and repair (LDAR)
programs. Three tiers of LDAR programs were defined based on leak definitions and inclusion
of connectors as follows:
1) Leak definition of 500 or 1,000 ppmv including connector monitoring
2) Leak definition of 500 or 1,000 ppmv; no connector monitoring
3) Leak definition of 10,000 ppmv
The emission factors for benzene were projected using the small and large model plant
equipment component counts and average benzene concentrations for various refinery process
units from Locating and Estimating Emissions of Benzene (USEPA, 1998). The equipment leak
emission factors are summarized in Table A.
Table A. Emission Factors for Benzene from Fugitive Equipment Leaks
Large
Kmissions of licn/ene (lons/yr per process unit)
Ucfinerv
Tier 1
Tier 2
Tier 3
Process I nil
( ul-olT
Small
Large
Small
Large
Small
Large
Crude Distillation
50,000
0.0146
0.0296
0.0326
0.0631
0.0628
0.1247
Vacuum Distillation
25,000
0.0018
0.0067
0.0038
0.0125
0.0074
0.0265
Catalytic Cracking
17,500
0.0108
0.0111
0.0218
0.0255
0.0456
0.0475
Catalytic Reforming
10,000
0.0409
0.0530
0.0858
0.1131
0.1688
0.2253
Hydrocracking
5,000
0.0180
0.0292
0.0382
0.0816
0.0741
0.1347
Thermal Cracking (coking)
10,000
0.0063
0.0110
0.0139
0.0278
0.0277
0.0481
Thermal Cracking
(visbreaking)
10,000
0.0103
0.0184
0.0192
0.0415
0.0432
0.0769
Hydrotreating/Hydrorefining
35,000
0.0130
0.0185
0.0283
0.0415
0.0545
0.0790
Alkylation (sulfuric acid)
5,000
0.0044
0.0044
0.0097
0.0092
0.0180
0.0187
Isomerization
2,500
0.0377
0.0298
0.0768
0.0653
0.1584
0.1295
Polymerization/Dimerization
1,000
0.0014
0.0015
0.0023
0.0033
0.0064
0.0062
Full-Range Distillation
5,000
0.0145
0.0254
0.0282
0.0557
0.0640
0.1069
Aromatics (as CRU)
5,000
0.0409
0.0530
0.0858
0.1131
0.1688
0.2253
Product Blending
5,000
0.0233
0.0282
0.0523
0.0573
0.1003
0.1195
P-16
-------
Table A. Emission Factors for Benzene from Fugitive Equipment Leaks
Large
Kmissions of lien/ene (lons/yr per process unit)
Uefinerv
Tier 1
Tier 2
Tier 3
Process I nil
( ul-olT
Small
Large
Small
Large
Small
Large
Hydrogen Plant (MMcfd)
10
0.0001
0.0004
0.0001
0.0007
0.0002
0.0021
Other Lube Oil Processes
5,000
0.0112
0.0094
0.0250
0.0204
0.0509
0.0412
MEK Dewaxing
5,000
0.0020
0.0056
0.0044
0.0128
0.0078
0.0206
Asphalt Plant
5,000
0.0005
0.0001
0.0011
0.0004
0.0022
0.0006
Sulfur Plant
75
0.0001
0.0001
0.0004
0.0002
0.0006
0.0004
Each refinery was assigned an equipment leak code based on its consent decree requirements or
State requirements so that appropriate equipment leak benzene emission factors were assigned to
each refinery. For refineries where this information was not available or applicable, the default
values for leak definition of 10,000 ppmv were used (Equipment Leak Code = 3). The emission
factors for benzene for each of the process units that are present at the refinery were summed to
calculate the facility's total benzene emissions from equipment leaks. The total benzene
emissions were subsequently multiplied by refinery-wide average process stream individual HAP
to benzene concentration ratios to calculate the emissions of other HAP at the refinery. These
concentration ratios were revised based on the relative volume of each processing or product
stream to crude input. The average concentration ratios used to estimate the fugitive equipment
leaks emissions for HAP other than benzene are summarized in Table B. These concentration
ratios were multiplied by the total mass fugitive equipment leak emissions calculated for benzene
to estimate the fugitive equipment leak mass emissions of the other HAP compounds.
Table B. Concentration Ratios Used for Equipment Leak Emission Estimates
CASUN
MAP
Average Uefinerv
Si ream Liquid
( oncenl ral ion" (\vl %)
Ualio of MAP lo lien/ene
Concenlralion1'
106-99-0
1,3-Butadiene
0.0007
0.0006
540-84-1
2,2,4-Trimethylpentane
2.27
1.97
71-43-2
Benzene
1.15
1
92-52-4
Biphenyl
0.040
0.034
1319-77-3
Cresols
0.29
0.25
98-82-8
Cumene
0.43
0.37
100-41-4
Ethylbenzene
1.02
0.88
110-54-3
Hexane
4.05
3.50
1634-04-4
Methyl tertiary butyl ether
0.67
0.58
91-20-3
Naphthalene
0.33
0.29
108-93-0
Phenol
0.21
0.18
100-42-5
Styrene
0.67
0.58
108-88-3
Toluene
3.86
3.34
P-17
-------
Table B. Concentration Ratios Used for Equipment Leak Emission Estimates
Average Ucfinery
Si ream Liquid
Uatio of MAP lo Ken/cue
CASUN
MAP
( 'oncenl rat ion" (\vl %)
Concentration1'
1330-20-7
Xylene
4.13
3.57
a Weighted average composition of all liquid process streams (Lucas, 2007a).
b Ratio of weighted average liquid concentration of selected HAP to weighted average liquid concentration for
benzene.
Finally, the source characteristics for the process equipment area were revised to reduce the
chance of the emission source area exceeding the dimensions of the refinery. The revised release
areas associated with the process equipment leaks are summarized in Table C.
Table C. Areas Assigned for Fugitive Equipment Leaks
Assigned Size
Assigned Lqiiipmcnl Leak
Ucfinery Crude Capacity (bbl/day)
Category
Process Area (MM ft2)
u to <125,uuu
Small
0.3
125,000 to <225,000
Medium
1.7
>225,000
Large
4
Emission Factors for Cooling Towers
For cooling towers, the emission estimates were developed for each refinery based on the
uncontrolled AP-42 emission factor of 6 lbs total hydrocarbon (THC)/million gallons (MMgal).
Cooling water flow rates were assumed to be 40 times the crude capacity. The HAP contents of
the organics in the cooling water were estimated based on the weighted average refinery stream
composition considering both liquid and gaseous streams as summarized by Lucas (2007a). The
resulting HAP emission factors normalized by crude throughput are summarized in Table D.
P-18
-------
Table D. Emission Factors for Cooling Towers
CASRN
HAP
Uncontrolled AP-42 Emission Factor (tpy/bbl/d)"
106-99-0
1,3-Butadiene
1.3E-08
540-84-1
2,2,4-Trimethylpentane
2.7E-05
71-43-2
Benzene
2.0E-05
92-52-4
Biphenyl
3.6E-07
1319-77-3
Cresols
2.7E-06
98-82-8
Cumene
4.1E-06
100-41-4
Ethylbenzene
1.0E-05
110-54-3
Hexane
9.7E-05
1634-04-4
Methyl tertiary butyl ether
1.2E-05
91-20-3
Naphthalene
3.1E-06
108-93-0
Phenol
1.9E-06
100-42-5
Styrene
6.4E-06
108-88-3
Toluene
4.4E-05
1330-20-7
Xylene
4.1E-05
" tpy/bbl/d = tons per year HAP emissions per barrel per day crude throughput.
Revised Emission Methodology and Source Areas for Storage Vessels
Emissions from storage vessels were originally developed based on emission estimates reported
in permit applications. While there are separate emission factors for crude oil, light distillates,
heavy distillates, and aromatics, only crude oil and light distillates are thought to be associated
with the MACT INESHAP. The "light distillates" originally included gasoline, naphtha, jet
fuel, and diesel fuel (i.e., No. 2 fuel oil). This category was divided into two categories: "light
distillates," which includes gasoline and jet naphtha, and "middle distillates," which includes
other jet fuels, kerosene, and diesel fuel.
Revised emission factors were developed to model external floating roof (EFR) crude oil storage
tanks, EFR light distillate (based on gasoline) storage tanks, and EFR middle distillate (based on
jet fuel) storage tanks based on TANKS v4.09 model estimates. For the REM analysis, slotted
guide poles and other openings or hatches with no fitting controls were assumed to be the level
of control at all facilities.
The results of the TANKS model runs are summarized elsewhere (Lucas, 2008). The specific
HAP composition of the volatile organic compound (VOC) emissions were estimated based on
the average liquid and vapor phase composition for crude oil, gasoline, and jet naphtha. It was
assumed that most of the VOC losses would be via gaseous losses, but that 20 percent of the
losses would be via liquid losses (e.g., liquid clinging to the sides of the wall or guide pole). The
average HAP concentrations of the VOC losses used in the analysis are presented in Table E.
These concentrations combined with the tank emission losses and throughputs yield the emission
factors presented in Table F.
P-19
-------
In addition to revising these emission factors, the area associated with the tank farm were
revised, again to limit the chance that the modeled emission source area would exceed the
boundaries of the facility. The revised tank farm release areas are presented in Table G.
Table E. HAP Concentration of VOC Storage Vessel Emissions
( rude Oil
(iasoline
.lei Naphtha
CASUN
MAP
(\M%)
(\vt%)
(w 1 %)
1U6-99-U
1,3-Buladiene
U%
U.U22" o
U%
540-84-1
2,2,4-Trimethylpentane
0.25%
1.71%
0.44%
71-43-2
Benzene
0.73%
0.82%
1.10%
92-52-4
Biphenyl
0.012%
0.002%
0%
1319-77-3
Cresols
0.044%
0.16%
0.004%
98-82-8
Cumene
0.034%
0.18%
0.21%
100-41-4
Ethylbenzene
0.12%
0.37%
0.37%
110-54-3
Hexane
6.18%
4.97%
9.72%
1634-04-4
Methyl tertiary butyl ether
0%
3.60%
0%
91-20-3
Naphthalene
0.045%
0.089%
0.081%
108-93-0
Phenol
0.067%
0.011%
0.013%
100-42-5
Styrene
0%
0.776%
0%
108-88-3
Toluene
0.56%
2.12%
2.05%
1330-20-7
Xylene
0.46%
1.62%
1.41%
P-20
-------
Table F. Storage Vessel HAP Emissions Factors
( rude Oil
Gasoline
.lei Naphtha
CASUN
MAP
(Ibs/MMbbl)
(Ibs/MMbbl)
(Ibs/MMbbl)
106-99-0
1,3-Butadiene
0.00
1.83
0.00
540-84-1
2,2,4-Trimethylpentane
3.39
144.62
22.00
71-43-2
Benzene
9.95
69.73
54.38
92-52-4
Biphenyl
0.17
0.17
0.00
1319-77-3
Cresols
0.60
13.42
0.19
98-82-8
Cumene
0.46
15.45
10.48
100-41-4
Ethylbenzene
1.62
31.48
18.23
110-54-3
Hexane
83.72
421.21
482.02
1634-04-4
Methyl tertiary butyl ether
0.00
305.44
0.00
91-20-3
Naphthalene
0.61
7.57
4.00
108-93-0
Phenol
0.91
0.94
0.67
100-42-5
Styrene
0.00
65.78
0.00
108-88-3
Toluene
7.55
179.30
101.44
1330-20-7
Xylene
6.19
137.47
70.11
Table G. Assumed Areas for Storage Vessel Tank Farms
Uellnerv Crude
Capacity
Storage Vessel l ank l-arm
(bbls/dav)
Assigned Size Category
Area (MM It2)
0 to <125,000
Small
0.5
125,000 to
Medium
4
<225,000
>225,000
Large
7
Revised Emission Methodology and Source Areas for Wastewater Treatment Systems
A simple correlation was previously used to estimate benzene emissions for wastewater systems
subject to the Benzene Waste Operations NESHAP (BWON; 40 CFR part 61, subpart FF) given
the total mass benzene loading rate to wastewater. The methodology used to estimate the
"controlled" BWON emissions were revised to better estimate the relative emissions from
wastewater collection systems and wastewater treatment systems and to evaluate different levels
of control.
The benzene loading rates to wastewater are estimated using the methodology from Locating and
Estimating Air Emissions from Sources of Benzene (US EPA, 1998) as was done previously. For
facilities that have benzene wastewater loadings (assumed to be the total annual benzene, or
P-21
-------
TAB, quantity) exceeding 10 Mg/yr, then the facility is assumed to be subject to BWON
requirements. For BWON facilities, the wastewater collection system is assumed to be 98
percent efficient, so that 2 percent of the TAB is released from the wastewater collection system.
It is assumed that approximately 50 percent of the remaining benzene is recovered in the oil
water separator and that 50 percent of the original TAB enters the enhanced biological unit
(EBU). Eighty percent control efficiency was assumed for the EBU. For wastewater systems
not subject to BWON, 85 percent of the benzene load is assumed to be emitted across the
refinery; 50 percent of these emissions were attributed to the wastewater collection area and 50
percent were assigned to the EBU. Emissions of other HAP were estimated from the calculated
benzene emissions using an adjustment factor based on the relative concentration of the HAP in
wastewater streams, its octanol-water partition coefficient, and WATER9 emission estimates as
was done previously (RTI, 2002).
As with the fugitive and storage tank farm release area parameters, the release areas for
wastewater treatment sources were reduced to reduce the likelihood that the wastewater sources
would exceed the boundaries of the facility. The revised release areas for wastewater treatment
sources are provided in Table H.
Table H. Assumed Areas for Wastewater Collection and Treatment
Uellnerv Crude
Capacity
(hhls/dav)
Assigned Size Category
Wastewater
Collection Area
(MM It2)
Wastewater
Treatment Area
(MM It2)
0 to <125,000
Small
0.10
0.10
125,000 to
<225,000
Medium
0.43
0.43
>225,000
Large
1.7
1.7
P-22
-------
Additional Data
Table I. Comparison of REM and RTR Modeled Maximum Individual Risks (MIR), By
Facility
l;icilil\ II)
MIK Kiilio
(KI.MiRIK)
KI M ( iincer MIK
(in 1 million)
KTK ( iincer MIK
(in 1 million)
1
PETNEI34057
0.1
2
14
2
PETNEI6095
0.1
1
13
3
PET_NEIPRT$64
0.2
2
12
4
PETNEI12711
0.3
9
28
5
PETNEI34898
0.3
7
21
6
PETNEI34050
0.4
5
12
7
PETNEI40371
0.4
4
10
8
PETNEI6475
0.4
5
13
9
PETNEI12480
0.5
3
6
10
PETNEI 12791
0.5
4
10
11
PETNEI12988
0.5
10
19
12
PETNEI20174
0.5
5
12
13
PETNEI40531
0.5
4
8
14
PETNEI46752
0.6
1
2
15
PET_NEIOKT$ 11009
0.6
6
9
16
PETNEI33031
0.7
10
15
17
PETNEI876
0.7
10
15
18
PETNEI 11449
1
5
9
19
PETNEI 12044
1
6
6
20
PETNEI12458
1
1
1
21
PETNEI12968
1
3
2
22
PETNEI18394
1
8
11
23
PETNEI19587
1
4
5
24
PETNEI32864
1
10
11
25
PETNEI33008
1
10
9
26
PETNEI33039
1
10
10
27
PETNEI34907
1
0.5
0
28
PETNEI41771
1
9
13
29
PETNEI42040
1
10
15
30
PETNEI42309
1
20
15
31
PETNEI42413
1
2
2
32
PETNEI6087
1
8
12
33
PETNEI6116
1
10
7
34
PETNEI6136
1
5
6
35
PETNEI6166
1
5
7
36
PETNEI6436
1
10
12
37
PETNEI6446
1
2
2
38
PETNEI6963
1
0.3
0
39
PETNEI 11200
2
2
1
40
PETNEI 11232
2
9
5
41
PETNEI 11663
2
8
4
P-23
-------
l-'iicilih II)
MIK Kiilio
(RIMiRIK)
KI M ( iincoi' MIK
(in 1 million)
KTK ( .nicer MIK
(in 1 million)
42
PETNEI 12460
2
1
1
43
PETNEI12486
2
9
4
44
PETNEI12969
2
2
1
45
PETNEI18372
2
0.8
0
46
PET NEI2CA314628
2
5
3
47
PET NEI2KS125003
2
5
3
48
PET_NEI32762
2
7
5
49
PETNEI32801
2
4
2
50
PETNEI33010
2
0.3
0
51
PETNEI34062
2
10
5
52
PETNEI34862
2
4
2
53
PETNEI34873
2
10
5
54
PETNEI42020
2
8
5
55
PETNEI42025
2
6
3
56
PETNEI42381
2
9
5
57
PETNEI42425
2
2
1
58
PETNEI43243
2
4
2
59
PET_NEI53702
2
5
2
60
PETNEI6022
2
20
10
61
PETNEI6062
2
10
5
62
PETNEI6123
2
10
5
63
PETNEI6519
2
10
5
64
PET_NEI7233
2
10
5
65
PET NEICA1910268
2
20
11
66
PETNEI 12464
3
4
1
67
PETNEI 19834
3
6
2
68
PETNEI26533
3
0.2
0
69
PETNEI41591
3
6
2
70
PETNEI6130
3
20
6
71
PET_NEI7781
3
20
6
72
PETNEI 11574
4
20
5
73
PETNEI13322
4
0.5
0
74
PETNEI13371
4
20
5
75
PETNEI363
4
3
1
76
PET_NEI40723
4
10
2
77
PETNEI415
4
6
2
78
PETNEI42016
4
9
2
79
PETNEI6127
4
10
2
80
PETNEI8139
4
6
1
81
PETNEI12084
5
2
0
82
PETNEI 19870
5
3
1
83
PETNEI26218
5
2
0
84
PETNEI32997
5
2
0
85
PETNEI42081
5
0.7
0
86
PETNEI46556
5
30
6
87
PET NEIWYT$ 12156
5
0.007
0.002
88
PETNEI 18406
6
20
9
P-24
-------
MIK Kiilio
KI M ( iincoi' MIK
K I K ( iinccr MIK
l-'iicilih II)
(RIMiRIK)
(in 1 million)
(in 1 million)
89
PETNEI26101
6
3
0
90
PETNEI34022
6
3
1
91
PET_NEI40732
6
30
5
92
PETNEI6375
6
8
1
93
PETNEI7441
6
5
1
94
PETNEI8612
6
4
1
95
PETNEI109
7
30
4
96
PETNEI12459
7
7
1
97
PETNEI41863
7
10
2
98
PETNEI53718
7
1
0
99
PET_NEINJT$891
7
8
1
100
PETNEI40625
8
3
0
101
PET_NEI42370
8
0.4
0
102
PETNEI6084
8
10
1
103
PETNEI889
8
8
1
104
PETNEI11119(B)
9
5
1
105
PETNEI41864
9
3
0
106
PET NEICA0379991
9
2
0
107
PETNEI11450
10
20
2
108
PETNEI21034
10
9
1
109
PETNEI34061
10
4
0
110
PETNEI49781
10
10
1
111
PET NEICA0370363
10
20
2
112
PETNEI41865
11
0.2
0
113
PETNEI113
20
9
1
114
PETNEI20103
20
10
1
115
PETNEI20467
20
20
1
116
PET NEI2CA254640
20
3
0
117
PETNEI32353
20
5
0
118
PETNEI34912
20
3
0
119
PETNEI42382
20
5
0
120
PETNEI6018
20
5
0
121
PETNEI7130
20
2
0
122
PETNEI20616
22
8
0
123
PETNEI11192
30
20
1
124
PETNEI11885
30
5
0
125
PETNEI19869
30
5
0
126
PETNEI20966
30
2
0
127
PET NEI2CA131003
30
10
0
128
PETNEI34069
30
0.6
0
129
PETNEI46764
30
4
0
130
PETNEI34863
40
1
0
131
PETNEI34872
40
30
1
132
PETNEI371
40
2
0
133
PETNEI42583
40
4
0
134
PETNEI18415
50
1
0
135
PETNEI6617
60
4
0
P-25
-------
l-'iicilih II)
MIR K;ilid
(RIM:RTR)
KI M ( iincoi MIK
(in 1 million)
KTK ( .nicer MIK
(in 1 million)
136
PET_NEI33007
70
0.9
0
137
PETNEI404
100
4
0.04
138
PETNEI25464
300
1
0
139
PETNEI55835
400
7
0
140
PETNEI 18408
500
7
0.01
141
PETNEI20154
3000
20
0.007
142
PETNEI21130
4000
10
0
143
PETNEI21466
5000
7
0
144
PETNEI26473
5000000
0.5
0.00000009
145
PETNEI 11715
10
146
PETNEI 18673
5
147
PET NEI2AK530001
0.02
148
PET NEI2AK560004
0.08
149
PET NEI2CA312611
2
150
PET NEI2NV110905
0.7
151
PETNEI33009
4
152
6
153
5
154
4
155
3
156
1
157
0.005
P-26
-------
Table J. Comparison o
fREM and R1
"R Modeled Annual Cancer Incidence, By Facility
r
-------
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
rear)
K I K Ciinccr Incidence
(excess c;inccr c;ises per >e;ir)
PET NEI6022
0.003
0.002
PET NEI12480
0.001
0.0006
PET NEI46752
0.00007
0.00003
PET NEI6130
0.0006
0.0003
PET NEI42025
0.0002
0.00008
PET NEI34062
0.00001
0.000005
PET NEI42425
0.00005
0.00002
PET NEI7233
0.002
0.0008
PET NEI26533
0.00002
0.00001
PET NEI41771
0.001
0.0005
PET NEI34873
0.002
0.0009
PET NEI12968
0.0002
0.00009
PET NEI18372
0.0003
0.0001
PET NEI13371
0.00005
0.00002
PET NEI6062
0.0006
0.0002
PET NEI11574
0.0004
0.0002
PET NEI6519
0.0001
0.00004
PET NEI2CA314628
0.0008
0.0003
PET NEI53702
0.001
0.0005
PET NEI7781
0.003
0.001
PET NEICA1910268
0.005
0.002
PET NEI41863
0.0005
0.0002
PET NEI46556
0.001
0.0004
PET NEI6116
0.0004
0.0001
PET NEI13322
0.00003
0.000008
PET NEI109
0.003
0.0008
PET NEI19870
0.0004
0.0001
PET NEI42081
0.0002
0.00005
PET NEI363
0.0001
0.00004
PET NEI415
0.001
0.0002
PET NEIWYT$ 12156
0.0000005
0.0000001
PET NEI41591
0.001
0.0003
PET NEI 12464
0.0001
0.00002
PET NEI42040
0.0006
0.0001
PET NEI42016
0.0004
0.00008
PET NEI8139
0.000008
0.000002
PET NEI6375
0.009
0.002
PET NEINJT$891
0.002
0.0005
PET NEI26218
0.0006
0.0001
PET NEI12084
0.0004
0.00007
PET NEI34862
0.003
0.0006
PET NEI19834
0.001
0.0002
PET NEI26101
0.0002
0.00003
PET NEI6087
0.0003
0.00005
PET NEI11119B
0.003
0.0004
PET NEI34022
0.002
0.0002
P-28
-------
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
rear)
RTR Ciinccr Incidence
(excess c;inccr c;iscs per \e;ir)
PET NEI42413
0.00009
0.00001
PET NEI40625
0.0006
0.00008
PET NEI53718
0.0005
0.00006
PET NEI889
0.0008
0.0001
PET NEI42370
0.0003
0.00003
PET NEI12459
0.0001
0.00001
PET NEI40732
0.0002
0.00002
PET NEI18406
0.0006
0.00007
PET NEI41864
0.0006
0.00006
PET NEI7441
10
0.0008
0.00009
PET NEI21034
10
0.008
0.0008
PET NEICA0370363
10
0.005
0.0004
PET NEI11192
10
0.001
0.00009
PET NEI49781
10
0.0003
0.00002
PET NEICA0379991
10
0.002
0.0002
PET NEI11450
10
0.002
0.0001
PET NEI41865
10
0.00005
0.000004
PET NEI7988
10
0.045
0.0005
PET NEI40723
10
0.01
0.0008
PET NEI6084
10
0.0003
0.00002
PET NEI20103
20
0.0002
0.00001
PET NEK 127
20
0.001
0.00008
PET NEI42382
20
0.0001
0.000006
PET NEB 3007
20
0.00003
0.000002
PET NEI34912
20
0.00006
0.000003
PET NEI20616
20
0.001
0.00007
PET NEI32997
20
0.00005
0.000002
PET NEB4061
20
0.00003
0.000001
PET NEI32353
20
0.0001
0.000006
PET NEI20966
20
0.001
0.00004
PET NEI113
20
0.002
0.00009
PET NEI7130
30
0.0004
0.00001
PET NEI42583
30
0.00009
0.000003
PET NEI46764
30
0.00007
0.000003
PET NEI6018
30
0.0004
0.00001
PET NEI2CA254640
30
0.00004
0.000001
PET NEI 11885
30
0.002
0.00004
PET NEI2CA131003
40
0.01
0.0003
PET NEI34069
40
0.00005
0.000001
PET NEI 19869
40
0.00006
0.000002
PET NEI34872
40
0.003
0.00009
PET NEI6617
40
0.0004
0.00001
PET NEB 71
40
0.00002
0.000001
PET NEI20467
50
0.01
0.0003
PET NEI34863
50
0.001
0.00002
PET NEI18415
60
0.00001
0.0000002
P-29
-------
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
rear)
RTR Ciinccr Incidence
(excess c;inccr c;ises per \e;ir)
PET NEI404
90
0.00003
0.0000004
PET NEI25464
300
0.0001
0.000001
PET NEI55835
500
0.001
0.000002
PET NEI18408
2000
0.00010
0.00000005
PET NEI20154
3000
0.0009
0.0000003
PET NEI21130
4000
0.004
0.000001
PET NEI21466
4000
0.004
0.000001
PET NEI26473
4000000
0.0001
0.00000000003
PET NEI11715
0.003
PET NEI18673
0.00002
PET NEB 3009
0.00005
PET NEI2CA312611
0.0006
PET NEI2NV110905
0.0000001
PET NEI2AK560004
0.0000004
PET NEI2AK530001
0.000000002
PET NEI33030
0.0002
PET NEI 12790
0.0002
PET NEI7134
0.0001
PET NEI25450
0.0002
PET NEI26489
0.00002
PET NEICA10578
0.00001
PET NEINMT$ 12478
0.00001
PET NEI2TX14199
0.000003
PET NEI20797
0.00001
PET NEI7973
0.0000002
P-30
-------
Table K. REM Input Assumptions for Each Facility
This table is in-progress and will be available upon request.
Table L. REM Emissions Estimates for Each Facility
This table is in-progress and will be available upon request.
Table M. RTR Emissions Estimates for Each Facility
This table is in-progress and will be available upon request.
P-31
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/R-09-006
Environmental Protection Health and Environmental Impacts Division June, 2009
Agency Research Triangle Park, NC
------- |