United States
Environmental Protection
Agency
EPA-452/R-98-001
March 1998
Air
THE TOTAL RISK INTEGRATED METHODOLOGY
IMPLEMENTATION OF THE TRIM CONCEPTUAL DESIGN
THROUGH THE TRIM.FaTE MODULE
A STATUS REPORT
Environmental Fate
6 Transport Module
(TRIM.FaTE)
Social,
Economics
& Political
Factors
Office of Air Quality Planning & Standards
-------
Disclaimer
This document was developed by the U.S. Environmental Protection Agency, in conjunction with
the Lawrence Berkeley National Laboratory (through Interagency Agreement DW89937866601),
f| Oak Ridge National Laboratory (through Interagency Agreement DW8993786601), IT
^ Corporation (through Contract No. 68-D-30094, Work Assignment No. 4-18), and TRJ
t&\
^j Environmental, Inc. (as a subcontractor to IT Corporation, under Contract No. 68-D-30094,
o Work Assignment No. 4-18). Any opinions, findings, conclusions, or recommendations are
those of the authors and do not necessarily reflect the views of reviewers, individuals named in
the acknowledgments, the U.S. Environmental Protection Agency, the U.S. Department of
Energy, Lawrence Berkeley National Laboratory, Oak Ridge National Laboratory, IT
Corporation, or TRJ Environmental, Inc. Comments on this document should be addressed to
Amy Vasu, U.S. Environmental Protection Agency, Office of Air Quality Planning and
Standards, MD-15, Research Triangle Park, North Carolina 27711.
U S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12in rwoi
Chicago. II 60604-3590
-------
Table of Contents,
Page
List of Tables iv
List of Figures v
List of Acronyms vi
Acknowledgment viii
1.0 Introduction 1-1
1.1 Clean Air Act Requirements 1-1
1.2 Recommendations for Improving Risk Assessment 1-2
1.2.1 National Research Council Report 1-2
1.2.2 CRARM Report 1-3
1.2.3 Other Risk Guidance 1-3
1.3 The Need for an Improved Risk Assessment Tool: TRIM 1-5
2.0 TRIM Conceptual Approach 2-1
2.1 Overall Goals and Objectives 2-1
2.2 Model Design 2-2
2.2.1 Environmental Fate and Transport (TRIM.FaTE) Module 2-3
2.2.2 Exposure Event Module 2-3
2.2.3 Pollutant Uptake Module 2-5
2.2.4 Biokinetics Module 2-5
2.2.5 Dose-Response Module 2-5
2.2.6 Risk Characterization Module 2-6
3.0 Conceptual Framework for TRIM.FaTE 3-1
3.1 Review of Existing Tools 3-1
3.2 The Need for an Improved Fate and Transport Modeling Tool: TRIM.FaTE ... 3-5
3.3 Uniqueness of TRIM.Fate 3-6
3.4 TRIM.FaTE Terminology and Basic Concepts 3-7
3.5 Governing Mass Balance Equations 3-11
3.6 Modeling Approach 3-16
3.6.1 Problem Definition 3-18
3.6.2 Link Setup 3-18
3.6.3 Simulation Setup 3-19
3.6.4 Simulation Implementation 3-19
-------
Table of Contents (Continued).
Page
3.6.5 Result Analysis 3-19
3.7 Sensitivity Analysis 3-19
3.8 Example Calculation of Transfer Factors 3-20
3.9 Summary of TRIM.FaTE Approach 3-23
4.0 TRIM.FaTE Prototype Development 4-1
4.1 Implementation of Prototypes 4-1
4.2 Prototype Development 4-2
4.2.1 Prototype 1 4-2
4.2.2 Prototype 2 4-4
4.2.3 Prototype 3 4-4
4.2.4 Prototype 4 4-7
4.3 Prototype Features 4-11
4.3.1 Abiotic Domains 4-11
4.3.2 Biotic Domains 4-13
4.3.3 Links 4-15
4.4 Fate and Transport Processes 4-17
5.0 Test Case 5-1
5.1 Data Inputs 5-1
5.2 Description of Model Runs 5-1
5.3 Results of Phase Calculations 5-2
5.4 Results of Constant Meteorology Runs 5-3
5.4.1 Results for No Precipitation, East Wind Direction Scenario 5-3
5.4.2 Results for Precipitation Scenario, East Wind Direction Scenario 5-5
5.4.3 Comparative Analysis 5-8
5.4.4 Results for All Runs - Constant Meteorology 5-8
5.4.5 Mass and Concentration Distribution in Biota 5-10
5.5 Variable Meteorology 5-16
6.0 Evaluation of TRIM.FaTE 6-1
6.1 Comparison with Other Models (SimpleBOX and CalTOX) 6-1
6.2 Sensitivity Analysis for TRIM.FaTE 6-2
6.3 Overall Capabilities 6-6
6.3.1 Time Scales 6-7
ii
-------
Table of Contents (Continued).
Page
6.3.2 Spatial Scales '. 6-7
6.3.3 Chemical Classes 6-7
6.4 Limitations 6-7
6.5 Conclusions from Developmental Work on Trim and TRIM.FaTE 6-9
6.5.1 Prototype Algorithms and Mathematical Structure 6-10
6.5.2 Input Data Needs, Verification, and Validation 6-10
7.0 Summary, Discussion, and Future Directions 7-1
7.1 Progress to Date 7-1
7.1.1 Conceptual Design of TRIM: Design Goals and Objectives 7-1
7.1.2 TRIM.FaTE Module Development 7-2
7.2 Limitations and Sensitivity 7-3
7.3 Future Developments 7-5
7.3.1 Overall TRIM Development 7-6
7.3.2 TRIM.FaTE Development 7-6
7.3.3 Exposure Event Module 7-8
7.3.4 Development of Uncertainty and Sensitivity Analysis Capabilities 7-8
8.0 References 8-1
Appendix A - Glossary
Appendix B - Algorithm Generalizations
Appendix C - Bibliography
111
-------
List of Tables-
Number Title Page
4-1 Databases Consulted in Developing Prototype 4 Ecosystem 4-12
4-2 Types of Abiotic Domains and Number of Volume Elements Modeled 4-13
4-3 Biotic Domains Modeled 4-14
4-4 Examples of Links Associated with Domains 4-16
5-1 Predicted Phase Distribution for B(a)P and Phenanthrene in Abiotic Media 5-2
5-2 Predicted Steady-State Results (No Precipitation, East Wind Direction Scenario) 5-5
5-3 Predicted Steady-State Results (Precipitation, East Wind Direction Scenario) 5-6
5-4 Predicted Distribution by Domain Type (No Precipitation Scenario) 5-9
5-5 Predicted Distribution by Domain Type (Precipitation Scenario) 5-9
5-6 Predicted Distribution in Biota by Domain Type (No Precipitation Scenario) 5-11
5-7 Predicted Distribution Biota by Domain Type (Precipitation Scenario) 5-12
5-8 Uptake Fractions for Specialized Fish Domains in Parcel Q (Precipitation,
East Wind Direction Scenario) 5-15
6-1 Parameters with High Sensitivity Scores for B(a)P 6-5
IV
-------
List of Figures.
Figure Title Page
2-1 Conceptual Overview of TRIM 2-4
3-1 Modeling Blueprint for TRIM.FaTE 3-8
3-2 Example of First Order Transfer Process for Two Cells 3-12
3-3 Simplified Ecosystem 3-15
3-4 Structure of TRIM.FaTE 3-17
4-1 Conceptual Site Model for Prototype 1 4-3
4-2 Conceptual Site Model for Prototype 2 4-5
4-3 Conceptual Site Model for Prototype 3 4-6
4-4 Plan View for Study Area in Prototype 4 4-9
4-5 Cross Sectional View for Land, River, and Lake Parcels in Prototype 4 4-10
5-1 Predicted Steady State Spatial Distribution of BAP and Phenanthrene,
Wind Due East, No Precipitation 5-4
5-2 Predicted Steady State Spatial Distribution of B(a)P and Phenanthrene,
Wind Due East, Precipitation 5-7
5-3 Steady State Concentrations of B(a)P in Biota and Soil in a Forested
Parcel, No Precipitation 5-13
5-4 Steady State Concentrations of Phenanthrene in Biota and Soil in a
Forested Parcel, No Precipitation 5-13
5-5 Wind Direction (degrees) Profile for TPJM.Fate Prototype, Clockwise from Due
North Towards Direction of Wind 5-17
5-6 24-Hour Precipitation (mm/hr) Profile for TRIM.Fate Prototype 5-17
5-7 B(a)P Mass Distribution for Parcels in TRIM.Fate Prototype, Variable
Meteorological Conditions 5-18
5-8 Phenanthrene Mass Distribution for Select Domains in TRIM.Fate Prototype,
Variable Meteorological Conditions 5-18
5-9 Phenanthrene Mass Distribution for Parcels in TRIM.Fate Prototype, Variable
Meteorological Conditions 5-19
5-10 B(a)P Mass Distribution in Biotic Domains for TRIM.Fate Prototype, Variable
Meteorological Conditions 5-19
6-1 Model Comparison for B(a)P 6-3
6-2 Model Comparison for Phenanthrene 6-3
-------
List of Acronyms.
B(a)P
CAA
CalTOX
CRARM
DOE
DISC
EPA
FY
g/day
GI
GIS
HAP
IEM2
ISMCM
kg/nr/day
km
LSODE
m/s
MCM
MEI
MEPAS
NAAQS
NAS
NRC
OAQPS
PI
P2
P3
P4
PAH
benzo(a)pyrene
Clean Air Act of 1990
California Department of Toxic Substance Control's Multimedia
Risk Computerized Model
Presidential/Congressional Commission on Risk Assessment and Risk
Management
U.S. Department of Energy
Department of Toxic Substance Control
U.S. Environmental Protection Agency
fiscal year
grams per day
gastrointestinal
Geographic Information System
hazardous air pollutant
Indirect Exposure Methodology
Integrated Spatial Multimedia Compartmental Model
kilograms per square meter per day
kilometer
Livermore Solver for Ordinary Differential Equation
meters per second
Multimedia Compartment Model
maximally exposed individual
Multimedia Environmental Pollutant Assessment System
micrograms per kilogram
micrograms per liter
National Ambient Air Quality Standards
National Academy of Science
National Research Council
Office of Air Quality Planning and Standards
Prototype 1
Prototype 2
Prototype 3
Prototype 4
polycyclic aromatic hydrocarbon
VI
-------
List Of Acronyms (Continued).
PBPK physiologically-based pharmacokinetic
PC personal computer
SAB Science Advisory Board
SMCM Spatial Multimedia Compartment Model
TRIM Total Risk Integrated Methodology
TRIM.FaTE TRIM Environmental Fate, Transport, and Exposure Model
vn
-------
Acknowledgments
The U.S. Environmental Protection Agency (EPA) would like to acknowledge the significant
contributions of individuals from the following organizations to the development of the Total
Risk Integrated Methodology (TRIM) and the TRIM.FaTE Module: Lawrence Berkeley National
Laboratory, Oak Ridge National Laboratory, International Technology Corporation, and TRJ
Environmental, Inc.
Vlll
-------
1.0 Introduction
The Office of Air Quality Planning and Standards (OAQPS) of the U.S. Environmental
Protection Agency (EPA) is responsible for the evaluation of health risks associated with air
pollutants and for the regulation of those pollutants, if needed. To date, OAQPS has not
consistently estimated multimedia impacts of air pollutants and has used distinctly different
methodologies to estimate risks from hazardous air pollutants (HAP) and criteria air pollutants.
While numerous models exist for use in risk assessment, there is no one model or modeling
system which meets the needs of OAQPS. As a result, OAQPS is developing the Total Risk
Integrated Methodology (TRIM), a multimedia, time-series simulation modeling system for the
assessment of human and ecological risks resulting from hazardous and criteria air pollutants.
TRIM represents an improved risk assessment tool which:
Meets the requirements of the Clean Air Act of 1990 (CAA).
Meets the scientific requirements/capabilities identified by the National Academy
of Sciences (NAS), the Presidential/Congressional Commission on Risk
Assessment and Risk Management (CRARM), and the EPA.
TRIM will provide a framework that is scientifically defensible, flexible, and user-friendly, for
characterizing human health and ecological risk and exposure to hazardous and criteria air
pollutants.
The purpose of this report is to summarize the work performed during the first developmental
phase of TRIM. The first phase included the conceptualization of TRIM and the implementation
of the TRIM conceptual approach through the development of the first TRIM module, the
environmental fate, transport, and exposure module, called TRM.FaTE. TRIM.FaTE
development focused on defining the mathematical structure of the module and initial testing and
evaluation of these concepts. This report provides detailed information about the overall
structure of TRIM and the development of the TRIM.FaTE module. The detailed technical
information (mathematical derivations, data inputs, justifications) supporting the testing and
implementation is provided in a separate document entitled The Total Risk Integrated
Methodology: Technical Support Document for TRIM.FaTE1.
1.1 Clean Air Act Requirements
The CAA contains several provisions that require EPA to evaluate effects to humans and the
environment caused by exposure to HAPs and criteria pollutants. In support of the CAA require-
1-1
-------
ments, and in response to recommendations of the NAS and the CRARM, and EPA risk
assessment policies and guidelines, EPA is developing TRIM. TRIM will provide a framework
for assessing human health and ecological risks attributable to HAP and criteria pollutants and,
thus, will include the capability to assess multimedia contamination (in air, water, soil, food, and
indoor environments) and multipathway exposure (via inhalation, ingestion, and absorption
exposure routes).
TRIM will be a useful tool for performing risk assessments needed by many CAA programs,
including: the Residual Risk Program (Section 112[fj); the Urban Area Source Program (Section
112[k]); the Special Studies (Sections 112[m] and 112[n]); petitions to delist source categories
and/or individual HAPs (Sections 112[b][3] and 112[c][9]); and review and setting of the
national ambient air quality standards (NAAQS) (Section 109).
1.2 Recommendations for Improving Risk Assessment
The risk assessment tools used by OAQPS must have maximum technical credibility and,
therefore, must address the recommendations of the NAS and the CRARM and be consistent
with EPA guidance and guidelines for risk and exposure assessment. Some of these
recommendations and guidelines are described in the following sections.
1.2.1 National Research Council Report
As required by Section 112(o) of the CAA, the EPA commissioned the National Research
Council (NRC) to perform a study of the risk assessment methods used by EPA for the
evaluation of HAP. The NRC created the Committee on Risk Assessment of Hazardous Air
Pollutants, within the Board on Environmental Studies and Toxicology, to: (1) review the risk
assessment methods used by EPA; (2) evaluate methods used for estimating the carcinogenic
potency of HAPs; (3) evaluate methods used for estimating human exposures; (4) evaluate risk
assessment methods for noncancer health effects for which safe thresholds may not exist; and (5)
indicate revisions needed in EPA's risk assessment guidelines. The resulting 1994 NRC report,
Science and Judgment in Risk Assessment,2 outlined the Committee's observations and
recommendations.
The NRC Committee observed that several themes that were common to all elements of the risk
assessment process also were usually the focal points for criticisms of specific risk assessments.
The themes discussed included the use of default assumptions; the lack of available data; the
need for a tiered, iterative approach to risk assessment; the need for quantification of uncertainty
and variability inherent in the risk assessment process; the assessment of multiple chemical
1-2
-------
exposures, multiple routes of exposure, and the potential for multiple adverse effects; and the
documentation of the steps taken to validate the methodologies used throughout the risk
assessment process.
The NRC concluded that EPA should retain its conservative, default-based approach to risk
assessment for screening analysis in standard setting. However, the Committee made specific
recommendations regarding ways that the process should be improved, including using defaults.
and explicitly identifying and better explaining all defaults; developing and using an iterative
approach to health risk assessments; and identifying the sources and magnitudes of uncertainty
associated with estimates of risk.
1.2.2 CRARM Report
The CRARM was also mandated by Congress, under the CAA, to: (1) assess uses and
limitations of risk assessment; (2) evaluate exposure scenarios for risk characterization; (3)
determine how to describe and explain uncertainties; (4) enhance strategies for risk-based
management decisions; and (5) review the desirability of consistency across federal programs.
The CRARM was also asked to make recommendations on the EPA peer review process and to
comment on the conclusions of the NRC's Science and Judgment in Risk Assessment. The
CRARM report3 identified several risk management contexts that may be relevant for a risk
assessment, including the consideration of multimedia, multisource, and/or multichemical
exposures, as well as multiple risks from different stressors.
The CRARM report recommended that risk assessments should take into consideration genetic
and other differences in receptor susceptibility, recognize the spectrum of interindividual
variations within the population, and identify subpopulations that are especially susceptible to
specific chemical exposures. In addition, CRARM identified the need for exposure assessments
to be designed to be commensurate with the needs of the risk management decisions. CRARM
also identified the need for more realistic exposure scenarios. The report recommended that
screening risk assessments should rely on more representative estimates, such as a maximally
exposed actual person, rather than on a hypothetical maximally exposed individual (MEI). Other
recommendations included identifying highly exposed populations or subpopulations, and
performing ecological risk assessments.
1.2.3 Other Risk Guidance
EPA has prepared numerous guidance on risk assessment and risk assessment methods, starting
with the 1986 risk assessment guidelines, which included Guidelines for Carcinogenic Risk
1-3
-------
Assessment4, Guidelines for Mutagenicity Risk Assessment5, Guidelines for the Health Risk
Assessment of Chemical Mixtures6, and Guidelines for Exposure Assessment1. These guidelines
have been subsequently augmented or revised. Most notably, the EPA has revised guidelines for
exposure assessment and has prepared new guidance on risk characterization and probabilistic
analysis.
EPA's 1995 Policy for Risk Characterization9 stresses the importance of risk characterization
and calls for all risk characterizations prepared at EPA, or submitted to EPA by outside parties,
to be done in a manner that is transparent, clear, reasonable, and consistent with other risk
characterizations of similar scope prepared across programs in the Agency.
Both the revised Guidelines for Exposure Assessment1 and Policy for Use of Probabilistic Risk
Analysis in Risk Assessment9 emphasize a distributional or probabilistic approach toward risk
assessment, moving away from the deterministic "point estimate" approach. Guidelines for
Exposure Assessment1 provides several descriptors of exposure and risk aimed at presenting a
fuller picture of risk that corresponds to the range of different exposure conditions encountered
by various individuals and populations exposed to environmental chemicals. The guidelines
recommend that central tendency and high-end exposures be characterized, as well as the
distribution of exposures and risks among the exposed population. EPA's Policy for Use of
Probabilistic Risk Analysis in Risk Assessment9 emphasizes the use of probabilistic analysis to
generate distributional estimates of risks to provide more information to the risk manager. In
response to the need for probabilistic analysis in risk assessment (also called for by NAS and
CRARM), EPA has prepared Guiding Principles for Monte Carlo Analysis10, which establishes
principles for how such analyses should be included in risk assessments.
The 1997 EPA document Guidance on Cumulative Risk Assessment11 discusses the Agency's
move to an emphasis on a more broadly based approach to risk assessment, characterized by
greater consideration of multiple endpoints, sources, pathways, and routes of exposure, as well as
flexibility in achieving goals and holistic reduction of risk, among other features. Cumulative
risk assessments are integrated assessments potentially involving multiple pollutants in several
media that may cause a variety of adverse effects to humans and other biota, or even to
ecosystems and their processes and functions. In planning a risk assessment, the guidance
recommends defining the dimensions of the assessment, including the characteristics of the
population at risk. These characteristics include individuals or sensitive subgroups that may be
highly susceptible to risks from stressors or groups of stressors.
1-4
-------
1.3 The Need for an Improved Risk Assessment Tool: TRIM
Based on the recommendations of the NAS and the CRARM, as well as the current EPA
guidelines and policies, in combination with the CAA requirements, OAQPS recognized the
need for improvements in risk and exposure assessment tools. OAQPS currently has a variety of
tools for HAP and criteria air pollutant exposure and risk assessments, though several significant
features were determined to be lacking in the current models. To be consistent with the
recommendations of the NAS and the CRARM, as well as EPA guidelines and policies, OAQPS
needs modeling tools that: (1) have multimedia assessment capabilities; (2) have ecosystem risk
and exposure modeling capabilities; (3) can perform multi-pollutant assessments (e.g., assess
mixtures of pollutants, track chemical transformations); (4) can explicitly address uncertainty and
variability; and (5) are readily available and user-friendly, so that they can be used by EPA, state
and local agencies, and other stakeholders. OAQPS also needs HAP exposure and risk models
that adequately estimate temporal and spatial patterns of exposures and that maintain mass-
balance. While many current OAQPS criteria air pollutant exposure and risk models have these
advanced features, the HAP models do not. Finally, OAQPS and others recognize the
importance of having modeling tools with the capability to model pollutant uptake, biokinetics,
and dose-response for HAPs and criteria air pollutants where possible and relevant.
A risk and exposure assessment model, or set of models, with all of the previously noted features
does not exist. Although individual models that perform individual functions do exist, none of
these, separately or in combination with other models, provide an integrated system that could
function to meet the modeling needs previously described. Therefore, to meet the specific
modeling needs of OAQPS, the conceptual framework for TRIM was developed. The TRIM
conceptual approach and the modular design of TRIM are described in Chapter 2.0. The fate and
transport module has been the focus of the first implementation phase of TRIM'S conceptual
approach. This multimedia model, TRIM.FaTE, is described in Chapters 3.0 through 6.0 of this
report. Future directions for the overall TRIM framework are presented in Chapter 7.0.
1-5
-------
2.0 TRIM Conceptual Approach
2.1 Overall Goals and Objectives
The goal in developing TRIM is to create a modeling system, and the components of that system,
to appropriately characterize human health and ecological risk and exposure in support of
hazardous and criteria air pollutant programs under the CAA. The goal in designing TRIM is to
develop a modeling system that is: (1) scientifically defensible, (2) flexible, and (3) user-
friendly.
(1) To be scientifically defensible, TRIM will be:
Mass-Conserving. Pollutant mass is conserved, within the system being
assessed.
Stochastic. Input data may be represented as ranges or probability distributions
rather than point estimates.
Able to characterize uncertainty and variability. Uncertainty and
variability in outputs are characterized using stochastic simulation and
distributional data.
Capable of assessing multiple pollutants, multiple media, and multiple
exposure pathways. Cumulative effects, due to multiple sources and/or
multiple pollutants affecting the same target organ or organism, may be estimated;
chemical and/or chemical species transformations are tracked.
Able to perform iterative analyses. The user may select the necessary level
of analysis, ranging from a screening level to a detailed risk assessment.
(2) To ensure flexibility, TRIM will be:
Modular in design. Only those model components necessary for evaluating the
particular pollutants of interest and/or endpoints of interest need to be selected and
used.
Flexible in temporal and spatial scale. Risk assessments are possible for a
wide range of temporal and spatial scales, including hourly to daily or yearly time
frames, temporally, and, spatially, from local scale (10 kilometers [km] or less) to
urban scale (approximately 100 km or less), or even greater.
Able to assess human and ecological endpoints. Impacts to humans
and/or biota, for individuals and/or populations, may be assessed.
2-1
-------
*
(3) TRIM must also be usable by O. .QPS, EPA Regions, states, and stakeholders. Therefore, the
goal is to develop a model that will be:
Easily accessible. TRIM will be accessible through a personal computer (PC)
and/or via the INTERNET and/or through an EPA model framework (e.g., Models-
vf 3).
Well-documented. Detailed instructions for use of the model will be provided
through a user's guide, with a focus on the modular aspects of the modeling system
and how to specify user options.
Clear and transparent. The graphical user interface will provide transparency
and clarity in the model function, and the risk characterization module will provide
information on model assumptions, limitations, and uncertainties.
2.2 Model Design
TRIM will provide a framework for assessing human health and ecological risks resulting from
multimedia contamination (in air, water, soil, and food) and multipathway exposure (via inhala-
tion, ingestion, and absorption exposure routes) to HAPs and criteria pollutants. TRIM will be a
dynamic modeling system that tracks the movement of pollutant mass through a comprehensive
system of compartments, providing an inventory of a pollutant throughout the entire system. The
compartments will be able to represent possible locations of the pollutant in the physical and
biological environments of a defined study area or species. Receptors may move through these
compartments for the estimation of exposure. Uptake, biokinetics, and dose-response models
may be used to determine dose and health impacts. The model will address uncertainty and
variability issues by evaluating a range of parameters.
The goal in developing TRIM is to create a modeling system that is complex enough to appro-
priately characterize human health and ecological risk and exposure, yet simple enough to be
useful in performing risk analyses for use in regulatory decision making. An extremely simple
modeling approach may be too restrictive to support risk and exposure assessments across the
CAA programs. An extremely complex model may be too difficult to initialize or may require
prohibitive amounts of data. The aim of developing TRIM is to suppress the less necessary
details and to focus on the processes that have the most significant impact on human health and
ecological risk.
For the development of TRIM, existing models and tools will be adopted, where possible. Incor-
porating existing models or model features into a modeling tool that meets OAQPS needs is
preferrable since it is the most efficient and cost-effective approach.
9-9
-------
As shown in Figure 2-1, TRIM is designed to be modular and will be an assembly of six primary
modules. Depending on the user's needs for a particular assessment, it may be possible to use
any one or more of these modules for an assessment. The first TRIM module, TRIM.FaTE,
accounts for movement of the pollutant mass through the ecosystem and determines the pollutant
concentration in media and biota. Exposures will be evaluated within the TRIM Exposure Event
Module by tracking small population groups of humans and/or other organisms, referred to as
"cohorts," through time and space. Also included in TRIM will be a Pollutant Uptake Module,
which will determine the quantity of a pollutant entering an organism during a specific exposure
event; a Biokinetics Module, which will determine the quantity of a pollutant delivered to a
target organ; and a Dose-Response Module, which will estimate health effects caused by the pol-
lutant quantity delivered to a target organ. The final module of TRIM, the Risk Characterization
Module, will present the risk estimates, assumptions, and uncertainties. A brief summary of each
module follows.
2.2.1 Environmental Fate and Transport (TRIM.FaTE) Module
TRIM.FaTE will estimate pollutant concentrations in multiple environmental media and biota, a
capability not currently available in other EPA air exposure models. TRIM.FaTE has been the
focus of current development efforts, and a TRIM.FaTE prototype has been developed. This
development has produced a library of algorithms that account for transfer of mass throughout an
environmental system, data to initialize these algorithms for a test site, and a working prototype.
The TRIM.FaTE module will model the movement of pollutant mass over time, through a user-
defined, bounded system, which includes both biotic and non-biotic (abiotic) components. Com-
plete details on the TRIM.FaTE module are presented in Chapters 3.0 through 6.0.
2.2.2 Exposure Event Module
The Exposure Event Module will be used to move a cohort of humans or other organisms
through locations where exposure can occur according to a specific activity pattern. The deve-
lopment of this module will take place primarily in 1998 and 1999. In a typical application,
TRIM.FaTE may be used to provide an inventory of pollutant mass across the ecosystem at
selected time intervals (i.e., days, hours). For pollutants believed to exhibit toxicity via direct
inhalation exposure only (i.e., those that are not persistent and/or bioaccumulative), monitoring
data or air dispersion modeling results may be used in place of data from a fate and transport
model. With these pollutant data as inputs, the Exposure Event Module may be used to define
2-3
-------
Figure 2-1. Conceptual Overview of TRIM
Air Quality Models
(e.g.. ISC3. AERMOD)
1 /
\ \
' Activity Data /
^ (e.g. CHAD) ^
(Population Data /
(e.g.. 1990 BOC) \
G\S Data [
>"»L
Environmental Fate 1
* ana i ranspon
(TRIM-FaTE)
1
Temporal and Spatial
Distribution of |~~
* Pollutant Concentration *
^^^^
r^
3
1
* rzxuuouiu L.VUIII
r
1
/ Physical, Chemical, /
\ rropenies \
/ /
1 Site Specific Data [
/ /
( GIS Data 1
Other Multimedia Models
(e.g.. MEND-TOX,...)
4
Temporal and Spatial
Distribution of
Exposure Level
within Exposed
Population
^ ^x
Biokinetics
Temporal and Spatial
Distribution of Target
Dose
Population Risk Estimates
Measure of Uncertainty
Limitations Description
Temporal and Spatial
Distribution of
Absorbed Dose
Dose/Response
©
Risk Characterization
Pollutant Uptake
Temporal and Spatial
Distribution of
Responses
2-4
-------
the movements of a human or other organism, or a population of organisms (e.g., a population of
an endangered species). The movements could be defined as an exposure event sequence that
corresponds to the time periods modeled by TRIM.FaTE. Each exposure event would place the
organism in contact with one or more of the cells in the ecosystem for a specified time interval.
For example, a particular event may assign the organism to an air cell and a water cell for
specified time intervals (i.e., 1 hour, 1 day). In addition to the location assignments, the event
would provide information relating to the potential for exposure, such as respiration rate (air cell)
and quantity of water consumed (water cell).
2.2.3 Pollutant Uptake Module
The Pollutant Uptake Module will be used to determine the quantity of a pollutant entering an
organism during a specific exposure event. Development of this module will occur in 1999 and
2000 using existing models, where possible. To more accurately estimate dose (and risk) within
an exposed population, the Pollutant Uptake Module will use exposure estimates to calculate the
uptake of a toxic chemical via inhalation (absorption through the lungs), ingestion (absorption
through the gastrointestinal [GI] tract), and dermal exposure (absorption through skin or plant
membrane).
2.2.4 Biokinetics Module
The Biokinetics Module will be used to determine the quantity of a pollutant delivered to a target
organ. Development of the Biokinetics Module will occur in 1999 and 2000 using existing
models, where possible. Since the toxicity of an agent is determined by the concentration of the
toxic chemical in the target organ or tissue, and since the concentration in the target organ or
tissue depends on the disposition of the chemical (i.e., absorption, distribution, biotransforma-
tion, and excretion), accounting for these processes will result in a more accurate estimation of
risk. This is the case because the amount of toxic chemical reaching the target organ may be
higher or lower than the amount that would be predicted if the chemical were assumed to be
uniformly distributed throughout the organism. The Biokinetics Module will use the dose
estimates generated by the uptake module to depict the range of target organ doses (and risks)
within an exposed population.
2.2.5 Dose-Response Module
The Dose-Response Module will estimate health effects caused by the pollutant quantity
delivered to a target organ. Development of the module will occur primarily in 1999 and 2000
using existing models, where possible. A module is desired that can address or apply EPA-
verified health benchmarks, such as cancer slope factors and reference concentrations, to assess
2-5
-------
risks associated with mixtures, evaluate the impacts of dose-rate on toxicity, and determine the
impacts of threshold and non-threshold mechanisms of action.
2.2.6 Risk Characterization Module
The Risk Characterization Module will present the risk estimates, as well as a description of
major assumptions, defaults, and uncertainties, from the risk analysis. The Risk Characterization
Module will be developed starting in 1998 and continuing into 1999. The results of any risk
assessment conducted using the TRIM are intended to support regulatory decision-making.
Therefore, it is critical that TRIM provide results in a manner that is meaningful to EPA risk
managers. This module will format and present risk estimates and related information in a
systematic manner that promotes regulatory decision making and meets the objectives outlined in
current EPA guidance and policy. Ideally, this module will present risk assessment information
in several formats (e.g., graphics, tables) such that risk managers can best interpret and
understand the risk assessment results.
2-6
-------
3.0 Conceptual Framework for TRIM.FaTE
An environmental fate, transport, and exposure model is needed by OAQPS to address non-
inhalation (indirect) routes of exposure not presently addressed in many current models.
TRIM.FaTE has been designed for this reason. TRIM.FaTE is a multimedia chemical mass
balance model being developed to assess contaminant transport among environmental
compartments (such as air, water, and soil) and link these compartments with ecosystem
components. This chapter describes the initial review of multimedia models and explains the
rationale for developing TRIM.FaTE. In addition, the overall logic and terminology for
TRIM.FaTE for expressing transport and transformation of chemical contaminants in a
multimedia environment is provided. This chapter also describes the processes being, simulated
in TRIM.FaTE, illustrates and discusses the mass balance approach and the resulting system of
differential equations for first-order systems, and demonstrates the application of the TRIM
mass-balance approach to a simple four-compartment environmental system.
3.1 Review of Existing Tools
The first step in the model development process was to evaluate EPA and non-EPA approaches
already existing in the fields of non-inhalation exposure assessment that may meet or contribute
to the needs of the TRIM approach. In April 1996, a review of existing models and approaches
was undertaken as part of the initial step in the TRIM development effort. The report, entitled
Evaluation of Existing Approaches for Assessing Non-Inhalation Exposure and Risk with
Recommendations for Implementing TRIM12, examined several multimedia models. Two
additional EPA studies10'11 conducted in 1997 have updated the 1996 study.
The literature searches identified several models/approaches for multimedia, multipathway
modeling for evaluation, including EPA's Indirect Exposure Methodology (IEM2), the California
Department of Toxic Substance Control's Multimedia Risk Computerized Model (CalTOX), the
Integrated Spatial Multimedia Compartmental Model (ISMCM), and the Multimedia
Environmental Pollutant Assessment System (MEPAS).
Efforts to assess human exposure from multiple media date back to the 1950s, when the need to
assess human exposure to global fallout led rapidly to a framework that included transport both
through and among air, soil, surface water, vegetation, and food chains13. Efforts to apply such a
framework to non-radioactive organic and inorganic toxic chemicals have been more recent and
have not as yet achieved the level of sophistication that exists in the radioecology field. In
3-1
-------
response to the need for multimedia models in exposure assessment, a number of multimedia
transport and transformation models have recently appeared. Thibodeaux14'15 proposed the term
"chemodynamics" to describe a set of integrated methods for assessing the cross-media transfers
of organic chemicals. The first widely used multimedia compartment modeling for organic
chemicals were the "fugacity" models proposed by Mackay16'17 and Mackay and Paterson18'19.
Cohen and his co-workers applied the concept of multimedia compartment modeling as a
screening tool by developing the Multimedia Compartment Model (MCM)20, followed by the
Spatial Multimedia Compartment Model (SMCM)21, and more recently the ISMCM, which
allows for non-uniformity in some compartments. Another multimedia screening model, called
GEOTOX22 was one of the earliest multimedia models to explicitly address human exposure.
The CalTOX program23'24-25 has been developed for the California EPA as a set of spreadsheet
models and spreadsheet data sets to assist in assessing human exposures for toxic substances
releases in multiple media. More recently, SimpleBOX, Version 2.026 has been developed for the
National Institute of Public Health and the Environment in the Netherlands in order to evaluate
the environmental fate of chemicals. Results can be for a level 3 (non-equilibrium, steady state)
or quasi-dynamic level 4 (non-equilibrium, non-steady state) system. All phases within the
compartments are assumed to be in a state of thermodynamic equilibrium at all times.
A brief summary of each of the multimedia models that were evaluated for its applicability to the
TRIM effort follows:
Indirect Exposure Methodology (IEM2). With an interim final document
completed in 199027 and with an addendum completed in 199328, the IEM
incorporates current EPA guidance. Descriptions of the fate and transport,
exposure pathways, and dose algorithms are presented in this methodology. This
methodology sets out procedures for estimating the indirect (i.e., non-inhalation)
human exposures and health risks that can result from the transfer of emitted
pollutants to soil, vegetation, and water bodies. The methodology addresses
exposures via inhalation; food, water, and soil ingestion; and, dermal contact.
There appear to be several shortcomings in the methodology. For example, the
methodology, as currently implemented, can be applied only to chemicals that are
emitted to the air. This methodology is not a comprehensive environmental audit,
but is best regarded as an evolving and emerging process that moves EPA beyond
the analysis of potential effects associated with only one medium (air) and exposure
pathway (inhalation) to the consideration of other media and exposure pathways.
Most importantly, it is crucial in the development of TRIM that a sense of
continuity be maintained between the existing (IEM2) and proposed (TRIM)
methodologies. IEM2 has undergone extensive scientific review.
3-2
-------
The Science Advisory Board (SAB) has identified several major limitations of
EM, which can be useful in focusing efforts in TRIM development. While EM
represents the current EPA guidance on multimedia multipathway modeling, it does
not meet the needs of OAQPS. One of the major limitations of EM is that it
consists of a one-way process through a "train" of linked models or algorithms and
is based on annual average air concentrations, wet and dry deposition values from
air dispersion modeling. As a result, it is not a truly coupled multimedia model and
thereby does not have the ability to model "feedback" loops or secondary emissions
and cannot provide time-series estimation of media concentrations and concomitant
exposure. In addition, the methodology does not provide for flexibility in site-
specific applications or in estimating population exposures. Significant site-
specific adjustment must be made to allow for spatially tracking differences in
concentrations and exposure. Much of the focus is on evaluating specific receptor
scenarios (e.g., recreational or subsistence fisher) that may be indicative of high-end
or average exposures but does not allow for modeling the range of exposure
scenarios within a population. Therefore, EM cannot estimate population exposure
distributions. More recent advances29 have addressed some of these limitations to
some degree but have not been fully implemented.
California Department of Toxic Substance Control's Multimedia Risk
Computerized Model (CalTOX). First issued in 1993 and updated in 1995,
with continual enhancements underway, CalTOX was developed as a spreadsheet
model for California's Department of Toxic Substance Control (DTSC), to assist in
human health risk assessments that address contaminated soils and the contami-
nation of adjacent air, surface water, sediment, and groundwater. CalTOX consists
of two component models: a multimedia transport and transformation (i.e., fate and
transport) model, which is based on both conservation of mass and chemical equili-
brium; and, a multipathway human exposure model that includes ingestion, inhala-
tion, and dermal uptake exposure routes. CalTOX is a fully mass balancing model
and also includes add-ins to quantify uncertainty and variability.
The multimedia transport and transformation model is a dynamic model that can be
used to assess time-varying concentrations of contaminants introduced initially to
soil layers or for contaminants released continuously to air, soil, or water. The
CalTOX multimedia model is a seven-compartment regional and dynamic
multimedia fugacity model. The seven compartments are (1) air, (2) ground surface
soil, (3) plants, (4) root-zone soil, (5) the vadose-zone soil below the root zone, (6)
surface water, and (7) sediment. The air, surface water, ground surface soil, plants,
and sediment compartments are assumed to be in quasi-steady state with the root-
zone soil, and vadose-zone soil compartments. Contaminant inventories in the
root-zone soil and vadose-soil zone are treated as time-varying state variables.
Contaminant concentrations in groundwater are based on the leachate from the
vadose-zone soil.
The multipathway exposure model encompasses 23 exposure pathways, which are
used to estimate average daily doses within a human population in the vicinity of a
3-3
-------
hazardous substances release site. The exposure assessment process consists of
relating contaminant concentrations in the multimedia model compartments to
contaminant concentrations in the media with which a human population has
contact (personal air, tap water, foods, household dusts/soils, etc.). The explicit
treatment of differentiating environmental media pollutant concentration and the
pollutant concentration to which humans are exposed favorably distinguishes
CalTOX from many other exposure models. In addition, all parameter values used
as inputs to CalTOX are distributions, described in terms of mean values and a
coefficient of variation, rather than as point estimates or plausible upper values
such as most other models employ. This stochastic approach allows both sensiti-
vity and uncertainty to be directly incorporated into the model operation. This
model does not conserve mass.
As indicated in the literature review reports, the CalTOX model appears to be the
most promising existing model for application to the TRIM effort. Several of the
mathematical concepts and derivations used by the developers of CalTOX can be
directly applied to the TRIM approach. However, CalTOX does have several
limitations that prevent it from being entirely imported into the TRIM approach.
These limitations result from going beyond intended applications for CalTOX; for
example, for landscapes in which there is a large ratio of land area to surface water
area, for a limited range of chemicals (e.g., non-ionic organic chemicals in a liquid
or gaseous state). As a result, the model does not provide adequate flexibility in
environmental settings and chemical classes (e.g., volatile metals such as mercury)
to be suitable for OAQPS needs. The most significant of these limitations, in terms
of application to TRIM, is the fact that the CalTOX model, as it currently exists,
does not allow spatial tracking of a pollutant as is required in the TRIM approach.
SimpleBox. SimpleBOX is a steady-state, non-equilibrium partitioning, mass
balance model. It consists of eight compartments, three of which are soils of
differing use and properties. It also produces quasi-dynamic (non-steady-state)
output by using an external numerical integrator. The model was developed as a
regional scale model for the Netherlands, so its default characteristics represent the
Netherlands30. SimpleBOX uses the classical concentration concept to compute the
mass balance. Its goals are comparable to TRIM to the extent that it simulates
regional systems31; however, its level of spatial and temporal complexity does not
match TRIM'S goals.
Integrated Spatial Multimedia Compartmental Model (ISMCM). ISMCM
has been under development with the School of Engineering and Applied Science at
University of California Los Angeles for the approximately 15 years. A newer
version of ISMCM, called MEND-TOX, is currently under evaluation by the EPA's
Office of Research and Development National Exposure Research Laboratory.
ISMCM considers all media, biological and non-biological, in one integrated
system. ISMCM includes both spatial and compartmental modules to account for
complex transport of pollutants through the ecosystem. Assuming mass
3-4
-------
conservation, ISMCM is able to predict transport based on a sound mechanistic
description of environmental processes, including estimation of intermedia transfer
factors. One of the limiting factors with the ISMCM system for use in the TRIM
system is that it is not structured to incorporate uncertainty/variability directly into
the model operation.
One of the limitations of the ISMCM model within the context of the goals for
TRIM (as described in 1995 thesis32) is the fact that the links and compartments
(spatial configuration) of this model are predetermined. ISMCM was apparently
not designed from start with the necessary flexibility. Having this flexibility is not
a trivial thing to request, if the system is to be fully integrated.
Multimedia Environmental Pollutant Assessment System (MEPAS).
MEPAS was developed at the U.S. Department of Energy's (DOE) Pacific North-
west Laboratory to assess risks from mixed wastes at DOE facilities. MEPAS is a
model that can be run on IBM-compatible PCs. This model consists of single-
media transport models linked together under appropriate boundary conditions.
The model considers four primary pollutant pathways (groundwater, overland,
surface water, and atmospheric) in evaluating human exposure and health effects.
The model also contains an exposure and risk module. The model's ability to
estimate multipathway risks for chemicals and radionuclides makes it unique. The
nature of its algorithms make it a screening tool, rather than a detailed assessment
tool. The model is updated periodically and the latest version of MEPAS (Version
3.1) also contains a uncertainty and variability analysis module (SUM)33. The
mathematical design of this model does not include mass balance and could not be
integrated into TRIM.
"e*
As with IEM2, MEPAS represents a "linked" model system that utilizes a one-way
process through a train of models that individually describe a specific environ-
mental process or media. These types of models are not mass conservative and do
not allow for appropriate temporal tracking of the pollutants and concomitant
exposure.
3.2 The Need for an Improved Fate and Transport Modeling Tool: TRIM.FaTE
Current OAQPS models for hazardous and criteria air pollutants do not address multimedia
exposures, and current OAQPS HAP models do not adequately estimate temporal and spatial
patterns of exposures. Adopting or incorporating existing models into a tool that meets OAQPS
needs represents the most cost-effective approach to developing the tools needed to support
regulatory decision making related to hazardous and criteria air pollutants. Based on the OAQPS
review of current multimedia models or modeling systems (described in Section 3.1), there is no
single model that meets the needs of OAQPS (outlined in Section 1.3 above) and that can be
adopted as part of TRIM. Most models are limited in the type of media and environmental
processes addressed. No single model can address the broad range of pollutants and environ-
3-5
-------
mental fate and transport processes anticipated to be encountered by OAQPS in evaluating risks
from hazardous and criteria air pollutants. In addition, it is also not likely that one individual
model could be developed to address this wide range of concerns. Therefore, the TRIM frame-
work emphasizes a modular design. The lack of a flexible multimedia fate and transport model
was identified as a major limitation and has become the focus of the first phase implementation
efforts for TRIM.
Current multimedia models can be divided into three basic categories, each with its own advan-
tages and disadvantages: "linked" model systems, fugacity models, and compartmental models.
However, the identified limitations were considered critical and, therefore, deemed unacceptable
for incorporating such models into TRIM. "Linked" model systems (e.g., DEM2, MEPAS) gene-
rally utilize a one-way process through a series of linked models that mathematically describe
distinct environmental media or processes (e.g., aquatic environment). These types of models can
never be truly mass conserving and cannot address feedback loops and secondary pollutant
movement (e.g., revolatilization and transport). Fugacity models (e.g., CalTOX) typically are
compartment modes without explicit spatial scale (zero dimensional); thus, they do not provide
the ability to spatially track pollutant movement. They are also applicable only to a limited range
of chemical classes (e.g., inappropriate to model volatile metals [e.g., mercury]). Compartmental
models (e.g., MCM) ar? ?.ISG zero dimer.Fior.al ?.nd do not allow for spatial tracking of pollutant
movement and concomitant exposures. Spatial compartmental models (e.g., ISMCM) represent
the closest current models to an integrated multimedia system. However, as previously described,
it does not meet the TRIM design goals for a flexible architecture.
In general, none of the current models is a sufficiently coupled multimedia model that accounts for
inherent "feedback" loops or secondary emissions (i.e., re-emission of deposited pollutants) or
releases to specific media, or that provides the temporal and spatial resolution critical in
estimating exposures. While it is unknown as to the degree to which modeled results would differ
between current models and a truly coupled multimedia model, models that are not truly coupled
have been considered to lack scientific credibility. Therefore, OAQPS determined it was
necessary to undertake efforts to develop a truly coupled multimedia model.
3.3 Uniqueness of TRIM.FaTE
Among the unique features of TRIM.FaTE are: (1) its flexibility to be formulated at different
spatial and temporal scales, (2) the ongoing development of an algorithm library, and (3) a full
accounting of all of the chemical mass that enters and leaves the environmental system.
TRIM.FaTE was developed to meet OAQPS modeling needs (Section 1.3) and fit the TRIM
design criteria (Section 2.2). To meet these goals requires a multimedia framework. Also
3-6
-------
required are true coupling of multiple media during a simulation (similar to Mackay-type models)
and a level of spatial and time-series resolution to date only obtained from linked single-media
numerical simulation models. The TRIM development team determined that TRIM must: (1)
address varying time steps (of one hour or greater) and provide sufficient spatial detail at varying
scales (site-specific to urban scale); (2) provide true "mass-conserving" results; (3) have the
transparency needed for use in a regulatory context; and, (4) be a truly coupled multimedia model
rather than a set of linked single media models. After reviewing currently available multimedia
models, the team determined that none of the available models offered all of these features. As a
result, the team engendered a new model framework that is distinct from other multimedia models
and unique among the current arsenal of EPA models.
TRIM.FaTE has a mathematical approach (Section 3.4), which makes possible: (1) different
mixes of compartment numbers, types, and links; (2) a unified approach to mass transfer based on
an algorithm library, which allows the user to change mass transfer relationships among
compartments without creating a new program; and, (3) the flexibility to match a simulation to the
spatial and temporal scales needed for a broad range of pollutants and geographical areas.
Although some applications of TRIM.FaTE may resemble a simple fugacity-based compartmental
model, it can be scaled to simulate time-series and spatial resolutions that current regional
fugacity-type models could not handle. The mathematical linking in TRIM.FaTE enables it to
simulate mass distribution within a system and attain a degree of precision not yet achieved by
other models.
3.4 TRIM.FaTE Basic Concepts and Terminology
The development of TRIM.FaTE began with a "conceptual blueprint" of the relationships and
processes that describe chemical transport within an ecosystem. This blueprint is shown in Figure
3-1. On this figure, the biota are represented by squares, biotic sinks are represented by
diamonds, and the abiotic media are represented by ovals. The various lines show possible
chemical transfers occurring between each of the components of the ecosystem. Any environment
can be thought of as a complex system, and thus can be represented using systems models that
follow from the principles of systems theory. Lines may represent transfers of energy or matter,
and in this case, the transfers represent chemical contaminants. All of the different locations,
geographical features, and ecosystems are then subsystems interacting with each other.
Because the terminology used in the world of multimedia modeling can have multiple meanings
and implications, it is critical in the conceptualization of any complex model that terminology used
be defined specifically within the framework of that model. Multimedia models by nature are
multidisciplinary. Terminology can be confusing because a single term will have dramati-
3-7
-------
Figure 3-1
Conceptual Blueprint for TRIM.FaTE
Sediment h- KSurface Water
-------
cally different meanings in different disciplines. To avoid confusion, discussion of TRIM.FaTE
terminology is presented in this section. Following the description of the terminology, a
summary is provided of the mathematical basis for TRIM.FaTE. In addition, a glossary is
included in Appendix A.
In the TRIM.FaTE model, the transport of multiple pollutant species in an ecosystem is set up as
a mass exchange among a set of systems used to represent spatial locations, collections of
environmental phases, and chemical species. The primary features of interest are the chemical
inventory and the chemical concentrations of the system as a function of time in the various
components of the modeled system. These values are called state variables since they describe
the state of the system while it is varying34.
The system being modeled is assumed to be partitioned into regions of three-dimensional space.
Each such region is referred to here as a volume element. Typically, only one type of abiotic
medium is contained in a volume element. This term is introduced as a matter of convenience
for organizing objects that have a natural spatial relationship. The region represented by a
volume element could be a cube or more complicated shape. A volume element usually shares a
surface with other volume elements. The spatial resolution of volume elements may vary from
application to application, and even within a single application.
Contained within volume elements are domains. The term "domain" is a loose equivalent of
what is referred to as "media" in environmental fate and transport modeling literature. However,
the term "media" was considered to be limited in its scope because it generally brings up images
of abiotic systems such as soil or air, while TRIM.FaTE includes both abiotic and biotic systems.
Therefore, the term domain was adopted for TRIM from the principles of systems theory to allow
for more flexibility in its definition. A domain is the material that contains a chemical(s). It is
currently assumed that, within any domain, a chemical is uniformly distributed throughout the
volume occupied by that domain. In addition, the various phases (gases, liquids, solids) that
make up a domain are assumed to be in equilibrium with respect to chemical partitioning.
Domains can be thought of in both a general and specific sense within the TRIM.FaTE modeling
structure. In a general sense, a domain type is defined to classify overall system components
such as soil, water, or mouse, or more specific components such as surface soil or vadose zone
soil. A specific manifestation of a domain type is a domain instance. Domain instances belong
to the same domain type with similar attributes. One domain instance is distinguished from
another by the values that define its composition attributes at a given location. For example, any
3-9
-------
"soil" domain type consists of gas, liquid, and organic- and mineral-solid phases. As a domain
instance, a surface-soil domain instance typically has more organic carbon than a vadose-soil
domain instance. Moreover, a vadose-soil domain instance will typically have a higher water-
volume fraction than a root-soil domain instance. Two different mouse domain instances may
differ by the population size attribute. There can be multiple domain instances within a volume
element, i.e., a worm domain instance may exist simultaneously with a surface soil domain
instance in a volume element. Typically, one abiotic domain type (soil, water, air), but multiple
biotic domain types (worm, plants), can exist within a volume element. When there is no need
for additional clarification between domain types and domain instances, these will be referred as
domains in TRIM.FaTE literature.
The set of all domain instances is assumed to contain all of the chemical mass within the eco-
system, excluding sources. A source is an external component that transfers chemical mass
directly into the domain instances. Examples of sources would include the factory emissions of a
chemical into an air domain, or the influx of chemical in a river from outside the modeled
system.
Associated with each domain instance is an inventory address or cell. A cell is a bin within the
computer code, and these cells collectively account for all potential locations of mass within the
ecosystem, and the pollutant sources and sinks outside the ecosystem that are required to balance
the overall mass flow. Each cell is uniquely defined by three indices. The first index is the
volume element. The second index identifies the domain containing the chemical at a given
location. The third index is the chemical species.
An important aspect that is tracked for each cell is the list of other cells in the system with which
it potentially exchanges chemical mass. It is necessary only to store in this list the cells from
which the cell receives mass. Elements of this list are referred to as links. With each link is
associated a sending cell and receiving cell. The sending cell is the cell from which the
chemical is potentially transported, and the receiving cell is the cell that receives the chemical.
Each specific link for any chemical may have unique properties, and hence must be considered as
an object separate from all other links. For example, a link between two particular soil cells may
contain information on the advective flow from the sending cell to the receiving cell. Another
example is the worm-to-soil cell link, which contains information on the ingestion rate of soil by
worms.
3-10
-------
The link between domains includes information on the potential exchange of chemical between
the two domains. This information includes a transfer factor, which is the instantaneous flux
from the sending domain to the receiving domain per unit chemical mass in the sending domain.
Transfer factors are calculated based on transport and fate processes such as advection, diffusion,
dispersion, reaction, and bioaccumulation. The mathematical basis for these transfer factors is
discussed in Section 3.4. The transfer factor is determined by use of the methods in a central
repository of algorithms, called an algorithm library. Algorithms in TRIM.FaTE are equations
that expresses the transfer factor as a function of a set of variables. This function is specific to
the locations, domains, and chemical species represented by the linked cells.
It is stressed that the algorithm library is not intended to consist of only documented methods;
instead, the methods must be properly entered in some standard manner so that they can be
accessed by other software. For first-order transfers, methods have been developed for
converting typically encountered concentration-based equations to mass balance form. Appendix
B presents generalizations for algorithm development. All major methods of pollutant
movement in the environment are frequently modeled with first-order methods. These include
advective processes, diffusive processes, and bioaccumulation.
3.5 Governing Mass Balance Equations
The TRIM.FaTE model is being developed with an emphasis on conserving chemical mass. This
means that the entire quantity of the chemical is tracked throughout the system being modeled.
When applied to a specific domain (e.g., soil or a mouse population), this implies that, over a
given time period, the amount of the chemical in the domain at the end of the period is equal to
the amount of the chemical in the domain at the beginning of the period, plus the gains of the
chemical that occurred during the time period, minus the chemical that was lost from the domain
during the time period.
Currently, the mass balance approach has been implemented primarily for first-order linear
processes. Therefore, the discussion here is limited to models of this type. It is important to note
that higher order non-linear methods can also be implemented within this structure.
A simplification of a transfer process is shown in Figure 3-2 for a system of two cells, where it is
assumed that the fluxes of chemical mass are first-order processes. Denoting by NJt) and Nb(t)
the mass of chemical in cells a and b, respectively (in units of mass), it can be seen that:
3-11
-------
Figure 3-2
Example of First Order Transfer Process for Two Cells
Source Flow = S,
U)
I
Cell a
Chemical mass in
cell = N.
Flow = Tah Na
Cell
Chemical mass in
;.; ' ':cell=lNb:
Flow = R, N.
d d
Flow = Rb Nb
Sink,
Sink,
-------
Gains for cell a = Sa +
Losses for cell a = TgbNa +
and
Gains for cell b = TghNa
Losses for cell b = TbaNb + RbNb
where:
Sa = chemical source in cell a, units of mass/time
Tab = transfer factor for movement of chemical from cell a to cell b during time
interval, units of /time
Tba = transfer factor for movement of chemical from cell a to cell b, units of /time
Ra = reaction loss of chemical in cell a, units of /time
Rb = reaction loss of chemical in cell b, units of /time.
The constraint that mass balance must be preserved means that, over any time interval, the
change in mass in a cell is equal to the gains minus the losses in mass over the time interval. The
instantaneous change in mass with respect to time is the derivative with respect to time, denoted
by dNc/dt. Thus, the mass balance constraint, when applied to the simple system discussed here,
yields a system of two linked differential equations:
dN
dN.
Additional terms are needed to properly account for the chemical mass. In particular, the fate of
the chemicals after reacting must be tracked. For this reason, two additional cells are added to
the system, and serve as the repository of the chemicals after reaction. These are referred to as
"sinks," since once the chemical is transferred into these cells, it no longer moves to any other
cells. While clearly the chemical would continue to move in its altered form throughout the
3-13
-------
system, for this example, this history is not of interest. Denoting by Sinka and Sinkb, the mass in
the reaction sinks for cells a and b, respectively, the complete system is:
dt
dN,
dt
dSinka
dt
dSinkb
dt
- (Ra + Tab)Na
- (*» + TJNb
or, in matrix form:
dN Idt '
a
dNJdt
dSinkJdt
dSinkJdt
1 ab
o
0
o o"
0 0
0 0
0 0
X"
^v
fy
Sinka
Sinkb
X"
0
0
0
Application of this methodology to a simplified ecosystem with ten cells, as shown in Figure 3-3,
yields a transition matrix equation of the form:
d
dl
N.,
N,2
N
'
Nw
Nf
Ns
N
no
N*.
N
Kl
NKf
-Kal 0 Tsal T^al 0 00000
Tala2 00 0 000000
T . 0 -K 0 000000
ajj I
Talw 0 Tw -Kw Tp 0 0 0 0 0
o o o r, -A:, ooooo
HJ /
0 0 Tjg 0 000000
R 00 0 000000
u
OtfH0 0 000000
0 Ofl 0 000000
J
0 0 0 Rf 000000
**.,
Na2
N
*
N»
Nf
N,
N
Ra
/v
Na
!
NRf
-f
5°;
0
0
0
0
0
0
0
o
0
3-14
-------
Figure 3-3
Simplified Ecosystem
The ecosystem consists of:
Two Air Cells
Soil Cell
Groundwater Cell (sink)
Surface Water Cell
Fish Cell
Four sinks
TOTAL =10 Cells
3-15
-------
where:
Sal = source team for air Cell 1
Tsj = .transfer factors for Cell I to j
NJ = mass of pollutant in Cell i
+7als +7ala2
K, = R, +
K, =Rf+7fw+7ro
K^ = R,, + 7WS +7wa
Applying this same approach to a general system with M cells (including all sinks), and allowing
the transfer factors and source terms to depend on time as well, results in a system of linked
differential equations of the form:
dN/dt = A(t)N + s(t), N(to) = N0
where:
N(t) = an Af-dimensional vector whose z'th entry is the mass in the /th cell
A(t) = an M x M time-dependent matrix
s(t) = an Af-dimensional vector accounting for the source terms in each cell.
The matrix A(t) is referred to as the transition matrix for the system. This term is borrowed from
Markov theory, although the model is not strictly a Markov process. The vector s accounts for
pollutant sources located within specific cells. The vector N0 is the initial distribution of mass
among the cells.
3.6 Modeling Approach
This section summarizes the general features of the application of the conceptual approach
previously described.
One of the primary features of the application of the TRIM approach is that it is to be an iterative
and flexible process. When the modeling process is first started, there is a general sequence that
must be followed. After the initial step, however, there is no fixed order in which the modeling
steps are necessarily performed. This process is shown in Figure 3-4. The boxes on the left side
of the figure represent a partitioning of the modeling sequence into five broad areas. These areas
include: basic problem definition, specification of links, setting up a run, performing a run, and
analysis of results. The particular division into five such areas is somewhat arbitrary, and in an
actual application, it may be that the progression is not quite as linear as that shown in the figure.
3-16
-------
Figure 3-4
Structure of TRIM.FaTE
PROCESS FLOW
PRIMARY TOOLS
fc
w
fe
w
\
Definition of problem
specify volume elements
specify domain instances
specify data or data source(s)
for domains
1
r
Specify links between domain
instances. For each link,
specify algorithm to use from
available list
specify data or data source(s)
for link
^
r
Set up run
set initial conditions
set source term(s)
set output time periods of
interest
T
r
Perform run
Call algorithm library for each
link to determine transfer
factors
Calculate mass distribution in
system of domain instances at
requested time periods
1
r
Analysis of results
DATA"
Spatial Data
Flow data
Chemical properties
Source terms
ALGORITHM
LIBRARY
GENERAL CALCULATION TOOLS
Differential equation solver
Partial differential equation solver
3-17
-------
H< wever, all of these steps are necessary. The vertical arrows between these boxes represent the
possible order of events in the modeling process. The arrows on the left side of the boxes
indicate the iteration that may be necessary or desired.
The shapes under the heading "Primary Tools" represent the primary tools used in the modeling
process. The arrows from these shapes to the flow boxes indicate where in the modeling process
these tools would be used. To focus on key aspects of the TRIM.FaTE approach, only selected
tools are shown. There are other tools that may be necessary that are not included in this figure.
Such tools would include pre/postprocessing software that may automate some aspects of the
process, and general user interface software.
3.6.1 Problem Definition
The first step requires the general problem definition. During this step, the chemical(s) to be
modeled and the initial spatial features of the ecosystem are determined. In the nomenclature
previously discussed, the volume elements and domain instances within the volume elements are
specified. For the first cycle through the simulation process, the spatial scales may be crude and
the number of domain instances may be small. It will be necessary at this step to specify various
types of data, or simply the sources of the data (e.g., a remote database). Data types include
spatial information about the ecosystem, chemical-specific environmental data (e.g., degradation
rates in various domain types), and data for the specified domain instances (e.g,. soil densities
and organic carbon content for soil domains).
3.6.2 Link Setup
The second step shown in Figure 3-4 specifies the links between the domain instances. Two
domains are considered "linked" if there is a direct means by which the chemical can be
exchanged. This definition does not include "indirect" links that result from a chain of direct
links (e.g,. chemical is transported in eroding soil to a water domain, and subsequently taken up
by a fish population). The system of links is one of the most critical components of the model.
By specifying a link between two domains, it is assumed that some method exists by which to
estimate the transfer of chemical through the link. If this method is already included in the
algorithm library, then it is only necessary to specify the data (or data source) for this link and
which algorithm to use. These data may depend on both of the domains in the link (e.g., erosion
flow rate for link between a soil domain and a water domain). These data do not include infor-
mation about the mass of the chemical, as tracking the inventory of chemical mass with time is
the purpose of the model. If the algorithm is not in the algorithm library, then it must be "added"
so that it can be accessed by the underlying software.
3-18
-------
3.6.3 Simulation Setup
The third step shown in Figure 3-4 is the preparation of a simulation after the volume elements,
domain instances, and links have been specified. This involves specifying the initial distribution
of chemical mass in the domains, specifying any source terms considered within domains, and
specifying the output time period(s) of interest. The initial conditions may be specified as
concentrations, which are then converted to mass form for the model. The "DATA" drum is
connected to this step because data are necessary for the initial conditions and source term(s).
The initial conditions and source terms may be estimated from monitoring data available, or from
the results of another model.
3.6.4 Simulation Implementation
The fourth step is the actual running of the model, where the movement of the chemical(s)
through the domains is simulated for the specified time period(s). The exact manner in which
this is performed depends on the algorithms chosen. For each link between domains, a call is
made to the algorithm library to determine the transfer factors that indicate the potential
exchange of chemical mass. If all algorithms involve only first-order processes, then movement
of the chemical will be simulated with a system of linked differential equations, the solution of
which would be found using a differential equation solver. For more complicated algorithms,
other tools would be necessary (e.g., a method of solving partial differential equations).
3.5.5 Result Analysis
The last step shown in Figure 3-4 is the analysis of the results generated for the modeled system.
These results include the time history of the chemical mass and associated concentrations in the
domains. This step would also include postprocessing analysis of the results and use of the
results in other parts of the TRIM.
3.7 Sensitivity Analysis
An important aspect of the TRIM is the integration of sensitivity and uncertainty analyses
methods into the model framework. The reasons for a sensitivity analyses are to identify
important inputs with respect to outcome variance in order to direct efforts related to:
Additional data collection
Additional research
Stratification of the population.
3-19
-------
Many of the parameters used in modeling of natural systems are uncertain or variable. It is
critical to confront sources and ranges of parameter variance for several reasons. Among them
are the need to determine the range of possible outcomes of the model, and the need to determine
what parameters are the important contributors to the range of outcome values generated by the
model.
The TRIM framework is designed to provide for a tiered uncertainty/sensitivity analyses in
several ways. All inputs to TRIM are entered in parameter tables where value distributions are
the default option and the labels "uncertain" or "variable" can be applied to make initial classifi-
cations. The capability to conduct a joint uncertainty and variability analysis is a goal of TRIM.
Currently, the capability exists to conduct simple sensitivity analyses. Ultimately, Monte Carlo
assessments and uncertainty importance assessment capabilities will be an integral part of
TRIM.FaTE. Some limited assessment of model uncertainty is provided through the option of
selecting from alternate transport/transformation algorithms from an algorithm library.
3.8 Example Calculation of Transfer Factors
In previous sections, the term "transfer factor" is used to describe the potential transfer of
chemical mass between two domain instances. This section shows an example of how these
transfer factors are determined for a first-order process, starting with a model for estimating the
concentration of a chemical in fish.
Thomann35 gives the following model for calculating the concentration in fish:
dC
where:
Cf = concentration in fish (micrograms per kilogram [ug/kg])
ku = uptake rate from water via the gills (1/kg-day)
CWD = dissolved chemical concentration in water (micrograms per liter [ug/L])
kD = chemical uptake from food (kg food/kg fish/day)
P, = proportion of the diet consisting of food item I
CD j = chemical concentration in food item I (ug/kg)
3-20
-------
kE = elimination via fecal egestion (I/day)
keg = elimination via the gills (1/day) -
RM = metabolic transformation of chemical (I/day)
ko = dilution contaminate concentration from growth (I/day).
This algorithm was derived to estimate concentrations in individual fish of a species. Initially, the
model is generalized for a population of two fish, and then for the case of an arbitrary number of
fish. The previous equation is further simplified by assuming that there is no uptake through
other food items. Also, the elimination via fecal egestion and the metabolic transformation
factors are neglected. Thus, for two fish with concentrations Cfl and Cn the previous equation
can be rewritten as:
dC
j
To convert the concentrations to masses it is assumed that:
N
r - *
<~WD y '
m,
where:
nij = mass of fish 1 (kg)
m2 = mass of fish 2 (kg)
N! = mass of contaminant in fish 1 (pg)
N2 = mass of contaminant in fish 2 (fig)
Vw = volume of surface water cell (L).
3-21
-------
Substituting results in:
dt
* Us.
IV i ^
N
dt
N.
k , L
k
Adding these equations yields the mass transfer equations for the total fish population consisting
of the two fish, as follows:
d ( N./m. + NJnu )
1 ' 2
N
N.
Making the simplifying assumptions that individual fish mass is represented by a population
average mf (m1=m2=inf), and that ku^lu^k,, and kegl=keg2=keg, yields:
m
/ / _
dt
This equation can be generalized from two to ^fish, with Nf (= N,+N2) being the total pollutant
mass in the fish domain to yield the following generalized mass transfer equation for a fish
domain:
dN
N
Implicit in the previous equation is the assumption that the mass of an individual fish is constant
over the time of the simulation. It may be-noted that the dilution due to growth factor (k^) is not
3-22
-------
included in this equation because RQ is based on concentrations not mass. Transfer factors for the
fish domain are now given by:
w
T = k
* K
where:
Twf = transfer factor for exchange of chemical mass from water to fish population
(/day)
Tfr = transfer factor for exchange of chemical mass from fish population to water
(/day).
3.9 Summary of TRIM.FaTE Approach
In this chapter, the TRIM.FaTE framework has been introduced by describing a unified concep-
tual approach to multimedia mass-balance models. The term "unified" refers to the fact that one
approach has been generalized to all components of a multimedia environment, including eco-
system components. The mass-balance approach for first-order systems reduces to a set of linear
ordinary differential equations was illustrated. However, the approach is not limited to first-
order linear methods. The modeling approach provides a flexible, iterative process of simulating
the movement of chemicals in a multimedia environment. This makes the approach useful for
addressing different types and aquatic and terrestial ecosystems and also for human exposure
assessment. It is important to note that the approach used is not based on linking different
models for different compartments or domain instances. Instead, the entire system is represented
in a single informational structure, i.e., a large matrix. In the next chapter, more specific
examples are presented of the multimedia models that can be constructed using this type of
flexible and iterative process.
3-23
-------
4.0 TRIM.FaTE Prototype Development
This chapter provides a description of the process of applying the TRIM.FaTE methodology
(Chapter 3.0) to cases of increasing complexity (referred to as "prototypes"). Section 4.1
discusses the implementation of the prototypes; Section 4.2 describes the development process
for each prototype; Section 4.3 addresses the features of the prototypes, including the types of
domains and links simulated; and Section 4.4 discusses the processes used to simulate links. The
goal of this chapter is to illustrate the flexibility of TRIM.FaTE for application at different levels
of spatial and temporal resolution. This chapter also serves to illustrate how different multimedia
configurations with TRIM.FaTE are set up.
4.1 Implementation of Prototypes
The concepts discussed in the previous chapter have been implemented using a combination of
Visual Basic, Fortran, and Microsoft Excel software. These implementations are documented
in detail in the technical support document'.
An object-oriented architecture, similar to that shown in Figure 3-2, was implemented using
Visual Basic 5 imbedded within Excel 97 to model the hierarchy of components of TRIM.FaTE.
This hierarchy includes volume elements, domain types, domain instances in the volume
elements, and links between the domains. The coding architecture is not tied to any specific
ecosystem configuration. A preliminary algorithm library that utilizes this coding architecture
was also implemented.
If all transport processes are simulated as first-order process, this results in a system of linear
ordinary differential equations. This system must be solved to determine the redistribution of
chemical mass as a function of time. For TRIM.FaTE, this system is solved using the Livermore
Solver for Ordinary Differential Equations (LSODE)36, a Fortran program freely available via
several online numerical algorithm repositories.
The LSODE subroutine solves systems of first order ordinary differential equations of the form37:
dy/dt = F(t,y), y(to) = y0
where y is an n-dimensional time-dependent vector, i.e.,
4-1
-------
The system of differential equations can be stiff or non-stiff. In the stiff case, it treats the
Jacobian matrix as either a full or banded matrix. It uses Adams methods (predictor-corrector) in
the non-stiff case, and backward differentiation formula methods in the stiff case. The linear
systems that arise are solved by direct methods (LU factorizations). LSODE supersedes the older
GEAR and GEARS packages.
The only restriction on the size of the system of differential equations is that imposed by
computer memory. This code was modified so that it could be accessed by Visual Basic 5 in
Excel 97. Another Fortran code was used, in a similar manner, to determine the steady state
solution to the system of linear differential equations38.
Microsoft Excel spreadsheets were used for general preprocessing, postprocessing, and data
storage (additional databases for spatial data were also created using Visual Basic and accessed
by Excel). Excel spreadsheets also served as a convenient interface to the Visual Basic and
Fortran subroutines.
The approach taken for testing the methodology made it possible to investigate the implications
of draft algorithms and to work on the development of a flexible system for addressing
conceptual site models with many domains. The pre- and postprocessing for the ultimate
implementation of TRIM.FaTE may require a more sophisticated platform. However, with some
modification, much of the Visual Basic code, and all of the Fortran code, can be used in other
computer programming languages.
4.2 Prototype Development
Multiple prototypes were developed with increasing complexity to model the movement of a
pollutant through an ecosystem. This section describes features of the prototypes in increasing
order of complexity.
4.2.1 Prototype 1
Prototype 1 (PI) was set up to test the mass transfer methodology (Chapter 3.0) and the LSODE
utility. Air, soil, groundwater, surface water, and fish domains were simulated in PI as seen in
the conceptual site model shown in Figure 4-1. PI includes a uniform volume source emission
of benzene into the air volume. Benzene was selected because most of its transfer factors were
readily available from CalTOX39.
4-2
-------
Figure 4-1
Conceptual Site Model for Prototype 1
1,000m
1,000m
Air Cell
5m
I
1
90m
Soil Zone
GW
-------
Some transfer factors were derived independently of CalTOX for the air to air sink, soil to
groundwater, fish to water, and water to fish transfers. The remaining factors were taken directly
from CalTOX. The dimensions of the terrain were adapted from CalTOX to facilitate com-
parison of results. Chemical reaction was not simulated in this prototype.
The runs produced estimates of benzene mass throughout the system, and no problems were
experienced in running the LSODE subroutine. The resulting mass distribution of benzene in
various domains was commensurate with theoretical expectations and showed that TRIM.FaTE
is relatively easy to implement for a simple ecosystem. These results prompted further testing of
the modeling approach on a more complex ecosystem.
4.2.2 Prototype 2
Prototype 2 (P2) includes a more spatial detail than PI and more sophistication than PI in both
the types and number of domains used. Unlike PI, P2 included multiple volume elements for
both the soil and air domain types and included the use of plant and sediment domains. In
addition, the links between cells had multiple-phase (i.e., gas, liquid, and solid) mass transfers.
P2 included a volume source emission of benzo(a)pyrene (B[a]P) into only one of the air
volumes. This made possible a very simple representation of spatial transport. B(a)P was
selected as a test chemical for this and subsequent prototypes because of its persistence in the
environment and the fact that it is a pollutant of interest to EPA's Risk and Exposure Analysis
Group. The derivation of the transfer factors are described in detail in the technical support
document1. The conceptual site model for P2 is shown in Figure 4-2.
Multiple-phase (liquid, gas and solid) transport within a domain was introduced in P2. The
phases are assumed to be at chemical equilibrium, with the ratios of the concentrations in the
individual phases constant.
4.2.3 Prototypes
The Prototype 3 (P3) code and input data are significantly more complex than either PI or P2.
P3 was developed both to incorporate lessons learned from P2, which has a refined set of abiotic
algorithms, and to set up the TRIM.FaTE model for the case study model run Prototype 4 (P4).
P3 includes a conceptual site that dimensionally represents the ecosystem of the test area for P4.
The conceptual site model for P3 is shown in Figure 4-3. The vertical dimensions of individual
air cells are not indicated because these dimensions were allowed to vary with time according to
4-4
-------
Turbulent
layer
5m
5m
Figure 4-2
Conceptual Site Model for Prototype 2
2.5km
2.5km
air over land
soil
plants
1m T
root
I
2m
t
2m
groundwater
air over water
5km
lake
mic olayered
fish
interstitial sediment
intersti
1 m^
sediment
1 km
5km
10m deep
-------
J
'varies^
with
time
varies
with
time
0.001m
0.999
m
2m
2m
5m
Figure 4-3
Conceptual Site Model for Prototype 3
-------
a set of specified meteorological conditions. The soil and surface water domains were split into
finer grid structures relative to P2, and several new biotic algorithms were added. The source
term simulated in P3 was a volume-source emission of B(a)P into only one of the four air volume
elements. This was used to make an approximation to a continuous point-source release.
The differences of P3 relative to P2 include:
Addition of terrestrial earthworms, kingfisher, and mouse domains
Addition of aquatic food-web system
Addition of cells with varying heights for the air domain to increase complexity
Division of soil cells horizontally to add complexity to soil domain
Introduction of "thermoclines" and refinement of mixing for surface water
Refinement of plant domain algorithms
Refinement of soil diffusion algorithms
Addition of erosion in the soil domain
Refinement of groundwater algorithm
Introduction of flexible code design
Introduction of temporal variation for a few key input parameters.
4.2.4 Prototype 4
Whereas PI through P3 used generic inputs and were intended for evaluation simulations, P4
was set up to be applied to an actual site. PI through P3 were used to develop and test the
TRIM.FaTE algorithms. P4 was developed and used to illustrate and evaluate the likely limits of
TRIM.FaTE with respect to the number of land parcels and length time steps used. This
prototype had the shortest plausible time step (1 hours), a large number of land units in the plan
view (20 parcels), and 21 different biotic domain types. This level of detail resulted in several
hundred cells, including abiotic and biotic domain instances, and the sinks needed to account for
transformation and transport losses outside of the system boundary. To test the model using a
realistic ecosystem, P4 was applied to an area in the northwestern region of the United States.
This prototype was developed as described in the setup methodology in Chapter 3.0. This
section provides a general description of the P4 study area and the process of mapping the case
study area into a form that is usable in TRIM.FaTE.
Description of Environmental Setting. In P4, the TRIM.FaTE model was applied to the
simulation of B(a)P and phenanthrene releases in a much more realistic test case: a mixed use
landscape surrounding an aluminum smelter. The circular region containing all land within 50
km of the facility was examined to define the boundaries of the study area. Precursory air
dispersion modeling was performed and results indicated that significant impacts of the
4-7
-------
emissions occurred within a radius of 5 km. The land use within this 5-km area was evaluated
from Geographic Information System (GIS) information and it was determined that an oval study
area approximately 8.5 by 9.0 km would provide an instructive test case for the TRIM.FaTE
model. Figure 4-4 is the plan view and Figure 4-5 displays the cross-sectional views of the study
area used for the test case.
The test case facility is located near a bay in an area that is predominantly industrial in nature.
Much of the area immediately surrounding the smelter is used for storage of timber prior to ocean
shipment. The nearest residents (human) are located approximately 800 meters east, north, and
northeast of the facility. Approximately 800 meters north of the aluminum smelter is a ridge
running in a southeast-northwest direction, with a maximum elevation of approximately 120
meters above sea level.
The only other major industrial facility in the vicinity of the aluminum smelter that has been
identified as having a significant potential to emit air pollutants is a paper mill approximately 5
km due west of the smelter and on the bay. Nearby sources of polyaromatic hydrocarbons (PAH)
may also include residential wood smoke, emissions from automobiles (to air), and boats (to
water), among others. For purposes of P4, it was assumed that the aluminum smelter was the
only source of phenanthrene and B(a)P within the study area.
Although an actual location in the northwestern region of the United States was used as a rough
guide for constructing this system, the application of TRIM.FaTE to this system was not intended
to provide pollutant estimates for any existing facility in the United States.
Mapping the Ecosystem for P4. The plan view of the map resulting from the mapping
process is shown in Figure 4-4. This process also involved setting up the domains and associated
links as described in Chapter 3.0.
Determining the appropriate grid scale to use in this modeling effort is based on tradeoffs
between the desired level of detail in the results and the data computational requirements
necessary to run a detailed model. When determining the grid scale to use in the model, it is
desirable to include as much detail as necessary to capture the spatial resolution, both in terms of
land use and in order to capture the spatial change in chemical concentration. On the other hand,
it is undesirable to have so much detail as to increase the complexity of the model to the point
where it is difficult to set up and run. Ideally, there should be enough grids to capture the details
needed for the required task and no more. Based on this tradeoff and as shown in Figure 4-4, the
4-8
-------
Figure 4-4
3.0km
Plan View for Study Area in Prototype 4
N
1.5km
1.5km
3.0km
1.8km
1.0km
2.0km
3.0km
8.5km
Source
-------
Figure 4-5
Cross Sectional View for Land, River, and Lake Parcels in Prototype 4
,r
t
t
/
/
/
/
f
/
Upper Air
Lower Air
FOREST/URBAN/
AGRICULTURAL
Surface Soil
Root Zone
Vadose Zone
Groundwater
/
i
/
' n
//
'
/
/
/
/
/
/
/
/
'l~~
f
/
/
Upper Air
r
/
/
/
/
Lower Air
^
*
/
/
/
Surface Water
/
,' Sediment
"4
t
/
T
*
t
'
'
i
t/
/j
/
Upper Air
,
/ "»» «. .
g
g
»
i
Lower Air
/!
\ Upper Surface Water
/ i
1 (MAinr Siirtec* Water
, Rflrtimunt
~-\
i\
'
i
...
/
/
/
/
/f
^
|T
Typical Land Parcel
Typical River Parcel
Typical Lake Parcel
-------
landscape was divided into 15 land units (parcels), and the water systems were divided into 5
water units (parcels). It is important to note that the embayment and other waters surrounding
the smelter were assumed to be fresh, not saline, for purposes of the modeling effort.
The GIS data for this area indicates that the land area is primarily forest and urban, with a patch
of agriculture and some very small parcels of grassy vegetation within the urban areas. GIS
databases accessed to characterize the study area are summarized in Table 4-1. Details on each
database are available in a separate report entitled Draft GIS and Spatial Data Report for the
Total Risk Integrated Model (TRIM)*0. The pattern of land use is irregular and parcels were
defined to be representative of the major land usage for the specific parcel selected.
Consequently, the location of each land use type does not correspond exactly with the actual
location of that type of land, but the total area of each type of land is representative of the total
actual areas. In locating the different land types, an effort was made to replicate the actual
locations of each usage as much as feasible. The grids are either urban, forest, or agriculture.
The small amount of grassy vegetation was accounted for by assuming that the urban parcels
consist of grassy and paved areas.
4.3 Prototype Features
The specific features simulated in the prototypes are discussed in this section. Section 4.3.1
presents the types of abiotic domains modeled; Section 4.3.2 includes the types of biotic domains
modeled; and Section 4.3.3 discusses the abiotic and biotic links associated with the prototypes.
4.3.1 Abiotic Domains
In PI (Figure 4-1), the air, soil, and surface water each consist of a single volume element.
Groundwater was simulated simply as a sink to the soil domain. P2, as shown in Figure 4-2,
consists of an air domain that contained 4 volume elements (2 upper air and 2 lower air layers);
the soil domain, which was divided into 4 volume elements (surface soil, root zone, and vadose
zones 1 and 2); and groundwater, surface water, and sediment, which were each simulated as a
single volume element. In P3, (Figure 4-3) the air domain consists of 6 volume elements (2
lower air and 2 upper air over soil, and a lower air and upper air over surface water); the soil
domain was divided into 32 volume elements (8 surface soil, 8 root zone, 8 vadose zone 1, and 8
vadose zone 2); groundwater and surface water were both simulated with 2 volume elements; and
sediment was simulated as a single volume element. P4 simulates 129 abiotic volume elements.
As shown in Figures 4-4 and 4-5, parcels were defined and divided vertically based on domain
type. The 129 abiotic domain instances associated with the parcels in P4 are summarized in
Table 4-2.
4-11
-------
Table 4-1
Databases Consulted in Developing Prototype 4 Ecosystem
Data type
Land use
and land
cover
Hydrology
Elevation
Soils
Database
USGS Land
Use and Land
Cover
Reach File
USGS1
Degree DEM
State Soil
Geographic
database
(STATSGO)
Characteristics of Database
Coverage
National
National
National
National
Data
source
High-
altitude
photo-
graphs
EPA
Maps
and
photo-
graphs
Soil
survey
maps,
remote
sensing
Scale
Minimum
interval =
40m to
160m
1:500Kto
1:1 OOK
70 m x 90
m cell size
map unit =
625
hectares =
2500 m x
2500m
square;
depth to
250 cm in
11 layers
Attributes
Anderson
detailed land
use
categories
Extensive
Elevation of
each cell as
integer
percent-ages
of soil types
in each map
unit (21
classes)
Notes
Pros: Easily aggregated into general categories.
Cons: Created in late 70s and early 80s (out of
date). Long narrow polygons are precluded due to
minimum width requirement.
Pros: National in extent.
Cons: Current version (RF1) not reliable for all
areas of the country. Complex routing (e.g., flow
direction not stated explicitly). RF3 file currently
under development may be improvement.
Pros: National in extent. Highest resolution dataset
available for whole nation.
Cons: Accuracy low in areas of low relief. Cell size
leads to smoothing of local relief.
Pros: National in extent.
Cons: Poor resolution. Not easily translatable for
use in spatial models. No mapping below 250 cm
depth. Sometimes discontinuous at state lines.
-------
Table 4-2
Types of Abiotic Domains and Number of Volume Elements Modeled
Domain
Mr
Soil
Surface Water
Sediment
TOTAL
DUMBER
Number of Volume Elements *
P1
1 - Air Layer
1 - Soil (general)
1 - Groundwater
1 - Surface
Water Layer
NA
4 Volume
Elements
P2
2 - Upper Air Layer
2 - Lower Air Layer
1 - Surface Soil
1 - Root Zone
1 - Vadose Zone 1
1 - Vadose Zone 2
1 - Groundwater
1 - Surface Water
Layer
1 - Interstitial Water
1 - Sediment
12 Volume Elements
P3
3 -Upper Air Layer
3 - Lower Air Layer
8 - Surface Soil
8 - Root Zone
8 - Vadose Zone 1
8 - Vadose Zone 2
2 - Groundwater
1 - Upper Surface
Water Layer
1 - Lower Surface
Water Layer
1 - Interstitial Water
1 - Sediment
44 Volume Elements
P4
20 -Upper Air Layer
20 - Lower Air Layer
14 -Surface Soil
14 -Root Zone
14 -Vadose Zone 1
14 - Vadose Zone 2
14 - Groundwater
1 - Upper Lake Layer
1 - Lower Lake Layer
5 - River Segments
6 - Interstitial Water
6 - Sediment
129 Volume Elements
*Reaction and advection sinks are not listed in this table.
4.3.2 Biotic Domains
In PI and P2, a single fish species is modeled and only uptake and loss of contaminant through
the gills is simulated. In the transition from P3 and P4, the number of biotic water column
domain instances was expanded from a single fish species to an aquatic food web represented by
several feeding trophic levels (domain instances). Bioaccumulation by herbivores, as well as
omnivores and carnivores, is accommodated within the P3 and P4 simulations. It is important to
note, however, that the trophic level representations were simplified to reflect primary uptake and
loss from a single representative species from each trophic level. No natural variability specific
to individual populations or communities is accounted for in P3. In P4, distribution ranges for
parameters such as lipid content, ventilation rate, and individual size are included. For example,
the aquatic carnivore community is represented by a single fmfish, the Largemouth Bass.
4-13
-------
Both P3 and P4 include terrestrial wildlife as domain instances. Wildlife may be exposed to
contaminants through food, soil, and water ingestion, and through inhalation of contaminants in
air. Elimination of contaminants from body tissues may occur through metabolic breakdown of
the contaminant and excretion through urine, feces, milk (mammals only), and eggs (birds and
reptiles only). Terrestrial and semiaquatic biota were not considered in PI and P2. Two species
were introduced in P3: a white-footed mouse (Peromyscus leucopus) and the belted kingfisher
(Ceryle alcyon). These species were selected because they are taxonomically dissimilar
(mammal versus bird) and represent differing domains (terrestrial omnivore and semiaquatic
piscivore, respectively). P4 simulated a more complex terrestrial, aquatic, and semiaquatic
system, as summarized in Table 4-3.
Table 4-3
Biotic Domains Modeled
Domain
P1
P2
P3
P4
Aquatic
Ecosystem
Single
Fish
Species
Single
Fish
Species
Macrophytes (Benthic
Herbivores)
Aquatic Herbivores
Aquatic Omnivores
Aquatic Carnivores
Macrophytes (Benthic
Herbivores)
Mayfly (Benthic Herbivores)
Bluegill (Modeled as
Herbtvore)
Channel Catfish (Omnivore)
Bass (Carnivore)
Mallard (Herbivore)
Raccoon (Omnivore)
Tree Swallow (Insectivore)
Terrestrial
Ecosystem
NA
NA
White-footed Mouse
(Omnivore)
Earthworm (Soil
Detritovore)
Plant Leaves, Roots,
Xylem and Stem
Semi-
Aquatic
Ecosystem
NA
NA
Belted Kingfisher
(Piscivore)
White-footed Mouse
(Omnivore)
Earthworm (Soil Detritovore)
Black-capped Chickadee
(Insectivore)
Red-tailed Hawk (Predator)
Long-tailed Weasel (Predator)
Black-tailed Deer (Herbivore)
Long-tailed Vole (Herbivore)
Mink (Piscivore)
Trowbridge Shrew (Ground
Invertebrate Feeder)
Insects
Plant Leaves, Roots, Xylem
and Stem
Belted Kingfisher (Piscivore)
Wetland Plant Leaves, Roots,
Xylem and Stem
4-14
-------
P3 and P4 also simulated pollutant transfer to earthworms. The concentration in earthworms was
assumed to be in equilibrium with the solid, liquid, and vapor-phase concentrations of the
chemical in the root zone volume elements.
The plant domain was introduced to the TRIM.FaTE framework in P2. The plant component of
the ecological model implemented for P2 and subsequent prototypes is comprised of leaves,
roots, xylem, and stem. Plants are divided into these components (volume elements) because:
(1) the literature suggests that concentrations of non-ionic organic contaminants in foliage are
primarily related to those in air and that concentrations in roots are generally related to those in
soil (with stems serving as the conduit between the two), and (2) herbivores may eat part but not
all of a plant. The xylem is added for future versions of the model that may address exchanges
between volume elements in which the xylem plays a critical role. Currently, each volume
element is assumed to be homogeneously-mixed. The plant algorithms implemented in P2
through P4 are applicable for mature plants only, and do not yet address plant growth.
4.3.3 Links
If mass can move from one cell to another cell without first moving through intervening cells,
then the two cells are considered "linked." Each linkage is associated with an algorithm that
determines the direction and rate of mass flow between the two cells. Linkages may be between
adjacent volume elements or within a volume element. At a given spatial location, and within a
single volume element, more than one domain may exist and linkages may exist between these
domains. The mass transfer algorithm specific to each linkage was based on review of the
appropriate scientific literature and is discussed in detail in the technical support document1.
Table 4-4 shows examples of generalized linkages applied to PI through P4. This table is
generic and can be used in conjunction with Tables 4-2 and 4-3 to define a specific link. For
example, in P2 through P4, transfer of a pollutant can occur from an upper air cell to adjacent
upper air cells and to a lower air cell. This is represented in Table 4-4 by the air (sending
domain) to air (receiving domain) link. A more complex example is the links associated with the
kingfisher from the semi-aquatic ecosystem. As a receiving domain, pollutant(s) can transfer to
the kingfisher from air (i.e., lower air), soil (i.e., surface soil), surface water (i.e., upper lake
layer), and aquatic (i.e., bluegill) ecosystems.
4-15
-------
Table 4-4
Examples of Links Associated with Domains
Sending Domain
Air
Soil
Groundwater
Surface Water
Sediment
Terrestrial Ecosystem
Aquatic Ecosystem
Semi-Aquatic Ecosystem
Receiving Domain
Air
Soil
Surface Water
Terrestrial Ecosystem
Semi-Aquatic Ecosystem
Air
Soil
Groundwater
Surface Water
Terrestrial Ecosystem
Semi-Aquatic Ecosystem
Groundwater
Surface Water
Surface Water
Sediment
Aquatic Ecosystem
Semi-Aquatic Ecosystem
Terrestrial Ecosystem
Surface Water
Aquatic Ecosystem
Terrestrial Ecosystem
Air
Soil
Aquatic Ecosystem
Semi-Aquatic Ecosystem
Terrestrial Ecosystem
Surface Water
Terrestrial Ecosystem
Air
Soil
Surface Water
The links from sending domains to sinks are not shown in Table 4-4. Sinks refer to the cells of
pollutant mass leaving the ecosystem through a reaction or physical process(es): Section 4.4
describes these processes.
4-16
-------
4.4 Fate and Transport Processes
One of the goals of the TRIM modeling framework was to develop underlying generalizations, or
"rules" for algorithms or estimation techniques. During the development of the transfer factors,
common rules underlying the development were observed and are presented in Appendix B.
These rules are based primarily on the physics and chemistry of the underlying transport proces-
ses rather than on any attribute of specific domain pairs. For example, because transport from
one cell to another always involves advection and/or diffusion processes, the mathematical form
of abiotic transport has a similar format for all domain-instance pairs.
Primary processes used to simulate pollutant movement in the abiotic domains are diffusion and
advection. These are key components of the overall transfer rates. The transport occurs both in
the gas and liquid phase for organic chemicals.
In the biotic domain, equilibrium relationships describing processes like bioaccumulation and
biomagnification were converted to a non-equilibrium form that could be used in the mass
transfer equations.
An advective process is one in which a chemical is transported within a given phase that is
moving from one cell to another (Mackay17 refers to this as a piggyback process, in which a
chemical is "piggybacking" on material that is moving from one place to another for reasons
unrelated to the presence of the chemical). Mathematically, all that is required to calculate the
advective flux is the velocity of the moving phase, and the amount of the chemical that is in the
moving phase. Examples of advective processes considered for transport of a chemical from the
soil domain to the surface water domain are erosion of surface soil, runoff from surface soil, and
recharge from groundwater.
In a diffusion process, a chemical is transported from one cell to another as a result of the magni-
tude and direction of the concentration differences between the two domain instances at the
interface between the two locations. This means that the direction of flux is not necessarily
constant with time. Estimates of effective diffusivity for a chemical species in gas and liquid-
phase diffusion were used to estimate the diffusive transfer rates.
Reaction and transformation processes are modeled using either a specified reaction/transfor-
mation rate or transformation half-life. Imall cases, the mass of chemical transformed in a given
cell is assumed to be lost from the system. To make possible a complete mass balance for the
4-17
-------
entire environmental system being modeled, this sink is modeled as an additional cell that
receives input only from the particular cell.
Reaction transformation processes include such processes as biodegradation, photolysis,
hydrolysis, oxidation/reduction, and radioactive decay. These are processes that transform a
chemical species into another chemical species; they do not involve a change of location or a
change of domain.
It is also possible that a chemical species transfers from one domain instance to another at the
same location. Possible examples include the non-equilibrium transfer of a chemical from the
fluid (liquid/gas) phases of soil to the solid phase, the uptake of a chemical by fish from water, or
the uptake of a chemical by worms from soil. These processes do not involve a change of loca-
tion or a change of chemical species. These processes are typically expressed in terms of the
half-time to equilibrium. The half-time to equilibrium is typically measured in one direction, i.e.,
from water to fish or from soil to worm.
4-18
-------
5.0 Test Case
This chapter presents data inputs and model results for various runs implemented in P4 for the
test site described in Section 4.2.4. Because P4 is a culmination of the knowledge acquired from
PI through P3 and is capable of simulating these three cases, P4 will henceforth be referred to as
the TRM.FaTE prototype.
5.1 Data Inputs
The data inputs used in this analysis are described in detail in the technical support document1. A
summary of the data inputs is provided in this section. The advective flows of media that
transport the chemical throughout the system are a critical factor in the application of the model.
Advective flows include wind, precipitation, erosion, runoff, and surface water. To realistically
model advective flows, site-specific parameters were used to the extent possible. Meteorological
data for 1 year were obtained from a nearby airport. These data indicated approximately
80 centimeters of precipitation annually, an average wind speed of 5.8 meters per second (m/s),
and a dominant wind direction toward the east. Soil properties were estimated with the
assistance of GIS maps of the area. Surface water flow rates were estimated using local river
flow data. Due to the absence of site-specific data for erosion and runoff flow, these parameters
were estimated based on reasonable assumptions. An erosion rate of approximately 1 kilogram
per square meter per day (kg/m2/day) was assumed when precipitation occurs. This value is
higher than would occur on a yearly average and is only used to demonstrate the type of results
obtained. This estimate was made based on information obtained from a data set41, which
indicated that there is 11 percent total soil loss for precipitation events less than 1.5 inches per
hour. The runoff flow was estimated as 80 percent of the hourly precipitation rate.
For most species, the population sizes were estimated using information on the density of biota
per area of habitat. For the purpose of investigating the possible impact of the biota on the
distribution of chemical, biota were assumed to be located in all but the urban cells. Wildlife
densities were assumed to be identical on forested and agricultural parcels of land.
5.2 Description of Model Runs
Using the assumed parameter values discussed previously, and those listed in the technical
support document1, numerous runs have been performed investigating the predicted behavior of
the modeled system. For brevity, selective results are presented and analyzed in this section.
The results reported here are categorized into three major divisions as follows:
5-1
-------
Theoretical phase calculations for predictive analyses (Section 5.3).
Constant meteorological conditions (Section 5.4). This category consists of
multiple runs under precipitation and no precipitation conditions
Variable meteorological conditions (Section 5.5).
To highlight key features of the model, results for constant meteorological conditions are
presented in greater detail than those obtained for variable conditions. By keeping
meteorological conditions fixed, it is easier to discern key trends and responses of the model.
For varying meteorological conditions, only the resulting apportionment of mass across the
domains are compared and summarized at this time.
5.3 Results of Phase Calculations
As previously discussed, B(a)P and phenanthrene are assumed to be released from one source, an
aluminum smelter, and initially there is no B(a)P or phenanthrene in the system. The general
phase distribution of B(a)P and phenanthrene in abiotic media if equilibrium is assumed is shown
in Table 5-1.
Table 5-1
Predicted Phase Distribution for B(a)P and Phenanthrene
in Abiotic Media for Equilibrium
Domain Type
Air
Surface Soil
Surface Water
Sediment
Sorbed
B(a)P
9.7E-1
1.0E+0
7.7E-1
1.0E+0
Phenanthrene
9.9E-1
9.9E-1
2.0E-2
9.8E-1
Dissolved
B(a)P
O.OE+0
5.0E-5
2.3E-1
1.5E-7
Phenanthrene
O.OE+0
8.5E-3
9.8E-1
2.0E-5
Vapor
B(a)P
3.3E-2
O.OE+0
O.OE+0
O.OE+0
Phenanthrene
5.5E-3
O.OE+0
O.OE+0
O.OE+0
Results presented in Table 5-1 calculated only from chemical properties of B(a)P and
phenanthrene and the assumed properties of the media, indicate that B(a)P and phenanthrene will
tend to be sorbed to solids. The main difference between B(a)P and phenanthrene concentrations
is in surface water, where almost all of the phenanthrene is predicted to be dissolved. The
fractions dissolved in sediment and surface water are approximately two orders of magnitude
higher for phenanthrene than for B(a)P.
5-2
-------
5.4 Results of Constant Meteorology Runs
Several test runs were performed to analyze the effects of precipitation and wind direction at
steady-state conditions. A total of eight runs were performed, each using a constant wind speed
of 5.8 m/s and a constant wind direction. Four of the runs did not include precipitation and had a
wind direction from either the east, west, north, or south. The additional four runs included a
precipitation scenario, with a wind direction from either the east, west, north, or south. An
emission rate of 216 grams per day (g/day) for B(a)P and 17,600 g/day for phenanthrene resulted
in different steady-state mass totals in the system for each run.
Results for runs with a due east wind direction are analyzed in detail in Sections 5.4.1 and 5.4.2
for purposes of discussing trends and model predictability relative to precipitation. A compa-
rative analysis of these runs is presented in Section 5.4.3. The results from all runs under
constant meteorological conditions are then compared and tabulated in Section 5.4.4. Separate
discussions on the ecological components are presented in Section 5.4.5.
5.4.1 Results for No Precipitation, East Wind Direction Scenario
For a due east wind direction and the given location of the source term, there is transport mecha-
nism by which the B(a)P or phenanthrene can enter the surface water if it is not raining. The
B(a)P and phenanthrene emitted in parcel I can accumulate only in the parcel with the smelter or
the parcels east of the facility (O and P). Because soil erosion or runoff is not assumed when
precipitation does not occur, the B(a)P deposited to soil can only be resuspended or flow verti-
cally through the soil layers. If resuspended, it is blown over the parcels to the east and/or out of
the system. Almost no vertical flow in soil is predicted due of the sorption properties of B(a)P.
The easterly wind flow will not bring the B(a)P over any surface water, and hence no dry depo-
sition to surface water will occur.
Spatially, as shown in Figure 5-1, the B(a)P and phenanthrene in the system are partitioned
relatively evenly among parcels I, O, and P, and there is an increase in the total mass of each
chemical as one moves east from the facility. The mass per unit area of B(a)P and phenanthrene
actually decreases as one moves east.
The presence of plants in parcel P (due to agricultural land use) is predicted to result in a magni-
fication of the B(a)P in the parcel. This behavior can be seen by analysis of select transfer factors
for parcel P (these transfer factors depend on the meteorological conditions and input parameters,
but are independent of any source terms or initial conditions assumed). Analysis of the select
transfer factors for parcel P (see technical support document1) shows that the interaction of the
5-3
-------
Figure 5-1
Predicted Steady State Spatial Distribution of B(a)P and Phenanthrene
Wind Due East, No Precipitation
Wind Direction East
8.5km
Source
|\\%| - B(a)P
xx%- Phenanthrene
Not to scale
5-4
-------
air cell with the soil cell, directly and via the plants, accounts for more than 30 percent of the
steady-state value of B(a)P in soil. This fraction is larger than the steady-state value in the plants
themselves. Plants are thus predicted to be a magnifier of B(a)P in the system. Plants them-
selves accumulate only approximately 3 percent of the total B(a)P in the system, but approxi-
mately 10 percent of the B(a)P in the system is directly due to the flux from the plant to surface
soil through litterfall. It can be seen from Figure 5-1 that 40 percent of total B(a)P is in parcel P
(mostly in soil) (Figure 5-1), 30 percent of which is accounted for by the litterfall flux from the
plants.
Table 5-2 summarizes the steady-state distribution of B(a)P and phenanthrene by domain type.
Table 5-2
Predicted Steady-State Results
(No Precipitation, East Wind Direction Scenario)
Distribution by
Domain Type
Total in System
Air
Soil
Sediment
Surface Water
Plants
Non-Plant Biota
B(a)P
Mass (g)
4.56-1-03
2.1e-i-00
1 .06+03
O.Oe+00
O.Oe+00
2.8e+01
<0.05
%
100
0.2
97.1
0
0
2.7
<0.01
Normalized
by Emission
Rate
(day)
4.66+00
9.20e-01
4.50e+00
O.OOe+00
O.OOe+00
2.00e-02
<4.6e-4
Phenanthrene
Mass (g)
3.9e+04
2.06+02
3.86+04
O.Oe+00
O.Oe+00
5.5e+02
<4
%
100
0.5
98.2
0
0
1.4
<0.01
Normalized
by Emission
Rate
(day)
2.26+00
1.0e-02
2.166+00
O.OOe+00
O.OOe+00
3.00e-02
<2.2e-4
5.4.2 Results for Precipitation, East Wind Direction Scenario
The predicted steady-state spatial distribution of chemicals is more complicated when precipita-
tion is occurring. When it is raining, there is enhanced atmospheric deposition, and erosion and
runoff transport the chemical to neighboring soil and water cells. As shown in Table 5-3, most
(98 percent) of the B(a)P is predicted to be in sediment (96 percent) and surface water (2
percent). Phenanthrene does not accumulate in sediment as much as B(a)P, and the total amount
of phenanthrene in the system, when normalized by the emission rate, is 20 times smaller than
that for B(a)P.
5-5
-------
Table 5-3
Predicted Steady-State Results (Precipitation, East Wind Direction Scenario)
Distribution by
Domain Type
Total in System
Air
Soil
Sediment
Surface Water
Plants
Other
B(a)P
Massig)
4.56+03
1.8e+00
7.66+01
4.3e+03
7.6e-i-01
1.8e+01
<0.2
%
100
0.04
1.7
96.2
1.7
0.4
<0.01
Normalized by
Emission Rate
(day)
2.0e+01
8.0e-03
3.5e-01
2.0e+01
3.4e-01
S.Oe-02
<2.0e-3
Phenanthrene
Mass
(a)
1.5e+04
1.7e+02
7.16+03
5.16+03
2.36+03
3.06+02
<0.15
%
100
1.1
47.3
33.9
15.5
2.0
<0.1
Normalized by
Emission Rate
(day)
8.0e-01
8.8e-03
3.50e-01
2.96-01
1.2e-01
1.6e-02
<8.0e-4
The spatial distribution of the chemicals is summarized in the Figure 5-2. The water bodies in M
and N receive the chemicals through erosion from parcels O and P, respectively. The waterway
in parcel L receives fluxes only through water flow from M. The water bodies north and south of
the facility, parcels H and J, have the smallest amount of both chemicals; this is because the
erosion from the parcel containing the smelter is split evenly between these two.
While much of the chemical is transported out of the system via wind or surface water outflow,
cycles of movement within the system are predictable. This result is illustrated by the fact that, at
steady-state, some mass of the chemical is predicted to be in the cells north of the facility, even
though, due to the easterly wind direction, there is no direct interchange between the cell contain-
ing the smelter (parcel I) and the cells north of it. This results from the mass-balance nature of
the model, as the cells to the north of the facility receive the chemicals through a chain of inter-
domain transfers. For example, the B(a)P emitted in parcel I is predicted to deposit in the soil
cells east of the facility (parcels I, O, and P). After deposition, the B(a)P is advected via erosion
into the waterways (H, J, M, and N), whereupon some is carried with the water flow into the
water bodies L and Q. Some of this B(a)P is then predicted to diffuse into the air column above
the water body. At this point, the chemical has been transported opposite to the wind direction,
and will be blown back across the region containing the cells north of the smelter. Once
deposited, it will undergo erosion and runoff back to the waterway H, to begin the cycle again.
Such cycling cannot be predicted by models that do not fully integrate mass-balance across
media. Environmental cycling, as discussed in Section 5.4.2, is expected under the precipitation
5-6
-------
Figure 5-2
Predicted Steady State Spatial Distribution of B(a)P and Phenanthrene,
Wind Due East, Precipitation
t
9km
T
Wind Direction East
5km
Source
\\%
-B(a)P
xx% - Phenanthrene
Not to scale
5-7
-------
scenario and is predicted by the simulation. For the no-precipitation, this is not expected or
predicted.
5.4.3 Comparative Analysis
Using normalized emission rates to compare results, both meteorological cases predict B(a)P to
accumulate in the environment more than phenanthrene. When precipitation is not occurring,
most of the B(a)P accumulates in soil. There is a significant difference when precipitation is
occurring due to accumulation in sediment. The difference between B(a)P and phenanthrene
accumulation in sediment is due to the difference in their sorption properties (Table 5-1) and the
difference of accumulation in soil is due to different half-life rates assumed (0.003/day for B(a)P
and 0.006/day for phenanthrene).
The precipitation scenario results in a higher mass accumulation of B(a)P in the system than for
the no precipitation scenario. The B(a)P mass in air, soil, and plants is higher in both magnitude
and as a fraction in the system when there is no precipitation. This result indicates the impor-
tance of washout on the amount of B(a)P contained in these media and within the system.
Phenanthrene, unlike B(a)P, is predicted to accumulate more mass in the no precipitation
scenario. The mass in air, soil, and plants is higher in magnitude when there is no precipitation.
For the precipitation scenario, both the magnitude and fraction of mass in the system increased in
the sediment and surface water.
5.4.4 Results for AH Runs - Constant Meteorology
Tables 5-4 and 5-5 show the steady-state distribution of B(a)P and phenanthrene by domain type
for different wind directions.
Qualitatively, almost all of the chemicals are predicted to be in soil or sediment, with the mass in
sediment positively correlated with precipitation. The differences in the results for different
wind directions are due to the spatial distribution of the domain types and assumed erosion and
runoff flows.
When the wind is blowing north, the forest and urban cells north of the facility accumulate the
chemicals in soil; the large fraction in soil in this case is due to the negligible erosion rates
assumed for the urban cell farthest north from the facility. The summary results for the western
and southern wind directions are similar; however, when the wind is blowing west, most of the
5-8
-------
Table 5-4
Predicted Distribution by Domain Type
(No Precipitation Scenario)
Wind Direction
Pollutant
Total Mass in Ecosystem (g)
Mass in system normalized
by emission rate
Soil
Air
Surface Water
Sediment
Plants
East
B(a)P
1000
4.6
97%
0%
0%
0%
3%
Phenanthrene
3.9E+04
2.2
98%
0%
0%
0%
1%
North
B(a)P
1100
5.1
84%
0%
0%
9%
7%
Phenanthrene
3.3E+04
1.9
95%
1%
0%
0%
4%
West
B(a)P
940
4.4
44%
0%
3%
54%
0%
Phenanthrene
1.7E+04
0.9
98%
1%
1%
0%
0%
South
B(a)P
8.8E+02
4.1
80%
0%
1%
19%
0%
Phenanthren*
2.9E+04
1.6
99%
t%
0%
0%
0%
Note: Non-plant biota is less than 0.1 percent of total mass in system.
Table 5-5
Predicted Distribution by Domain Type
(Precipitation Scenario)
Wind Direction
Pollutant
Total Mass in Ecosystem (g)
Mass in system normalized
by emission rate
Soil
Air
Surface Water
Sediment
Plants
East
B(a)P
4.5E+03
20.8
2%
0%
2%
96%
0%
Phenanthrene
1.5E+04
0.8
47%
1%
16%
34%
2%
North
B(a)P
4900
22.7
53%
0%
1%
44%
2%
Phenanthrene
1.2E-f05
6.8
95%
0%
1%
2%
2%
West
B(a)P
2500
11.6
1%
0%
4%
95%
0%
Phenanthrene
7.2E+03
0.4
41%
2%
27%
30%
0%
South
B(a)P
2500
11.6
2%
0%
3%
95%
0%
Phenanthrene
9.9E+03
0.6
52%
1%
19%
28%
0%
Note: Non-plant biota is less than 0.1 percent of total mass in system.
i
B(a)P is transported to parcels J and Q, while when the wind is blowing south, most of the B(a)P
is transported to parcels L and Q. Since parcel J is located south of parcel I, it may seem unusual
that B(a)P would be transported to parcel J when the wind is blowing due west; however, the
B(a)P is homogeneously distributed in parcel I, and the wind speed from one cell to another
depends on the angle of the boundary with respect to the wind direction.
5-9
-------
For all wind directions except east, B(a)P is predicted to accumulate in sediment even in the
absence of precipitation. In contrast, little phenanthrene accumulates in sediment unless
precipitation is occurring. This difference is due to a combination of factors. First, the estimated
transfer factors from air to water is approximately six times larger for B(a)P than for phenan-
threne. This ratio is approximately the same as the ratio of the vapor-phase fraction of B(a)P
(0.033) to that of phenanthrene (0.0055). This indicates that the diffusion to water from air is
being predicted to be an-important process. Another factor that accounts for these differences is
that the reaction rate in water is approximately four times larger for phenanthrene than that for
B(a)P. Finally, the calculated transfer factors for deposition of B(a)P to the sediment bed is more
than 30 times larger than that for phenanthrene. This is due to the predicted phase distribution of
the chemicals in the water body. Seventy-seven percent of the B(a)P is sorbed to suspended
sediment, and hence is susceptible to deposition, while only approximately 2 percent of phenan-
threne is sorbed.
5.4.5 Mass and Concentration Distribution in Biota
The mass distribution of B(a)P and phenanthrene within biota domains is based on steady-state
conditions. The steady-state distributions of B(a)P and phenanthrene for four wind directions is
shown in Tables 5-6 and 5-7 for conditions of precipitation and no precipitation, respectively. It
should be noted that the mass distribution presented in these tables represent only the biotic por-
tion of the total mass in the ecosystem. The mass in a particular biotic domain depends on the
size of the population as well as its diet. A few general terrestrial and aquatic biota trends are
discussed separately in the following paragraphs.
Most of the chemical mass is predicted to accumulate in plants when the wind direction blows
towards the parcels containing plants. The mass in the system is highest when the wind is
blowing north toward the forested areas, and lowest when blowing south toward the urban areas
without precipitation. After plants, fish and macrophytes accumulate most of the B(a)P and
phenanthrene. Relatively little of either chemical is predicted to accumulate in the terrestrial
species, although terrestrial wildlife species that consume fish (e.g., raccoon) accumulate the
most chemical mass. Note that this analysis is based on mass, not concentration. High concen-
trations of pollutants may be seen in some wildlife domains due to low population biomass.
5-10
-------
Table 5-6
Predicted Distribution in Biota by Domain Type
(No Precipitation Scenario)
Wind Direction
Pollutant
Mass total (g)
in Biota
Mass in system
normalized by
emission rate
Black-capped
Chickadee
Redtailed Hawk
Tree Swallow
Mule Dee
Blacktailed
Deer
Longtailed Vole
Longtailed
Weasel
Mink
Raccoon
Trowbridge
Shrew
Mallard
Sluegill,
Herbivore
Catfish,
Omnivore
.argemouth
Bass, Carnivore
Mice
Kingfisher
Insect
Insect, Mayfly
Macrophyte
Plant, Leaf
East
Bia^P
2.84E+01
1.32E-01
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
Phenanttirene
5.28E+02
3.00E-02
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
North
BfalP
7.36E+01
3.41 E-01
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
Phenanthrene
1.40E+03
7.95E-02
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
Wast
Bfa^P
4.27E-01
1.98E-03
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
3%
1%
69%
0%
0%
0%
0%
27%
0%
Phenanthrana
2.64E-01
1.50E-05
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
20%
1%
71%
0%
0%
0%
1%
7%
0%
s
Bia^P
1.06E-01
4.93E-04
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
3%
1%
70%
0%
0%
0%
0%
26%
0%
auth
Phenantttrane
8.73E-02
4.95E-06
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
20%
2%
71%
0%
0%
0%
1%
6%
0%
5-11
-------
Table 5-7
Predicted Distribution in Biota by Domain Type
(Precipitation Scenario)
Wind Direction
Total Mass (g)
in Biota
Mass in system
normalized by
emission rate
Black capped
Chickadee
Redtailed Hawk
Tree Swallow
Mule Deer
Blacktailed
Deer
Longtailed Vole
Longtailed
Weasel
Mink
Raccoon
Trowbridge
Shrew
Mallard
Bluegill,
Herbivore
Catfish,
Omnivore
Largemouth
Bass, Carnivore
Mice
Kingfisher
Insect
Insect, Mayfly
Macrophyte
Plant, Leaf
East
2.08E+01
9.62E-02
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
5%
0%
0%
0%
0%
2%
93%
Phenanthrana
3.03E+02
1.72E-02
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
1%
0%
2%
0%
0%
0%
0%
0%
97%
h
7.73E+01
3.58E-01
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
1%
0%
0%
0%
0%
0%
98%
orth
Phananthrtma
1.93E+03
1.10E-01
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
West
1.70E+00
7.89E-03
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
3%
1%
69%
0%
0%
0%
0%
27%
0%
Phananthrana
8.09E+00
4.S9E-04
0%
0%
0%
0%
0%
0%
0%
1%
0%
0%
19%
2%
70%
0%
0%
0%
1%
6%
0%
South
1.57E+00
7.26E-03
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
3%
1%
70%
0%
0%
0%
0%
26%
0%
Phananthrene
9.52E+00
5.40E-04
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
20%
2%
71%
0%
0%
0%
1%
6%
0%
Concentrations in Terrestrial Biota. Estimated steady-state concentrations for B(a)P and
phenanthrene in the terrestrial ecosystem are presented in Figures 5-3 and 5-4, respectively. The
concentrations do not include background concentrations of the contaminants in any media. The
wind direction in the scenario depicted is due north, and there is no precipitation. As previously
stated, the emission rate for B(a)P is 216 g/day (1,000 g at steady-state), and the emission rate for
phenanthrene is 17,600 g/day (39,000 g at steady-state). The concentration of B(a)P in surface
soil (the top 1 millimeter) is approximately six orders of magnitude higher than that in root zone
5-12
-------
Figure 5-3
Steady State Concentrations of B(a)P in Biota and Soil in a Forested
Parcel, No Precipitation
Figure 5-4
Steady State Concentrations of Phenanthrene in Biota and Soil in a
Forested Parcel, No Precipitation
5-13
-------
soil (the next 1 m) (Figure 5-3). The concentration of phenanthrene in root zone soil is substan-
tially higher than that of B(a)P, as would be expected given the higher emissions rate of the
former chemical.
Even though wildlife domains do not contribute significantly to the mass balance of PAHs,
concentrations of the chemicals in wildlife are generally within two orders of magnitude of the
concentrations in plant leaves and surface soil. The transfer from surface soil is the largest
contributor to the mass in wildlife. Because of the large difference in contaminant concentrations
in surface and root zone soil, the wildlife results are sensitive to the fraction of soil assumed to be
incidentally ingested from each of these domains. Future modifications of the model will focus
on improving the accuracy or expressing the uncertainty in the soil-to-wildlife transfers as
represented here. Because of the low concentrations of PAH in earthworms and plant roots,
these domains are not represented in the figures. For example, the estimated concentrations of
B(a)P in earthworms and plant roots are 4 x 10~16 and 4 x 10'11, respectively.
Despite the higher emission rate of phenanthrene, the estimated steady state concentrations of the
chemical in individual domains are often lower than those for B(a)P. Specifically, the concen-
trations of phenanthrene in surface soil, plant leaf, deer, vole, weasel, shrew, and mouse are
somewhat lower than those for B(a)P. The estimated concentration of phenanthrene in the insect
is approximately three orders of magnitude lower than that of B(a)P in the insect.
Concentrations in Aquatic Biota. At this point in its development, TRIM.FaTE represents
very simplified transfers between abiotic and biotic domains. Although the model is being tested
on two PAHs emitted from an aluminum smelter located in a coastal northwestern setting, the
aquatic system was assumed to be unstratified freshwater and not estuarine with complex salinity
and density regimes. It was also assumed that the freshwater bluegill population would feed
exclusively on plant matter (algae) and thus represent herbivorous creatures. In reality, these fish
are omnivorous; however, they represented a suitable species to occupy the water column
herbivore trophic level.
Table 5-8 shows the contribution to the steady-state values for the fish species in the water body
in parcel Q. Comparing the mass normalized by emission rate, more B(a)P than phenanthrene is
predicted to accumulate in all fish species. For carnivores and herbivores, the dominant uptake
pathway is through interaction with the water column for both chemicals. There is a marked
difference in the accumulated mass of each chemical for the omnivores (catfish). Most of the
5-14
-------
B(a)P is accumulated from the sediment, while most of the phenanthrene is taken up through the
water column.
Table 5-8
Uptake Fractions for Specialized Fish Domains in Parcel Q
(Precipitation, East Wind Direction Scenario)
Species
Uptake Fractions
(% of total in specialized domain)
B(a)P
Phenanthrene
Carnivore (Largemouth Bass)
Normalized mass of chemical in fish
(g/g emission/day)
Water
Herbivore
Omnivore
2.0E-3
61
38
1
5.6E-5
86
14
0.2
Herbivore (Bluegill)
Normalized mass of chemical in fish
(g/g emission/day)
Water
Macrophyte
9.7E-5
97
3
1.6E-5
100
0
Omnivore (Catfish)
Normalized mass of chemical in fish
(g/g emission/day)
Water
Sediment
Herbivore
Macrophyte
1.2E-5
1.3
99
0.1
0.1
6.8E-7
90
9
1
0
It is also important to note that the distribution within biotic domains of B(a)P and phenanthrene
predicted by the model suggests a surprisingly high percentage of PAH mass concentrated within
the carnivorous largemouth bass (Micropterus salmoides) population as opposed to the other
terrestrial and aquatic receptors (Tables 5-6 and 5-7). A detailed description of the assumptions
and algorithms adopted for the aquatic transfers are provided in the technical support document1.
A few factors contributing to the apparent imbalance in PAH distribution among biotic receptors
include diet and associated biomagnification tendencies, as well as the fact that the model
segment Q was assumed to be a large lake or slow-moving embayment with a significant bass
population of 50 to 100 individuals per hectare. While this density range is supported in the
5-15
-------
literature for slower moving waters such as lakes42, when combined with lipid level estimates and
assumptions of 80 percent of its diet being bluegills, the bass is inaccurately being predicted as the
ultimate biotic sink. Further sensitivity analysis on the model and subsequent adjustments should
result in a more accurate prediction of PAH distribution within the aquatic biota domains.
In addition, Table 5-8 indicates that the bottom-dwelling catfish uptakes 99 percent of its B(a)P
from the sediment. Given the partitioning characteristics of B(a)P and the bottom-scouring habits
of catfish, this may be fairly accurate. This uptake is driven by the assumption that 96 hours are
required to reach steady state, as described in the technical support document1. This assumption
may be inaccurate; the sensitivity of the model estimates to this assumption will be tested in
subsequent model runs.
5.5 Variable Meteorology
In this Section, the results for time-dependent meteorology are discussed. The emitted chemical
is modeled for a 24-hour period. The factors that depend on time are the wind speed, wind
direction, and precipitation. The wind direction and precipitation time profiles used in this
analysis are shown in Figures 5-5 and 5-6.
Figures 5-7 through 5-10 show the predicted mass of each chemical in various domain types for
the variable meteorological conditions. After 12 hours, each chemical is predicted to begin
accumulating to some degree in almost all domain types. An exception to this is seen for
phenanthrene in surface water. When precipitation stops (starting approximately the 19th hour),
there is no longer any erosion or runoff load to the water bodies. It can be seen in Figure 5-8 that
phenanthrene mass in SW starts decreasing marginally. At this point, the phenanthrene in the
water bodies begins to either settle into the sediment or be flushed out, eventually reaching the
outflow sink for the water domain in parcel Q.
The air domain is predicted to be most sensitive to the hourly fluctuations in meteorological
conditions. This can be seen by looking at the first few hours, where the rapidly fluctuating
profile of precipitation is reflected in the clearly observable oscillations in the chemical mass in the
air domains (Figure 5-8). When precipitation is occurring, the chemical is removed from the
atmosphere; when precipitation is not occurring, less chemical is removed, resulting in the small
"peaks" during the second and fourth hours. After the fourth hour, the precipitation rate, when
precipitation occurs, is lower than that in the first four hours. This results in smaller oscillations.
Further, the chemical in the air domains will begin cycling as the wind direction changes from
hour to hour.
5-16
-------
Figure 5-5
Wind Direction (degrees) Profile for TRIM.FaTE Prototype
Clockwise from Due North Towards Direction of Wind
rtrt
^m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time(br)
Figure 5-6
24-Hour Precipitation (mm/hr) Profile for TRIM.FaTE Prototype
Precipitation (mm/hr)
3 -i (V) 00 *> O1 0
3 8 8 8 8 8 i
H
I
1 |
1 1 1 1
1 1 llB pllll
I 1 III lllll.llll
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hme(hr)
5-17
-------
Figure 5-7
B(a)P Fraction of Mass Distribution for Parcels in TRIM.FaTE Prototype, Variable
Meteorological Conditions
0.2
0.1
6 Time(hr) 12
-A -»-B
-c -aD
-E - F
-Q H
-I »J
-K -* L
-M -N
-O IP
-Q R
-S - T
Figure 5-8
Phenanthrene Mass Distribution for Select Domains in TRIM.FaTE Prototype, Variable
Meteorological Conditions
1 OOE-05
1 OOE-06
0 OOE+00
6 OOE+00
1 20E+01
1 80E+01
2 40E+01
-Surface
Water
-Sediment
- Plants
-Fish
- Macrophyte
-Sinks
-Worm
-Birds.
mammals
Time (hr)
5-18
-------
Figure 5-9
Phenanthrene Fraction of Mass Distribution for Parcels in TRIM.FaTE Prototype, Variable
Meteorological Conditions
02
01
Time (hr)
-OIP
-Q R
-S-*-T
Figure 5-10
B(a)P Mass Distribution in Biota for TRIM.FaTE Prototype, Variable Meteorological Conditions
10 -
001
0001
00001
000001
0000001
BleckcappedChickadee
RedtaitedHawk
TreeSwallow
MuleDeerBtacitaHeSDf
r
LonQtailedVo4e
LonolBiledWeuel
Mink
Racoon
TrowbridgeShrew
Multord
Fnh. Heibivore
Fish, Omnivore
ish. Carnivore
Mice
Kingfisher
Insect
Insect. Mayfly :
Plant, Leaf
Time(hr)
5-19
-------
The plant domains are predicted to have a noticeable increase in chemical mass at the fourth hour
(Figure 5-10). This is a result of the wind blowing north for the first time, as most of the plants
are located north of the facility. The increase is more gradual for later hours. Similarly, there is
a sharp increase for the aquatic domains (water column, sediment, fish) during the first few hours
(Figure 5-10), with the mass in sediment and fish following the same general trend as that of the
water column. There is comparatively little chemical mass predicted to accumulate in the biotic
domains.
5-20
-------
6.0 Evaluation of TRIM.FaTE
As a process for iterative discovery, science does not reward advocates; it rewards those who find
truth. Many models are assembled through dialectic interactions among advocacy groups, i.e^
those who favor complex models versus those who favor simple models; those who want to
represent the highest plausible exposure versus those who want to represent central tendency;
those who favor an environmental perspective versus those who favor an industry perspective,
etc. TRIM and TRIM.FaTE aspire to a science-based model process, which requires that the goal
be to model what is real. If the reality is that exposure cannot be modeled without large uncer-
tainties, then this should be reflected in the model process. TRIM is designed to provide EPA
and other users with a tool that can be used in a flexible and iterative manner to explore human
and ecosystem exposures and provide insight in addition to risk estimates. The statistician
George Box has noted that "All models are wrong, but some models are useful"43. To make
TRIM "useful" and scientifically defensible, it has been designed with flexibility, iterative
analyses, and explicit treatment of sensitivity and uncertainty as core components of the model
building and model implementation process.
In this chapter, conclusions and evaluations based on the current work on TRIM and TRIM.FaTE
are provided. The chapter is divided into five major sections that compare the TRIM.FaTE
prototype to other multimedia models; identify the capabilities, limitations, and important
sensitivities of the current TRIM.FaTE prototype; and summarize the important conclusions that
derive both from prototype development process and from the application of the prototype to a
study site.
6.1 Comparison with Other Models (SimpleBOX and CalTOX)
To compare the results of TRIM.FaTE to other models, for illustrative purposes, two models
were applied with the same landscape and chemical input data that were used for the TRIM.FaTE
model case study. The models used for this comparison were CalTOX (Version 2.3)23 and
SimpleBOX (Version 2.0)26. Comparisons among the three models were made for the distribu-
tion of mass in multiple environmental media for the chemicals B(a)P and phenanthrene. The
descriptions of CalTOX and SimpleBOX are provided in Section 3.1.
To make the comparison tractable, and to make TRIM.FaTE results consistent with the type of
information produced by the much less complex CalTOX and SimpleBOX models, an emission
6-1
-------
rate of 9 g/day to the air compartment of the case study site was considered. A steady-state mass
distribution was obtained from each model
Figures 6-1 and 6-2 compare the results of mass distribution predictions for B(a)P and phenan-
threne obtained for similar landscape data sets using CalTOX, SimpleBOX, and TRIM.FaTE.
From Figure 6-1, it can be seen that for B(a)P, TRIM.FaTE, SimpleBOX, and CalTOX all give
similar distributions of mass in soil, water, sediment, and plant compartments. In the air
compartment, TRIM.FaTE and CalTOX produce similar results and SimpleBOX is a factor of 10
lower. The results in Figure 6-2 show that for phenanthrene, which has a higher vapor pressure
than B(a)P, all three models give similar distributions of mass in soil. For water and sediment,
TRIM.FaTE and CalTOX produce similar results. For air and plants, all three models appear to
yield a large variation in results. This variation appears to be due in large part to the differences
in air/plant uptake factors among the models. This comparison indicates that TRIM.FaTE yields
similar results to CalTOX and SimpleBOX for some media, but different results for others, based
on different algorithms. However, without actual measured concentrations in a controlled
system, it cannot be determined which model more accurately reflects reality.
6.2 Sensitivity Analysis for TRIM.FaTE
Five factors that determine the precision or reliability of an environmental transfer model are44:
specification of the problem (scenario development); formulation of the conceptual model (the
influence diagram); formulation of the computational model; estimation of parameter values; and
calculation and documentation of results, including uncertainties.
It should be recognized that there are some important inherent uncertainties in the TRIM.FaTE
multimedia approach. Parameter uncertainties and model sensitivities are addressed in detail in
the technical support document1.
At this time, only a simplified sensitivity analysis for TRIM.FaTE has been completed. The
method used considers the range of uncertainty in the parameter value and the linear elasticity of
predicted organism concentration with respect to each input parameter. This method identifies
parameters with both relatively high sensitivity and a large range of uncertainty. The method is
used to identify parameters for which decreasing uncertainty would have the largest impact on
reducing output uncertainty.
6-2
-------
Figure 6-1
Model Comparison for B(a)P
co
u
M
O)
o
0)
M
w
a
o
100.00% g
10.00% =
1.00% =
0.10% i
0.01% :
0.00%
air
soil
w ater
sediment
plants
media
QTRIM.FaTE
(SimpleBox
DCalTOX
Figure 6-2
Model Comparison for Phenanthrene
n
u
-------
The technique used in this preliminary method uses a sensitivity score, defined as:
BY Xt°
sensitivity score = x CV
3X, ' Y°
where:
dY/ dXj - change in output Y per change in input X
CV, = coefficient of variation of i* input
X,°/Y° = ratio of nominal values of input and output.
The sensitivity score was calculated for all of the inputs to the TRIM.FaTE model, and the
sensitivity to the change in inputs was determined for the following outputs: chemical concen-
trations in a carnivorous fish, macrophytes, a vole, a chickadee, and a hawk. The calculation was
made for B(a)P in a steady-state condition. The coefficients of variation used were estimated
based on both reasonable judgment and coefficients of variation developed for the California
EPA for similar parameters used in the CalTOX model. Of the 400 inputs, Table 6-1 presents
the 20 inputs that have relatively large sensitivity scores.
As discussed in Section 5.4.5, the mass distribution in the carnivorous fish (bass) was predicted
to be unusually high. The results of this analysis shows that the parameters with high sensitivity
scores for the macrophytes and fish appear to be reasonable, relative to our expectations. B(a)P
binds to particles in the air; therefore, exposure is influenced by how much B(a)P is transported
to the surface water during wet deposition, based on the wash-out ratio, thus increasing the
surface water concentration. Parameters that influence the concentration in the surface water also
have a strong effect on the results because aquatic exposure is through surface water. Decay
constants are highly uncertain because the model is sensitive to the decay content and the effects
of varying them score high. B(a)P is likely to partition into the organic carbon in suspended
sediment in water; thus, the amount of organic carbon suspended in water is an important factor.
The carnivorous fish is also dependent on sediment properties, as its food comes primarily from
he sediment. The assimilation efficiencies are highly uncertain for wild species.
6-4
-------
Table 6-1
Parameters with High Sensitivity Scores for B(a)P
Parameter
Washout- ratio
Octanol-water partition coefficient
Organic carbon partition coefficient
Decay constants in air
Decay constants in surface water
Decay constants in sediment
Decay constants in fish
Suspended sediment in surface
water
Organic carbon in suspended
sediment
Sediment organic carbon fraction
Porosity of sediment zone
Assimilation efficiencies
Accumulation factor
Fraction of lipids in fish
Fish diet
Water ingestion rate for the
chickadee
Inhalation rate of the chickadee
Water ingestion rate of the vole
Food ingestion rate of the vole
inhalation rate of the vole
Carnivorous
Fish
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Macrophytes
X
X
X
X
X
X
X
X
X
X
Chickadee
X
X
X
X
X
X
X
X
X
Vole
X
X
X
X
X
Hawk
X
X
X
X
X
X
X
The parameters with high sensitivity scores for the three terrestrial species (chickadee, vole, and
hawk) included many of the same parameters as the aquatic species. The terrestrial species are
sensitive to the chemical concentration in the surface water because they use the surface water in
6-5
-------
large part as their drinking water supply. The hawk also eats fish, whose chemical concentration
is dependent on the concentration in surface water. Additionally, the chickadee and hawk are
sensitive to the water ingestion rate and inhalation rate of the chickadee. The chickadee is
obviously sensitive to its own intake rates, but the hawk is also sensitive to them because it feeds
on the chickadee. Similarly, both the vole and hawk are sensitive to the water ingestion rate,
food ingestion rate, and inhalation rate of the vole.
The model results were found to be highly dependent on the chemical properties of the chemical
species being modeled. Nonetheless, in all cases, the model was very sensitive to source terms.
All model predictions were directly proportional to the initial inventory or input rates used. For
many applications of a model such as TRIM.FaTE, source data are variable and/or uncertain,
particularly for contaminant measurements in soils. For most chemicals, the model is sensitive to
the magnitude of the transformation rates in soils, air, surface water, and/or sediment. These rate
constants can have a large impact on the predicted persistence of any chemical species and are
often the most uncertain inputs to the model. For volatile chemicals, the model is sensitive to the
magnitude of the air-water partition coefficient. For semivolatile chemicals and inorganic
species, the model is more sensitive to the soil-water partition coefficients. Researchers typically
assume that these partition processes are linear and reversible. When this assumption is not
valid, the reliability of the model is reduced because of the uncertainties about the degree to
which soil partition processes diverge from ideal behavior. The transformation of contaminants
in the environment can have a profound effect on their potential for persistence.
6.3 Overall Capabilities
TRIM.FaTE is a model that explicitly represents time and spatial resolution by the number of
cells and links among its compartments. In descending order of reliability, the model will be
capable of handling non-ionic organic chemicals, radionuclides, fully dissociating organic and
inorganic chemicals, and solid-phase metal species. Limitations in reliability derive from
relevance and availability of data. With careful attention to inputs and selection of the approp-
riate algorithms, the mathematical structure of TRIM.FaTE can be used to model partially
dissociated organic and inorganic species. As better data and scientific understanding become
available, TRIM.FaTE will be capable of assessing such difficult-to-model agents as surfactants,
inorganic chemical species with high vapor-pressure-to-solubility ratios, and volatile metals. As
a result, TRIM.FaTE will be applicable to most chemicals of concern from a multimedia,
multipathway perspective.
6-6
-------
6.3.1 Time Scales
The TRIM.FaTE model was designed to be applied over time periods ranging from 1 hour to 1 or
more days, months, or years, when seasonally and yearly averaged partition factors apply.
6.3.2 Spatial Scales
Spatial resolution is implicitly linked to the time-step size selected. When short time steps are
selected, TRIM.FaTE can provide spatial information on scales of hundreds of meters. The
assumption that compartments are well-mixed requires that compartment dimensions be less than
the distance traveled by a chemical in one time-step. Because TRIM.FaTE is a compartmental-
type model, there are no explicit vertical or horizontal dimensions in the cells used to represent
various components of the environment.
In addition to the time-step considerations, other factors should determine the appropriate
horizontal cell size. These include: (1) resolution of input datasets and (2) similarity of habitat
(e.g., vegetation cover) and soils within a cell.
6.3.3 Chemical Classes
There are many classes of chemicals that must be addressed in environmental transport/trans-
formation models, including organic chemicals, metals, inorganic chemicals, and radionuclides.
These chemical species can also be categorized according to the physical state in which they are
introduced to the environment (gas, liquid, or solid), according to whether they dissociate in
solution (ionic or nonionic), and the charge distribution on the molecule (polar or nonpolar). The
traditional fugacity-type approach is most appropriate for nonionic organic chemicals in a liquid
or gaseous state. However, with modifications for condensation of solids on air particles, this
approach can be made appropriate for solid-phase organic chemicals. Additional adjustments
make possible the treatment of inorganic species, metals, and fully ionized organic species.
Metals (such as mercury) and inorganic chemicals with a relatively large vapor pressure pose
special problems not addressed in most multimedia models, but TRIM.FaTE provides the
potential for addressing such species. In addition, TRIM.FaTE can handle special modeling
problems, such as those that occur with mixed polarity and dissociating organic species, such as
surfactants.
6.4 Limitations
TRIM.FaTE is being designed to simulatef>ollutant movement within these complex ecosystems.
Given the complexity of processes dictating the transfer of pollutants within these systems, it
must be understood that the model's predictive capability is presently limited to gross transfers of
6-7
-------
pollutants between sources, receptors, and sinks. The model's overall predictive capabilities
depend on: (1) the explicit transfer links built into the model; (2) available databases for which it
is possible to derive distributions of parameters; and (3) the current understanding of ecological
and abiotic transfers.
The model is designed to accommodate new information in scientific understanding, so that its
precision and usefulness will improve with time. Factors that contribute to the uncertainty in
outputs of TRIM.FaTE include:
Limitations on the number of receptor species representing terrestrial and aquatic
trophic levels and the mass (number) associated with those species;
Limitations on our understanding of pollutant synergistic interactions, and their
effect on transfer, uptake, and loss rates;
Limitations on our understanding of pollutant biotransformation processes and our
ability to quantify such processes;
Limitations on our understanding of biotic interactions at the population,
community, and ecosystem level;
Limitations on our understanding of pollutant assimilation processes, as well as
depuration/egestion rates for aquatic and terrestrial receptors; and
Limitations on our understanding of population dynamics and seasonal biomass
fluctuation for certain receptor species.
Because of the complexity, the current TRIM.FaTE generates enormous amounts of output.
There are more than 1,500 links in the TRIM.FaTE prototype in which transition rates are
calculated for each time period. Each link contains important information regarding the process
being simulated. Proper evaluation of the model requires that the generated information be
explored and assessed. This information includes:
The contribution of various components to the total transition rate (e.g., diffusion
versus advection, solid phase advection versus liquid phase advection);
The contribution of intermediate processes to specific components of the transition
rate (e.g., cuticle conductance versus stomatal conductance in calculating diffusion
component of the air-to-plant transition rate); and
The partitioning of the flow of chemical through particular domains.
6-8
-------
Limited model verification has been performed to date, but more verification is needed. This
may best be accomplished with simple applications that focus on a particular subset of domains.
The driving forces in TRIM.FaTE are the flows of air, water, and solids throughout the system.
The modeled chemical(s) will be transported primarily via these mechanisms. While such flows
are considered external to the basic structure of the model, rather than part of the model itself, it
would be worthwhile to allow the inclusion of additional flow models within the model
framework. Numerous models exist for long-term flow (e.g., the Universal Soil Loss Equation
for erosion flow), but their application in a dynamic, multi-compartment context must be
carefully investigated. Currently, the flow model for air transport is also overly simplistic and
other appropriate models or algorithms need to be investigated.
6.5 Conclusions from Developmental Work on TRIM and TRIM.FaTE
Detailed single-media models, such as Gaussian-plume models, subsurface transport models, and
surface water models, exist for a variety of applications. Some multimedia models are based on
the linking of these detailed single media models; MEPAS is one example45. However, it is
extremely difficult to impose strict mass balance relationships, implement thermodynamics of
partition processes, and carry out comprehensive sensitivity and uncertainty analyses with these
"linked-model" systems15. In contrast, the Mackay-type multimedia models provide strict mass
balance relationship, use fugacity capacities (that is the capacity of a compartment to contain a
chemical on a unit volume-basis) to define the kinetics and limits of mass transfer, and provide a
tractable and scientifically defensible framework for assessing pollutant behavior in complex
systems17'31.
TRIM.FaTE was designed to fill the middle ground between the more spatially complex single-
media models and the comprehensive but often low-resolution multimedia mass-balance models.
TRIM.FaTE is a Mackay-type multimedia model based on the following criteria: it uses a series
of fully interacting compartments to represent all components of an environmental system; and it
is fully mass-balancing, and uses fugacity-based relationships to define the kinetics and
limitations of mass transfer processes.
However, unlike any Mackay-type multimedia model to date (for examples see The Multimedia
Fate Model: A Vital Tool for Reducing the Fate of Chemicals3'), TRIM.FaTE is designed to
accommodate relatively short time steps and high spatial resolution. In addition, unlike most
multimedia models, TRIM.FaTE has the capability of simulating non-reversible liquid-solid
6-9
-------
sorption processes in soil and of simulating the coupled transport and transformation of multiple
chemical species.
TRIM.FaTE currently has a spreadsheet interface with compiled FORTRAN modules used as
equation-solving routines. This arrangement facilitates sensitivity and uncertainty analyses and
makes possible the analysis of alternate algorithms for linking compartments.
6.5.1 Prototype Algorithms and Mathematical Structure
Tests of a prototype version of TRIM.FaTE indicate that the multimedia approach produces
realistic mass distributions in ecosystems containing representative air, water, soil, plant, and
animal compartments.
TRIM uses a dynamic mass-balance approach to provide estimates of the exposure and dose
profile received by selected receptors. The TRIM.FaTE module accounts for the movement of
pollutant mass through a user-defined, bounded systems model that includes both biotic and
nonbiotic (abiotic) compartments. The compartments have index addresses that represent the
spatial location, domain type, and chemical species of the pollutant. The model uses mass
balance relationships, fugacity capacities, and biokinetics to determine the movement of pollutant
mass among the compartments. A system of linked differential equations describing pollutant
mass transfer rates between pairs of addresses is at the heart of this model.
The features that make the mathematical structure of TRIM.FaTE (Chapter 3.0) unique are: (1)
its system of linked differential equations across all locations, environmental domains, and
chemical species; and (2) the estimation of transfer factors between cells based on a library of
algorithms. These features provide flexibility in defining the complexity of a simulation.
6.5.2 Input Data Needs, Verification, and Validation
Data sets needed to carry out TRIM.FaTE assessments include: chemical properties data,
including basic chemical properties and transformation rates; landscape data, including eco-
system, land use, hydrology, and climate data; and nonchemical-specific biotic parameters.
TRIM.FaTE concentration estimates are to be as spatially and temporally explicit as is feasible.
The data needed for the spatial explicitness of TRIM.FaTE will be provided by a CIS containing
readily available national or regional data sets for required model parameters such as land cover,
soil characteristics, roads, water bodies, presence and abundance of species or biomass, and
climate variables. A default set of spatial data for biotic and abiotic parameters has been
identified.
6-10
-------
7.0 Summary, Discussion, and Future Directions
The OAQPS has the regulatory responsibility for hazardous and criteria air pollutant programs
under the C AA. A broad range of risk assessments are needed to support regulatory decision
making for these programs. The OAQPS has embarked on a 4-year effort to develop a total risk
modeling system, and the components of that system, for use in assessing human health and
ecological risk and exposure in support of HAP and criteria pollutant programs under the CAA.
TRIM will provide a framework for assessing human health and ecological risks resulting from
multimedia, multipathway exposure to HAPs and criteria pollutants. The goal is to develop a
framework that is scientifically defensible, flexible, and user-friendly, and that meets the broad
range of risk assessments required under the various CAA programs and supports regulatory
decision-making for these programs. This section discusses the progress made to date on the
development of TRIM, the limitations and uncertainties identified with respect to TRIM, and
plans for future development.
7.1 Progress to Date
Progress on TRIM, to date, has been in two main areas: (1) the definition of TRIM, including the
overall conceptual design and detailed plans for development; and (2) the development of a
prototype of TRIM.FaTE, the first module within TRIM (Figure 2-1).
7.1.1 Conceptual Design of TRIM: Design Goals and Objectives
TRIM is intended to be the next generation of environmental risk and exposure models for
OAQPS. Developing a predictive environmental model of chemical transfers to human and
ecological endpoints that is flexible and applicable to both criteria pollutants and HAPs, while
incorporating multimedia, multipathway exposures, is a complex problem. To be successful,
TRIM must address the range of spatial and temporal scales, endpoints, and pathways of interest
to specific CAA programs. TRIM is expected to be a very complex model that can depict a
range of environmental and physical processes. Balance is needed in the design of the model.
One of the most critical mistakes a system developer can make is to create a system that is too
complex. Therefore, clarity must be maintained on what model processes and outputs are really
needed, how those outputs are going to be used, and how precise they need to be. A quality
systems approach demands that the endpoints be considered carefully and analyzed rigorously at
the beginning of a project so that the planning and design will satisfy the true project objectives.
As a result, efforts have been made to clearly establish the overall objectives of TRIM and to
7-1
-------
identify specific design features that can be used to measure progress and performance of both
the overall modeling system and its individual components.
TRIM consists of six individual modules (see Chapter 2.0), each addressing a major element
affecting risks associated with environmental pollutants. These six components allow for a
phased approach for development. These six modules are intended to address specific processes
affecting environmental fate and transport, exposure, and dose (absorbed and target organ). It
should be noted that there is no clear delineation between the modules. Actual boundaries may
be blurred between the components for specific applications. For example, TRIM.FaTE is
intended to estimate pollutant concentrations in various environmental compartments that serve
to define, temporally, exposures for individuals as they move through the modeled area over
time. However, TRIM.FaTE essentially is a complete model for stationary ecological receptors
(e.g., worms), incorporating exposure and uptake for these receptors. Exposure of mobile
ecological receptors to contaminants may be modeled by also using the Exposure Event Module.
Furthermore, the use of six modules allows for flexibility in both the development and
application of TRIM. Modules can be developed in a phased approach as science permits while
using standard default parameters and algorithms until more detailed information becomes
available. For example, standard assumptions regarding pollutant uptake can be used until a
model is developed or chemical-specific data become available. The development of
TRIM.FaTE, a prototype environmental fate, transport, and exposure model, demonstrates the
practicality of both phased development and flexible modular design. Specific details of how
future modules will be developed and how they will be integrated are still under development.
7.1.2 TRIM.FaTE Module Development
TRIM.FaTE demonstrates the practicality of a phased modular development, which allows
integration of the most current scientific advances. How TRIM.FaTE meets the major design
goals for TRIM of scientific defensibility and flexibility are discussed throughout the report and
are summarized in the following paragraphs.
Scientific Defensibility. TRIM.FaTE (discussed in detail in Chapters 3.0
through 5.0) represents an innovative model that addresses many of the concerns
with previous multimedia models. TRIM.FaTE is a true coupled multimedia model
rather than a linked model system. As a result, it is truly mass conserving and
allows for the estimation of both spatial and time-step analyses, thereby allowing
for non-equilibrium analysis. The case study using two organic pollutants has
demonstrated the model's flexibility in addressing the fate and transport of multiple
7-2
-------
pollutants. In general, TRIM.FaTE was developed in an iterative manner through a
series of phases where the model complexity was increased with each project phase.
Flexibility. TRIM.FaTE has demonstrated flexibility in its modular nature, spatial
and temporal resolution. Its ability to assess both human and ecological risks will
be further demonstrated by the other TRIM modules. In addition, TRIM.FaTE
estimates time series concentrations in various environmental media, which would
allow for moving (human or ecological) receptors through these components and .
allow for estimating exposure and concomitant risk to human and ecological
receptors. As previously stated, prototypes can be used individually to support
analysis of varying levels of complexity. Another major feature of TRIM.FaTE is
the algorithm library, which allows for flexibility by readily integrating future data
or scientific advances into the model as they become available.
TRIM.FaTE also appears to provide the basic foundation to support analysis in subsequent
modules. By estimating pollutant mass distributions as a function of time and space in various
environmental domains, the foundation is laid for moving human and ecological receptors in
time and space through these domains to estimate exposure. The ability to address seasonal and
other temporal dimensions of these movements also exists. Therefore, TRIM.FaTE integrates
well with the latter TRIM modules, specifically the Exposure Event Module.
7.2 Limitations and Sensitivity
One of the greatest concerns in developing TRIM'S individual modules is the complexity of the
model. Complexity of the model is of concern because it leads to: excessive computation, due
to its mathematical structure; burdensome data needs; and requirement of a sophisticated user.
A multimedia model describing the world in true detail would be highly complex. It is unlikely
that a model could be developed to simulate the real world. Whether all processes are defined is
not known, and to account for all processes would be an ambitious undertaking. Obscure and
seemingly unimportant relationships may actually provide some critical feedback mechanism
and, if not accounted for, may result in errors in long-term projections. Attempting to include a
broad range of mathematical relationships increases the potential for inconsistencies between
parameters that may not have mathematical solutions. Furthermore, with complex modeled
systems, there is a magnified importance in initialization settings for the model. With complex
systems, minor differences in initialization settings can lead to vast differences in final modeled
estimates. The lack of well-defined initialization settings (e.g., due to lack of data, definition of
background) is to be expected for any setting to be modeled and, therefore, may be a problem in
TRIM if the model is too complex. The sensitivity of the model to these effects will continue to
be addressed in future model evaluations.
7-3
-------
Complex models include a multitude of parameters. The value of such models is dependent on
the quality and quantity of data for each parameter. The model may only be as good as the
lowest quality data available for an individual parameter. Also, given the large number of
parameters, large quantities of data may be needed to run the model. TRIM.FaTE represents only
one component of TRIM and the test case application has shown the need for large data sets.
The developers of TRIM have focused attention on developing a model that is dependent on
basic physical and chemical properties of pollutants. These data are generally available in the
literature, if somewhat uncertain. However, there are numerous other data needed, some of
which may be difficult to obtain on the temporal and spatial scales required for site-specific
assessments. The success of TRIM.FaTE, future TRIM components, and the overall approach is
dependent on obtaining critical parameters and furnishing these data in a readily accessible
database. Requiring the user to collect voluminous data for every new application may prohibit
TRIM.FaTE usability. However, it is unlikely that a library of default data distributions can be
developed that would be appropriate for all applications.
Multimedia risk assessment models by nature are very complex and draw on a broad range of
disciplines (e.g., from meteorology to plant physiology in TRIM.FaTE). It is unreasonable that
any one individual has a fundamental understanding of all parameters and relationships contained
in the model. Therefore, the goal of TRIM is to develop a system that minimizes the detail to
which a user would be required to make judgements. To this end, the model and possible
structures must be fully evaluated to determine the appropriate detailed relationships and the user
would be required only to establish general relationships. Well-studied generic settings may also
prove useful. The model must be designed to prevent the user from proposing invalid
relationships or scenarios. Therefore, a detailed evaluation of the TRIM.FaTE (and future
modules) structure, uncertainty, and sensitivity is essential to determine the appropriate model
structure and level of user defined options. For example, TRIM.FaTE could be designed to allow
the user to be responsible for defining all linkages, or could be set up to develop detailed linkages
based on generic relationships defined by the user.
One of the key limitations in multimedia models is the lack of model validation opportunities.
Ideally, detailed environmental compartment monitoring data would be available for comparing
model results to monitoring data. However, very limited data exist in a detailed enough manner
to allow for evaluating environmental concentrations specific to a single modeled source on
regional scales. As a result, model-to-model comparisons are the best validation processes
available, although the magnitude of how they describe reality is not fully known. Comparisons
of TRIM.FaTE to other modules is further complicated by the fact that it is a truly coupled
7-4
-------
multimedia modeling system. TRIM.FaTE has been shown to describe processes of pollutant
movement that appear reasonable based on the understanding of environmental systems (i.e., the
re-emission from water and deposition on the forested land).
Current model development has focused on site specific applications or generic settings. A
detailed analysis has not been made that evaluates the contribution of site-specific data to the
overall variability and uncertainty of risk estimates. If site-specific data can be shown to be a
small relative contributor to overall uncertainty as compared to more ubiquitous parameters, the
use of generic environmental settings may be feasible. Generic settings would minimize the need
for user input and judgement. Regardless of whether this is the case, there may also be the
question of establishing initialization (baseline) settings for any model application.
The uncertainties and limitations associated with TRIM.FaTE may apply to other also specific
modules (e.g., data, mathematic complexity, user sophistication). There may be other overall
limitations that are applicable to TRIM. Future modules of TRIM may not be supported by
current science and their development may have to be in stages. For example, physiologically-
based pharmacokinetic (PBPK) models are currently being developed and validated on a
chemical-specific basis. Whether current science allows for generic PBPK models is not known.
Likewise, the dose-response module is limited by current agency policy and guidance for
evaluating risk to noncarcinogens, focusing on a threshold hazard quotient approach as opposed
to true dose-response. Therefore, progress may be limited on several of the future modules.
However, it should be noted that these limitations or delays may only be in the near future, and
that the design of TRIM (i.e., modular design) allows for readily integrating the most current
science as it becomes available.
Given the complexity and limitations of multimedia modeling, some questions have been raised
regarding applicability of multimedia risk assessment results. The Science Advisory Board
(SAB), in their review of the Mercury Report to Congress46, has stipulated that, given the
uncertainties associated with such multimedia models, their accuracy is questionable and,
therefore, can only be used as a qualitative comparative measure of risk and not for quantitative
purposes. Therefore, until TRIM can be validated and detailed sensitivity and uncertainty
analysis be conducted, its usefulness will most likely be limited to qualitative analyses.
7.3 Future Development
TRIM is intended to be developed in a phased approach. A total of six modules have been
identified for development. Once these are completed, risk assessments can be conducted in a
7-5
-------
comprehensive manner. However, defaults assumptions or scenarios can be used in their place
until all modules are completed. As a result, the usefulness of TRIM is not contingent upon
completion of all modules. Each module only increases the overall confidence of the risk
estimates and allows for evaluating variability and uncertainty in greater detail. For example, if
uptake and biokinetic models are not available, concentration response relationships could be
used to support the risk assessment.
7.3.1 Overall TRIM Development
To maximize the usefulness of TRIM, the phased approach will be emphasized, and the focus of
these efforts will be module development and platform integration. Modules will be developed
according to their perceived overall importance and the current status of applicable science. In
addition, to facilitate the integration of individual modules, computer platforms will be evaluated
to determine which is most appropriate for addressing the existing modules and for directing
future module development. The schedule proposed for further TRIM module development and
release of TRIM is as follows:
Fiscal year (FY) 1998: SAB review of TRIM approach and TRIM.FaTE
TRIM.FaTE refinement (see Section 6.6)
Exposure event model prototype
TRIM computer framework design
FY 1999: SAB review of exposure event model
Risk characterization module prototype
Other TRIM modules
FY 2000 and beyond: TRIM beta testing.
Release of TRIM for public use.
TRIM was intended as a 4-year effort. It is unlikely that the complete TRIM modeling system, as
conceptualized, will be completed in that time frame. Rather, efforts will focus on maximizing
the utility of the overall modeling system. It is anticipated that refinement of the TRIM.FaTE
module will be completed, together with prototypes for the exposure-event model and risk
characterization model by FY 1999. It is anticipated that these prototypes will be evaluated and
improved, and that a complete TRIM prototype will be available for beta testing in FY 2000.
7.3.2 TRIM.FaTE Development
TRIM.FaTE currently consists of multiple prototypes under refinement. Development continues
to focus on refinement of TRIM.FaTE and features development of a version programmed in
7-6
-------
optimized computer language and in a form that can be easily used and widely distributed.
Future development and testing of TRIM.FaTE will be focused on four main areas:
Expansion of ecological algorithms and features
Uncertainty and sensitivity analysis
Model verification and validation
Database development.
Expansion of Ecological Algorithms and Features. One of the key areas of emphasis in
the future development of TRIM.FaTE will be improvement of the ecological links in the model.
Several new algorithms will be added to the model, and the parameterization of existing
algorithms for ecological receptors will be improved. Additional algorithms will be required for
many chemical transfers to allow the fate of inorganic chemicals to be modeled.
Distributions or ranges of several wildlife parameters will be added to the database for
TRIM.FaTE, including ingestion and inhalation rates, elimination rate constants (or half-lives)
for particular chemicals, and population densities for particular ecosystems. Data contributing to
distributions of lipid levels in fish will be collected. Algorithms for chemical transfers to fish
and benthic organisms will be parameterized for additional representative species. The creation
of new wildlife domains will not be a priority in the near-term; representative species have been
selected for all major food-web groups.
Domains for seeds and fruits (and perhaps wood and bark) will be created, and all plant cells will
be linked. A domain for algae will be created. Adjustments of algorithms and parameters based
on plant taxa (e.g., waxy versus non-waxy leaves) will be considered. Spatial data sets for
parameters such as leaf-area index and vegetation biomass will be obtained.
Seasonal processes, such as litterfall, plant growth and senescence, crop harvesting, migration of
wildlife, wildlife food habits, and winter sleep will be incorporated into TRIM.FaTE, to the
extent possible. Other changes in biomass, such as organism and population growth and death,
will be incorporated into the model, either through growth functions or through changes in
monthly biomass input parameters. The movement and reproduction of wildlife and fish must be
represented in a simplified manner, since individual-based models for many species would not be
feasible.
Model outputs will be compared to empirical data, to the extent possible. Assumptions will be
based on the best available science. At this time, for example, it is clear that the masses of
7-7
-------
contaminants estimated in wildlife, earthworms, and plant roots are very sensitive to the choices
of the soil depth interval from which incidental ingestion or accumulation of chemicals in soil
occurs.
Significant effort will be focused on estimating uncertainty bounds for all half-life terms, both in
biota and abiotic media.
7.3.3 Exposure Event Module
Work on the Exposure Event Module is ongoing. This module will be used to translate pollutant
sources (indoor and outdoor) into quantitative estimates of the amount of contaminant that comes
in contact with the human-environment boundaries, that is, the lungs, the GI tract, and the skin
surface of individuals within a specified population. Human exposures within TRIM.FaTE will
be modeled through the use of "cohort" models within reference indoor/outdoor environments.
Each cohort will consist of a group of persons with similar physical and demographic
characteristics who follow a common activity pattern. The activity pattern of each cohort will
consist of a realistic, time-ordered series of exposure events. Each exposure event will be
defined by a start time, an end time, and a list of compartments in the study area to which the
cohort is exposed during the event (e.g., air and soil associated with the indoor domain). An
uptake equation specific to each compartment in the list is applied to the current pollutant
concentration of the compartment to determine the potential dose received by the cohort. The
total dose received during the exposure event is the sum of the doses received from the individual
compartments.
7.3.4 Development of Uncertainty and Sensitivity Analysis Capabilities
An important component of the TRIM program in the next 2 years will be to focus on uncertainty
and sensitivity analyses. Initially, focus will be on parameter uncertainty. As previously
discussed, each model input can be entered as a distribution such that Monte Carlo simulations
can be completed to examine the range of possible outputs. Although parameters can be entered
as distributions at this time, the methods for constructing and entering parameter distributions
need to be optimized such that Monte Carlo simulations can be processed more efficiently.
Methods for conducting and evaluating model uncertainty analysis need to be better defined,
carried out, and evaluated. The current approach is to select among alternative algorithms from
the algorithm library for each transport and transformation process. By looking at the change in
7-8
-------
resulting output, the effects of model uncertainty and the influence of each process can be
evaluated. Although this can be done manually in the TRIM framework, a system to automate
this process needs to be developed.
The system to evaluate spatial and temporal variability needs to be automated. Changing the
spatial step size or temporal step size can change the results of the model. Evaluating the effect
of this change helps the researcher decide what scale is appropriate for their needs.
Ultimately, TRIM will provide a framework that facilitates probabilistic analyses, including the
use of correlation, rank correlation, or regression to examine the degree to which outcome
variance is attributable to particular inputs or assumptions, and helping the user to better focus
data gathering efforts. Additionally, completing an uncertainty analysis facilitates the
comparison of model predictions to limited multimedia environmental data. Implementing
output ranges and parameter uncertainty dependence will help the model validation stage of
TRIM.FaTE.
7-9
-------
8.0 References
1. U.S. Environmental Protection Agency (EPA), 1997, Development of the Total Risk
Integrated Methodology: Draft Technical Support Document for TRIM.FaTE,
December.
2. National Research Council (NRC), 1994, Science and Judgment in Risk Assessment,
National Academy Press, Washington, DC.
3. Commission of Risk Assessment and Risk Management (CRARM), 1997, Risk
Assessment and Risk Management in Regulatory Decision-Making, Final Report,
Volume 2.
4. U.S. Environmental Protection Agency (EPA), 1986, Guidelines for Carcinogenic Risk
Assessment, Federal Register 51:33992-34003.
5. U.S. Environmental Protection Agency (EPA), 1986, Guidelines for Mutagenicity Risk
Assessment, Federal Register 51:34006-34012.
6. U.S. Environmental Protection Agency (EPA), 1986, Guidelines for the Health Risk
Assessment of Chemical Mixtures, Federal Register 51:34014-34025.
7. U.S. Environmental Protection Agency (EPA), 1986, Guidelines for Exposure
Assessment, Federal Register 51:34042-34054.
8. U.S. Environmental Protection Agency (EPA), 1995, Policy for Risk Characterization.
9. U.S. Environmental Protection Agency (EPA), 1995, Policy for Use of Probabilistic
Analysis in Risk Characterization.
10. U.S. Environmental Protection Agency (EPA), 1997, Guiding Principles for Monte
Carlo Analysis, Office of Research and Development, Washington, DC, EPA/630/R-
97/001.
11. U.S. Environmental Protection Agency (EPA), 1997, Guidance on Cumulative Risk
Assessment.
12. Mozier, J. W. and T. R. Johnson, 1996, Evaluation of Existing Approaches for
Assessing Non-Inhalation Exposure and Risk with Recommendations for
Implementing TRIM, Contract No. 68-D-30094, Work Assignment 2-9, Prepared by IT
Corporation for EPA, April 1996.
13. Whicker, F. W. and T. B. Kirchner, 1987, PATHWAY: A Dynamic Food-Chain Model
to Predict Radionuclide Ingestion After Fallout Deposition, Health Phys. 52, 717-737.
8-1
-------
14. Thibodeaux, L. J., 1979, Chemodynamics, Environmental Movement of Chemicals in
Air, Water, and Soil, John Wiley and Sons, New York.
15. Thibodeaux, L. J., 1996, Environmental Chemodynamics: Movement of Chemicals in
Air, Water, and Soil, 2nd ed., New York: J. Wiley & Sons.
16. Mackay, D., 1979, Finding Fugacity Feasible, Environ. Sci. Technol. 13 , 1218-1223.
17. Mackay, D, 1991, Multimedia Environmental Models: The Fugacity Approach, Lewis
Publishers, Chelsea, Michigan.
18. Mackay, D. and S. Paterson, 1981, Calculating Fugacity, Environ. Sci. Technol. 15,
1006-1014.
19. Mackay, D. and S. Paterson, 1982, Fugacity Revisited, Environ. Sci. Technol. 16, 654-
660.
20. Cohen, Y. and P. A. Ryan, 1985, Multimedia Modeling of Environmental Transport:
Trichloroethylene Test Case, Environ. Sci. Technol. 9, 412-417.
21. Cohen, Y., W. Tsai, S. L. Chetty, and G. J. Mayer, 1990, Dynamic Partitioning of
Organic Chemicals in Regional Environments: A Multimedia Screening-Level
Approach, Environ. Sci. Technol. 24, 1549-1558.
22. McKone, T. E. and D. W. Layton, 1986, Screening the Potential Risk of Toxic
Substances Using a Multimedia Compartment Model: Estimation of Human
Exposure, Regul. Toxicol. Pharmacol. 6, 359-380.
23. McKone, T. E., 1993, CalTOX, A Multimedia Total-Exposure Model for Hczard.ius-
Wastes Sites Part I: Executive Summary, Lawrence Livermore National Laboratory,
Livermore, California, UCRL-CR-111456PtI.
24. McKone, T. E., 1993, CalTOX, A Multimedia Total-Exposure Model for Hazardous-
Wastes Sites Part II: The Dynamic Multimedia Transport and Transformation Model,
Lawrence Livermore National Laboratory, Livermore, CA, UCRL-CR-111456Ptn.
25. McKone, T. E., 1993, CalTOX, A Multimedia Total-Exposure Model for Hazardous-
Wastes Sites Part HI: The Multiple-Pathway Exposure Model, Lawrence Livermore
National Laboratory, Livermore, CA, UCRL-CR-111456PtHI.
26. Brandes, L. J., H. den Hollander, and D. van de Meent, 1997, SimpleBOX, v 2.0,
National Institute of Public Health and the Environment (RIVM), the Netherlands.
27. U.S. Environmental Protection Agency (EPA), Office of Health and Environmental
Assessment, 1990, Methodology for Assessing Health Risks Associated with Indirect
8-2
-------
Exposure to Combustor Emissions, Interim Final, EPA/600/6-90/003, Washington DC,
January.
28. U.S. Environmental Protection Agency (EPA), 1993, Addendum to Methodology for
Assessing Health Risks Associated with Indirect Exposure to Combustor Emissions,
Office of Health and Environmental Assessment, External Review Draft, EPA/600/AP-
93/003, Washington DC, November.
29. Rice, G., R. Hetes, J. Swartout, Z. Pekar, and D. Layland, 1997, Methods for Assessing
Population Exposures to Combustor Emissions, Presentation at the 1997 Society for
Risk Analysis Annual Meeting, Washington, DC.
30. van de Meent, D., 1993, SIMPLEBOX: A Generic Multimedia Fate Evaluation Model,
National Institute of Public Health and Environmental Protection (RIVM) Report No.
672720 001, Bilthoven, The Netherlands.
31. Cowan, C. E., D. Mackay, T. C. J. Feijtel, D. van de Meent, A. Di Guardo, J. Davi, and
N. Mackay, 1995, The Multimedia Fate Model: A Vital Tool for Predicting the Fate of
Chemicals, Society of Environmental Toxicology and Chemisty, SET AC Press,
Pensacola, Florida.
32. van de Water, R. B., 1995, Modeling the Transport and Fate of Volatile and Semi-
Volatile Organics in a Multimedia Environment, M. S. Thesis, University of California,
Los Angeles.
33. Buck, J. W., G. Whelan, J. G. Droppo, Jr., D. L. Strenge, K. J. Castleton, J. P. McDonald,
C. Sato, and G. P. Streile, 1995, Multimedia Environmental Pollutant Assessment
System (MEPAS) Application Guidance, Guidelines for Evaluating MEPAS Input
Parameters for Version 3.1, PNL-10395, Pacific Northwest Laboratory, Richland,
Washington.
34. Odum, H. T., 1983, Systems Ecology, An Introduction, John Wiley and Sons, Inc.
35. Thomann, R. V. 1989, Bioaccumulation Model of Organic Chemical Distribution in
Aquatic Food Chains, Environmental Science Technology, Vol. 23, pp.699-707.
36. Radhakrishnan, K. and A. C. Hindmarsh, 1993, Description and Use ofLSODE, the
Livermore Solver for Ordinary Differential Equations, LLNL UCRL-ID-113855.
37. Hindmarsh, A. C., 1983, ODEPACK, A Systematized Collection of Ode Solvers,
Scientific Computing, R. S. Stepleman, et al. (eds.), North-Holland, Amsterdam, pp.
55-64.
38. Barrodale, I. and G. F. Stuart, 1981, ACM Transactions on Mathematical Software,
September.
8-3
-------
39. Maddalena, R. L., T. E. McKone, D. W. Layton, and D. P. H. Hsieh, 1995, Comparison
of Multi-Media Transport and Transformation Models: Regional Fugacity Mode vs.
CalTOX, Chemosphre, 30, 869-889.
40. Hunsaker, C. and A. Simcock, 1997, Draft CIS and Spatial Data Report for the Total
Risk Integrated Model (TRIM), Oak Ridge National Laboratory, Oak Ridge, Tennessee.
41. Purdue University, web site, nhp://AGEN521 .WWW.ecn.purdue.edu/AGEN521/
epadir/erosion/ raindrops.html.
42. Maceina, M., W. Wrenn, and D. Lowery, 1995, "Estimating Havestable Largemouth Bass
Abundance in a Reservoir with an Electrofishing Catch Depletion Technique," North
American Journal of Fisheries Management, 15(1) 103-109.
43. Box, G. E. P. and G. C. Tiao, 1973, Bayesian Inference in Statistical Analysis, Wiley
Classics, New York, New York.
44. International Atomic Energy Agency (IAEA), 1989, Evaluating the Reliability of
Predictions Made Using Environmental Transport Models, Safety Series 100,
International Atomic Energy Agency, Vienna.
45. Whelan G., J. W. Buck, D. L. Strenge, J. G. Droppo, 1992, Overview of the Multimedia
Environmental Pollutant Assessment System (MEPAS), Hazardous Waste & Hazardous
Materials, V9N2:191-208.
46. U.S. Environmental Protection Agency (EPA), 1997, Mercury Study Report to Congress,
Vol. I-Vffl, EPA-452/R-97-003 through EPA-452/R-97-010.
8-4
-------
APPENDIX A
GLOSSARY
-------
Glossary
Advection - Process in which a chemical is transported within a given moving phase that is
moving from one cell to another. Calculation of advective flux requires velocity of and amount
of chemical in the moving phase. MacKay 1991 refers to advection as the "piggyback" process,
in which a chemical "piggybacking" on material moving from one place to another for reasons
unrelated to the presence of the chemical.
Atmospheric Half-Life - The time required for one-half of the quantity of an air pollutant to
react and/or break down in the atmosphere.
Bioaccumulation - Progressive increase in amount of chemical in an organism or part of an
organism that occurs because the rate of intake exceeds the organism's ability to remove the
substance from the body.
Bioconcentration - Same as bioaccumulation; refers to the increase in concentration of a
chemical in an organism.
Biological Half-Life - The time required for the concentration of a chemical present in the
body or in a particular body compartment to decease by one-half through biological processes
such as metabolism and excretion.
Carcinogenic - Able to produce malignant tumor growth. Operationally, most benign tumors
are usually included also.
Cell - A uniquely defined address, within the computer code, which accounts for all potential
locations of mass within the ecosystem. The three indices that make up an inventory address in
TRIM.FaTE are volume element, domain, and species. There is no concentration gradient within
an inventory address; the chemical is uniformly mixed. Also referred to as inventory address.
Clean Air Act of 1990 - This amendment to the Clean Air Act of 1970 contains several
provisions requiring the EPA to evaluate effects to humans and the environment caused by
exposure to hazardous air pollutants and criteria air pollutants.
A-l
-------
Cohort Study- A study of a group of persons sharing a common experience (e.g., exposure to
a substance) within a defined time period: this experiment is used to determine if an increased
risk of a health effect (disease) is associated with that exposure.
Compartmental Systems Model- A model that is represented by a series of inventory
addresses, each with a state variable, which interact through transfer factors. In this model, the
transport of multiple pollutant species in a multimedia region is set up as a mass exchange among
a set of systems used to represent spatial locations, collections of environmental phases, and
chemical species.
Criteria Pollutant - Six common pollutants, used as indicators of air quality, regulated by EPA
on the basis of human and/or environmental adverse effects.
Dermal Uptake - Absorption through the skin membrane.
Diffusion - Movement of a chemical substance from areas of high concentration to areas of low
concentration. Biologically, diffusion is an important means for toxicant deposition for gases
and very small particles in the pulmonary region of the lungs.
Dispersion Model - A mathematical model or computer simulation used to predict the
movement of airborne pollution. Models take into account a variety of mixing mechanisms
which dilute effluents and transport them away from the point of emission.
Domain - The domain refers to the composition of material in which the chemical is dissolved,
sorbed, or otherwise held. What distinguishes one domain from another at a given location is the
requirement that all phases of.a single domain must attain equilibrium within a single calculation
time step. There is a hierarchical system of domain, i.e., different levels of domain. At a more
general level, a domain is a collections of volume elements, such as all of the root zone soil. For
example, an instance of the domain is the root zone soil within a volume element.
Dose - The amount of chemical absorbed by an organism usually expressed as mass of
substance per unit body weight of organism per unit time.
Dose Response - Determination of the magnitude of toxic response to dose.
A-2
-------
Dry Deposition - A transfer process from air to soil. Dry deposition velocity is the ratio of
contaminant flux (mol/[m2-h]) to contaminant concentration in air (mol/m3). -
Environmental Fate The destiny of a chemical or biological pollutant after release into the
environment. Environmental fate involves temporal and spatial considerations of transport,
transfer, storage, and transformation.
Equilibrium - The state in which opposing forces are exactly counteracted or balanced. Types
of equilibrium include acid-base, colloid, dynamic, homeostatic, and chemical. Used in risk
assessment of toxic air pollutants to generally describe the chemical equilibrium between a
pollutant in the inhaled air and the level in the body.
Exposure Assessment - Measurement or estimation of the magnitude, frequency, duration
and route of exposure of animals or ecological components to substances in the environment.
The exposure assessment also describes the nature of exposure and the size and nature of the
exposed populations, and is one of four steps in risk assessment.
Hazardous Air Pollutant - Any air pollutant listed pursuant to Section 112(b) of the Clean
Air Act of 1990. Those pollutants known or suspected to cause serious health problems.
Half-Life - See atmospheric half-life and biological half-life. Also, the period of time
characteristic of a radionuclide in which one-half of the activity has decayed.
Indices - There are three indices used in the inventory address and state variable used as a
tracking and auditing system to clarify the volume element, domain, and species in the
TRTM.FaTE framework.
Indirect Exposure Methodology- This methodology sets out procedures for estimating the
indirect (i.e., non-inhalation) human exposures and health risks that can result from the transfer
of emitted pollutants to soil, vegetation, and water bodies. This methodology is not a
comprehensive environmental audit, but is best regarded as an evolving and emerging process
that moves the EPA beyond the analysis of potential effects on only one medium (air) and
exposure pathway (inhalation) to the consideration of other media and exposure pathways.
Ingestion Uptake - Intake via the mouth, with transfer to the GI tract.
A-3
-------
Inhalation Uptake - Intake via the nose and mouth, with transfer to the lungs
Inventory Address - A uniquely defined address, within the computer code, which accounts
for all potential locations of mass within the ecosystem. The three indices that make up an
inventory address in TRIM.FaTE are volume element, domain, and species. There is no
concentration gradient within an inventory address; the chemical is uniformly mixed. An
informal synonym for this term is "cell."
Links - If mass moves without first moving through intervening cells, the two cells are
considered linked. Each linkage is associated with and algorithm determining the direction and
rate of mass flow between the two cells.
Multi-Pathway Exposure - Exposure through inhalation, ingestion, and adsorption routes.
Microenvironment - The immediate local environment of an organism.
Model - A mathematical representation of a natural system intended to mimic the behavior of
the real system, allowing description of empirical data, and predictions about untested states of
the system.
Models-3 - EPA's third generation of air quality management/modeling system to be used as a
tool for decision making by federal, state and the industry environmental analysts. It is a layered
modeling system providing various kinds of services at various levels of complexity. The most
important layers are the User, Management, Computational Modeling, and Data Access layers.
Multimedia Contamination - Contamination in air, water, soil, and food.
Pharmacokinetics - The field of study concerned with defining, though measurement or
modeling, the absorption, distribution, metabolism, and excretion of drugs or chemicals in a
biological system as a function of time.
Phase - The building blocks of which everything else is composed (gas, liquid, solid, lipid, and
other material). Each domain consists of multiple phases.
A-4
-------
Reaction or "R" Factor-Used for chemical reactions or transformation processes. Used for
transfer from one species to another (e.g., radioactive decay), or for transfers out of the system
(e.g., chemical degradation to a chemical not being tracked). As an example, consider an address
that represents a soil layer contaminated with trichloroethylene (TCE). This address will be at
the same location and represent the same domain, but differs in the last entry from the address
that represents the inventory of vinyl chloride (VC), which is a decay product of TCE. Both
addresses are at the same location and domain, but TCE must undergo a transformation to move
from the address with the TCE/soil layer to the VC/soil layer address.
Reaction Transformation - Process that transforms a chemical species into another chemical
species, but does not include a change of location or domain. Biodegradation, photolysis,
hydrolysis, oxidation/reduction, radioactive decay, etc., are reaction transformation processes.
Risk Assessment - The scientific activity of evaluating the toxic properties of a chemical and
the conditions of human exposure to it in order both to ascertain the likelihood that exposed
humans will be adversely affected, and to characterize the nature of the effects they may
experience. May contain some or all of the following four steps:
Hazard Identification - The determination of whether a particular chemical is or is
not causally liked to particular health effect(s).
Dose-Response Assessment - The determination of the relation between the
magnitude of exposure and the probability of occurrence of the health effects in
question.
Exposure Assessment - The determination of the extent of human exposure.
Risk Characterization - The description of the nature and often the magnitude of
human risk, including attendant uncertainty.
Risk Characterization - The final step of a risk assessment, which is a description of the
nature and often the magnitude of human risk, including attendant uncertainty.
Risk Management - The decision-making process that uses the results of risk assessment to
produce a decision about environmental action. Risk management includes consideration of
technical, scientific, social, economic, and political information.
A-5
-------
Sink - A domain that receives input from a domain, but does not output the chemical. Once a
chemical enters a sink, its mass is not tracked in the model; it is a final "resting place" for the
chemical.
Species - Chemical compound state/phase (i.e., form of the chemical and phase of the
chemical-physical properties of the chemical) This index is useful for representing radioactive
decay or chemical decay, (e.g., TCE to vinyl chloride).
State Variable - Values that describe the state of the system as a function of time in the various
components of the modeled system.
Transfer Factors - The rate at which a chemical will be transferred from one inventory address
to another inventory address for a given mass in that inventory address. The units of this factor
are fraction of total inventory per unit time. When a transfer rate factor is multiplied by an
inventory expressed as mass, we obtain from this product the mass transferred per unit time.
There are three types of transfer factors, "T" factor and "R" factor, for transport, reaction, and
exchange, respectively. It should be noted that the magnitude of transfer rates do not define the
magnitude of the transfer because if there is very little mass in a cell, there will be very little
transferred.
Transformation - Alteration of a chemical substance from one chemical form to another
through a chemical, physical, or biological reaction.
Transport or "T" Factor - This transfer rate can be used in several ways. First, it can be used
for transport from one location to another. This is further broken down into transport processes
involving the change of location by advection or by diffusion. At a minimum, there will be a
change in the first index, the spatial location for this type of process, i.e., diffusion in the soil.
Sometimes, there will be both a change of domain and spatial location, i.e., diffusion from air to
soil. An alternate use is for transfer from one domain to another at the same spatial location, in
which case the second index changes. For example, if we decide that sediment particles should
not be in equilibrium with water, then the water and the particles would have to be separate
domains at the same volume element and a "T" factor would be used to express the exchange
between these phases.
A-6
-------
Transport - To move or be conveyed from one place to another. In the context of
environmental contamination, a containment is transported from one location to another by
dispersion, advection (e.g., wind), or diffusion(e.g., dilution in air) processes.
Volume Element - An entity characterized by a total spatial volume (m3) completely enclosed
by a contiguous surface. This surface may be shared by one or more neighboring volume
elements. The volume element has a unique spatial location, which can be defined with a set of
x-y-z coordinates. The volume is occupied by all domains at this location. For example, if there
are two domains at a volume element, the sum of the volumes of these two domains must sum
to the volume of the volume element. When two domains occupy the same volume location,
these two domains are assumed to be well mixed.
Wet Deposition - Transfer process from air to soil. Occurs during precipitation and is
proportional to the rate of precipitation (i.e., rain in m/h), but differs in both the relative
magnitude and nature between particles and gas-phase chemicals.
A-7
-------
APPENDIX B
ALGORITHM GENERALIZATIONS
-------
APPENDIX B
ALGORITHM GENERALIZATIONS
One of the goals of the TRIM modeling framework is to develop underlying generalizations,
estimation techniques or "rules", for application of algorithms. During the development of the
transfer factors for the prototypes, common rules were observed. These rules were functions of
the physics and chemistry of the transport processes rather than the domains. For example,
abiotic transport from one cell to another has the same mathematical form for all domains.
Some of these rules were refined after study of their documentation in the literature (McKone,
1996, Mac Kay, 1992). This appendix documents the underlying rules for use in subsequent
prototypes in order to simplify algorithm development. These rules were observed as a
consequence of building the transfer factors for the different abiotic domains and are presented
before the domain specific algorithms because the rules are common across all the abiotic
domains.
B. 1 Multiple Phase Calculations
This section describes how multiple phases within a domain are modeled in P4. Phases
considered in P4 are liquid, gas, and solid and are assumed to be at chemical equilibrium.
Because chemical equilibrium is assumed, the ratios of the concentrations in the individual
phases are constant, and mass balance need only be tracked for the total amount of the chemical
in all phases in a cell. The amount of chemical in the cell in a particular phase can be determined
from the total amount in the cell (this is described below). It is possible that, in later prototypes,
chemical equilibrium will not be. assumed, in which case the amount of chemical in different
phases will need to be tracked as separate cells.
In any cell, the total amount of chemical in a given cell is made up of the sum of the amounts in
the different phases:
"" mount n g
,-, water., waer .-, so,,
+ L V C K
Nt ""= Amount in gas phase + Amount in aqueous phase + Amount in solid phase
i water.. water .-, solid,. solid
where:
Total _ totaj amount of chemical in domain/cell, units of g [chemical]
B-l
-------
C,gas = concentration of chemical in gas phase in domain/cell, units of g [chemical]/
m}[gas] in domain ~~
Vf" = volume of gas in domain/cell, units of m3[gas] in domain
£*>aifr _ concentration of chemical in aqueous phase in domain/cell, units of
g [chemical]/m3[water] in domain
V*ater = volume of aqueous matter in domain/cell, units of volume [aqueous] in
domain
(solid _ concentration of chemical in solid phase in domain/cell, units of
g [chemical]/m3[solid] in domain
yiond _ voiume of soijd jn domain/cell, units of m3[solid] in domain
If it is desired that the units of N*olal be in units of moles [chemical], then the preceding equation
must be multiplied by the molecular weight of the chemical (which has units of moles
chemical]/g [chemical])
Since chemical equilibrium is assumed, the ratios of the concentrations are constant. However,
care must be used in specifying what the units of the concentrations are. This is because, in
practice, it is more common to define notation for ratios of concentrations on a mass basis other
than that of mass by volume basis.
B. 1.1 Normalization to Liquid Phase
This section describes the relevant formulas when the concentrations are normalized to the
concentration in the liquid phase. This normalization is utilized in P4 for all soil, surface water,
and sediment cells. Using the equilibrium assumptions, we have that:
,
f~, solid _ . ,, _. _, water
= (HIRT)C"a"r
where:
- density of solid phase in cell, units of kg [solid phase ]/m3'[solid phase]
Kd = equilibrium partition coefficient; ratio of concentration in solid phase (units of
kg[chemical]/kg[solid phase]) to that in liquid phase (units of
kg I'chemical)'/Liters[liquid phase])
Cf = 10'3 m3/L, conversion factor to convert m3 [liquid phase] to Liters [liquid
phase]
H = Henry's law constant for chemical, units of Pa-m3/mol
R = ideal gas constant, 8.314 m3-Pa/mol-K
T = temperature, units of degrees K
B-2
-------
Applying these relationships to the general equation in the beginning of this section yields:
Total = c»aJ H_ygas + y water + solid}
i ' I ny ' ' "solid a f i I
The volumes of each phase in the domain can be expressed as fractions of the total volume of the
cell, in which case the above equation yields:
»i Total _ ,-, water,, Total
Tf gas ,, water y solid
H ' + - + P.-^C-i--
d J yTotal
where:
,Total _ i7gas ,,water ,,solid
j - V + Vj + Vj
The term CITola'=N^c"al^om' is the total concentration of the chemical in the cell. Using the
assumed equilibrium relationships, the concentrations in the individual phases can be recovered
from the total amount of mass in the cell, as follows:
c.
,T Total ,, , Total
* r
,,
'
solid
gas
ftf ,,Total ,,Total
/^ wate r .
C* .
RT '
Tola,
(HIRT) N.
y
n Y i
r Solid
Total ,,Total
i * i
£ _^j
P solid d f Total
C
solid
r r*\ j.r
solid KdC}>N.
y
Total,,, Total
T.
water
RT yT«'al y
Total
p ,., K, Cf
"solid a t
solid
solid d f jomi
B-3
-------
For cases in which the concentration in the water phase is negligible (e.g., when domain is the
atmosphere, or the chemical has a very low solubility), the concentrations must be normalized to
another phase.
The preceding equation can be simplified by using the notation of fugacity. The fugacity
capacities for the pure phases are defined by Mackey, 1991.
= 1'H
' 1-J \~f I J /V I l^<-^
solid " solid a f water
The total fugacity capacity Z,Tl"a' for the cell is defined as:
rr gas ,, water ,, solid
r-r total _ 'i 'i "i
z =z . + z + z ,.j
' airvTo,al wa'er ,.Total sohd
i '/
phase,
'-'
phase ,
phase, 2
where phase is either the solid, liquid, or gas phase.
Applying these relationships shows that:
Z N'"'"' Z
,-, water _ water i _ water ^ total
' y total ,, Total total '
Zi Vi A
T'"at
Z
^
total
Z N T'
^gas_ oir 1Tf
(_, . - . -- (^
,-r total ,. Total 7 total
*-i Vi Li
7 M T»tat f
solid _ ^solid i _ ^solid ^ total
"~
~ total ,. Total total '
B-4
-------
where C,T<"al is the total concentration of the chemical in the cell (units of g [chemical]/m3[total
cell]. From these relationships, we have that, in general, the amount of mass in the different
phases is given by:
7 fJ °'a 7 7
>, water _, , water.-, water _ ,, water water i _ , 7 water water water ^ total
' ' ' ' total ,, Total ' total total '
,,
"
Total
7 \T
\r gas _,~,gasy gas _ y gas air i _ ,7gas gas gas £ total
' ' ' ' 7 total ,, Total ' 7 total 7 total '
Li Vj Z, Z,-
ry \j Total i-j
j, solid _.~, solid,, solid _ y solid 'solid i _-,7 solid 'solid solid .-. total
' ' ' ' ' '
total T r Total ' total total '
Vi
where N""1", N,ga3, and TV,10'"' are the mass in the water, gas, and solid phases, respectively.
In the following sections, these general equations are applied to the soil and surface water
domains. These "applications" involve only adhering to notation commonly used in the literature
for the different media.
B.1,2 Application to Soil, Surface Water, and Sediment Domains
For soil, surface water, and sediment domains, the concentrations are normalized to the
concentration in the water phase, and the same notation is used to represent the relevant
parameters. In a soil cell, the solid phase consists of the soil particles. In a surface water cell, the
solid phase consists of the sediment suspended in the water column. In a sediment cell, the solid
phase consists of the benthic sediments. Following common practice, the volume fractions of
each phase are denoted as follows:
, . water
11 - =6,.
TT total
'
V,
total '
»f
'i
solid
total
B-5
-------
where:
0, = water
, = gas
l-e,-,.= 1-4),
where <# is the total porosity of the cell (= 6, + e). The equations for the total mass of chemical
in the cell and in the different phases are then given by:
H
and the total fugacity for the cell is given by:
For the groundwater, surface water, and sediment domains, the volume fractions of the gas phase
(£,) are assumed to be zero.
In P4, the partition coefficients, soil - water, Kd in each cell (soil, surface water, and sediment)
are calculated in a manner applicable for nonionic organic chemicals (Karickhoff 1981, as cited
in California Department of Toxic Substance Control's model [CalTOX] 1993, p.25) by:
where:
Koc = organiB-carbon partition coefficient
f,,c = fraction of organic carbon in the cell/domain
B. 1.3 Multi-Phase Partitioning in the Air Domain
Since the volume of water in an air domain is so small relative to the volume of the solid and the
gas phase, there has not been a historical development of AT/s (i.e., ratio of concentration in solid
B-6
-------
phase to that in dissolved phase) for the atmosphere, although the concept still applies. Instead,
only the solid and gas phases are usually addressed. If chemical equilibrium is assumed-between
the phases, then a normalization other than to the liquid concentration is required. In an air cell,
the solid phase consists of the paniculate matter in the atmosphere.
In P4, the volume fractions of chemical in each phase are given by:
r^7=zVP«fr«
where:
DL = atmospheric dust load in air cell, kg[aerosol]/m3[air cell]
Pdus, = density of aerosols, kg[aerosol]/m3 [aerosol]
The dust load and aerosol density are specified in P4. In order to normalize to either the gas or
solid phase, the equilibrium ratio of the concentrations in the two phases must be estimated. In
P4, the fraction of the contaminant bound to particles, denoted by
-------
C
3oltd
The total mass of chemical in the air cell is then:
Total
c,
solid
The fugacity capacity in the solid phase can be determined by use of the relationship below (see
Section 2.1.1).
solid air~
solid
gas
=z
The total fugacity in the air cell is then given by:
B-8
-------
y gas TT solid
, total_ "i _ "i
'' ~ airy Total + solid y Total
=Zair(l-DL/pJ + ZsolidDL/pdust
B. 1.4 Calculation of the Fraction of Contaminant Bound to Aerosol
In P4, the fraction of contaminant bound to paniculate in the air cell, denoted by T
VP - {
|exp[6.79(7V/r-l)] if
B.2 Converting Equations with Equilibrium Relationships to Dynamic Form
In the course of converting equations to a form that is suitable for use within the intended
framework, it is possible to convert some algorithms that represent steady-state equilibrium
relationships into time-dependent ones. This can be performed if an estimate of the time required
for the concentration to reach some fraction of the steady-state value is available. In particular, if
B-9
-------
the concentration in one domain/cell C, is related to the concentration in another domain/cell C2
by an equilibrium relationship of the form C}= K C2, where K is unknown and it is known that it
takes time ta in order to reach 100a% of the steady-state value when C2 is approximately
constant, then we have that:
where K2 and K are defined below:
Indeed, the solution of the above differential equation with initial condition C,(0)=0 is given by:
The steady-state solution is C,(t)= (k/k2) C2, and so we have that K=k/k2. This assumption that
100a% of the steady-state value is reached at time ta means that:
l-£
Solving for k2 yields:
When K2 is determined, K[ = K2 from which the general result follows.
In P4, this conversion is performed only for the xylem, stem and root cells of the plant domain.
B.3 Advective Processes
An advective process is one in which a chemical is transported within a given phase that is
moving from one cell to another (MacKay 1991 refers to this as a piggyback process, in which a
chemical is "piggybacking" on material that is moving from one place to another for reasons
B-10
-------
unrelated to the presence of the chemical). All that is required to estimate the advective flux is
the velocity of the moving phase, and the amount of the chemical that is in the moving phase. In
general, the advective flux in a given phase (e.g.,attached to particles, or dissolved in water) from
cell i to cell; is given by:
Advective flux from cell i to cellj= (Volume of phase that moves from cell i to cell j per unit time)
x (Amount of chemical in phase per volume of phase in cell i)
or
N.(t) xffrhase)
Advective flux Cell i-'Cell j = Q(phase) x
Vfohase)
xN,(0
where:
Q (phase) = volumetric flux of phase from cell i to cell;, m3[phase]/day
N,(t) = amount of chemical in cell / at time t, moles chemical]
f,(phase) = fraction of chemical in cell / that is in the moving phase, moles chemical in
phase]/moles chemical in cell /],
Vfphase) = volume of phase that is in cell /, m3[phase],
T,_j'(phase) = transition probability for advective flux from cell i to receiving cell, /day,
given by:
-a*, , . Q(phase) xf (phase)
7,_ (phase) = :
; V (phase)
This formula for the transition probability is valid for all advective processes from one cell to
another, and does not rely on the fugacity concept. Application of the concept of fugacity shows
that (see Section 2.1).
Z (phase) V.(phase)
f (phase) = x
Z.(Total) V (Total)
where:
B-ll
-------
Z,(phase) = fugacity capacity for moving phase, mol/m3[phase]-Pa ^
ZJTotal) = total fugacity capacity for celli, mol/m3 [sending cell /]-Pa
VJTotal) = total volume of cell i (sum of volumes of each phase in cell), m3[cell i]
Applying this shows that the fugacity-based form for the transition probability for advective flux
is:
,«*,,. . Q (phase) xZfrha
,(phase) =
' J
v (phase) x Ar x.Zfphase)
VffotatyxZ/JotaT)
In most applications, the volumetric flow rate Q(phase) of the phase is calculated as the product
of a relevant area (A,j) and the volumetric flow rate per unit area, or a flow velocity (v,j). Usually
the relevant area is the interfacial area between the sending and receiving cells, but this is not
always the case; e.g., erosion from surface soil to surface water is usually reported in units of
mass[soil]/area[soil layerj-time, in which case the relevant area is the area of the surface soil
layer. Table B-l summarizes the implementation of all volumetric flows for domains in P4.
These flows are discussed in more detail in the sections describing the specific domains.
B.4 Diffusive Processes
The net flux from diffusion from one cell to another cell depends on the difference in the
concentrations in the two cells. This means that the direction of flux is not necessarily constant
with time. However, it is possible to derive the general form for diffusion from one cell to
another and then break up defining net diffusion into a part proportional to the mass in one cell
and a part proportional to the mass in another cell. Although it is not known beforehand which is
the "sending" cell, if one is dealing with a fixed cell, then these terms take the mathematical form
of a "sending" component to the other cell and a "receiving" component from the other cell.
In all cases, diffusion across compartment boundaries is modeled in P4 as a two-resistance model
through the boundary layers on either side of the domain interface (as discussed in CaJTOX, Vol.
n, pp 36-41). This is first done in a general manner that simplifies the presentation, as
characterization of diffusion between two cells reduces to specifying an interfacial area between
the two cells and specifying the algorithm for calculating the mass transfer coefficient for a
particular domain.
B-12
-------
In the two-resistance model for molecular and turbulent diffusion, the mass transfer between a cell
/ to cell j depends on mass transfer through two distinct layers: the boundary in cell / and the
boundary in cell/ It is assumed that the net flux is equal on both sides of the boundary between the
two cells. This flux is assumed to be proportional to the difference in the bulk concentration in the
cell and the concentration at the cell-side of the boundary. The constant of proportionality has units
of m/day, and is called the mass transfer coefficient. Determination of the mass transfer coefficient
depends on the domain type that the cell is, and in some cases on the domain type of both cells.
The general form for the net diffusive flux between two cells is given by:
\u
Net diffusive flux between cell i and cell j = Fr - A A IJ
cf-c;
c;-c}b
B-13
-------
Table B-1
Summary of Volumetric Advective Flow Velocities Considered in Prototype 4
Source/
Sending Cell
Soil
Receiving
Cell
Soil
Air
Moving
Phase
Liquid
Gas
Solid
Description of
Phase Velocity
Precipitation driven
percolation
Gas Discharge
Resuspension
Units
m3[water]/day
m3[gas]/day
m3[soil]/day
Calculated or Specified in Prototype 4
Specified
= A*Darcy_Liq
where:
A = Area of soil-soil interface, m2
DarcyJJq = Darcy velocity of water in sending soil cell,
m3[water]/m2[area]-day. i
Specified
= A*Darcy_Gas
where:
A = Area of soil-soil interface, m2
Darcy_Gas = Darcy velocity of gas in sending soil cell, j
m3[gas]/m2[area]-day. i
It is assumed that volumetric flow of particles from soil is the same as
that to soil. Volumetric resuspension rate is then
= Vol. Flow TO soil = A*vd * pA/ pp
where:
A = Area of soil-soil interface, m2[area]
vd = Dry deposition velocity of particles, m/day
pA = Atmospheric dust load in air domain (concentration of dust in air),
kg[particles]/m3[atmosphere]
pp = Density of air particles, kg[particles]/m3[particles) \
-------
Table B-1
Summary of Volumetric Advective Flow Velocities Considered in Prototype 4
Source/
Sending Cell
Soil/Groundwater
Soil/Groundwater
Air
Receiving
Cell
Surface Water
Surface Water
Soil and Surface Water
Moving
Phase
Solid
Liquid
Solid
Liquid
(Vapor?)
Description of
Phase Velocity
Erosion
Runoff
Recharge
Wet & Dry
deposition of
particles
Wet deposition of
liquid
Units
m3[soil]/day
m3[water]/day
m3[water]/day
m3[particles]/day
|
Calculated or Specified In Prototype 4 '
Calculated from mass-based areal erosion rate and soil density:
= A'E/ps
where:
A = Area of soil layer, m2
E = erosion rate to surface water, kg [soil]/m2[area]-day
ps = density of eroding soil, kg[soil]/m3[soil]
= A'Runoff
where:
A = Area of soil layer, m2
Runoff = Amount of runoff that reaches waterbody per units area of
watershed, m3[water]/m2[area]-day
=A*Recharge
where:
A = Area of soil-surface water interface, m2
Recharge = Volume of water flow per unit area of interface,
m3[water]/m2[area]-day
Calculated:
= A'vd * pA/ pp
where:
A = Area of soil layer, m2
vd = Dry deposition velocity of particles, m/day
pA = Atmospheric dust load in air domain (concentration of dust in air), '
kg[particles]/m3[atmosphere]
pp= Density of air particles, kg[particles]/m3[particles]
\
Not implemented
-------
Table B-1
Summary of Volumetric Advective Flow Velocities Considered in Prototype 4
Source/
Sending Cell
Receiving
Cell
Moving
Phase
Description of
Phase Velocity
Units
Calculated or Specified In Prototype 4
Air or Air Advection
Sink
Total
Wind advection
m3[air]/day
= A*u
where:
A = Area of air-air interface, m2
u = Wind velocity from sending to receiving air cell, m/day
Air
Plant Leaf
Solid
Wet deposition of
particles
m3[particles]/day
Calculated:
=AT (wa, * pA/ Pp) * fL
where:
A = Area of soil layer containing plant
I = Interception fraction (see Section 7 for description of algorithm)
vd= Deposition velocity of particles, m/day
PA= Atmospheric dust load in air domain (concentration of dust in air),
kg[particles]/m3[atmosphere]
pp = Density of air particles, kg[particles]/m3[particles]
(L = Fraction of deposition adhering that is taken up by plant
Surface Water
Sediment
Solid
Sediment
deposition
m3[suspended
sedimentyday
River to river
Total
River flow
m3[air]/day
Calculated:
-A'Sdep/Pss
where
A = Area of surface water-sediment interface, m2
S^ = Deposition rate of suspended sediment to sediment bed,
kg[suspended sediment]/m2[area]-day
pss = Density of suspended sediment, kg[suspended
sediment]/m3[suspended sediment]
= A*ur (
where:
A = Area of river parcel interface, m2
ur = River velocity from sending to receiving river cell, m/day
-------
Table B-1
Summary of Volumetric Advective Flow Velocities Considered in Prototype 4
Source/
Sending Cell
Sediment
Receiving
Cell
Surface water advection
sink
Surface Water
Moving
Phase
Total
Solid
Description of
Phase Velocity
Outflow
Sediment
resuspension
Units
m3[water column]/day
m3[benthic sediment]/day
'
Calculated or Specified in Prototype 4
Calculated so that net flux of water into surface water is 0:
= (Runoff'ASSoll + Recharge* ASW.GW + Inflow) + (Rain - EV)*A
where:
A = Area of surface water, m2
Runoff = runoff from surface soil, m3[water reaching surface
water]/m2[surface soil area]-day
ASSoll = Area of surface soil, m2[surface soil area]
Recharge =Recharge from groundwater to surface water,
m3[water]/m2[interface]-day
ASW GW = Area of surface water-ground water interface, m2[interface]
Rain = Precipitation rate, m3[water]/m2[surface water area]-day
EV = Evaporation rate, m3[water]/m2[surface water area]-day
Inflow = inflow of water to water body cell, m3[water]/day
Calculated:
= A'SrasuS(,/pbs
where:
A = Area of sediment-surface water interface, m2
S,esusp = Resuspension rate of benthic sediment to water column,
kg[benthic sediment]/m2[area]-day
pbs = Density of benthic sediment, kg[benthic sediment]/m3[benthic
sediment]
-------
Table B-1
Summary of Volumetric Advr re Flow Velocities Considered in Prototype 4
Source/
Sending Cell
Receiving
Cell
Moving
Phase
Description of
Phase Velocity
Units
Calculated or Specified in Prototype 4
Sediment burial sink
Solid
Sediment burial
m3[sediment]/day
Calculated so that amount of sediment buried is equal to maximum of 0
and amount deposited minus amount resuspended:
= A*max/-0, Sdep/pss - Sresus(/pbs}
where:
A = Area of sediment-surface water interface, m2
Sdep = Deposition rate of suspended sediment to sediment bed,
kg[suspended sediment]/m2[area]-day
pss = Density of suspended sediment, kgfsuspended
sediment]/m3[suspended sediment]
SIBSUSP= Resuspension rate of benthic sediment to Water column,
kg[benthicsediment]/m2(area]-day
pbs = Density of benthic sediment, kg[benthic sediment]/m3[benthic
sediment]
-------
Bibliography
Agency for Toxic Substances and Disease Registry (ATSDR), 1993, lexicological Profile for
Polycyclic Aromatic Hydrocarbons (PAHs), U.S. Dept. Health Human Services, Atlanta,
Georgia, 273 pp.
Alexander, G. R., 1977, Food of Vertebrate Predators on Trout Waters in North Central Lower
Micron, The Michigan Academician, 10: 181-195.
Bacci, E., D. Calamari, C. Gaggi, and M. Vighi, 1990, Bioconcentration of Organic Chemical
Vapors in Plant Leaves: Experimental Measurements and Correlation, Environ. Sci. Technol.
24:885-889.
Bacci, E., M. J. Cerejeira, C. Gaggi, G. Chemello, D. Calamari, and M. Vighi, 1990,
Bioconcentration of Organic Chemical Vapors in Plant Leaves: The Azalea Model,
Chemosphere 21:525-535.
Baes, C. F., R. D. Sharp, A. L. Sjoreen and R. W. Shor, 1984, A Review and Analysis of
Parameters for Assessing Transport of Environmentally Released Radionuclides through
Agriculture, ORNL-5786, Oak Ridge National Laboratory, Oak Ridge, Tennessee.
Barber, M. C., L. A. Suarcz and R. R. Lassiter, 1988, Modeling Bioconcentration ofNonpolar
Organic Pollutants by Fish, Environmental Toxicology and Chemistry, Vol. 7, pp. 545-558.
Batzli, G. O., 1977, Population Dynamics of the White-Footed Mouse in Flood Plain and
Upland Forests, Am. Midi. Nat., 97: 18-32.
Beall, M. L. and R. G. Nash, 1971, Organochlorine Insecticide Residues in Soybean Plant
Tops: Root vs. Vapor Sorption, Agronomy J. 63:460-464.
Belfroid, A., M. van den Berg, W. Seinen, J. Hermens and K. van Gestel,. 1995, Uptake,
Bioavailability, and Elimination ofHyrophobic Compounds in Earthworms (Eisenis andrei)
in Field-Contaminated Soil, Environ. Toxicol. Chem., 14(4):605-612.
Belfroid, A., M. Sikkenk, W. Seinen, K. van Gestel and J. Hermens, 1994, The Toxicokinetic
Behavior of Chlorobenzenes in Earthworm (Eisenis andrei) Experiments in Soil, Environ.
Toxicol. Chem., 13:93-99.
Belfroid, A., J. Meiling, D. Sijm, J. Hermens, W. Seinen and K. van Gestel, 1994, Uptake of
Hydrophobic Halogenated Aromatic Hydrocarbons for Food by Earthworms, Arch. Environ.
Contam. Toxicol., 27:260-265.
Bevan, D. R. and M. R. Ulman, 1991, Examination of Factors That May Influence Disposition
of Benzo(a)pyrene in Vivo: Vehicles and Asbestos, Cancer Lett, 57: 173-179.
C-l
-------
where:
Ftj
C'°'al =
Cj""al
= net diffusive flux from cell i to cell j, mol/day _
= interfacial area between cells i andy through which diffusion occurs, m2
= mass transfer coefficient for combined turbulent and molecular diffusion on "i"
side of boundary between cells i andy, m/day ( ={ mol/m2 [area]-day}
/{mol/m3[cell/]})
= mass transfer coefficient for combined turbulent and molecular diffusion on "j"
side of boundary between cells i andy, m/day ( ={ mol/m2[area]-day}
/{mol/m3 [celly]})
total concentration of chemical in cell i, mol/m3[cell z]
= total concentration of chemical in cell i at the boundary with celly but in cell /,
mol/m3[cell /]
= total concentration of chemical in celly, mol/m3[celly]
= total concentration of chemical in celly at the boundary with cell i but in celly,
mol/m3[cell /]
In order to derive the general form for diffusion, the concept of fugacity is applied. Dividing the first
equation by Z/°'a/ and the second by Z/0'"' yields:
total
c:
to,al
, total
total
Fu
J
total
If it is assumed that, at the boundary between the cells, the fugacities on both sides of the boundary
are equal; i.e., iffl'=C'/Zl""al=C/ZJ'n""=fJ', where/ is the fugacity of a cell, then:
> total
F.
, total
total
total
Solving for Ft) shows that:
F =
~ total
L/
-7 total
Nt A
Z V Z
1 1 J
~ total
J
~? lutal
^J )
/
V
J
(
{ \Z>
1
A ^ total
1
total j j
uu A
i
A.Z^'U^
\
total j,
U J J' /
-1
B-19
-------
where:
U..
N,
V,
r total
= net diffusive flux from cell i to cell j, mol/day
= mass transfer coefficient for combined turbulent and molecular diffusion on "i"
side of boundary between cells / andy, m/day ( ={ mol/m2[area]-day}
/{mol/m3[celli]})
= mass transfer coefficient for combined turbulent and molecular diffusion on "j"
side of boundary between cells j andy, m/day (={ mol/m2[area]-day}
/{mol/m3[celly]})
= amount of chemical in cell /, moles chemical]
= total volume of cell /, m3[cell i]
= total fugacity capacity for cell /, moles chemical]/m3[cell t]-Pa
The general form for the transfer factors can now be derived. In particular, the differential equations
for the amount of chemical in cells i and j are:
dN I N N.
\Y(ij)A. + other gains and losses
totaly ztolaty \ IJ
dt
, total, , rj total,,
'i Vi Li Vj
Y(iJ) A. + other gains and losses
-i
. Rearranging shows that:
dt
dt
- N
total*
+ other gains and losses
y('t/M,
- N: + other gains and losses
, lutal,.
From these equations we see that the general form for the transition probability for diffusive
transport from cell i to cell; is:
*ff _
f total, ,
"i i
7 total,, _ total,
-i U,j Ls (
-^ total,,
A v,
J'/
B-20
-------
where:
T*ff = transition probability for diffusive transport from cell i to celly, /day ._
U,j = mass transfer coefficient for combined turbulent and molecular diffusion on "i"
side of boundary between cells / andj, m/day (={ mol/m2[area]-day}
/{moym3[celli]})
Ujj = mass transfer coefficient for combined turbulent and molecular diffusion on "j"
side of boundary between cells i and j, m/day ( ={ mol/m2 [area]-day}
/{mol/m3[celly]})
Vt = total volume of cell i, m3 [cell /]
Z'"'"1 = total fugacity capacity for cell i, moles chemical]/m3[cell f]-Pa
This general form is used to model diffusive transport in P4. All that must be determined for each
such diffusive link between cells are the mass transfer coefficients and the interfacial area between
the cells through which diffusion occurs.
Table B-2 summarizes how the mass transfer coefficients are estimated for all diffusive transfers
considered in P4. These are discussed in more detail in the sections describing the specific domains.
B.5 Reaction and Transformation Processes
In P4, reaction and transformation processes are modeled using either a specified reaction/
transformation rate or chemical half-life for each cell. In all cases, the mass of chemical transformed
in a given cell is assumed to be lost from the system into a sink cell which receives input only from
the particular cell. For a given cell / and an associated sink, the transition probability T"nk that
represents the transformation/reaction process is simply the specified or calculated reaction/
transformation rate.
B-21
-------
Table B-2
Summary of Diffusion Mass sfer Coefficients Considered in Prototype 4
Sending Cell
Air
Surface Water
Air
Receiving
Cell
Surface
Water
Air
Soil
u..ndm».>R.c.ivinB (Mass transfer coefficient, m/day)
= D"'a,,/Saw
= 0.00316 * WindSpeed * (18/MW)"2, if WindSpeed>0.5 m/s'
= 0.00162 * WindSpeed * (18/MW)"2, if WindSpeed<0.5 m/s
where:
D'^MZ./Z^.KDJ
Da,r = Chemical diffusivity in air, mZ/day
6aw = Boundary layer thickness in air above surface water, m
Windspeed = Wind speed ; m/day
MW = Molecular weight of chemical, g/mol
Specified (0.24 m2/s)
= De".i,/6aw
where:
De''alr2 = (Zalr/Z10,alair2)(Dair)
Dalr = Chemical diffusivity in air, m2/day (Specified)
Sas = Boundary layer thickness in air above soil, m (Specified)
Reference
Southworth(1979)
See p. 42 of CalTox
manual
CalTox p. 41
'The general form of this equation includes a term for the velocity of the surface water. In Prototype 2.0 we are assuming thai this velocity is 0
!This is also based on a slow current velocity.
-------
Table B-2
Summary of Diffusion Mass Transfer Coefficients Considered in Prototype 4.0
Sending Cell
Soil
Soil
Surface Water
Receiving
Cell
Air
Soil
Sediment
U..ndin9->R«eivina (Mass transfer coefficient, m/day)
= De,,/5sa
where:
Def, = Effective diffusivity in soil, m2/day
= Zwale/Z,0,a,)Dwate,e'°'3/ct>2 +
(Zac/Z10,al)Dalre1(V3/(t)2
where:
Zwa,e, = Fugacity capacity for water, mol/m3-Pa
ZM = Fugacity capacity in air,mol/m3-Pa
Z,0,al = Total fugacity capacity for sending soil layer, mol/m3-Pa
Dwale, = Diffusion coefficient in pure water, m2/day
Dai, = Diffusion coefficient in pure air, m2/day
8 = Volume fraction of water in sending soil layer, m3[water]/m3[soil layer]
e = Volume fraction of air in sending soil layer, m3[air]/m3[soil layer]
= Total void fraction in sending soil layer, 4>= 0 + e
= Boundary layer thickness in sending soil with air, m.
= 0.108 ' D6ll(ss)0229(for surface soil)
= 318.4 * Den(I2)06B3(for root zone)
= Depth of layer / 2 (for vadose zones 1 and 2)
Same as for Soil -> Air
- De" / 5
u walei ' uwd
where:
no" -17/7 \ n
^ waler \*-waler *-lotal surface waler/ ^ water
Dwaler = Diffusion coefficient in water, m2/day
6wd = Boundary layer thickness in surface water above sediment, m. Specified.
Reference
CalToxvol.il, p.35-36,
based on Jury et al.
(1983)
CalTox
CalTox, p.36, based
on Jury et al. (1983).
\
-------
APPENDIX C
BIBLIOGRAPHY
-------
Beyer, W. N., E.Conner and S. Gerould, 1994, Estimates of Soil Ingestion by Wildlife,J.
Wildl., Mgmt. 58: 375-382.
Boersma, L., C. McFarland and F. T. Lindstrom, 1991, Mathematical Model of Plant Uptake
and Translations of Organic Chemicals: Application to Experiments, J. Environ. Qual.
20:137-146.
Boese, B. L., 1984, Uptake Efficiency of the Gills of English Sole (Parophrys vetulus) for Four
Phthalate Esters, Fish. Aquat. Sci., Vol. 41, pp. 1713-1718.
Briggs, G. G., R. H. Bromilow, A. A. Evans, and M. R. Williams, 1983, Relationships between
Upophilicity and the Distribution of Non-ionised Chemicals in Barley Shoots following
Uptake by the Roots, Pestic. Sci. 14:492-500
Briggs, G. G., R. H. Bromilow and A. A. Evans, 1982, Relationship between Lipophilicity and
Root Uptake and Translocation of Non-ionised Chemicals by Barley, Pestic. Sci. 13:495-504.
Bromilow, R. H. and K. Chamberlain, 1995, Principles Governing Uptake and Transport of
Chemicals, pp. 37-68, In S. Trapp and J. C. McFarlane, eds., Plant Contamination: Modeling and
Simulation of Organic Chemical Processes, Lewis Publishers, Boca Raton.
Calder, W. A. and E. J. Braun, 1983, Scaling of Osmotic Regulation in Mammals and Birds,
Am. J. Physiol., 224: R601-R606.
California Environmental Protection Agency, 1993, CalTOX: A Multimedia Total Exposure
ft '-'for Hazardous-Waste Sites, Technical Reports, Department of Toxic Substances
C trol, The Office of Scientific Affairs, December.
Capel, P. D., C. Leuenberger, and W. Giger, 1991, Hydrophobic Organic Chemicals in Urban
Fog, Atmos. Environ. A, 25:1335-1346.
Chamberlain, A. C., 1970, Interception and Retention of Radioactive Aerosols by Vegetation,
Atmos. Env. 4:57-78.
Chang, L. H, 1943, The Fecal Excretion ofPolycyclic Hydrocarbons following their
Administration to the Rat, J. Biol. Chem., 151: 93-99.
Computer Sciences Corporation Systems Engineering Division and Ogden Professional Services,
1995, ECOTOX - ECOTOXicology Database System, A User's Guide, Prepared for the Office
of Research and Development, U.S. Environmental Protection Agency, Mid-Continent Ecology
Division, Duluth, Minnesota, June.
Connell, D. W., 1990, Bioaccumulation ofXenobiotic Compounds, CRC Press, Boca Raton,
Florida.
C-2
-------
Connell, D. W, 1990, Bioconcentration ofLipophilic and Hydrophobic Compounds by Aquatic
Organisms, Bioaccumulation of Xenobiotic Compounds, pp. 97-144. -
Connell, D. W, 1990, Biomagnification ofLipophilic Compounds in Terrestrial and Aquatic
Systems, Bioaccumulation of Xenobiotic Compounds, pp. 145-185.
Connell, D. W. and R. D. Markwell, 1990, Bioaccumulation in the Soil to Earthworm System,
Chemosphere, 20(1-2):91-100.
Connell, D. W. and G. Schuurmann, 1988, Evaluation of Various Molecular Parameters as
Predictors of Bioconcentration in Fish, Ecotoxicology and Environmental Safety, Vol. 15, pp.
324-335.
Creelman, L. W., V. Pesinova, Y. Cohen and A. M. Winter, 1994, Intermedia Transfer Factors
for Modeling the Environmental Impact of Air Pollution from Industrial Sources, J.
Hazardous Materials 37:15-25.
Dahl, A. R., D. S. Coslett, J. A. Bond, and G. R. Hesseltine, 1985, Metabolism of
Benzo(a)pyrene on the Nasal Mucosa of Syrian Hamsters: Comparison to Metabolism by
other Extrahepatic Tissues and Possible Role of Nasally Produced Metabolites in
Carcinogenesis, J. Nat. Cancer List., 75: 135-139.
Davis, W. J., 1982, Territory Size in Megaceryle alcyon Along a Stream Habitat, The Auk, 99:
353-362.
Devillers, J., S. Bintein and D. Domine, 1996, Comparison ofBCF Models Based on Log P,
Chemosphere, Vol. 33, No. 6, pp. 1047-1065.
DiToro, D. M., C. S. Zarba, D. H. Hansen, W. J. Berry, R. C. Swartz, C. E. Cowan, S. P. Pavlou,
H. E. Allen, N. A. Thomas and A. P. R. Paquin, 1991, Technical Basis for Establishing
Sediment Quality Criteria for Nonionic Organic Chemicals using Equilibrium Partitioning,
Environ. Toxicol. Chem., 10:1541-1583.
Dreicer, M., T. E. Hakonson, G. C. White and F. W. Whicker, 1984, Rainsplash as a
Mechanism for Soil Contamination of Plant Surfaces, Health Phys. 46:177-187.
Dunning, J. B., 1993, CRC Handbook of Avian Body Masses, CRC Press, Boca Raton, FL., 371
pp.
Eddy, W. F., 1977, Algorithm 523: CONVEX, A New Convex Hull Algorithm for Planar Sets,
ACM TOMS 3, 411-412.
Edwards, N. T., B. M. Ross-Todd, and E. G. Carver, 1982, Uptake and Metabolism of'4C
Anthracene by Soybean (Blycine max), Environmental and Experimental Botany 22:349-357.
C-3
-------
Erickson, R. J. And J. M. McKim, 1990, A Simple Flow-Limited Model for Exchange of
Organic Chemicals at Fish Gills, Environmental Toxicology and Chemistry, Vol. 9, ppl 159-
165.
Fries, G. F., and G. S. Marrow, 1981, Chlorobiphenyl Movement from Soil to Soybean Plants,
J. Agric. Food Chem, 29:757-759.
Gifford, F. A., and S. R. Hanna, 1973, Modeling Urban Air Pollution, Atmos. Environ. 7,
131-136.
Gobas, F., 1993, A Model for Predicting the Bioaccumulation of Hydrophobic Organic
Chemicals in Aquatic Food-Webs: Application to Lake Ontario, Ecological Modeling, Vol. 69,
pp. 1-17.
Green, D. A., and J. S. Millar, 1987, Changes in Gut Dimensions and Capacity ofPeromyscus
maniculatus Relative to Diet Quality and Energy Needs, Can. J. Zool., 65: 2159-2162.
Hamas, M. J., 1994, Belted Kingfisher (Ceryle alcyon). In The Birds of North America, No. 84
(A. Poole and F. Gill, Eds.), Philadelphia; The Academy of Natural Sciences; Washington, DC:
The American Ornithologist Union.
Hanna, S. R., G. A. Briggs and R. P. Hosker, Jr., 1982, Handbook on Atmospheric Diffusion,
U.S. Department of Energy report DOE/TIC-11223
Harms, H. H., 1996, Bioaccumulation and Metabolic Fate of Sewage Sludge Derived Organic
X notics in Plants, Sci. Total. Environ. 185:83-92.
Harms, H., and C. Langebartels, 1986, Standardized Plant Cell Suspension Test Systems for an
Ecotoxicologic Evaluation of the Metabolic Fate of Xenobiotics, Plant Sci. 45:157-165.
Hawker, D., 1990, Description of Fish Bioconcentration Factors in Terms of Solvatochromic
Parameters, Chemosphere, Vol. 20, No. 5, pp. 467-477.
Hawker, D. W. and D. W. Connell, 1985, Relationships Between Partition Coefficient, Uptake
Rate Constant, Clearance Rate Constant and Time to Equilibrium for Bioaccumulation,
Chemosphere, Vol. 14, No. 9, pp. 1205-1219.
Hecht, S. S., W. Grabowski, and K. Groth, 1979, Analysis of Feces for Benzo(a)pyrene After
Consumption of Charcoal-broiled Beef by Rats and Humans, Food and Cosmet. Toxicol., 17:
223-227.
Hope, B. K., 1995, A Review of Models for Estimating Terrestrial Ecological Receptor
Exposure to Chemical Contaminants, Chemosphere 30:2267-2287.
C-4
-------
Hsu, F. C., R. L. Marxmiller, A. Y. S. Yang, 1990, Study of Root Uptake and Xylem
Translocation of Cinmethylin and Related Compounds in Detopped Soybean Roots using a
Pressure Chamber Technique, Plant Physiol. 93:1573-1578.
IT Corporation, 1997, Overview and Key Features of the Integrated Spatial Multimedia
Compartmental Model (ISMCM) An Addendum to Evaluation of Existing Approaches for
Assessing Non-Inhalation Exposure and Risk with Recommendations for Implementing
TRIM, Contract No. 68-D-30094, Work Assignment 3-11, Prepared by IT Corporation for EPA,
April 1997.
IT Corporation, 1997, Model Comparison Update for TRIM, Contract No. 68-D-30094, Work
Assignment 4-18, Task 2 Report, Prepared by IT Corporation for EPA, November 1997.
Keymeulen, R., N. Schamp and H. Van Langenhove, 1993, Factors Affecting Airborne
Monocyclic Aromatic Hydrocarbon Uptake by Plants, Atmos. Environ. 27A: 175-180.
Landrum, P. F., M. D. Reinhold, S. R. Nihart and B. J. Eadie, 1985, Predicting the
Unavailability of Organic Xenobiotics to Pontoporeia Hoyi in the Presence ofHumic and
Fulvic Materials and Natural Dissolved Organic Matter, Environmental Toxicology and
Chemistry, Vol. 4, pp. 459-467.
Lasiewski, R. C. and W. A. Calder, Jr., 1971, A Preliminary Allometric Analysis of Respiratory
Variables in Resting Birds, Resp. Physiol., 11: 152-166.
Lien, G. J. and J. M. McKim, 1993, Predicting Branchial and Cutaneous Uptake of 2,2 ',5,5'-
Tetrachlorobiphenyl in Fathead Minnows (Pimephales promelas) and Japanese Medaka
(Oryzias latipes): Rate Limiting Factors, Aquatic Toxicology, Vol. 27, pp. 15-32.
Lokke, H., 1994, Ecotoxicological Extrapolation: Tool or Toy, pp. 411-425, M. Donker, H.
Eijsackers, and F. Heimbach (eds.), Ecotoxicology of Soil Organisms, Lewis Publishers, Boca
Raton, Florida.
Mackay, D., M. Joy and S. Paterson, 1983, A Quantitative Water, Air, Sediment Interaction
(QWASI) Fugacity Model for Describing the Fate of Chemicals in Lakes, Chemosphere 12,
981-997.
Mackay, D., S. Paterson and M. Joy, 1983, A Quantitative Water, Air, Sediment Interaction
(QWASI) Fugacity Model for Describing the Fate of Chemicals in Rivers, Chemosphere 12,
1193-1208.
Mackay, D., 1982, Correlation of Bioconcentration Factors, Environmental Science
Technology, 16, pp. 274-278.
Matzner, E., 1984, Annual Rates of Deposition of Polycyclic Aromatic Hydrocarbons in
Different Forest Ecosystems, Water, Air, Soil, Pollut. 21:425-434.
C-5
-------
McCarthy, J. F. and B. D. Jimenez, 1985, Reduction in Unavailability to Bluegills ofPolycyclic
Aromatic Hydrocarbons Bound to Dissolved Humic Material, Environmental Toxicology and
Chemistry, Vol. 4, pp. 511-521.
McCrady, J. K., 1994, Vapor-phase 2,3,7,8-TCDD Sorption to Plant Foliage-A Species
Comparison, Chemosphere 28:207-216.
McCrady, J. K. and S. P. Maggard, 1993, Uptake and Photodegradation of 2,3,7,8-
tetrachlorodibenzo-p-dioxin Sorbed to Grass Foliage, Environ. Sci. Technol. 27:343-350.
McKim, J., P. Schmieder and G. Veith, 1985, Absorption Dynamics of Organic Chemical
Transport Across Trout Gills as Related to Octanol-Water Partition Coefficient, Toxicology
and Applied Pharmacology, Vol. 77, pp. 1-10.
McKone, T. E, 1993, The Precision ofQSAR Methods for Estimating Intermedia Transfer
Factors in Exposure Assessments, SAR and QSAR in Environ. Res., 1: 41-51.
McKone, T. E. and P. B. Ryan, 1989, Human Exposures to Chemicals Through Food Chains:
An Uncertainty Analysis, Environ. Sci. Technol. 9:1154-1163.
Muir, D., W. L. Fairchild, A. L. Yarechewski and M. D. Whittle, 1992, Derivation of
Bioaccumulation Parameters and Application of Food Chain Models for Chlorinated Dioxins
and Furans, Chemical Dynamics in Fresh Water Ecosystems, pp. 187-210.
Nakaiima, D., Y. Yoshida, J. Suzuki and S. Suzuki, 1995, Seasonal Changes in the
C itration ofPolycyclic Aromatic Hydrocarbons in Azalea Leaves and Relationship to
Ai... ,pheric Concentration, Chemosphere 30:409-418.
National Research Council (NRC) 1991, Frontiers in Assessing Human Exposure to
Environmental Toxicants, National Academy Press, Washington, DC.
National Research Council (NRC) 1991, Human Exposure Assessment for Airborne
Pollutants: Advances and Opportunities, National Academy Press, Washington, DC.
Nelson, D. W. and L. E. Sommers, 1982, Total Carbon, Organic Carbon, and Organic Matter,
pp. 574-576, Methods of Soil Analysis, Part 2, ASA/SSSA, Madison, Wisconsin.
O'Neill, I .K., M. T. Goldberg, F. ElGhissassi and M. Rojas-Moreno, 1991, Dietary Fibre, Fat
and Beef Modulation ofColonic Nuclear Aberrations and Microcapsule-Trapped
Gastrointestinal Metabolites of Benzo(a)pyrene-Treated C57/B6 Mice Consuming Human
Diets, Carcinogenesis, 12: 175-180.
Pasquill, F., 1961, "The Estimation of the Dispersion of Windborne Material," Meteorol. Mag.
90, 33-49.
C-6
-------
Paterson, S., D. Mackay and C. McFarlane, 1994, A Model of Organic Chemical Uptake by
Plants from Soil and the Atmosphere, Environ. Sci. Technol. 28:2259-2266.
Paterson, S., D. Mackay, E. Bacci and D. Calamari, 1991, Correlation of the Equilibrium and
Kinetics of Leaf-air Exchange of Hydrophobic Chemicals, Environ. Sci. Technol. 25:866-871.
Paterson, S., D. Mackay and A. Gladman, 1991, A Fugacity Model of Chemical Uptake by
Plants from Soil and Air, Chemosphere 23:539-565.
Petridou-Fischer, J., S. L. Whaley and A. R. Dahl, 1988, In Vivo Metabolism of Nasally
Instilled Benzo(a)pyrene in Dogs and Monkeys, Toxicology, 48: 31-40.
Riederer, M., 1995, Partitioning and Transport of Organic Chemicals Between the
Atmospheric Environment and Leaves, pp. 153-190, In S. Trapp and J. C. McFarlane, eds.
Plant Contamination: Modeling and Simulation of Organic Chemical Processes, Lewis
Publishers, Boca Raton.
Riederer, M., 1990, Estimating Partitioning and Transport of Organic Chemicals in the
Foliage/Atmosphere System: Discussion of a Fugacity-based Model, Environ. Sci. Technol.
24:829-837.
Ryan, J. A., R. M. Bell, J. M. Davidson and G. A. O'Connor, Plant Uptake of Non-ionic
Organic Chemicals from Soils, Chemosphere 17:2299-2323.
Sabljic, A., H. Gusten, J. Schonherr and M. Riederer, 1990, Modeling Plant Uptake of Airborne
Organic Chemicals. 1. Plant Cuticle/Water Partitioning and Molecular Conductivity,
Environ. Sci. Technol. 24:1321-1335.
Saito, S., A. Tanoue and M. Matsuo, 1991, The i/o Characters to Describe Bioconcentration of
Organic Chemicals in Fish, Chemosphere, Vol. 23, No. 6, pp. 789-799.
Schnoor, J. L., 1981, Fate and Transport ofDieldrin in Coraville Reservoir: Residues in Fish
and Water Following a Pesticide Ban, Science 211, 840-842.
Schnoor, J. L. and D. C. MacAvoy, 1981, A Pesticide Transport and Bioconcentration Model,
J. Environ. Eng. Div, American Society of Civil Engineers 107, 1229-1245.
Sijm, D. and A. Van Der Linde, 1995, Size-Dependent Bioconcentration Kinetics of
Hydrophobic Organic Chemicals in Fish Based on Diffusive Mass Transfer and Allometric
Relationships, Environmental Science Technology, Vol. 29, pp. 2769-2777.
Silva, M. and J. A. Downing, 1995, CRC Handbook of Mammalian Body Masses, CRC Press,
Boca Raton, Florida, 359 pp.
Simonich, S. L. and R. A. Hites, 1994, Importance of Vegetation in Removing Polycyclic
Aromatic Hydrocarbons from the Atmosphere, Nature 370:49-51.
C-7
-------
Simonich, S. L. and R. A. Kites, 1994, Vegetation-Atmosphere Partitioning ofPolycyclic
Aromatic Hydrocarbons, Environ. Sci. Technol. 28:939-943. -
Southworth, G. R., 1979, The Role of Volatilization in Removing Polycyclic Aromatics from
Aquatic Environments, Bull. Environ. Contamin. Toxicol. 21, 507-514.
Stahl, W. R., 1967, Scaling of Respiratory Variables in Mammals, J. Appl. Physiol., 22(3): 453-
460.
Suter, G. W., H, 1993, Ecological Risk Assessment, Lewis Publ. Co., Boca Raton, Florida, 538
pp.
Thomann, R. V., J. P. Connolly and T. Parkerton, 1992, Modeling Accumulation of Organic
Chemicals in Aquatic Food Webs, Chemical Dynamics in Fresh Water Ecosystems, pp. 153-
186.
Thomann, R. V., 1989, Bioaccumulation Model of Organic Chemical Distribution in Aquatic
Food Chains, Environmental Science Technology, Vol. 23, pp. 699-707.
Thomas, W., Uber die Verwendung von Pfanzen zur Analyse raumlicher Spurensubstanz-
Immisionsmuster, Staub-Reinhalt, Luft 43:141-148.
Thomas, W., A. Ruhling and H. Simon, 1984, Accumulation of Airborne Pollutants (PAH,
Chlorinated Hydrocarbons, Heavy Metals) in Various Plant Species and Humus, Environ.
PoMut. (Ser. A) 36:295-310.
\ j. E. and H. R. Meyer, 1983, Radiological Assessment, A Textbook on Environmental
Dose Analysis, NUREG/CR-3332, ORNL5968, September 1983.
Topp, E., I. Scheunert, A. Attar and F. Korte, 1986, Factors Affecting the Uptake ofl4-C-
Labeled Organic Chemicals by Plants from Soil, Ecotoxicol. Environ. Safety 11:219-228.
Total Human Exposure Related Database and Advanced Simulation Model (THERDBASE),
1996, A Modeling and Database System for Making Total Human Exposure Assessments,
(computer software), Prepared for EPA, National Exposure Research Laboratory, Las Vegas,
Nevada 89193, Prepared for University of Nevada Las Vegas (UNLV) Exposure Modeling &
Validation Group, Harry Reid Center for Environmental Studies.
Trapp, S., 1995, Model for Uptake ofXenobiotics into Plants, pp. 107-151, In S. Trapp and J. C.
McFarlane, eds., Plant Contamination: Modeling and Simulation of Organic Chemical Processes,
Lewis Publishers, Boca Raton, Florida.
Trapp, S., M. Matthies, I. Scheunert and E. M. Topp, 1990, Modeling the Bioconcentration of
Organic Chemicals in Plants, Environ. Sci. Technol. 24:1246-1252.
C-8
-------
Travis, C. C. and A. D. Arms, 1988, Bioconcentration ofOrganics in Beef, Milk, and
Vegetation, Environ. Sci. Technol., 22: 271-274.
Turner, D. B., 1970, Workbook of Atmospheric Dispersion Estimates, Office of Air Programs,
U.S. Environmental Protection Agency, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency (EPA), 1996, Technical Background Document for Soil
Screening Guidance, EPA 540/R-95/128, May.
U. S. Environmental Protection Agency (EPA), 1994, Revised Draft of Risk Assessment
Implementation Guidance for Hazardous Waste Combustion Facilities, Memorandum from M.
H. Shapiro, U. S. EPA Office of Solid Waste to Waste Management Division Directors, Regions
I-X, May 5, as cited in [WTI report].
van Gestel, C. A. M., W. Ma and C. E. Srnit, 1991, Development ofQSARs in Terrestrial
Ecotoxicology: Earthworm Toxicity and Soil Sorption of Chlorophenols, Chlorobenzenes, and
Chloroaniline, Sci. Total. Environ., 109/110:589-604.
van Gestel, C. A. M. and W. C. Ma, 1988, Toxicity and Bioaccumulation of Chlorophenols in
Earthworms in Relation to Bioavailability in Soil, Ecotox. Environ. Saf., 15:289-297.
Weyand, E. H. and D. R. Bevan, 1987, Species Differences in Disposition ofBenzo(a)pyrene,
Drug Metabolism and Disposition, 15: 442-448.
Weyand, E. H. and D. R. Bevan, 1986, Benzo(a)pyrene Disposition and Metabolism in Rats
Following Intratracheal Instillation, Cancer Res., 46: 5655-5661.
Whitaker, J. O., Jr., 1966, Food ofMus Musculus, Peromyscus Maniculatus Bairdi, and
Peromyscus Leucopus in Vigo County, Indiana, J. Mamm., 47: 473-486.
Wild, S. R. and K. C. Jones, 1995, Polynuclear Aromatic Hydrocarbons in the United Kingdom
Environment: A Preliminary Source Inventory and Budget, Environ. Pollut. 88:91-108.
Wolff, J. O., R. D. Dueser and K. S. Berry, 1985, Food Habits ofSympatric Peromyscus
Leucopus and Peromyscus Maniculatus, J. Mamm., 66: 795-798.
Yang, S. Y. N., D. W. Connell, D. W. Hawker and S. I. Kiyal, 1991, Poly cyclic Aromatic
Hydrocarbons in Air, Soil, and Vegetation in the Vicinity of an Urban Roadway, Sci. Total
Env. 102:229-240.
C-9
-------
TECHNICAL REPORT DATA
(Please read Instructions on reverse before completing)
1. REPORT NO.
EPA-452/R-98-001
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
5. REPORT DATE
March 1998
The Total Risk Integrated Methodology: Implementation of the
TRIM Conceptual Design through the TRIM.FaTE Module
6. PERFORMING ORGANIZATION CODE
Office of Air and Radiation
7. AUTHOR(S)
U.S. EPA
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
AQSSD/REAG (MD-15)
Research Triangle Park, NC 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Final
Office of Air Quality Planning and Standards
Office of Air and Radiation
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
14. SPONSORING AGENCY CODE
EPA/200/04
15 SUPPLEMENTARY NOTES
U.S. EPA Project Officer: Amy B. Vasu
16 ABSTRACT
The report summarizes work performed during the first developmental phase of the Total Risk Integrated
Methodology (TRIM). TRIM is a modeling system for the assessment of human health and ecological risk
resulting from exposure to hazardous and criteria air pollutants. The first phase of TRIM development
included the conceptualization of TRIM and the implementation of the TRIM conceptual approach through
the development of the first TRIM module, the environmental fate, transport, and exposure module, called
TRIM.FaTE. This report provides detailed information about the overall structure of TRIM and the
development of the TRIM.FaTE module.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSAT1 Field/Group
Multimedia modeling; exposure; risk assessment
Air pollution
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (Report)
Unclassified
21. NO. OF PAGES
152
20. SECURITY CLASS (Page)
Unclassified
22. PRICE
EPA Form 2220-1 (ReT. 4-77) PREVIOUS EDITION IS OBSOLETE
------- |