May 8, 2000
EPA SAB-EC-ADV-00-004
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, NW
Washington, DC 20460
Subject: Advisory on the Agency's "Total Risk Integrated Methodology" (TRIM)
Dear Ms. Browner:
The Environmental Models Subcommittee (EMS), hereinafter referred to as the
"Subcommittee", met December 13-14, 1999 to review the "Total Risk Integrated Methodology
(TRIM) Status Report", the draft "TREVI.Expo Technical Support Document" and the two volume
"TREVLFaTE Technical Support Document" developed by EPA's Office of Air Quality Planning and
Standards (OAQPS). The Subcommittee conducted this advisory in order to provide the Agency with
advice and insights on the adequacy of the proposed TRIM approach to predict exposures and risks
posed by air pollutants.
In general, we find the OAQPS' ongoing efforts to develop TRIM as a flexible, state-of-the-art
system for evaluating multimedia chemical fate, transport, exposure and risk, to be effective and
innovative. The OAQPS staff and its consultants have been responsive to our previous advice on
components of TREVI.FaTE that required further consideration and improvement. While current
knowledge and available methods continue to limit the rigor and accuracy with which Hazardous Air
Pollutant (HAP) impacts can be assessed, OAQPS is developing a system capable of recognizing and
characterizing these limitations and attendant uncertainties. Much of our review focuses upon ongoing
efforts that can help to promote further advancement and reduction of these uncertainties. Specific
findings and recommendations are as follows:
The modular design and open architecture of TRIM provide a flexible and appropriate means
of accommodating new scientific information. However, significant user guidance and support will be
necessary to ensure that appropriate model setups and configurations are selected. Early workshops
and beta testing of the integrated TRIM system by the affected user community are recommended to
ensure the appropriate level of stakeholder input and a well-designed product.
Proposed plans for addressing uncertainty and variability within TRIM are innovative and
consistent with the current state-of-the-art in the risk sciences. Significant user guidance and interface
capabilities will be needed to ensure that these tools are utilized in an informed and well-documented
-------
manner. The OAQPS should provide examples where uncertainty analysis and model evaluations are
presented within the context of intended regulatory applications, such as the calculation of residual risk
following implementation of emission reductions. Additional effort is also recommended to develop
methods capable of capturing the effects of local, sub-compartmental variations in landscape
properties, chemical concentrations and exposures.
Evaluation plans for TRUVI.FaTE, including code validation, mass balance checks, model-to-
model comparisons, and comparisons against observed data in field applications are appropriate,
though further review of the results of these evaluations in the context of regulatory applications is
necessary to ensure proper implementation and inferences. Evaluation plans should consider
simultaneous test applications of the TRIM modules (FaTE, Expo and Risk) to ensure that model
linkages are properly specified, with outputs from the upstream modules of the appropriate type and
resolution for input to the subsequent module(s).
The Subcommittee is concerned that the planned field comparison for mercury may not be
effective, because of the high degree of complexity and uncertainty in mercury chemistry, the large
number of adjustable parameters that must be fit for the model, and the difficulty in obtaining adequate
data to properly test the model. As such, planned field comparisons of mercury should be replaced or
augmented by considering sites with other metals that undergo simpler fate and transport processes in
the environment, so as to ensure that a successful model specification can be achieved.
Tests of the TRIM system at larger scales, such as for a metropolitan air basin with multiple
stationary and mobile sources, are recommended to evaluate the ability of the TRIM system to operate
at this scale of application and to utilize emerging electronic/GIS databases with geographic,
meteorological, emissions, population and environmental resource information. Evaluation of the TRIM
system should be a continuing process that encourages ongoing data collection and evaluation for local,
site-specific applications, with iterative updates to the model structure, database and characterization of
uncertainty. The application protocol for TRIM should provide incentives for the development of
improved data collection methods (with improved sensitivity, accuracy and precision) and improved
databases for model input.
Efforts to improve the TRUVI.FaTE module to address local-scale atmospheric dispersion
through the use of alternative air pollution models, to develop appropriate transport parameters for
surface waters and air-soil mass transfer, to include kinetic, reversible reactions among metal species,
and to incorporate seasonal effects in biotic processes, are appropriately focused. Specific
recommendations are provided in our report to allow further improvements and focus. In each case,
we emphasize the need for an effective user-interface that guides users in the selection of alternative
modules and input parameters given the chemical(s), landscape properties, and types of exposures and
risks addressed.
The design of TRDVLExpo and its use of existing modeling science for inhalation and ingestion
exposures are, in general, consistent with OAQPS' programmatic and regulatory needs and the state-
-------
of-the-art of exposure assessment. System flexibility is needed to allow the selection of time-activity
patterns and exposure factors consistent with the necessary level of spatial and temporal detail,
distinguishing between long-term chronic health endpoints and short-term acute effects.
The design of the TRUVI.Risk module appears to be appropriate, though more details and
example applications will be necessary for rigorous review. Particular care will be needed to ensure
that the time scales for exposure and risk calculation are appropriately matched (in particular, for acute,
non-cancer health effects associated with short-term and/or peak exposures, where current knowledge
and guidance are lacking) and in providing options for incorporating emerging knowledge on pollutant
interactions which may result in non-additive risk. Ecological risk endpoints and metrics are not well
defined in TRIM (or in any other risk assessment framework currently available), and further work will
be needed to provide appropriate options. The multi-chemical, multi-media design of the TRIM system
provides the opportunity to explore methods for addressing new and emerging issues in risk assessment
and the risk sciences, including the effects of mixtures, population susceptibility and cumulative risk; as
well as metrics for environmental equity and ecological impacts at the population level. OAQPS should
take advantage of this opportunity to suggest and illustrate possible approaches to these issues for
broader Agency consideration and evaluation.
We thank the Agency for the opportunity to provide technical advice on this important effort at
the interface of environmental science, modeling and regulatory need. The Science Advisory Board
and its Environmental Models Subcommittee look forward to your response to our report, with a
particular emphasis on the points raised in this letter to you, and for the opportunity for further focused
reviews in the future.
Sincerely,
/s/
Dr. Morton Lippmann, Interim Chair
Science Advisory Board
/s/
Dr. Mitchell Small, Chair
Environmental Models Subcommittee
Science Advisory Board
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a public
advisory group providing extramural scientific information and advice to the Administrator and other
officials of the Environmental Protection Agency. The Board is structured to provide balanced, expert
assessment of scientific matters related to problems facing the Agency. This report has not been
reviewed for approval by the Agency and, hence, the contents of this report do not necessarily
represent the views and policies of the Environmental Protection Agency, nor of other agencies in the
Executive Branch of the Federal government, nor does mention of trade names or commercial products
constitute a recommendation for use.
Distribution and Availability: This Science Advisory Board report is provided to the EPA
Administrator, senior Agency management, appropriate program staff, interested members of the
public, and is posted on the SAB website (www.epa.gov/sab). Information on its availability is also
provided in the SAB's monthly newsletter {Happenings at the Science Advisory Board). Additional
copies and further information are available from the SAB Staff.
-------
ABSTRACT
The Environmental Models Subcommittee of the EPA Science Advisory Board (SAB)
reviewed the Agency's development of the Total Risk Integrated Methodology (TRIM) for predicting
multimedia exposures and risks posed by hazardous air pollutants. The Subcommittee found the EPA
TRIM model to be an innovative, flexible, state-of-the-art system for evaluating multimedia chemical
fate, transport, exposure and risk. Specific recommendations are provided on efforts to improve the
TREVLFaTE module, planned field comparison studies of the TRIM system, and the design and
implementation of the exposure and risk modules.
The Subcommittee determined that there is a need for OAQPS to better specify its plans and
timeline for use of the TRIM system within the Agency and subsequent release to a broader user
community. Early workshops and beta testing of the integrated TRIM system by the affected user
community are recommended to help in the development of user guidance and support. The
application protocol for TRIM should provide incentives for the development of improved data
collection methods and improved databases for model input. For all current risk assessment models,
including the TRIM system, new methods are needed to address emerging issues including: the effects
of mixtures; population susceptibility and cumulative risk; and metrics for environmental equity and
ecological impacts at the population level.
Keywords: TRIM, hazardous air pollutants (HAPs), multimedia environmental models, exposure
assessment, risk assessment, validation, sensitivity analysis, uncertainty
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
ENVIRONMENTAL MODELS SUBCOMMITTEE OF THE EXECUTIVE
COMMITTEE
CHAIR
Dr. Mitchell Small, Professor, Department of Civil Engineering & Public Policy, Carnegie Mellon
University, Pittsburgh, PA
MEMBERS
Dr. Steven M. Bartell, Senior Associate, Cadmus Group, Inc., Oak Ridge, TN
Dr. Calvin Chien, Senior Environmental Fellow, E.I. DuPont Company, Wilmington, DE
Dr. Kai-Shen Liu, Epidemiologist, California Department of Health Services, Environmental Health
Laboratory Branch, Berkeley, CA
Dr. Paulette Middleton, Associate Director, Environmental Science and Policy Center, RAND
Corporation, Boulder, CO
Dr. Ishwar Murarka, Chief Scientist and President, ISH Inc., Cupertino, CA
CONSULTANTS
Dr. M. Bruce Beck, Professor & Eminent Scholar, Warnell School of Forest Resources, University
of Georgia, Athens GA
Dr. Linfleld Brown, Professor, Department of Civil and Environmental Engineering, Tufts University,
Medford, MA
Dr. Arthur J. Gold, Professor, Department of Natural Resources Science, University of Rhode Island,
Kingston, RI
Dr. Helen Grogan, Research Scientist, Cascade Scientific, Inc., Bend, OR
Dr. Wu-Seng Lung, Professor, Department of Civil Engineering, University of Virginia, Charlottesville,
VA
Dr. Jana Milford, Associate Professor, Department of Mechanical Engineering, University of
Colorado, Boulder, CO
Dr. Thomas Theis, Professor & Chair, Department of Civil and Environmental Engineering, Clarkson
University, Potsdam, NY
iii
-------
SCIENCE ADVISORY BOARD STAFF
Dr. John R. Fowle HI, Deputy Staff Director/Designated Federal Officer, Environmental Protection
Agency, Science Advisory Board (1400A), 1200 Pennsylvania Avenue, NW, Washington, DC
20460
Ms. Melanie Medina-Metzger, Designated Federal Officer, Environmental Protection Agency,
Science Advisory Board (1400A), 1200 Pennsylvania Avenue, NW, Washington, DC 20460
Dr. K. Jack Kooyoomjian,, Designated Federal Officer, Environmental Protection Agency, Science
Advisory Board (1400A), 1200 Pennsylvania Avenue, NW, Washington, DC 20460
Mrs. Dorothy M. Clark, Management Assistant, Environmental Protection Agency, Science
Advisory Board (1400A), 1200 Pennsylvania Avenue, NW, Washington, DC 20460
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 4
3. CHARGE TO SUBCOMMITTEE 5
3.1 Overall TRIM System 5
3.2 TRIM.FaTEModule 5
3.3 TREVLExpo 6
3.4 TRIM.Risk 6
4. RESPONSE TO CHARGE 7
4.1 Overall TRIM System 7
4.2 TRIM.FaTE Module 10
4.3 TREVLExpo 16
4.4 TRIM.Risk 17
5. CONCLUSIONS 20
REFERENCES R-l
APPENDIX A - ILLUSTRATION OF POTENTIAL MISREPRESENTATION OF DISPERSION
INTRIM.FaTE A-l
APPENDIX B - MORE DETAILED COMMENTS ON THE PARAMETERIZATION OF
DISPERSION AND ADVECTION FOR SURFACE WATER COMPARTMENTS
IN TREVLFaTE B-l
APPENDIX C - SPECIFIC EDITORIAL COMMENTS ON THE TREVl.Expo and
TRIMRiskMODULES C-l
APPENDIX D - UPDATED LIST OF TRIM PRESENTATIONS AND PUBLICATIONS ... D-1
APPENDIX E - ACRONYMS E-l
-------
1. EXECUTIVE SUMMARY
The Environmental Models Subcommittee (EMS), hereinafter referred to as the
"Subcommittee", met December 13-14, 1999 to review the "Total Risk Integrated Methodology
(TRIM) Status Report", the draft "TREVI.Expo Technical Support Document" and the two volume
"TREVLFaTE Technical Support Document" developed by EPA's Office of Air Quality Planning and
Standards (OAQPS). The Subcommittee conducted this review in order to provide the Agency with
advice and insights on the adequacy of the proposed Total Integrated Risk Methodology approach to
predict exposures and risks posed by air pollutants.
The Subcommittee found the OAQPS' TRIM model to be an innovative, flexible, state-of-the-
art system for evaluating multimedia chemical fate, transport, exposure and risk. The Office of and its
consultants were responsive to our previous advice on components of TREVI.FaTE that required further
consideration and improvement. While current knowledge and available methods continue to limit the
rigor and accuracy with which HAP impacts can be assessed, OAQPS is developing a system capable
of recognizing and characterizing these limitations and attendant uncertainties. Much of the
Subcommittee's review focused upon ongoing efforts to help promote further advancement and
reduction of these uncertainties. Specific findings and recommendations are as follows:
The modular design and open architecture of TRIM provide a flexible and appropriate means
of accommodating new scientific information. However, significant user guidance and support will be
necessary to ensure that appropriate model setups and configurations are selected. Early workshops
and beta testing of the integrated TRIM system by the affected user community are recommended to
ensure the appropriate level of stakeholder input and a well-designed product.
Proposed plans for addressing uncertainty and variability within TRIM are innovative and
consistent with the current state-of-the-art in the risk sciences. Significant user guidance and interface
capabilities will be needed to ensure that these tools are utilized in an informed and well-documented
manner. The OAQPS should provide examples where uncertainty analysis and model evaluations are
presented within the context of intended regulatory applications, such as the calculation of residual risk
following implementation of emission reductions. Additional effort is also recommended to develop
methods capable of capturing the effects of local, sub-compartmental variations in landscape
properties, chemical concentrations and exposures.
Evaluation plans for TREVI.FaTE, including code validation, mass balance checks, model-to-
model comparisons, and comparisons against observed data in field applications are appropriate,
though further review of the results of these evaluations in the context of regulatory applications is
necessary to ensure proper implementation and inferences. Evaluation plans should consider
simultaneous test applications of the TRIM modules (FaTE, Expo and Risk) to ensure that model
linkages are properly specified, with outputs from the upstream modules of the appropriate type and
resolution for input to the subsequent module(s). The Subcommittee is concerned that the planned field
comparison for mercury may not be effective, because of the high degree of complexity and uncertainty
1
-------
in mercury chemistry, the large number of adjustable parameters that must be fit for the model, and the
difficulty in obtaining adequate data to properly test the model. As such, planned field comparisons of
mercury should be replaced or augmented by considering sites with other metals that undergo simpler
fate and transport processes in the environment, so as to ensure that a successful model specification
can be achieved. Tests of the TRIM system at larger scales, such as for a metropolitan air basin with
multiple stationary and mobile sources, are recommended to evaluate the ability of the TRIM system to
operate at this scale of application and to utilize emerging electronic/GIS databases with geographic,
meteorological, emissions, population and environmental resource information. Evaluation of the TRIM
system should be a continuing process that encourages ongoing data collection and evaluation for local,
site-specific applications, with iterative updates to the model structure, database and characterization of
uncertainty. The application protocol for TRIM should provide incentives for the development of
improved data collection methods (with improved sensitivity, accuracy and precision) and improved
databases for model input.
Efforts to improve the TRDVI.FaTE module to address local-scale atmospheric dispersion
through the use of alternative air pollution models, to develop appropriate transport parameters for
surface waters and air-soil mass transfer, to include kinetic, reversible reactions among metal species,
and to incorporate seasonal effects in some biotic processes, are appropriately focused. Specific
recommendations are provided in our report to allow further improvements and focus. In each case,
we emphasize the need for an effective user-interface that guides users in the selection of alternative
modules and input parameters given the chemical(s), landscape properties, and types of exposures and
risks addressed.
The design of TRDVLExpo and its use of existing modeling science for inhalation and ingestion
exposures are, in general, consistent with OAQPS' programmatic and regulatory needs and the state-
of-the-art of exposure assessment. System flexibility is needed to allow the selection of time-activity
patterns and exposure factors consistent with the necessary level of spatial and temporal detail,
distinguishing between long-term chronic health endpoints and short-term acute effects.
The design of the TREVLRisk module is appropriate for analysis of human health and ecological
risks, though more details and example applications will be necessary for rigorous review. Particular
care will be needed to ensure that the time scales for exposure and risk calculation are appropriately
matched (in particular, for acute, non-cancer health effects associated with short-term and/or peak
exposures, where current knowledge and guidance are lacking) and in providing options for
incorporating emerging knowledge on pollutant interactions which may result in non-additive risk.
Ecological risk endpoints and metrics are not well defined in TRIM (or in any other risk assessment
framework currently available), and further work will be needed to provide appropriate endpoints and
options for assessing them. The multi-chemical, multi-media design of the TRIM system provides the
opportunity to explore methods for addressing new and emerging issues in risk assessment and the risk
sciences, including the effects of mixtures, population susceptibility and cumulative risk; as well as
metrics for environmental equity and ecological impacts at the population level. OAQPS should take
advantage of this opportunity to suggest and illustrate possible approaches to these issues for broader
Agency consideration and evaluation.
-------
2. INTRODUCTION
The Environmental Models Subcommittee (EMS) met December 13 and 14, 1999 to review
the "Total Risk Integrated Methodology (TRIM) Status Report", the draft "TREVI.Expo Technical
Support Document" and the two volume "TRUVLFaTE Technical Support Document". The
Subcommittee conducted this review in order to provide the Agency with advice and insights on the
adequacy of the proposed Total Integrated Risk Methodology (TRIM) approach to predict exposures
and risks posed by air pollutants.
The charge to the Subcommittee contained nine questions. In general the Subcommittee finds
that the OAQPS efforts to develop the TRIM system are innovative and effective, given the significant
challenges and the relatively new and rapidly evolving state of science for multimedia fate, transport,
exposure and risk models potentially applicable to hazardous air pollutants (HAPs) and criteria air
pollutants. Indeed, development of the proposed TRIM system demands much more than piecing
together existing models for these processes; rather new methods, approaches and advancement of the
state-of-the-art are needed to ensure effective predictive capability and linkage of the various
components. Given this, current knowledge and methods continue to limit the ability of such models to
provide accurate and meaningful predictions of environmental concentrations, exposures and risks for
many of the HAPs and criteria air pollutants that are the focus of the TRIM effort. Our report focuses
on key issues and concerns for continued development of a scientifically sound and acceptable
modeling system that remains cognizant of these limitations and attendant uncertainties, while
encouraging further research and data collection to address them.
-------
3. CHARGE TO SUBCOMMITTEE
3.1 Overall TRIM System
Charge 1: Does TRIM'S current design and modular approach provide us with an
appropriate design for accommodating new scientific information and model flexibility?
Charge 2: Are the proposed plans for addressing uncertainty and variability scientifically
defensible? If not, what would you propose?
Charge 3: An important feature of TRIM is its open architecture (e.g., as demonstrated by the
inclusion of an "algorithm library" developed for TRIMFaTE). This architecture allows the
user to add new processes, input data, and approaches to the TRIM modules. Given that a
User's Guide will be developed to provide guidance on the use of these features, does this
provide TRIM, and specifically TREVLFaTE, with a significant technical asset in the area of
user flexibility and model transparency? If not, what would be the preferred alternative?
3.2 TRIM.FaTE Module
Charge 1: In response to specific recommendations by the previous SAB panel, we
(OAQPS) have investigated certain issues. As a result, TRIM.FaTE has been modified (item
a), below), or the model documentation has been updated to address specific panel concerns
(items b) and c), below).
a) Do the modifications described in items (i) and (ii) below demonstrate an improvement
in the ability of TREVLFaTE to address these two issues? If not, what specific
recommendations could you offer to do that?
(i) As a demonstration of the capability of TRIM.FaTE to incorporate output from
external models, the Agency has developed and implemented methods for
TRIM.FaTE to accept air concentrations and deposition output from an air
quality model.
(ii) TRIM.FaTE is revised with regard to how it addresses certain kinds of
chemical mass movement between certain compartments, specifically: (1)
dispersive transport between surface water compartments, and (2) diffusion
and advection between soil compartments and from air compartments to
surface soil compartments.
b) The incorporation of both horizontal and vertical atmospheric dispersion/diffusion
algorithms directly within TRIM.FaTE using both lateral and vertical Pasquill-Gifford
plume dispersion coefficients was pursued. After further review and consultation, it was
-------
concluded that incorporating dispersion terms into the TRIM.FaTE air model was not
appropriate at this time. Do you concur with this conclusion (discussed more fully in the
model documentation)? What alternate methods for incorporating atmospheric
dispersion/diffusion directly within TREVLFaTE would you recommend?
c) An evaluation plan for TRIM.FaTE has been developed and is being implemented. Is
the proposed evaluation plan scientifically defensible? If not, what would you propose?
Charge 2: Do the newly added TREVLFATE abilities to: a) follow transformation products
temporally through the study period and allow for 'reverse transformation'; and b) track metals
in a multimedia environment provide notable improvement in TRDVLFATE's usefulness for
Agency air pollutant risk assessments? Have we adequately captured the salient technical
features in these additions? If not, what should be included?
Charge 3: Does the addition to TRIM.FaTE of certain seasonal processes (i.e., litterfall and
changes in plant uptake) represent an improvement to the model's ability to predict different
media pollutant concentrations for the time frames of interest for risk assessment? Are there
other processes related to seasonality which play an important role in pollutant transfers and
resulting media concentrations for relevant time frames, and thus would be a valuable addition
to the model?
3.3 TRIM.Expo
Charge 1: Does the proposed conceptual design for TRIM.EXPO capture the salient
technical features necessary to characterize inhalation and ingestion exposures given our
programmatic and regulatory needs? If not, what salient features are missing and how should
they be included?
Charge 2: Have we drawn appropriately from the existing modeling science on inhalation and
ingestion exposures and are the specific algorithms structured in a way that is consistent with the
modeling framework?
3.4 TRIM.Risk
Charge 1: Does the overall conceptual approach described for the TRIM.RISK module
adequately capture the outputs from the other TRIM modules and provide quantitative human
health and ecological risk characterization information that is consistent with Agency risk
characterization policy?
-------
4. RESPONSE TO CHARGE
4.1 Overall TRIM System
Charge 1: Does TRIM's current design and modular approach provide an appropriate design
for accommodating new scientific information and model flexibility? and
Charge 3: An important feature of TRIM is its open architecture (e.g., as demonstrated by the
inclusion of an "algorithm library" developed for TRIMFaTE). This architecture allows the
user to add new processes, input data, and approaches to the TRIM modules. Given that a
User's Guide will be developed to provide guidance on the use of these features, does this
provide TRIM, and specifically TREVLFaTE, with a significant technical asset in the area of
user flexibility and model transparency? If not, what would be the preferred alternative?
The Subcommittee concurs that the current design and modular approach provide a flexible and
appropriate means of accommodating new scientific information. Furthermore, the open architecture of
TRIM, which allows users to add new processes, input data and approaches, is a significant technical
asset in promoting user flexibility and model transparency. To help motivate and clarify these features
and their advantages, the technical support document (TSD) for TRIM should include a review of
alternative options for promoting model flexibility and ease of use, such as the use of existing model-
building software systems. The TSD should also elaborate on why the current design was preferred
and ultimately chosen over these other options.
Key issues affecting the appropriate level of model flexibility and user-adaptability involve who
will be using the model, for what purpose, and when. There is a need for OAQPS to better specify its
plans and timeline for use of the TRIM system within the Agency and subsequent release for use by a
broader user community. While the Subcommittee recognizes that much further work is needed before
the model will be ready for use for OAQPS's regulatory needs (and that the deadlines for these
applications are rapidly approaching), early interactions with, and involvement of, affected stakeholders
are essential steps in the process. Early workshops and beta testing by the affected user community
will help identify software errors and limitations, and allow the Agency to gather useful information on
the user community.
The need for involving a broad range of affected stakeholders in model development, testing
and fmalization, was emphasized in a previous consultation of the Environmental Modeling
Subcommittee. In its review of the Model Acceptability Use Criteria (MAC White Paper), the
Subcommittee recommended that "EPA model development can benefit greatly from targeted
stakeholder participation...". In her response to this recommendation Dr. Norine Noonan, the Assistant
Administrator for Research and Development, indicated that:
-------
"Follow-up actions in the MAC White Paper include developing and utilizing a model
clearinghouse to inform internal and external stakeholders on model evaluation results,
availability, and application experience... We have also included the development of a
communication strategy for the public and other external model users to provide
feedback to EPA through Internet sites in Section 6. " (EPA, 1999)
While a flexible modeling system has many advantages, a drawback to this approach is that it
may be easy for users to mis-configure or otherwise misuse the program and obtain inappropriate or
misleading results. Users may have difficulty integrating new processes, data, and/or approaches
within the TRIM framework, unless the TRIM modules are properly designed and the User's Guide is
clear and specific on how to add new elements to the TRIM modules. Good user guidance and an
eventual user interface to guide model setup and configuration will thus be necessary. Such an interface
would, for example, identify and help facilitate the setup of a "standard regulatory configuration" for
specific problem applications and inform the user that alternative assumptions must be justified (e.g.,
shown to be as good or better than the standard assumptions) when selected for a particular
application. Eventually, an expert-system type user interface to help in module selection, configuration,
and input data specification would be of great benefit to the users of TRIM to ensure appropriate and
consistent application. This need for pre-screening and an effective user interface is identified a number
of times in our report in response to specific questions regarding options for alternative fate and
transport, exposure and risk assumptions and modules in the TRIM system.
Linking and interfacing the three proposed TRIM modules on fate and transport, exposure, and
risk is a major and difficult task. The first module, TREVI.FaTE, and the second module, TREVLExpo,
have thus far been developed in a parallel and while interactions and communication between these
efforts have taken place, some degree of mismatch is apparent. The major input data needed by the
second module do not in many cases correspond to the output data from the first module. The
modules' reliance on, and use of, data and parameters of differing spatial and temporal scales may
create a number of unforeseen problems. With the rapid advance of GIS, it is quite possible that in the
near future detailed meteorological data and three-dimensional geographic maps will be available
electronically at local, regional and national levels. It is not clear whether TRIM can handle and
integrate the large amount of data this will entail at different temporal and spatial scales in an efficient
and accurate manner. Early experimentation with model linkage and the use of available data-sets for
emissions, meteorological inputs, census information and GIS resource inventories should take place
soon, and not await finalization of the individual modules.
Effective presentation of the TRIM components and their interactions is essential to
communicate the key elements of this complex modeling system. Further use of conceptual diagrams is
recommended for the TSD to demonstrate model calculations, dependencies and logic.
Charge 2: Are the proposed plans for addressing uncertainty and variability scientifically
defensible? If not, what would you propose?
-------
The proposed plan for addressing uncertainty and variability is innovative and consistent with
the current state-of-knowledge in the risk sciences. The proposed methods for separating variability
and uncertainty are appropriate and current. The proposed tiered method for pre-screening the
sensitivity of model parameters to eliminate those to which the model is insensitive from consideration in
further, more-detailed uncertainty analyses, is appropriate and necessary to provide for feasible and
effective model evaluation and insights. However, caution and appropriate user guidance is needed to
ensure that important model uncertainties are not eliminated in this pre-screening phase. Users need to
be aware that the results of sensitivity and uncertainty analysis are themselves often very sensitive to the
methods used and the assumptions adopted. For example, local (or derivative-based) sensitivity
coefficients can be highly sensitive to the base-case around which the derivatives are computed. The
inclusion of parameter correlations in an uncertainty analysis may cause predicted uncertainties and
resulting inferences to change, however, our knowledge of correlations is often very limited. Users
should be provided with a number of options (and accompanying guidance) for sensitivity and
uncertainty analysis. The assumptions and input functions for any uncertainty analysis (e.g., the
distribution of model inputs, assumed correlation structure, etc.) should be clearly presented and
documented as part of the model output, allowing the user to check and document their evaluations.
The TRUVLFaTE model utilizes physical compartments for the environment which are often
highly aggregated. Local variations in compartment properties (e.g., soil texture, soil pH or redox
conditions) may result in smaller-scale variations in chemical concentrations and exposure which are
missed with such aggregation. The Agency should explore options for considering and characterizing
the effects of local variability of this type in the TRIM model.
A number of the most interesting results of an exposure and risk assessment occur at the "tails
of the distributions." This may occur at the upper tail of the variability distributions, where variations in
media concentrations or exposure factors lead to an individual or subpopulation with very high
predicted exposure and risk; or at the tails of the uncertainty distributions, where certain possible
(though perhaps unlikely) combinations of parameter values or assumptions lead to significant exposure
and risk to all or some of the target population. Traditional methods for sampling variability and
uncertainty distributions may not provide adequate coverage of these tails. The Agency should thus
explore alternative methods, such as importance sampling, the use of mixtures of distributions, or the
decomposition of exposure and risk distributions into further subpopulations, for proper characterization
of the extremes of the distributions.
The routing and tracking of uncertainty across multiple, linked modules is not a simple task, with
few reported examples in either the basic or applied risk literature. Special attention should be given to
those uncertain parameters which affect more than one model component or module. Inferences about
the relative sensitivity and uncertainty of the fate and transport, exposure and risk modules, and the
interactions between their uncertainties, should be an important part of the learning that takes place as a
result of TRIM system applications.
-------
A key issue in uncertainty analysis, widely recognized though rarely addressed in a meaningful
manner, is that uncertainties in model formulations and conceptual assumptions are often much more
significant and important than uncertainties in model input values. To help promote the evaluation of
alternative model formulations and assumptions, the flexible architecture of the TRIM system should be
exploited to allow alternative formulations to be compared as part of the model sensitivity and
uncertainty analysis.
The issues of model uncertainty analysis and model evaluation are highly interrelated. While the
Subcommittee addresses specific issues related to plans for TRUVI.FaTE evaluation later in this report
(see Charge Question Ic under the TRUVI.FaTE Module), the more general relationship between
model evaluation and uncertainty should be recognized. Model evaluation must acknowledge and lead
to a characterization of the uncertainties that remain even after a model is properly formulated, coded,
fitted and tested to the fullest extent possible. Model uncertainty analysis can help guide critical needs
for further data collection and research to improve the model.
As acknowledged in the TRIM Status Report (p. 3-6), data are not typically measured or
collected in a way that allows for the separation of variability and uncertainty for most parameters.
Uncertainty is reduced primarily by good measurement methods that minimize and/or adjust bias,
increase sensitivity (by lowering detection or quantification limits), and improve precision. Model
evaluation and uncertainty analysis should thus be linked to ongoing laboratory- and field-data
collection efforts to better specify key model assumptions and input parameter values, both for the
model in general and for use in site-specific applications. The Subcommittee encourages the Agency to
provide ongoing guidance and support on the experimental and field-data collection efforts needed to
execute an effective application of TRIM, and to promote the development of an improved database
that can address long-term needs for model improvement and uncertainty reduction. The protocol for
use of the TRIM system should include an appropriate mix of requirements and incentives to see that
such critical data are collected and utilized. With this, the OAQPS FIAPs and criteria air pollution
program using the TRIM system could very well serve as a test case for studying the use of regulatory
models in applications with complex, rapidly evolving science, and the associated challenges of model
evaluation and uncertainty characterization. With the lessons learned, the effort could also serve as a
benchmark for other regulatory programs where similarly complex and uncertain models are considered
for use.
4.2 TRIM.FaTE Module
Charge 1: In response to specific recommendations by the previous SAB panel, we
(OAQPS) have investigated certain issues. As a result, TRUVI.FaTE has been modified (item
a, below), or the model documentation has been updated to address specific panel concerns
(items b and c, below).
-------
a) Do the modifications described in items (i) and (ii) below demonstrate an improvement
in the ability of TRUVLFaTE to address these two issues? If not, what specific
recommendations could you offer to do that?
(i) As a demonstration of the capability of TRIM.FaTE to incorporate output from
external models, the Agency has developed and implemented methods for
TRIM.FaTE to accept air concentrations and deposition output from an air
quality model.
(ii) TRIM.FaTE is revised with regard to how it addresses certain kinds of
chemical mass movement between certain compartments, specifically: (1)
dispersive transport between surface water compartments, and (2) diffusion
and advection between soil compartments and from air compartments to
surface soil compartments.
b) The incorporation of both horizontal and vertical atmospheric dispersion/diffusion
algorithms directly within TRIM.FaTE using both lateral and vertical Pasquill-Gifford
plume dispersion coefficients was pursued. After further review and consultation, it was
concluded that incorporating dispersion terms into the TRIM.FaTE air model was not
appropriate at this time. Do you concur with this conclusion (discussed more fully in the
model documentation)? What alternate methods for incorporating atmospheric
dispersion/diffusion directly within TREVLFaTE would you recommend?
The ability of TRIM.FaTE to use other air pollution models, better able to characterize near-
field (spatial) and short-term (temporal) variations in air concentrations and deposition, will be
important for a number of pollutants and applications where these effects are critical. Examples include
lead (or other metal or particle-associated chemical) concentrations and deposition near smelters, or
chemical plants or refineries where there are steep concentration gradients near the facility, or for which
acute impacts from toxic chemical spills, accidental releases or extreme meteorological conditions are of
concern. For these types of problems, the large-scale, compartmental mixing of the TRIM.FaTE
model is inappropriate. With the typical air-compartment sizes of the multi-media TRIM.FaTE model,
local scale diffusion is greatly overestimated; and it is very difficult to represent the smaller-scale spatial
and temporal variations that are important. Example calculations to illustrate this are presented in
Appendix A. Incorporation of diffusion within the TRDVLFaTE model (by the inclusion of exchange
flow between adjacent air compartments) will not address this issue, since the problem is not one of too
little local diffusion, but rather too much diffusion due to the inherent assumption of complete mixing
within each box of the air model. The use of existing and newer Gaussian-plume based air pollution
models, such as the Industrial Source Complex (ISC) family of models and its successors, can address
this problem for the air concentration and deposition calculations of TRIM.FaTE, but this approach
raises difficult issues of matching mass transfer between media (i.e., between the air and the adjacent
water, soil and biotic compartments). There is no immediately clear method for simultaneously fitting
10
-------
and executing the imported air pollution model and the rest of the TRUVLFaTE program to ensure
overall mass balance.
The OAQPS and its consultants have considered the difficult problems and tradeoffs associated
with using a separate air pollution model, which may make it impossible to ensure mass balance, versus
the use of the current TRUVLFaTE multimedia compartment approach, with its limited ability to
represent near-field and short-term concentration gradients. The OAQPS should continue to work this
problem from both ends, exploring creative mechanisms for linking separate air models to TRUVI.FaTE
while providing checks and corrections for mass balance where appropriate, and providing options to
use the standard TRUVI.FaTE compartment model with finer grid boxes and finer temporal resolution of
wind fields. However, assuming that drawbacks will remain with either approach, the ability to
prescreen problems, determine which approach is preferable, and guide users on the selection of an
appropriate option, or mix of options, is essential.
For now, the TRIM status report should better distinguish between artificial dilution associated
with complete mixing assumptions and numerical dispersion that results from discrete treatment of
advection. The status report should also distinguish between horizontal and vertical dispersion in its
discussion of the treatment of dispersion in traditional photochemical grid models. Many models use
fine vertical resolution and treat vertical dispersion relatively well. Horizontal dispersion is more difficult
to capture over varying spatial and temporal scales. For long-term model development, user guidance
and preprocessing tools should clearly distinguish between alternative problem types. For problems
where broad dispersal in the environment and intermedia transfers and partitioning control the
predominant pathways of chemical transport and exposure and chronic risks are the focus of concern,
the standard TRTM.FaTE module is appropriate. When significant near-source gradients occur, or
when short-term temporal concentration excursions in the air result in acute health risks, a separate and
more specialized air pollution model will be required. The eventual TRIM system preprocessor should
thus screen the chemical and landscape properties as well as the health and ecological effects of
concern to recommend to the user the most effective module options and configurations.
The chemical transport and intermedia mass transfer algorithms for TRUVI.FaTE can have
important implications for the distribution of chemicals in the environment (TREVLFaTE Module Charge
Question 1 (a) (ii)). While the fundamental algorithm for dispersive transport between surface water
compartments, tailored after that of the WASP model and based on exchange flow between adjacent
compartments, is correct, the TSD lacks adequate guidance for properly differentiating between, and
selecting dispersion coefficient values for horizontal and vertical dispersion in the water column and
vertical dispersion between the water column and adjacent sediment pore waters. Furthermore, by
assigning the same default value dispersion coefficient for surface water to surface water transport
(horizontal or vertical) and surface water-sediment exchange, the current documentation is misleading
and incorrect. Since the compartment model imparts numerical dispersion even when no exchange flow
is included, users also need guidance on the magnitude of this numerical dispersion and its tradeoff with
user-assigned dispersion; the model output should include a report on computed values of numerical
and total dispersion. More guidance is also needed for selecting appropriate advective flow rates for
11
-------
surface water systems. The advection and dispersion assumptions/parameters for a surface water
system are linked, since finer resolution of the advection field (in time and space) may preclude the need
for some dispersion, and because the intensity of mixing and the corresponding magnitude of the
dispersion coefficient may be a function of flow rates and water velocities. Finally, the standard
compartment mass balance model does not include a continuity calculation or check to ensure that a
fluid balance (water for the water compartments; air for the atmospheric compartments) is maintained,
and user inputted flows can thus, when in error, be unbalanced. Flow balance calculations should be
performed and users alerted when water and/or air balances are violated for model compartments.
This feature, along with the guidance noted above on the selection of appropriate, context-specific
dispersion coefficients and flow-dispersion relationships, should be a basic part of the user interface and
the TKDVI.FaTE technical support document (TSD). Citations should also be provided on accepted
approaches for determining more detailed and accurate advection and dispersion parameterizations for
surface water system, including the use of hydrodynamic models and the use of conservative tracer
measurements (such as salinity in coastal areas) to estimate the transport terms. Additional guidance on
dispersion coefficients appropriate for use in surface water systems is provided in Appendix B.
The proposed approach for coupling advection and diffusion within soil columns and air-surface
soil mass transfer, through the use of an effective penetration depth, is innovative. The method is
currently formulated assuming transport from the air to the soil; a similar parameterization should be
developed for cases where soil to air transport ensues. The current model does not appear to fully
account for impervious soils or the impervious built environment (roads, buildings, etc.). Appropriate
inclusion of such a compartment could be important for mass balance calculations in urban areas, and
may also allow eventual evaluation of FLAP impacts on materials and infrastructure damage.
c) An evaluation plan for TREVI.FaTE has been developed and is being implemented. Is
the proposed evaluation plan scientifically defensible? If not, what would you propose?
The overall evaluation plan for TRDVLFaTE, including code validation, mass balance checks,
model to model comparisons, and comparisons against observed data in field applications, is
appropriate. The Subcommittee has not yet received detailed reports on the results of these evaluation
efforts (some are not yet completed or are only in the planning phase1), so we cannot as yet comment
on whether each of these steps have been implemented in a proper manner and whether the inferences
drawn are justified. The Subcommittee looks forward to future presentation of the model evaluation
results and findings. In so doing, the OAQPS is encouraged to frame and present their evaluations in
the context of specific regulatory applications, e.g., an evaluation of the model in its intended use for
estimating residual risks after meeting specific emission reduction requirements. Also, the OAQPS
should be very explicit about the hypotheses that are being tested relative to the a priori expectations
or needs for model performance in these contexts. This will in most cases dictate tests involving
1 In response to a request from the Subcommittee at the review meeting, OAQPS did subsequently provide
a more detailed list of presentations and publications on the TRIM system. This list is provided in Appendix D. The
Subcommittee has not had the opportunity to evaluate these presentations and publications.
12
-------
combined and integrated applications of all three TRIM modules (fate and transport, exposure, and
risk), not just a single module or component.
The proposed evaluation tests should also include application of the sensitivity and uncertainty
analysis tools of the TRIM systems. The performance of these tools in these applications must
themselves be tested and evaluated and, as noted above, sensitivity and uncertainty analyses are (or at
least should be) closely integrated with the model evaluation exercise. Sensitivity analysis of model
parameters and component assumptions helps to identify where further improvements in model
formulation or input data reliability are essential. The model evaluation exercise should not just be
viewed as a one-time effort necessary to finalize an "approved" regulatory model. Rather, it should be
viewed as an ongoing process of system learning and model improvement. Similarly, many components
of the TRIM system model will require location-specific calibration and evaluation when used for site-
specific applications. As previously noted, the proposed plans for TRIM model evaluation, sensitivity
and uncertainty analysis are predicated on the availability of appropriate (unbiased, sensitive and
precise) field measurements and model inputs. The OAQPS should thus endeavor to develop a
protocol for "local" model evaluation as a standard and requisite part of these site-specific applications.
This protocol will likely include an outline of necessary steps for basic QA/QC of local applications to
ensure proper module selection and configuration, and the use of appropriate and justifiable values for
input parameters. It should also indicate which site-specific data are necessary for model evaluation
and provide an appropriate framework and incentive for collecting data that promotes long-term model
development and improvement.
The proposed TREVLFaTE evaluation effort, for a chlor-alkali plant with mercury emissions
impacting its surrounding landscape and ecosystem, provides an opportunity for the development of
such an exemplary protocol. As such, the Subcommittee recommends that the plan for TRUVI.FaTE
evaluation at this or other sites be expanded to include demonstration and testing of TRUVI.FaTE linked
with the other TRIM (exposure and risk) modules, and the use of sensitivity and uncertainty analysis
tools. The Subcommittee notes that selection of a mercury test case for evaluation imparts significant
challenges and difficulties. It is unlikely that sufficient field data are available to allow identification and
parameter estimation of the complex, kinetic reactions and processes that impact the multiple chemical
species of mercury included in the model. If data for other metals which are subject to simpler
transport and reaction mechanisms are available for the site, simultaneous model evaluation for these
constituents is recommended. If not, another site or sites where such chemicals are present should be
added to the evaluation exercise. This will increase the chances that a successful model specification
can be achieved, and help to identify the relationship between specific chemicals chosen for assessment,
environmental conditions, and the amount and quality of data necessary to calibrate, apply and evaluate
the model in a confident manner.
TRIM is intended for application to FLAPs and criteria air pollutants evaluation on a variety of
spatial scales, ranging from local, plant-specific estimates to large regional airsheds (e.g., the Los
Angeles metropolitan area). The prototype case study of the chlor-alkali plant involves only a few
emission sources over a relatively small spatial scale. It will be a significant challenge to apply TRIM to
13
-------
large metropolitan areas such as the LA basin with numerous stationary and mobile emission sources,
and planning should be initiated for testing such applications. In all cases, evaluation should include an
assessment of whether the TRIM.FaTE output is appropriately scaled and defined for input to
TRIM.Expo.
Charge 2: Do the newly added TRIM.FATE abilities to
a) follow transformation products temporally through the study period and allow for
'reverse transformation', and
b) track metals in a multimedia environment provide notable improvement in
TREVLFATE's usefulness for Agency air pollutant risk assessments? Have we
adequately captured the salient technical features in these additions? If not, what should
be included?
While reversible, kinetically controlled reaction processes can be important for some metals,
and the inclusion of this capability in TREVLFaTE should eventually provide a notable improvement in
modeling capability, our knowledge of, and ability to specify the parameter values for these processes
in real environmental settings is quite limited. As noted above, the metal for which such processes are
most likely to be important, mercury, has been the subject of significant study in recent years, but is still
poorly characterized. The current version of TREVI.fate incorporates a set of five parallel
transformation reactions for mercury oxidation-reduction and alkyl complexation that are described
mathematically using first-order rate expressions.
There are two principal issues of concern in the mercury speciation model. First, it is not
obvious that a rate-dependent approach is required in all cases. Generally speaking, trace inorganic
substances display variable rates of oxidation/reduction (which may be relatively slow), but rather rapid
complexation (seconds to minutes). While mercury may be an exception (reported methylation rate
constants in the Status Report are around 0.00 I/day), this is usually because of a rate limiting
elementary step in the reaction sequence, such as the generation of the methyl radical (which itself is
dependent on its source), rather than in the complexation step itself. Methylation is not especially
important for most inorganics (although complexation is), indicating that the current model includes a
degree of complexity that is not generally necessary. Since first-order kinetics are used, this issue can
be addressed using a straightforward Damkohler analysis, as is done effectively in the Vertical
Transport Algorithms (TSD Vol. El). One question the OAQPS may wish to address is whether or not
TREVI.fate should contain a generic description of inorganic fate, or instead utilize element-specific
submodules (if the uses of TRIM will encompass only a few elements, then perhaps the latter approach
preferred).
The rate coefficients presented in the TSD for the transformations between mercury species
appear to be chosen independently. In reality, significant interdependences exists among them (and
these interdependences can and should be evaluated as part of the model sensitivity and uncertainty
14
-------
analysis). The input table (Appendix C of the Status Report) shows four coefficients for six separate
compartments, for a total of twenty-four constants needed to execute the mercury submodel. It seems
unlikely that a complete set of such coefficients will ever be available for a given application, creating
the danger that users will choose coefficients from many different literature sources that, in aggregate,
are neither internally consistent (i.e., consistent with Hg equilibria), nor descriptive of Hg transformation
rates at the site of interest. This relates to the problem noted above of the impossibility of being able to
establish an unambiguous set of Hg coefficients. Given this, a more parsimonious model is probably
called for in this case. While efforts to better characterize reversible reactions for metals such as
mercury should continue, similar effort should be exerted in characterizing the simpler, equilibrium
partitioning behaviors of other metals and their dependence on environmental conditions and settings
related to factors such as pH and redox conditions.
Charge 3: Does the addition to TRUVI.FaTE of certain seasonal processes (i.e., litterfall and
changes in plant uptake) represent an improvement to the model's ability to predict different
media pollutant concentrations for the time frames of interest for risk assessment? Are there
other processes related to seasonality which play an important role in pollutant transfers and
resulting media concentrations for relevant time frames, and thus would be a valuable addition
to the model?
The inclusion of seasonal processes related to plant growth cycles is a valuable addition to
TRUVI.FaTE, and other seasonal processes may be important as well, especially related to water
balances and flows through soils and surface waters. Seasonal variations in precipitation, runoff,
infiltration and evapotranspiration can affect soil moisture and subsequently impact air-soil mass
transfers and chemical processes in the subsurface. Seasonal (or intermittent) variations in streamflow
can affect water column-sediment transport and chemical partitioning, as well as surface water
concentrations for uptake by biota and waterborne exposures. At a minimum, user guidance is needed
to ensure that seasonal and/or intermittent patterns of soil moisture and streamflow are consistent with
inputted values of temperature and precipitation. If temporal variations in soil moisture and streamflow
are found to be important for certain chemicals in certain applications, it may be necessary to interface
TRUVI.FaTE with a water balance model (i.e, a soil and/or stream hydrologic model) to ensure that
these inputs are properly specified.
The ability of the TRIM system to interface with, and accept output from, others model should
facilitate this approach; OAQPS may wish to begin to identify appropriate water balance models for an
application of this type. For seasonal processes of a more fundamental nature, affecting the structure
and rate of chemical uptake or transformation processes in specific environmental media or settings
(such as nonlinear algal growth or food chain uptake in surface waters, or more complex soil-to-plant
uptake), the ability of TRUVLFaTE to interface with more detailed, media-specific models, can be
exploited. For example, in cases where surface water processes have an important influence on
chemical fate, ecological risk and human exposure, an integrated surface water modeling system, may
be appropriate. Again, the ability to prescreen problems and guide users to appropriate component
modules and system configurations will be critical.
15
-------
4.3 TRIM.Expo
Charge 1: Does the proposed conceptual design for TRIM.EXPO capture the salient
technical features necessary to characterize inhalation and ingestion exposures given our
programmatic and regulatory needs? If not, what salient features are missing and how should
they be included? and
Charge 2: Have we drawn appropriately from the existing modeling science on inhalation and
ingestion exposures and are the specific algorithms structured in a way that is consistent with the
modeling framework?
The Subcommittee finds that the conceptual design of TREVLExpo and its use of existing
modeling science for inhalation and ingestion exposures are, in general, consistent with OAQPS'
programmatic and regulatory needs, the overall TRIM modeling framework, and the state-of-the-art of
exposure assessment. However, the Subcommittee also notes that the exposure assessment discipline
is young and rapidly evolving, and that a "one-size-fits-all" approach (even one that is very flexible) may
not work for all chemicals in all exposure settings. Very different approaches may be needed for
assessing long-term, chronic exposures vs. short-term acute impacts. These may significantly impact
the type and level of detail necessary in characterizing individual and subgroup variability in time-activity
patterns and exposure factors. In certain cases, more refined differentiation of cohort characteristics
may be appropriate, while in others, greater emphasis is needed on individual-to-individual variation
within a cohort. Greater aggregation of time-activity patterns is acceptable when computing chronic
exposures, a finer definition of exposure scenarios may be necessary to identify the upper tail of acute
exposures and risks. Once again, an effective pre-screening of problem features, including chemical
properties, landscape-exposure characteristics, and the time and spatial scales of health impacts, is
needed to best specify the level of temporal and spatial aggregation in the exposure model. A
mechanism for such pre-screening and user guidance is needed in the TRIM system.
A number of significant challenges and uncertainties remain in collecting the necessary
information for a meaningful and reliable exposure assessment. OAQPS and it consultants are clearly
aware of these challenges. These include difficulties in extrapolating short-term time-activity studies to
estimate longer-term behavior (and exposure); a lack of information on diet source and the relative role
of home-grown, local and national or international sources of foods for a given local pollution scenario;
and the appropriate chemical concentrations of different food sources to assign, particularly when issues
of cumulative risk begin to be addressed. The relative importance of these issues will hopefully be
illuminated as evaluation studies of the overall TRIM system (and TREVLExpo) evolve. Test
applications are encouraged for a problem such as mercury or other metal exposure occurring through
a variety of environmental pathways, as well as a HAP problem with a predominant air exposure
pathway, perhaps using a NHEXAS study area (and its associated data set) for model calibration and
evaluation.
4.4 TRIM.Risk
16
-------
Charge 1: Does the overall conceptual approach described for the TRIM.Risk module
adequately capture the output from the other TRIM modules and provide quantitative human
health and ecological risk characterization information that is consistent with Agency risk
characterization policy?
The overall plan and conceptual approach for TRIM.Risk appears to be appropriate; since it is
in the early stage of development, more details and example applications will be necessary for a more
rigorous evaluation. There are some issues that the Subcommittee has identified as possible concerns
for the ongoing module development. These include:
a) Risk Analysis of Acute Health Effects: For human cancer risk assessment, the time
frame is defined as the life span of 70 years. The 70-year risk estimate is divided by 70
to generate the annual incidence rate. For non-cancer acute health effects caused by
many hazardous and criteria air pollutants involving short-term or peak exposures, no
time scale is defined for performing the risk calculation. It is intended that TREVLFaTE
will generate hourly, daily and yearly concentration estimates. There is currently a lack
of guidance on how to integrate short-term and/or peak exposure estimates for
calculating the risk of acute health effects.
b) Adequacy of Additive Model: An additive model is chosen to estimate risk for
simultaneous exposures to several carcinogens (page 9-8 of the TRIM Status Report).
This is a reasonable and logical choice for general representation. However, there are
important cases where synergistic effects are believed to occur when there is
simultaneous exposure to two carcinogens. Examples include smokers exposed to
asbestos or radon. Flexibility to allow for synergistic or antagonistic effects of multiple
contaminants will be important for such cases (see also our more general comments on
chemical mixtures and cumulative risk effects below).
c) Interpretation of Hazard Index (HI} and Hazard Quotient (BO): The HQ is defined as
the ratio of computed exposure to a reference level and the HI (for a mixture of
compounds) is calculated as HI = HQ1 + HQ2 + ...+HQi. As such, an additive model
is again assumed for multiple chemical effects. Both the HQ and the additive HI have
been challenged as appropriate indicators for noncancer risks. As acknowledged on
page 9-8 of the TRIM Status Report, if comparisons of HQs across substances may
not be valid, then use of the HI, defined as the summation of HQs, is even more
questionable.
d) Indicators and Descriptors for Ecological Risks: No general guidance is given in the
conceptual TRIM.Risk module to calculate, analyze, express and interpret ecological
risks. It is unclear what kind of outputs will be generated by TREVLRisk to characterize
17
-------
ecological risks. This problem is not unique to the TRIM system, since (similar to
methods for noncancer human health risk assessment) current methods and accepted
protocols for ecological risk assessment are limited. It will thus be difficult to define
endpoints and metrics of impact and risk that are generally and widely accepted.
The Subcommittee does commend OAQPS' ongoing efforts to link with other multimedia
exposure and risk assessment activities in the Agency, such as the Hazardous Waste Information Rule
(HWIR) being developed in the Office of Solid Waste and Emergency Response (OSWER) and the
Multi-media Integrated Modeling System (MEVIS) being developed by the National Exposure
Research Laboratory (NERL) in the Office of Research and Development. These collaborations are
important to ensure that maximal benefit, consistency and coordination is achieved between these
efforts.
The TRIM system provides the opportunity to begin to seriously address new and emerging
issues in risk assessment and characterization, including the evaluation of mixtures, cumulative risk,
environmental equity and methods for considering more-susceptible subpopulations. This opportunity is
especially applicable to the assessment of the residual risk associated with multiple HAPs under
different emission control scenarios. While the Agency continues to struggle with how best to address
and incorporate these new and emerging ways of looking at risk, OAQPS should be proactive in
suggesting and exploring options within its integrated modeling system. Often the efficacy of new ways
of addressing problems cannot be judged until they are first tried and illustrated. The OAQPS should
utilize this opportunity to see whether the TRIM system in the context of HAP assessment can begin to
provide such examples for broader Agency consideration and evaluation.
The framework for ecological risk assessment in TREVI.Risk (and the state-of-the-art in
general) is quite new and undeveloped. Clear measures of ecological impact must be identified and
evaluated. Currently, these focus on inferences of toxicity and hazard extrapolated from limited tests on
individual organisms in isolation. As new knowledge emerges on population-level ecosystem effects,
attempts should be made to incorporate this science. To begin to explore such options, OAQPS
should review available population-level ecosystem models to see which may be able to utilize output
from, or be compatible with, the TRIM system.
18
-------
5. CONCLUSIONS
The Environmental Models Subcommittee (EMS, or "the Subcommittee") of the Science
Advisory Board's (SAB) Executive Committee met on December 13 & 14, 1999 to review the
Agency's "Total Risk Integrated Methodology (TRIM) Status Report," as well as the draft
"TREVI.Expo Technical Support Document," and the two-volume TRUVLFaTE Technical Support
Document" developed by the EPA's Office of Air Quality Planning and Standards. The Subcommittee
conducted this review in order to provide the Agency with Advice and insights on the adequacy of the
proposed TRIM approach to predict exposures and risks posed by air pollutants.
The Subcommittee concluded that:
1) The Overall TRIM Efforts to Date Have Been Innovative and Effective - The EPA
OAQPS' efforts to develop TRIM as a flexible, state-of-the-art system for evaluating multimedia
chemical fate, transport, exposure and risk, have to date been innovative and effective. The modular
design and open architecture of TRIM provide a flexible and appropriate means of accommodating
new scientific information;
2) There is a Significant Need for User Guidance and Support - Significant user guidance
and support will be necessary to ensure that appropriate model setups and configurations are selected;
3) There is a Need to Better Specify Plans and Timeline - There is a need for OAQPS to
better specify its plans and timeline for use of the TRIM system within the Agency and subsequent
release for use by a broader user community;
4) Early Interactions and Involvement with Stakeholders . Workshops and Beta
Testing is Essential - While the Subcommittee recognizes that much further work is needed before
the model will be ready for use for OAQPS's regulatory needs, early interactions with, and involvement
of, affected stakeholders are essential steps in the process. Early workshops and beta testing of the
integrated TRIM system by the affected user community are thus recommended to ensure the
appropriate level of stakeholder input and a well-designed product;
5) Efforts to Improve TRCVLFaTE Module are Appropriately Focused - Efforts to
improve the TRTM.FaTE module to address local-scale atmospheric dispersion through the use of
alternative air pollution models, to develop appropriate transport parameters for surface waters and air-
soil mass transfer, to include kinetic, reversible reactions among metal species, and to incorporate
seasonal process in biotic processes, are appropriately focused. The planned field comparison study of
mercury should be reconsidered or augmented to include metals that undergo simpler fate and transport
processes in the environment to ensure that a successful model specification can be achieved;
19
-------
6) Testing of TRIM System at Larger Scales is Recommended - Tests of the TRIM
system at larger scales, such as for a metropolitan air basin with multiple stationary and mobile sources,
are also recommended to evaluate the ability of the TRIM system to operate at this scale of application
and to utilize emerging electronic/GIS databases with geographic, meteorological, emissions,
population and environmental resource information.
7) Design of TRCVLExpo is Consistent with OAOPS' Programmatic and Regulatory
Needs and the Current State-of-the-Art of Exposure Assessment - The design of TREVI.Expo
and its use of existing exposure science for inhalation and ingestion modeling are, in general, consistent
with OAQPS' programmatic and regulatory needs and the current state-of-the-art of exposure
assessment;
8) Design of TRLVLRisk Module Appears Appropriate. But Needs More Example
Applications for Rigorous Review - The design of the TREVI.Risk module appears to be
appropriate, though more details and example applications will be necessary for rigorous review;
9) Need to Ensure That Model Linkages are Appropriately Evaluated - Evaluation
plans should consider simultaneous test applications of the TRIM modules (FaTE, Expo and Risk) to
ensure that model linkages are properly specified; and
10) Should Motivate Improved Data Collection Methods. Improved Data Bases for
Model Input and New Approaches to Emerging Risk Issues - The application protocol for TRIM
should provide incentives for the development of improved data collection methods and improved
databases for model input. Because the multi-chemical, multi-media design of the TRIM system
provides an opportunity to explore methods for addressing new and emerging issues in risk assessment
and the risk sciences, including the effects of mixtures, population susceptibility and cumulative risk; as
well as metrics for environmental equity and ecological impacts at the population level, OAQPS should
suggest and illustrate possible approaches to these issues for broader Agency consideration and
evaluation.
20
-------
REFERENCES
BASIN. 1998. Better Assessment Science Integrating Point and Nonpoint Sources, EPA-823-B-98-
006
BIER, V.M., Y.Y. Haimes, J.H. Lambert, N.C. Matalas and R. Zimmerman. 1999. A survey of
approaches for assessing and managing the risk of extremes. Risk Analysis, 19: 83-94.
CHNHYD . 1982. Channel Hydrodynamic Model for simulating flows and water surface elevation in a
stream river network, ORNL 5701
CHNTIN . 1983. Channel Transport Model for simulating sediments and chemical distribution in a
stream river network, ORNL 5882
Clark, C.E. 1961. Importance sampling in Monte Carlo analyses. Operation Research, 9:603-620.
EPA. 1999. Letter to Dr. Daisey in response to the Science Advisory Board's July 30, 1999 report on
the White Paper on the Nature and Scope of Issues on Adoption of Model Use Acceptability
Criteria (MAC White Paper) from Assistant Administrator Norine Noonan, October 22, 1999.
ICM .1981. An integrated compartment method for numerically solving partial differential equations,
ORNL 5684
MATTUM. 1983. A Multidimensional Model for simulating moisture and thermal transport in
unsaturated porous media, ORNL, 1983.
Ripley, B.D. 1987. Stochastic Simulation. John Wiley, New York.
Titterington, D.M., A.F.M. Smith and U.E. Makov. 1985. Statistical Analysis of Finite Mixture
Distributions. John Wiley, Chichester, UK.
R-l
-------
APPENDIX A - ILLUSTRATION OF POTENTIAL
MISREPRESENTATION OF DISPERSION IN TRIM.FaTE
The TRIM.FaTE model is designed to be applied with low resolutions in both space and time,
similar to related multimedia compartment models, such as CalTOX and ISMCM. The numerical
algorithms effectively employ a first-order upwind finite-difference method with respect to the advective
transport. As a result, there is a constraint on the sizes of spatial grids and time steps. The algorithm
yields a numerical diffusion coefficient, DN, given by:
DN = kV AZ [1 - ?(Ai / AS)]
where k is a constant depending on the dimensionality (for 1-D problems, k = %), V = velocity, AT =
Ti(j,s-aTs:r ai^s, av5 AS = spatial grid size. Clearly, the mesh Courant number Cr = VAt/AS = Tt (
where Tt is the transfer factor defined in TRIM.FaTE) must be less than or equal to 1 to make the
numerical diffusion coefficient positive. In practice, it should not be much greater than 1 so that the
simulation results look reasonable (whether the simulations are accurate or not is another issue). If
explicit integration is used, then the Courant number must be less than or equal to 1 to ensure numerical
stability. Furthermore, to make numerical simulations close to the mathematical representation of the
advective processes, DN should be less than the physical diffusion/dispersion coefficient. This implies
that the users must design the time-step and spatial-grid sizes such that the mesh Courant number is
close to 1.0 for all compartments (an extremely difficult, if not impossible task, under space- and time-
varying velocity fields).
Consider an air compartment as an example. Assume a typical wind velocity of 10 m/s. If we
use a time-step size of 1 year, 1 month, 1 day, and 1 hour, the compartment size must be greater than
310,000 km, 26,000 km, 864 km, and 36 km, respectively. The use of large compartment sizes would
yield very large numerical diffusion coefficients, unless the mesh Courant number can be made exactly
equal to 1.0 for all compartments (an impossible task). If we are able to design the compartment sizes
such that the mesh Courant number on average is 0.5, then the numerical diffusion coefficients would be
on average equal to 3,100,000,000 m2/s, 260,000,000 m2/s, 8,640,000 m2/s, and 36,000 m2/s,
respectively. Thus, even for a time-step size of one hour, the numerical diffusion would be significantly
overestimated (short-term diffusion coefficients in Gaussian plume models increase with spatial distance
and scale, but are typically on the order of 100 - 5000 m2/s), yielding simulations that may appear
reasonable, yet are very inaccurate (predicting concentrations possibly several orders-of-magnitude
smaller than the true solutions in terms of peak concentrations). Faced with this problem, users can
either (1) use high resolution in both spatial-grid and time-step sizes to yield accurate simulations (this
would require excessive computational resources with TRIM.FaTE); or (2) use a coarse resolution in
compartment sizes and time-step sizes as done in CalTox or ISMCM, overlooking the resulting
inaccuracy in predicted concentrations.
A-l
-------
In summary the inability of TREVLFATE to numerically produce an appropriate mathematical
representation when coarse compartments and large time-step sizes are used is a major limitation. The
predicted concentrations in simulations using TRIM.FaTE could be several orders-of-magnitude smaller
(with respect to peak concentrations) or higher (with respect to small concentrations) than the true
solutions.
A-2
-------
APPENDIX B - MORE DETAILED COMMENTS ON THE
PARAMETERIZATION OF DISPERSION AND ADVECTION FOR
SURFACE WATER COMPARTMENTS IN TRIM.FaTE
The dispersion between surface water compartments in TRIM.FaTE is modeled after the
methodology in WASP (Ambrose, 1993). This appears adequate for TRIM.FaTE simulations. The
importance of including a physical dispersion process in the exchange between surface water
compartments will depend directly on the size (spatial resolution) and type (river vs. estuary) of the
compartments and whether they contain external sources (i.e., whether the numerical dispersion
overwhelms the physical dispersion). The suggested tiered approach to determining model sensitivity
can be used to assess the importance of the contribution of physical dispersion to the overall distribution
of mass among the various model compartments.
Dispersion Coefficients in Surface Water Compartments and Between Surface Water and Sediment
Compartments:
On pp. 4-9 to 4-11 of Vol. II of TRIM.FaTE TSD, the default dispersion coefficient values for
these two processes (i.e., water to water dispersion, and water to sediment dispersion) are the same,
listed as 2.25 x 10"4 m2/day. Actually, there are three dispersive processes involved: horizontal
dispersion and vertical dispersion in the water compartment and the dispersion between the sediment-
water interface. Normally, horizontal dispersion coefficients in the water column are several orders of
magnitude greater than the dispersion coefficients in the vertical direction, which in turn are several
orders of magnitude greater than the dispersion coefficients across the sediment-water interface.
Typical values are:
Horizontal dispersion in water column: 1-300 m2/s
Vertical dispersion in water column: 10"5 - 10"3 m2/s
Dispersion across sediment-water interface: 10"10 - 10"9 m2/s
Horizontal dispersion coefficients include longitudinal and lateral dispersion. Usually, longitudinal
dispersion coefficients are greater than lateral dispersion coefficients.
For the TRIM.FaTE model, we suggest that ranges of dispersion coefficient values should be provided
for use as default values.
Derivation ofAdvective Transport Coefficients:
The TSD document does not indicate how to derive advective transport coefficients such as the
bulk transport. In general, two methods can be used: direct field measurements and using a
hydrodynamic model. Note that being a box-model, the WASP modeling framework requires the
assignment of the mass transport coefficients by the user. Hydrodynamic models are available for 1-D,
B-l
-------
2-D, or 3-D configurations of a waterbody to develop mass transport coefficients, however, this results
in a significant increase in model complexity and input data requirements.
In estuaries and coastal waters, tidal currents are at least one order of magnitude greater than
tidally averaged velocities.
B-2
-------
APPENDIX C - SPECIFIC EDITORIAL COMMENTS ON THE
TRIM.Expo and TRIM.Risk MODULES
Page 4-3, on one side of equation 4-1 the exposure variable seems to only be dependent on (t, z, and
m) while on the other side of the equation the other two variables include a number of other
"indices". Perhaps using the "conditional" notation "|" could clarify?
In Sections 4.1.2 to 4.1.4, how would acute exposure scenarios be described?
See Table 4-1 page 4-13. Is this too ambitious?
Page 4-29: Do we lose the potential dependency between the original parameters and outputs of
subsequent modules if the outputs of the intermediate modules are summarized into percentile
distributions and used as inputs into the subsequent modules?
Using short-term data to project to longer term: "Blanket statement" at bottom of page 4-29: "OAQPS
believes that these data are the best available and extrapolating to generate long-term
factors or creating a longer term sequence of such data is the best approach " This should
be referenced with a description of the studies conducted and any efforts to "validate" this
statement.
Page 5-3, different exposure event sequences are selected each day, how about repetitive tasks, (e.g.,
wake up, shower, breakfast, commute to work, etc...)? Is the same "exposure" used for all
people in a cohort?
Chapter 6:
There appears to be some inconsistency in the description of contaminant exposure pathways in
the TRIM.Expo module as compared to exposure pathways described in the TKDVI.FaTE module
description. This probably reflects the earlier stage of development of the TREVLExpo module as
compared to the TREVI.FaTE module. However, it is important that the modules are consistent.
Examples include:
Sections 6.2.3 and 6.2.4: In the discussion of ingestion pathways for foodstuffs it is not clear if
root-uptake of pollutants, and subsequent translocation to above-ground plant parts is factored
into the exposure pathway model. There is a short description of the parameter, Kow, the
octanol/water partition coefficient, but is not stated if this parameter is used or not. The
implication is that it is not. However, in the TREVLFaTE Technical Support Document
(Volume n) biotic transport algorithms for root-uptake are defined, suggesting this pathway is
considered
C-l
-------
Page 6-7. Table 6-2: In the taxonomy of food types categorized as fruits, vegetables and grains
(Table 6-2), beans are listed as protected, above-ground crops. This is not true for all types of
beans. Some commonly eaten beans such as, french beans and mange-tout, are essentially
unprotected because the pod is also consumed. Later in section 6.2.3. 1 string beans are listed
in the exposed produce category.
Page 6-15: The exposure pathways shown in Figure 6-4, and the descriptions in the supporting
text appears somewhat inconsistent. It would be helpful to insert a section that deals with
pasture because this is a key component to the livestock and animal produce exposure
pathways. Currently, in section 6.2.3.2 on dairy products, pasture is defined as all foodstuffs
grown on the farm to feed the animals (e.g., open pasture grass, grain, corn). These are a
combination of different food categories that are expected to have different contaminant
concentrations. Frequently, there are seasonal differences in the relative contributions of these
categories to the livestock's feed. Therefore, it would seem more appropriate to explicitly
identify the different categories of fodder in the model.
In Figure 6-4. the exposure pathway for chickens is shown to come from pasture, rather than
grains that represent a more realistic exposure pathway. Using the current definition for
pasture, pasture may in fact refer to grains, which is confusing because grains are shown as a
separate category in the figure and supporting text. A more transparent approach, as described
above, is recommended.
C-2
-------
APPENDIX D - UPDATED LIST OF TRIM PRESENTATIONS AND
PUBLICATIONS
1) Peer Reviewed Publications
Palma, Ted., Amy B. Vasu, and Robert G. Hetes, 1999. "Total Risk Integrated Methodology
(TRIM)", Air and Waste Management Association - EM Magazine, March 1999, Pg 30-34.
2) External Scientific Meetings
20th Annual Meeting of the Society of Environmental Toxicology and Chemistry. 14-18
November, 1999. Philadelphia, PA.
Efroymson, R.A., Jones, D.S., Vasu, A. An Ecological Risk Assessment Methodology for
Hazardous Air Pollutants.
International Life Sciences Institute's "Aggregate Exposure Assessment Model Evaluation
and Refinement Workshop". 19-21 October, 1999. Baltimore, Maryland.
Richmond, H., McKone, T. Total Risk Integrated Methodology (TRIM).
19th Annual Meeting of the Society of Environmental Toxicology and Chemistry. 15-19
November, 1998. Charlotte, NC.
Vasu, A.B., Hetes, R.G., Palma, T., McKone, T.E. TRIM: A Multimedia, Multipathway
Framework for Assessing Human and Ecological Exposure and Risk.
91st Annual Conference of the Air and Waste Management Association. 14-18 June, 1998.
San Diego, CA.
Palma, Ted, Amy B. Vasu, and Robert G. Hetes. An Introduction to the Total Risk
Integrated Methodology (TRIM).
Annual Meeting of the Society for Risk Analysis. December 7-10,1997. Washington, DC.
A.B. Vasu, T.R. Johnson, T.E. McKone, T. Palma, M.G. Dusetzina. Introduction to the
Total Risk Integrated Model.
T. R. Johnson, G.W. Suter H, T. Palma. EcoFaTE: A TRIM Module for Linking
Multimedia Environmental Systems with Biotic Domains.
D-l
-------
S. Guha T.E. McKone, D.H. Bennett, B.F. Lyon^4 Generalized, Mass-Conserving
Multimedia Transport and Transformation Model: Development of Transfer Factors.
B.F. Lyon, S. Guha, T.E. McKone. TRIM: Mathematical and Numerical Aspects.
D.H. Bennett, T.E. McKone, M.G. Dusetzina. Building Uncertainty and Sensitivity Analysis
into the TRIM Framework.
R.A. Efroymson, B.E. Sample, B.F. Lyon, G.W. Suter II, C.T. Hunsaker, A.W. Simcock, and
D.H. Bennett. A Dynamic Model for Terrestrial Ecological Exposure to Toxic Air
Pollutants.
R.D. Zimmer, J.Dee, D.S. Jones. Predicting Pollutant Mass Transfers within Aquatic
Systems using the EcoFaTE Model..
T.E. McKone, Ted Johnson, G.W. Suter II. Estimating Multi-pathway human and ecosystem
exposures within the Total Risk Integrated Model (TRIM).
C.T. Hunsaker, B.F. Lyon, T.E. McKone, V. Kilaru, A. W. Simcock. EcoFaTE Model
Parameterization and Verification.
18th Annual Meeting of the Society of Environmental Toxicology and Chemistry. 16-20
November, 1997. San Francisco, CA.
T. McKone, S. Guha, T. Johnson, B. Lyon, G. Suter, A. Vasu. A Multimedia Health and
Ecological Risk Assessment Methodology for Air Pollutants.
R. Zimmer, J. Dee, D. Jones, B. Sample, G. Suter. Use of EcoFaTE in Modeling Pollutant
Mass Transfers and Subsequent Risks within Aquatic Systems.
R. Efroymson, B. Sample, C. Hunsaker, B. Lyon, A. Simcock, and G. Suter. A Dynamic
Model for Terrestrial Ecological Exposure to Toxic Air Pollutants.
D-2
-------
APPENDIX E - ACRONYMS
BASIN Better Assessment Science Integrating Point and Nonpoint Sources
CHNHYD Channel Hydrodynamic Model for Simulating Flows and Water Surface Elevation in a
Stream River Network
CalTox A California EPA Total Exposure Model for Hazardous Waste Sites
CHNTIN Channel Transport Model for Simulating Sediments and Chemical Distribution In a
Stream River Network
cr
D
Dn
EC
EMS
EPA
HAP
Hg
ffl
HQ
HWIR
ICM
ISC
ISMCM
Courant Number
Dimensions (1-D, 2-D, or 3-D) (Configurations for models)
Diffusion Coefficient
Executive Committee (U.S. EPA/SAB/EC)
Environmental Models Subcommittee (of the U.S. EPA/SAB/ EC)
Environmental Protection Agency (U.S. EPA, or EPA)
Hazardous Air Pollutant
Mercury
Hazard Index
Hazard Quotient
Hazardous Waste Identification Rule
Integrated C ompartment Method
Industrial Source Complex family of models and its successors
Integrated Spatial Multimedia Compartmental Model (USCLA/Sch
and Applied Science)
E-l
-------
km
Kilometer
MAC
Model Acceptability Use Criteria
MATTUM Multidimensional Model for simulating moisture and Thermal Transport in Unsaturated
Porous Media
m
Meter
MIMS Multi-Media Integrated Modeling System
NERL National Exposure Research Laboratory (U.S. EPA/NERL)
OAQPS Qfficeof Air Quality Planning and Standards (U.S. EPA/OAQPS)
SAB Science Advisory Board (U.S. EPA/SAB)
S Spatial Grid Size
TRIM Total Risk Integrated Methodology (Includes TRIM.Expo, TRIM.FaTE, and
TREVLRisk)
TRIM.Expo Total Risk Integrated Methodology Exposure Module
TREVLFaTE Total Risk Integrated Methodology Fate Module
TRIM.Risk Total Risk Integrated Methodology Risk Module
TSD Technical Support Document
V Velocity
WASP Waste Area S P model (- see TRIM.FaTE discussion)
E-2
-------
United States Science Advisory EPA-SAB-EC-ADV-00-004
Environmental Board (1400A) May 2000
Protection Agency Washington DC iviviv.epa.gov/sab
&EPA AN SAB ADV|SORY ON
THE AGENCY'S "TOTAL
RISK INTEGRATED
METHODOLOGY" (TRIM)
Prepared by the Environmental
Models Subcommittee of the
Science Advisory Board
------- |