EPA530-D-03-001a
July 2003
SAB Review Draft
/ A Multimedia, Multipathway, and
Multireceptor Risk Assessment
(3MRA) Modeling System
Volume I: Modeling System and Science
-------
EPA530-D-03-001a
July 2003
SAB Review Draft
Multimedia, Multipathway, and
Multireceptor Risk Assessment
(3MRA) Modeling System
Volume I: Modeling System and Science
prepared by
U.S. Environmental Protection Agency
Office of Research and Development Office of Solid Waste
National Exposure Research Laboratory Washington, D.C.
Athens, GA
and
Research Triangle Park, NC
-------
This document is the first volume of a five-volume set. This volume describes the
conceptual design, scientific rationale, and supporting data that are the foundation for the 3MRA
modeling system. Volume II describes the data developed and used to run the 3MRA modeling
system. Volume III describes the approach to quality assurance, including verification and
validation activities ranging from extensive peer reviews to multimedia model comparisons.
Volume IV describes the methodology used to evaluate sensitivity of model parameters and
characterize different types of uncertainty in the 3MRA modeling system. Volume V describes
the technology design and includes a user's guide.
-------
Table of Contents
Table of Contents
Section Page
List of Figures viii
List of Tables x
1.0 Introduction 1-1
1.1 Background 1-1
1.2 Overall Design Goals of the 3MRA Modeling System 1-4
1.3 Overview of the 3MRA Modeling System 1-9
1.4 Overview of the 3MRA Science Module Architecture 1-12
1.5 Overview of Results Generated by the 3MRA Modeling System 1-15
1.6 Organization of This Document 1-19
1.7 References 1-21
2.0 Modeling Approach 2-1
2.1 Overview and Conceptual Approach 2-1
2.2 Spatial and Temporal Scale 2-6
2.2.1 Model Spatial Scale 2-6
2.2.2 Model Temporal Scale 2-8
2.3 Design of the 3MRA Modeling System Multipathway Modules 2-11
2.3.1 Chemicals in the 3MRA Modeling System Database 2-11
2.3.2 Sources (WMUs) 2-15
2.3.3 Transport Media, Fate Processes, and Intermedia Contaminant Fluxes 2-19
2.3.4 Food Chain/Food Web Components 2-21
2.3.5 Human Exposure and Risk 2-24
2.3.6 Ecological Receptors, Exposure Pathways, and Risk Measures 2-28
2.4 References 2-32
3.0 Spatial Aspects of Environmental Settings and Receptors 3-1
3.1 Overview of Spatial Layout for Environmental Settings and Receptors 3-1
3.1.1 Settings and Areas of Interests 3-1
3.1.2 Site Layout and Spatial Data Layers 3-1
3.2 Watersheds and Waterbodies 3-5
3.2.1 Definitions for Watershed and Waterbody Layout 3-7
3.2.2 Assumptions for Waterbodies and Regional Watersheds 3-8
3.2.3 Assumptions for Local Watersheds 3-8
3.2.4 Watershed Soils 3-9
3.3 Human Receptors, Farms, and Wells 3-9
3.3.1 Resident and Home Gardener Locations 3-10
3.3.2 Farm Locations 3-11
3.3.3 Well Locations (Private Wells Only) 3-11
3.3.4 Recreational Fisher Locations 3-11
in
-------
Table of Contents
Table of Contents (continued)
Section Page
3.4 Habitats and Ecological Receptor Placement 3-12
3.4.1 Habitat Delineation 3-12
3.4.2 Assignment of Receptors to Habitats 3-16
3.5 Summary 3-17
3.6 References 3-17
4.0 Wastewater Source Modules 4-1
4.1 Purpose and Scope 4-1
4.2 Conceptual Approach 4-2
4.2.1 Description of Waste Management Units 4-3
4.2.2 Calculate Constituent Concentrations within the Unit 4-6
4.2.3 Calculate Solids Concentrations within the Unit 4-10
4.2.4 Calculate Volatile Emission Rates 4-12
4.2.5 Estimate Resuspension, Sedimentation, and Burial Velocities 4-13
4.2.6 Estimate Constituent Release in Leachate
(Surface Impoundments only) 4-19
4.2.7 Adjust for Temperature Effects 4-23
4.3 Module Discussion 4-25
4.3.1 Strengths and Advantages 4-25
4.3.2 Uncertainty and Limitations 4-26
4.4 References 4-27
5.0 Land-based Source Modules 5-1
5.1 Purpose and Scope 5-1
5.2 Conceptual Approach 5-5
5.2.1 Description of WMUs 5-5
5.2.2 The Generic Soil Column Model 5-9
5.2.3 Local Watershed Model 5-15
5.2.4 Particulate Emissions Model 5-22
5.3 Module Discussion 5-24
5.3.1 Strengths and Advantages 5-24
5.3.2 Uncertainty and Limitations 5-26
5.4 References 5-28
6.0 Air Module 6-1
6.1 Purpose and Scope 6-1
6.2 Conceptual Approach 6-2
6.2.1 Characterize Source-specific Parameters 6-5
6.2.2 Calculate Receptor Locations (polar-grid or site-specific) 6-6
6.2.3 Calculate Receptor-specific Concentration and Deposition Rates .... 6-6
6.2.4 Calculate Constituent-specific Annual Average Concentrations and
Deposition Rates 6-8
-------
Table of Contents
Table of Contents (continued)
Section Page
6.3 Module Discussion 6-8
6.3.1 Strengths and Advantages 6-8
6.3.2 Uncertainty and Limitations 6-9
6.4 References 6-9
7.0 Watershed Module 7-1
7.1 Purpose and Scope 7-1
7.2 Conceptual Approach 7-2
7.2.1 Calculate Soil Contaminant Concentrations and Surface Water
Loadings 7-3
7.2.2 Calculate Hydrological and Soil Erosion Inputs 7-5
7.3 Module Discussion 7-6
7.3.1 Strengths and Advantages 7-6
7.3.2 Uncertainty and Limitations 7-7
7.4 References 7-8
8.0 Surface Water Module 8-1
8.1 Purpose and Scope 8-1
8.2 Conceptual Approach 8-2
8.2.1 Construct Waterbody Network 8-2
8.2.2 Route Hydraulic Flow Through the Waterbody Network 8-4
8.2.3 Construct and Solve the Mass Balance Equations Describing
Contaminant Fate and Transport throughout the Waterbody Network . 8-6
8.3 Module Discussion 8-7
8.3.1 Strengths and Advantages 8-7
8.3.2 Uncertainty and Limitations 8-7
8.4 References 8-8
9.0 Vadose Zone and Aquifer Modules 9-1
9.1 Purpose and Scope 9-1
9.2 Conceptual Approach 9-2
9.2.1 Vadose Zone Module 9-3
9.2.2 Aquifer Module 9-5
9.2.3 Chemical Reaction Modeling 9-9
9.3 Module Discussion 9-11
9.3.1 Strengths and Advantages 9-11
9.3.2 Uncertainty and Limitations 9-12
9.4 References 9-14
10.0 Farm Food Chain Module 10-1
10.1 Purpose and Scope 10-1
10.2 Conceptual Approach 10-4
-------
Table of Contents
Table of Contents (continued)
Section Page
10.2.1 Calculate Contaminant Concentrations in Plants due to
Contaminants in Air 10-5
10.2.2 Calculate Contaminant Concentrations in Plants due to
Contaminants in Soil 10-8
10.2.3 Calculate Total Contaminant Concentrations in Plants 10-9
10.2.4 Calculate Contaminant Concentrations in Beef and Milk 10-9
10.3 Module Discussion 10-11
10.3.1 Strengths and Advantages 10-11
10.3.2 Uncertainty and Limitations 10-12
10.4 References 10-13
11.0 Terrestrial Food Web Module 11-1
11.1 Purpose and Scope 11-1
11.2 Conceptual Approach 11-3
11.2.1 Calculate Contaminant Concentrations in Soil 11-5
11.2.2 Calculate Total Contaminant Concentrations in Plants 11-7
11.2.3 Calculate Contaminant Concentrations Soil Invertebrates 11-7
11.2.4 Calculate Contaminant Concentrations in Vertebrate Prey Categories 11-8
11.3 Module Discussion 11-9
11.3.1 Strengths and Advantages 11-9
11.3.2 Uncertainty and Limitations 11-10
11.4 References 11-12
12.0 Aquatic Food Web Module 12-1
12.1 Purpose and Scope 12-1
12.2 Conceptual Approach 12-3
12.2.1 Select Food Web Appropriate for Each Waterbody 12-3
12.2.2 Construct Dietary Matrix for Food Web 12-6
12.2.3 Calculate Contaminant Concentrations in Food Web 12-7
12.2.4 Report Contaminant Concentrations for Fish 12-14
12.3 Module Discussion 12-15
12.3.1 Strengths and Advantages 12-15
12.3.2 Uncertainty and Limitations 12-16
12.4 References 12-17
13.0 Human Exposure Module 13-1
13.1 Purpose and Scope 13-1
13.2 Conceptual Approach 13-4
13.2.1 Calculate Ambient Air Concentrations for Inhalation Exposures .... 13-5
13.2.2 Calculate Shower Air Concentrations for Inhalation Exposures 13-5
13.2.3 Calculate Dose from Inhalation of Carcinogens 13-6
13.2.4 Calculate Dose from Ingestion of Contaminants in Media or Food .. 13-7
vi
-------
Table of Contents
Table of Contents (continued)
Section Page
13.2.5 Calculate Dose from Ingestion of Contaminants in Breast Milk 13-9
13.3 Module Discussion 13-10
13.3.1 Strengths and Advantages 13-10
13.3.2 Uncertainty and Limitations 13-11
13.4 References 13-12
14.0 Human Risk Module 14-1
14.1 Purpose and Scope 14-1
14.2 Conceptual Approach 14-3
14.2.1 Calculate Risk Measures 14-3
14.2.2 Process Results for Decision Making 14-7
14.3 Module Discussion 14-9
14.3.1 Strengths and Advantages 14-9
14.3.2 Uncertainty and Limitations 14-10
14.4 References 14-11
15.0 Ecological Exposure Module 15-1
15.1 Purpose and Scope 15-1
15.2 Conceptual Approach 15-2
15.2.1 Criteria for the Ecological Exposure Module 15-3
15.2.2 Construct a Dietary Matrix 15-12
15.2.3 Calculate Applied Dose for Receptors in Terrestrial Habitats 15-15
15.2.4 Calculate Applied Dose for Receptors in Margin Habitats 15-18
15.3 Module Discussion 15-19
15.3.1 Strengths and Advantages 15-19
15.3.2 Uncertainty and Limitations 15-20
15.4 References 15-22
16.0 Ecological Risk Module 16-1
16.1 Purpose and Scope 16-1
16.2 Conceptual Approach 16-5
16.2.1 Development of EBs and CSCLs 16-5
16.2.2 Calculate Hazard Quotients 16-12
16.2.3 Process HQ Results for Decision Making 16-15
16.3 Module Discussion 16-18
16.3.1 Strengths and Advantages 16-18
16.3.2 Uncertainty and Limitations 16-19
16.4 References 16-21
vii
-------
Table of Contents
List of Figures
Figure Page
1-1 3MRA modeling system design 1-11
1-2 Linkages among the source, fate, transport, exposure, and risk modules for the
3MRA modeling system 1-13
1-3 Conceptual layout of typical 3MRA modeling system site 1-14
1-4 Probability that percent protection is less than P for a given waste concentration
and target risklevel 1-16
1-5 Percent of receptors protected for different waste concentrations and risk levels ... 1-17
1-6 Protective Summary Output figure for human risk 1-18
1-7 Protective Summary Output figure for ecological HQ 1-18
1-8 Document organization 1-20
2-1 Conceptual framework for human receptors 2-2
2-2 Conceptual framework for ecological receptors 2-2
2-3 Conceptual framework of the 3MRA modeling system 2-3
2-4 Current area of interest and concentric distance rings for human risk 2-6
2-5 Example of watershed delineation for a typical site 2-6
2-6 Example of human receptor placement for a typical site 2-8
2-7 Example of representative ecological habitats delineated for typical site 2-8
2-8 Illustration of concurrent time aggregation of risks 2-10
3-1 Site-based spatial overlays for 3MRA modeling system spatial framework
(GIS analysis) 3-2
3-2 Example of transfer of polygons to 100 x 100 m template grid 3-4
3-3 GIS data coverages for waterbody and watershed delineation 3-5
3-4 Regional watershed subbasin delineation 3-6
3-5 Local watershed delineation 3-8
3-6 Example site layout for human receptors 3-10
3-7 Preprocessed habitat codes 3-14
4-1 Information flow for the Wastewater Source Module in the 3MRA modeling system . 4-2
4-2 Conceptual model schematic for Wastewater Source Modules 4-4
4-3 Surface impoundment cross-section showing sediment and soil layers modeled by
the Surface Impoundment Module infiltration rate algorithms 4-20
5-1 Information flow for the Land-based Source Modules in the 3MRA modeling system 5-1
5-2 Interaction of the models that form the Land-based Source Modules 5-4
5-3 Illustration of landfill (shown with six cells and three waste layers) 5-5
5-4 Illustration of a waste pile in local watershed 5-7
5-5 Illustration of LAU in local watershed 5-8
5-6 Conceptual diagram of the Generic Soil Column Model (GSCM) 5-10
5-7 Daily water balance model 5-15
5-8 Runoff/erosion conceptual model 5-19
6-1 Information flow for the Air Module in the 3MRA modeling system 6-1
6-2 Conceptual diagram of dispersion and deposition 6-3
7-1 Information flow for the Watershed Module in the 3MRA modeling system 7-1
7-2 Illustration of watersheds within an AOI 7-2
-------
Table of Contents
List of Figures (continued)
Figure Page
8-1 Information flow for the Surface Water Module in the 3MRA modeling system 8-1
9-1 Information flow for the Vadose Zone and Aquifer Modules in the 3MRA modeling
system 9-1
9-2 Conceptual diagram of Vadose Zone and Aquifer Module 9-3
10-1 Information flow for the Farm Food Chain Module in the 3MRA modeling system . 10-1
10-2 Release, exposure, and uptake mechanisms of contaminants in plants 10-4
11-1 Information flow for the Terrestrial Food Web Module in the 3MRA modeling
system 11-1
11-2 Hypothetical forest habitat with four home ranges shown 11-4
12-1 Information flow for the Aquatic Food Web Module in the 3MRA modeling system 12-1
12-2 Example of simplified food web for freshwater lake (Gobas et al., 1993) 12-4
12-3 Relationship between essential metal concentration and organism health (adapted
from Chapman et al., 1996) 12-13
13-1 Information flow for the Human Exposure Module in the 3MRA modeling system . 13-1
14-1 Information flow for the Human Risk Module in the 3MRA modeling system 14-1
15-1 Information flow for the Ecological Exposure Module in the 3MRA modeling
system 15-1
15-2 Simple terrestrial food web showing example receptors 15-7
15-3 Simple margin food web showing both aquatic and terrestrial components 15-8
16-1 Information flow for the Ecological Risk Module in the 3MRA modeling system .. 16-1
IX
-------
Table of Contents
List of Tables
Table Page
2-1 Constituents in the 3MRA Database 2-12
2-2 Methodology and Data Sources for 3MRA Chemical Properties 2-14
2-3 WMU Types and Source Term Characteristics 2-16
2-4 WMU Types and Release Mechanisms Modeled 2-16
2-5 Human Exposure Pathways by Receptor Type 2-27
2-6 Ecological Exposure Routes Evaluated by Receptor Type 2-31
3-1 Ecological Risk Assessment Representative Habitats for Terrestrial Receptors 3-13
5-1 Summary of Mechanisms of Release of Particulate Emissions for Each WMU 5-22
11-1 Terrestrial Plant and Prey Categories in the Terrestrial Food Web 11-2
11-2 Example Exposure Concentrations for Contaminant^ Calculated by the Terrestrial
Food Web Module 11-5
12-1 Matrix of Biota in Food Webs for Freshwater Systems in 3MRA 12-6
13-1 Default Pathways Considered by Receptor Type 13-3
14-1 Example HQ Counts for Hypothetical Sites 14-8
14-2 Example Cumulative Frequency at Hypothetical Site 14-9
15-1 Representative Habitats for 3MRA 15-4
15-2 Representative Habitats for Receptor Species 15-10
15-3 Categories of Dietary Items for Ecological Exposure Assessment 15-13
15-4 Example of Dietary Preferences for Raccoon 15-14
15-5 Example Dietary Preferences for Short-Tailed Weasel 15-16
15-6 Example Exposure Concentrations to Contaminant^ for Short-Tailed Weasel .... 15-17
16-1 Assessment Endpoints and Measures of Effects for the 3MRA Modeling System ... 16-3
16-2 Example HQ Counts for Mammals at Hypothetical Site 16-16
16-3 Example Cumulative Frequency HQ Histogram for Mammals at Hypothetical Site 16-16
x
-------
Section 1.0
Introduction
1.0 Introduction
1.1 Background
The U.S. Environmental Protection Agency (EPA) Office of Solid Waste (OSW) is
responsible for managing solid and hazardous waste as specified by the Resource Conservation
and Recovery Act (RCRA) of 1976 and subsequent legislation, such as the Hazardous and Solid
Waste Act (HSWA) of 1984. These acts and the programs developed to implement them were
designed to protect human health and the environment. Thus, many of the regulatory decisions
within the RCRA programs are based, at least in part, on the human health risk and
environmental impacts of the regulatory options under consideration.
In recent years, as the RCRA program has evolved, and as new risk assessment methods
have been developed, EPA's need for improved risk assessment models has greatly increased.
The RCRA programs initially addressed only releases to ground water from land disposal
operations and releases to air from waste incinerators and other types of boilers and industrial
furnaces. However, the RCRA programs have expanded in scope over the years to encompass
hundreds of constituents, thousands of waste streams, and many types of waste management
practices, ranging from recycling and reuse to disposal and destruction techniques. Thus, new
risk assessment models were needed to assess the types and magnitude of risks that fall under the
broad purview of the RCRA programs.
In addition, in the mid-1990s, several groups within and outside of EPA came forward
with recommendations or guidance for improving risk assessment methods. In 1996, EPA issued
new guidelines for conducting exposure assessments and risk assessments (U.S. EPA, 1996f,g).
These guidelines focused on improving the science underpinning the risk or exposure
assessments that were being conducted, as well as improving the methods for characterizing the
uncertainty in the risk estimates that are generated. In 1997, the Presidential/Congressional
Commission on Risk Assessment and Risk Management (CRARM, 1997) issued a report on
improving risk assessment methods used by the federal government. Also, EPA's Science
Advisory Board (SAB) reviewed and commented on a number of EPA risk assessments and
models, including the dioxin and mercury risk assessments (U.S. EPA, 1996e, 2000).
One of the assessments that the SAB reviewed was a national-level risk assessment effort
conducted by OSW to support the proposed exit levels in the 1995 Hazardous Waste
Identification Rule (HWIR). The proposed HWIR (60 FR 66344, December 21, 1995) was, in
part, designed to establish contaminant-specific exit levels for low-risk solid wastes. Under this
proposal, generators of listed hazardous wastes that could meet the new concentration-based
criteria defined by the HWIR methodology would no longer be subject to the hazardous waste
management system specified under Subtitle C of RCRA. This established a risk-based "floor"
1-1
-------
Section 1.0 Introduction
for low-risk hazardous wastes that would encourage pollution prevention, waste minimization,
and the development of innovative waste treatment technologies. The rule also sought to reduce
possible overregulation arising from the "mixture" and "derived-from" rules promulgated earlier.
The SAB concluded that the methodology proposed for the HWIR assessment "lacks the
scientific defensibility for its intended regulatory use," and the subcommittee made the following
recommendations for establishing a scientifically defensible risk-based methodology applicable
at the national level for the waste program (U.S. EPA, 1996a):
Develop a true multipathway risk assessment in which a receptor receives a
contaminant from a source via all pathways concurrently, a receptor is exposed to
the contaminant via different routes, and the dose corresponding to each route is
accounted for in an integrated way;
Use a methodology that maintains mass balance;
Before implementing the model, focus a validation effort on a few key exposure
pathways of concern and those parameters that have a major impact on risks to
human health and the environment;
Examine model parameters systematically to ensure a consistent and uniform
application of the proposed approach, and further, to ensure that the full suite of
uncertainties is addressed in the final methodology;
Discard the proposed screening procedure for selecting the initial subset of
contaminants for ecological analysis and instead require that a minimum data set
be satisfied before ecologically based exit criteria are calculated;
Seek substantive participation, input, and peer review by EPA scientists and
outside peer-review groups as necessary to evaluate the individual components of
the methodology in much greater detail; and
Reorganize and rewrite the documentation for both clarity and ease of use.
The SAB review findings led to a joint decision between OSW and EPA's Office of
Research and Development (ORD) to develop an integrated and improved source, fate and
transport, exposure, and risk modeling tool that could be used to support national assessments
and regulatory actions. In 1997, OSW and ORD began working together on the development of
such a tool: the Multimedia, Multipathway, and Multireceptor Risk Assessment (3MRA)
modeling system. The design of the 3MRA modeling system
¦ Follows the risk paradigm and recent EPA guidance and scientific
recommendations;
Addresses major review comments of the HWIR analysis in 1995, specifically
those concerning the need for a true multipathway risk assessment methodology,
the conservation of mass, the validation of the methodology and its components,
1-2
-------
Section 1.0
Introduction
and the substantive participation of EPA scientists and outside peer-reviewer
groups in evaluating the methodology and its components;
¦ Is based on current science and a state-of-the-art modeling framework that
facilitates consistent use of sound science models, controls model sequencing,
facilitates data exchange, and provides data analysis and a results visualization
tool;
¦ Has multimedia and multipathway exposure and risk assessment capabilities;
¦ Has human health and ecological exposure and risk characterization capabilities;
¦ Is based on a mass balance approach within the waste management units
(WMUs);
¦ Represents variability in environmental fate and transport, exposure, and risk;
¦ Can quantitatively define the degree of protectiveness for a specific risk value
(e.g., protective of 95 percent of the exposed population at a risk level of 10"6);
¦ Uses site-based data (i.e., actual geographic locations and associated
environmental and population characteristics) to the extent available;
¦ Is applicable to multiple scales of analysis, including regional and national;
¦ Is capable of assimilating new science and component modules in the software
system;
¦ Reflects quality assurance and control protocols and is reproducible; and
¦ Has a verified approach and components that have been compared with other
analytical solutions, numerical models, and/or field data.
In addition to the above discussion on the background of the 3MRA modeling system,
this section is intended to provide the necessary context to review the science modules (Sections
2 through 16 of this Volume), the databases developed to support the modeling system
(Volume II), the quality assurance approach for the technology, science modules, and data
(Volume III), and the sensitivity and uncertainty analyses conducted to date (Volume IV). The
3MRA technology is discussed in detail in Volume V of this report, Technology Design and
User's Guide. The remainder of Section 1 provides an overview of the overall design goals,
system technology and architecture of the science modules, and output generated by the
modeling system to support national-scale, risk-based decision making. In keeping with EPA's
goals for easy accessibility and clear and transparent documentation, EPA has released Version 1
of the 3MRA modeling system available on CD ROM from OSW, and on ORD's Center for
Exposure Assessment Modeling (CEAM) Web site (http://www.epa.gov/ceampubl/mmedia/
index.htm), along with all technical documentation and data files (http://www.epa.gov/epaoswer/
haz waste/i d/h wirwste/ri sk. htm).
1-3
-------
Section 1.0
Introduction
1.2 Overall Design Goals of the 3MRA Modeling System
The overall design goals for the 3MRA modeling system reflect one of the most
important regulatory goals for OSW: determining when the chemical constituent concentrations
in a waste stream, as managed, may pose unacceptable risks to human health and the
environment. For example, certain generic waste streams are currently being managed as
hazardous waste because of the way they were captured by the hazardous waste regulations
(under Subtitle C of RCRA). However, some subset of these waste streams may not pose
significant health or ecological risks if they are disposed of as non-hazardous industrial wastes
(under with Subtitle D regulations). To quantify specific concentration-based criteria for
determining which waste streams may "safely" exit the hazardous waste "cradle-to-grave"
program, EPA could assess the potential health and ecological risks related to the management in
Industrial Subtitle D units of certain wastes with low concentrations of hazardous contaminants.
Waste streams containing concentrations below EPA-specified thresholds could exit the
hazardous waste system; those containing concentrations above the thresholds would remain in
the hazardous waste program.
Using the 3MRA modeling system, constituent-specific distributions could be generated
of cancer risks or hazards to humans and hazards to ecological receptors living in the vicinity of
Industrial D waste sites that might manage exempted wastes. The 3MRA modeling system can
produce national-level statistics that characterize risks and provide exit-level waste
concentrations for a contaminant that meet specific criteria of policy makers. These policy
criteria might include the following:
¦ The level of acceptable risk (e.g., 1 in 1 million),
¦ The probability of protection (e.g., 95 percent of the exposed population at a site),
¦ The probability that a particular site is protective at the risk level and population
protection level specified,
¦ The risk to the exposed population at various distances from the site (e.g.,
1,000 m),
¦ The risk to various receptor types (e.g., children, farmers, rare and endangered
species), and
¦ The risk to each receptor type by each exposure pathway (e.g., inhalation of
ambient air, ingestion of ground water used as drinking water).
To ensure that the modeling system could provide this type of information, OSW and
ORD engaged in a series of discussions to identify the key functional requirements for the
model, as well as the requirements for scientific defensibility that would shape the 3MRA
modeling framework. The requirements are summarized below represent the core design
decisions for the 3MRA modeling system.
1-4
-------
Section 1.0
Introduction
1.2.1 Key Functional Requirements
The 3MRA modeling system is intended to be one of EPA's next generation of
multimedia exposure and risk models to support regulatory decisions. EPA designed the
modeling system specifically to meet the needs of OSW programs, but the model has the
flexibility to be used for many EPA applications. In particular, it was designed to provide risk
managers with information, at a national level, on exposure and risk to human and ecological
receptors from the release of hazardous contaminants from the management of industrial wastes.
Specific functional requirements for the design of the 3MRA modeling system include the
following:
¦ Multiple Contaminants. Many different constituents may be present in wastes
regulated under RCRA. Therefore, the 3MRA modeling system was designed to
use current science to model more than 400 constituents with very diverse
physical-chemical properties and effects on humans and the environment. The
science on which the fate and transport of contaminants in the environment is
based differs for various types of constituents. In addition, the nature of national
assessment methodologies requires the ability to adjust some physical-chemical
properties that are dependent on environmental variables such as temperature.
Within specified constraints, constituents were grouped in the following five
categories: metals, organic chemicals, dioxin-like chemicals, mercury, and other
special case chemicals, such as those that are miscible in water or are metabolized
quickly. Even within these categories, distinctions must be made regarding
bioaccumulation, metabolism, and transformation for organics and regarding
congener-specific properties for dioxin-like chemicals.
¦ Multimedia. Traditionally, land disposal (i.e., landfilling) and destruction by
combustion sources have been the predominant waste management scenarios used
in the RCRA program. Release of a constituent to ground water is typically the
primary pathway modeled for landfilling, and releases to air are the primary
pathway modeled for combustion sources. However, a large quantity of
hazardous and industrial wastes are managed as wastewaters in surface
impoundments and tanks. Releases to air can be significant for some constituents
managed in these types of WMUs. Similarly, the ash from combustion and the
sludge from tanks and impoundments are sometimes applied to land, as are some
organic wastes. For the 3MRA modeling system, new source models for surface
impoundments, tanks, landfills, land application units (LAUs), and waste piles
were developed to provide the ability to model releases to all environmental
media (air, soil, and ground water).
¦ Multipathway. Once a contaminant is released to the environment, it is important
to follow the transport and fate of the contaminant through all environmental
media in order to capture all relevant pathways of exposure. In a national
application of such a model, what may be a driving pathway in one type of
environmental setting may be a small contributor to exposure in another. Also,
given the broad range of chemical properties being considered, the ground water
ingestion pathway may be important for one contaminant, the inhalation of
1-5
-------
Section 1.0
Introduction
ambient air for another contaminant, the ingestion of contaminated food crops or
prey for another, and the ingestion of fish from contaminated streams for another.
Therefore, the modeling system needed to include as many pathways as was
feasible and scientifically defensible.
¦ Multireceptor. The 3MRA modeling system was designed specifically to
quantify the risk to ecological and human receptors. The 3MRA modeling system
uses a site-based approach, in which various types of human and ecological
receptors are spatially delineated around a WMU. This spatial component of the
3MRA modeling system is critical in analyzing the differential effects on human
and ecological receptors in all media across the study area.
¦ Variability and Uncertainty. Quantifying variability and uncertainty in exposure
and risk estimates is an important capability of any modeling system. The 3MRA
modeling system was designed with a two-stage Monte Carlo analysis capability,
which enables users to distinguish between variability and uncertainty in input
variables. In addition, EPA has conducted extensive sensitivity analyses and
benchmarking against other similar models or model in order to understand the
limitations and uncertainty of this modeling system.
¦ Programmatic Needs. The 3MRA modeling system was designed with several
specific science and technology requirements agreed upon by OSW and ORD.
Key requirements included scientific defensibility (see Section 1.2.2) and a
technology design that was adaptable to a wide range of future applications and
the capability to incorporate into the system legacy codes that have been
extensively peer reviewed and used in support of regulatory activities at EPA (see
Volume V of this report). The following regulatory models or components of
regulatory models have been incorporated into Version 1 of the 3MRA modeling
system: Industrial Source Complex Short-Term Model, Version 3 (ISCST3) (U.S.
EPA, 1995) for air dispersion and deposition; EPA's Composite Model for
Leachate Migration with Transformation Products (EPACMTP) (U.S. EPA,
1996b,c,d, 1997) for subsurface transport; and EPA's Exposure Analysis
Modeling System II (Exams II) (Burns et al., 1982; Burns, 1997) for surface
water transport.
1.2.2 Scientific Defensibility Requirements
To address scientific defensibility, EPA implemented a systematic quality assurance
program throughout the conceptual design, development, and application of the 3MRA modeling
system. The program was designed to build confidence in the underlying science and
technology, ensure that the system could produce reliable risk results, and to characterize the
uncertainty and variability inherent in multimedia modeling. The major components of that
program are described briefly below.
¦ Technical Review. The development of the 3MRA modeling system evolved
from the ORD/OSW Integrated Research and Development Plan for the
Hazardous Waste Identification Rule (U.S. EPA, 1998). This report represented
1-6
-------
Section 1.0
Introduction
the collaboration between ORD and OSW to define (1) the assessment strategy
for national-scale risk analyses and (2) the design specifications of the 3MRA
modeling framework. A critical component of this blueprint for the 3MRA
modeling system was the technical review cycles required for the databases,
science modules, and system technology. This iterative process ensured that the
conceptual approach and implementation of the 3MRA modeling system was
consistent and defensible across all data sources and science modules, as well as
the software technology.
Peer Review. The overall technical approach and each science-based module
included in the 3MRA modeling system were peer reviewed. Teams of peer
reviewers (at least three per module) provided critical feedback about the science-
based modules. More than 45 independent experts reviewed the science modules
to ensure that the theoretical concepts describing the processes within the release,
fate, transport, uptake, exposure, and risk components were adequate
representations of the processes to be evaluated. The peer review cycle has been
an ongoing process since 1999, and is described in Volume III of this report. The
process has recently included the acceptance of The 3MRA Risk Assessment
Framework - A Flexible Approach for Performing Multimedia, Multipathway,
andMultireceptor Risk Assessments Under Uncertainty (in press) by the
International Journal of Human and Ecological Risk Assessment (Marin, et al.,
n.d.).
Verification. All software components and databases underwent a series of tests
to verify that the software and data were performing properly. At the heart of this
protocol was the requirement that each component of the modeling system
include a designed and peer-reviewed test plan that was executed by both the
module developer and a completely independent modeler (i.e., someone who did
not participate in the original module development). Prior to testing, each of the
test plans was reviewed and revised, as appropriate, to ensure that the tests were
comprehensive and addressed all of the major functions of the software and
supporting databases. These procedures, test plans, test packages, and test results
are fully documented and available to the public. Verification efforts for the
3MRA modeling system are described in Volume III of this report.
Validation. True validation of the 3MRA modeling system for a national
application would require validation over the full range of environmental settings
that are relevant to the application. However, determining whether the modeling
system is valid for the full range of settings was not possible because EPA was
unable to find such a data set. Instead, individual modules and data sets were
validated when appropriate data could be identified. A number of science
modules were based on existing models/methods that had already been validated
using field data and, therefore, prior validation studies and data were evaluated to
determine their relevance to the 3MRA application. The use of legacy models
developed by EPA was considered to be an important aspect of the overall
validation efforts. A description of the validation efforts of the 3MRA modeling
1-7
-------
Section 1.0
Introduction
system, including the system, science modules, and supporting data, is
documented in Volume III of this report.
Model Comparison. The 3MRA modeling system is undergoing a comparative
analysis with EPA's Total Risk Integrated Methodology (TRIM) (U.S. EPA,
2002), which is currently under development. The objective of the model
comparison effort is to increase confidence that the 3MRA modeling system
produces estimates consistent with other multimedia models. A description of the
model comparison efforts is documented in Volume III of this report.
Environmental Data Comparison. Modeled results from the 3MRA modeling
system are being compared with a multimedia data set for an actual industrial site
for which media monitoring data and field data (e.g., fish concentrations) were
available for mercury. These data represent a snapshot of mercury in the
surrounding environment rather than a continuous measure of its presence during
the operation of an industrial facility. Although the data set does not vary as a
function of time or include estimates of exposure and risk, the comparison will
yield important insights into the performance of multimedia models. A
description of this comparison effort is documented in Volume III of this report.
Representative National Data Set. A comprehensive data collection approach
was developed to parameterize the modeling system for 201 sites in accordance
with the site-based approach described in this document. The site-based data are
intended to provide a representative data set for a national-level assessment and,
to a large degree, serve as the test data set for the 3MRA modeling system. This
data collection plan described the general collection methodology for the major
types of data (for example, facility location, land use, soil characteristics, and
receptor locations), including quality assurance and quality control procedures
and references for data sources. The data collection effort for the representative
national data set is documented in Volume II of this report.
Sensitivity and Uncertainty Analyses. As the level of assessment complexity
grows, it becomes both more difficult and more important to establish
comprehensive and quantitative expressions of uncertainty. Uncertainty analysis
can be a difficult task, especially for complex, integrated, multimedia models
such as the 3MRA modeling system. Thus, a formal program focusing on
sensitivity and uncertainty analysis for complex modeling systems was initiated at
ORD. The early focus of this program is the investigation of parameter
sensitivities and system uncertainties within the 3MRA modeling system. To
facilitate this evaluation, ORD has recently developed a Windows-based parallel
computer cluster. This supercomputer for Modeling Uncertainty and Sensitivity
Evaluation (SuperMUSE), comprising 176 client PCs and supporting software
infrastructure that allows exhaustive experimentation of the 3MRA modeling
system. A complete description of the computational and software framework for
conducting evaluation strategies for the 3MRA modeling system, the SuperMUSE
system, and initial results of the evaluation efforts are presented in Volume IV of
this report.
1-8
-------
Section 1.0
Introduction
1.3 Overview of the 3MRA Modeling System Technology
The 3MRA modeling system was developed as a predictive tool to provide risk
assessment support for the types of risk management decisions that are made within OSW. OSW
applies risk assessment modeling tools in a variety of situations; one application is the conduct
of site-based national-level risk assessments to support rulemaking for the identification of
hazardous waste. Consequently, the 3MRA modeling system needed to be able to model waste
management environmental settings that are representative of the range of environmental
settings found in the United States, and within this broad range of settings, to simulate the
release, fate, and transport of many contaminants in waste undergoing a range of physical and
biochemical processes. More than 400 constituents are regulated under the RCRA programs.
EPA needs to consider the impacts of these released contaminants on humans and the
environment within the broad range of environmental settings. This requires a modeling tool
that encompasses releases to all media, transport within those media, uptake in terrestrial and
aquatic food webs, and exposure of specific receptors to contaminated media and food items.
The 3MRA modeling system predicts human and ecological risks at a statistically
selected number of waste management sites across the United States, and accumulates statistics
on the estimated risks at each of these sites in order to generate national distributions of risks.
This site-based modeling is applied for each combination of site, WMU, chemical, and
concentration level in a given waste stream. The assessment methodology is national in scale
and site-based; that is, risks are assessed at individual sites across the United States and rolled up
to represent a national distribution of risks. The resulting national distribution of risks forms the
basis for determining waste stream constituent concentrations that satisfy criteria reflecting the
percentage of nationwide receptors and sites that are "protective." When a simulation is
complete, the risk estimates are organized into expressions of the probability of protection for
different constituent concentration levels in waste streams. These risk outputs can be expressed
as a function of several dimensions, for example, waste unit type, human receptor type, or the
distance from the WMU. This ability to express risks and hazard as a function of different
attributes provides the decision makers significant flexibility to ask "what if' questions leading
up to final decisions. The primary technology requirements related to the application of the
3MRA modeling system to national level human and ecological risk assessments are as follows:
¦ Facilitate "plug and play" functionality throughout the modeling system to allow
new science, modules, and data to be integrated into the system as the state of
multimedia modeling science and software development continues to evolve;
¦ Conduct simulations using the 3MRA modeling system science modules and
databases; each simulation representing a combination of site, chemical, WMU,
and waste constituent concentration;
¦ Generate and store risk matrices, i.e., risk estimates as a function of site, exposure
area, exposure pathway, exposure route, contact medium, receptor, receptor type,
and Monte Carlo realization; and
1-9
-------
Section 1.0
Introduction
¦ Provide a flexible framework that can accommodate alternate policy formulations
including different measures of protection, and both waste and leachate
concentration regulatory limits.
Figure 1-1 provides an overview of the 3MRA modeling system design. As suggested in
this figure, the system is constructed of components that manage the processing, flow, and
storage of information through the system, including input/output files and a variety of
supporting databases. At the top of Figure 1-1, the looping structure used to conduct national-
scale analyses is summarized, including the site location loop, the WMU loop, the number of
iterations, and the number of chemicals (which are simulated individually). The function of each
component that serves to manage, process, and store information is summarized below. For
additional information and details on the 3MRA technology, see Volume V of this report,
Technology Design and User's Guide.
¦ System User Interface (SUI). This processor represents the user access point to
the technology. Via the SUI, the user selects which combinations of sites,
WMUs, constituents, and constituent concentrations in waste streams to be
simulated, and the number of Monte Carlo simulations to be executed per site.
The SUI also provides the user with the ability to configure the computer
directory structure where individual components of the system are stored.
Finally, the SUI manages the overall execution of the user-defined national
assessment.
¦ Site Definition Processor (SDP). This processor performs all data retrieval from
the site, regional, national, and chemical databases and organizes the data into a
series of "site simulation files" that contain the input data for each of the
seventeen science models. The site definition in the figure includes both the
selection of site characteristics from data sources at multiple spatial scales (i.e.,
local, regional, and national), as well as the estimation and selection of chemical
properties.
¦ Multimedia Simulation Processor (MMSP). This processor manages the
invocation, execution, and error handling associated with the seventeen individual
science models that simulate source release, multimedia fate and transport, food
web dynamics, and human/ecological exposure and risk. The multimedia,
multipathway simulation includes all of the science modules linked together to
predict behavior of constituents from source release through exposure and risk.
¦ Chemical Properties Processor (CPP). This processor accesses the chemical
properties database and either transfers or calculates all requested data. The CPP
provides a single location within the modeling system where chemical data is
available.
¦ Exit Level Processor I (ELP I). This processor assimilates the individual site
risk results and builds a risk summary database containing data used to assess
national protection criteria.
1-10
-------
System User Interface (SUI)
Waste Management Facility Loop (201 National Sites)
Key
I I User Interface
Waste Management Unit Loop (5 Source Types)
Sampled Input Data Iteration Loop (nr)
~
Data File
o
Processor
~
Database
Header Info from SUI
Chemical Loop (43 Metals & Organics)
Cw Loop
C Waste stream concentration
Warnings/Errors to SUI
of Sites
Exit
Level
Processor
II
Site-Based
Database
Multimedia
Multipathway
Simulation
Processor
Exit
Level
Processor
I
Site
Definition
Processor
Regional
Database
Risk
Visualization
Processor
National
Database
List of Chemicals
Chemical
Properties
Database
Chemical
Properties
Processor
Metal
Isotherms
MET
Database
"Y
Site Input Data
Site Definition
-A y V
Multimedia Multipathway
Simulation
Exit Level Processing
Figure 1-1. 3MRA modeling system design.
-------
Section 1.0
Introduction
¦ Exit Level Processor II (ELP II). This processor reads the risk summary
database created by the ELP I and generates, based on regulatory criteria, specific
national exemption levels.
¦ Risk Visualization Processor (RVP). This processor is identical to the ELP II
except that it presents results in graphical form.
1.4 Overview of the 3MRA Science Module Architecture
The MMSP can be viewed as an integration of science-based modules within an
assessment strategy, i.e., procedures by which the modules are combined and applied to perform
the risk assessment. The 3MRA modeling system simulates contaminant releases from a WMU
to the various media (air, water, soil) based on the physical-chemical properties of the
constituent, the characteristics of the modeled WMU, and the environmental setting (e.g.,
meteorological region) in which the facility is located. Once released from the WMU, the
constituent is transported through environmental media and into biological compartments such
as produce, beef, and fish. Human and ecological receptors included in the simulation may be
exposed concurrently to contaminated media and food through multiple pathways and routes of
exposure. For each receptor that is included in the simulation, the 3MRA modeling system
performs risk/hazard calculations based on aggregate exposures modeled through space and
time. The linkages among the science modules are depicted in Figure 1-2.
Figure 1-3 illustrates the conceptual layout of a typical 3MRA modeling system site. As
suggested by this figure, the 3MRA site model represents a comprehensive multimedia approach
to assessing the potential impacts of chemical releases from land-based WMUs. Shown in
Figure 1-3 are the primary site layout features including the WMU at the center of the Area of
Interest (AOI), which extends 2 kilometers from the unit boundary by default. The concentric
distance "rings" are used to characterize risk as a function of distance. Other physical features of
the site layout include watersheds, surface water networks, aquifers, ecological habitats, home
ranges of resident ecological species, and human population distributions. These features are
explicitly delineated and the relative connectivity determined for each site included in the
assessment.
To the extent possible, the site layout data are based on site-specific information.
However, site data were not available for all sites, and resource limitations associated with
collecting the data are prohibitive. Consequently, the supporting data for the representative
national data set are based on a tiered approach to data collection that includes site-specific-,
regional-, and national-level data. During any execution of the modeling system, the most
preferred data source is site-specific followed by regional. Finally, lacking either a site-specific
or regional source of data, a national-scale statistical distribution of the variable is sampled and
assigned to the site. Each of these databases is available to the system and, as new data become
available, the system automatically acts on this hierarchy. In all, several hundred variables are
required to model any given site.
1-12
-------
Terrestrial Food
Web
Surface
Water
Watershed
Ecological
Risk
Aquatic Food
Web
Ecological
Exposure
Human
Exposure
vadose
Zone
Aquifer
Farm FoodChain
Human
Risk
Landfill
Waste Pile
Surface
Impoundment
Aerated
Tank
Land
Application
Unit
O
Sources Transport Foodchain Exposure/Risk
S?
Tlie dashed line indicates that soil concentrations for the local (land-based source) and regional watersheds may be added together to estimate g
total soil concentrations for areas (e.g., habitats) that include both regional and local watershed components. §-
Figure 1-2. Linkages among the source, fate, transport, exposure, and risk modules fo the 3MRA modeling system. §
-------
Section 1.0
Introduction
'S Sub 1
3
WBN Rch 3
|-+r
Src 1
Hab
WS Sub 4
Range
^ange
LAKE
BGil
WBN Rch 4
^ange 3
BG2
Fami 1
Farm 2
N.,
Src 1
*15
I Rch 3
*14 ,
Hab :
WS Sub 4
*16
Farm 2
*19
*20
Rang
-------
Section 1.0
Introduction
¦ Estimate human and ecological risk per receptor per pathway, and aggregate
pathways and receptors as appropriate; and
¦ Repeat this sequence for each of a series of waste concentrations (Cw) to establish
a quantitative relationship between Cw and risk or hazard.
1.5 Overview of Results Generated by the 3MRA Modeling System
The 3MRA modeling system produces two measures of protection that can be used in
regulatory decision making. The first measure of protection is the nationwide distribution of
risks for receptors of concern. Specifically, a concentration limit can be defined by the percent
of nationwide receptors of concern that exceed a given risk level falling below a specific percent
(e.g., 95 percent of all receptors across all sites, all pathways, and all WMU types within 2 km of
the WMU incur a risk of 10"6 or less). Because the risk/HQ data at the site level is stored by
indices including receptor type, exposure pathway, exposure ring distance, and WMU, it is
possible to construct "views" of the national-scale protectiveness that reflect varying
combinations of the indices. For example, protection measures can be applied to individual
receptor types, combinations of receptor types, individual WMUs, etc. The second measure of
protection is the nationwide distribution of sites that are protected. The limit can be defined
under this protection measure by the percentage of protected sites nationwide that is greater than
a given target level. A site is protective if the percentage of site-based receptors incur a risk/HQ
less than a specified target value.
These measures of protection are combined in the 3MRA modeling system to allow a
user to specify both the percentage of receptors nationwide as well as the percentage of sites that
are to be protected (e.g., 95 percent of the sites are protective of 99 percent of the site-based
receptors). To transform the site-based risk/HQ data into these national measures of
protectiveness requires that the population counts be converted to population percentages and
accumulated across all sites. With the risk/HQ indices preserved from the site results (i.e.,
receptor, pathway, ring distance, WMU, etc.), one can query the database containing results and
generate the desired estimates of national protectiveness. Figure 1-4 presents an example
corresponding to a query for a target risk level of 10"6 from the iterations corresponding to a
waste concentration of 10"3 mg/kg. The figure indicates that there is a 5 percent chance that the
level of protection (percent of receptors that would be protected at the target risk level for the
given waste concentration) would be less than or equal to 85 percent. Similarly, there is a 25
percent chance that less than or equal to 93 percent of the receptors would be protected at the
target risk level for the given waste concentration. Querying the output data base for different
waste concentrations can produce the set of graphs such as those shown in Figures l-5(a), (b),
and (c). The figures show how the percent protection varies as a function of the target risk, the
waste concentration, and the confidence limit, and can be used to select the waste concentration
that meets a specified protection measure. These types of figures could also be produced for
subsets of receptors to investigate the effects of selecting a waste concentration on secondary
protection measures.
These risk results are produced by three processors (ELP I, ELP II, and RVP) that
collectively accumulate, transform, and present information generated by the individual site-
based risk assessments in the form of measures of site and population protectiveness at the
national level. The ELP I reads individual site-based human-health risk and hazard and
1-15
-------
Section 1.0
Introduction
100-1-
(98%, 95%)
80%
85%
90%
95%
100%
P (% Receptor Protected, Cw = 10"3, TR = 10"6)
Figure 1-4. Probability that percent protection is less than P for a given
waste concentration and target risk level.
ecological hazard results from the Human and Ecological Risk Modules, transforms the
population counts into percentages of receptor populations protected, and stores the resulting risk
and hazard information in a series of national Risk Summary Output Files (RSOFs). The RVP
and ELP II processors query the RSOF in response to specific criteria for protective measures
and produce graphical and tabular views of the trade off between levels of constituent
concentration in waste streams and levels of protectiveness. With the combination of the RSOF
database and the RVP/ELP II, the following type of question can be asked: "What is the
maximum allowable constituent concentration for waste streams entering landfills such that at
least 90 percent of all receptors at 95 percent of the sites nationwide incur a risk less than 10"6?"
In addition, the RVP allows the user to query and summarize the information stored in the
RSOFs and graphically view the results of such queries. The user is prompted for a set of
scenario attributes (that is, WMU type, constituent, distance, receptor type, cohort, etc.) and a
level of protection (for example, risk of 1.0 x 10"6 or HQ of 1.0, etc.). The RVP uses this
information to construct plots showing the probability of protecting human or ecological
receptors (in the group of receptors defined by the scenario attributes) as a function of Cw. Using
information obtained from the RSOFs, the probability of protecting a given receptor is
determined by taking the group of receptors defined by the scenario attributes across all sites and
plotting the percentage of those receptors that are protected (that is, have a risk or hazard value
equal to or below the level of protection). Because each site is simulated with multiple
realizations, several probability curves are plotted, each corresponding to the chosen confidence
levels (for example, 5 percent, 50 percent, 95 percent). These plots are called Protective
Summary Output curves. The RVP creates a Protective Summary Output curve for human risk,
human hazard quotient (HQ), and ecological HQ, depending on the appropriateness and
existence of the data for the constituent chosen. Figures 1-6 and 1-7 illustrate the human risk
protective summary and ecological hazard protective summary plots produced by the RVP,
respectively.
1-16
-------
Section 1.0
Introduction
100%
3 log(1/Cw | Risk =10^)
100%
log(l/Cw | Risk = 10*)
100%
3 log(1/Cw | Risk = 10^)
Figure 1-5. Percent of receptors protected for different waste concentrations and risk levels.
1-17
-------
Section 1.0
Introduction
File Options
Selections i Human Risk =| Ecological HQ ]
Cw Interpolator"—
Population percentile
Probability of protection
jJ Interpolated Ow [~
"3
Set Defaults
Set Individual Defaults
E5TBZ1EH
iwinci [jptoi
Human Risk Protective Summarv
100
90
Hil
c
o
ts
70
m
o
fin
CL
Ml
*
40
JS
o
30
CL
20
10
0
CWs in ug/g
Description
Chemical constituent:
~ 25% pop.
| 95% pop.
WMU source type:
wu
Distance:
2000.000000
Critical year method:
Maximum
Exposure pathway:
liSummation of alMngestion art
Receptor type:
HAH^Receptors
Cohort type:
All Cohorts
1E-6
Figure 1-6. Protective Summary Output figure for human risk.
HW1R - Risk Visualization Processor
File Options
Selections | Human Risk [ Ecological HQ
rCw Interpolator
Population percentile (
Probability of protection I
Interpolated Cw
"1
Set Defaults
Set Individual Defaults
sum s
100
90
Sd
70
40
30
20
10
0
Ecological HQ Protective Summary
1
*
SI
\
8
)
\
N
)
6
>
X
"'X4
)
?
0.001 0.1 1 100 1000
CWs in ug/g
~ 25% pop
95% pop
1:0.001
4:100
'1IJUU
-Description
Chemical constituent:
Benzene
WMU source type:
Lau
Rpllup option:
By Ring and Receptor Group
Radius ring distance
Eco Ring #2
Receptor group
MMammal
Hazard quotient:
0.1
Figure 1-7. Protective Summary Output figure for ecological HQ.
1-18
-------
Section 1.0
Introduction
1.6 Organization of This Document
The first three sections of this document (including this introduction) discuss the
modeling system as a whole. Section 2 presents an overview of the modeling approach from a
conceptual viewpoint and describes the assessment strategy and the general approach for its
implementation. Section 3 provides a discussion of the development and use of spatial data in
this model, which are fundamental to developing variability in exposure and risk estimates
across a site.
The remaining sections describe the WMU source, fate and transport, exposure, and risk
modules. Figure 1-8 shows which section describes each module. The five source modules are
covered in two sections, one for land-based sources (LAUs, landfills, and waste piles) and one
for wastewater systems (surface impoundments and tanks).
1-19
-------
Section 1.0
Introduction
Surface Impoundment/Aerated Tank Source Modules
(Section 4)
Description of Waste Management Units (4.2.1)
Calculate Constituent Concentrations within the Unit (4.2.2)
Calculate Solids Concentration within the Unit (4.2.3)
Calculate Volatile Emission Rates (4.2.4)
Estimate Resuspension, Sedimentation, and Burial Velocities (4.2.5)
Estimate Constituent Release in Leachate (Surface Impoundment Only) (4.2.6)
Adjust for Temperature Effects (4.2.7) J
Land-Based Source Modules
(Section 5)
Description of WMUs (5.2.1)
Generic Soil Column Model (5.2.2)
Local Watershed Model (5.2.3)
Particulate Emissions Model (5.2.4)
Media Fate & Transport
{ Air Module
( Watershed Module
f Surface Water Module^
( Vadose Zone and ¦
(Section 6)
(Section 7)
(Section 8)
Aquifer Modules
• Characterize Source-Specific
• Calculate Soil
• Construct Waterbody
(Section 9)
Parameters (6.2.1)
Concentrations and
Network (8.2.1)
• Model Vadose Zone Flow
• Calculate Receptor Locations
Surface Water Loadings
• Route Hydraulic Flow
and Transport (9.2.1)
(6.2.2)
(7.2.1)
through the Waterbody
• Model Groundwater Flow
• Calculate Receptor-Specific
• Calculate Hydrological
Network (8.2.2)
and Transport (Aquifer
Concentration and Deposition
and Soil Erosion Inputs
• Construct and Solve
Model) (9.2.2)
Estimates (6.2.3)
(7.2.2)
Mass Balance Equations
• Model Subsurface
• Calculate Constituent-Specific
v y
Describing Contaminant
Chemical Reactions
Annual Average
Fate and Transport
V (9.2.3) )
Concentrations and Deposition
V <8-2-3) y
Rates (6.2.4) J
Food Webs
Farm Food Chain Module
(Section 10)
Calculate Contaminant
Concentrations in Plants Due to
Contaminants in Air (10.2.1)
Calculate Contaminant
Concentrations in Plants Due to
Contaminants in Soil (10.2.2)
Calculate Total Contaminant
Concentrations in Plants (10.2.3)
Calculate Contaminant
Concentrations in Beef and Milk
^ (10-2.4) ,
Terrestrial Food Web Module
(Section 11)
Calculate Contaminant
Concentrations in Soil (11.2.1)
Calculate Total Contaminant
Concentrations in Plants (11.2.2)
Calculate Contaminant
Concentrations in Soil Invertebrates
(11.2.3)
Calculate Contaminant
Concentrations in Vertebrate Prey
Categories (11.2.4)
Aquatic Food Web Module
(Section 12)
Select Food Web Appropriate
for Each Waterbody (12.2.1)
Construct Dietary Matrix for
Food Web (12.2.2)
Calculate Contaminant
Concentrations in Food Web
(12.2.3)
Report Contaminant
Concentrations for Fish
Consumed by Wildlife and
Humans (12.2.4)
Ecological Risk Module
(Section 16)
Development of Ecological Benchmarks
and Chemical Stressor Concentration
Limits (16.2.1)
Calculate Hazard Quotients (16.2.2)
Process the HQ Results for Decision
Making (16.2.3)
Human Risk Module
(Section 14)
Calculate Risk Measures (14.2.1)
Process Results for Decision Making
(14.2.2)
Ecological Exposure Module
(Section 15)
Criteria for the Ecological Exposure
Module (15.2.1)
Construct a Dietary Matrix for Each
Receptor (15.2.2)
Calculate Applied Doses for Animals in
Terrestrial Habitats (15.2.3)
Calculate Applied Doses for Animals in
Margin Habitats (15.2.4)
Human Exposure Module
(Section 13)
Calculate Ambient Air Concentrations
(13.2.1)
Calculate Shower Air Concentration
(13.2.2)
Calculate Dose from Inhalation of
Carcinogens (13.2.3)
Calculate Dose from Ingestion of
Contaminants in Media or Food (13.2.4)
Calculate Dose from Ingestion of
Contaminants in Breast Milk (13.2.5)
Figure 1-8. Document organization.
1-20
-------
Section 1.0
Introduction
1.7 References
Burns, L. A. 1997. Exposure Analysis Modeling System (EXAMS II): User's Guide for
Version 2.97.5. EPA-600/R-97/047. Athens, GA: U.S. Environmental Protection
Agency.
Burns, L. A., D.M. Cline, and R.R. Lassiter. 1982. Exposure Analysis Modeling System
(EXAMS): User Manual and System Documentation. EPA-600/3-83-023. Athens, GA:
U.S. Environmental Protection Agency, Environmental Research Laboratory.
CRARM (Presidential/Congressional Commission on Risk Assessment and Risk Management).
1997. Risk Assessment and Risk Management in Regulatory Decision Making. Final
Report, Volume 2. Washington, DC: Government Printing Office.
Hazardous Waste Identification Rule (HWIR). 1995. Federal Register 60:245. December 21,
p. 66344.
Marin, C.M., V. Guvanasen, and Z.A Saleem. n.d. The 3MRA risk assessment framework - a
flexible approach for performing multimedia, multipathway, and multireceptor risk
assessments under uncertainty. International Journal of Human and Ecological Risk
Assessment (in press; scheduled for publication December 2003).
Resource Conservation and Recovery Act. 1976. 42 U.S.C. s/s 6901 et seq. Available online:
http://www4.law.cornell.edu/uscode/42/ch82.html.
U.S. EPA (Environmental Protection Agency). 1995. User's Guide for the Industrial Source
Complex (ISC 3) Dispersion Models. Volume II: Description of Model Algorithms. EPA-
454/B-95-003b. U.S. Environmental Protection Agency, Emissions, Monitoring, and
Analysis Division, Office of Air Quality Planning and Standards, Research Triangle
Park, NC. September.
U.S. EPA (Environmental Protection Agency). 1996a. An SAB Report: Review of a
Methodology for Establishing Human Health and Ecologically Based Exit Criteria for
the Hazardous Waste Identification Rule (HWIR). EPA-SAB-EC-96-002. Washington,
DC: Science Advisory Board. May.
U.S. EPA (Environmental Protection Agency). 1996b. EPACMTP Sensitivity Analysis.
Washington, DC: Office of Solid Waste. March.
U.S. EPA (Environmental Protection Agency). 1996c. EPA 's Composite Model for Leachate
Migration with Transformation Products (EPACMTP): Background Document.
Washington, DC: Office of Solid Waste. September.
U.S. EPA (Environmental Protection Agency). 1996d. EPA 's Composite Model for Leachate
Migration with Transformation Products, EPACMPT. Background Document for Finite
Source Methodology for Chemicals with Transformation Products. R09-96-588.
Washington, DC: Office of Solid Waste. October.
1-21
-------
Section 1.0
Introduction
U.S. EPA (Environmental Protection Agency). 1996e. Mercury Study Report to Congress,
Volumes V, VI, VII, and Supplement. SAB Review Draft. EPA-452/R-96-001a-h.
Washington, DC: Office of Air Quality Planning and Standards and Office of Research
and Development. June.
U.S. EPA (Environmental Protection Agency). 1996f. Proposed Guidelines for Carcinogen
Risk Assessment. Notice of Availability and Opportunity to Comment on Proposed
Guidelines for Carcinogen Risk Assessment. Federal Register 61 FR 17960-18011.
April 23.
U.S. EPA (Environmental Protection Agency). 1996g. Proposed Guidelines for Ecological Risk
Assessment. Risk Assessment Forum, Washington, DC. Web site at
http://www.epa.gov/ORDAVebPubs/ecorisk.
U.S. EPA (Environmental Protection Agency). 1997. EPA 's Composite Model for Leachate
Migration with Transformation Products (EPACMTP): User's Guide. Office of Solid
Waste. Washington, DC.
U.S. EPA (Environmental Protection Agency). 1998. ORD/OSW Integrated Research and
Development Plan for the Hazardous Waste Identification Rule (HWIR). Washington,
DC.
U.S. EPA (Environmental Protection Agency). 2000. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlorobinzo-p-Dioxin (TCDD) and Related Compounds.
Part III: Integrated Summary and Risk Characterization for 2,3,7,8-Tetrachorodibenzo-
p-Dioxin (TCDD) and Related Compounds (SAB Review Draft). EPA/600-P-00/00Bg.
Washington, DC: National Center for Environmental Assessment, Office of Research
and Development. September.
U.S. EPA (Environmental Protection Agency). 2002. Total Risk Integrated Methodology: TRIM
Fate Documents. Available online: http://www.epa.gov/ttn/atw/urban/trim/trimpg.html.
1-22
-------
Section 2.0
Modeling Approach
2.0 Modeling Approach
The 3MRA modeling system is intended to be one of EPA's next generation of
multimedia exposure and risk models, capable of modeling multimedia, multipathway, and
multireceptor exposures. The 3MRA modeling system models multimedia exposures by
simulating releases of a constituent to air, soil, and ground water, and then modeling the
transport and fate of the constituent in each of these media and in the food webs associated with
them. It models multipathway exposures by calculating simultaneous exposures of a receptor
through multiple pathways, such as the ambient air, soil, food items, and drinking water, and
summing them, when appropriate. It models multireceptor exposures by simulating a set of
human receptors and a set of ecological receptors that characterize the receptor populations and
behaviors of those receptors within an area of interest (AOI).
This section
¦ Presents an overview of the 3MRA framework—the conceptual approach and
science that is implemented in the 3MRA modeling system;
¦ Describes the spatial and temporal scales that provide the context for the 3MRA
framework; and
¦ Describes the design of the 3MRA modeling system as it pertains to constituents
assessed and sources, exposure pathways, receptors, and human health and
ecological exposures and risks modeled.
2.1 Overview and Conceptual Approach
Figures 2-1 and 2-2 illustrate the conceptual framework for human receptors and
ecological receptors, respectively, in the 3MRA framework. This conceptual framework is
further depicted in the flow diagram in Figure 2-3. These figures show the conceptual models
for human and ecological receptors, beginning with the multimedia release of constituents from a
waste management unit (WMU). Once released, constituents travel through environmental
media that determine the exposure pathways for the analysis: air, ground water, watershed soils,
and surface water. In addition, plants and animals take up contaminants either directly from the
different media or through bioaccumulation of constituents in terrestrial and aquatic food webs.
Eventually, human receptors are exposed to contaminants through inhalation or ingestion of
contact media (i.e., contaminated air, soil, ground water, homegrown produce, beef and milk
produced on a farm, and fish). The ecological receptors are exposed to contaminants through
direct contact or ingestion of contact media (i.e., contaminated air, soil, surface water, and flora
and fauna).
2-1
-------
Section 2.0
Modeling Approach
^grfcultural Exposures
Figure 2-1. Conceptual framework for human receptors.
resemative Terrestrial8*0
Figure 2-2. Conceptual framework for ecological receptors.
2-2
-------
Section 2.0
Modeling Approach
Inputs
Source
Release
Modules
• Source data
• Physical-chemical
properties data
• Meteorological data
• Site-specific data
Inputs
Site-specific GIS data
Physical-chemical
properties data
Meteoroiogical data
Regional hydrogeologic
and soils data
Receptor locations
Media Fate and Transport
Air
Module
Watershed
Module
Surface
Water
Module
Subsurface
Modules
Farm
Food Chain
Module
Food Webs
Terrestrial Food
Web Module
Aquatic Food
Web Module
Inputs
• Site-specific GIS data
• Physical-chemical
properties data
• Habitat data
• Prey preference data
• Farm animal diet data
Inputs
• Site-specific GIS dati
receptor locations
• Physical-chemical
properties data
• National exposure
factors data
• Shower Model data
• Breast Milk Model da
Human Exposure
Module
Ecological Exposure
Module
Inputs
• Site-specific GIS data-
habitat locations
• Habitat data
• Prey preference data
Inputs
1 Site-specific GIS
Census and farm
population data
1 Human health
benchmarks
Human Risk
Module
Ecological Risk
Module
Inputs
Site-specific GIS data
Ecological benchmarks
Inputs
Exit Level
• Risk characterization
format
• Decision criteria
Processors 1 & 2
Figure 2-3. Conceptual framework of the 3MRA modeling system.
2-3
-------
Section 2.0
Modeling Approach
Several major strengths of the 3MRA modeling system and associated components are
that they provide true multipathway functionality, maintain mass balance within the WMUs,
model sorption and degradation processes, account for timing of exposures, use a regional site-
based approach, and capture variability and uncertainty. These are discussed below.
True Multipathway Functionality. An overriding consideration in the development of
the 3MRA modeling system was to provide a system with true multipathway functionality in
which a receptor receives constituents from a source via all relevant pathways simultaneously.
Constituents are released from a source to various media, depending on the characteristics of the
source, the waste, and the surrounding environment. Once the constituents are in the air, soil,
and ground water, they are transported through these media to the farm food chain, terrestrial
food web, and aquatic food web, or they come into direct contact with a receptor via air, soil, or
drinking water. At any given time, the contaminant concentrations to which a receptor is
exposed in each medium or food item are evaluated simultaneously.
Mass Balance. The 3MRA modeling system maintains mass balance for a constituent
within each type of WMU source. The system starts with a concentration of a constituent in the
WMU and partitions the mass to vapor, liquid, and sorbed phases. The 3MRA modeling system
continues this partitioning over the life of the WMU as wastes are added and constituents
released. Constituent mass released via each phase is no longer available for partitioning to and
release through other phases. The data used in partitioning include a range of constituent-
specific partition coefficient (Kd) values representing various waste forms that are intended to
reflect the full range of waste types (see Sections 4 and 5 for further discussion).
Sorption. The mobility of metals in the environment is dependent on the geochemical
properties of the soil, surface water, and ground water to which they are released. To account for
the metal-specific interactions with various environments modeled (wastes, soil, surface water,
and ground water), the 3MRA modeling system uses nationwide distributions of key
geochemical parameters, including metal sorption coefficients. See Volume II of this report for
a discussion of the methods and data used to generate the sorption coefficients.
Degradation Processes. Both the source and the fate and transport modules in the
3MRA modeling system were designed to simulate biodegradation and constituent degradation.
Full implementation of these processes is limited, primarily because of the limited availability of
constituent-specific data on aerobic and anaerobic biodegradation rates. EPA has invested in the
development of additional biodegradation rate data specifically to enhance the use of the 3MRA
modeling system to support specific RCRA programs. Because of the large number of
constituents regulated under RCRA, some constituents were designated as high priority because
of their presence in high-volume wastes. Those high-priority constituents were addressed first.
Volume II of this report provides a full discussion of the constituent-specific degradation data
developed for the 3MRA model (see Sections 4 through 9 for further discussion of the
degradation processes included in each of the modules).
Time Series Management. The 3MRA modeling system is designed to estimate the
potential risk or hazard quotient (HQ) associated with a constituent that will be managed in a
WMU throughout the WMU's operational life and beyond. Average annual exposures are
specific to the constituent and environmental setting, and are estimated to occur for up to
2-4
-------
Section 2.0
Modeling Approach
10,000 years in some cases. The source modules simulate release of a constituent until the
concentration in the WMU decreases to 1 percent of the maximum concentration or until a
period of 200 years has been simulated; however, media concentrations are simulated until the
constituent concentration in a particular medium (e.g., ground water) decreases to 1 percent of
the maximum concentration for that medium or until a period of 10,000 years has been
simulated. Therefore, modeling simulations can range from a few hundred years (for
contaminants that move quickly in the environment, are not persistent, and do not
bioaccumulate) to 10,000 years (for the most persistent and least mobile contaminants, such as
some metals).
The risk results are evaluated over the entire modeling period, which could be up to
10,000 years. Evaluating peak doses over this time horizon allows the system to capture the
slow movement of certain contaminants through the subsurface. Although the time frame for
such travel may be long, such contamination could be a serious problem when the contaminant
reaches the receptor wells. Particularly for contaminants that do not degrade, it is important to
determine the magnitude of risk that would be experienced once the contamination does reach a
drinking water well. The selection of a long time frame assumes that peaks are more likely to be
considered in the assessment (see Section 2.2.1 for further discussion).
Regional Site-Based Data. The 3MRA modeling system was designed to be
implemented using a site-based approach. In this approach, site-based data are used as inputs to
the system when available. When site-based data are not available, data collected on a regional
level, followed by data collected on a national level, are used for the evaluation. Using site-
based and regional data to create data sets that represent actual site and environmental conditions
ensures that implausible exposure pathways and scenarios are not included.
Detailed documentation regarding the methodology for gathering data, what data were
collected, where they were obtained, how they were collected and processed, and other issues
and uncertainties is provided in Volume II. Examples of the types of data collected as site-based
characteristics include facility locations and the physical and environmental characteristics of the
sites and surrounding areas (e.g., land use, human receptor locations, and ecological habitats).
Regional data collected include meteorological data, some soil characteristics, aquifer data, and
types of ecological receptors. Data collected at the national level include human exposure
factors, ecological exposure factors, human health toxicity values, and ecological toxicity values.
Variability and Uncertainty Analysis. The 3MRA modeling system was designed
specifically to support the analysis of uncertainty and variability in the risk outputs. The site-
based approach provides a measure of intrasite variability in risk or hazard in terms of both
spatial and temporal variables. With the current representative national data set, the
representative sites provide a measure of nationwide intersite variability across industrial waste
management sites; that is, the sites cover the range of different land-based WMU types and
environmental settings. In addition, the 3MRA modeling system has been designed for a two-
stage Monte Carlo analysis. For many inputs, distributions were developed to represent their
variability.
2-5
-------
Section 2.0
Modeling Approach
2.2 Spatial and Temporal Scale
The 3MRA framework uses a site-based concept to address spatial issues. The data
requirements for a multimedia modeling system that includes human and ecological receptors are
considerable. Developing a realistic data set for modeling purposes requires consistency among
all of the input data, such that the input data set represents a realistic, internally consistent
environmental setting. To accomplish this, EPA adopted a site-based approach for the 3MRA
modeling system, in which site-specific data, supplemented by regional and national data, are
used to construct the environmental setting to be modeled. The inputs to the modeling system
can be represented by distributions to reflect variability, and multiple distributions for selected
variables can be developed to reflect uncertainty in those variables; however, the uncertainty and
variability with respect to the modeling system inputs are all accounted for within the construct
of a specific site. To conduct a national-scale assessment, multiple sites that either account for
all relevant sites or statistically represent the population of such sites may be used as inputs to
the 3MRA modeling system.
The 3MRA modeling system also manages time series data to address temporal issues.
The multimedia, multipathway aspect of the modeling system results in exposures through the
various pathways occurring at different times. Sometimes exposures can occur hundreds of
years apart, especially when considering a ground water pathway versus an aboveground
pathway. Thus, the ability to track contaminant concentrations and resulting exposures across all
pathways on a common time scale is required. Although this time scale can vary depending on
the needs of the analysis, a 1-year time step is used for outputs from all modules because the
3MRA modeling system was developed primarily for chronic exposures. Some of the modules
use much shorter time steps to produce these annual outputs. For example, the Air Module uses
an hourly time step, but provides annual averages as outputs.
2.2.1 Model Spatial Scale
The spatial scale for the 3MRA modeling system is defined in terms of the AOI
surrounding a WMU or other source. The size of the AOI will depend on the goals of a
particular application, as well as the ability to collect the needed data for the AOI. For most
RCRA WMUs other than incinerators and other types of combustors, the AOI will be relatively
constrained because of the operational nature of these sources. For the representative 201 sites
in the modeling system, a distance of 2 km was selected for the AOI. If analysts or risk
managers are interested in larger AOIs or in areas closer in to the WMU, site-based data sets can
be created to provide additional spatial coverage or resolution for the risk results. By default, the
3MRA modeling system estimates risks for the following distance rings, as shown in Figure 2-4:
¦ Human risks are totaled within the 0 to 0.5 km, 0 to 1 km, and 0 to 2 km rings
around the WMU.
¦ Ecological risks are totaled within the 0 to 1 km, 1 to 2 km, and 0 to 2 km rings
around the WMU.
2-6
-------
Section 2.0
Modeling Approach
Within the AOI, two key spatial
considerations are the location of watersheds and
associated waterbodies, and the location of
receptors (both human and ecological). These
spatial considerations, in combination with other
location-specific information, such as WMU
characteristics, meteorology, hydrogeology, and
land uses, constitute the environmental setting
that is modeled. Figure 2-5 illustrates the
delineation of watershed subbasins in the 3MRA
modeling system. The watershed subbasins
define the areas over which deposition rates and
soil concentrations are averaged and assumed to
be uniform. Although watershed size is
determined mainly by the topography and
hydrography at a site, the size of the subbasins
within the watersheds is defined during
watershed delineation. Currently, these are
defined in the 3MRA modeling system so that
there are generally about 10 to 12 watershed subbasins within the AOI at a site, with an average
subbasin area within the AOI of about 1 million square meters. This provides the spatial
resolution needed to map soil concentration gradients across the site while keeping the total
number of watershed subbasins at a given site to a reasonable number.
1,000 m
Site Location
2,000 m
AOI
~ WMU Boundary ^
I I Watershed Subbasin Boundary
Figure 2-5. Example of watershed delineation for a typical site.
WMU
0.5 km
1 km
Figure 2-4. Current area of interest and
concentric distance rings for human risk.
2-7
-------
Section 2.0
Modeling Approach
Figure 2-6 illustrates
human receptor locations at an
example site. Within the current
representative national data set,
human receptor locations at a site
are defined by the centroids of
Census blocks from the 1990 U.S.
Census. If a distance ring crosses
a Census block, then that block is
divided into two components,
each with a centroid, and the
population associated with the
Census block is apportioned to
the two centroids based on the
area of each of the two
components. The density of
receptor points varies with
population density because
Census blocks are sized by the
population they contain.
WMU
— Concentric Distance Rings
— Census Blocks
• Human Receptor Points
Figure 2-6. Example of human receptor
placement for a typical site.
Ecological receptor
locations are defined in terms of habitats and home range areas. Figure 2-7 presents a site
example with five habitats delineated. The representative habitats are delineated for each site
based on a variety of geographic information system (GIS) coverages of topography, land use,
wetlands, and waterbodies. Within each habitat, receptor home ranges are placed at random.
For each receptor, concentrations in contaminated media (soil or surface water), plants, and prey
items are averaged over the home range. Various types of ecological receptors are placed within
a home range, based on the available data for that particular receptor.
2.2.2 Model Temporal Scale
The 3MRA modeling system is designed to evaluate chronic exposures, with individual
module results reported as a time series of annual average concentrations or fluxes. However,
the time scale can be adjusted to be shorter or longer depending on the needs for a particular
assessment. Individual modules may use input data on a shorter time scale to capture physical
processes that vary within a year, and thus provide a more realistic estimate of the annual
average. For example, annual average precipitation data do not provide details on storm events,
which have a major effect on runoff and erosion processes that occur over a year. Therefore, the
Land Application Unit (LAU) and Watershed Modules use daily precipitation data to more
accurately estimate precipitation-driven runoff and erosion events. Similarly, for the Surface
Impoundment Module, monthly temperature data can be used to capture temperature extremes
across seasons, which can affect volatilization.
2-8
-------
Section 2.0 Modeling Approach
Forest
Residential (2)
Permanently
Flooded
Forest (2)
Permanently
F^HIl
Forest (1)
u
¦SI
K
Figure 2-7. Example of representative ecological habitats
delineated for typical site.
The time frame for estimating exposure and risk depends on a number of variables,
including distance of receptors from the WMU, direction of receptors from the WMU, and
physical-chemical properties affecting the transport and fate of contaminants in the receiving
media. For most media (i.e., air, surface water, soil), the exposure and risk occur in the same
time frame as the release from the WMU. For ground water, where the medium and chemical
properties attenuate the transport process, the exposure and risk time frame can be hundreds to
thousands of years after the release. The time frame, therefore, varies for each contaminant and
environmental medium considered for each specific WMU. The 3MRA modeling system
calculates concentrations for each pathway for each human receptor or habitat location until less
than 1 percent of the maximum contaminant concentration in the medium is left in the medium,
up to a maximum of 10,000 years. This is intended to capture the large majority of exposures
from most contaminants that would be included in an analysis. A few metals and dioxin-like
substances may move so slowly in the subsurface that exposures within the AOI may not be fully
captured even in the 10,000-year time frame.
A given receptor is subject to exposures from various (but not necessarily all) pathways
simultaneously, depending on whether there is a contaminant concentration in each pathway.
The aggregate risk to any individual receptor is defined as the sum of the risks from each
pathway over a given time period. In the 3MRA modeling system, all exposures are calculated
on an annual basis. However, the risk can be based on an exposure duration defined for the
analysis. All exposure durations greater than 1 year would be the average daily exposure (or
average daily dose) over consecutive years that make up the duration. For example, if there were
2-9
-------
Section 2.0
Modeling Approach
100 years of exposure output and the exposure duration was defined as 10 years, there would be
91 10-year averages from which to calculate risk or hazard, beginning with years 1 to 10, 2 to
11, and 3 to 12 and ending with 91 to 100.
Given that the exposure in the different media can occur over significantly different time
periods for each receptor location, aggregation of risk is performed for exposures that occur at
the same time. For instance, exposures and risks due to contaminated air occurring during years
1 to 10 are not aggregated with exposures and risks due to contaminated ground water occurring
during years 91 to 100. Figure 2-8 illustrates how risks during the same time periods are
overlaid and aggregated across exposure pathways for a given receptor and contaminant. Risks
are aggregated across different exposure routes (i.e., ingestion, inhalation) only after considering
current EPA practices for combining exposures through different routes.
Risk due
to Pathway 1
Time
Risk due
to Pathway 2
Time
Risk due
to Pathways
1 &2
Combined
Time
Figure 2-8. Illustration of concurrent time aggregation of risks.
2-10
-------
Section 2.0
Modeling Approach
Once risk estimates are calculated for each receptor at a location within the AOI, the
critical time period during which the maximum risk and/or HQ occurs across the population for
each receptor/cohort combination and for each exposure pathway and pathway aggregation is
determined. The critical time period is determined in a similar fashion in the Ecological Risk
Module for receptor types, groups, and distances. Types of outputs for the critical period include
population-weighted risk or HQ; media-specific concentrations at receptor location; pathway-
specific exposures for each receptor/cohort or indicator species at each receptor location; and
various aggregations of media concentrations, pathways, or receptors.
2.3 Design of the 3MRA Modeling System Multipathway Modules
The design of the 3MRA modeling system takes into consideration how constituents are
released to the environment, transported through various media, and result in exposure to
humans and ecological receptors, and how these exposures are characterized in terms of risk
measures. The major considerations in determining the design of the science modules in the
3MRA modeling system were
¦ Chemicals in the 3MRA database;
¦ Sources (WMUs);
¦ Transport media, fate processes, and intermedia contaminant fluxes;
¦ Food chain/food web components;
¦ Human exposure and risk; and
¦ Ecological exposure and risk measures.
2.3.1 Chemicals in the 3MRA Modeling System Database
The RCRA hazardous waste program
covers more than 400 constituents. These
constituents have a large range of physical-
chemical properties, toxicological properties,
and behavior in the environment. Some of
these constituents are well characterized with
respect to these properties and others are not.
In developing the 3MRA modeling system,
different approaches (e.g., algorithms) were
required to appropriately model some groups of
constituents versus others because of
differences in properties. Five broad categories were established to differentiate the behavior of
contaminants in the environment: organic chemicals, metals, mercury, dioxin-like chemicals,
and special chemicals (such as those that readily metabolize or those that are completely soluble
in water). Mercury and dioxin-like chemicals are specific subcategories for which EPA has
developed special approaches for modeling release and fate and transport.
Table 2-1 shows the 46 constituents selected to develop and test the 3MRA modeling
system. These constituents were selected to ensure coverage of the range of contaminant
behavior in the environment. To facilitate this selection process, all RCRA-regulated
Five categories used in the 3MRA
modeling system to differentiate the
behavior of contaminants in the
environment:
¦ Organic chemicals
¦ Metals
¦ Mercury
¦ Dioxin-like chemicals
¦ Special chemicals (e.g., readily metabolizable)
2-11
-------
Section 2.0
Modeling Approach
Table 2-1. Constituents in the 3MRA Database
3MUA
( onsiiiiicnl l \|)o
C cinsl it noil 1 ( hiss
( onsliliionl Niimo
CASUS
organic (1)
organonitrogen (1)
Acetonitrile
75-05-8
organic (1)
organonitrogen (1)
Acrylonitrile
107-13-1
organic (1)
organonitrogen (1)
Aniline
62-53-3
organic (1)
aromatic hydrocarbon (4)
Benzene
71-43-2
organic (1)
organosulfur (7)
Carbon disulfide
75-15-0
organic (1)
chlorinated aromatic (8)
Chlorobenzene
108-90-7
organic (1)
chlorinated hydrocarbon (9)
Chloroform
67-66-3
organic (1)
chlorinated pesticide (10)
2,4-Dichlorophenoxyacetic acid
94-75-7
organic (1)
brominated hydrocarbon (11)
Ethylene dibromide
106-93-4
organic (1)
misc. halogenated (12)
Hexachloro-1,3 -butadiene
87-68-3
organic (1)
chlorinated pesticide (10)
Methoxychlor
72-43-5
organic (1)
carbon/hydrogen/oxygen (6)
Methyl ethyl ketone
78-93-3
organic (1)
carbon/hydrogen/oxygen (6)
Methyl methacrylate
80-62-6
organic (1)
chlorinated hydrocarbon (9)
Methylene chloride
75-09-2
organic (1)
organonitrogen (1)
Nitrobenzene
98-95-3
organic (1)
nonhalogenated phenolic (15)
Phenol
108-95-2
organic (1)
organonitrogen (1)
Pyridine
110-86-1
organic (1)
chlorinated hydrocarbon (9)
T etrachloroethylene
127-18-4
organic (1)
carbamate group (17)
Thiram
137-26-8
organic (1)
aromatic hydrocarbon (4)
Toluene
108-88-3
organic (1)
chlorinated hydrocarbon (9)
Trichloroethylene
79-01-6
organic (1)
chlorinated hydrocarbon (9)
1,1,1 -Trichloroethane
71-55-6
organic (1)
oxoanion metal (2)
Vanadium
7440-62-2
organic (1)
chlorinated hydrocarbon (9)
Vinyl chloride
75-01-4
metal (2)
oxoanion metal (2)
Antimony
7440-36-0
metal (2)
oxoanion metal (2)
Arsenic
7440-38-2
metal (2)
cationic metal (3)
Barium
7440-39-3
metal (2)
cationic metal (3)
Beryllium
7440-41-7
metal (2)
cationic metal (3)
Cadmium
7440-43-9
metal (2)
oxoanion metal (2)
Chromium (hexavalent)
18540-29-9
metal (2)
oxoanion metal (2)
Chromium (total)
7440-47-3
metal (2)
oxoanion metal (2)
Chromium (trivalent)
16065-83-1
metal (2)
cationic metal (3)
Lead
7439-92-1
metal (2)
cationic metal (3)
Nickel
7440-02-0
metal (2)
oxoanion metal (2)
Selenium
7782-49-2
metal (2)
cationic metal (3)
Silver
7440-22-4
(continued)
2-12
-------
Section 2.0
Modeling Approach
Table 2-1. (continued)
3MUA
C onsiiiiienl l \|K'
C cinsl it noil 1 ( hiss
( onsiiiiionl Niimo
CASUS
metal (2)
oxoanion metal (2)
Thallium
7440-28-0
metal (2)
cationic metal (3)
Zinc
7440-66-6
mercury (3)
organometallic (13)
Methyl mercury
7439-97-6m
mercury (3)
cationic metal (3)
Mercury (elemental)
7439-97-6e
mercury (3)
cationic metal (3)
Mercury (total)
7439-97-6
dioxin-like (4)
dioxin/furan (16)
2,3,7,8-Tetrachlorodibenzo-p-dioxin
1746-01-6
special (5)
polynuclear aromatic (5)
Benzo[a]pyrene
50-32-8
special (5)
carbon/hydrogen/oxygen (6)
Bis-(2-ethylhexyl)phthalate
117-81-7
special (5)
polynuclear aromatic (5)
Dibenz [a,h] anthracene
53-70-3
special (5)
chlorinated phenol (14)
Pentachlorophenol
87-86-5
constituents were sorted into 17 representative constituent classes with similar physical-chemical
properties. Only constituents with adequate constituent-specific toxicity and physical-chemical
properties data were considered for model testing.
The specific properties used to establish these 17 constituent classes included
¦ Degree of aromaticity (the number and arrangement of benzene rings);
¦ Volatility;
¦ Presence of halogens, such as bromine and chlorine;
¦ Presence of other key elements, such as oxygen, nitrogen, sulfur, and/or
phosphorus;
¦ Use (e.g., pesticides);
¦ Presence of organic functional groups, such as phenols and carbamates; and
¦ Similarities in ionic behavior (e.g., anionic and cationic metals).
Candidate constituents were selected from each of the 17 classes by considering
toxicology, fate and transport properties, waste chemistry, and other considerations, such as
¦ Total number of RCRA constituents within a group,
¦ Range of expected toxicity of the constituents within a group,
¦ Availability of defensible physical-chemical property data and analytical
methods,
2-13
-------
Section 2.0
Modeling Approach
¦ Differences in constituent structures within a group,
¦ Differences in degree or type of halogenation (chlorinated or brominated),
¦ The mix of isomers represented in the toxicity data,
¦ The potential for constituents to accumulate in food or prey,
¦ The degradation products required to understand the toxicity for a constituent,
¦ The significance of constituents to other EPA programs or their
representativeness of a given class of chemicals (e.g., 2,3,7,8-TCDD for
halogenated dioxins and furans), and
¦ The frequency or expectation of finding the constituent in many process waste
streams.
The chemical properties were obtained through a combination of modeling, existing
databases, and literature review. Many properties were modeled using SPARC (System
Performs Automated Reasoning in Chemistry) and MINTEQA2. These models are described
briefly below. Table 2-2 summarizes how the various chemical property values were obtained.
Table 2-2. Methodology and Data Sources for 3MRA Chemical Properties
I'l-iipi-m
Kill'mill's
Organic Chemicals
Thermodynamic properties and
partition coefficients3
SPARC-calculated values, adjusted by
CPP for temperature and pH
U.S. EPA (2003)
U.S. EPA(1999f)
Hydrolysis rates
Measured or estimated rate constants
adjusted by CPP for temperature and
PH
U.S. EPA(1996d)
U.S. EPA(1999f)
Aerobic biodegradation rates
Measured values from literature,
grouped by pH and temperature
regimes
Aronson et al. (1999)
Anaerobic biodegradation rates
Measured values from literature,
grouped by pH, temperature, and redox
regimes
U.S. EPA(1999d)
Soil/water partition coefficients
Calculated from Kow by CPP
U.S. EPA(1999f)
Metals
Partition coefficients (waste, soil,
surface water, sediment)
Measured or estimated values,
presented as national distributions
U.S. EPA(1999e)
Sorption isotherms
MINTEQA2 model
U.S. EPA (1998, 1999c)
a Molecular weight, density, volume, vapor pressure, boiling point, air and water diffusion coefficients, solubility, Henry's law
constant, octanol/water partition coefficient (Kow), ionization coefficient.
2-14
-------
Section 2.0
Modeling Approach
SPARC. EPA developed the predictive modeling system SPARC to help meet the
growing need for constituent-specific inputs for multimedia, multipathway, multireceptor risk
assessment tools such as 3MRA. SPARC calculates a large number of physical and chemical
parameters (such as solubility, vapor pressure, Henry's law constant, octanol/water partition
coefficient, air diffusivity, water diffusivity, and ionization potential) from pollutant molecular
structure and basic information about the environment (media, temperature, pressure, pH, etc.)
For more information on SPARC, see U.S. EPA (2003).
MINTEQA2. The MINTEQA2 model was used to develop metal sorption isotherms
that are contained within the 3MRA Vadose Zone and Aquifer Modules and used to provide the
pH and concentration-adjusted soil/water partition coefficients needed to estimate sorption of
metal contaminants in the subsurface. MINTEQA2 is an equilibrium metals speciation model
that can be used to calculate the equilibrium composition of dilute aqueous solutions in the
laboratory or in natural aqueous systems. The model can calculate the equilibrium mass
distribution among dissolved species, adsorbed species, and multiple solid phases under a
variety of conditions including a gas phase with constant partial pressure. For more information
on MINTEQ, see U.S. EPA (1998) and U.S. EPA (1999c).
The 3MRA chemical database can be expanded as applications using this modeling
system require and as constituent data become available.
2.3.2 Sources (WMUs)
The WMUs included in the 3MRA modeling system represent the major management
practices where wastes are put into or on the land for recycling, recovery, reuse, treatment, or
disposal. The various types of WMUs manage different types of industrial waste in different
ways. For example,
¦ Surface impoundments are used to treat and dispose of liquid industrial waste
and generate a sludge that requires further management;
¦ Aerated tanks are used to actively treat industrial liquid waste and generate a
sludge requiring further management;
¦ Landfills are a common disposal site for many nonliquid industrial wastes;
¦ Waste piles are temporary storage areas on the ground for nonliquid industrial
waste, such as ash or slag; and
¦ LAUs are used to reuse, treat, or dispose of industrial waste in liquid, semiliquid,
or solid form. Some wastes are used as a soil amendment, which is a reuse
practice; some wastes are applied to land for treatment through biological
degradation; and some wastes are applied to land as a disposal method.
As illustrated in Table 2-3, each WMU source is modeled to conserve mass. That is, the
WMU source boundaries are defined, and mass balance equations are derived to describe the
various transport and loss mechanisms that affect the amount (mass or concentration) of
2-15
-------
Section 2.0
Modeling Approach
Table 2-3. WMU Types and Source Term Characteristics
Source Term ( hiiriiclcrislics
WMl Tjpe
Miiss ISiiliiucc
Piirliliouiu^
l-irsl-ordcr
l)cur;idiilioii
l-iuilc Source
Surface Impoundment
~
~
~
~
Aerated Tank
~
~
~
~
Landfill
~
~
~
~
Waste Pile
~
~
~
~
LAU
~
~
~
~
constituent within the defined WMU. Each of the WMU source mass balances provides and
accounts for partitioning of the constituent of interest among solids (sediment or soil particles),
water, and air. First-order degradation terms are also included to account for constituent losses
via hydrolysis and biodegradation.
Each of the WMUs included in the 3MRA modeling system is shown in Table 2-4, along
with release mechanisms and directly affected media. Surface impoundments, landfills, waste
piles, and LAUs were selected because they are the most likely destinations for industrial
nonhazardous waste, according to an EPA industrial waste screening study (Westat, 1987).
Aerated tanks were selected because hazardous waste management has been shifting from
surface impoundments to aerated tanks since the 1987 study was conducted.
Table 2-4. WMU Types and Release Mechanisms Modeled
Kcleiisc Mccliiiiiism (rccci\iu» medium)
KimolT/crosioii
l.ciichiu^ (ground Yohilili/iiliou \\ iudhlowii riuM (soil ;ind miiT;icc
WMl Tjpe \\iiler) (;iir) ciir) «;iler)
Surface Impoundment
~
~
Aerated Tank
~
Landfill
~
~
~
Waste Pile
~
~
~
~
LAU
~
~
~
~
As shown in Table 2-4, constituents enter four environmental media after direct release
from a WMU: (1) air, where volatile and particulate releases disperse and are deposited onto
soil, plants, and surface water; (2) soil, which receives runoff and eroded solids from WMUs;
(3) surface water, which receives runoff and eroded solids from WMUs; and (4) ground water,
which receives dissolved constituents leached from wastes in land-based units by infiltrating
precipitation.
2-16
-------
Section 2.0
Modeling Approach
Certain characteristics were assumed for each of the five WMU types to complete the
waste management scenario and to capture the variability in management practices. These
characteristics drive the design and parameterization of the 3MRA source modules; they are
described briefly below and in more detail in Sections 4 and 5 of this volume.
Surface Impoundment. The surface impoundment is used for the management of liquid
wastes in an earthen basin. Constituents are released during the lifetime of the unit. The
impoundment does not receive or contribute runoff to the watershed. No liner other than native
soils is assumed to be present, and there is no cover to reduce volatile emissions. The unit is
assumed to contain two well-mixed phases: liquid and sediment. The solids in the liquid waste
settle to the bottom of the impoundment over time and fill pore space in the native soils. This
attenuates the transport or leaching of constituents to underlying soils and ground water. In
addition, a fraction of each surface impoundment is aerated, which enhances biodegradation and
increases volatilization of some constituents. Each surface impoundment is assumed to operate
for 50 years and then undergo clean closure—that is, all waste is removed from the unit.
The surface impoundment releases leachate to the unsaturated zone and volatile
emissions to air. The surface impoundment is assumed to have a berm to prevent overrun;
therefore, runoff is assumed not to occur. However, volatilized releases can be deposited on the
surrounding soils, and erosion and runoff from these soils are modeled. The model accounts for
hydrolysis, volatilization, and sorption, as well as settlement, resuspension, growth and decay of
solids, and activated aerobic biodegradation (that is, a higher rate based on the amount of
biomass present) in the liquid phase, and hydrolysis and anaerobic biodegradation in the
sediment.
Aerated Tank. The aerated tank is used for the treatment of wastewaters. Only tanks the
size of a drum or larger are modeled. Each aerated tank has a maximum lifetime of 20 years;
therefore, the operating lifetime assumes a tank replacement every 20 years.
The aerated tanks are assumed not to fail or leak liquids to soil or surface runoff. As a
result, the only release pathway for an aerated tank is volatile emissions to air. Constituent loss
within the tank is simulated through hydrolysis and aerobic biodegradation.
Landfill. The design of the landfill assumes a series of vertical cells of equal volume that
are filled sequentially, with each cell being filled in one year. The landfill model is based on the
assumption that the constituent mass in the landfill cells may be linearly partitioned into
aqueous, vapor, and solid phases. Releases from the landfill can occur over the active lifetime of
the landfill (30 years) and continue until virtually all of the constituent mass is released, or, for
immobile constituents, releases can continue for a specified number of years.
The landfill modeled in the 3MRA system assumes minimal controls and is constructed
below grade. In particular, the unit has no liner, and the cover at closure is a layer of natural
soil. The design of the landfill allows volatilization and particle emissions, as well as
infiltration, which leaches constituents from the waste. The belowgrade design prevents runoff
and erosion. In addition, anaerobic biodegradation and hydrolysis processes can degrade
constituents within the landfill.
2-17
-------
Section 2.0
Modeling Approach
Waste Pile. The waste pile manages wastes in a storage pile on the ground (above
grade). The wastes are removed and the pile site closed at the end of its active life (assumed to
be 30 years). Constituents are released only during the operating lifetime of the pile. Waste pile
height and area are set and held constant at each site based on the volume of waste stored. The
waste in the pile is assumed to be completely removed annually and replaced with new waste.
The waste pile is assumed to be uncovered, so that constituents can be released through
infiltration and leaching, volatilization, particle entrainment by wind, and erosion and runoff. In
addition, constituents can be degraded by hydrolysis and aerobic degradation at the surface of
the pile and by hydrolysis and anaerobic degradation within the pile. Although the waste pile
design does not incorporate management controls, the pile is assumed to be located within a
local watershed in such a way that there is no run-on of uncontaminated soil onto the pile.
For the waste pile, after runoff and erosion have occurred for some period of time,
downslope land areas will be contaminated and their soil concentrations can approach the
residual constituent concentrations in the waste pile itself. Thus, after extensive runoff and
erosion from a waste pile, the entire downslope surface area can be considered a source and
becomes an extended source area in the model construct. Consequently, a holistic modeling
approach was taken with the waste pile to incorporate it into the watershed of which it is a part.
Land Application Unit. The LAU allows for the placement of wastes in an open field for
degradation, treatment, or disposal. Some industrial wastes are applied to land for purposes of
soil amendment, thus constituting a reuse of the waste. The waste is applied to the surface soil
periodically and then tilled into the top layer of the soil during each of the assumed 40 years of
operation. Releases from the LAU can occur over the active lifetime of the LAU and beyond,
and continue until virtually all of the constituent mass is released or, for immobile constituents,
for a specified number of years. Other than tilling into the soil and a crop cover, no management
controls are present to limit releases from the LAU.
Constituents are released from the LAU by leaching to the unsaturated zone,
volatilization and particulate emissions to the air, runoff of dissolved constituents, and erosion
and runoff of particles. Constituent losses include hydrolysis and aerobic biodegradation. Also,
LAUs are on the land surface and are an integral part of their respective watersheds.
Consequently, they receive run-on and erosion from upslope land areas and affect downslope
land areas through runoff and erosion.
For the LAU, after runoff and erosion have occurred for some time, downslope land areas
will be contaminated, and their soil concentrations can approach the residual constituent
concentrations in the LAU (or conceivably even exceed them, long after operation ceases).
Thus, after extensive runoff and erosion from the LAU, the entire downslope surface area can be
considered a source and becomes an extended source area in the model construct. Consequently,
a holistic modeling approach was taken with LAU to incorporate it into the watershed of which it
is a part.
2-18
-------
Section 2.0
Modeling Approach
2.3.3 Transport Media, Fate Processes, and Intermedia Contaminant Fluxes
The 3MRA modeling system has five
main environmental transport modules: the Air
Module, the Watershed Module, the Surface
Water Module, the Vadose Zone Module, and the
Aquifer Module. Four of these modules (Air,
Surface Water, Vadose Zone, and Aquifer) are
based on existing well-established, peer-reviewed
regulatory models. These legacy models provide
both state-of-the-art environmental transport and
fate modeling and well-established acceptability
through various peer-review cycles and
applications that have undergone stringent
scrutiny. They have been incorporated into the
3MRA modeling system through the use of pre-
and postprocessors that integrate the specific
input and output requirements of these legacy
models into the 3MRA modeling system. Within
the modules, all functionality of the original
model is maintained, including degradation
processes, transformation processes, and release
processes. Each of the transport modules is
summarized below.
Air Module. The Air Module is based on
the Industrial Source Complex Short-Term
Model, Version 3 (ISCST3). ISCST3 uses a
steady-state Gaussian plume modeling approach,
which assumes that the plume of emissions
follows a straight-line path in the direction of the
mean wind flow, and the contaminant
concentrations in the horizontal and the vertical
directions (both orthogonal to the mean wind
direction) follow a normal distribution. The
model provides estimates of contaminant air concentration, dry deposition rates (particles only),
and wet deposition rates (particles and vapors) for user-specified averaging periods (i.e.,
annually). It also has the capacity to model area sources or point sources. A point source is
usually considered a source with a small area, such as a stack or vent. An area source is a source
that covers some defined area that is large enough to have an impact on the dispersion across the
source. The 3MRA modeling system sources are all area sources. There is also a regulatory
definition of area source that refers to the quantity of emissions from a particular source. That
definition does not apply to this discussion.
In the Air Module, ISCST3 is used to model the transport and diffusion of contaminants
in the form of volatilized gases or fugitive dust emitted from area sources. The air
concentrations are used to estimate bio-uptake from plants, and human exposures due to direct
3MRA Modeling System Transport
Media, Fate Processes, and
Intermedia Contaminant Fluxes
Transport Media
• Atmosphere
• Watershed
• Subsurface (vadose zone and aquifer)
• Surface water
Fate Processes
• Chemical/biological transformation (and
associated products)
• Linear partitioning (water/air, water/soil,
air/plant, water/biota)
• Nonlinear partitioning (metals in vadose
zone)
• Chemical reactions/speciation (mercury in
surface waters)
Intermedia Contaminant Fluxes
• Air
Watershed/farm habitat
soil (wet/dry deposition,
vapor diffusion)
• Air
Surface water (wet/dry
deposition, vapor
diffusion)
• Watershed soil
Surface water (erosion,
runoff)
• Surface water
Sediment
(sedimentation)
• Vadose zone
Ground water
(infiltration)
• Watershed soil
Air (volatilization)
• Ground water
Surface water
2-19
-------
Section 2.0
Modeling Approach
inhalation. The predicted deposition rates are used to determine contaminant loadings to
watershed soils, terrestrial habitats, farms, and surface waters.
ISCST3 is used as legacy code in the 3MRA modeling system. That is, the model was
left intact and the necessary interfacing to the modeling system is handled using pre- and
postprocessors. Together, the EPA air quality model (ISCST3) and the pre- and postprocessing
code that integrates ISCST3 into the 3MRA modeling system are referred to as the Air Module.
The pre- and postprocessing code also provides additional functionality to support other 3MRA
modeling system requirements. Additional detail on the Air Module can be found in Section 6 of
this document.
Watershed Module. Contaminant mass can be released from a WMU in the form of
volatile and particulate emissions. These emissions can then be transported and deposited onto
the soils of nearby land areas as wet or dry deposition (estimated by the Air Module). Once
deposited, a contaminant is then subject to fate and transport processes, such as runoff, erosion,
and degradation. Contaminants in the soil of the AOI are available either for direct exposure to
human or ecological receptors or for indirect exposure through uptake in a food web. The
Watershed Module accounts for these fate and transport processes across the AOI.
The loss mechanisms for contaminants deposited onto soils are volatilization, leaching,
and biological and/or contaminant degradation. The fate and transport mechanisms for
contaminants that are deposited onto soils include runoff and erosion into adjacent waterbodies.
Because the surface transport processes are hydrologically related, the land areas surrounding the
WMU are disaggregated into watershed subbasins. A watershed subbasin can vary in size from
a portion of a hillside to much larger areas encompassing regional stream or river networks. In
all cases, a watershed subbasin is treated as a single, homogeneous area with respect to soil
characteristics, runoff and erosion characteristics, and contaminant concentrations in soil. No
spatial disaggregation below the watershed subbasin level is made; that is, no spatial
contaminant concentration gradients are estimated within a given watershed subbasin.
Additional detail on the Watershed Module can be found in Section 7 of this document.
Surface Water Module. The Surface Water Module has the ability to model streams,
lakes, ponds, and wetlands at a site. Constituent mass released from a WMU can enter a nearby
surface waterbody network through runoff and erosion directly from the WMU, atmospheric
deposition to the water surface, runoff and erosion from adjoining watershed subbasins, and
interception of contaminated ground water. The chemical is then subject to transport and
transformation processes occurring within the waterbody network, resulting in variable chemical
concentrations in the water column and underlying sediments. These chemical concentrations
are the basis for direct exposure to human and ecological receptors and indirect exposure through
uptake in the aquatic food web.
The Surface Water Module consists of a core model, Exposure Analysis Modeling
System II (EXAMS II) (Burns, 1997; Burns et al., 1982), and the pre- and postprocessors for the
model. The Surface Water Module estimates annual average total and dissolved chemical
concentrations in the water column and sediments in stream reaches, ponds, and lakes.
Transport and transfer processes modeled include advection, vertical diffusion, volatilization,
deposition to the sediment bed, resuspension to the water column, and burial to deep sediments.
2-20
-------
Section 2.0
Modeling Approach
Transformation processes include hydrolysis and biodegradation as first-order reactions
influenced by temperature and pH. Outputs from the Surface Water Module include water
column and sediment concentrations that are used by the Aquatic Food Web Module and the
Ecological Exposure Module.
Exams II is used as legacy code in the 3MRA modeling system. That is, the model is left
intact and the necessary interfacing to the framework is handled using pre- and postprocessors.
Together, the Exams II model and the pre- and postprocessing code that integrates Exams II
into the 3MRA modeling system are referred to as the Surface Water Module. Additional detail
on the Surface Water Module can be found in Section 8 of this document.
Subsurface (Vadose Zone and Aquifer) Modules. The Vadose Zone and Aquifer
Modules consist of components from the EPA Composite Model for Leachate Migration with
Transformation Products (EPACMTP) (U.S. EPA, 1996a,b,c) and pre- and postprocessors to
provide compatibility with the 3MRA modeling system. The Subsurface Modules account for
the fate and transport of constituents released from WMUs into the underlying unsaturated or
vadose zone (soil) and saturated zone (surficial aquifer). The Subsurface Modules can consider
the formation and transport of transformation products, the impact of ground water mounding on
ground water velocity, finite source and continuous source scenarios, and metal transport.
The Vadose Zone Module consists of a one-dimensional (1-D) model that simulates
infiltration and dissolved chemical transport through the unsaturated zone. The Vadose Zone
Module accounts for sorption and attenuation of a chemical moving through the soils under the
WMU. The Aquifer Module consists of a 3-D model that simulates transport through the
saturated zone. The Aquifer Module consists of a 3-D ground water flow submodel and a 3-D
contaminant transport submodel. The flow submodel accounts for the effects of leakage from a
land disposal unit and regional recharge on the magnitude and direction of ground water flow.
The contaminant transport submodel accounts for 3-D advection and dispersion and linear or
nonlinear equilibrium sorption.
The Vadose Zone Module receives infiltration and solute mass fluxes from the source
modules. The migration of chemicals in the vadose zone is terminated at the water table where
the chemical fluxes, in the form of concentrations, are transferred to the Aquifer Module. The
Aquifer Module also receives areal recharge from the Watershed Module. The Aquifer Module
provides time-dependent, annual average chemical concentrations at receptor wells and annual
average chemical fluxes at an intercepting stream, when present, in the AOI.
Both the Vadose Zone and Aquifer Modules can be described as legacy code in the
3MRA modeling system. The modules were left intact and the necessary interfacing to the
framework is handled using pre- and postprocessors. The pre- and postprocessing code also
provides additional functionality to support other 3MRA modeling system requirements.
Additional detail on the Subsurface Modules can be found in Section 9 of this document.
2.3.4 Food Chain/Food Web Components
The 3MRA modeling system has three food web components: the Farm Food Chain
Module, the Terrestrial Food Web Module, and the Aquatic Food Web Module. The processes
2-21
-------
Section 2.0
Modeling Approach
in each component include the uptake by plants and
animals of chemicals released from the sources and
transported through the environmental media. The
Farm Food Chain Module provides estimates of
chemical concentrations in homegrown produce and
crops, beef, and milk produced on a farm. The
Terrestrial and Aquatic Food Web Modules provide
estimates of chemical concentrations in plants and
animals that are used to evaluate the hazards to
ecological receptors. The Aquatic Food Web
Module also provides estimates of chemical
concentrations in fish that may be eaten by
recreational fishers.
Farm Food Chain Module. The Farm Food
Chain Module calculates the concentration of a
chemical in homegrown produce (fruits and
vegetables), farm crops for cattle (forage, grain, and
silage), beef, and milk. The concentrations in homegrown produce, beef, and milk are inputs to
the Human Exposure Module and are used to calculate the applied dose to human receptors who
consume them. The modeling construct for the Farm Food Chain Module is based on recent and
ongoing research conducted by EPA ORD and presented in Methodology for Assessing Health
Risks Associated with Multiple Exposure Pathways to Combustor Emissions (U.S. EPA, 2000).
The Farm Food Chain Module is designed to predict the accumulation of a chemical in
the edible parts of a plant from uptake of chemicals in soil and through transpiration and direct
deposition of the chemical onto the plant surface. Concentrations are predicted for four main
categories of food crops presumed to be eaten by humans: exposed fruits and vegetables (i.e.,
those without protective coverings, such as lettuce), protected vegetables (e.g., those with
protective covering, such as corn), protected fruits (i.e., those in which the outer skins or rinds
are not eaten, such as melons or bananas), and root vegetables (e.g., potatoes). In addition, the
chemical concentrations in beef and milk are estimated from the biotransfer of chemical in feed
(i.e., forage, grain, and silage), soil, and drinking water to beef and dairy cattle through
ingestion.
The Farm Food Chain Module predicts the concentration of chemicals in produce grown
by home gardeners and in food crops, beef, and milk produced on farms. The methodology for
home gardeners uses point estimates of air and soil concentrations at the residential receptor
location assigned to each Census block. In contrast, the methodology used for farms calculates
an area-weighted average soil concentration for the farm and the average air concentration across
the area of the farm. Thus, the predicted concentrations in farm food crops reflect the spatial
average for the farm. Similarly, feed concentrations for cattle are derived using spatial averages.
In predicting concentrations in beef and milk, the contribution from contaminated drinking water
sources, such as farm ponds or wells on the farm, is also considered. However, irrigation of
crops and home gardens is not modeled.
Food Chain/Food Web Fluxes
• Air Vegetation (particulate
deposition; vapor
diffusion)
• Farm/habitat/
garden soil Vegetation (root uptake,
translocation,
deposition)
• Vegetation, soil,
surface water,
ground water Animals (uptake)
• Surface water,
sediment Aquatic organisms
(uptake)
2-22
-------
Section 2.0
Modeling Approach
Because the behavior of each chemical is affected by its physical-chemical properties, the
module takes into account whether the chemical is organic, metal, mercury, dioxin-like, or
special. For most organic chemicals, the module calculates chemical-specific values for the
biotransfer factors used in the various equations, including air-to-plant biotransfer factor, root
concentration factor, and soil-to-plant biotransfer factor. For metals, dioxin-like chemicals, and
special chemicals, the module generally uses empirical values from the literature for the various
biotransfer factors, when available. If empirical values are not available for dioxin-like
compounds and special chemicals, the biotransfer factors are calculated in the same way as for
organics. Additional detail on the Farm Food Chain Module can be found in Section 10 of this
document.
Terrestrial Food Web Module. The Terrestrial Food Web Module calculates chemical
concentrations in soil, terrestrial plants, and various prey items consumed by ecological
receptors, including earthworms, other soil invertebrates, and vertebrates. These concentrations
are used as input to the Ecological Exposure Module to determine the applied dose to each
receptor of interest (e.g., deer, kestrel). The module is designed to calculate spatially averaged
soil concentrations in the top layer of soil (i.e., surficial soil), as well as deeper soil horizons
(i.e., depth-averaged over approximately 20 cm). The spatial averages are defined by the home
ranges and habitats that are delineated within the AOI at each site. Once the average soil
concentrations are calculated, these values are multiplied by empirical bioconcentration factors
(for animals) and biotransfer factors (for plants) to predict the tissue concentrations for items in
the terrestrial food web.
The Terrestrial Food Web Module was designed to predict a range of concentrations in
plants and prey items to which a given receptor may be exposed. The predator and various prey
are represented in the site layout by allowing the respective home ranges to overlap. For plants
and soil fauna, the Terrestrial Food Web Module estimates concentrations based on the spatially
averaged soil and air concentrations across each home range. Receptors that ingest plants and
soil invertebrates as part of their diet are presumed to forage only within that part of the home
range that is contained within the AOI at a given site. Consequently, the home range defines the
spatial scale for concentrations in soil, plants, and prey (both mobile and relatively immobile) to
which a given receptor is exposed.
As with the Farm Food Chain Module, the Terrestrial Food Web Module modeling
construct is based on recent and ongoing research conducted by EPA ORD and presented in
Methodology for Assessing Health Risks Associated with Multiple Exposure Pathways to
Combustor Emissions (U.S. EPA, 2000). The model can distinguish among different types of
chemicals, using empirically derived algorithms for the various uptake factors for some
chemicals and biouptake data from field or greenhouse studies for other chemicals. The
Terrestrial Food Web Module accounts for uptake via root-to-plant translocation, air-to-plant
transfer for volatile and semivolatile chemicals, and particle-bound deposition to edible plant
surfaces.
To estimate the concentrations in other categories of terrestrial prey items (e.g.,
earthworms, small birds), the Terrestrial Food Web Module relies on soil-to-organism
bioconcentration factors identified from empirical studies and/or generated using regression
methods developed by the Oak Ridge National Laboratory (see, for example, Sample et al.,
2-23
-------
Section 2.0
Modeling Approach
1998a,b). Additional detail on the Terrestrial Food Web Module can be found in Section 11 of
this document.
Aquatic Food Web Module. The Aquatic Food Web Module calculates chemical
concentrations in aquatic organisms that are consumed by humans (e.g., game fish), and in
ecological receptors (e.g., emergent aquatic plants). These concentrations are used as input to
the Human and Ecological Exposure Modules to determine the applied dose to receptors of
interest. The module is designed to predict concentrations in aquatic organisms in both
coldwater and warmwater aquatic habitats.
The underlying framework for the Aquatic Food Web Module is the development of
representative freshwater habitats for warmwater and coldwater systems. Four basic types of
freshwater systems were included for the two temperature categories: streams/rivers,
permanently flooded wetlands, ponds, and lakes. Simple food webs were constructed for each of
the eight freshwater habitats (four coldwater and four warmwater) that specify: (1) the predator-
prey interactions, (2) the physical and biological characteristics of the species that are assigned
to each habitat (e.g., size, lipid content), and (3) the dietary preferences for fish in trophic
levels 3 (TL3) and 4 (TL4). For each freshwater habitat, the feeding guilds (i.e., groups of
species that use environmental resources, such as food, in a similar way) for various types and
sizes of fish were used to construct a simple food web and to map dietary preferences for
organisms in each habitat (U.S. EPA, 1999a). The food web structure and species assignments
are critical in determining concentrations of hydrophobic contaminants in aquatic organisms.
Additional detail on the Aquatic Food Web Module can be found in Section 12 of this document.
To estimate the concentrations in aquatic prey items, the Aquatic Food Web Module
relies on bioconcentration factors identified from empirical studies and/or generated using
regression methods developed by the Oak Ridge National Laboratory (see, for example, Sample
et al., 1998a,b).
2.3.5 Human Exposure and Risk
Within the 3MRA modeling system, human exposure and risk are estimated using two
modules: one for exposure and one for risk. The Human Exposure Module provides an annual
average dose (i.e., in mg/kg-d) and focuses on receptor types, exposure pathways, and routes.
The Human Risk Module calculates risk or hazard and determines the number of people
associated with these risk levels.
The conceptual approach used in developing the human exposure assessment and risk
calculations within the 3MRA modeling system accounts for the major sources of variability in
human exposures and risk. In particular, the approach considers variability through (1) a suite of
human receptors that reflect different behaviors that lead to exposure; (2) a suite of age cohorts
that reflect variability in exposure factors, such as body weight and consumption or intake rates
across ages, as well as some behavior patterns such as showering; (3) the use of distributions for
most exposure factors to account for the variability within a receptor/cohort category (e.g., adult
resident, farmer child aged 12 to 18); (4) the location of receptors that reflect the spatial
variability in contaminant concentration across an AOI, as well as the existence or nonexistence
of some pathways, such as ground water being used as drinking water; (5) the variability in
2-24
-------
Section 2.0
Modeling Approach
contaminant concentration in various pathways
over time; and (6) the variability in the number of
people exposed at various locations within an AOI.
This site-based approach is used on multiple sites
in a national assessment to address the variability
both within an AOI for an individual site and
across multiple sites.
Human Exposure Module. The Human
Exposure Module calculates the applied dose
(milligram of contaminant per kilogram of body
weight) to human receptors from media and food
concentrations calculated by other modules in the
3MRA modeling system. The applied dose is used
for calculating carcinogenic risk for ingestion and
inhalation exposure pathways and for calculating
HQs for ingestion pathways. For inhalation
pathways for noncarcinogens, the air concentration
to which a receptor is exposed (expressed in
milligrams of contaminant per cubic meter of air)
is used. These calculations are performed for each
receptor, cohort, exposure pathway, and year
within each AOI.
Human Receptors. Human receptor types
considered in the 3MRA modeling system are
intended to cover the likely human receptor types
evaluated at the national level. The 3MRA
modeling system can model five different receptor
types: residents, gardeners, beef farmers, dairy farmers, and fishers. Receptor types are used to
differentiate behavior that leads to different profiles of exposure. For example, the exposure
pathways evaluated for a resident include ingestion of soil and ground water and inhalation of
airborne vapors and particulates. For home gardeners and farmers, contaminated foodstuffs are
considered in addition to the pathways for the resident. Census and land use data are used to
identify receptor types and populations that are potentially exposed within an AOI.
Age Cohorts. Each receptor type is divided into five age cohorts: infants, children aged
1 to 5 years, children aged 6 to 11 years, children aged 12 to 19 years, and adults. The only
exposure pathway evaluated for the infant is the breast milk pathway, which is derived from the
mother's exposure. The other age cohorts provide a way of estimating applied dose to different
age groupings that exhibit different distributions of intake rates and body weights. For the
3MRA modeling system, statistical distributions were developed for these exposure factors for
each age cohort using data available in the Exposure Factors Handbook (U.S. EPA 1997). The
3MRA modeling system samples from these distributions using a Monte Carlo approach.
Human Exposure Pathways. The approach used in human exposure assessment is to
predict the type, timing, and magnitude of exposures that receptors may experience as a result of
3MRA Modeling System Human
Receptors, Exposure Routes, and
Risk Measures
Human Receptors*
• Resident
• Home gardener
• Dairy farmer
• Beef farmer
• Fisher**
* For each human receptor type, includes
5 age cohorts
** All other receptor types can also be
fishers.
Age Cohorts
• Infants
• Child aged 1 to 5 years
• Child aged 6 to 11 years
• Child aged 12 to 19 years
• Adult (over 19 years)
Exposure Pathways
• Ingestion (plant, meat, milk, fish, water,
soil, breast milk)
• Inhalation (gases, particulates)
Human Risk Measures
• Cancer (risk probability)
• Noncancer (hazard quotient)
• Population weighted risk distribution
2-25
-------
Section 2.0
Modeling Approach
contact with the contaminants of potential concern. An exposure pathway describes the course
that a contaminant takes from a source to an exposed individual. Exposures are evaluated for all
potentially complete exposure pathways. An exposure pathway is complete when there is a
route by which a human receptor takes up a constituent that was released from the source of
concern. For example, if ground water is predicted to be contaminated, and there are no private
wells located in the contaminated plume or the receptors across the AOI are connected to a
municipal water supply that is not contaminated, the drinking water pathway (private ground
water wells) and the shower inhalation pathway would not be complete.
Exposure routes include uptake mechanisms such as ingestion, dermal contact, and
inhalation. When modeling human exposure, the exposure routes that are included in the Human
Exposure Module include
¦ Ingestion of soil,
¦ Ingestion of contaminated ground water (private ground water wells only),
¦ Inhalation of contaminated shower air (private ground water wells only),
¦ Inhalation of volatile emissions,
¦ Inhalation of particulate emissions,
¦ Ingestion of homegrown produce (gardeners, farmers) and beef and milk
produced on farms (farmers only),
¦ Ingestion of fish caught by recreational fishers, and
¦ Exposure of an infant through ingestion of breast milk from an exposed mother.
These routes define the exposure media to be modeled in the risk analysis (i.e., ground
water, soil, air, vegetables, beef, milk, and fish). EPA considered the inclusion of the dermal
route of exposure but decided that health benchmarks for dermal toxicity are not sufficiently
developed at this time for use in analyses that could support regulatory decisions.
The exposure pathways and routes that can be evaluated for each receptor type are shown
in Table 2-5. Residents can be exposed to contaminants in the air (inhalation) and in soil
(incidental ingestion) and are assumed to be exposed to potentially contaminated ground water
(inhalation and ingestion) if the house is not connected to a public water supply. Home
gardeners are residents who also grow some portion of their fruits and vegetables. Farmers have
the same exposure pathways as home gardeners with the additional exposure to either
contaminated beef or milk (depending on the type of farms present at a site). Recreational
fishers are any of the above receptors with the added pathway of ingestion of contaminated fish
from local streams or lakes. Thus, some fraction of residents, home gardeners, and farmers are
also fishers.
2-26
-------
Section 2.0
Modeling Approach
Table 2-5. Human Exposure Pathways by Receptor Type
Nome Beef Diiin
Resident Homo (>;irdener Beef l-';irmer Ihiin I'nrmer
P;il hw ;i> Resident Kisher (>;irdener Kislier l-'iirmer Kislier l-";i rinoi* Kisher
Air inhalation
~
~
~
~
~
~
~
~
Shower air
inhalation
~
~
~
~
~
~
~
~
Soil ingestion
~
~
~
~
~
~
~
~
Water ingestion
~
~
~
~
~
~
~
~
Crop ingestion
~
~
~
~
~
~
Beef ingestion
~
~
Milk ingestion
~
~
Fish ingestion
~
~
~
~
The Human Exposure Module aggregates exposures across exposure pathways and
routes, when appropriate (e.g., daily doses of beef contaminated by uptake from forage, silage,
grain, soil, and drinking water), and provides estimates of total exposure for the eight routes
listed above.
Within the 3MRA modeling system, the evaluation of human exposure is accounted for
by a contaminant's distribution across a site, a contaminant's concentration profile with time,
and also the variability and uncertainty in exposure factors for each receptor type. The exposure
for each receptor type is estimated at each receptor location within the study area to capture
spatial variability, and for every year over the modeling time frame to capture temporal
variability. Additional detail on the Human Exposure Module can be found in Section 13 of this
document.
Human Risk Module. The Human Risk Module calculates the population-weighted risk
or hazard for each receptor location, receptor type, age cohort, pathway, and exposure period.
Hazard or risk distributions are developed across the site by doing these calculations at each
receptor location, providing a measure of the spatial variability at a site. These same
distributions are developed for each exposure period in the modeling time frame providing a
measure of temporal variability. At each location, the number of people for each receptor
type/cohort is matched with the risk or hazard estimates, providing a population weighting for
each receptor location.
Once the risk or hazard calculations are complete, the model generates the risk
distribution across a site for each exposure period in the modeling time frame. Given the
resulting pathway-specific, receptor-type/cohort-specific, and location-specific risks, HQs, or
margins of exposure (MOEs) (for the breast milk pathway), a time series of population-weighted
risk data is generated for each. This can be thought of as a set of cumulative frequency
histograms. The histogram for any given year is constructed as a series of risk bins defined by
risk or HQ ranges. For any given year, the histogram contains the pathway-specific risk or HQ
distribution of the number of people (corresponding to a given receptor type/cohort) across
locations. Figure 2-9 shows an illustration of this concept of a cumulative frequency histogram
2-27
-------
Section 2.0
Modeling Approach
1800
1600
1400
C
i 1200
_Cu
I 1000
800
600
400
200
<108 1 07 1 06 1 05 1 04 1 03
Figure 2-9. Example cumulative population risk histogram
for the inhalation from ambient air pathway,
single time period, single receptor.
of risk for one pathway, one receptor type, and one time period at one site. These types of
histogram can be used to assess the variability of risk to the receptors across the AOI. The risk
for this illustration ranges from below 10"8 for about 25 percent of the receptors within the AOI
to 10"4 for about 15 percent of receptors within the AOI. Because the data sets are available for
each pathway and each receptor type, a similar assessment can be made by looking at the
contribution of each pathway to the total risk for a receptor type. When the time series for each
data set is examined, the variability in risk from year to year can be assessed.
Given the resulting time series of pathway-specific and receptor-type/cohort-specific risk
and/or HQ, the years during which the total risk or HQ across all receptors within the AOI is a
maximum is determined. The exposure period with the highest total risk is termed the critical
exposure period.
The distributions for the critical exposure period are the outputs for each site. For any
given site, there is a distribution for each pathway independently, for all ingestion pathways
together, for all inhalation pathways together, for the ground water-only pathways, and, where
appropriate, for all ingestion and inhalation pathways together for each receptor type/cohort, all
cohorts within a receptor type, and all receptor types together at various distances from the
source and across the entire AOI. Additional detail on the Human Risk Module can be found in
Section 14 of this document.
2.3.6 Ecological Receptors, Exposure Pathways, and Risk Measures
Ecological exposure and risk are estimated using two modules: one for exposure and one
for risk. The Ecological Exposure Module focuses on receptors, habitats, and exposure
pathways. The Ecological Risk Module estimates HQs for representative species.
2-28
-------
Section 2.0
Modeling Approach
Ecological Exposure Module. The Ecological Exposure Module calculates the applied
dose (in mg/kg-d) to ecological receptors that are exposed to contaminants via ingestion of
contaminated plants, prey, and media (i.e., soil, sediment, and surface water). However, just as
in the Human Exposure Module, some exposures are expressed as media concentrations and are
passed directly to the Ecological Risk Module to account for the way in which toxicity data are
used. These dose estimates are then used as inputs to the Ecological Risk Module. The
Ecological Exposure Module calculates the applied dose for each receptor placed within a
terrestrial or freshwater aquatic habitat. Exposure is a function of (1) the habitat to which the
receptor is assigned, (2) the spatial boundaries of the species' home range, (3) the food items
(plants and prey) that are available in a particular home range, (4) the dietary preferences for
food items that are available, and (5) the media concentrations in the receptor's home range. In
essence, the module estimates an applied dose for birds, mammals, and selected herpetofauna
that reflects the spatial and temporal characteristics of the exposure (i.e., exposure is tracked
through time and space). The home range is the area within a habitat that is needed for each
receptor.
The conceptual approach for the ecological
exposure assessment within the 3MRA modeling
system accounts for the major sources of variability
in ecological exposures. In particular, the approach
considers variability through (1) the development of
representative habitats; (2) selection of receptors
based on ecological region; (3) the recognition of
opportunistic feeding and foraging behavior using
probabilistic methods; (4) the creation of a dietary
scheme specific to region, habitat, and receptor; and
(5) the application of appropriate graphical tools to
capture spatial variability in exposure. The
underlying framework for the Ecological Exposure
Module is based on a representative habitat scheme
to increase the resolution of general terrestrial and
freshwater systems.
Depending on the type of habitat and
contaminant-specific uptake and accumulation,
animals may be exposed through the ingestion of plants (both aquatic and terrestrial), soil
invertebrates, aquatic invertebrates, fish, terrestrial vertebrates, media, or any combination that is
reflected by the dietary preferences of the particular species. For example, an omnivorous
vertebrate that inhabits a freshwater stream corridor habitat may ingest fish, small terrestrial
vertebrates found in the stream corridor, terrestrial and aquatic plants, surface water, and soil.
The dietary preferences are independent of the contaminant type; therefore, contaminant
concentrations in some food items may be near zero for contaminants that do not bioaccumulate.
The dietary preferences for each receptor are supported by an extensive exposure factors
database containing information on, for example, dietary habits and natural history for more than
50 representative species of interest. The module includes an innovative approach to
characterizing the diet: a probabilistic algorithm that cycles through the database on minimum
and maximum prey preferences to simulate dietary variability.
Representative habitats currently
delineated in the 3MRA modeling
system include
Terrestrial Habitats
• Grasslands
• Shrub/scrub
• Forest
• Crop fields and
pastures
• Residential
Waterbody Margin
Habitats
• Rivers/streams
• Ponds
• Lakes
Wetland Margin Habitats
• Permanently or intermittently flooded
grassland
• Permanently or intermittently flooded
shrub/scrub
• Permanently or intermittently flooded
forest
2-29
-------
Section 2.0
Modeling Approach
The Ecological Exposure Module focuses on
two areas:
¦ Representative ecological receptors,
and
¦ Potential and relevant exposure
pathways.
Representative Ecological Receptors. The
ecological receptors considered in the 3MRA
modeling system include the following:
¦ Terrestrial wildlife using habitats at or
near WMUs; this wildlife can be
directly exposed to waste or indirectly
exposed by consuming plants, soil, or
prey items that bioaccumulate
constituents released from the WMU.
¦ Aquatic plants and other biota; these
may be exposed to contaminants that
are transported from a WMU to nearby
aquatic habitats.
Representative ecological
receptors for the following groups
of taxa were used to populate
habitats:
Terrestrial Habitats Wetland Margin
• Terrestrial plants
Habitats
• Soil biota
• Wetland plants
• Reptiles
• Hydric
• Birds
soil-associated
• Mammals
invertebrates
• Aquatic
Waterbody Margin
invertebrates
Habitats
• Fish
• Aquatic plants
• Amphibians
• Aquatic
• Reptiles
invertebrates
• Birds
• Benthos
• Mammals
• Fish
• Amphibians
• Birds
• Mammals
Vascular plants and other terrestrial biota; these may be exposed to contaminants
that are transported from the WMU to nearby terrestrial habitats (e.g., surficial
soils).
Because all potentially affected species cannot be assessed in a national-level assessment,
potentially affected plants and wildlife were organized into guilds of taxonomically and
functionally related organisms (e.g., herbivorous birds, insectivorous birds, carnivorous
mammals). Receptors were selected to represent each guild based on taxonomic relatedness,
function in the ecosystem, and availability of wildlife exposure factors and toxicity data. For
example, the American robin may be selected to represent insectivorous birds at a WMU because
(1) it is a bird, (2) it eats insects and worms, (3) it is one of many thrushes observed at or near the
WMU, and (4) wildlife exposure factors have been established (U.S. EPA, 1993).
Receptors were assigned to habitats within the AOI based on site-based and regional
data. Representative ecological receptors were assigned to appropriate habitats based on
documented foraging and feeding behavior and habitat usage.
Ecological Exposure Pathways. Table 2-6 lists the media and exposure routes that are
evaluated by the Ecological Exposure Module. Direct contact with contaminants by terrestrial
wildlife was not included in the model because dense undercoats or down effectively prevents
contaminants from reaching the skin of wildlife species and significantly reduces the total
surface area of exposed skin (Peterle, 1991; U.S. ACE, 1996). Also, results of exposure studies
indicate that exposures due to dermal absorption are insignificant compared exposures due to
ingestion for terrestrial wildlife (Peterle, 1991). Similarly, inhalation of volatile organic
chemicals was not included in the model because concentrations of volatile chemicals released
2-30
-------
Section 2.0
Modeling Approach
Table 2-6. Ecological Exposure Routes Evaluated by Receptor Type
l)ii
•eel ( onlacl
Ingestion
Rcccplor
Surface \\ aler
Sedimenl
Soil
Surface \\ aler
Soil
l-'ood
Plants
Aquatic
~
~
Terrestrial
~
Invertebrates
Aquatic
~
~
~
~
Terrestrial
~
~
~
Wildlife
Fish
~
Amphibians
~
Reptiles
~
~
~
Birds
~
~
~
Mammals
~
~
~
from soil to air are drastically reduced, even near the soil surface (U.S. ACE, 1996), and
inhalation toxicity data for laboratory rats and mice show that volatile organic chemical
concentrations in soils would have to be great to induce noncarcinogenic effects in wildlife
(U.S. ACE, 1996). In addition, availability of inhalation benchmarks for ecological receptors is
limited. Additional detail on the Ecological Exposure Module can be found in Sectionl5of this
document.
Ecological Risk Module. The Ecological Risk Module calculates HQs for a suite of
ecological receptors assigned to habitats delineated at a site. These receptors fall into eight
receptor groups: mammals, birds, herpetofauna, terrestrial plants, soil community, aquatic plants
and algae, aquatic community, and benthic community. The spatial resolution of the Ecological
Risk Module is largely determined by both the home ranges and the habitats delineated at each
site. The habitats are intended to represent habitats across the United States that may be found at
or near WMUs and that support wildlife receptors.
The habitat area is important in assessing risks to several receptor groups (e.g., benthic
community); exposures and associated risks are considered across the entire habitat rather than
for one or more home ranges. For example, contaminant concentrations to which the aquatic
community is exposed are represented by a habitat-wide average that may include multiple
stream reaches. The temporal resolution is based on annual average applied doses (for
comparison with ecological benchmarks [EBs]) and media concentrations (for comparison with
chemical stressor concentration limits [CSCLs]).
The 3MRA modeling system calculates HQs for all receptors assigned to the study site.
The HQs are used in developing cumulative distributions of HQs. Each of the HQs calculated by
the Ecological Risk Module has a series of attributes associated with it that allows ecological
risks to be interpreted in a number of ways. For instance, distance from the source (i.e., 1 km, 1
to 2 km, or across the entire site) is important in understanding the spatial character of potential
2-31
-------
Section 2.0
Modeling Approach
ecological risks. Other attributes considered relevant to ecological risks and regulatory decision
making include the following:
¦ Habitat type (e.g., grassland, pond, permanently flooded forest),
¦ Habitat group (i.e., terrestrial, aquatic, and wetland),
¦ Receptor group (e.g., mammals, amphibians, soil community), and
¦ Trophic level (i.e., producers, TL1, TL2, TL3 top predators).
The maximum HQ across the site is also reported along with its ecological risk attributes.
In calculating receptor-specific HQs, the Ecological Risk Module does all of the
necessary accounting to develop distributions based on the specific receptor and habitat
groupings of interest. The Ecological Risk Module calculates HQs based on the EB or CSCL
and the chemical exposure information, and provides summaries of ecological risk information
from the simulation to determine when critical years with maximum HQs are experienced.
Outputs are provided for all attributes associated with each receptor. Additional detail on the
Ecological Risk Module can be found in Section 16 of this document.
2.4 References
Aronson, D., M. Citra, K. Schuler, H. Printup, and P.H. Howard. 1999. Aerobic Biodegradation
of Organic Chemicals in Environmental Maedia: A Summary of Field and Laboratory
Studies. SRC TR 99-002. Syracuse Research Corporation. Syracuse, NY.
Burns, L.A. 1997. Exposure Analysis Modeling System (EXAMS II): User's Guide for Version
2.95.5. EPA/600/R-047. U.S. Environmental Protection Agency, Athens, GA.
Burns, L.A., D.M. Cline, and R.R. Lassiter. 1982. Exposure Analysis Modeling System
(Exams): User Manual and System Documentation. EPA/600-3-82/0231. U.S.
Environmental Protection Agency, Environmental Research Laboratory, Athens, GA.
Peterle, T.J. 1991. Wild Toxicology. Van Nostrand Reinhold, New York, NY.
Sample, B.E., J.J. Beauchamp, R.A. Efroymson, G.W. Suter, II, and T.L. Ashwood. 1998a.
Development and Validation of Bioaccumulation Models for Earthworms. ES/ER/TM-
220. Lockheed Martin Energy Systems Inc., Oak Ridge, TN. February.
Sample, B.E., J.J. Beauchamp, R.A. Efroymson, and G.W. Suter, II. 1998b. Development and
Validation of Bioaccumulation Models for Small Mammals. Prepared for the U. S.
Department of Energy under contract DE-AC05-840R21400.
U.S. ACE (Army Corp of Engineers). 1996. Environmental Quality Risk Assessment
Handbook. Volume II: Environmental Evaluation. EM 200-1-4. Department of the
Army, Washington, DC. June 30.
2-32
-------
Section 2.0
Modeling Approach
U. S. EPA (Environmental Protection Agency). 1993. Wildlife Exposure Factors Handbook.
Volumes I and II. EPA/600/R-93/187. Office of Health and Environmental Assessment
and Office of Research and Development. Washington, DC. December.
U.S. EPA (Environmental Protection Agency). 1996a. An SAB Report: Review of a
Methodology for Establishing Human Health and Ecologically Based Exit Criteria for
the Hazardous Waste Identification Rule (HWIR). EPA-SAB-EC-96-002. Washington,
DC: Science Advisory Board. May.
U.S. EPA (Environmental Protection Agency). 1996b. EPACMTP Sensitivity Analysis.
Washington, DC: Office of Solid Waste. March.
U.S. EPA (Environmental Protection Agency). 1996c. EPA 's Composite Model for Leachate
Migration with Transformation Products (EPACMTP): Background Document.
Washington, DC: Office of Solid Waste. September.
U.S. EPA (Environmental Protection Agency). 1996d. Environmental Fate Constants for
Organic Chemicals under Consideration for EPA 's Hazardous Waste Identification Rule.
Office of Research and Development. Athens, GA.
U.S. EPA (Environmental Protection Agency). 1997. Exposure Factors Handbook..
EPA/600/P-95/002F a, b, c. Office of Research and Development. August.
U.S. EPA (Environmental Protection Agency). 1998. MINTEQA2/PRODEFA2, A Geochemical
Assessment Model for Environmental Systems: User Manual Supplement for Version 4.0.
Prepared for the U.S. Environmental Protection Agency by HydroGeoLogic, Inc.,
Contract No. 68-C6-0020.
U.S. EPA (Environmental Protection Agency). 1999a. Data Collection for the Hazardous
Waste Identification Rule. Section 11. Aquatic Food Web Data. Office of Solid Waste,
Washington, DC. October.
U.S. EPA (Environmental Protection Agency). 1999b. Sensitivity ofISCST3 model estimates to
distance from source. Memorandum to Stephen Kroner, October 27, 1999, at:
www.epa.gov/epaoswer/hazwaste/id/hwirwste/risk.htm.
U.S. EPA (Environmental Protection Agency). 1999c. Changes in theMINTEQA2Modeling
Procedure for Estimating Metal Partition Coefficients in Ground Water. Prepared by
HydroGeoLogic, Inc. for the U.S. Environmental Protection Agency, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999d. Anaerobic Biode gradation of Organic
Chemicals in Groundwater: Summary of Field and Laboratory Studies. Office of Solid
Waste. Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999e. Surface Water, Soil, and Waste Partition
Coefficients for Metals. Office of Solid Waste. Washington, DC.
2-33
-------
Section 2.0
Modeling Approach
U.S. EPA (Environmental Protection Agency). 1999f. Chemical Database for HWIR99. Office
of Solid Waste. Athens, GA.
U.S. EPA (Environmental Protection Agency). 2000. Methodology for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions. Update to
EPA/600/6-90/003 Methodology for Assessing Health Risks Associated with Indirect
Exposure to Combustion Emissions. National Center for Environmental Assessment,
Cincinnati, OH.
U.S. EPA (Environmental Protection Agency). 2003. Prediction of Chemical Reactivity
Parameters and Physical Properties of Organic Compounds from Molecular Structure
Using SPARC. Internal Report. Ecosystems Research Division, National Exposure
Research Laboratory, Athens, GA. March.
Westat, Inc. 1987. Screening Survey of Industrial Subtle D Establishments. Draft Final Report.
U.S. Environmental Protection Agency, Westat, Inc., Rockville, MD. December 29.
2-34
-------
Section 3.0
Spatial Layout for Environmental Settings
3.0 Spatial Aspects of Environmental Settings
and Receptors
The 3MRA modeling system uses site-based modeling to provide the spatial distribution
of contaminants and receptors in a specified area of interest (AOI) around each of the sources
being modeled. To support this site-based strategy, a geographic information system (GIS) is
used to provide a spatial frame of reference and associated attributes around each site or facility
within the AOI. This section describes the spatial layout and the methods and data used to
characterize the various spatial attributes of an AOI in the 3MRA modeling system. The spatial
aspects of an AOI and the associated data collection process are discussed in greater detail in
Volume II of this report, Data Collection for the 3MRA Modeling System (U.S. EPA, 2003).
3.1 Overview of Spatial Layout for Environmental Settings and Receptors
The spatial layout of an AOI used in the 3MRA modeling system consists of the source
of constituent releases, the surrounding environment, and the location of receptors. The sources,
such as a landfill, surface impoundment, industrial facility, or abandoned site, are stationary, and
the AOI is defined by a distance from the source. This distance could be a few hundred meters
up to 50 km. The 50 km limit is imposed by the limits of the air dispersion model (ISCST3)
used in the 3MRA modeling system. The distance defining the AOI is wholly dependent on the
areal extent needed for a specific analysis. For the sites and environmental settings currently in
the 3MRA representative national data set, the AOI distance is set at 2 km from the boundary of
the source.
3.1.1 Settings and Areas of Interests
The 3MRA modeling system incorporates a number of attributes that characterize the
environmental setting of the AOI. A setting is defined as the source or sources of pollutant
releases being modeled plus the characteristics of the environment in the AOI. The 3MRA
modeling system uses more than 700 total variables to describe a site's environmental setting.
The values for many of these variables are based on site-specific information. Others are based
on regional or national data sets, when site-specific data are not available. A GIS was used to
overlay layers of spatial data to construct a complete description of the environmental setting
within the AOI.
3.1.2 Site Layout and Spatial Data Layers
Several key spatial data layers for each setting are shown in Figure 3-1. The layers are
defined as follows:
3-1
-------
Section 3.0
Spatial Layout for Environmental Settings
Human Receptors
(census and land use data)
Ecological Habitats and Receptors
(land use, wetlands, T&E species,
etc., data)
Watersheds
(DEMS)
Waterbodies
Streams, Lakes, and Wetlands
(DEMs Reach Files,GIRAS, NWI)
Base Grid, Area of Interest
(AOI)
(specified: 100m x 100m grid
cells; 2 km radius AOI)
x = facility centroid
~ = waste management unit
Figure 3-1. Site-based spatial overlays for 3MRA modeling system spatial framework
(GIS analysis).
3-2
-------
Section 3.0
Spatial Layout for Environmental Settings
Human receptor points are defined by U.S. Census block centroids for residents,
and by randomly placed farms based on Census block group, agricultural census,
and land use data for farmers (U.S. EPA, 2000b). These locations are used to
calculate exposure concentrations in various media and to estimate risks to human
receptors. A detailed explanation of how these data are used is provided in
Volume II (U.S. EPA, 2003).
Ecological habitats and receptors are defined by land use and other ecologically
relevant data (U.S. EPA, 2002). These locations are used to calculate exposure
concentrations in various media and to estimate risks to ecological receptors. A
detailed explanation of how these data are used is provided in Volume II
(U.S. EPA, 2000a).
Watersheds and watershed subbasins are delineated either electronically, using
digital elevation models (DEMs) of topography, or manually, based on Reach File
3 (RF3) stream networks (U.S. EPA, 2002). Watershed subbasins provide the
spatial context and connectivity necessary to model chemical deposition, erosion,
overland transport, and resultant soil concentrations in the Waste Pile, Land
Application Unit (LAU), and Watershed Modules. Typically, a watershed is the
entire drainage area for a particular stream network in the AOI, and is subdivided
into subbasins. A watershed subbasin can vary in size from a portion of a hillside
to much larger areas encompassing regional stream or river networks. In all
cases, a subbasin is treated as a single, homogeneous area with respect to soil
characteristics, runoff and erosion characteristics, and constituent concentrations
in soil.
Waterbodies (lakes, streams, and wetlands) are defined by DEMs, RF3 data, land
use data contained in the Geographic Information Retrieval and Analysis System
(GIRAS), and/or National Wetlands Inventory (NWI) (see U.S. EPA, 2002).
Waterbodies provide the spatial context and connectivity necessary to model
contaminant deposition, fate, transport, and the resulting water column and
sediment concentrations in streams, lakes, and wetlands.
Subsurface features include the surficial aquifer (the unconfined ground water
source nearest the surface) and the vadose zone (the soil zone between the ground
surface and the surficial aquifer). The vadose zone and surficial aquifer are
assumed to directly underlie the source. The State Soil Geographic Data Base
(STATSGO) provides vadose zone characteristics. Aquifer properties are
obtained from regional data sets. Ground water flow is assumed to follow the
topography of the land surface, as determined from the DEMs.
3-3
-------
Section 3.0
Spatial Layout for Environmental Settings
In the GIS, each of these spatial data layers is composed of 2-D polygons, except for
streams, which are defined by 1-D vectors (stream reaches), and human receptor locations
(residences and drinking water wells), which are defined as points. Because these polygon
coverages could not be exported directly to the 3MRA modeling system, each spatial data layer
was defined in terms of a base grid composed of 100 >< 100 m cells, which roughly correspond to
the minimum resolution of several site-specific data types (i.e., land use, topographic, and soil).
This base grid (or xj; coordinate system) serves as the basis for defining receptor points at which
the Air, Vadose Zone, and Aquifer Modules in the 3 MR A modeling system produce constituent
concentrations (and deposition rates for the Air Module) in terms of distance and direction from
the constituent source. To provide the 3MRA modeling system with the data necessary to
specify air points and well locations, spatial data are passed to the model using this site
coordinate system for the following data layers:
¦ Watersheds,
¦ Waterbodies,
¦ Farms,
¦ Human receptor points,
¦ Wells (human receptor points with drinking water wells), and
¦ Ecological habitats.
The site coordinate system is described using metric xy coordinates relative to the
ground surface at the facility or site centroid (the georeference point). The 3MRA modeling
system requires that the georeference point be specified using a latitude and longitude in the
Universal Transverse Mercator (UTM) coordinate system. Using GIS, a coverage of spatial data
(e.g., watersheds) is overlaid on the standard grid. If the centroid of a grid cell intersects a
polygon, it receives the identifier for that polygon (see Figure 3-2).
Watershed
polygon >
Figure 3-2. Example of transfer of polygons to 100 x 100 m template grid.
3-4
-------
Section 3.0
Spatial Layout for Environmental Settings
3.2 Watersheds and Waterbodies
To develop the watershed/waterbody layout, GIS programs were used to
¦ Compile available hydrologic, DEM land use, and wetlands data sets (see
Figure 3-3);
¦ Extract site-specific data from these data sets; and
¦ Delineate the watershed subbasins, waterbodies, and local watersheds.
The watershed layout
includes the watersheds that
contribute flow to waterbodies
within the AOI. Although the
waterbody/watershed layout is
important for several modules
in the 3MRA modeling system,
the layout primarily supports
the surface water, regional
watershed, and local watershed
components. These
components serve the
following purposes:
GIRAS
Land Use
NWI Wetlands
RF3 Reaches
Streams
The Surface
Water Module
estimates water
column and
sediment
chemical
concentrations
for use in the Aquatic Food Web and Ecological Exposure Modules.
Figure 3-3. GIS data coverages for waterbody and
watershed delineation.
¦ The Watershed Module estimates surficial (top 1 cm) and depth-averaged soil
concentrations within the AOI for the Terrestrial Food Web, Farm Food Chain,
and Human and Ecological Exposure Modules; generates hydrologic inputs
(runoff, baseflow, soil loads) for the Surface Water Module; and generates annual
average infiltration estimates for use as recharge for the Aquifer Module.
¦ The Waste Pile and LAU Modules estimate runoff, soil loads, and contaminant
loadings for upgradient and downgradient subbasins. The modules simulate
contaminant movement due to water erosion and overland transport from the
LAU or waste pile to the nearest downslope waterbody.
Within the context of the 3 MR A modeling system, certain assumptions and definitions
were required to develop the procedures for delineating the watersheds, subbasins, and
3-5
-------
Section 3.0
Spatial Layout for Environmental Settings
waterbodies to be modeled. Figure 3-4 illustrates the conceptual framework for the waterbody
and watershed layout, which is detailed below.
.*
WBN 1
WBN 2 *
SB 7
SB 1
SB 6
SB 10
SB 4
SB 9 n
wmu
SB 3
SB 8
SB 2
SB 5
SB 12
-7
SB 11
SB 11
WBN 3
• •
SB 13
Area of Interest
Watershed Subbasins (SB)
Modeled Waterbody Reaches
Unmodeled Waterbody Reaches
Reach Order
Waterbody Network
1,2 etc.
WBN
Figure 3-4. Regional watershed subbasin delineation.
3-6
-------
Section 3.0
Spatial Layout for Environmental Settings
3.2.1 Definitions for Watershed and Waterbody Layout
The 3MRA modeling system uses the following definitions for watersheds and
waterbodies:
¦ A waterbody reach is a stream segment between tributaries, a lake (or pond),1 a
wetland, or an estuary. A waterbody reach is the basic modeling unit for the
Surface Water Module.
¦ A headwater reach (or first-order reach) is a reach that flows into another reach
downstream but has no upstream reaches draining into it.
¦ An exiting reach is a nonheadwater reach that flows out of the AOI.
¦ A waterbody network is a series of connected (possibly branching) reaches that
defines a stream network (three waterbody networks are shown in Figure 3-4,
labeled WBN 1, WBN2, and WBN3).
¦ A regional watershed is the entire drainage basin associated with an individual
modeled waterbody network and can extend outside of the AOI, as necessary.
Upstream watersheds are not delineated for any waterbody reaches greater than
order 52 entering the AOI because the volume of water in this size stream or river
would dilute the concentration of a constituent from a source. However, tributary
lands for reaches greater than order 5 lying within the AOI are delineated as
regional watershed subbasin(s) (see definition below) to simulate soil
concentrations. Such subbasin delineations only include land within the AOI.
The upstream boundary of any regional watershed is its natural, upstream
boundary (i.e., its headwater basin, except, as noted above, for those greater than
order 5).
¦ A watershed subbasin is a portion of the regional watershed. A regional
watershed is composed of a set of nonoverlapping subbasins. The set of all
subbasins for a given regional watershed will completely cover the regional
watershed, except for watersheds greater than order 5. For example, in Figure 3-
4, Subbasins 2 through 10 make up the regional watershed for Waterbody
Network 2. Depending on the waterbody layout, one or more subbasins can drain
into a reach, or a subbasin can drain into one or more reaches. The subbasin is
the basic spatial modeling unit for the Watershed Module.
¦ A local watershed is a drainage area that contains a source, such as an LAU. A
local watershed extends from the upslope drainage divide downslope to the first
1 All lakes and ponds were designated as lakes in the waterbody layout data.
2 As defined by the Strahler ordering system (Strahler, 1957).
3-7
-------
Section 3.0
Spatial Layout for Environmental Settings
defined drainage channel, lake, or pond. It is divided into subareas upslope and
downslope of the source and the source itself. These subareas are the basic
modeling units for the local watershed model.
3.2.2 Assumptions for Waterbodies and Regional Watersheds
The following assumptions were used to develop and collect data associated with the
regional watershed and waterbody layout:
¦ The Surface Water Module models streams of order 1 through 5, ponds, lakes,
and wetlands. All lakes, ponds, and wetlands connected to a waterbody network
are modeled regardless of order. Only reaches lying completely or partially within
the AOI are modeled.
¦ Streams of order 1 through 5, ponds, lakes, and wetlands within the AOI are
modeled until they exit the AOI. Headwater reaches (order 1) that exit the AOI
are not modeled.
¦ Order 3 and higher stream reaches, along with lakes, ponds, and certain wetlands,
are assumed to be fishable (i.e., support fish populations suitable for recreational
anglers). Order 1 and 2 stream reaches within the AOI that feed into an order 3
reach within the AOI contribute to the contaminant concentration in the order 3
stream reach.
¦ Surface waters in the waterbody network are subject to contaminant loads from
indirect runoff (i.e., from aerial deposition of a contaminant on the watershed).
Some are subject to both indirect and direct loads (i.e., direct runoff/erosion from
the source).
3.2.3 Assumptions for Local Watersheds
Some sources, such as waste piles and LAUs, release pollutants directly through runoff
and erosion. Assumptions associated with the local watershed component of these sources
include the following:
If the source overlaps a drainage
divide, multiple local watersheds
are modeled.
The local watershed will contain
at least two subareas—the source
(or portion thereof) and a 30.5 m
wide (100 ft wide) buffer between
the source and the stream (see
Figure 3-5). The local watershed
may also include a third, upslope
subarea from the drainage divide
to the source.
/
~
~
~
~
~
~
Upstream
x Subarea
It
i \
' *
¦ V
%
>
"N
V
%
%
WMU
%
s
%
t Source
Subarea ±
Buffer y
Subarea \
%
__
+ m
+ #
+ a
\
*
+ f
~ f
~ I
> ¦
Figure 3-5. Local watershed delineation.
3-8
-------
Section 3.0
Spatial Layout for Environmental Settings
3.2.4 Watershed Soils
The 3MRA modeling system estimates surface soil concentrations of a contaminant in
each of the watershed subbasins within the AOI. It also provides estimates of the vadose zone
soil concentrations, but only under the source, not in the whole watershed. The depth of the
surface soil is defined by the user in the 3MRA modeling system. Vadose zone soil extends from
the ground surface to the water table. Depending on the location of the AOI, average (area- and
depth-weighted) or predominant soil properties are derived for the soil depth zone of interest
across each watershed subbasin or the waste management unit (WMU). All soil parameters are
either site-specific or derived from site-specific data using national relationships and thus are
considered site-based. Parameters derived from site-specific data include soil hydrologic
properties derived from site-specific soil texture or hydrologic class, and other properties derived
from a combination of soil texture or class and site-specific land use.
3.3 Human Receptors, Farms, and Wells
Human receptor locations, which include residences and farms, are one of the primary
spatial data layers in the 3MRA modeling system. They enable human risk or hazard to be
calculated spatially within the AOI where people are likely to be located. GIS data are used to
locate these points and collect human receptor numbers and characteristics (e.g., receptor types,
age cohorts). Using this information, the 3MRA modeling system can generate risk distributions
around a source that are based on the population within the AOI. These data provide a level of
specificity for a national-level assessment that accounts for how many people are affected and at
what levels of risk.
As discussed in Section 2, human receptors are subdivided into groups that reflect
differences in exposure patterns due to either location or behavior. Resident human receptor
points are located and populated by 1990 Census block data (U.S. Bureau of the Census, 1987,
1992; U.S. EPA, 1995a,b). Farms are located and populated using Census block group
boundaries, subdivided by farm land use, along with county-level agricultural census data.
Figure 3-6 illustrates many of the spatial data elements related to human receptors for a
sample AOI. This AOI is relatively simple in terms of receptor population density. It includes
20 residential locations, one for each (populated) Census block, and one farm. Contaminant
concentrations are estimated at the single point at the centroid of each Census block to evaluate
the exposure and risk or hazard to residential receptors. For example, the soil concentrations are
an areal average of the watershed subbasin soil concentrations within the Census block. The
ambient air concentration is the average of air concentrations at points modeled within a Census
block. The ground water concentration is evaluated at a well at the Census block centroid when
some or all of the receptors in the Census block use private drinking water wells. The estimated
risk or hazard quotient (HQ) for each receptor type and age cohort at a Census block centroid is
assigned to all individuals in that Census block that are the same receptor type and age cohort.
The farm is randomly located, but its location is constrained to be both within a Census block
group that includes farming activity and also within an agricultural land use area within that
block group. For this example, a single representative farm is placed within one block group to
3-9
-------
Section 3.0
Spatial Layout for Environmental Settings
Source
Farm
Census
" block
centroid
Figure 3-6. Example site layout for human receptors.
represent the exposures for all farms within that block group. The farm population by receptor
type and age cohort in the block group is assigned the risk or HQ associated with the
representative farm.
The following subsections summarize the steps used to place and process human receptor
locations, farms, and wells. Details of these procedures are presented in U.S. EPA (2002). The
total number of people within an AOI is divided into receptor types and age cohorts. These are
further subdivided by whether or not households have a private drinking water well. Measures
are taken with respect to numbers of people at a site in order to maintain the correct total
population across the site.
3.3.1 Resident and Home Gardener Locations
Census block coverages are used to locate residents and home gardeners. A resident or
home gardner receptor is placed at the centroid of every Census block within the AOI. The
characteristics within each Census block coverage are linked to the other spatial data (regional
watersheds, local watersheds, and distance rings) within the system. Area weightings are used to
calculate the number of receptors/cohorts and the environmental characteristics within the AOI.
For example, if Census blocks extend across distance rings, the block data are fractioned into
each ring/block combination based on the relative areas in each.
3-10
-------
Section 3.0
Spatial Layout for Environmental Settings
3.3.2 Farm Locations
Farms within the AOI are located by overlaying the Census block groups with the land
use coverages that identify areas with farm-type land use (i.e., crop and pastureland) in block
groups that have farmers. A farm is placed in each of these block group/farm land use areas as
follows:
¦ The median farm size for the county is determined based on a census of
agriculture data (U.S. Bureau of the Census, 1987, 1992). These data include the
median farm size and fractions of farms that are beef or dairy farms by county.
The fraction of beef or dairy farms is calculated as the number of beef or dairy
farms divided by the total number of farms in a county.
¦ If the block group/farm land use area is larger than the median farm size, then a
random point inside the farm land use area of the block group is picked. A circle
is created to represent the area of the farm. To keep the farm within the block
group/farm land use area, the farm is clipped at the farm land use boundary. The
radius of the circle is increased incrementally until a polygon is produced that
approximately matches the median farm area and is completely contained within
the farm land use area of the block group. If the farm land use area within a block
group is smaller than the median farm size, then the entire area of farm land use in
the block group is used as the farm.
¦ Once the farms are placed within the AOI, the fractions of beef farms and dairy
farms are applied to the total number of farms to calculate the number of beef and
dairy farmers in each block group with a farm.
3.3.3 Well Locations (Private Wells Only)
Census block group data include the number of households with private wells. These
data are used to calculate the fraction of people with private water supplies within the AOI.
This block group fraction is applied to block-level human receptor points within each block
group to identify points with wells. Private wells are located at the centroid of each Census
block that contains people on private wells. Wells are also located at the centroid of each farm.
Every farm is assumed to be on a private well. Not all private wells are necessarily affected by
the ground water contamination from the source, only those that are predicted to be within the
plume of contamination given the characteristics at each site.
3.3.4 Recreational Fisher Locations
All receptor types may also be recreational fishers. The location of fishers is the same as
for residents, gardeners, and farmers. For example, a recreational fisher may be a resident
located at the centroid of a Census block. However, the fisher may catch fish in up to three
streams within the AOI. If there are more than three fishable streams in the AOI, the location of
the three streams in which a particular receptor fishes is randomly selected.
3-11
-------
Section 3.0
Spatial Layout for Environmental Settings
The number of recreational fishers in each category of receptor types is based on 1996
National Survey of Fishing, Hunting, Wildlife, and Recreation (NSFHWAR) data, which include
the percentage of residences with recreational fishing licenses in urban, rural, and rural farm
areas by state (U.S. DOI and U.S. DOC, 1997). For each state, recreational fisher percentages
can be estimated using the NSFHWAR urban and rural population breakdowns and splitting the
rural population into farm and nonfarm populations using farm/nonfarm population fractions
calculated from 1990 U.S. Census data (U.S. Bureau of the Census, 1987, 1992). This provides
state-by-state estimates of urban, rural-farm, and rural-nonfarm recreational fishers as a
percentage of the total state population. These percentages are used to calculate the recreational
fisher population within an AOI.
3.4 Habitats and Ecological Receptor Placement
Representative habitats were developed that can be used in the 3MRA modeling system
to describe the biological conditions within an AOI. Receptors were assigned to habitats based
on wildlife-habitat relationships documented in the literature. The habitat was chosen as the
appropriate level of resolution for the spatial element of the ecological risk assessment used for a
national-level assessment. In this context, the term habitat implies a level of detail and
specificity that is meaningful for estimating exposure but that does not require extensive
biological inventory or field investigation for identification or delineation.
Fourteen representative terrestrial, wetland, and margin habitats have been developed for
use in the 3MRA modeling system. Table 3-1 presents an overview of the representative
habitats. The representative habitats address areas inhabited by land-based receptors. In
addition to terrestrial mammals, birds, and herpetofauna, terrestrial receptors also include some
species that spend significant time in the water, such as the bullfrog and snapping turtle, and
some that derive all or most of their food from the water, such as the osprey and muskrat. In
order to fully assess exposure, the margin habitats are defined and delineated to include both the
waterbody and its adjacent terrestrial areas, such as stream corridors and pond margins, in which
all these receptors are potentially exposed. Section 15 of this document describes how the
representative habitats were selected, characterized, and delineated.
3.4.1 Habitat Delineation
The GIS data layers needed for habitat delineation are
¦ Facility/site location,
¦ AOI,
¦ GIRAS land use,
¦ Delineated waterbodies—rivers, lakes, and wetlands,
¦ NWI wetlands,
¦ RF3 waterbodies,
¦ Managed areas,
¦ 2 m DEM contours, and
¦ Preprocessed habitat grid.
3-12
-------
Section 3.0
Spatial Layout for Environmental Settings
Table 3-1. Ecological Risk Assessment Representative
Habitats for Terrestrial Receptors
Torres! rfcil Ihihiliils
Grasslands
Shrub/scrub
Forests
Crop fields and pastures
Residential
Wsilerhodv M:ir«in Ihihilnls
Rivers/streams
Lakes
Ponds
Wclhind M:ir»in lliihiliils
Intermittently flooded grasslands
Intermittently flooded shrub/scrub
Intermittently flooded forests
Permanently flooded grasslands
Permanently flooded shrub/scrub
Permanently flooded forests
Habitats are delineated by assigning each grid cell to one of the 14 representative
habitats. A grid cell may also be designated as "no habitat" if it consists of highly developed or
impervious areas (such as parking lots or train tracks) not likely to support wildlife. Figure 3-7
shows an example AOI with preprocessed habitats.
Terrestrial Habitat Delineation. To delineate the terrestrial habitats at each site, the
representative habitat types were correlated with Anderson land use categories. The Anderson
Land Use Classification, developed by the U.S. Geological Survey (USGS), assigns land use
descriptors to areas that are distinguishable in satellite and other remotely sensed data. Digitized
Anderson land use data are readily available through GIRAS and, therefore, provide a useful tool
for locating habitats. GIRAS land use/land cover data delineate land use patterns wit respect to
vegetation, human activity, and waterbodies. Although these data are 15 to 25 years old and
therefore do not reflect current conditions in some locations, the GIRAS data set is the most
complete and current national data set available. A full description of GIRAS data and methods
for use in the 3MRA modeling system is presented in Volume II (U.S. EPA, 2003) of this report.
Delineation of Waterbody Margin Habitats. Some terrestrial receptor species depend on
aquatic systems for some or all of their food. These exposure scenarios are assessed in the
waterbody margin habitats. Stream corridor, pond, and lake margin habitats were delineated as
being adjacent to the waterbodies generated as part of the waterbody and watershed layout data
processing. Data sources for delineation of streams include DEM data (USGS, 1999), RF3 data
3-13
-------
Section 3.0
Spatial Layout for Environmental Settings
Gather
Land Use Code and Habitat
11 Residential
12 Commercial and services
(nonhabitat)
13 Industrial (nonhabitat)
14 Transportation,
communications, and
utilities (included in
largest adjacent habitat)
17 Other urban or builtup
land (residential)
41 Deciduous forest land
(forest)
21 Cropland and pastures
(crop fields/pastures)
75 Strip mines, quarries,
gravel pits (nonhabitat)
76 Transitional areas
(nonhabitat)
Figure 3-7. Preprocessed habitat codes.
(U.S. EPA, 1994b), and GIRAS data (U.S. EPA, 1994a). Data sources for delineation of ponds
and lakes include GIRAS and the NWI (U.S. FWS, 1998). Once the streams, ponds, and lakes
were delineated for an AOI, their respective corridor and margin habitats were added.
There is no simple correlation between waterbody characteristics and wildlife distribution
within corridors and margins. The vegetation, topography, and land use of the adjacent land, as
well as the size, depth, flow, and aquatic food web of the waterbody, are but a few of the more
prominent variables that affect wildlife use of waterbody margin habitats. Elevation contours
were assumed to be the best indicators of stream corridors and pond and lake margins. Using
DEM contour data, an attempt was made to determine a visual natural limit for the corridor or
margin. Because waterbodies occur in the landscape along elevation contours, natural
boundari es were frequently evident. If no contour-based boundaries were apparent, surrounding
land use was used instead. For example, if GIRAS data indicated a forest buffer running parallel
to a stream and a commercial or industrial area adjacent to the forest, the stream corridor would
consist of the forest buffer. When neither contours nor land use indicated corridor or margin
boundaries, a default minimum margin was delineated. Waterbody margin habitats at a single
site were combined, or bridged, as were terrestrial habitat patches.
Delineation of Wetland Margin Habitats. National Wetlands Inventory (NWI) data
provide the most complete and readily available digitized data on location and type of wetlands
on a national scale (U.S. FWS, 1998). NWI data, however, have not yet been digitized for the
3-14
-------
Section 3.0
Spatial Layout for Environmental Settings
entire United States. When digitized NWI data were not available for an AOI, wetland data from
GIRAS were used.
Wetland identification and classification in the NWI are based on photo interpretation of
remote imagery at a 1:24,000 scale. Each wetland polygon identified is classified and coded
according to the Cowardin et al. (1979) classification system, which indicates substrate type,
dominant vegetation, and hydrologic regime, as well as additional details regarding soil and
water chemistry, where relevant. The GIRAS wetland data are on a 1:80,000 or smaller scale
and distinguish only between wetlands dominated by woody vegetation and those not dominated
by woody vegetation (Anderson et al., 1976). GIRAS data do not show wetlands smaller than
16 hectares (39.5 acres). Consequently, there is significant variation in the level of detail in the
wetland data for sites with and without NWI coverage. Because the NWI provides high-quality
data that readily distinguish the different wetland habitat types, it is the preferred source for
wetland data.
Wetland Flood Regime. The 3MRA representative habitats include three intermittently
flooded and three permanently flooded wetland types. The primary criterion for this division is
the wetlands' ability to support fish populations. When NWI data were available, this distinction
was made based on water regime modifiers in the NWI code. Based on the descriptions in
Cowardin et al. (1979), a decision was made about whether each water regime would be
expected to support an aquatic food web that includes TL3 or TL4 fish. Wetland types with
flood regimes that indicate the presence of sufficient flooding to support fish populations are
delineated as permanently flooded wetlands. Those wetlands in which flooding is infrequent or
of short duration are delineated as intermittently flooded wetlands.
All GIRAS-identified wetlands are delineated as permanently flooded because GIRAS
generally does not recognize wetland ecosystems at the drier end of the wetland flood regime
continuum.
Treatment of Intermittently Flooded Wetlands. Wetlands designated as intermittently
flooded were then further classified according to the representative wetland habitats as
intermittently flooded grassland, shrub/scrub, or forest wetlands. These determinations also
were made based on NWI codes.
For the 3MRA modeling system, the three intermittently flooded wetland habitat types
were identified and designated, but their boundaries were not delineated. Many intermittent
wetlands occur as small, scattered habitat patches and, even when combined with similar patches
in an AOI, often do not constitute large enough areas for placement of home ranges. However,
these areas are known to provide important, albeit highly fragmented, habitat. Therefore, during
site layout data processing, intermittent wetlands were identified as inclusions within
surrounding upland habitats (e.g., grassland, forest), and their unique receptor species are
included in the data passed to the Ecological Exposure Module. In this manner, the receptors
expected to occur in intermittent wetlands are included in the exposure assessment.
An exception was made to this approach for certain intermittent wetlands adjacent to
streams of order 3 and higher. Water regimes for seasonally flooded and seasonally
flooded/saturated imply seasonal flooding for extended time periods; in wetlands adjacent to
3-15
-------
Section 3.0
Spatial Layout for Environmental Settings
fishable stream reaches (stream order 3 and higher), these water regimes could potentially
support an aquatic food web. Therefore, these intermittent wetlands, whether grassland-,
shrub/scrub-, or forest-dominated, were delineated when they occurred adjacent to streams of
order 3 or higher. These habitats were treated in a manner identical to the permanently flooded
wetlands, as described in the following section.
Delineation of Permanently Flooded Wetlands. Permanently flooded wetlands were
identified at each site based on the NWI codes. When NWI data were not available, GIRAS data
were used. GIRAS data classify wetlands as forested or nonforested and do not include any
information on the flood regime. Because most national data sets generally apply the term
wetlands to tidal and other aquatic habitats and do not recognize noninundated areas as wetlands,
it is assumed that wetlands identified in GIRAS data fall within the permanently flooded wetland
habitats. The forested GIRAS wetlands are delineated as permanently flooded forested
wetlands; the nonforested GIRAS wetlands are delineated as permanently flooded grasslands.
Although some of the wetlands included in the GIRAS data are undoubtedly dominated by
shrub/scrub vegetation, the data do not allow this distinction to be made. In the absence of better
data, the intermittently flooded grassland habitat is considered the most appropriate alternative.
Permanently flooded wetlands frequently occur in association with streams, rivers, lakes,
and ponds. Thus, the potential arises for areas adjacent to waterbodies to include both wetlands
and waterbody margin habitats. The most effective and straightforward approach to handling
this situation appeared to be to default to the wetland habitat when wetlands and waterbody
margin habitats overlapped. In fact, many wildlife receptors probably forage across both
waterbody margin and wetland habitats, whereas other species show a preference for or tend to
avoid the wetland habitat. Thus, wetland habitats were delineated whenever they were indicated,
including within a waterbody margin. The wetlands occurring near waterbodies were not
subsumed in the stream corridor or lake and pond margin habitat.
3.4.2 Assignment of Receptors to Habitats
Ecological risk is assessed for the wildlife receptors expected to be present within an
AOI. A receptor is placed within the AOI if the AOI is located in the receptor's geographic
range and includes suitable habitat to support the receptor species.
The 3MRA modeling system was designed to support national analyses; therefore, the
list of receptor species was developed to represent ecological regions throughout the contiguous
United States. Within each AOI, the receptor species were assigned to represent all of the faunal
classes, trophic levels, and feeding strategies that are typical of terrestrial and margin habitats,
respectively. The simple food webs created for the terrestrial and margin habitats provided the
context for receptor selection and were used to define the relationships between predators and
prey. National applicability was achieved primarily by selecting species that are widely
distributed throughout the contiguous United States, and then adding species to cover as many
ecological regions as possible. Several criteria were established as part of the selection process,
including geographical distribution, availability of data pertinent to exposure (e.g., body weight,
dietary preferences), and representation of the faunal classes and functional niches represented in
the terrestrial and margin food webs. In some instances, receptor species were added to
represent food web components in regions where "common" species were not likely to be found.
3-16
-------
Section 3.0
Spatial Layout for Environmental Settings
For example, the mule deer was selected as a large herbivore representing regions where the
white-tailed deer is not distributed. Similarly, the eastern cottontail rabbit and the black-tailed
jackrabbit were both included in the grasslands receptor group in order to provide a small
mammalian herbivore in temperate regions of the eastern United States and the drier western
regions. Based on information describing Bailey's ecological regions (Bailey, 1994), receptor
species were assigned to habitats within the AOI only if (1) the species was documented to occur
in the ecological region containing the site, and (2) the species represented a component of the
food web.
In assigning ecological receptors to habitats within a given AOI, it is important to
recognize the implicit assumption with respect to species occurrence. Specifically, it is assumed
that all receptor species occur in the representative habitats to which they are assigned regardless
of the AOI's position in the landscape. That is, if a forest habitat is delineated at a site, all the
species included in the terrestrial food web for the forest habitat that occur in that particular
region are assumed to be present. Although the simple food webs represent the major trophic
elements and feeding strategies that are likely present in a representative habitat, the food webs
are simplifications of what may be very different structures. In fact, it is unlikely that all of the
receptor species would be present, particularly those that are less adapted to human impacts and
development. For example, the black bear is included in the forest habitat receptor group and is
assessed at sites within its geographic range where forest habitat is indicated. However, the
black bear requires tracts of land significantly larger than the AOI, and this pattern may not
occur to an appreciable extent in areas where the land uses indicate an industrial base. As a
result, the list of receptors assigned to each habitat delineated within the AOI does not reflect
differences in species diversity related to habitat quality. Although small habitat patches were
connected within the AOI to simulate typical habitat use by wildlife, "habitat bridging" reflected
in the site layout files does not address the issue of habitat quality and species occurrence.
3.5 Summary
The delineation of habitats allows for the placement of receptor species within the AOI.
A substantial amount of data processing is required, including the assignment of receptor species
based on habitat and region, and the placement of the receptor species' home ranges at each site.
These steps are described in detail in Volume II of this report (U.S. EPA, 2003) and in
Section 15 of this document.
The spatial delineation of habitats throughout an AOI and the assignment of species to
these habitats provides an overall approach for the national assessment of ecological risk. The
variability in species affected by a contaminant release, both within a site and between sites, is
captured by this approach.
3.6 References
Anderson, J.R., E.E. Hardy, J.T. Roach, and R.E. Witmer. 1976. A Land Use and Land Cover
Classification System for Use with Remote Sensor Data. Geological Survey Professional
Paper 964. Pp. 1-34 in U.S. Geological Survey Circular 671. United States Geological
Survey, Washington, DC. Web site at
http://mapping.usgs.gov/pub/ti/LULC/lulcpp964/lulcpp964.txt. February 24.
3-17
-------
Section 3.0
Spatial Layout for Environmental Settings
Bailey, R.G., P.E. Avers, T. King, and W.H. McNab. 1994. Ecoregions and Subregions of the
United States (Bailey's Ecoregion Map). U.S. Department of Agriculture, Forest
Service, Washington, DC. Web site at http://www.epa.gov/docs/grdwebpg/bailey/.
Cowardin, L.M., V. Carter, and F. C. Golet. 1979. Classification of Wetlands andDeepwater
Habitats of the United States. FWS/OBS-79/31. Fish and Wildlife Service, Washington,
DC. December.
Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Transactions,
American Geophysical Union 38(6):913-920. December.
U.S. Bureau of the Census. 1987. Census of Agriculture: Geographic Area Series State and
County Data. Washington, DC.
U.S. Bureau of the Census: 1992. Census of Agriculture: Geographic Area Series State and
County Data. Washington, DC.
U.S. DOI and U.S. DOC (Department of the Interior and Department of Commerce). 1997.
1996 National Survey of Fishing, Hunting, and Wildlife-Associated Recreation. FHW/96
NAT. Fish and Wildlife Service and Bureau of the Census, Washington, DC. Web site
at http://www.nctc.fws.gov/library/pubs3.html. November.
U.S. EPA (Environmental Protection Agency). 1994a. 1:250,000 Scale Quadrangles of
Landuse/Landcover GIRAS Spatial Data in the Conterminous United States: Metadata.
Office of Information Resources Management, Washington, DC, pp. 1-9. Web site at
http://www.epa.gov/ngispgm3/nsdi/projects/giras.htm.
U.S. EPA (Environmental Protection Agency). 1994b. U.S. EPA Reach File. Version 3.0
Alpha Release (RF3-Alpha), Technical Reference, First Edition. Web site at
http://gispcl4/projects/hwir/rf3.htm. December.
U.S. EPA (Environmental Protection Agency). 2000a. Background Document for the
Ecological Exposure and Ecological Risk Modules for the Multimedia, Multipathway,
and Multiple Receptor Risk Assessment (3MRA) Software System. Office of Solid Waste,
Washington, DC. November.
U.S. EPA (Environmental Protection Agency). 2000b. Background Document for the Human
Exposure and Human Risk Modules for the Multimedia, Multipathway, and Multiple
Receptor Risk Assessment (3MRA) Model. Office of Solid Waste, Washington, DC.
June.
U.S. EPA (Environmental Protection Agency). 2003. Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) Modeling System, Volume II: Site-based,
Regional, and National Data. Office of Solid Waste, Washington, DC. July.
3-18
-------
Section 3.0
Spatial Layout for Environmental Settings
U.S. FWS (Fish and Wildlife Service). 1998. National Wetlands Inventory (NWI) Metadata.
National Wetlands Inventory, U.S. Fish and Wildlife Service, St. Petersburg, FL. Web
site at ftp://www.nwi.fws.gov/metadata/nwi_meta.txt. August.
USGS (Geological Survey). 1999. 1-Degree USGS Digital Elevation Models. U.S. Geological
Survey, Web site at http://edcwww.cr.usgs.gOv/glis/hvper/guide/l dgr dem.
3-19
-------
Section 3.0 Spatial Layout for Environmental Settings
3-20
-------
Section 4.0
Wastewater Source Modules
4.0 Wastewater Source Modules
4.1 Purpose and Scope
The Wastewater Source Modules simulate wastewater management. Two Wastewater
Source Modules were developed for the 3MRA modeling system to represent the common
management practices for industrial wastewaters. These management practices include flow
equalization, storage, treatment (typically biological treatment or neutralization), and solids
settling (clarification). The two waste management units (WMUs) for wastewaters are
¦ Surface Impoundments, which may be either aerated or quiescent and are used
to treat, store, or dispose of many industrial wastewaters; and
¦ Aerated Tanks, which are aerated or mixed tanks used to treat or store many
industrial wastewaters.
The two Wastewater Source Modules use a common set of algorithms with similar mass
balance and transport equations. Both modules estimate volatile emissions to air. Leaching
losses from the bottom of the unit are modeled only for surface impoundments; tanks are
assumed to have impervious bottoms. Both modules simulate degradation and solids settling,
although solids settling and accumulation are more significant for quiescent units.
The two Wastewater Source Modules were designed to provide estimates of annual
average volatilization rates to air, which are used by the Air Module. In addition, the Surface
Impoundment Module outputs annual average infiltration rates and leachate constituent flux
rates, which are used by the Vadose Zone and Aquifer Modules, and annual average surface
impoundment water concentrations, which are used by the Ecological Exposure Module to
estimate exposure to wildlife receptors that may drink or consume organisms from the
impoundment. Figure 4-1 shows the relationship and information flow between the Wastewater
Source Modules and the 3MRA modeling system.
The Wastewater Source Modules have six major functions, as follows:
1. Calculate constituent concentrations within the unit. The Wastewater Source
Modules use a mass-balance, temperature-adjusted approach to estimate
constituent concentrations in the WMU. This approach considers constituent
diffusion between wastewater and sediments, and constituent removal by
volatilization, biodegradation, hydrolysis, partitioning to solids, solids settling,
and, for surface impoundments only, infiltration through the bottom of the unit.
4-1
-------
Section 4.0
Wastewater Source Modules
Key Data Inputs
Operating lifetime
Unit depth
Number of impellors
Henry's law constant
Surface
Impoundment
and Aerated
Tank Modules
Emission Rates
(volatile only)
Chemical Fluxes Infiltration Rates
Air Module
(SI only)
Vadose Zone
and Aquifer
Modules
Wastewater Concentrations
Ecological
Exposure
Module
(SI only)
Figure 4-1. Information flow for the Wastewater Source Module
in the 3MRA modeling system.
2. Calculate solids concentrations within the unit. The Wastewater Source
Modules use a mass-balance approach to estimate solids concentrations in the
WMU.
3. Calculate volatile emission rates. The modules calculate volatile emission rates
for both aerated and quiescent surfaces.
4. Estimate resuspension, sedimentation, and burial velocities within the units.
5. Estimate constituent release in leachate. The Surface Impoundment Module
calculates infiltration rates and constituent leachate flux rates for use in the
Vadose Zone and Aquifer Modules. The Aerated Tank Module does not calculate
leachate release, as tanks are assumed to have impervious bottoms.
6. Adjust for temperature effects. The modules account for the effect of
temperature on air viscosity and density, water viscosity, chemical properties, and
sediment biodegradation rates.
4.2 Conceptual Approach
This section first describes the two different WMUs and the processes modeled for them,
and then describes the major model functions listed above.
4-2
-------
Section 4.0
Wastewater Source Modules
4.2.1 Description of Waste Management Units
The Aerated Tank Module was developed to model tanks that are aerated or mixed; thus,
it uses a well-mixed, steady-state mass balance solution. Tanks are assumed to have an
impervious bottom; therefore, leaching is not modeled for tanks.
The Surface Impoundment Module was developed to model both aerated and quiescent
impoundments using either a well-mixed, steady-state mass balance solution for well-mixed
impoundments or a time-dependent mass balance solution for plug flow, batch, or disposal
impoundments.
The following assumptions were made in developing both modules:
¦ The WMU is divided into three distinct compartments: impoundment liquid,
unconsolidated sediment, and consolidated sediment. Each compartment has a
fixed volume for a given monthly solution and volumes are readjusted to account
for solids accumulation. The three-compartment model is described in more
detail later in this section.
¦ Aerated tanks are well mixed; surface impoundments may be well mixed or have
plug or batch flow. For well-mixed units, a steady-state mass balance solution is
used. For plug or batch flow impoundments, a time-dependent mass balance
solution is used.
¦ Volatilization in the liquid compartment follows first-order kinetics.
¦ Hydrolysis in both the liquid and sediment compartments follows first-order
kinetics.
¦ Aerobic biodegradation in the liquid compartment follows first-order kinetics
with respect to both constituent concentration and biomass concentration.
¦ Biomass growth rate follows Monod kinetics with respect to total biological
oxygen demand loading.
¦ Solids settling follows first-order kinetics.
¦ Anaerobic biodegradation of constituent in the sediment compartments follows
first-order kinetics.
¦ The biomass decay rate within the accumulating sediment compartment is first-
order.
¦ There is no constituent in precipitation (i.e., rain or snow).
¦ Constituent partitioning among adsorbed solids, dissolved phases, and vapor
phases is linear.
4-3
-------
Section 4.0
Wastewater Source Modules
¦ A moisture-dependent infiltration rate is calculated using Darcy's law and Van
Genuchten moisture relationships.
Figure 4-2 illustrates the model construct for both Wastewater Source Modules, which
are divided into three primary compartments:
¦ The liquid compartment includes the influent and effluent wastewater streams
and the aqueous liquid layer above the sediment at the bottom of the unit. Liquid
enters the compartment via the influent waste stream and rainfall, and leaves the
compartment via evaporation and the effluent waste stream. Processes modeled in
this compartment include volatilization, aerobic biodegradation, constituent
hydrolysis, biomass solids growth, and solids settling and resuspension.
¦ The unconsolidated sediment compartment is a layer of loose sediment
immediately below the liquid compartment. Processes modeled in this layer
include anaerobic biological degradation and burial of solids into the consolidated
sediment compartment.
¦ The consolidated sediment compartment is a layer of compacted sediment at
the bottom of the unit. Processes modeled in this compartment include anaerobic
degradation and removal of solids by cleaning and dredging and by
decomposition due to anaerobic digestion.
II Evap
Influent
platile
issions
aerated or nonaerated liquid surfa
le
(SI
ach§
on
Q[fl ui ion
te
fl U:
Liquid Compartment
Volatilization
Aerobic biodegradation
First-order chemical hydrolysis
Biomass growth
Solids settling and resuspension
V
Effluent \
Unconsolidated Sediment Compartment
Anaerobic degradation/decay
Solids burial
Sediment Removal
Consolidated Sediment Compartment
Anaerobic degradation
Figure 4-2. Conceptual model schematic for Wastewater Source Modules.
4-4
-------
Section 4.0
Wastewater Source Modules
As shown in Figure 4-2, sediment is deposited and resuspended across the boundary
between liquid and unconsolidated sediment, but only deposition (burial) occurs between the
unconsolidated to consolidated sediment compartments. The modules also allow constituents to
diffuse between adjacent compartments. In addition, the Surface Impoundment Module models
leachate loss from the liquid layer through the sediment layers to the underlying soil.
For each compartment, the Wastewater Source Modules perform mass balances at time
intervals small enough that the hydraulic retention time in the liquid compartment is not
significantly affected by the solids settling and accumulation. In the liquid compartment, there is
flow both in and out of the WMU, with constituent loss through volatilization, constituent decay
(hydrolysis), aerobic biodegradation, and particle settling and burial (net sedimentation).
The following processes are modeled in the liquid compartment:
¦ Volatilization using a two-film model and mass transfer correlations from Air
Emissions Models for Waste and Wastewater (U.S. EPA, 1994).
¦ Constituent loss due to hydrolysis and biodegradation using first-order rate
constants for natural soil systems. These are adjusted to WMU conditions by
assuming an effective system biomass concentration of 2.Ox 10"6 Mg/m.3
¦ Particle removal using particle settling velocities to estimate the projected
sediment removal efficiency of the unit.
The modules estimate solids generation in the liquid compartment according the relationship of
biological growth to the decomposition of organic chemicals (as biological oxygen demand) in
the influent. These solids settle in the liquid to form the two underlying sediment compartments.
The following processes are modeled in the sediment compartments:
¦ Constituent loss due to hydrolysis and (anaerobic) biodegradation using first-
order degradation rate constants. Hydrolysis and biodegradation rate constants are
assumed to apply to the total constituent concentration (both dissolved and sorbed
constituent) in each compartment.
¦ Constituent mixing between the compartments through constituent diffusion as
well as particle sedimentation and resuspension.
¦ Solids destruction due to sludge digestion. It is assumed that solids
decomposition is limited to the fraction of solids that are biologically active.
¦ Sediment compaction in the consolidated compartment, estimated by calculating
the vertical effective stress across the consolidated sediment. Sediment
compaction impacts the depth, volumetric water fraction, and the effective
hydraulic conductivity of the consolidated sediment compartment.
4-5
-------
Section 4.0
Wastewater Source Modules
Using the well-mixed assumption, the model assumes that the suspended solids
concentration within the WMU is constant throughout the unit. However, some stratification of
sediment is expected across the length and depth of the WMU so that the effective total
suspended solids concentration within the unit is assumed to be a function of the WMU's total
suspended solids removal efficiency rather than equal to the effluent total suspended solids
concentration. The liquid (dissolved) phase constituent concentration within the unit, however,
is assumed to be equal to the effluent dissolved phase concentration for the well-mixed model
solution.
4.2.2 Calculate Constituent Concentrations within the Unit
The governing constituent mass balance equations used to calculate the constituent
concentrations in the liquid and sediment compartments of the impoundment or tank are
presented in the box starting on the next page. Detailed equations and solutions used by the
Surface Impoundment and Aerated Tank Modules can be found in U.S. EPA (1999), including
all assumptions and algorithms used in the modules.
The following sections describe the calculations for each of the three compartments, and
the calculations for diffusion between the liquid and sediment compartments.
4.2.2.1 Liquid Compartment. The change in the constituent mass in the liquid
compartment is a function of the constituent loss through volatilization, aerobic biodegradation,
hydrolysis, and infiltration (denoting liquid flow of both liquid and entrained solids) into the
sediment compartment. In addition, constituent is transported across the liquid/sediment
compartment interface by constituent diffusion and by solids settling and resuspension. These
processes are affected by whether the unit is aerated or nonaerated, and by constituent-specific
properties.
For the time-dependent solution, the initial liquid compartment concentration is equal to
the influent waste concentration.
4.2.2.2 Unconsolidated Sediment Compartment. The change in constituent mass
within the unconsolidated sediment compartment is dependent on constituent infiltration from
the liquid compartment (which includes entrained sediment), "filtered" leachate out the bottom
of the compartment, and constituent loss through hydrolysis and anaerobic biodegradation. In
addition, constituent is transported across the liquid/sediment compartment interface by
constituent diffusion and by solids settling and resuspension. Constituent is transported across
the unconsolidated/ consolidated compartment interface by constituent diffusion.
For aerated tanks, the initial sediment depth is zero. For surface impoundments, the
initial sediment depth is 20 cm (unconsolidated). When the simulation begins, this initial
sediment mass is partitioned equally into the unconsolidated and consolidated sediment
compartments. The sediment mass is fixed for each month. At the end of the month, the gross
mass of sediment accumulated during the month is calculated and the mass of sediment that is
decomposed is calculated, resulting in a net mass of accumulated sediment. This net mass of
accumulated sediment is partitioned equally into the unconsolidated and consolidated sediment
compartments prior to simulation of the next month.
4-6
-------
Section 4.0
Wastewater Source Modules
Aerated Tank and Surface Impoundment Constituent Mass Balance Equations
(variables listed on next page)
Liquid Compartment
Time-dependent constituent mass balance (surface impoundments only):
Q^cw + f
-------
Section 4.0
Wastewater Source Modules
Constituent mass balance equation variables
(x denotes the compartment: 1 = liquid, 2 = unconsolidated sediment, 3 = consolidated sediment)
^tot, X
Ctot, infl
c
^tot, out
Qleach
Qinfl
Qout
K-ol
A
V,
V2
V3
di
^hyd
^bm
^ba
[TSS]X
[MLVSS]]
^sed
Vres
Vb
vdiffl2
Vdiff23
fix
total constituent concentration in compartment x (mg/L = g/m3)
total constituent concentration in influent (mg/L = g/m3)
total constituent concentration in effluent (mg/L = g/m3)
leachate flow rate from WMU (m3/s)
influent flow rate into WMU (m3/s)
effluent flow rate out of WMU (m3/s)
overall volatilization mass transfer coefficient (m/s)
total surface area of WMU (m2)
volume of liquid compartment in WMU= d, A (m3)
volume of unconsolidated sediment compartment in WMU (m3)
volume of consolidated sediment compartment in WMU (m3)
depth of liquid compartment (m)
hydrolysis rate (1/s)
complex first order biodegradation rate constant (m3 /Mg-s)
ratio of biologically active solids to the total solids concentration (i.e., kba =
[MLV S S] /[TS S] j)"
anaerobic biodegradation decay rate of constituent (1/s)
concentration of total suspended solids (TSS) in compartment x (g/cm3 =
Mg/m3)
concentration of biomass as mixed liquor volatile suspended solids (MLVSS)
liquid compartment (g/cm3 = Mg/m3)
solids settling or sedimentation velocity (m/s)
solids resuspension velocity (m/s)
solids burial velocity (m/s)
mass transfer coefficient between liquid compartment (1) and unconsolidated
sediment compartment (2) (m/s)
mass transfer coefficient between unconsolidated sediment compartment (2)
and consolidated sediment compartment (3) (m/s)
dissolved constituent fraction in compartment x:
f =
J d,x
a
liq,x
1
C
tot,x
Ap,x
= particulate constituent fraction in compartment x:
f =¦
J p,X
Uresl,
c— "(sw+UresL)
liq,x
CliqjX
r
^SOljX
kds
Koc
f
volumetric liquid fraction of compartment x (m3/m3)
liquid-phase constituent concentration in compartment x (mg/L = g/m3)
solid-phase constituent concentration in compartment x (mg/kg = g/Mg)
solid-water partition coefficient (m3/Mg) = Kocxfoc for organics
soil-water partitioning (m3/Mg)
fraction organic carbon in the waste (mass fraction).
4-8
-------
Section 4.0
Wastewater Source Modules
The initial sediment compartment concentrations are calculated by assuming the
sediment compartments' void space is filled with waste at the influent concentration and that the
initial sediment particles do not contain any contaminant. For each subsequent month, the initial
sediment compartment constituent concentration (for either sediment compartment) is estimated
based on the previous month's sediment compartment concentration. For a plug-flow unit, it is
estimated as the log mean average sediment compartment concentration across the unit from the
previous month. For batch units, it is the final sediment compartment concentration at the end of
the previous month.
4.2.2.3 Consolidated Sediment Compartment. The change in the constituent mass
within the consolidated sediment compartment is a function of "filtered" leachate flow from the
unconsolidated sediment compartment and out the bottom of the unit, and constituent loss
through hydrolysis and biodegradation. In addition, constituent is transported across the
unconsolidated/consolidated compartment interface by constituent diffusion.
4.2.2.4 Diffusion between Liquid and Sediment. The models estimate effective
diffusion velocity between the liquid and sediment compartments using the following two-
resistance model based on the liquid phase mass transfer coefficient for quiescent surfaces and
the porosity of the sediment compartment:
vdiff
< 1 1 X
+
k It
Khq Keff,2
(4-1)
where
vdifr = effective diffusion velocity between liquid and sediments
kj q = liquid phase mass transfer coefficient for quiescent surface areas as calculated
in Section 4.2.3 (m/s)
kcfK2 = effective liquid mass transfer coefficient in sediment compartment (m/s).
To determine the effective liquid mass transfer coefficient in the sediment compartment,
the models first calculate the effective liquid diffusion rate from the porosity of the sediment
layer using the Millington-Quirk tortuosity model (Millington and Quirk, 1961):
- e,i x a,, <4-2>
where
Deff,2 = effective liquid diffusion rate (cm2/s)
0liq,2 = volumetric porosity (assumed to be liquid filled) of sediment compartment =
1 - [TSS]2 / pxss, where pxss = density of total suspended solids
Du = diffusivity in liquid (water) (cm2/s).
4-9
-------
Section 4.0
Wastewater Source Modules
In most cases, the sediment accumulating at the bottom of the WMU will be more of a
viscous sludge layer than a rigid mass of particles. Therefore, the top layer of sediment is
expected to be affected by the bulk currents within the WMU (caused by wind shear, aeration, or
mixing) similar to the liquid phase mass transfer coefficient for quiescent surfaces. The liquid
phase quiescent mass transfer coefficient is primarily a function of the liquid diffusivity raised to
the two-thirds power; therefore, the models estimate the effective liquid mass transfer coefficient
from the liquid compartment to the sediment layer as follows:
eff,2
= k.
D*,i
ll
A,i
8
= k 0 9
Kl,q liq,2
(4-3)
4.2.3 Calculate Solids Concentrations within the Unit
The time-dependent solution for estimating total solids concentration within the unit is
based on the assumption of constant total suspended solids (TSS) concentration. In reality, the
solids concentration varies, either increasing as a result of biomass production from the
consumption of organic material in the waste stream or decreasing as a result of solids settling.
The solids mass balance for the liquid compartment is
rBOD^ + vreA12^4 - Q!each[TSS]l - vsecjA\TSS\l (4-4)
- acrjrasy
dt
where
[TSS], = concentration of total suspended solids in liquid compartment (g/cm3 =
Mg/m3)
Vj = volume of liquid compartment (m3)
rBOD = normalized biodegradation rate of BOD5 (g-BOD/g-biomass/sec)
X = biomass yield (g-biomass (dry basis)/g-BOD consumed)
kba = ratio of biologically active solids to the total solids concentration
(i.e.,kba = [MLVSS]1/[TSS]1)
vres = solids resuspension velocity (m/s)
[TSS]2 = concentration of total suspended solids in unconsolidated sediment
compartment (g/cm3 = Mg/m3)
A = total surface area of WMU (m2)
Qieach = leachate flow rate from WMU (m3/s)
vsed = solids settling or sedimentation velocity (m/s).
The normalized BOD5 biodegradation rate is estimated using the Monod equation as
follows:
Kbmax CBOD , , _
rBOD (V- + r \ )
°BOD)
4-10
-------
Section 4.0
Wastewater Source Modules
where
CBoD,infi = BOD5 concentration in the influent to the WMU (g/cm3 or Mg/m3)
CBod,i = BODj concentration in the liquid compartment (g/cm3 or Mg/m3)
Kbmax = maximum BOD5 biodegradation rate (g-BOD/g-biomass/sec or Mg/Mg/sec)
= 6.94x 10"6 x Tcorr (the value of 6.94x 10"6 comes from the maximum rate
of 0.6 g-BOD/g-biomass/hr ^ 86,400 sec/hr)
Kb2 = half-saturation constant = 0.00005 (g/cm3 or Mg/m3)
Tcorr = temperature correction factor for biodegradation rate constants (see Section
4.2.7.5).
The models estimate the maximum BOD5 degradation rate constant based on a typical
design value for F/M (a 0.6 food-to-biomass ratio) for activated sludge systems based on values
in Eckenfelder et al. (circa 1984) and Hermann and Jeris (1992). The model uses a typical half-
saturation rate constant (Kb2) of 50 mg/L (0.00005 g/cm3) selected from values reported in the
literature (Tabak et al., 1989; Gaudy and Kincannon, 1977; Goldsmith and Balderson, 1989;
Rozich et al., 1985) .
The integration period time steps in the overall time-dependent solution depend on how
quickly the total suspended solids and BOD5 concentrations vary. The model selects an initial
time step based on the initial concentrations of TSS and BOD5 so that BOD5 concentrations are
effectively constant. The TSS concentration is calculated at the end of the first time step and
compared to the starting total suspended solids concentration; if it changes by more than a factor
of 5, additional time steps are added until the starting and ending TSS concentrations differ by
less than a factor of 5. This allows the model to determine an effective average TSS
concentration across a given time step that can be used as a constant value in the constituent
mass balance solution equations.
For well-mixed systems, the model calculates the effluent TSS concentration from the
predicted solids removal efficiency of the unit (see Section 4.2.5) and the predicted BOD5
removal efficiency. The TSS mass balance for the liquid compartment can be written as
[mo. = (1 - <*s> (PSSG,„ + * £I0D cw) (4.6)
yout
where
= total suspended solids mass removal efficiency in WMU (unitless)
= biomass yield (g-biomass (dry basis)/g-BOD)
= biological oxygen demand removal efficiency of WMU (unitless)
CBoD,infi = biological oxygen demand of influent (Mg/m3)
[TSS]infl = concentration of total suspended solids in the influent (g/cm3 = Mg/m3).
-BOD
4-11
-------
Section 4.0
Wastewater Source Modules
[TSS]0Ut = concentration of total suspended solids in the effluent (g/cm3 = Mg/m3).
Qinfl = influent flow rate into WMU (m3/s)
Qout = effluent flow rate out of WMU (m3/s).
Although the liquid compartment is assumed to be well mixed, solids removal and
growth are not expected to be instantaneous. To account for gradients of TSS concentration
along the WMU length and depth, the model estimates the effective TSS concentration within
the WMU as the log-mean average between the influent and effluent total suspended solids
concentrations (based on first-order sedimentation). Given the influent and effluent total
suspended solids concentrations, the effective (mean) total suspended solids concentration in the
liquid compartment is
[TSS] j = exp
(in[rayw) + H[TSS]0Ut)
(4-7)
The model calculates BOD5 removal efficiency based on the BOD5 biodegradation rate
using the same biodegradation rate model used for the time-dependent solution. Because the
model uses BOD5 degradation rate primarily to determine the production rate of biological solids
within the impoundment, it neglects decreases in BOD5 concentrations due to dilution by
precipitation (i.e., the influent flow rate is assumed to be equal to the effluent flow rate plus the
leachate flow rate). With this simplification, the BOD5 removal efficiency can be written as
follows:
[res], Fi
BOD n r (4-8)
xlinfl BODJnfl
Because BOD5 removal efficiency depends on the effective TSS concentration, the model
calculates the effective TSS concentration using iterative calculations between estimating the
BOD5 removal efficiency and the effective TSS concentration.
4.2.4 Calculate Volatile Emission Rates
The Wastewater Source Modules use an overall mass transfer coefficient that determines
the rate of volatilization based on a two-resistance model: a liquid-phase mass transfer
resistance and a gas-phase mass transfer resistance. The liquid- and gas-phase mass transfer
resistances for turbulent surfaces are very different from those for quiescent (laminar flow)
surfaces. Therefore, the overall mass transfer coefficient is a composite of the coefficients for
the turbulent surface area and the quiescent surface area. The overall coefficient is based on an
area-weighted average, as follows:
Knr A + Kn, A
Kol = °u 1 (4-9
4-12
-------
Section 4.0
Wastewater Source Modules
where
Kol = overall mass transfer coefficient for the WMU (m/s)
KOL,t = overall mass transfer coefficient for turbulent surface areas (m/s)
At = turbulent surface area = faer A(m2)
faer = fraction of total surface area affected by aeration (unitless)
A = total surface area (m2)
K0L,q = overall mass transfer coefficient for quiescent surface areas (m/s)
Aq = quiescent surface area (m2) = (l-faer) x A (Note: At + Aq must equal A).
The overall mass transfer coefficient for turbulent surface areas based on the two-
resistance model is
V/
( i 1 V1
+
ht h' k.
g>tj
(4-10)
where
kljt = liquid-phase mass transfer coefficient for turbulent surface areas (m/s)
H' = dimensionless Henry's law constant = H/RTh
k t = gas-phase mass transfer coefficient for turbulent surface areas (m/s).
Similarly, the overall mass transfer coefficient for quiescent surface areas is
K.
OL,q
Kt
1_
H' k
-1
(4-11)
%>i /
where
kl q = liquid-phase mass transfer coefficient for quiescent surface areas (m/s)
kg q = gas-phase mass transfer coefficient for quiescent surface areas (m/s).
The mass transfer correlations used to estimate the liquid- and gas-phase mass transfer
coefficients for turbulent and quiescent surfaces are the same as those used in the WATER8 and
CHEMDAT8 emission models developed by EPA. Basic equations are provided in U.S. EPA
(1999) with a more detailed treatment in Chapter 5 of the CHEMDAT8 model documentation
(U.S. EPA, 1994).
4.2.5 Estimate Resuspension, Sedimentation, and Burial Velocities
To solve the constituent and sediment mass balance equations, the Wastewater Source
Modules must estimate the transfer rate of the sediment (and its associated constituent content)
between the liquid and sediment compartments in the WMU. Key parameters in this calculation
include
4-13
-------
Section 4.0
Wastewater Source Modules
¦ Sedimentation velocity, which establishes the rate at which particles in the liquid
compartment enter the unconsolidated sediment compartment;
¦ Resuspension velocity, which establishes the rate at which particles in the
unconsolidated sediment compartment enter the liquid compartment; and
¦ Burial rate, which is the net accumulation rate of sediment in the two sediment
compartments.
Sediment movement between the liquid and sediment compartment should vary primarily
with the dimensions and flow characteristics of the WMU, and with the relative surface area
affected by turbulent mixing. The general approach used to estimate the various sediment
transport rates is to first estimate the suspended solids mass removal efficiency of the WMU.
Given this removal efficiency, the Wastewater Source Modules estimate the resuspension,
sedimentation, and burial velocities based on the characteristics of the mean-sized particles in
the WMU.
4.2.5.1 Estimate Design Sediment Removal Efficiency. The WMU quiescent surface
area and flow rate are used to calculate the vertical (or "upflow") velocity of the impoundment as
follows:
Q n
V upflow ~ ~~4 (4-12)
Aq
where
vuPfiow = upflow velocity (m/s)
Qinfl = influent flow rate into WMU (m3/s)
Aq = quiescent surface area (m2).
The upflow velocity is assumed to act on the liquid compartment to cause an upward flux of
particles. The model estimates sediment removal efficiency in the WMU from WMU flow rate,
surface area (i.e., the upflow velocity), and particle size distribution characteristics (mean
particle size and relative standard deviation) by considering the terminal settling velocity of the
particles. Particles with a terminal settling velocity greater than the upflow velocity settle within
the WMU, while particles with a terminal velocity less than the upflow velocity remain
suspended and are entrained in the effluent.
The model assumes that suspended solids are spherical when calculating the terminal
velocity (or critical particle diameter) and the mass-to-volume ratio of the particles. The model
calculates the terminal velocity of a sphere using Stoke's Law (see U.S. EPA 2001), and
determines the particle diameter that has a terminal velocity equal to the upflow velocity. The
mass sediment removal efficiency of the WMU is then calculated from the particle size
distribution (model input parameters assuming lognormal distribution) and the mass of particles
of a given diameter (based on spherical particles). The lognormal distribution density function is
4-14
-------
Section 4.0
Wastewater Source Modules
=
dpart ° (2*)"
exp
Mdpgr/dmegn)]:
2 o2
(4-13)
where
-------
Section 4.0
Wastewater Source Modules
resuspended will equal the mass settling when the average TSS concentration in the liquid
compartment equals the target effluent concentration. The burial rate is then calculated for each
individual time step based on the difference in the sedimentation and resuspension rates at the
average liquid compartment TSS concentration for that time step. The equations for the
resuspension, sedimentation, and burial velocities are
v,
sed
= 100 x v , (4-16^
part,mean v ^ 1 vj
[i®U,
Iresf <4"17>
where
vb =
Q,
leach
+ V
sed
[TSS]2
- v„
(4-18)
part,mean
[TSS]target
[TSS]ave
particle settling velocity of a mean-diameter particle (cm/s)
target effluent TSS concentration = [TSS]infl (1- eXSSo) (g/cm3 = Mg/m3)
average TSS concentration in the liquid compartment for a given time
interval (g/cm3 = Mg/m3)
solids burial velocity (m/s).
As constructed, the time-dependent solution assumes the mass rate of sediment
resuspension will equal the mass rate of sediment settling at the target or design effluent TSS
concentration. The rate at which the target TSS concentration is reached is dependent on the
particle characteristics as well as the growth rate of biomass (i.e., the BOD5 consumption rate).
The actual effluent TSS concentration predicted by the model may not reach the target TSS
concentration at very low hydraulic residence times or where significant quantities of biosolids
are produced. As sediment accumulates in the WMU, the corresponding change in the hydraulic
residence time may also affect the predicted effluent TSS concentration.
4.2.5.3 Estimate Resuspension. Sedimentation, and Burial Velocities for Well Mixed
Model Solution. In the well-mixed model, mass balance consideration of the sediment requires
that the suspended solids burial (or accumulation) rate be determined from the predicted
sediment removal efficiency. As constructed, the design sediment removal efficiency is
independent of WMU depth, and therefore does not change as sediment accumulates in the
WMU. This will generally be true for large depths, but for shallower depths, the increased
lateral flow rates tend to cause "short-circuiting" flow patterns, which decrease the sediment
removal efficiency of the WMU. To take this phenomenon into account, it is assumed that the
sediment removal efficiency remains constant at the design efficiency (i.e., eTSS = eTSS o) at liquid
depths of 1.2 meters (4 feet) or more based on design considerations of settling chambers. As
4-16
-------
Section 4.0
Wastewater Source Modules
the liquid depth becomes less than 1.2 meters, it is assumed that the sediment removal efficiency
will decrease as a function of the liquid retention time. A first-order sedimentation rate constant
is estimated based on the design sediment removal rate and the WMU retention time at a liquid
depth of 1.2 meters. This first-order sedimentation rate constant is calculated as
^ ~ hi (1 - eBS J
sed (1.2 m) A (4-19)
Qinfl
where
ksed = apparent first-order sedimentation rate at a liquid depth of 1.2 meters (1/s).
For liquid depths less than 1.2 meters, the removal efficiency is estimated using this first-
order sedimentation rate constant and the hydraulic retention time as
^sed A
e = i - e\~^T) (4"2°)
^TSS 1 e
where
6tss = predicted mass sediment removal efficiency of the WMU as sediment
accumulates (mass fraction)
d, = depth of liquid compartment.
The predicted mass sediment removal efficiency is assumed to apply equally to influent
sediment and sediment generated within the unit. The net rate of sediment transfer or burial
from the liquid compartment to the sediment compartment can be calculated based on a mass
balance of sediment in the liquid compartment, which can be rearranged to calculate the burial
velocity (defined in terms of the sediment concentration in the sediment compartment) as
follows:
v,
Qinfi (ITSS\infl + A eBOD CBOD) (1 eTSS)
h A [TSS]7
(4-21)
The resuspension velocity acts on the sediment compartment, and it is assumed to affect
the same upward flux of sediment as the upflow velocity. Therefore, the resuspension velocity
can be calculated from the upflow velocity and the relative concentrations of particles in the
liquid and sediment compartments as follows:
4-17
-------
Section 4.0 Wastewater Source Modules
[TSSJ
Vres Vupflow J (4-22)
The sedimentation rate is calculated from the mass balance of sediment in the sediment
compartment (Equation 4-4), which can be rearranged as follows.
v,
[^2 Qleack
sed
- (V"* + Wt - A (4"23)
where [TSS], is calculated from [TSS]infl and [TSS]0Ut using Equation 4-7.
4.2.5.4 Estimate Sediment Decomposition. The burial rate is the total sediment
accumulation rate for the time step. To account for the reduction in solids typically associated
with anaerobic digestion, a sediment decomposition rate (or sludge digestion rate) is included in
the burial (accumulation) compartment. If the entire sediment compartment included this
anaerobic digestion term, a more rigorous accounting of the biological (organic) versus inert
solids would be required, but, ultimately, the sediment compartment will reach a steady state
(i.e., biomass growth equals biomass decay). By including it only in the burial (accumulation)
compartment, sediment reduction (which includes a constituent reduction associated with the
sediment) by digestion can be included without significantly complicating the model. The net
accumulation of sediment over a time step is estimated as
Ad2 = v„ 4/ [1 - kba (1 - e"'*-")] (4-24)
where
Ad2 = change in depth of the unconsolidated sediment compartment (m)
At = time step (s)
kba = ratio of biologically active solids to the total solids concentration - assumed to
be the same ratio as present in the liquid compartment
kdec = anaerobic digestion/decay rate of the organic sediment (1/sec).
Prior to the next time step calculations, Ad2 is added to d2 (and subtracted from d,).
Additionally, the total amount of sediment in the tank or impoundment and the total time since
the last cleaning or dredging action is compared to the input cleaning and dredging parameter
(i.e., the fraction of the WMU that can be filled with sediment before the WMU is cleaned or
dredged). The module will also automatically run the dredge subroutine in the event that the
sediment settling for the next time step (based on the sediment settling for the current time step)
would completely fill the WMU. The removed sediment and the contaminant associated with
the removed sediment is recorded; this removal acts as a sink for the overall system.
4-18
-------
Section 4.0
Wastewater Source Modules
4.2.6 Estimate Constituent Release in Leachate (Surface Impoundments only)
The Surface Impoundment Module estimates leachate infiltration rate from liquid depth
and from the hydraulic conductivities and thicknesses of three layers: the sediment
compartment, a clogged soil layer, and the underlying native soil. This procedure follows the
method outlined in EPA's Composite Module for Leachate Migration with Transformation
Products (EPACMTP) background document (U.S. EPA, 1996), except that the liquid depth is
known and there is a sediment layer between the impoundment liquid and the underlying soil
layer.
The Surface Impoundment Module treats the unconsolidated sediment layer as free liquid
to calculate the pressure head on the consolidated sediment layer and underlying soil. The model
calculates the infiltration rate in an iterative manner. It makes an initial estimate of the
infiltration rate, calculates the associated pressure profile in the underlying soil, and compares
the calculated pressure head at the ground water surface with the boundary condition (i.e.,
pressure head of zero). Based on this comparison, the model revises the infiltration rate estimate
and iterates until the boundary conditions are met.
Based on Darcy's law, the leaching (infiltration) rate for a given soil sublayer is:
T = K k
n s,n rw,n
% ~ 1
s,n
(4-25)
where
In = infiltration rate (m/d)
Ks n = hydraulic conductivity of the nth soil sublayer (m/d)
k^ H = relative permeability of the nth soil sublayer (unitless)
Yn = pressure head at top of the nth soil sublayer (m)
| = pressure head at base of the nth soil sublayer (m)
ds n = depth of the nth soil sublayer (m).
The relative permeability is a function of the effective saturation and can be expressed by
soil class parameters using relationships developed by Van Genuchten (1980) as follows:
if <1> * 0 = 1 (4-26a)
. (i - (-«„*,/•"'[i+ (-
-------
Section 4.0
Wastewater Source Modules
Solution methods for these equations can be found in the EPACMTP background
document (U.S. EPA, 1996). As shown in Figure 4-3, the Surface Impoundment Module applies
these equations to the sediment layer, a liner or soil layer clogged with sediment particles
immediately below the surface impoundment, and another soil layer under that. The sediment
layer is assumed to be saturated and modeled as a single layer using Darcy's law. The underlying
soil layers are partially saturated and are modeled with five sublayers using the Van Genuchten
relationships and algorithms developed for EPACMTP.
Top of Liquid Compartment
Topographical Level
//&/&//$ 1
A
¦
///$//&//$
Liquid Compartment
"
1 Loose Sediment DSJ
Dciog *
Compacted Sediment (Filter Cake) „ Of-Cv '
, , , ,
Clogged Native Material
Native Material
Groundwater
Grahics:\EPA\definrtion_sketah.cdr
Water Table
Figure 4-3. Surface impoundment cross-section showing sediment and soil layers modeled by the
Surface Impoundment Module infiltration rate algorithms.
4.2.6.1 Leachate Infiltration Rate. The Surface Impoundment Module calculates a
leachate infiltration rate through three compartments: consolidated sediment, clogged soil, and
subsoil. Initially, the consolidated sediment is saturated and the soil layers are unsaturated. The
model divides each of the soil sublayers into three sublayers. The unconsolidated sediment layer
is loose (fluid) so that the effective pressure head for the consolidated sediment layer is simply
the liquid depth plus the depth of the unconsolidated sediment. The model assumes that the
system is at steady state; therefore, a water balance dictates that the infiltration rate is the same
for all compartments and compartment sublayers. Assuming the pressure head at the ground
water interface is zero, the general solution for the infiltration rate is shown in Equation 4-27.
4-20
-------
Section 4.0
Wastewater Source Modules
/= (4 * p.* 1 dj
(4-27)
4, ^
y ^s,rftrw,n j
where
I = leachate infiltration rate (m/d)
Ks n = saturated hydraulic conductivity of the nth soil sublayer (m/d)
n = relative permeability of the nth soil sublayer (unitless)
ds n = thickness of the nth soil sublayer (m).
The relative permeabilities (krw) of the clogged soil and native soil sublayers are a function of
whether the previous sublayer is saturated. The model uses an interative, steady-state method to
solve this equation across the sediment and soil compartments, as described in U.S. EPA (1999).
The infiltration rate is then set equal to the lowest of the calculated rates across the three
compartments.
The Surface Impoundment Module calculates the volumetric leachate flow rate from the
calculated infiltration rate as follows:
Qieach ~ 24 x 3600 (4'28)
where
Qieach = leachate infiltration flow rate (m3/s)
A = surface impoundment area (m2).
Leachate flow rates and leachate contaminant concentrations (calculated as the liquid
contaminant concentration in the consolidated sediment compartment) are output from the
Surface Impoundment Module as a time series of annual-average values.
4.2.6.2 Hydraulic Conductivity of Consolidated Sediment. As sediment accumulates
at the bottom of the impoundment, the weight of the liquid and upper sediments tends to
compress (or consolidate) the lower sediments. This consolidated sediment acts as a filter cake,
and its hydraulic conductivity may be much lower than the unconsolidated sediment. The
Surface Impoundment Module assumes that the sediment compartment is at pseudo-steady-state
and that all sediment layer thicknesses are nearly stationary and approximately constant. The
Surface Impoundment Module sets the initial sediment depth at 20 cm to account for sediment
and compaction created during the excavation of the impoundment.
4-21
-------
Section 4.0
Wastewater Source Modules
The Surface Impoundment Module simulates the effective stress of the overlying liquid
and unconsolidated sediment on the compacted sediment layer, and the effects of this stress on
the porosity and hydraulic conductivity of the filter cake. The filter cake thickness is capped at
one-half of the total sediment depth and is assumed to have a minimum thickness of 100 cm.
Equations for the algorithms used to estimate sediment consolidation and the resulting hydraulic
conductivity of the compacted sediment layer can be found in U.S. EPA (1999).
4.2.6.3 Hydraulic Conductivity of Clogged Native Material. The Surface
Impoundment Module simulates the effect of sediment particles that enter and clog the soil layer
immediately underlying the surface impoundment by assuming that the saturated hydraulic
conductivity of the clogged soil zone is one-tenth of the hydraulic conductivity of the native soil.
The model also constrains the hydraulic conductivity of the clogged layer to be less than or equal
the underlying soil layer and greater than or equal to the consolidated sediment filter cake. Based
on observed penetration depths of up to about 0.45 m, the depth of the clogged layer is fixed at
0.5 m.
4.2.6.4 Limitations on Maximum Infiltration Rate. If the surface impoundment
infiltration rate exceeds the rate at which the aquifer can transport ground water, the ground
water level will rise into the vadose zone, and the assumption of zero pressure head at the base of
the vadose zone would be violated. This ground water "mounding" will reduce the effective
infiltration rate to a maximum infiltration rate. The model estimates this maximum rate as one
that does not cause the ground water mound to rise to the bottom elevation of the surface
impoundment unit, using the following equation (U.S. EPA, 1999; HydroGeoLogic, 1999):
2K D ID , - H)
j ^ aqsatr aqsatv vadose '
Max ~ n
Rl In— <4"29'
where
Imix = maximum allowable infiltration rate (m/d)
Kaqsat = hydraulic conductivity of the aquifer (m/d)
Daqsat = aquifer thickness (m)
Dvadose = vadose zone thickness (m)
H = the effective head in the WMU (see Figure 4-3)
R(l = equivalent source radius (m)
R = length between the center of the source and the downgradient boundary
where the boundary location has no perceptible effects on the heads near the
source (m).
Under certain conditions of high soil hydraulic conductivity and long residence time in
the surface impoundment, the leachate flow rate may exceed the influent flow rate. Rather than
reiterating the infiltration rate calculation with liquid depth as a variable, the leachate rate is
limited to 99 percent of the influent flow rate. This limit is based on a volumetric balance on the
WMU and an assumption that the effluent flow is never less than 1 percent of the influent flow.
4-22
-------
Section 4.0
Wastewater Source Modules
4.2.6.5 Surface Impoundment Effluent Flow Rate. The Surface Impoundment
Module calculates effluent flow based on a volumetric water balance on the WMU:
Qout Qinfl Q leach+ ^ rain ^evap) (4-30)
where
Qou, =
effluent flow rate (m3/s)
Qinfl —
influent flow rate (m3/s)
Qleach
leachate infiltration rate (m3/s)
A
surface impoundment surface area (m2)
P
rain
precipitation rate (m/s)
P
evap
evaporation rate (m/s).
Under certain conditions of influent flow rate, impoundment dimensions, infiltration,
precipitation, and evaporation, there may be months that Equation 4-30 predicts a negative or
zero effluent rate, which would violate the pseudo-steady-state assumption. Therefore, if
Equation 4-30 produces an effluent flow rate of less than 1 percent of the influent flow rate, the
effluent flow rate is calculated as
Qout = 001 Qinfl (4-31)
When the infiltration rate is capped at 99 percent of the influent rate, the model uses this
equation only if the evaporation rate exceeds the precipitation rate. Depending on the various
rates, Equation 4-31 can be triggered even if the infiltration rate is not capped at 99 percent of
the influent rate. The Surface Impoundment Module uses Equation 4-31 to limit (or cap) the
evaporation rate to prevent a zero or negative effluent rate.
4.2.7 Adjust for Temperature Effects
Temperature can affect a number of the inputs used by the Wastewater Source Modules,
including air density and diffusivity, biodegradation rate, liquid viscosity, and Henry's law
constant. Some of the equations employed by the modules already include a temperature
correction factor. For example, the liquid-phase, turbulent surface mass transfer coefficient
includes a temperature correction term of 1,024T"20. The modules use the ambient air
temperature to adjust the air-side properties (air diffusivity, air density, etc.). Liquid-side
properties (liquid diffusivity, liquid viscosity, etc.) are adjusted using the wastewater
temperature within the tank or surface impoundment.
4.2.7.1 Estimating Temperature in the Waste Management Unit. The Wastewater
Source Modules use a simplified energy balance around the aerated tank or surface
impoundment to estimate the liquid temperature in the WMU from the liquid temperature of the
influent, the monthly ambient air temperature, and the liquid residence time in the WMU. The
model uses coefficients for free and forced convective heat transfer coefficients for both water
4-23
-------
Section 4.0
Wastewater Source Modules
and air (Kreith and Black, 1980) to estimate the average overall heat transfer coefficient. The
model assumes that there is forced convection on the air side (windspeed greater than 0 m/s),
free convection on the quiescent liquid side, and forced convection on the turbulent liquid side.
Assumptions and equations used to estimate tank and surface impoundment temperature are
detailed in U.S. EPA (1999).
The surface impoundment and aerated tank temperature estimates do not take into
account the heat of fusion from ice formation, and can yield liquid temperatures below 0°C.
When this happens, the models set the liquid temperature to 0.1°C and estimate the amount of
ice formed based on the specific heat capacity and density of water and ice. The models translate
the additional heat loss in taking the water from 0°C to below 0 into a mass of ice formation, and
estimate the volume or depth of ice formed using the following equation:
dice = depth of ice layer formed (m)
A = area of the unit (m2)
T, = temperature of liquid (C).
This equation tends to overestimate ice formation because convective heat transfer from
the surrounding soil is not included in the heat balance. Also, although a small amount of ice
formation will not significantly impact the emission estimates and other parameters estimated by
the module, if a solid crust of ice forms over the entire impoundment for a prolonged period of
time, the model would overstate the potential for volatile emissions because it does not consider
volatilization through an ice layer. Therefore, when the depth of the projected ice layer is 10 cm
or more for 3 consecutive months, the model generates a warning message that significant ice
formation is projected.
4.2.7.2 Temperature Effects on Air-Side Properties. Air-side properties include
density and viscosity. The model estimates air density at a given temperature using the ideal gas
law (U.S. EPA, 1999). Because the viscosity of air is only slightly affected by temperatures in
the temperature range of interest, with a range from 1.75><10"4 to 2.17><10"4 g/cm-s from 0°C to
100°C (Kreith and Black, 1980), it is not adjusted for temperature in the Wastewater Source
Modules.
4.2.7.3 Temperature Effects on Liquid-Side Properties. The density of water is
basically insensitive to temperature and no temperature adjustments are used in the Wastewater
Source Modules. The viscosity of water varies by more than a factor of 5 over the temperature
range of interest (0°C to 100°C). This temperature dependency is important not only for mass
transport, but also for its effect on the solids settling rate (terminal velocity) at lower Reynolds
numbers. Using the data from Kreith and Black (1980), the modules use the a correlation
developed using a log-log least squares linear regression to adjust the viscosity of water (U.S.
A x d x (- 7\)
i wmu \ V
ice = OA /A m ,
80 (0.9) A
(4-32)
where
4-24
-------
Section 4.0
Wastewater Source Modules
EPA, 1999). The values for the viscosity of water calculated from this equation agree well with
the values reported by Liley and Gambill (1973) for temperatures between 0°C and 100°C.
4.2.7.4 Temperature Effects on Constituent-specific Properties. Air diffusivity,
water diffusivity, Henry's law constant, and aerobic and anaerobic biodegradation rates are
constituent-specific properties used by the Wastewater Source Modules to estimate constituent
volatilization, degradation, and release to air and, for the surface impoundment, through
leachate. In the 3MRA modeling system, these properties are supplied by the Chemical
Properties Processor (CPP). Because the Wastewater Source Modules operate on monthly time
steps, they call the CPP for these properties according to the temperature in the unit at the
beginning of each month. Details on the temperature correction routines used by the CPP can be
found in the documentation for the 3MRA CPP (Pacific Northwest National Laboratory, 1998).
4.2.7.5 Temperature Effects on BOD and Sediment Biodegradation Rates. The
Wastewater Source Modules assume that the BOD and sediment decay rates (kb, kdec) are
relatively unaffected by temperature over a reasonably wide range of temperature. At
temperatures above 50°C and at temperatures near freezing, the decay rate is assumed to drop
rapidly. The modules incorporate a simple temperature correction factor for these decay rates
based on these assumptions: between 7°C and 40°C, the biodegradation rate temperature
correction factor is assumed to be 1 (i.e., no correction). If temperatures fall below 3°C or above
60°C, the model stops biodegradation of BOD and the sediment mass by setting the temperature
correction factor to 0. A linear extrapolation is used to determine the temperature correction
factor between 3°C and 7°C and between 40°C and 60°C.
4.3 Module Discussion
4.3.1 Strengths and Advantages
The Wastewater Source Modules are based on sound engineering principles and
algorithms, many of which have been tested and peer reviewed. Some of the strengths and
advantages of these modules include the following:
¦ Volatilization model. The volatilization component of the Wastewater Source
Modules employs the mass transfer correlations recommended by the Office of Air
Quality Planning and Standards (OAQPS) as developed though the CHEMDAT and
WATER series of models. These correlations have been developed, tested, and
validated through many years of EPA research and public and industry use.
¦ Leachate infiltration model. The leachate infiltration flow rate component of the
Surface Impoundment Module was adapted from the equations and algorithms
developed by the Office of Research and Development (ORD) in the EPACMTP
model, which has undergone peer review by SAB. The infiltration rate component
also considers the effect of sediment consolidation within the impoundment and
sediment impregnation (or clogging) of the underlying soil layer on the infiltration
rate. These components were recommended by ORD based on their on-going
research and development efforts.
4-25
-------
Section 4.0
Wastewater Source Modules
¦ Solids mass balance. The modules account for sorption using correlations for Kd
and Koc developed jointly by OSW and ORD. The modules also perform an explicit
mass balance on the solids within the unit, including production of new solids
through biomass growth, transport of solids between compartments due to
infiltration (for the Surface Impoundment Module), sedimentation, resuspension and
burial, and decay of sediment due to anaerobic digestion. Therefore, the modules
evaluate total, dissolved, and sorbed phase concentrations needed for various
potential exposure mechanisms.
¦ Loss mechanisms. The modules account for a variety of loss mechanisms,
including volatilization, sorption, hydrolysis, biodegradation, and infiltration (for the
Surface Impoundment Module). Furthermore, the modules account for changes in
the rates of these loss mechanisms due to monthly changes in ambient temperature,
the predicted effect of these ambient temperature changes on the average wastewater
temperature, and monthly evaporation and precipitation rates. Therefore, the
modules can elucidate differences in constituent fate through seasonal variations.
Additionally, the modules account for the accumulation and consolidation of
sediment within the units, and thus can be used to predict the transient nature of the
constituent fate as a unit fills with sediment (thereby reducing the hydraulic retention
time and perhaps the sediment removal efficiency).
¦ Site-specific data. The modules have a numerous input variables that allow
modeling of site-specific units using detailed site-specific information. Without this
site-specific data, the modules can be used with Monte-Carlo-derived inputs to
derive a reasonable expectation of the expected fate of the constituent and the
extremes of the potential exposure concentrations. The modules employ
predominantly analytical solutions so that numerous module runs, such as those
needed when performing a Monte Carlo analysis, can be completed in a short time
frame.
¦ Plug-flow and batch operations. The Surface Impoundment Module provides
solution algorithms for modeling plug-flow and batch operations in addition to the
well-mixed solution algorithm. The Aerated Tank Module was designed specifically
for aerated or mixed tanks, and therefore assumes a well-mixed liquid compartment.
However, non-aerated, quiescent tanks that operate in a plug-flow or batch mode can
be effectively be modeled using the Surface Impoundment Module by setting a very
low sediment hydraulic conductivity (driving the infiltration rate essentially to zero).
4.3.2 Uncertainty and Limitations
The most significant uncertainties and limitations of the Wastewater Source Modules
include the following:
¦ Applicable only to dilute aqueous wastes. By using a simple biodegradation rate
model together with Henry's law partitioning coefficients, the modules are most
applicable to dilute aqueous wastes. High constituent concentrations can reduce or
inhibit biodegradation of toxic constituents. Also, if constituents exceed solubility
4-26
-------
Section 4.0
Wastewater Source Modules
limits and free phase is present, Henry's law can overestimate volatilization (the
modules output a warning when this occurs).
¦ Reaction byproducts not considered. Daughter products from hydrolysis and
biological degradation of the constituents are not addressed by the modules (i.e.,
volatile emissions or leaching are not considered for reaction intermediates or end
products).
¦ No oxygen balance. The modules assume that the liquid compartment is aerobic
and contains adequate oxygen for the degradation of influent carbon (as BOD).
However, high influent carbon loadings for certain low-aeration surface
impoundments can result in anoxic (low oxygen) conditions that can limit biological
degradation. In such cases, the modules will tend to overestimate constituent and
BOD removal through biodegradation and sorption, and underestimate volatile
emissions and leachate flux.
¦ No delineation of organic solids. The modules maintain suspended solids
characteristics in the influent throughout the simulation. However, as influent carbon
loading converts to biomass, the characteristics of the suspended solids within the
WMU could change. For example, the fraction organic carbon, average particle
density, and fraction of biologically active solids can change dramatically from
influent solid values within a biologically active unit. Such changes can affect the
solids balance (e.g., the sedimentation rate) as well as the partitioning of organic
chemicals. This limitation creates uncertainty in the output variables for WMUs with
high biomass generation rates relative to the influent solids loading. For units with
relatively high solids loading rates and low biomass generation rates, changes in
suspended solid particle characteristics due to biological growth would be limited.
¦ No spatial variations in sediment depths. For all units (well-mixed, plug-flow or
batch; aerated or nonaerated), the modules assume that sediment compartment
depths are uniform throughout the impoundment. For plug-flow systems, it is likely
that the sediment will accumulate fastest near the influent point. For aerated units,
sediment depths directly beneath high-speed aerators are expected to be less than in
less agitated parts of the WMU. Because including areas of different sediment depth
would greatly complicate the model construct and solution, the even distribution of
sediment is a reasonable simplifying assumption. However, the modules could
overestimate leachate flux for plug-flow units that have significant sediment
accumulation near the influent (i.e., lower infiltration rates where the concentrations
are the highest), and could underestimate leachate flux for units that have low
sediment accumulation areas below mechanical aerators.
4.4 References
Bird, R.B., W.E. Stewart, and E.N. Lightfoot. 1960. Transport Phenomena. John Wiley and
Sons, Inc., New York, NY. pp. 190 through 196.
Bryant, C.W. 1985. Lagoons, ponds, and aerobic digestion. Journal WPCF 57(6): 531-533.
4-27
-------
Section 4.0
Wastewater Source Modules
Eckenfelder, W.W., M.C. Goronszy, and T.P. Quirk, circa 1984. The activated sludge process:
state of the art. Chapter in: CRC Critical Reviews in Environmental Control. Volume
15, Issue 2. CRC Press, Inc., Boca Raton, FL. pp. Ill through 177.
Gaudy, A.F., Jr., and Kincannon, D.F. 1977. Comparing design models for activated sludge.
Water and Sewage Works, February 1977, 66-70.
Goldsmith, C.D., and Balderson, R.K. 1989. Biokinetic constants of a mixed microbial culture
with model diesel fuel. Hazadous Waste and Hazardous Materials 6(2): 145-154.
Hermann, J. D., and Jeris, J. 1992. Estimating parameters for activated sludge plants. Pollution
Engineering 24(21):56-60.
HydroGeoLogic. 1999. Additional Components in the HWIR99 Surface Impoundment Module.
Prepared for the Office of Solid Waste, U.S. Environmental Protection Agency,
Washington, DC. EPA Contract No. 68-W7-0035. Herndon, VA.
Kreith, F., and W.Z. Black. 1980. Basic Heat Transfer. Harper & Row Publishers, New York,
NY. pp. 15, 514, and 520.
Liley, P. E., and W.R. Gambill. 1973. Chapter 3: Physical and chemical data. In. Perry's
Chemical Engineers' Handbook, 5th Edition, Robert H. Perry and Cecil H. Chilton (eds.).
McGraw-Hill, Inc., New York, NY. Pp.3-1.
Millington, R.J., and J.P. Quirk. 1961. Permeability of porous solids. Transactions of the
Faraday Society. 57(7): 1200-1207. July.
Pacific Northwest National Laboratory. 1998. Documentation for the FRAMES-Technology
Software HWIR System, Volume 13: Chemical Properties Processor. Prepared for the
Office of Research and Development and Office of Solid Waste, U.S. Environmental
Protection Agency, Washington, DC. EPA Contract No. DE-AC06-76RLO 1830.
(PNNL-11914, Vol. 13).
Rozich, A.F., A.F. Gaudy, Jr., and P.C. D'Adamo. 1985. Selection of growth rate model for
activated sludges treating phenol. Water Res. 19(4):481-490.
Tabak, H.H., S. Desai, and R. Govind. 1989. The determination of biodegradability and
biodegradation kinetics of organic pollutant compounds with the use of electrolytic
respirometry. Presented at the 15th Annual Research Symposium: Remedial Action,
Treatement, and Disposal of Hazardous Waste, April 10-12, 1989, Cincinnati, Ohio.
U.S. EPA (Environmental Protection Agency). 1994. Air Emissions Models for Waste and
Wastewater. EPA-453/R-94-080A. Office of Air Quality Planning and Standards,
Research Triangle Park, NC.
4-28
-------
Section 4.0
Wastewater Source Modules
U.S. EPA (Environmental Protection Agency). 1996. EPA 's Composite Model for Le achate
Migration with Transformation Products. EMCMTP. Background Document. Office of
Solid Waste, Washington, DC. pp. 1-7 through 1-12.
U.S. EPA (Environmental Protection Agency). 1999. Source Modules for Tanks and Surface
Impoundments: Background and Implementation for the Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) for HWIR99. Office of Solid Waste,
Washington, DC. October.
Van Genuchten, M. Th. 1980. A closed-form equation for predicting the hydraulic conductivity
of unsaturated soils. Soil Sci. Soc. J. 44:892-898.
Weber, A.S., E.K. Russell, J.E., Alleman, J.H. Sherrard, R.O. Mines, and M.S. Kennedy. 1985.
Activate sludge. Journal WPCF 57(6): 517-526.
4-29
-------
Section 4.0 Wastewater Source Modules
4-30
-------
Section 5.0
Land-based Source Modules
5.0 Land-based Source Modules
5.1 Purpose and Scope
The Land-based Source Modules simulate partitioning and emission of constituents from
land-based waste management units (WMUs). Three Land-based Source Modules were
developed for the 3MRA modeling system to represent the major management practices where
wastes are put into or on the land for recycling, recovery, reuse, treatment, or disposal. These
modules simulate waste management practices in the following types of WMUs:
¦ Landfills, which are a common disposal unit for many nonliquid industrial
wastes;
¦ Waste piles, which are temporary storage areas on the ground for nonliquid
industrial waste, such as ash or slag; and
¦ Land application units (LAUs), which are used to reuse, treat, or dispose of
industrial waste in liquid, semiliquid, or solid form. Some wastes are used as a
soil amendment, which is a reuse practice; some wastes are applied to land for
treatment through biological degradation; and some wastes are applied to land as
a disposal method.
The three Land-based Source Modules were designed to provide estimates of annual
average constituent concentrations in surface soil and constituent mass emission rates to air,
surface water, and ground water, and to maintain mass balance between the source and all
release routes. The emission rate estimates are then used in the 3MRA modeling system, which
links source modules with environmental fate and transport modules. Figure 5-1 shows the
relationship and information flow among the Land-based Source Modules in the 3MRA
modeling system.
Each of the three Land-based Source Modules provides some similar and some different
features in terms of the ways constituents of concern can be released to the environment. All
three modules have the potential to release constituents to the air by volatilization or particle
entrainment, and to the subsurface and ground water by leaching. Waste piles, because they are
elevated and have more surface area exposed, are assumed to have a greater potential for air
emissions than the other two source types. Waste piles and LAUs have the additional release
mechanism of erosion and runoff to the surrounding watershed and the nearest stream or other
waterbody. Landfills are assumed to be below grade and, thus, do not release constituents
through erosion or runoff.
5-1
-------
Section 5.0
Land-based Source Modules
Key Data Inputs
• Precipitation rate
• Waste fraction
• Depth to aquifer
• Kd
Emission Rates
Air Module
Vadose Zone and
Aquifer Modules
Chemical Fluxes
Infiltration Rates
Surface Water
Module
Chemical Loadings, Soil Loadings
(LAU, WP only)
Landfill,
Waste Pile,
LAU
Modules
Farm Food Chain
Module
Soil Concentrations
Terrestrial Food
Web Module
Soil Concentrations
Soil Concentrations
Human Exposure
Module
(LAU, WP only)
Ecological
Exposure Module
Soil Concentrations
(LAU, WP only)
Figure 5-1. Information flow for the Land-based Source Modules
in the 3MRA modeling system.
The Land-based Source Modules contain the following three models:
1. The Generic Soil Column Model (GSCM) was developed to describe the
constituent fate and transport in a porous medium, such as soil or waste. It
provides the vertical concentration profile in the soil/waste column at various
times. The GSCM provides the following outputs:
- Annual average constituent concentrations in surficial soil (top 1 cm), as
well as depth-averaged concentrations over greater depths in the WMU
- Annual average constituent concentrations in surficial soils, as well as
depth-averaged concentrations over greater depth in the buffer zone
located downslope from the WMU (LAU and waste pile only)
5-2
-------
Section 5.0
Land-based Source Modules
- Annual average constituent fluxes in leachate draining from the WMU
into the vadose zone
- Annual average leachate flow rates (infiltration rates) draining from the
WMU into the vadose zone
- Annual average constituent fluxes volatilizing from the WMU surface into
the atmosphere.
2. The Local Watershed Model (for LAUs and waste piles) is based on mass
balances of solids and constituents in the runoff and the top layer of the soil
column modeled by the GSCM. The "local watershed" comprises the land area
between the waterbody and the top of the hillside containing the WMU and may
include an upslope area, the WMU, and a downslope or buffer area between the
WMU and the waterbody. The Local Watershed Model provides the following
outputs:
- Annual average constituent loadings in surface runoff and erosion that
enter the nearest surface waterbody downslope of the WMU
- Annual average eroded soil loadings in surface water runoff that enter the
nearest surface waterbody downslope of the WMU
- Annual average runoff that enters the nearest surface waterbody
downslope of the WMU.
3. The Particulate Emissions Model was designed to provide estimates of the
annual average emission rate of constituent mass adsorbed to particulate matter
less than 30 |im in diameter. The release mechanisms considered differ for each
WMU, but may include wind erosion, vehicular activity, unloading operations,
tilling, and spreading/compacting operations. The Particulate Emissions Model
provides the following outputs:
- Annual average particulate constituent fluxes emitted into the atmosphere
from the WMU surface sorbed to soil particles via wind erosion or other
surface disturbances
- Annual average particulate fluxes (the particles themselves) emitted into
the atmosphere from the WMU surface via wind erosion or other surface
disturbances.
- Particle size distributions for airborne soil particles.
5-3
-------
Section 5.0
Land-based Source Modules
Figure 5-2 illustrates how these three models interact within the Land-based Source
Modules. The GSCM models vertical movement of constituents within and from the waste/soil
(with the exception of particulate emissions, which are modeled by the Particulate Emissions
Model). The Local Watershed Model simulates lateral movement and mass balance between
waste/soil and runoff. In addition to the constituent fluxes indicated in the figure, various media
fluxes are also calculated (e.g., leachate flux, by the GSCM; eroded soil load, by the Local
Watershed Model; and particle size distribution, by the Particulate Emissions Model). The three
models interact to maintain mass balance between the compartments. The GSCM calculates the
depth-averaged soil concentration in the buffer zone for several additional layers below the
surficial layer only in order to maintain mass balance between the surficial layer and the deeper
layers. The deeper layers represent a sink, and the soil concentrations in those layers are not
used. Similarly, the GSCM calculates volatilization from the buffer zone only to maintain mass
balance—the volatilization flux from the buffer zone is not used in the Air Module. The
omission of volatilization and leachate flux from the buffer zone from subsequent modeling
results in an underestimation of air and ground water concentrations. However, these emissions
likely have a less significant impact on the results than do direct releases from the WMU.
Upslope
WMU
Downslope
r
Air
Volatilization
(GSCM)
Particulate
Emissions
Volatilization
(GSCM) (Sink)
(PEM)
Surficial Waste/
Soil Layers
Runoff Layer
Waste/
Soil Layers
Runoff Cone (LWS)
Runoff (LWS)
Soil Erosion (LWS)
Soil Cone (GSCM) Soil Cone (GSCM)
Soil Cone (GSCM)
Soil Cone (GSCM)
(Sink)
Surface
Water
Soil Cone (GSCM)
Vadose
Zone
Leaching
(GSCM)
Figure 5-2. Interaction of the models that form the
Land-based Source Modules.
5-4
-------
Section 5.0
Land-based Source Modules
5.2 Conceptual Approach
The physical processes modeled differ among the three land-based sources. As a result,
not all of the models listed above are used for all three of the WMUs. This section first describes
the three different WMUs and the processes modeled for each and then describes the three
models that are contained in the Land-based Source Modules in greater detail.
5.2.1 Description of WMUs
Landfills. Figure 5-3 shows the physical processes being modeled in the 3MRA
modeling system for landfill operations. The Landfill Module was developed to approximate the
effects of the gradual filling of active landfills. The landfill is divided into vertical cells of equal
volume running from the site surface to the bottom of the landfill, each sized so that they require
1 year to fill. Waste mass is added gradually, forming layers of waste. After 1 year, the cell is
full and the waste may be covered with a clean soil cover (this is optional). This process is
repeated with the next cell and continues until the landfill reaches maximum capacity. The
landfill can be sized to operate for a specified number of years (such as a 20-year operational
life). Releases of a constituent to the air or ground water are modeled for up to 200 years or until
the concentration in the landfill is 1 percent of the maximum concentration during the operating
life of the landfill.
Volatilization
Volatilization
*
lization
Yd
Cover soil
'articulati
emission
Bio/chemicl
degradation
Leaching
Liner soil
Leaching
Leaching
Figure 5-3. Illustration of landfill
(shown with six cells and three waste layers).
The following assumptions were made in developing the Landfill Module:
¦ The empty landfill is approximated as an excavated volumetric rectangle. The
volume is assumed to be completely below grade; therefore, it is assumed that no
constituent mass is lost due to runoff or erosion.
5-5
-------
Section 5.0
Land-based Source Modules
¦ Each landfill cell is approximated as a soil column consisting of three
homogeneous zones: soil cover, landfill waste, and subsoil. Each zone is
modeled as a homogeneous porous medium whose properties are uniform in
space and time within the zone. The soil cover zone and subsoil zones are
optional.
¦ The concentration of constituents is constant in the incoming waste, and waste is
added at a constant rate. The concentration of constituents in the waste can be
adjusted to account for other wastes entering the landfill that do not contain the
constituent of interest.
¦ There is no lateral transport of constituents between cells.
¦ Waste is added to the landfill cell in layers. A waste layer, for the purposes of the
model, is simply a zone within which initial concentrations are assumed to be
uniform. Waste layers are conceptualized as being formed over time by the
dumping of loads of waste next to one another in the landfill cell until eventually
a waste layer of uniform depth is formed. At this point, a new layer is started.
¦ At the start of the landfill cell simulation, one waste layer is assumed to be
present. After each time period, another waste layer is laid down until the landfill
cell is full. At that time, it may be covered with clean soil or left exposed.
¦ The first-order constituent and biological loss processes modeled for the entire
landfill (including cover soil, waste, and liner material) are anaerobic
biodegradation and hydrolysis.
¦ The first-order loss rate due to particulate emissions from an active landfill cell
includes losses due to wind erosion, vehicular activity, and spreading and
compacting. Only losses due to wind erosion are modeled from inactive landfill
cells.
Waste Piles. Figure 5-4 shows the physical processes modeled in the 3MRA modeling
system for waste piles, which may manage ash, slag, or other similar types of waste. The waste
pile has a constant height and constant area equal to the footprint of the waste pile. At the start,
the waste pile is filled to capacity. After each period of time, such as 1 year, the entire waste pile
is removed and instantaneously refreshed (i.e., replaced with fresh waste). In reality, waste is
added and removed from a waste pile incrementally; the assumption that the waste is
instantaneously refreshed is made to simplify modeling. Because the waste surface is not
refreshed between placements of new waste, the Waste Pile Module may underestimate volatile
losses relative to a model where waste is built up gradually. This is unlikely to affect emission
estimates for nonvolatile constituents.
5-6
-------
Section 5.0
Land-based Source Modules
Volatilization
irticulate emissions
Bio/chemical
degradation
Volatilization
Leaching
Leaching
Figure 5-4. Illustration of a waste pile in local watershed.
The following assumptions were made in developing the Waste Pile Module:
¦ The waste pile site is used for a finite number of years, after which it is assumed
that the final waste pile is removed and clean soil is placed on the surface where
the waste pile was.
¦ The first-order constituent and biological loss processes modeled for the waste
pile are aerobic biodegradation and hydrolysis in the surface waste layer, and
anaerobic biodegradation and hydrolysis in all subsurface layers of the waste pile.
¦ The first-order loss rate due to particulate emissions from an active waste pile is
applied only to the surface layer and includes losses due to wind erosion,
vehicular activity, and spreading and compacting operations from an active waste
pile.
¦ For purposes of modeling runoff and erosion processes, the following
assumptions were made:
- The waste pile is conceptualized as having side slopes (from which
increased runoff and erosion would occur) that are insignificant (i.e., the
sloped surface area is small relative to the surface area of the top of the
waste pile).
- The top surface of the waste pile has the same slope as the average slope
of the local watershed.
- No run-on occurs to the waste pile subarea from upslope subareas in the
local watershed.
5-7
-------
Section 5.0
Land-based Source Modules
- Following removal of the waste pile, there is no subsequent runoff/erosion
transport pathway from the waste pile subarea of the local watershed
downslope to the surface water, only from the buffer subarea. That is,
there is assumed to be no remaining surficial contamination in the subarea
that contained the waste pile because clean soil is assumed to be placed
there after waste pile removal.
¦ The topmost layer of waste serves as the "soil" compartment in the runoff and
erosion algorithms used in the local watershed model. For the purposes of
applying these algorithms, it is assumed that the surface layer is 0.01 m deep.
Land Application Unit. Figure 5-5 shows the physical processes being modeled in the
3MRA system for LAUs. Such units are used to manage liquid, semi-solid, and solid wastes.
Certain types of waste can be used to amend agricultural soils, thus constituting a reuse of the
waste.
Volatilization
¦articulate
5io/chemical
legradation
Leaching
emissions
Volatilization
Bio/chemical
Figure 5-5. Illustration of LAU in local watershed.
The following assumptions were made in developing the LAU Module:
¦ Waste is applied to the soil surface periodically at even intervals (e.g., quarterly)
and then tilled into the soil to a specified depth, such as 0.2 m.
¦ The till zone is completely mixed upon each application of waste to soil.
¦ The modeled surface layer consists of one homogeneous zone—the till zone—
which consists of a soil/waste mixture. The till zone properties can be estimated
5-8
-------
Section 5.0
Land-based Source Modules
as the depth-weighted average of the soil and waste properties according to the
depth of soil and waste in the till zone.
¦ The water contained in the wet waste added to the LAU increases the annual
average infiltration rate.
¦ The constituent mass is concentrated in the solids portion of the waste and is re-
partitioned among the solid, aqueous, and gas phases in the soil column.
¦ Waste applications do not result in significant buildup of the soil surface, nor does
erosion significantly degrade the soil surface (i.e., the vertical distance from the
site surface). As a result, there is no naturally occurring limit to the modeled
concentration of constituents in the soil other than the limit for nonaqueous-phase
liquids (NAPLs). As a result, the modeled concentration in the till zone could
exceed the concentration in the waste. This is physically possible for highly
immobile constituents if the waste matrix is organic and decomposes, leaving
behind the constituent to concentrate over multiple applications.
¦ The LAU is operated for a specified number of years (e.g., 40 years).
¦ The first-order constituent and biological loss processes in the till zone include
aerobic biodegradation and hydrolysis.
¦ The first-order loss rate due to particulate emissions from an active LAU is
applied to the surface layer of the till zone only and includes losses due to wind
erosion, vehicular activity on the surface of the LAU, and tilling operations. The
particulate emission loss rate from an inactive LAU includes wind erosion only.
¦ The topmost waste/soil layer serves as the soil compartment in the runoff and
erosion algorithms used in the Local Watershed Model. For the purposes of
applying these algorithms, it is assumed that the surficial soil layer is 0.01 m
deep.
5.2.2 The Generic Soil Column Model
The GSCM was developed to describe the dynamics of constituent mass fate and
transport within land-based WMUs and near-surface soils in watersheds. Figure 5-6 illustrates
the processes modeled by the GSCM. The GSCM solves the 1-D partial differential equation
describing constituent fate and transport in a porous medium in space and time. This solution
represents the vertical concentration profile in the soil/waste column at various times, where the
soil/waste column can represent either the top portion of the vadose zone of a land application
site (LAU Module), the waste pile (Waste Pile Module), a landfill cell (Landfill Module), or the
top portion of the vadose zone of an entire watershed subbasin (Watershed Module). In addition
to describing vertical constituent concentration gradients in the soil column, the GSCM is
coupled with a runoff/erosion model (either the Local Watershed Model, described in
Section 5.2.3, or the Watershed Module, described in Section 7) that describes constituent
5-9
-------
Section 5.0
Land-based Source Modules
Governing equation:
dCT JCT dCT
— = A-— -
Volatilization
Waste loadings
Particulate emissions
Erosion/runoff
Erosion/run-on
Infill
! )ecay, hyd
rolysis
Solid/i
partiti
queou«7 gaseous
jniug
Leachate
Figure 5-6. Conceptual diagram of the Generic Soil
Column Model (GSCM).
concentration in surface runoff across the soil column's surface and interactions between this
runoff compartment and the soil compartment.
The GSCM was originally designed for the 3MRA modeling system to make some
modest improvements to the well-known Jury model (Jury et al., 1983, 1990), which had been
used historically to simulate constituent fate and transport in land-based sources. Similar to the
GSCM, the Jury model describes vertical concentration gradients over time of total (sorbed,
dissolved, and gaseous) constituent concentration. The Jury model was limited for 3MRA
purposes for two reasons. First, it assumes an infinitely deep soil column. The 3MRA Land-
based Source Modules output leachate fluxes to the Vadose Zone Module, so the infinite depth
assumption is inappropriate. Second, the Jury model's initial conditions cannot accommodate
the vertical gradient resulting from periodic waste loadings, such as from repeated land
application of waste, over all or a portion of the soil column depth. The governing equations for
the GSCM are similar to those used by Jury et al. (1983, 1990) and Shan and Stephens (1995),
but have been applied in a manner that avoids these limitations. Use of the GSCM allows
¦ Constituent mass balance;
¦ Waste additions and removals to simulate active facilities; and
5-10
-------
Section 5.0
Land-based Source Modules
¦ Joint estimation of
- Volatilization of gas-phase constituent mass from the surface to the air,
- Leaching of aqueous-phase constituent mass by advection or diffusion
from the bottom of the WMU or vadose zone,
- Concentrations of constituents within the soil or waste column, and
- First-order losses, which can include hydrolysis and anaerobic and aerobic
biodegradation.
The following assumptions were made in the development of the GSCM for use in all of
the Land-based Source Modules:
¦ The constituent partitions to three phases: adsorbed (solid), dissolved (liquid), and
gaseous (as in Jury et al., 1983, 1990). The governing equation is
cT=Pbcs + Q + QaQ? (5-i)
where
Cx = total constituent concentration in soil (g/m3 of soil)
pb = soil dry bulk density (g/cm3)
Cs = adsorbed-phase constituent concentration in soil (jug/g of dry soil)
0W = soil volumetric water content (m3 soil water/m3 soil)
CL = aqueous-phase constituent concentration in soil (g/m3 of soil water)
0a = soil volumetric air content (m3 soil air/m3 soil)
CG = gas-phase constituent concentration in soil (g/m3 of soil air).
¦ The constituent undergoes reversible, linear equilibrium partitioning between the
adsorbed and dissolved phases (as in Jury et al., 1983, 1990):
Cs = Kdx CL (5-2)
where
Kd = linear equilibrium partitioning coefficient (cm3/g).
For organic constituents,
Kd=f.y-K„ (5-3)
where
foc = organic carbon fraction in soil or waste (unitless)
Koc = equilibrium partition coefficient, normalized to organic carbon (cm3/g).
5-11
-------
Section 5.0
Land-based Source Modules
Alternatively, Kd can be specified as an input parameter for inorganic
constituents.
It is implicit in this linear equilibrium partitioning assumption that the sorptive
capacity of the soil column solids is considered to be infinite with respect to the
total mass of constituent over the duration of the simulation, i.e., the soil column
sorptive capacity does not become exhausted.
Constituents in the dissolved and gaseous phases are assumed to be in equilibrium
and to follow Henry's law (as in Jury et al., 1983, 1990):
CG= H'x CL (5-4)
where
H' = dimensionless Henry's law coefficient.
The Henry's law coefficient is adjusted for seasonal variations in temperature.
The total constituent concentration in soil can also be expressed in units of mass
of constituent per mass of dry soil (|ig/g) by dividing by the soil dry bulk density.
Using the linear equilibrium approximations in Equations 5-2 through 5-4, Cx can
be expressed in terms of CL, Cs, or CG as follows:
CT=KTLxCL=^xCs=^-xC0 (5-5)
Kd H
where
Kti = dimensionless equilibrium distribution coefficient between the total and
aqueous-phase constituent concentrations in soil.
The total water flux or infiltration rate is constant in space and time (as in Jury et
al., 1983, 1990) and greater than or equal to zero. It is specified as an annual
average.
Material in the soil column (including bulk waste) can be approximated as
unconsolidated homogeneous porous media (i.e., one whose basic properties,
which include dry bulk density, fraction organic carbon, soil volumetric water
content, soil volumetric air content, and total soil porosity, are uniform in space).
Waste/soil properties are specified as annual average values.
Constituent mass may be lost from the soil column through one or more first-
order loss processes.
The total constituent flux is the sum of the vapor flux and the flux of the dissolved
solute (as in Jury et al., 1983, 1990).
5-12
-------
Section 5.0
Land-based Source Modules
¦ The constituent is transported in one dimension through the soil column (as in
Jury etal., 1983, 1990).
¦ The effective diffusivity in soil may be calculated by the model of Millington and
Quirk (1961) (as in Jury et al., 1983, 1990).
¦ The modeled spatial domain of the soil column remains constant in volume and
fixed in space with respect to a vertical reference, such as the water table.
Under the above assumptions, the governing mass fate and transport equation for the
GSCM can be written as follows:
dCT d2CT dCT
—- = De—y~ - VE—^ - kCT (5-6)
dt E dz1 E dz T y '
where
t = time
k = total first-order loss rate (1/d)
De = effective diffusivity in soil (m2/d)
z = verticle direction (m), depth
VE = effective solute convection velocity (m/d).
De can be considered to be the algebraic sum of the effective gaseous and water diffusion
coefficients in soil. VE is equal to the water flux corrected for the constituent partitioning to the
solid phase (m/d).
A solution of the complete convective-diffusive-decay concentration model
(Equation 5-6) was undertaken to evaluate, in a soil/waste column,
¦ Total constituent concentration as a function of time and depth below the surface,
and
¦ Constituent mass fluxes across the upper and lower boundaries of the soil column.
A quasi-analytical approach was developed that allows for relative computational speed
and significantly reduces concern about numerical diffusion and lack of stability associated with
fully numerical solution techniques. The tradeoff is a loss of ability to evaluate short-term trends
in concentration and diffusive flux profiles. The method was developed to allow estimation of
long-term (i.e., annual average) constituent concentration profiles and mass fluxes.
The solution for the simplified case, in which the soil column consists of one
homogeneous zone whose properties are uniform in space and time, is described below.
Adaptations of the solution technique to account for variations from this simplified case (e.g.,
more than one homogeneous zone, as for a landfill with cover soil zone atop the waste zone) are
5-13
-------
Section 5.0
Land-based Source Modules
described in the module-specific sections of the 1999 background documents at
www.epa.gov/epaoswer/hazwaste/id/hwirwste/risk.htm.
The quasi-analytical approach is a step-wise solution of the three components of the
governing equation (Equation 5-6) on the same grid. That is, the following equations are solved
individually and then added (under the principle of superposition for linear differential
equations) to yield the complete solution:
dCT <92Ct , N
—^=de—^ (5-7)
a E dz2 v '
dCT
<5C1
dt ^'E dz
(5-8)
dCT
—r~=-kCT
a T
(5-9)
Equations 5-7 through 5-9 each have an analytical solution that can be combined to
obtain a pure diffusion solution that moves with velocity VE through the porous medium (Jost,
1960). The solution of the general differential equation then has the form of the solution of the
diffusive portion with its time dependence, translating in space with velocity VE, and decaying
exponentially with time.
The following boundary conditions are assumed:
¦ Zero concentration is assumed at the upper boundary of the soil column.1 This is
consistent with the assumption that the air is a sink for volatilized constituent
mass, but requires the approximate method for estimating the mass flux of volatile
emissions.
At the lower boundary of the soil column, the flexibility exists with this solution
technique to specify a value between 0 and 1 for the ratio of the total constituent
concentration in the soil directly below the modeled soil column and in the soil
column. A ratio of 1 corresponds to a zero gradient boundary condition. A ratio
of 0 corresponds to a zero concentration boundary condition. For the 3MRA
modeling system, a boundary condition of 0 was assumed.
1 The Cp = 0 "boundary condition" is not a boundary condition in the usual sense of it determining the
functional form of the resulting analytical solution to the underlying diffusion differential equation (Equation 5-7).
That solution is based on an implicit boundary condition of Cp = 0 at +/- infinity, similar to heat transfer in an
infinitely long plate. The CT = 0 boundary condition is here intended to reflect the fact that there is no back-
diffusion from the overlying air column into the soil column, i.e., the concentration in the overlying air column is
assumed to be zero so that no back-diffusion results.
5-14
-------
Section 5.0
Land-based Source Modules
5.2.3 Local Watershed Model
The Local Watershed Model is a holistic model used in the Waste Pile and LAU Modules
to incorporate them into the watershed of which they are a part. These WMUs are on the land
surface, so they are integral land area components of their respective watersheds and,
consequently, are not only affected by runoff and erosion from upslope land areas, but also affect
downslope land areas through runoff and erosion. After some period of operation during which
runoff and erosion have occurred from these WMUs, the downslope land areas will have been
contaminated and their surface concentrations could approach the residual constituent
concentrations in the WMU itself (or conceivably even exceed them long after the WMU ceases
operation). Thus, after extensive runoff and erosion from a WMU, the entire downslope surface
area can be considered a source, and it becomes important to consider these extended source
areas in the model.
As stated earlier, the watershed that includes the LAU or waste pile is termed here the
"local watershed" (see Figures 5-4 and 5-5). A local watershed is defined as the drainage area
that contains the WMU (or a portion thereof) in the lateral direction (perpendicular to runoff
flow) and the area in which runoff occurs as overland flow (sheet flow) only. Thus, a local
watershed extends downslope only to the point that runoff flows and eroded soil loads would
enter a well-defined drainage channel (e.g., a stream, lake, or some other waterbody). The sheet-
flow-only restriction is based on the assumption that any subareas downslope of the WMU
subarea are subject to constituent contamination from the WMU through overland runoff and soil
erosion.
The local watershed is conceptualized as a 2-D, two-medium system. The dimensions
are longitudinal (i.e., downslope or in the direction of runoff flow), and vertical (i.e., through the
soil column). The media are the soil column and, during runoff events, the overlying runoff
water column. The local watershed is assumed to be made up (in the longitudinal direction) of
an arbitrary number of land subareas2 that may have differing surface or subsurface
characteristics (e.g., land uses, soil properties, and constituent concentrations).
The Local Watershed Model comprises the following three components:
¦ A hydrology model,
¦ A soil erosion model, and
¦ A constituent fate and transport model.
These are described below.
Hydrology Model. Hydrologic modeling is performed to simulate watershed runoff and
ground water recharge (termed here "infiltration"). The hydrology model is based on a daily soil
water balance performed for the root zone of the soil column as shown in Figure 5-7.
2 The conceptual approach allows for an arbitrary number of subareas, but in the implementation of the
3MRA modeling system, that number has been limited to a maximum of 3: upslope, WMU, and downslope/buffer.
5-15
-------
Section 5.0
Land-based Source Modules
Evapotranspiration, ETt
Precipitation, Pt
Soil Moisture, SMt
Runoff, Qt
1/
V Infiltration, I,
SMt= SMt j + P, - Q, - ETt - I,
Figure 5-7. Daily water balance model.
At the end of a given day, the soil moisture in the root zone of an arbitrary watershed subarea is
given as
SMt = + Pt-Ot ~ ETt - It
(5-10)
where
SMj = soil moisture in root zone at end of day t for subarea i (cm)
SMj.j = soil moisture in root zone at end of previous day for subarea i (cm)
Pt = total precipitation on day l (cm)
Qt = net of storm runoff on day l leaving the subarea and storm runoff entering
the subarea (cm)
ETt = evapotranspiration from root zone on day l for subarea i (cm)
I, = infiltration (ground water recharge) on day l for subarea i (cm).
Precipitation is undifferentiated between rainfall and frozen precipitation; that is, frozen
precipitation is treated as rainfall.
The daily runoff estimate is based on the Soil Conservation Service's (SCS's) widely
used "curve number" procedure (USDA, 1986) and is a function of land use and current and
antecedent precipitation. Land use is considered empirically by application of the curve
numbers, which are catalogued by land use or cover type (e.g., woods, meadow, impervious
surfaces), treatment or practice (e.g., contoured, terraced), hydrologic condition, and hydrologic
soil group.
Curve numbers are typically presented in the literature assuming average antecedent
moisture conditions (AMC II), but can be adjusted for drier (AMC I) or wetter (AMC III)
5-16
-------
Section 5.0
Land-based Source Modules
conditions (Chow et al., 1988). These three categories are used in the hydrology model; a
distinction is made within them between the dormant season and the growing season. The
growing season is assumed to be June through August (Julian Day 152 to 243) throughout the
country. These adjustments have the effect of increasing runoff under wet antecedent conditions
and decreasing runoff under dry antecedent conditions, relative to average conditions.
Potential evapotranspiration (PET) is the demand for soil moisture from evaporation and
plant transpiration. When soil moisture is abundant, actual evapotranspiration equals the PET.
When soil moisture is limiting, actual evapotranspiration will be less than PET. The extent to
which it is less under limiting conditions has been expressed as a function of PET, available soil
water, and available soil water capacity (Dunne and Leopold, 1978).
The more theoretically based models for daily evapotranspiration (e.g., the Penman-
Monteith equation [Monteith, 1965]) rely on the availability of significant daily meteorological
data, including temperature gradient between surface and air, solar radiation, windspeed, and
relative humidity. For 3MRA, it is assumed that all of these variables will not be readily
available for all applications. Therefore, a less data-demanding model, the Hargreaves equation
(Shuttleworth, 1985), is used. The Hargreaves method, which is primarily temperature-based,
has been shown to provide reasonable estimates of evaporation (Jensen, et al., 1990)—
presumably because it also includes an implicit link to solar radiation through its latitude
parameter (Shuttleworth, 1993).
Soil moisture in excess of the soil's field capacity, if not used to satisfy
evapotranspiration, is available for gravity drainage from the root zone as infiltration to subroot
zones (Dunne and Leopold, 1978). The resulting infiltration rate will, however, be limited by
the root zone soil's saturated hydraulic conductivity.
In the event that infiltration is limited by the saturated hydraulic conductivity (rather than
evapotranspiration), the hydrology model includes a mechanism to increase the previously
calculated runoff volume by the amount of excess soil moisture (i.e., the soil water volume in
excess of the field capacity). This adjustment is made to preserve water balance and is based on
the assumption that the runoff curve number method, which is only loosely sensitive to soil
moisture (through the antecedent precipitation adjustment) has admitted more water into the soil
column than can be accommodated by evapotranspiration, infiltration, and/or increased soil
moisture. After the runoff is increased by this excess, the evapotranspiration, infiltration, and
soil moisture are updated to reflect this modification and preserve the water balance.
Soil Erosion Model. The soil erosion model used in the Local Watershed Model is
based on the Universal Soil Loss Equation (USLE), an empirical methodology (see, e.g.,
Wischmeier and Smith, 1978) derived from measured soil losses at small, experimental field-
scale plots throughout the United States. The USLE predicts sheet and rill erosion from hillsides
upslope of defined drainage channels, such as streams. It does not predict streambank erosion.
The eroded soil flux from a hillside area averaged over a specific time period is predicted by the
USLE as the product of the six variables discussed below.
The rainfall factor (R, 1/time) accounts for the erosive (kinetic) energy of falling
raindrops, which is essentially measured by rainfall intensity. The kinetic energy of an
5-17
-------
Section 5.0
Land-based Source Modules
individual storm multiplied by its maximum 30-minute intensity is sometimes called the
erosivity index factor. Rainfall factors have been compiled throughout the United States on a
long-term annual average basis. These rainfall factors were developed by cumulating individual
storm erosivity index factors.
The soil erodibility factor (K, kg/m2) is an experimentally determined property that is a
function of soil type, including particle size distribution, organic content, structure, and profile.
Soil erodibility factor values are reported by soil type in the literature.
The cropping management factor (C, unitless) varies between 0 and 1. It accounts for the
type of cover (e.g., sod, grass type, fallow) on the soil. The cropping management factor is used
to correct the USLE prediction relative to the cover type for which the experimentally
determined soil erodibility values were measured (fallow).
The practice factor (P, unitless) accounts for the effect of erosion control practices, e.g.,
contouring or terracing. The practice factor is never negative, but could be greater than 1 if land
practices actually encourage erosion relative to the original experimental plots on which soil
erodibility factors were measured.
The combined length-slope factor (LS, unitless) is given by U.S. EPA (1985b). The
length-slope factor increases with increasing flow distance because runoff quantity and erosive
energy generally increase with downslope distance. LS increases with slope because runoff
velocity generally increases with slope.
The sediment delivery ratio (Sd, unitless) estimates the fraction of on-site eroded soil that
reaches a particular downslope or downstream location in the subbasin (Shen and Julien, 1993).
The sediment delivery ratio is used here to account for deposition of eroded soil from the local
watershed in ditches, gullies, or other depressions before reaching a downslope stream or
waterbody. Vanoni (1975) developed the relationship between the sediment delivery ratio and
the watershed drainage area.
The USLE is implemented within the Local Watershed Model on a storm event basis,
i.e., the "modified" USLE (MUSLE) is used. This implementation requires determining a
rainfall factor value for each daily storm event that specifies the erosivity of that individual
storm.
Constituent Fate and Transport Model. Constituent and suspended solids
concentrations in storm-event runoff are simulated using a two compartment model for the soil
surface layer: a runoff compartment and the soil compartment. Figure 5-8 presents the
conceptual runoff /erosion model showing the two compartments and the fate and transport
processes considered. Hydrolysis, volatilization, and biodegradation processes are not simulated
in the runoff compartment. These processes are continuously simulated in the surface layer of
the soil column by the GSCM. The percentage of time that runoff is actually occurring is
sufficiently short that any additional constituent losses in the runoff water due to these processes
should be minimal.
5-18
-------
Section 5.0
Land-based Source Modules
Runon Flow
Partitioning
Dissolved ^ Particulate
Runoff Flow
RUNOFF
Diffusion
k Settling
r >
^ Resuspension
k
SURFACE SOIL
Burial/erosion
Figure 5-8. Runoff/erosion conceptual model.
Runoff Compartment. The runoff compartment model is based on mass balances of
solids and constituent in the runoff water and the top soil column layer. A simplifying
assumption is made that solids and constituent concentrations in the runoff are at instantaneous
steady-state equilibrium during each individual runoff event, but can vary among runoff events
(i.e., a quasi-dynamic approach is used). While assumption of instantaneous, steady-state
equilibrium for each storm event is not strictly accurate, it was deemed appropriate for the
following reasons:
¦ Data will not generally be available at the temporal scale to accurately track
within-storm event conditions (e.g., rainfall hyetographs).
¦ The actual time to steady-state may not differ significantly from the 1 day or less
implicitly assumed here because of the anticipated relatively small surface areas
of the watershed subareas and the associated relatively small runoff volumes. A
sensitivity analysis was performed, using a dynamic form of the runoff
compartment model, which suggested relatively little difference in calculated soil
concentrations as a function of the steady-state versus dynamic assumption.
¦ To the extent that the actual time to steady-state is greater than 1 day, the model is
biased toward overestimating downslope concentrations and waterbody loads.
A steady-state mass balance of solids in the runoff, i.e., suspended solids from erosion, is
used:
0 = Qi-1 mL, i " Q, mu ~ vs, Amu + vr, A,m2 (5-11)
5-19
-------
Section 5.0
Land-based Source Modules
where
QVi =
total run-on flow volume (water plus solids) from subarea z'-l (m3/s)
(see Equation 5-12)
Q'i =
total runoff flow volume (water plus solids) leaving subarea z (m3/s)
(see Equation 5-12)
mi,i-l =
solids concentration in runon from subarea z'-l (g/m3)
ml,i
solids concentration in runoff from subarea z (suspended solids) (g/m3)
m2
solids concentration in the top soil column layer of subarea z (g/m3)
vs;
settling velocity (m/s)
vr;
resuspension velocity (m/s)
Ai =
surface area of subarea z (m2).
Q\ = Q, + -y- (5-12)
where
Qi = runoff flow leaving subarea i (m3/s)
CSL; = cumulative soil load leaving subarea / (g/s)
p = particle density (g/m3) (i.e., 2.65 g/m3).
The first term in Equation 5-11 is the flux of soil across the upslope interface of subarea i. The
second term is the flux of soil across the downslope interface. The third term is an internal sink
of soil due to settling, and the fourth term is an internal source due to resuspension.
Solids mass transport from or to the soil compartment within any given watershed
subarea is assumed to occur only in a vertical direction, i.e., no downgradient advection of the
top soil column layer itself is considered. The downslope mass transport of soil occurs because
of vertical erosion or resuspension of soil followed by advective transport of the soil in the
runoff water as suspended solids. The transport is described in terms of three parameters:
settling, resuspension, and burial/erosion velocities. Under the assumption of no advective
transport of the soil column layer, the steady-state mass balance equation for the surficial soil
layer is
0 = vst mu AI - vr m2 i At - vbi m2 i At (5-13)
where
vb; = burial/erosion velocity (m/s).
The first term of Equation 5-13 is a source of soil mass to the surficial soil column layer
due to settling from the overlying runoff water. The second term is a sink from resuspension.
The third term is either a source or a sink depending on the sign of the burial/erosion velocity as
described subsequently.
5-20
-------
Section 5.0
Land-based Source Modules
A steady-state mass balance of constituents in the runoff is achieved, as illustrated in
Figure 5-8. The governing equation for this mass balance is as follows:
Fd7 . Fd, .
0 - 2 i-ici,;-r Q fi,r vsiAiFPucu+ vrlA,FP2,iEr,c2,i+ vdiH-^c2,r (5-14)
2 ^1,/
where
cu =
total constituent concentration (particulate + dissolved) in runoff in subarea i
(g/m3)
C2,i —
total constituent concentration in soil (g/m3)
Vu =
subarea-specific (not cumulative) runoff volume for subarea i (m3)
FPJ,i =
fraction of constituent sorbed to particulate in runoff in soil compartment j
(unitless)
Fdy =
fraction of constituent sorbed to dissolved in runoff in soil compartment j (1-
FPj,i)
vd; =
diffusive exchange velocity (m/s)
Er; =
enrichment ratio (unitless)
=
porosity of the runoff (unitless)
Note that c|)2 is equivalent to porosity (r|) in the GSCM.
Equation 5-14 can be used to express chi as a function of cLl_, and c2h as shown in
Equation 5-15:
Q 'j- lCl,i- 1 + \yriAiFP2,firi+ vdiAi(Fd2j^C2J
Cu Q'i+ vsiAiFPij+ vdiAi(FdJ%)
where c2i is determined by the GSCM, as described in Section 5.2.2.
An enrichment ratio is used to account for preferential erosion of finer soil
particles—with higher specific surface areas and more sorbed constituent per unit area—as
rainfall intensity decreases. That is, large (highly erosive) runoff events may result in average
eroded soil particle sizes and associated sorbed constituent loads that do not differ much from
the average sizes/loads in the surficial soil column layer. However, less intense runoff events
will erode the finer materials, and resulting constituent loads could be significantly higher than
represented by the average soil concentration. U.S. EPA (1985b) gives the storm-event-specific
enrichment ratio as a power function of sediment discharge flux.
Soil Compartment. From the discussion of the GSCM, the governing differential
equation for the surface soil layer of subarea i is
dc7. re. re.
—^=de --VE—^L-lkiC2t + sst (5-16)
dt E dz2 E dz ] %t v '
5-21
-------
Section 5.0
Land-based Source Modules
where kj represents first-order rate constant due to not including runoff/erosion processes, i.e.,
biological decay and hydrolysis and wind/mechanical action. The last term, .v.v„ is a source/sink
term representing the net effect of runoff and erosion processes on the concentration in the
surface soil. The term sst comprises the following terms respectively: a source of constituent due
to settling from the overlying runoff water, a sink of constituent due to resuspension, and a
source or sink (depending on the relative values of Cu and C2 i) due to constituent diffusion
from/to the runoff.
Referring back to the governing equation for the surface soil column layer, the first two-
component equations remain the same, while the third is revised to
—fL = - A-'C, + Id, , (5-17)
dt -
With this equation, the constituent mass balance is maintained in the GSCM with the inclusion
of erosion and runoff.
5.2.4 Particulate Emissions Model
The Land-based Source Modules have been designed to provide estimates of the annual
average, area-normalized emission rate of constituent mass adsorbed to particulate matter less
than 30 |im in diameter (PM30), as well as annual average particle size distribution information in
the form of the mass fractions of the total particulate emissions in four aerodynamic particle size
categories—30 to 15 |im, 15 to 10 |im, 10 to 2.5 |im, and <2.5 |im.
A variety of release mechanisms are considered. The release mechanisms considered
differ for each WMU and may include wind erosion, vehicular activity, unloading operations,
tilling, and spreading and compacting operations. Table 5-1 summarizes the mechanisms
considered for each WMU.
Table 5-1. Summary of Mechanisms of Release of Particulate Emissions for Each WMU
Mechanism of Release
WMU Type3
Algorithm
Reference
LAU
LF
\\T
Active
Inact.
Active
Inact.b
Active
Inact.
Wind erosion from open area
/
/
/
/
Cowherd et al. (1985)
Wind erosion from waste pile
/
/
U.S. EPA (1985a)
Vehicular activity
/
/
/
U.S. EPA (1995a)
Unloading
/
/
U.S. EPA (1995a)
Spreading and compacting or tilling
/
/
/
U.S. EPA (1985a)
a Active = Operating WMU. Inact. = Inactive WMU where no additional constituent mass is being added.
b Inactive (full) and uncovered landfill cell. Assume no emissions from a covered landfill cell.
5-22
-------
Section 5.0
Land-based Source Modules
This section describes the algorithms and assumptions used to estimate the following for
each mechanism of release:
¦ Annual average PM30 emission rate due to each mechanism considered for each
WMU, and
¦ Annual particle size range mass fractions, i.e., the mass fractions of PM30 in the
aerodynamic particle size categories identified above.
For each WMU, the following are estimated:
¦ The total annual average PM30 emission rate due to all release mechanisms,
¦ The annual average emission rate of constituent sorbed to PM30,
¦ Annual average particle size range mass fractions of the total annual average
PM30 emission rate, and
¦ Annual average first-order loss rate from the soil surface due to constituent mass
losses caused by particulate emissions.
Wind Erosion from Open Fields. The algorithm for the estimation of PM30 emissions
due to wind erosion from an open field is based on the procedure developed by Cowherd et al.
(1985). The annual average emission rate of PM30 per unit area of the contaminated surface is
estimated in the LAU and Landfill Modules.
To account for differing degrees of vegetation, surface roughness height, and frequency
of disturbances per month in active versus inactive WMUs, different values are assigned to these
parameters according to whether the WMU is active or inactive.
The methodology of Cowherd et al. (1985) was originally developed to estimate the
emission rate of PM10. Emission rates for PM30 can be approximated from those for PM10 with
knowledge of the ratio between PM30 and PM10 emissions for wind erosion. According to
Cowherd (1998), a good first approximation of this ratio is provided by the particle size
multiplier information presented in U.S. EPA (1995) for wind erosion from open fields, where
PM30/PM10 is equal to 2. Therefore, a factor of 2 has been incorporated into Cowherd et al.'s
(1985) equations for PM10 emission to allow estimation of PM30 emissions.
Wind Erosion from Waste Piles. The equation used in the Waste Pile Module to
estimate the annual average PM30 emission rate per unit area of contaminated surface due to
wind erosion is an adaptation of the empirical equation developed for total suspended particulate
matter (TSP) from active sand and gravel waste piles presented in U.S. EPA (1985a, referred to
here as AP-42; see Equation 3, p. 11.2.3-5). TSP is defined as those particulates measured by a
high-volume sampler, and the effective cutoff commonly assigned to standard high-volume
samplers is 30 |im (U.S. EPA, 1985a). A dust-control efficiency factor is added.
5-23
-------
Section 5.0
Land-based Source Modules
A more recent version of AP-42 (U.S. EPA, 1995) recommends the use of an event-based
algorithm for estimating wind emissions from a waste pile. The updated algorithm was
evaluated for use in 3MRA, but it was determined that it requires detailed site-specific
information, typically not available for site-based analyses. The algorithm used in the Land-
based Source Modules will tend to overestimate emissions relative to the event-based algorithm
(Meyers, 1998).
Vehicular Activity. To estimate the PM30 emissions from vehicular travel on the surface
of the WMU, an algorithm was derived from an empirical equation presented in U.S. EPA (1995;
Equation 1, p. 13.2.2-1) for the kilograms of size-specific particulate emissions emitted per
vehicle per kilometer traveled on unpaved roads. (In the Land-based Source Modules, the EPA
parameter "fraction of waste on unpaved roads" is equal to 1 because travel is on the surface of
the WMU.) EPA's equation has been adapted for the Land-based Source Modules to provide
emissions normalized to the contaminated surface area and to account for the control of
emissions with a dust-control efficiency factor.
Unloading Operations. The equation for estimating the PM30 emission rate due to
unloading operations at waste piles and landfills was adapted from U.S. EPA (1995, Equation 1,
p. 13.2.4-3). The EPA equation was adapted by multiplying it by the average annual loading
rate, normalizing the emissions for the contaminated surface area, and applying the particle size
multiplier for <30 |im.
Spreading and Compacting or Tilling Operations. The equation for estimating the
rate of PM30 emissions due to spreading and compacting or tilling operations was adapted from
an equation in U.S. EPA (1985a, Equation 1, p. 11.2.2-1) that was developed for estimating
emissions due to agricultural tilling in units of kilograms of particulate emissions per hectare per
tilling (or spreading and compacting) event. The particle size multiplier for <30 |im is applied in
the Land-based Source Modules.
5.3 Module Discussion
5.3.1 Strengths and Advantages
Relative to other models that might have been candidates for the functionality provided
by the Land-based Source Modules, the strengths and advantages of the modules include the
following:
¦ The GSCM extends the functionality of the Jury model. The GSCM was
developed to make modest improvements to known limitations of the Jury-type
approach for the 3MRA modeling system. Specifically, the GSCM simulates
changes over time in vertical soil constituent concentration profiles given (1)
periodic waste applications that result in non-uniform vertical concentration
profiles (this addresses the Jury limitation of uniform initial condition); and (2) a
finite vertical boundary is formed by an underlying vadose zone so that the lower
boundary depth is not infinite (this addresses the Jury limitation of a zero
concentration boundary condition at the lower soil column boundary. (A zero
concentration gradient boundary condition is assumed instead.)
5-24
-------
Section 5.0
Land-based Source Modules
The "local watershed" algorithm gives the LAU and Waste Pile Modules
spatial resolution appropriate for the 3MRA modeling system exposure
scenarios. The conceptual site models for both the LAU and waste pile WMUs
assume that both WMUs are adjoined by a buffer area (non-WMU land area).
(The landfill is not assumed to have off-site runoff or erosion, so it is not
considered part of a "local watershed".) The buffer area is situated between the
WMU and the nearest surface waterbody. Overland flow runoff is assumed to
occur from the WMU, across the buffer area, and into the nearest surface
waterbody. This set of land areas (WMU, buffer, and a possible third land area
upslope of the WMU, which contributes stormwater runoff and eroded soil, but
not constituent loads, to the WMU area) located on the "hillside" adjacent to a
surface waterbody constitutes the "local" watershed. Because constituent
concentrations over time will be different in the buffer area than in the WMU
area, and receptors can be located in the buffer area, it was important for the
purposes of the 3MRA modeling system to have this spatial resolution. Prior to
developing the Land-based Source Modules, which contain this "local watershed"
functionality, a review of candidate watershed/vadose zone models was
performed, and none were found to have this spatial resolution. For example,
EPA's PRZM model, widely used for pesticide contamination studies in
agricultural fields, is "field-scale" only (i.e. it considers the agricultural field only
and does not consider a downslope, adjacent buffer area). On the other end of the
spectrum, the HSPF watershed model is "watershed-scale"(i.e., it considers entire
drainage basins), so that what might be happening on a two-land-area hillside in
that larger drainage area is considered by its resolution. Thus, the "hillside-scale"
of the local watershed algorithm is intermediate between these two extremes and
needed to be custom-developed.
A carefully selected hybrid of empirical and mechanistic algorithms and
approaches are used to maximize overall suitability for the purposes of the
3MRA modeling system. From a model development perspective, the 3MRA
framework presented an interesting challenge with regard to selection of
appropriate levels of mechanistic detail. As a general rule, the more scientifically
based (mechanistic) the model, the greater potential that model has to provide
realistic simulations under a variety of environmental conditions. However, that
potential carries with it the price of including many parameters that must be
estimated from site-specific data before these highly parameterized models reach
their potential. This essentially requires site-specific model calibration. In
contrast, the 3MRA modeling system is applied to many sites with limited site-
specific data. Thus, site-specific model calibration was not an option and the
modeling approach needed to be based on a judicious balance of theoretical and
empirical methods. For example, the GSCM is based on theoretical, mass balance
principles, but involves parameters that are readily available (e.g., soil properties,
constituent properties), or can be assigned from national distributions. Other
algorithms (e.g., the Universal Soil Loss Equation for eroded soil losses, and the
"curve number" method for estimating stormwater runoff) are fundamentally
empirical, yet are widely used and acknowledged to provide reasonable
predictions for planning-level applications.
5-25
-------
Section 5.0
Land-based Source Modules
5.3.2 Uncertainty and Limitations
The major limitations of the Land-based Source Modules and their implications are
summarized below, by model.
Limitations of the GSCM. The GSCM was originally designed for the 3MRA modeling
system to make some modest improvements to the well-known Jury model, which had been used
historically to simulate constituent fate and transport in land-based sources. The GSCM has the
following limitations:
¦ Series solution of diffusion/decay/advection processes is problematic for some
constituents. The GSCM solves the mass balance differential equation
sequentially in time in the following order: diffusive transport, internal decay,
advective transport. This approach is justified conceptually on the basis of
linearity and the principle of superposition; however, the approach can lead to
simulation errors when relatively large time steps are taken. The magnitude of
the errors would depend on the relative loss rates associated with the three
processes. For example, if the first-order loss rate due to degradation were high
and the calculation time step were sufficiently large, then leachate flux and the
mass of constituent remaining for the advection phase would be underestimated.
¦ Leachate flux postprocessing algorithm is artificial. The leachate flux
postprocessing algorithm postprocesses the constituent mass in the leachate over
the total period of the simulation and attempts to allocate it over the individual
years in the simulation in a reasonable manner. This postprocessing is needed for
persistent constituents (highly sorbed, low decay) because of the combination of
choice of the GSCM state variable (total concentration) and the diffusion/decay/
advection series solution methodology. Patterned after the Jury model, the
GSCM's state variable is total (sorbed + dissolved + gaseous) constituent
concentration. The "effective solute convection velocity," VE, is not the actual
leachate infiltration velocity, but rather (because total constituent is the state
variable, not dissolved) is corrected to account for sorption. When that sorption is
high (high Kd), the VE is small and can be much smaller (slower velocity) than the
true infiltration rate. "Convective transport," i.e., advection, of the constituent
from one computational layer in the GSCM down to the next (or from the bottom
layer out as leachate) is performed only when the constituent has traversed the
depth of a computational layer (typically 1 cm) as measured by VE, not by the
actual infiltration velocity. For very small values of VE, this travel time can
become excessive, and vertical movement due to advective transport may not
occur within a year; indeed, it may not occur for the duration of the simulation.
In reality, dissolved constituent would be more or less continuously moving
downward due to advective transport in the infiltration. The postprocessing
algorithm attempts to mitigate this limitation.
¦ Treatment of different zones with appropriate boundary conditions is not
possible. The GSCM solution methodology does not allow a rigorous treatment
of scenarios where different zones with different sorption characteristics exist in
5-26
-------
Section 5.0
Land-based Source Modules
the soil column. For example, if the landfill soil column had a liner with less
sorptive capacity than the waste zone, then the liner's sorptive capacity would be
exhausted relatively quickly relative to the waste zone's. At this point, since
dissolved constituent drains from the waste zone into the liner, there would be no
further reduction by sorption, and a boundary condition should be applied at this
waste/liner interface that requires a zero concentration gradient for dissolved
constituent. The GSCM cannot accommodate such a boundary condition (or an
analogous waste/cover boundary condition), superficially because the GSCM is
simulating total (not dissolved) constituent, but more fundamentally because the
underlying solution methodology is simply not designed to deal with internal
boundary conditions. This limitation has been addressed by imposing the relevant
boundary condition, but it is not a mathematically rigorous treatment of the
physical scenario. A more elegant solution would be to include nonlinear
sorption kinetics in the algorithm, so that the relative differences in sorption
among different zones are intrinsically accounted for, rather than imposed by
boundary conditions.
¦ Simulation of volatilization out of the surface of soil column is artificial. The
GSCM's simulation of diffusive transport over the depth of the soil column
transports total constituent in accordance with the constituent's water diffusivity
(corrected for sorption). This diffusion is treated as diffusion in an infinitely deep
column, so that total constituent flux is effectively diffused across the surface of
the soil column. The amount of constituent mass that is not in the gaseous phase,
but has been (erroneously) diffused across the surface, is then calculated and
"replaced" back into the top computational layer to better reflect reality. It is
unclear to what extent this overall treatment biases the volatile emissions
themselves, but the re-introduction of the escaping, nongaseous constituent into
the surficial soil layer is artificial and could lead to an erroneous estimate of the
vertical concentration gradient.
Limitations of the Local Watershed Model. The Local Watershed Model has the
following limitations:
¦ Burial/erosion introduces minor mass balance error. The burial/erosion
mechanism introduces a minor mass balance error into the model. The model for
surface soil/runoff fate and transport is based on a conceptual model originally
developed for use in a stream/sediment application where the sediment
compartment location relative to a reference point below the surface can move
vertically ("float") as burial and erosion occur. In that moving frame of reference,
burial/erosion of a constituent does not introduce a mass balance error. However,
in this application, the frame of reference is not allowed to float, but is fixed by
the elevation of the lower boundary (e.g., top of the vadose zone). Thus, if sorbed
constituent is eroded from the surface computational cell, that surface cell, which
is vertically fixed, must have a "source" that is internal to the modeled soil
column to compensate for this sink, or its internal mass balance is not maintained.
The magnitude of this mass balance error is equal to the mass of eroded soil from
the surface over the duration of the simulation multiplied by its average sorbed
5-27
-------
Section 5.0
Land-based Source Modules
constituent concentration. In most cases, this error as a percentage of the total
constituent mass in the modeled WMU will be quite small, and that has been
confirmed in multiple executions of the model.
¦ Potential hot spots in the local watershed buffer subarea are diluted. The
local watershed buffer is considered as a single, homogeneous subarea by the
model to facilitate watershed delineation and decrease run time. This will result
in spatial averaging of concentrations in the buffer.
¦ Sheet flow is assumed across buffer subarea. The conceptual Local Watershed
Model construct assumes that runoff and erosion occur as sheet flow from the
WMU across the buffer subarea to the waterbody. This implicitly assumes no
short-circuiting of constituent loads directly from the WMU to the downslope
waterbody, such as might actually occur in runoff/erosion-created ditches or
swales. To the extent that such short-circuiting might occur, the model will
underestimate waterbody constituent loadings. Mass balance is maintained;
consequently, any such underestimation would come at the expense of
overestimating soil concentrations in the (bypassed) buffer zone.
¦ Hydrological responses are limited by the available record. The hydrology
model uses available historical records of meteorological data. When the number
of years simulated exceeds the number of years in the record, the record is
repeated. Thus, unusual hydrological events (e.g., major storms) are limited to
those actually observed. Events not observed, but possible nonetheless, that could
result in increased source releases or media transport will not be included in the
simulation.
Limitations of the Particulate Emissions Model. The particulate emissions models
were incorporated without changes from the sources discussed (mostly AP-42) and have the
same limitations as those models, as described in the source documentation.
5.4 References
Chow, Ven Te, David R. Maidment, and Larry W. Mays. 1988. Applied Hydrology. New York,
NY: McGraw-Hill, Inc.
Cowherd, C., G.E. Muleski, P.J. Englehard, and D.A. Gillette. 1985. Rapid Assessment of
Exposure to Particulate Emissions from Surface Contamination Sites. EPA/600/8-
85/002. U.S. Environmental Protection Agency, Office of Research and Development,
Office of Health and Environmental Assessment, Washington, DC. February.
Cowherd, C.H. 1998. Personal communication. Midwest Research Institute, Kansas City,
Missouri, February 27.
Dunne, Thomas, and Luna B. Leopold. 1978. Water in Environmental Planning. New York:
W.H. Freeman and Company.
5-28
-------
Section 5.0
Land-based Source Modules
Jensen, M.E., R.D. Burman, and R.G. Allen. 1990. Evapotranspiration and irrigation water
requirements. ASCEManual 70:332.
Jost, W.A. 1960. Diffusion in Solids, Liquids, Gases. Third Printing (with Addendum). New
York, NY: Academic Press, Inc.
Jury, W.A., W.F. Spencer, and W.J. Farmer. 1983. Behavior assessment model for trace
organics in soil: I. Model description. Journal of Environmental Quality 12(4):558-564.
October.
Jury, William A., David Russo, Gary Streile, and Hesham El Abd. 1990. Evaluation of
volatilization by organic chemicals residing below the soil surface. Water Resources
Research 26(1): 13-20. January.
Myers, R. 1998. Personal communication. Office of Air Quality and Planning. U.S. EPA,
Research Triangle Park, NC, January 8.
Millington, R.J., and J.P. Quirk. 1961. Permeability of porous solids. Transactions of the
Faraday Society 57(7): 1200-1207. July.
Monteith, J.L. 1965. Evaporation and Environment. Pp. 205-234 in Symposia of the Society for
Experimental Biology: Number XTX. New York, NY: Academic Press, Inc.
Richardson, C.W., G.R. Foster, and D.A. Wright. 1983. Estimation of erosion index from daily
rainfall amount. Transactions of the ASAE 26(1): 153-156.
Shan, Chao, and Daniel B. Stephens. 1995. An analytical solution for vertical transport of
volatile chemicals in the vadose zone. Journal of Contaminant Hydrology 18:259-277.
Shen, Hsieh Wen, and Pierre Y. Julien. 1993. Chapter 12: Erosion and sediment transport.
Pp. 12-12 in Handbook of Hydrology, David R. Maidment (ed.). NewYork,NY:
McGraw-Hill, Inc.
Shuttleworth, W. James, and J. S. Wallace. 1985. Evaporation from sparse crops - an energy
combination theory. Quart. J. R. Met. Soc., Ill, 839-855.
Shuttleworth, W. James. 1993. Chapter 4: Evaporation. Pp. 4-4 in Handbook of Hydrology,
David R. Maidment (ed.). New York, NY: McGraw-Hill, Inc.
USDA (Department of Agriculture). 1986. Urban Hydrology for Small Watersheds. TR-55.
U.S. Department of Agriculture, Engineering Division, Soil Conservation Service,
Washington, DC. p. 2-5. June.
U.S. EPA (Environmental Protection Agency). 1985a. Compilation of Air Pollutant Emission
Factors. Volume L: Stationary Point and Area Sources (Fourth Edition). AP-42. U. S.
Environmental Protection Agency, Office of Air and Radiation and Office of Air Quality
Planning and Standards, Research Triangle Park, NC. September.
5-29
-------
Section 5.0
Land-based Source Modules
U.S. EPA (Environmental Protection Agency). 1985b. Water Quality Assessment. A Screening
Procedure for Toxic and Conventional Pollutants in Surface and Ground Water-Part I.
(Revised). EPA/600/6-85/002a. Office of Research and Development, Environmental
Research Laboratory, Athens, GA. September.
U. S. EPA (Environmental Protection Agency). 1995. Compilation of Air Pollutant Emission
Factors Volume 1: Stationary Point and Area Sources, 5th Edition. AP-42. PB95-
196028INZ, Office of Air Quality Planning and Standards, Research Triangle Park, NC.
U.S. EPA (Environmental Protection Agency). 1999. Source Modules for Non-Wastewater
Waste Management Units (Land Application Units, Waste Piles, and Landfills),
Background and Lmplementation for the Multimedia, Multipathway, andMultireceptor
Risk Assessment (3MRA) for HWLR99. Office of Solid Waste, Washington, DC.
October.
Vanoni, Vito A. (ed.). 1975. Sedimentation Engineering. New York, NY: American Society of
Civil Engineers.
Wischmeier, W.H., and D.D. Smith. 1978. Predicting rainfall erosion losses. A guide to
conservation planning. In Agricultural Handbook. 537 Edition. Washington, DC: U.S.
Department of Agriculture.
5-30
-------
Section 6.0 Air Module
6.0 Air Module
6.1 Purpose and Scope
The Air Module estimates the annual average air concentration of dispersed constituents
and the annual deposition rates of vapors and particles at various receptor points in the area of
interest (AOI). This module simulates the transport and diffusion of constituents in the form of
volatilized gases or fugitive dust emitted from area sources into the air. The predicted air
concentrations are used to estimate biouptake into plants, and human exposures due to direct
inhalation. The predicted deposition rates are used to determine constituent loadings to farm
crops and soils, watershed soils, and surface waterbodies. Figure 6-1 shows the relationship and
information flow between the Air Module and the 3MRA modeling system.
Surface
Impoundment/
Aerated
Tank Source
Modules
Key Data Inputs
Source height
Wind speed
Annual precipitation rate
Land-based
Source
Modules
Emission
Rates
Emission
Module
Rates
Air Concentrations
Farm Food 1
Chain Module 1
Deposition Rates
Deposition Rates ^
Watershed 1
Module 1
W
Air Concentrations ^
Human |
Exposure 1
Module 1
Deposition Rates
Surface Water 1
Module 1
Air Concentrations
Terrestrial Food 1
Web Module 1
Deposition Rates
Figure 6-1. Information flow for the Air Module in the 3MRA modeling system.
6-1
-------
Section 6.0
Air Module
The Air Module performs four major functions, as follows:
1. Characterize source-specific parameters. For each AOI, the Air Module
characterizes emission sources in terms of waste management unit (WMU)
dimensions, emission rate, and the site-layout. As an option, the module can
calculate the source-specific, long-term average particulate mass fraction
distribution from the outputs of the Land-based Source Modules (Landfill, Waste
Pile, or Land Application Unit [LAU] Module).
2. Calculate receptor locations (polar grid or site specific). The Air Module
provides the option to model directly to all site-specific output coordinates needed
by the 3MRA modeling system or to model to a set of polar coordinates and then
use a two-dimensional (2-D) cubic spline method to interpolate from the polar set
to the larger set of interest. The spline interpolation can be used to reduce the
ISCST3 run time.
3. Calculate receptor-specific contaminant concentration and deposition rates.
The Air Module calculates annual average air concentration and deposition rates
for each receptor location specified. Concentrations and deposition rates
calculated include
- Air concentration of vapors,
- Air concentration of particles,
- Wet deposition rate for vapors and particles, and
- Dry deposition rate for particles.
4. Calculate constituent-specific annual average concentrations and deposition
rates. The Air Module converts the receptor-specific concentrations and
deposition rates based on unit emission rates (e.g., 1 g/m2-s) to constituent-
specific estimates by multiplying the values by the constituent-specific emission
rate for each year.
ISCST3 may be implemented by the 3MRA modeling system during a run, or it may be
run outside of the 3MRA modeling system and the results called by the Air Module.
6.2 Conceptual Approach
Figure 6-2 illustrates the dispersion and subsequent deposition of vapors and particles
from a WMU into the environment. Constituents are released to the air as vapors (by
volatilization) or sorbed to particulates (by wind erosion and mechanical disturbances) and move
through the air to locations around the AOI. The constituents can then deposit on soil and plant
surfaces. The Air Module is used to estimate air concentrations and deposition rates for each
contaminant/site/WMU combination.
6-2
-------
Section 6.0
Air Module
Dispersion
w
Volatilization
Bioaccumulation
Infiltration
Wind
Particulates
• * v; Deposition
Groundwater
Well
Figure 6-2. Conceptual diagram of dispersion and deposition.
The Air Module is based on EPA's legacy model Industrial Source Complex Short-Term
Model, Version 3 (ISCST3), which has been used extensively by EPA in regulatory applications.
ISCST3 was extended for use in the Air Module through the development of pre- and
postprocessors that serve as an interface between the ISCST3 model and the rest of the 3MRA
modeling system. The preprocessor reads initial settings from files generated by the 3MRA
modeling system. Next, the preprocessor determines if the core model needs to be run or if
results from previous runs can be reused. If the ISCST3 model needs to be run, the preprocessor
builds an input file that controls the ISCST3 run. The ISCST3 model is run using unit
constituent emissions to produce receptor-specific annual average air concentrations and
deposition rates normalized to emission rate. The postprocessor converts these normalized
concentration and deposition estimates to constituent-specific annual averages by multiplying the
values by the constituent-specific emission rates for the WMU for each year. The resulting
annual averages are written to files that are read by other modules. A summary of the ISCST3
model—the science and the computational core of the Air Module—is presented in the
accompanying text box. The full capabilities of the ISCST3 model are explained in U.S. EPA
(1995).
6-3
-------
Section 6.0
Air Module
Summary ofISCST3
ISCST3 (U.S. EPA, 1999a,b), a recommended dispersion model inEPA's Guideline on Air Quality Models (U.S. EPA,
1999c), is a steady-state, Gaussian plume dispersion model. A steady-state model is one in which the model inputs and
outputs are constant with respect to time. The term "Gaussian plume" refers to the kind of mathematical solution used to
solve the air dispersion equations. It essentially means that the constituent concentration is dispersed within the plume
laterally and vertically according to a Gaussian distribution, which is similar to a normal distribution. These assumptions
and solutions hold for each hour modeled. The results for each hour are then processed to provide values for different
averaging times depending on the user's needs (e.g., annual average).
ISCST3 is capable of simulating dispersion of pollutants from a variety of sources, including point, area, volume, and line
sources. ISCST3 can account for both long- and short-term air concentration of particles and vapor, and wet and dry
deposition of particles and vapor. In addition to deposition, wet and dry plume depletion can be selected to account for
removal of matter by deposition processes and to maintain mass balance. Receptor locations can be specified in polar or
cartesian arrays or can be set to discrete points as needed. Flat or rolling terrain can be modeled, but only flat terrain can be
used for area sources. ISCST3 considers effects of the environmental setting on dispersion by allowing the user to set
urban or rural dispersion parameters. Dispersion around the centerline of the constituent plume is estimated by empirically
derived dispersion coefficients that account for horizontal and vertical movement of constituents. In addition, constituent
movement from the atmosphere to the ground is also modeled to account for deposition processes driven by gravitational
settling and removal by precipitation.
For area sources, the general equation for ground-level concentration at a receptor is given by a double integral in the
upwind (x) and crosswind (y) directions:
X =
Qak
2
f
-
( \
y
2 '
\
VD
r
exP
- 0.5
dy
dx
J
\y
K ayJ
(6-1)
where
X
Qa
K
x
y
z
V
D
concentration at x, y (mass per volume)
area source emission rate (mass per unit area per unit time)
scaling coefficient to convert calculated concentrations to desired units (default value of 1 x 106 for
Q in g/s and concentration in ng/m3)
distance from source in wind direction (positive is downwind) (m)
distance from centerline of plume across the wind direction (m)
vertical distance from ground (m)
vertical term accounting for vertical distribution of the plume (includes effects of source elevation,
receptor elevation, plume rise, limited mixing in the vertical direction, and gravitational settling and
dry deposition of particulates)
decay term accounting for pollutant removal by physical or chemical processes
standard deviation of lateral and vertical concentration distribution (m)
mean wind speed at release height (m/s).
Key meteorological data required for the ISCST3 model include
Wind Direction: Determines the direction of the greatest impacts.
Windspeed: Affects ground-level air concentration. The lower the windspeed, the higher the concentration.
Stability Class: Affects rate of lateral and vertical diffusion. The more unstable the air, the greater the
diffusion.
Mixing Height: Determines the height to which constituents can be diffused vertically.
Deposition Parameters: Additional site-specific meteorological parameters are required to make estimates of
dry and/or wet deposition. These parameters vary based on land-use types and seasons.
6-4
-------
Section 6.0
Air Module
6.2.1 Characterize Source-specific Parameters
For each site, ISCST3 input files can be created that contain source-specific information
such as WMU type, dimensions, location, and source variables for settling, removal, and
deposition. Values for most of these data are obtained from other modules or the 3MRA
modeling system databases.
The WMU types modeled by the 3MRA modeling system include surface impoundments,
aerated tanks, landfills, waste piles, and LAUs. All of these unit types except for aerated tanks
and waste piles are modeled as ground-level area sources. Aerated tanks and waste piles are
modeled as elevated area sources.
For the 201 sites in the representative national data set, the location and the size of the
sources is determined from the 3MRA modeling system site-based database. The sources for
these data are the 1985 Screening Survey for Industrial Subtitle D Establishments (Westat, 1987)
and the 1986 Survey of Hazardous Waste Treatment, Storage, and Disposal Facilities (U.S. EPA,
1987). Neither of these data sources includes data on height; therefore, the waste pile heights
were calculated based on volume of waste and area of the pile, and tank height was randomly
selected. These heights are stored in the representative national data set.
When modeling particulates, ISCST3 requires a long-term average particle size
distribution. As an option, the Air Module can calculate the source-specific mass fraction
distribution. The Land-based Source Modules (i.e., the Landfill, Waste Pile, and LAU Modules)
generate a time series of particle size distributions (i.e., one distribution per year), as well as a
time series of particulate mass fluxes for particles less than or equal to 30 |im in diameter. From
these inputs, the Air Module estimates the long-term average particle size distribution.
Other source-specific information included in the ISCST3 input file include specification
of rural versus urban setting and meteorological dataset information.
Rural vs. Urban. The rural vs. urban setting in ISCST3 allows the user to account for
differences between rural and urban environments. In urban environments, the built
environment (e.g., buildings, roads, and parking lots) alters the dispersion character of the
atmosphere, particularly at night as a result of building-induced turbulence and reduced
nighttime cooling. Thus, there is greater nighttime mixing of constituents in urban areas
compared with rural areas. For the purposes of ISCST3 modeling, the urban classification
applies mainly to large cities; even small cities and suburban areas are more appropriately
classified as rural.
Meteorological data. The 3MRA modeling system database contains meteorological
data collected at regional meteorological stations. Each of the 201 sites contained in the 3MRA
modeling system representative national data set was assigned to the nearest station with similar
weather conditions and adequate data. In making these assignments, EPA considered all
available data from 218 meteorological stations across the United States to find the best data for
each site. Each of the meteorological data sets included in the 3MRA modeling system database
contains a minimum of 10 years of data.
6-5
-------
Section 6.0
Air Module
6.2.2 Calculate Receptor Locations (polar-grid or site-specific)
The user has the option of specifying receptor locations for all cartesian coordinates as
requested by the 3MRA modeling system, or for a set of polar grid coordinates. If the second
option is chosen, a two-dimensional cubic spline method (discussed below) is applied to
interpolate from the polar grid coordinates to the larger receptor location set requested by the
3MRA modeling system.
The run time for ISCST3 can be long, particularly for large area sources with many
receptor locations. As an option to minimize run time of ISCST3 in the Air Module, a 2-D cubic
spline interpolation algorithm was developed. Under this option, run-time savings can be
achieved by using ISCST3 to model directly to a smaller number of points than the total number
of receptor locations for the site. In general, the points making up the spline data set are not a
subset of the points representing the receptor locations; rather, they are uniformly distributed
across the AOI. A requirement of the spline algorithm is that the data provided to the spline
must be "complete" in the sense that, no matter how the underlying 2-D grid is configured (e.g.,
cartesian, polar), data must be provided at each intersection of the two dimensions. A polar grid
is used as the basis for the spline because, under the "complete" data set requirement of the
spline algorithm, a polar grid can achieve a higher resolution in the region of steepest output
gradients (near the WMU) with fewer total points than can a cartesian grid.
The spline method can only interpolate values. Consequently, value estimates cannot be
made for any points located outside of the outermost grid circle.
6.2.3 Calculate Receptor-specific Concentration and Deposition Rates
Both vapors and particles are modeled for landfills, waste piles, and LAUs; only vapors
are modeled for surface impoundments and aerated tanks. The type of model output selected
depends on the WMU being modeled. Concentration, dry deposition rate, and wet deposition
rate are calculated for landfills, waste piles, and LAUs. Only concentration and wet deposition
rate are calculated for surface impoundments and aerated tanks because the pollutants are only
emitted in vapor phase. Dry deposition of vapors is not modeled in the Air Module.
Consequently, dry deposition of vapors is calculated as part of the Farm Food Chain Module (see
Section 10.2.1) and Terrestrial Food Web Module (see Section 11.2.1). These modules estimate
the dry deposition rate of vapors based on the vapor-phase concentration in air and a vapor-phase
dry deposition velocity (default value of 1 cm/s).
The Air Module calculates the following concentrations and deposition rates:
¦ Air concentration of particles and vapors. ISCST3 estimates air concentrations
of particles and vapors, accounting for downwind movement of the plume. It also
accounts for dispersion of vapors and particles around the centerline of the plume
as the plume travels in a downwind direction. Removal of constituent mass from
the plume occurs as a result of wet and dry deposition.
¦ Wet deposition of particles and vapors. Wet deposition is the loss of material
from a plume onto a surface as a result of precipitation. A scavenging ratio
6-6
-------
Section 6.0
Air Module
approach is used to model the deposition of gases and particles through wet
removal. The amount of material removed from the plume by wet deposition is a
function of the scavenging rate coefficient, which is based on particle size
(U.S. EPA, 1995) for particles. For vapors, ISCTS3 is typically run with
constituent-specific scavenging coefficients. These scavenging coefficients are
read in by the Air Module. When multiple constituents are run, this method
proportionately increases the number of runs that need to be made. As an
alternative to applying constituent-specific coefficients, a single set of gas
scavenging coefficients stored in the representative national data set can be
applied. These coefficients are based on approximating the gases as very small
particles.
¦ Dry deposition of particles. Dry deposition refers to the loss of material from a
plume that has been deposited onto a surface (e.g., ground, vegetation) as a result
of processes such as gravitational settling, turbulent diffusion, and molecular
diffusion. Dry deposition of particles is calculated as the product of air
concentration and dry deposition velocity.
To reduce the computational burden on the 3MRA modeling system, several new features
were added to ISCST3 to create the Air Module. A complete description of the technical
algorithms for these features is provided in U.S. EPA (1999b). Operational instructions are
provided in U.S. EPA (1999a).
Revised Plume Depletion Scheme. The version of ISCST3 distributed by EPA's
Office of Air Quality Planning and Standards contains the Horst (1983) plume
depletion algorithm. This algorithm was found to be computationally intensive.
A new plume depletion and settling algorithm developed by Venkatram (1998)
was implemented for the Air Module, resulting in a faster, more robust approach.
This approach is based on depleting material in a surface-based internal boundary
layer that grows with distance from the source. In conjunction with this change,
the deposition velocity algorithm was also modified by removing the inertial
impaction term. The inclusion of this term appears to provide deposition velocity
estimates that are too high for some particle sizes.
Sampled Chronological Input Model (SCIM). To reduce model run time, an
option was added to the Air Module to sample the long-term meteorological
record at regular, user-specified intervals and scale the model results at the end of
the run to produce annual average estimates. This method is called the Sampled
Chronological Input Model (SCIM). An advantage of this method is that it uses
hourly meteorological data that maintain the serial correlation between wet
deposition rate and concentration. The user specifies two sampling intervals.
Using the first interval, the meteorological data are sampled, ignoring any
recorded precipitation, and the air concentration and dry deposition rate are
calculated for each receptor location of interest. The second interval specifies the
sampling rate for the hours of meteorological data during which precipitation was
recorded. This latter sampling rate is used to determine the air concentration, dry
deposition rate, and wet deposition rate. The estimates from these separate
6-7
-------
Section 6.0
Air Module
schemes are combined at the end of the model run using a weighted average
based on the number of hours sampled in each interval.
¦ Output by Particle Size. The Human Risk Module requires air concentrations
for pollutants with particle sizes <10 |im to calculate inhalation risks. ISCST3
does not output concentrations by particle size. Therefore, an option was added
to output concentration and deposition rate by particle size.
6.2.4 Calculate Constituent-specific Annual Average Concentrations and Deposition
Rates
To reduce the number of required runs, the ISCST3 model is executed using unit
emissions. The output from the model is normalized receptor-specific annual average
concentrations and deposition rates. These estimates are converted to constituent-specific annual
average concentrations and deposition rates by multiplying the values by the yearly emissions
provided as output from the source modules.
6.3 Module Discussion
6.3.1 Strengths and Advantages
The Air Module has the following strengths and advantages:
¦ ISCST concentration algorithms. The concentration algorithms have been
extensively reviewed and evaluated. The model has a long history of use by the
EPA for fine-scale modeling and has been promulgated in the Guideline on Air
Quality Modeling (U.S. EPA, 1999c).
¦ ISCST particle deposition algorithms. The particle deposition algorithms were
selected for inclusion based on an extensive comparison against other algorithms
and field data.
¦ 10-year period of record for the meteorological data. The 10-year period of
record ensures that all conditions typical of a meteorological station are modeled.
EPA guidance for air quality applications requires only 5 years of off-site
meteorological data to establish representativeness.
¦ Urban or rural conditions. The Air Module is appropriate for urban or rural
dispersion conditions, depending on the location of the source.
¦ Flexible receptor locations. Receptors may be placed anywhere in the area of
interest.
¦ Options for use of meteorological data. The Air Module provides the flexibility
to either model all hours of meteorological data or use SCIM to model only
selected hours if runtime is a constraint, which decreases runtime considerably.
6-8
-------
Section 6.0
Air Module
6.3.2 Uncertainty and Limitations
The Air Module includes the following limitations or uncertainties:
¦ Meteorological data gaps. A data quality review uncovered missing data within
the various meteorological data time series required for the 3MRA modeling
system. Programs were written to automatically find and correct these data gaps,
within the technical constraints established by Atkinson and Lee (1992). Missing
data were typically less than 2 percent of all hours, so this correction should have
little effect on the results.
¦ Lack of data on shapes and heights of WMU. Currently, none of the national-
level data sets contain data on the shape, height, or orientation of the sources. In
the absence of these data, units can be characterized as a square, rectangle, or
20-sided polygon shaped to approximate a circle.
¦ Wet deposition of vapors. Modeling of wet deposition of vapors requires
constituent-specific scavenging coefficients and separate runs for each
constituent. When modeling multiple constituents, the number of runs, as well as,
the availability of constituent-specific data may be an issue. As an alternative, a
set of gas scavenging coefficients that can be used for all constituents is stored in
the representative national data set. These coefficients are based on
approximating the gases as very small particles. This approach may lead to
underprediction of wet deposition for some gases and overprediction of others
depending on the Henry's law coefficient for the gas.
¦ Dry deposition of vapors. The Air Module is not designed to allow modeling of
dry deposition of vapors using ISCST3. Modeling of dry deposition of vapors
requires input of several contaminant-specific variables, which would require
separate runs for each chemical. Alternatively, dry deposition of vapors is
estimated in the Farm Food Chain Module (see Section 10.2.1) based on vapor-
phase concentration in air and a default vapor-phase dry deposition velocity.
Because dry deposition is calculated external to ISCST3, the plume is not
depleted within the model. However, the impact to the modeling results is not
expected to be significant.
6.4 References
Atkinson, D., and R.F. Lee. 1992. Procedures for Substituting Values for Missing NWS
Meteorological Data for Use in Regulatory Air Quality Models. U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina.
Horst, T.W. 1983. A correction to the Gaussian source-depletion model. In Precipitation
Scavenging, Dry Deposition andResuspension, H.R. Pruppacher, R.G. Semonin, W.G.N.
Slinn, eds., Volume 2. Elsevier Science Publishing Co., Inc., New York, NY, pp. 1205 -
1217.
6-9
-------
Section 6.0
Air Module
U.S. EPA (Environmental Protection Agency). 1987. 1986 National Survey of Hazardous
Waste Treatment, Storage, Disposal, and Recycling Facilities (TSDR) Database. Office
of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1995. User's Guide for the Industrial Source
Complex (ISC 3) Dispersion Models. Volume II: Description of Model Algorithms. EPA-
454/B-95-003b. U.S. Environmental Protection Agency, Emissions, Monitoring, and
Analysis Division, Office of Air Quality Planning and Standards, Research Triangle
Park, NC. September.
U.S. Environmental Protection Agency. 1999a. User's Guide for the Industrial Source Complex
(ISC3) Dispersion Models for use in the Multimedia, Mulipathway and Multireceptor
Risk Assessment (3MRA) for HWIR99. Volume I: User Instructions. Office of Solid
Waste. June.
U.S. EPA (Environmental Protection Agency). 1999b. User's Guide for the Industrial Source
Complex (ISC3) Dispersion Models for Use in the Multimedia, Multipathway and
Multireceptor Risk Assessment (3MRA) for HWIR99. Volume II: Description of Model
Algorithms. Office of Solid Waste, Washington, DC. June.
U.S. EPA (Environmental Protection Agency). 1999c. Guideline on Air Quality Models.
U.S. EPA (Environmental Protection Agency). 1999d. Air Module Pre- and Post-Processor,
Background and Implementation for the Multimedia, Multipathway, and Multireceptor
Risk Assessment (3MRA) for HWIR99. Office of Solid Waste, Washington, DC.
October.
Venkatram, A. 1998. A Simple Model for Dry Deposition and Particle Settling. Subcontractor
Progress Report 2 (including addendum). EPA Contract No. 68D70002, Work
Assignment No. 1-001.
Westat, Inc. 1987. Screening Survey of Industrial Subtitle D Establishments. Draft Final Report.
U. S. Environmental Protection Agency Contract No. 69-01-7359. Rockville, MD.
December 29.
6-10
-------
Section 7.0
Watershed Module
7.0 Watershed Module
7.1 Purpose and Scope
The Watershed Module estimates contaminant concentrations in soil resulting from aerial
deposition of contaminants throughout the area of interest (AOI) around each modeled site and
the resulting contaminant loadings to surface waterbodies from runoff and erosion. It also
estimates some hydrological inputs for the Surface Water Module (flows, eroded soil loads) and
the Vadose Zone and Aquifer Modules (infiltration rates). Figure 7-1 shows the relationship
and information flow between the Watershed Module and the 3MRA modeling system .
Key Data Inputs
• Annual infiltration rate
• Number of subbasins
• Erosivity factor
Chemical Loads, Flows
Eroded Soil Loads
Infiltration Rates
Deposition
Soil Concentrations
Rates
Soil Concentrations
Soil Concentrations
Air Module
Watershed
Module
Farm Food
Chain
Module
Terrestrial
Food Web
Module
Human
Exposure
Module
Surface
Water
Module
Aquifer
Module
Figure 7-1. Information flow for the Watershed Module in the 3MRA modeling system.
7-1
-------
Section 7.0
Watershed Module
The Watershed Module has two major functions:
1. Calculates soil contaminant concentrations and surface water loadings. The
Watershed Module predicts the fate and transport of contaminants in the
watershed subbasins to estimate soil concentration in the regional watershed and
contaminant loadings to surface water from runoff and erosion.
2. Calculates hydrological inputs. The Watershed Module calculates several
hydrological inputs for the Surface Water Module and the Aquifer Module.
7.2 Conceptual Approach
The AOI around a site consists of a set of contiguous watershed subbasins. Excluding
surface waterbody areas, all land surfaces within the AOI fall within one of the watershed
subbasins. Depending on the watershed delineation criteria, the number of watershed subbasins
constituting an AOI varies; the typical range is 6 to 25 for the current site-based data set.
Figure 7-2 shows a sample AOI that consists of all or part of 12 watersheds.
The watersheds are
delineated so that surface
runoff in one subbasin is
modeled independently of
surface runoff in other
subbasins; therefore, the
3MRA modeling system
assumes no contaminant
transport occurs from one
watershed subbasin to
another via erosion or
runoff, or via wind
suspension or
volatilization and
subsequent deposition.
Thus, the Watershed
Module models each of
the watershed subbasins at
a site independently of the
other subbasins at that
site.
The Watershed Module simulates contaminant fate and transport related only to loads
that result indirectly from the WMU (i.e., through the process of wind erosion or other
particulate suspension or volatilization from the WMU into the atmosphere and through
subsequent deposition from the atmosphere onto the surrounding regional watersheds).
Contaminant loads to a waterbody resulting from direct runoff and erosion from a WMU are
simulated by the LAU or Waste Pile Module. Watershed soil concentrations and waterbody
concentrations are a functi on of both direct and indirect contaminant loads. Thus, if a receptor is
HWIRGI3 Docket Maps
Waterbody/Water shed
Data for Site 0730407
Largest WMU type at site: If
Legend
• Sites
~ AOI
_ J WMU
IJpj1 Lakes and Wetlands
A/Streams/Rivers
I | Local Watersheds
] Watersheds
Inset show:
August 3.19991
Figure 7-2. Illustration of watersheds within an AOI.
7-2
-------
Section 7.0
Watershed Module
located in a buffer area between a WMU and the downslope waterbody (i.e., in the WMU's local
watershed), the total soil concentration to which the receptor is exposed is the aerial deposition-
related concentration estimated by the Watershed Module plus the WMU runoff and erosion-
related concentration estimated by the relevant source module.
There is significant overlap between the models used in the Watershed Module and those
used in the Land-based Source Modules described in Section 5. This section discusses only
aspects of the Watershed Module that are different from the Land-based Source Modules.
7.2.1 Calculate Soil Contaminant Concentrations and Surface Water Loadings
The Watershed Module is based on conceptual and mathematical models similar to those
used in the LAU and Waste Pile Modules, such as the Generic Soil Column Model (GSCM)
algorithm described in Section 5. The GSCM provides the solution to a one-dimensional (1-D),
partial differential equation that describes the spatial and temporal distribution of contaminants
in a porous medium subject to advective/dispersive transport and first-order losses. The
governing equation is
dC7 ctC? dC0
2 = De—2- - V> - k'C2 + Id• (7-1)
8t c dz2 dz
where
total contaminant concentration in the soil layer (g/m3)
vertical diffusion coefficient (m2/d)
infiltration rate (m/d)
vertical distance (m)
annual average wet plus dry deposition contaminant mass loading rate
(g/m3-d)
lumped first-order decay rate (d"1)—equal to the sum of the hydrolysis loss
rate, the aerobic biodegradation loss rate, and two additional first-order rate
constants that quantify the rainfall runoff and erosional loss processes.
As described in Section 5.0, Equation 7-1 is disaggregated into three component
equations—diffusion, convection, and first-order losses, each solved individually on the soil
column's numerical grid. For the Watershed Module, while the first two component equations
remain the same as in the GSCM, the third is revised to Equation 5-17 (Section 5.2.3), with
watershed parameter lddep replacing the local watershed subarea run-on load, IdThe solution
to Equation 5-17 is the same as that described for the LAU or Waste Pile Module, with the same
substitutions noted above.
C2 =
De =
VE =
z =
l^dep
k'
7-3
-------
Section 7.0
Watershed Module
After C2 in the surface layer of the soil column is determined at the end of a given time
step, Cl3 the contaminant concentration in the runoff water, averaged over the time step, is
determined using Equation 5-15 (Section 5.2.3), where all the parameters are annual averages
determined from annual average runoff flow and cumulative soil load (or mass of eroded soil).
The time-step-averaged contaminant concentration in the soil compartment, C2 in Equation 5-15,
is calculated using the following equation:
where T is the averaging time period, which is the same as the computational time step here, and
C2° is the contaminant concentration in the soil compartment at the start of the averaging period
(which is the same as at the end of the previous time step).
At the end of each year's simulation, annual average C, (g/m3) is determined and
multiplied by the annual average runoff rate (m3/d) to determine the annual average contaminant
mass load to the waterbody due to runoff and erosion (g/d).
In the Watershed Module, the depth of the soil column is a user-specified input, set at a
default of 5 cm. Each soil column layer is 1 cm thick. The surficial soil column layer (top 1 cm)
is linked to the runoff compartment during runoff events using the local watershed/soil column
algorithm described in Section 5. Similar to the presentation in Section 5, the runoff water
during a runoff event is considered as Compartment 1 and the surficial soil column layer as
Compartment 2 of the two-compartment conceptual model for the watershed/soil column
algorithm. The total (particulate-sorbed plus dissolved) contaminant concentration in the
watershed runoff is coupled to the total concentration in the soil layer.
In the subsurface layers of the soil column, the contaminant mass fate and transport
governing equation is also given by Equation 7-1; however, the total first-order loss rate (k') is
equal to the sum of the input first-order loss rates due to hydrolysis and anaerobic biodegradation
only.
The solution technique for Equation 7-1 is identical to that used for the LAU, Waste Pile,
and Landfill Modules and is described completely in Section 5. The concentration in the surface
layer of the soil column is determined at the end of each time step by solving Equation 7-1; its
average value over the time step is calculated by integrating over the time step. The contaminant
concentration in the runoff water averaged over the time step is calculated as a function of the
surface soil concentration averaged over the time step, as described in Section 5. At the end of
each year's simulation, the annual average contaminant concentration in the runoff water is
determined and multiplied by the annual average runoff rate to determine the annual average
contaminant mass load to the waterbody due to runoff and erosion.
r° id
1- exp(- k'T))+j^;(k'T-l+exp(- k'T)) k' > 0
(7-2)
k' = 0
7-4
-------
Section 7.0
Watershed Module
7.2.2 Calculate Hydrological and Soil Erosion Inputs
Streamflow. The Watershed Module uses the identical hydrology submodel described in
detail in Section 5 to estimate stormwater runoff and ground water infiltration. The hydrology
submodel is applied to entire, individual watersheds where no further spatial disaggregation
occurs.
Streamflows are assumed to be made up of both stormwater runoff and baseflow.
Baseflow is streamflow occurring during nonrunoff periods and is derived from ground water
discharge to streams or interflow (shallow infiltration flowing parallel to the ground surface).
For a given stream reach, baseflow can vary seasonally, or even near-continuously, as ground
water levels and/or interflow varies and can be estimated for a given time period by analyzing
runoff hydrographs that include runoff as well as pre- and post-runoff flows. For the purposes of
the 3MRA modeling system, within-year variability in baseflows is not estimated. Rather, a
single estimate is sought that reasonably characterizes annual average baseflow conditioned on
stream reach order (or tributary drainage area), year, and hydrologic region.
The single flow statistic that best represents annual average baseflow for a given region,
reach order, and year is an important issue. The widely available annual average streamflow
would, in general, tend to overestimate baseflow. Conversely, the common-low flow statistic,
7Q10 (the minimum 7-day average flow expected to occur within a 10-year return period, i.e., at
least once in 10 years), would tend to underestimate baseflow. Therefore, the 30Q2 low flow,
i.e., the minimum 30-day average flow occurring, on average, at least once every other year, was
selected as a reasonable estimate of annual average baseflow for any given year.
The Watershed Module estimates the 30Q2 flow based on the area of the regional
watershed, using a regression equation developed for each of the 18 hydrologic unit codes
(HUCs) in the United States. The general equation is
where
Q = a x WSA1
(7-3)
Q = 30Q2 baseflow
a = HUC-specific regression parameter
WSA = watershed area
b = HUC-specific regression parameter.
Watershed Slope for Soil Erosion. The Watershed Module uses the modified Universal
Soil Loss Equation (MUSLE), as described in Section 5, to predict soil erosion from entire
watersheds. To do so, the slope parameter of the length-slope factor presented in Section 5 must
reflect subbasin average conditions, rather than local watershed conditions as in the source
modules application. Sheet-flow slope for a subbasin is not the slope of the stream network
draining the watershed; rather, it is the average slope of the (essentially infinite) individual sheet-
flow paths that form the land surfaces of that subbasin. As presented in Williams and Berndt
7-5
-------
Section 7.0
Watershed Module
(1977), the watershed-subbasin-average slope is estimated from the following equation, which
was first proposed by Horton (1914):
•Z (LC2C + LCcq + LCyc)
S = 25 55 (7-4)
44 v 7
where
S = watershed-subbasin-average slope (percent)
Z = difference in the subbasin's maximum and minimum elevations (m)
A = total surface area of the subbasin (m2)
LC25 = total length of the contour line at the 25th percentile of Z (m)
LC50 = total length of the contour line at the 50th percentile of Z (m)
LC75 = total length of the contour line at the 75h percentile of Z (m).
7.3 Module Discussion
7.3.1 Strengths and Advantages
The Watershed Module has two overall objectives: (1) to simulate contaminant
concentrations in soils surrounding WMUs over time, and (2) to generate needed hydrological
inputs and contaminant loads required by other modules. Relative to other known models that
might have been candidates to satisfy these two objectives, the strengths and advantages of the
Watershed Module include the following:
Appropriate level of spatial resolution. Time-varying soil concentrations
surrounding the WMU arise only as a result of deposition of airborne particles
and vapors originating from the WMU. The threshold issue for designing the
spatial resolution for the Watershed Module was, how fine did this resolution
need to be, given such competing objectives such as minimizing runtime and data
collection? Although it could reasonably be expected that soil concentrations
resulting from atmospheric deposition are, on average, quite low, it is also
possible that land areas in relative proximity to the WMU might have loadings of
some concern. Therefore, the balancing act for spatial resolution was to not have
so much spatial resolution as to unduly affect runtimes and data collection, but to
also avoid as much as possible the "diluting-out" of contaminant hot spots in
close proximity to the WMU. This balance led to the approach of assuming that
each watershed subbasin being modeled has uniform spatial concentration (a
"completely-mixed" approach), which tends to "dilute" hot spots, but having the
ability to delineate "watershed subbasins" such that those in close proximity to
the WMU are relatively small, so that hot spots are reasonably well represented,
with larger (and presumably less important from a contamination standpoint)
subbasins resulting at greater distances. Thus, although each watershed subbasin
is modeled identically, the spatial resolution is highly flexible. Within a
watershed subbasin, concentrations at any time are uniform, but they can vary
among watershed subbasins. This level of spatial resolution control by the model
7-6
-------
Section 7.0
Watershed Module
user has proven to be very advantageous in modeling a wide variety of different
sites.
¦ Ability to generate needed hydrological inputs for other modules. In addition
to having the functionality to simulate different time-varying soil concentrations
(and depth profiles) within each watershed subbasin (to be used for subsequent
exposure calculations), the Watershed Module also generates hydrological inputs
and contaminant loadings for other modules over a variety of meteorological and
environmental conditions representing the continental United States. For the
Surface Water Module, subbasin-specific stormwater runoff, eroded soil loads,
stream baseflow (dry weather streamflow), and contaminant loadings are
generated as time series outputs. Of particular note, baseflows are estimated in
addition to stormwater runoff, so that the total streamflow (not just the surface
runoff portion of streamflow) is available to the Surface Water Module. The
baseflows were estimated by fitting 18 different regression models to USGS low-
flow streamflow data, each model being specific to one of 18 Hydrologic Units
(HUCs) in the conterminous United States. This ability to simulate region-
specific baseflow is extremely important as a Surface Water Module input.
¦ Consistency between the GSCM and hydrology algorithms used for Land-
based Source Modules. The algorithms used to estimate watershed subbasin soil
concentrations over time and depth (the GSCM), as well as the hydrological
algorithms (soil erosion, stormwater runoff) are identical to those used in the
Land-based Source Modules. This commonality provided not only economies
with respect to model development and ease-of-use, but, more importantly, it
ensures that underlying assumptions are consistent across these modules.
7.3.2 Uncertainty and Limitations
The Watershed Module includes the following limitations or uncertainties:
¦ GSCM limitations. As mentioned previously, the GSCM is the computational
engine for the Watershed Module. Accordingly, all uncertainties or limitations
inherent to the GSCM, and described for the LAU, Waste Pile, and Landfill
Modules (Section 5), also apply to the Watershed Module.
¦ Spatial dilution of intra-watershed hot spots. Because each watershed
subbasin is assumed to be uniform with respect to contaminant concentrations in
soil, hot spots resulting from nonuniform aerial deposition within a watershed
subbasin will not be detected.
¦ 30Q2 equivalent to baseflow. There is uncertainty in the baseflow estimates
with regard to whether the module uses the correct low flow statistic (e.g., 30Q2),
and in representing the variability in the baseflow for any given watershed. For a
given watershed, the 30Q2 estimate of constant baseflow is a point estimate,
generated from a regression model. Thus, the same baseflow will always be
estimated for a given hydrologic region and watershed size.
7-7
-------
Section 7.0
Watershed Module
¦ Sheetflow runoff assumption. The MUSLE application to estimate eroded
solids loads and associated sorbed contaminant loads in each watershed subbasin
assumes that sheet-flow runoff and erosion apply across the entire area.
Watersheds are delineated, in part, to support that assumption, but it is unlikely
that true sheetflow runoff occurs over all portions of all watershed subbasins. To
the extent that channelized flow occurs and can "short-circuit" contaminant loads
directly to adjacent waterbodies without first traversing downslope land surfaces,
the estimated contaminant loadings to waterbodies may be underestimated by the
Watershed Module. Conversely, this assumption would tend to overestimate
average soil concentrations across the watershed subbasin, leading to
overestimated soil exposures.
7.4 References
Horton, R.E., 1914. Discussion of rainfall and run-off Transactions of the American Society of
Civil Engineers, 77:369-375. December.
Williams, J.R., and H.D. Berndt. 1977. Determining the universal soil loss equation's length-
slope factor for watersheds. In: A National Conference on Soil Erosion, May 24-26,
1976. Perdue University, West Lafayette, IN. pp. 217-225, Soil Conservation Society of
America, Ankeny, 10.
7-8
-------
Section 8.0
Surface Water Module
8.0 Surface Water Module
8.1 Purpose and Scope
The Surface Water Module simulates contaminant concentrations in surface waterbodies
throughout the area of interest (AOI) around each site modeled. Inputs to the Surface Water
Module include contaminant loadings from direct air deposition onto surface waters,
contaminant loadings from runoff and soil erosion from land areas associated with sources (LAU
and waste piles only), contaminant loadings from contaminated ground water plumes that are
intercepted by surface waters, contaminant loadings in runoff and eroded soil from the
watersheds in the AOI, and hydrological inputs (flows, soil loads) from the watersheds. Surface
Water Module outputs include water column and sediment contaminant concentrations, which
are then used by the Aquatic Food Chain Module, Farm Food Chain Module, and Ecological
Exposure Module. All inputs and outputs are annual average time series. Figure 8-1 shows the
relationship and information flow between the Surface Water Module and the 3MRA modeling
system.
Key Data Inputs
• Flow Rate
• Total suspended solids
Deposition Rates
Water Column and
Sediment Concentrations
Chemical Loadings
Water Column
Soil Loadings
Concentrations
Chemical Loads, Flows,
Water Column and
Eroded Soil Loads
Sediment Concentrations
Chemical Loads from
Water Column and
Groundwater Interception
Sediment Concentrations
Surface
Water
Module
Ecological
Risk Module
Vadose Zone
and Aquifer
Modules
Ecological
Exposure
Module
Watershed
Module
Farm Food
Chain
Module
Land-based
Source
Modules
Air Module
Aquatic Food
Web Module
Figure 8-1. Information flow for the Surface Water Module in the 3MRA modeling system.
8-1
-------
Section 8.0
Surface Water Module
The major tasks performed by the Surface Water Module to simulate contaminant
concentrations throughout the surface waterbody network in the AOI are as follows:
1. Constructs waterbody network. For each AOI, the Surface Water Module
identifies the streams, wetlands, ponds, lakes, and bay reaches to be modeled, and
assigns lengths, areas, depths, and volumes to them.
2. Routes hydraulic flow and solids through the waterbody network. The
Surface Water Module conducts water and solids balances for each waterbody in
each year of the simulation.
3. Constructs and solves the mass balance equations describing contaminant
fate and transport throughout the waterbody network. The Surface Water
Module calculates the contaminant concentration in each waterbody for each
year. Outputs include total water column concentration and dissolved
concentration.
8.2 Conceptual Approach
The Surface Water Module is based on the EPA legacy model Exposure Analysis
Modeling System (Exams II), which has been thoroughly verified and validated in numerous
applications (see Volume III of this report). Exams II was extended for use in the Surface Water
Module primarily through development of the pre- and postprocessor, EXAMS 10. EXAMS 10 is
the interface between Exams II and the rest of the 3MRA modeling system. It reads data from
other 3MRA modules and databases, builds the Exams input files describing the waterbody
environment and chemical properties, builds a command file that specifies the contaminant
loading history, and controls the Exams simulation. Exams 10 passes control to Exams II,
which conducts the simulation and produces intermediate results files. Exams 10 then processes
the intermediate files and passes the output data back to the proper 3MRA modeling system
database. The complete system consisting of Exams II and Exams 10 is denoted in this
discussion as the Surface Water Module. A summary of Exams II—the science and
computational core of the Surface Water Module—is presented in the accompanying text box.
The full capabilities of Exams II are explained in Burns et al. (1982) and Burns (1997).
8.2.1 Construct Waterbody Network
Each Monte Carlo iteration of the 3MRA system represents a unique site covering the
AOI around a waste management unit (WMU). This AOI is defined by a radius of 2 km in the
sample 3MRA data set. One or more contiguous waterbody networks may be located within
each AOI. For each of the 201 sites currently contained in the 3MRA modeling system database,
waterbody networks were defined using the EPA Reach File system supplemented with digital
elevation modeling. Each waterbody network is divided into reaches, which are characterized
and stored in the site database.
8-2
-------
Section 8.0
Surface Water Module
Summary of Exams II
Exams II combines contaminant loadings, transport, and transformations into a set of differential
equations using the law of conservation of mass as an accounting principle. It accounts for all the contaminant
mass entering and leaving a system as the algebraic sum of external loadings, transport processes that export the
compound from the system, and transformation processes within the system that convert the contaminant to
daughter products. (Transformation from parent to daughter products is implemented only for mercury in the
Surface Water Module.)
Exams II represents each waterbody by a set of segments or distinct zones in the system. The program
is based on a series of mass balances for the segments that give rise to a single differential equation for each
segment. Working from the individual transport and transformation process equations, Exams II compiles an
overall equation for the net rate of change of contaminant concentration in each segment. The resulting system
of differential equations describes the mass balance for the entire system, which is then solved by the method of
lines.
Exams II includes process models of the physical, chemical, and biological phenomena governing the
transport and fate of compounds. Each of the unit process equations used to compute the transformation
kinetics of contaminants accounts for the interactions between the chemistry of a compound and the environ-
mental forces that shape its behavior in aquatic systems. This "second-order" or "system-independent"
approach allows one to study the fundamental chemistry of compounds in the laboratory and then, based on
independent studies of the levels of driving forces in aquatic systems, evaluate the probable behavior of the
compound in systems that have never been exposed to it. Most of the process equations are based on standard
theoretical constructs or accepted empirical relationships. The user can specify reaction pathways for the
production of transformation products of concern, whose further fate and transport can then be simultaneously
simulated by Exams II.
Exams II contains process modules for several chemical reactions. Equilibrium reactions are used for
sorption and ionization. Kinetic reactions are used for volatilization, hydrolysis (acid, base, and neutral),
biodegradation (water column and sediments), photolysis, oxidation, and reduction. (The Surface Water
Module does not consider photolysis or oxidation/reduction.) Exams II uses these modules as determined by
the input chemical properties. Exams II has been designed to accept standard water quality parameters and
system characteristics that are commonly measured by limnologists throughout the world and contaminant data
sets conventionally measured or required by EPA regulatory procedures.
The contaminant fate algorithms in Exams II model sorption to suspended solids, biotic solids, and
sediment solids, but Exams II does not simulate a solids balance. Solids concentrations are specified as input
data. The effects of settling and resuspension on contaminant fate are accounted for in a bulk sediment-water
exchange term.
Exams II incorporates a few major assumptions. The model was designed to evaluate the
consequences of longer-term, primarily time-averaged contaminant loadings that ultimately result in trace-level
contamination of aquatic systems. Exams II generates a steady-state, average flow field (long-term or monthly)
for the ecosystem. The program cannot therefore fully evaluate the transient, high concentrations that arise
from chemical spills, although spills under average hydrological conditions can be studied. An assumption of
trace-level contaminant concentrations was used to design the process equations. The contaminant is assumed
not to radically change the environmental variables that drive its transformations. Exams II uses linear sorption
isotherms and second-order (rather than Michaelis-Menten-Monod) expressions for biotransformation kinetics,
which are known to be valid for low concentrations of pollutants. Sorption is treated as a thermodynamic or
constitutive property of each compartment in the system; that is, sorption-desorption kinetics are assumed to be
rapid compared with other processes. Although this assumption may be violated by extensively sorbed
contaminants, these contaminants tend to be captured by benthic sediments, where their release to the water
column is controlled by benthic exchange processes.
8-3
-------
Section 8.0
Surface Water Module
From the 3MRA modeling system site database, stream reaches are assigned lengths, and
wetland, pond, lake, and bay reaches are assigned surface areas. Site-specific reach depths and
volumes were not available. Instead, depths were assigned to water column and sediment
compartments for wetlands, ponds, lakes, and bays based on national distributions. The Surface
Water Module calculates volumes from the site-specific surface areas and the estimated depths.
For stream reaches, the Surface Water Module calculates widths and depths using discharge
coefficients based on empirical observations of stream hydrogeometric relationships (Leopold
and Maddox, 1953) and then calculates volumes using site-specific reach lengths.
8.2.2 Route Hydraulic Flow Through the Waterbody Network
Water Balance. The Surface Water Module conducts a water balance for each simulated
year, routing incoming flows from the headwater reaches down through the network and out of
the exiting reach. For a given reach in a given year, the outflow is the sum of all the inflows
minus evaporation, as shown in the water balance equation:
NumTribr NumWSr (_ jr |
Qr,t = X QTrib,,r,t + X QRun°ff,,r,t x WSFracir + QBase r + —'1QQ^ 365 (8-!)
where
Qr t = average outflow from reach r during year l (m3/d)
NumTribj. = number of upstream reaches flowing into reach r
QTribi r t = average upstream outflow from reach tributary to reach r during
year t (m3/d)
NumWSr = number of adjacent watersheds flowing into reach r
QRunoff; r t = annual average surface runoff inflow from watershed i
immediately adjacent to reach r during year l (m3/d)
WSFracir = area fraction of adjacent watershed i contributing to reach r
QBaser = average constant baseflow flow to reach r (m3/d)
Rt = precipitation for year l (cm/yr)
Et = evaporation for year l (cm/yr)
ASr = surface area for reach r under baseflow conditions (m2).
The Surface Water Module also checks and ensures positive flow in every reach for
every year. If evaporation exceeds the total inflow to a reach for a year, the evaporation in that
reach is reduced to a level supporting an outflow equivalent to precipitation of 0.5 mm/yr over
the entire contributing watershed.
Long-term average baseflow from each adjacent watershed is calculated in the Watershed
Module using empirical correlations based on watershed surface area and regional location (see
Section 7). The input data map these watersheds to the appropriate reaches and also specify the
fraction of each contributing watershed's baseflow that should be associated with each reach.
Total reach baseflow is then calculated by the fraction-weighted sum over all contributing
watersheds. This baseflow estimate is derived from observed stream flow data and accounts for
both seepage inflow and evaporation loss during dry periods.
8-4
-------
Section 8.0
Surface Water Module
This water balance is conducted for each simulated year on a reach-specific basis by
Exams 10. The resulting internal flows are used to calculate stream hydrogeometry and the
waterbody solids balance. Information on stream widths and depths and all hydrologic inflows
is used to recalculate the water balance on a compartment-specific basis for use in the chemical
advection calculations.
Solids Balance. Exams 10 conducts a solids balance for each simulated year, routing
incoming solids from the headwater reaches down through the network and out of the exiting
reach. The total suspended solids (TSS) in a reach may originate in erosion from adjacent
watershed subbasins, from internal bank erosion, or from internal production of biotic solids.
Solids deposition and burial are not simulated in this version of the Surface Water Module;
consequently, TSS is modeled as a conservative substance. Therefore, for a given reach in a
given year, the average solids flux per year through the reach is the sum of all the solids mass
inflows, including upstream loads, loads delivered from adjacent surface erosion, and loads
generated internally to the reach:
NumTribr NumWSr
TSSr,t = Z QTribi,y,t X TSSTribi,r,t + Z TSSRUn°ffi,r,t X WSFrClC^ + Qr , X
/=l ' ' /=1
where
TSS,t
= average solids load exiting reach r in year t (g/day)
QTribir,
= average upstream outflow from reach tributary to reach r during
year t (m3/d)
TSSTribirt
= TSS concentration exiting upstream reach tributary to reach r, in
year t (g/m3)
TSSRunoffirt
= solids load delivered from erosion of watershed r, immediately
adjacent to reach r, in year l (g/day)
SolBnk,.
= TSS concentration from incremental bank erosion in reach r (g/m3)
Phyt (TIr)
= biotic carbon concentration for trophic index for reach r (g-C/m3)
TIr
= trophic index for reach r (unitless)
FocBio
= organic carbon fraction of biotic solids (g-C/g-solids)
The first right-hand term in this equation accounts for upstream tributary solids loads.
The second term accounts for adjacent watershed erosion loads. This erosion load is generated
by the Watershed Module and already incorporates an area-dependent sediment delivery ratio.
The third term accounts for bank erosion loading, where the TSS concentration from incremental
bank erosion is the resulting incremental concentration in reach r. This is set to 5 g/m3 for
stream reaches and 1 g/m3 otherwise. The final term accounts for internal biotic production of
solids. The trophic index is related to the net biotic production.
I'hylilJ )
SolBnk + (8-2)
FocBio v '
8-5
-------
Section 8.0
Surface Water Module
8.2.3 Construct and Solve the Mass Balance Equations Describing Contaminant Fate and
Transport throughout the Waterbody Network
The primary function of the Surface Water Module is to simulate the spatial and temporal
contaminant concentrations for both fate within the individual waterbody segments (or
compartments) and transport among these compartments that make up the overall waterbody
network.
Contaminant fate and transport are calculated on a compartment-specific basis, by
accounting for a set of transport and transformation processes applied within a mass balance
framework. The Surface Water Module is based on a series of mass balances, which give rise to
a single differential equation for each compartment. Working from the transport and
transformation process equations, the Surface Water Module generates an equation for the net
rate of change of contaminant concentration in each compartment. The resulting system of
differential equations describes a mass balance for the entire waterbody network. These
equations have the following general form:
- L.+ L,- r*K*C (8-3)
where
V = volume of water in the compartment (m3)
C = total contaminant concentration (mg/L)
Le = total external loading to the compartment (g/h)
L; = total internal loading to the compartment resulting from contaminated flows
among system compartments and the formation of chemical reaction products
(g/h)
K = overall pseudo-first-order loss constant that expresses the combined effect of
transport and transformation processes (i.e., advection, dispersion,
volatilization, biodegradation) that decrease contaminant concentration (h"1)
The Surface Water Module provides for a total of five kinds of compartment loadings:
¦ Point source or stream-borne loadings,
¦ Nonpoint source loadings,
¦ Contaminated ground water seepage,
¦ Precipitation washout from the atmosphere, and
¦ Spray drift (or miscellaneous) loading.
The inputs include monthly and annual average loadings to all appropriate compartments.
The Surface Water Module evaluates the status of individual loadings entering the
compartments to prevent conditions outside of its numerical operating range. For example, the
Surface Water Module makes no provision for crystallization of the compound from solution,
nor for dissolution of a compound from a solid or liquid phase. In addition, reaction
nonlinearities potentially present at high contaminant concentrations are not incorporated into
8-6
-------
Section 8.0
Surface Water Module
the model. The Surface Water Module calculates the dissolved contaminant concentration
within each input flow, using receiving compartment pH and temperature to account for
ionization and using entrained sediment loads to account for sorption. If the dissolved
concentration for any of the species is greater than half its solubility, then the simulation is not
used by the 3MRA modeling system.
The system of mass balance differential equations is solved numerically, using the
method of lines algorithm. The solution to the system of mass balance equations yields the
distribution of contaminant concentration throughout the surface waterbody network (water
column and sediments) over time. Three temporal options are available for Exams II:
¦ Steady-state (no change over time),
¦ Quasi-dynamic with time-constant environmental data, and
¦ Quasi-dynamic with monthly-varying environmental data.
The Surface Water Module implements the quasi-dynamic mode with time-constant
environmental data. In this mode, the dynamic simulations consider annually-varying flows and
loadings but assume other environmental conditions are at their long-term average values.
8.3 Module Discussion
8.3.1 Strengths and Advantages
The Surface Water Module has the following strengths and advantages:
¦ Use of a proven chemical fate model. The Surface Water Module uses Exams
to simulate chemical fate for the scenarios presented by the 3MRA framework.
Exams is an EPA legacy model that has been used for two decades to assess the
fate of organic chemicals in aquatic systems. It has been used routinely to
evaluate pesticide exposure scenarios for the Office of Pesticide Programs.
During its extensive application history, EXAMS has been tested in a wide variety
of waterbodies for a wide range of organic chemicals. The code has been
verified, and its predictions have been confirmed in several site-specific
evaluations published in the technical literature.
¦ Applicable to a variety of conditions. Exams is a robust model that solves the
governing differential equations effectively under a variety of conditions.
8.3.2 Uncertainty and Limitations
The following uncertainties and limitations are inherent in the Surface Water Module:
¦ Limited temporal resolution of numerical solution. The current 3MRA
modeling system methodology calls for repeated, relatively long simulations (e.g.,
200 or more years). Accordingly, run time must be relatively fast, which
precludes small time steps that might otherwise give insight into contaminant
behavior on a finer temporal scale.
8-7
-------
Section 8.0
Surface Water Module
¦ Limited resolution of loadings and flows. Use of annual-average loadings and
flows rather than daily loadings and flows leads to calculated annual-average
concentrations that are biased high, depending on the correlation between flow
and loading at a particular site. This bias is somewhat mitigated for reactive and
volatile contaminants where the loss rate is proportional to the concentration.
¦ Limited site-specific data. Use of national distributions rather than site-specific
environmental data could cause calculated concentrations to be low or high at a
given location, with no known general bias.
¦ Simplistic solids balance. The simple solids balance overestimates suspended
solids concentrations slightly in streams and more significantly in ponds,
wetlands, and lakes. Calculated total water column contaminant concentrations
will be high, whereas the dissolved contaminant fraction will be low. The net
result for dissolved water column contaminant concentrations, which are used for
fish exposure, is not expected to be biased significantly high or low.
¦ Procedure for preventing drying of surface water reaches. For sites that
experience periodic drying (due to limited inflows), a small default positive flow
equivalent to 0.5 mm per year of direct precipitation onto the waterbody surface is
maintained to keep the model functioning. This procedure conducts contaminant
loads downstream within a remnant waterbody reach rather than within runoff
over a dry bed or subsurface flow within the bed. Although the mass balance is
maintained, the contaminant and solids concentrations will tend to be elevated
within the remnant reach. These elevated concentrations are probably realistic for
years in which evaporation exceeds all hydrologic inflows.
8.4 References
Burns, L. A., D.M. Cline, and R.R. Lassiter. 1982. Exposure Analysis Modeling System
(EXAMS): User Manual and System Documentation. U.S. Environmental Protection
Agency, Athens, GA. EPA-600/3-83-023.
Burns, L.A. 1997. Exposure Analysis Modeling System (EXAMS II): User's Guide for
Version 2.97.5. U.S. Environmental Protection Agency, Athens, GA. EPA-600/R-
97/047.
Leopold, L.B. and T. Maddox. 1953. The Hydraulic Geometry of Stream Channels and Some
Physiographic Implications. Professional Paper 252, U.S. Geological Survey,
Washington, DC.
-------
Section 9.0
Vadose Zone and Aqui fer Modules
9.0 Vadose Zone and Aquifer Modules
9.1 Purpose and Scope
The Vadose Zone and Aquifer Modules simulate the subsurface movement of
contaminants in leachate from surface impoundments, landfills, land application units (LAUs),
and waste piles to downgradient drinking water wells and waterbodies. The modules are not
used for aerated tanks, because tanks are assumed not to leak. Detailed information on the
Vadose Zone and Aquifer Modules can be found in the technical background document (U.S.
EPA, 1999e). Figure 9-1 shows the relationship and information flow between the Vadose Zone
and Aquifer Modules and the 3MRA modeling system.
Watershed
Module
Infiltration Rates
Source
Modu es
Chemical Fluxes
Infiltration Rates
Key Data Inputs
Well location
Fraction organic carbon
K
Vadose Zone
and Aquifer
Modules
Chemical Loads from
Subsurface Interception
Ground Water
Concentrations
Subsurface
Concentrations
Surface
Water
Module
Human
Exposure
Module
Farm Food
Chain
Module
Figure 9-1. Information flow for the Vadose Zone and Aquifer Modules
in the 3MRA modeling system.
The Vadose Zone and Aquifer Modules simulate the fate and transport of dissolved
contaminants from a point of release at the base of a WMU, through the underlying soil, and
through a surficial aquifer (or ground water source). Module outputs include ground water
contaminant concentrations in wells, which are used by the Human Exposure Module to estimate
9-1
-------
Section 9.0
Vadose Zone and Aquifer Modules
exposures through drinking water and showering, and by the Farm Food Chain Module to
estimate contaminant concentrations in beef and milk from farm well use; and contaminant
fluxes into waterbodies, which are used by the Surface Water Module, along with contaminant
fluxes from atmospheric deposition and overland flow, to estimate contaminant concentrations in
streams, lakes, and wetlands.
The Vadose Zone and Aquifer Modules are used by the 3MRA modeling system only if
there are wells or downgradient streams, lakes, or wetlands at a site. Waterbodies are
downgradient if they are in the direction of ground water flow away from the WMU.
The Vadose Zone and Aquifer Modules perform the following functions:
1. Model vadose zone flow and transport. The one-dimensional (1-D) Vadose
Zone Module simulates infiltration and dissolved contaminant transport, by
advection and dispersion, leaching from the bottom of a WMU through the soil
above the water table (i.e., the vadose zone) to estimate the contaminant and
water flux to the underlying ground water.
2. Model ground water flow and transport. The pseudo-3-D Aquifer Module
simulates ground water flow and contaminant transport, by advection and
dispersion, from the base of the vadose zone to estimate contaminant
concentrations in drinking water wells and contaminant discharge fluxes to
intercepted waterbodies.
3. Model subsurface chemical reactions. Both the Vadose Zone and Aquifer
Modules simulate sorption to soil or aquifer materials and biological and
chemical degradation, which can reduce contaminant concentrations as they move
through soil and ground water. In cases where degradation of a contaminant
yields other contaminants that are of concern, the modules can account for the
formation and transport of up to six different daughter and granddaughter
degradation products. For metals, the modules use sorption isotherms that allow
adjustment of sorption behavior to account for varying metal concentrations and
geochemical conditions.
The Vadose Zone and Aquifer Modules in the 3MRA modeling system were extracted
from EPA's Composite Model for Leachate Migration with Transformation Products
(EPACMTP) (U.S. EPA, 1996 a,b,c; 1997). EPACMTP is used in EPA regulatory efforts by
OSW and has been subject to extensive peer review and public comment. EPACMTP is a tool
used routinely to predict potential ground water pathway exposure at a downstream receptor well
for regulatory development purposes.
9.2 Conceptual Approach
Figure 9-2 illustrates how the Vadose Zone and Aquifer Modules work together to
calculate receptor well contaminant concentrations and contaminant fluxes to downgradient
waterbodies. The Vadose Zone Module and the Aquifer Module are described below, followed
by a discussion of the chemical reaction modeling that operates in both modules. Additional
9-2
-------
Section 9.0
Vadose Zone and Aqui fer Modules
Receptor Well Concentration
Mass Flux and Infiltration from
Source Module
Receptor
Well \
Vadose Zone Module
Areal
Recharge
yyyy
Concentration at Water Table
Gaining Stream
Screen
Saturated Zone Module
Figure 9-2. Conceptual diagram of Vadose Zone and Aquifer Modules.
details on the Vadose Zone and Aquifer Modules, including governing equations and detailed
input/output specifications, can be found in the 3MRA modeling system background documents
(U.S. EPA, 1999a-e).
9.2.1 Vadose Zone Module
The Vadose Zone Module simulates the flow and transport of contaminants in leachate
from the upper boundary of the vadose zone at the base of the WMU source to the lower vadose
zone boundary at the water table. The Vadose Zone Module estimates contaminant flux to the
Aquifer Module given leachate flows and fluxes from the 3MRA source modules. Its outputs are
the long-term, steady-state infiltration rate and the annual time series of contaminant
concentrations that are used as inputs to the Aquifer Module.
Vadose Zone Flow. The Vadose Zone Module performs 1-D analytical and numerical
solutions for water flow and contaminant transport in unsaturated soil underlying a WMU
source. The model assumes that flow in the vadose zone is at steady-state and flows vertically
from underneath the source toward the water table.1 The Vadose Zone Module receives a time
series of annual average infiltration rates and contaminant mass fluxes from the source modules.
1 Because water flow in the vadose zone is predominantly gravity driven, the vertical flow component
accounts for most of the fluid flux between the WMU source and the water table.
9-3
-------
Section 9.0
Vadose Zone and Aquifer Modules
Because the Vadose Zone Module assumes steady-state flow, it calculates a long-term average
infiltration rate from the full series of annual average infiltration rates. At the same time, the
model calculates an annual average time series of leachate concentrations, while conserving
contaminant mass with respect to the full series of contaminant mass fluxes output from the
source modules.
The vadose zone flow model assumes that the unsaturated soil is a uniform porous
medium and that the infiltrating flow rate (7) is governed by Darcy's law (Freeze and Cherry,
1979):
I = -Ks k™ (di|//dz - 1) (9-1)
where
i|/ = pressure head (cm)
z = depth in the soil column (positive downward) (cm)
Ks = saturated hydraulic conductivity (cm/h)
k^, = relative permeability (unitless).
Solution of this equation for unsaturated soil conditions requires stipulation of the
relationships between
¦ The relative permeability and the volumetric water content of the porous medium
and
¦ The volumetric water content and the pressure head.
The Vadose Zone Module assumes that these relationships follow the equations given by Van
Genuchten (1980). The governing equations for these relationships are given in U.S. EPA
(1999e). Solution methods for these equations can be found in the EPACMTP background
document (U.S. EPA, 1996a-c).
Vadose Zone Transport. The Vadose Zone Module simulates transient2 contaminant
transport through the unsaturated zone by advection and dispersion. The model assumes that the
unsaturated zone is initially free of contamination and that contaminants migrate downward
along with the leachate flow from the WMU. Chemical reactions modeled include single- or
multiple-chemical chain decay reactions, and linear or nonlinear sorption. The chemical reaction
modeling is described in Section 9.2.3.
The vadose zone transport model simulates the 1-D transport of contaminants through the
soil column using the advection-dispersion equation of Huyakorn and Pinder (1983). This
equation estimates contaminant degradation using a first-order decay constant that calculates
mass fractions of both parent and daughter compounds. The effect of equilibrium sorption of a
species is expressed by a retardation coefficient that can simulate sorption using either linear or
nonlinear (Freundlich) sorption isotherms. The governing equations for the vadose zone
2 Contaminant transport can vary year to year as leachate concentration changes.
9-4
-------
Section 9.0
Vadose Zone and Aquifer Modules
transport model can be found in U.S. EPA
(1999e). Solution methods for these
equations, which are semi analytical for
organic chemicals and analytical for
nonlinear metals, are described in detail in the
EPACMTP background documents
(U.S.EPA, 1996a-c).
Vadose Zone Module Assumptions,
Inputs, and Outputs. Key assumptions for
the Vadose Zone Module are summarized in
the text box. Vadose Zone Module inputs
include
¦ Contaminant fluxes and
infiltration rates from the
3MRA source modules;
¦ Soil hydrologic properties
(saturated hydraulic
conductivity, total porosity,
residual water content, and
Van Genuchten water-retention parameters);
¦ Soil bulk density, pH, temperature, and organic matter content;
¦ Vadose zone thickness (depth to ground water); and
¦ Contaminant-specific degradation and sorption variables (see Section 9.2.3).
The module produces a time series of contaminant concentrations in the infiltrate passed to the
aquifer, a long-term average infiltration rate, and the duration of the WMU source of
contamination, all of which are used as input to the Aquifer Module. Detailed specifications of
these inputs and outputs can be found in U.S. EPA (1999e).
9.2.2 Aquifer Module
The Aquifer Module simulates 1-D ground water flow and pseudo-3-D contaminant
transport to calculate ground water contaminant concentrations at downgradient3 drinking water
wells and intercepting surface waterbodies (i.e., streams, rivers, lakes, and wetlands). The
Aquifer Module estimates these concentrations and fluxes given infiltration flow rate and
contaminant concentrations from the Vadose Zone Module. Its primary outputs include time
series of annual average contaminant concentrations at each downgradient receptor well (used by
Key assumptions of the Vadose Zone Module
¦ Contaminants are released from a square WMU
source; there is no contaminant flux outside the
WMU
¦ Flow and transport are steady-state and are 1-D,
with year-to-year changes in leachate
concentration from the source.
¦ Flow is vertical with no horizontal component.
¦ The soil is initially free of contamination.
¦ Soil is a uniform porous medium with uniform
properties; there are no soil layers or preferential
pathways such as fractures or soil macropores.
¦ Contaminant transport is by advection and
dispersion only; there is no facilitated transport by
colloids or nonaqueous-phase liquid (NAPL).
¦ Contaminants are in the aqueous and sorbed
phases only; there are no mass transfer processes
between phases other than adsorption onto soil
particles.
¦ There is no volatilization from the unsaturated
zone (i.e., no gas-phase release or transport).
¦ NAPLs (e.g., oil) are not present.
3 "Downgradient" refers to features (wells, waterbodies) that are downstream from the WMU with respect
to the ground water flowpath.
9-5
-------
Section 9.0
Vadose Zone and Aquifer Modules
the Human Exposure and Farm Food Chain Modules) and a time series of contaminant fluxes at
any downgradient intercepting waterbody (used by the Surface Water Module).
Ground Water Flow. The Aquifer Module simulates ground water flow in a surficial
(unconfined) aquifer of constant thickness using a 1-D ground water flow submodel and a
pseudo-3-D transport submodel. The model accounts for the effects of infiltration from the
WMU and regional recharge downgradient of the WMU on the magnitude and direction of
ground water flow.4 The concept accounts for regional flow in the horizontal direction driven by
a regional hydrologic gradient, along with vertical disturbance of this regional flow by water
infiltrating to the surficial aquifer from the overlying vadose zone and WMU.
The Aquifer Module uses a 1-D, steady-state solution for predicting hydraulic head and
Darcy velocities that assumes that the aquifer is composed of a uniform porous media, and that
ground water flow is governed by Darcy's law (Bear, 1972). This solution begins with the
classic 3-D governing equation for ground water flow that is used in EPACMTP:
Kx (32H/3x2) + Ky (32H/3y2) + Kz(32H/3z2) = 0 (9-2)
where
H = hydraulic head (cm)
Kx = hydraulic conductivity in the longitudinal direction (cm/hr)
Ky = hydraulic conductivity in the horizontal-transverse direction (cm/hr)
Kz = hydraulic conductivity in the vertical direction (cm/hr).
The 1-D flow equation used in the Aquifer Module is derived by setting the transverse
hydraulic conductivity, Ky, equal to zero and invoking the Dupuit- Forchheimer assumption
(Bear, 1972), in conjunction with vertical boundary conditions to account for infiltration and
recharge. Details on this flow equation, its derivation, and its solution can be found in U.S. EPA
(1999e).
The ground water flow model accounts for ground water mounding beneath the WMU by
allowing different recharge rates underneath and outside the source area. Within the model,
ground water mounding beneath the source is represented in the flow system by increased
hydraulic head values at the top of the aquifer. This approach is reasonable so long as the height
of the mound is small relative to the thickness of the saturated zone.
Ground Water Transport. The Aquifer Module simulates contaminant transport
through ground water by use of a pseudo-3-D advection and dispersion submodel. The model
assumes that ground water is initially free of contamination; that is, initial aquifer contaminant
concentration and concentration gradients along the downstream and upstream boundaries are set
to zero. Contaminants enter the aquifer only from the vadose zone immediately underneath the
4 Recharge is provided by the 3MRA Watershed Module as a time series of annual average recharge rates
with units of m/d. Because the aquifer model requires a steady-state recharge rate with units of m/yr, the model
calculates an effective long-term recharge rate as the average of the time series recharge rates received from the
Watershed Module.
9-6
-------
Section 9.0
Vadose Zone and Aquifer Modules
WMU, which is modeled as a square, horizontal plane source. Chemical reactions modeled
include single- or multiple-chemical chain decay reactions, and linear or nonlinear sorption. The
chemical reaction modeling is described in Section 9.2.3.
The Aquifer Module simulates the advective-dispersive transport of dissolved
contaminants in one dimension with dispersion in the other two dimensions added analytically
(pseudo-3-D). The model is based on the 3-D advection-dispersion governing equation from
Huyakorn and Pinder (1983), with a simplified solution to model advection in one dimension
(horizontally) and dispersion in three dimensions. Although this may not be as accurate as a
fully 3-D solution, it is much more computationally efficient and therefore more suitable for the
large-scale Monte Carlo simulations for which the 3MRA modeling system was designed.
The ground water transport equation estimates contaminant degradation using a first-
order decay constant that calculates mass fractions of both parent and daughter compounds. The
effect of equilibrium sorption of a species is expressed by a retardation coefficient, R, that can
simulate sorption using linear sorption isotherms.5 Governing equations for the ground water
transport equation can be found in U.S. EPA (1999e), which also describes the solution methods
applied to develop the Aquifer Module.
Aquifer Fractures and Heterogeneity. The basic Aquifer Module described above
assumes that aquifers are homogeneous and isotropic, and it therefore does not simulate flow and
transport under the fractured or heterogenous subsurface conditions that are common across the
United States. Because modeling fractures and heterogeneity is complex, EPA developed an
indirect approach to address fracture flow, which tends to increase the rate of migration
compared to nonfractured settings, and heterogeneity, which can cause a plume to break into
fingers of higher concentration ground water that can increase the concentration at a well.
To address fracture flow conditions, the Aquifer Module uses an equivalent porous media
(EPM) approach, which applies uniform porous media analogues to a fractured aquifer flow
field. The approach involves developing "fracture multiplier" distributions for several classes of
fractured hydrogeologic settings. For a particular Monte Carlo realization at a fractured site, the
3MRA modeling system selects a multiplier from the appropriate distribution and applies it to
increase the porous media hydraulic conductivity selected for the site. Additional details on the
development and implementation of this approach can be found in U.S. EPA (1999d,e).
To incorporate effects of heterogeneity in aquifers, EPA first assessed the effects of
heterogeneity on receptor well concentrations (U.S. EPA, 1999b). Based on this study, EPA
developed an algorithm and database to estimate the effects of heterogeneity on well
concentrations output by the Aquifer Module. For a particular Monte Carlo realization, the
3MRA modeling system selects input values from the database and applies the algorithm to
adjust the well concentration outputs. Additional details on the development and application of
this method can be found in U.S. EPA (1999b,e).
5 The aquifer model differs from the vadose zone model in that it does not allow the use of nonlinear
Freundlich sorption isotherms.
9-7
-------
Section 9.0
Vadose Zone and Aquifer Modules
Aquifer Module Assumptions,
Inputs, and Outputs. Key assumptions for
the Aquifer Module are summarized in the
text box. Aquifer Module inputs include
¦ WMU length, width, and
location;
¦ A steady-state, long-term
annual infiltration rate and a
time series of annual average
contaminant concentrations
from the Vadose Zone
Module;
¦ A time series of annual-
average recharge rates from
the Watershed Module;
¦ Aquifer properties, including
saturated hydraulic
conductivity, regional
gradient, thickness, pH,
organic carbon content, and
temperature;
¦ The ratio of horizontal to
vertical hydraulic
conductivity;
¦ Longitudinal, horizontal-
transverse, and vertical
dispersivities;
¦ Receptor well locations and
screen depths below the water
table surface;
Key assumptions of the Aquifer Module
¦ The aquifer is unconfined and subject to recharge
from the ground surface.
¦ The upper aquifer boundary is the water table; the
lower aquifer boundary is an impermeable
confining layer.
¦ The aquifer has a uniform (regional) hydraulic
gradient and a constant saturated thickness.
¦ Aquifer materials are uniform porous media that
are isotropic except for hydraulic conductivity,
which can vary between the vertical and
horizontal directions.
¦ The downgradient intercepting waterbody does
not fully alter the ground water flow pattern.
¦ The ground water is initially free of
contamination.
¦ Contaminants are released from a square WMU
source; there is no contaminant flux outside the
WMU.
¦ Recharge of contaminant-free water occurs from
the watershed outside of the WMU.
¦ Ground water flow is steady-state, but
contaminant concentrations can vary year to year.
¦ Preferential flow through fractures and aquifer
heterogeneities is addressed by adjusting aquifer
hydraulic conductivity or well concentrations.
¦ The model does not apply to solution openings in
karst limestone.
¦ Contaminants are transported by advection and
dispersion only; there is no facilitated transport by
colloids orNAPL.
¦ Contaminants are in the aqueous and sorbed
phases only; there are no mass transfer processes
between phases other than adsorption onto soil
particles.
¦ There is no volatilization from the water table
(i.e., no gas-phase release or transport).
¦ NAPLs (e.g., oil, halogenated solvents) are not
present.
Location coordinates for downgradient intercepting waterbodies;
Direction of regional ground water flow, measured clockwise from due north;
Contaminant-specific degradation and sorption variables (see Section 9.2.3); and
Termination criteria for ending the model simulation, including the maximum
simulation time.
9-8
-------
Section 9.0
Vadose Zone and Aquifer Modules
The Aquifer Module outputs time series of annual average ground water contaminant
concentrations at downgradient receptor wells within the ground water plume, which are used by
the Human Exposure and Farm Food Chain Modules. It also outputs a time series of
contaminant fluxes into a downgradient intercepting waterbody that is read by the Surface Water
Module. Detailed specifications of aquifer model inputs and outputs can be found in U.S. EPA
(1999e).
9.2.3 Chemical Reaction Modeling
The Vadose Zone and Aquifer Modules include code to simulate sorption to soil or
aquifer materials and biological and chemical degradation, processes that can reduce
contaminant concentrations as they move through soil and ground water. The Vadose Zone and
Aquifer Modules use very similar methods to model these processes. The modeling approach
differs between organic chemicals, which can degrade but follow simple sorption relationships,
and metals, which do not degrade but require a more complicated approach to model sorption.
Organic Chemical Reactions. For organic chemicals, the Vadose Zone and Aquifer
Modules simulate decay reactions for single compounds or multiple-compound chains, along
with sorption to solid soil and aquifer components. Degradation reactions addressed include
both chemical and biological transformation processes, with all transformation reactions
represented by first-order decay processes. The models account for chemical and biological
transformations by combining first-order degradation rates derived for chemical hydrolysis and
biological degradation.
The transport of organic chemicals is influenced in part by hydrolysis. Chemical
hydrolysis is modeled using acid-catalyzed, base-catalyzed, and neutral contaminant-specific
hydrolysis rates (i.e., the Arrhenius equation). These rate constants are all influenced by ground
water temperature, while the acid- and base-catalyzed rate constants are also influenced by pH.
If chemical degradation by-products are hazardous and their contaminant-specific parameters are
known, they can also be modeled in the simulation as part of a decay chain.
To model biological degradation, the Vadose Zone Module uses aerobic biodegradation
rates and the Aquifer Module uses anaerobic rates. Both modules combine biological and
chemical degradation rates to determine an overall decay rate. As a result, the modules cannot
explicitly consider the separate effects of hydrolysis and biodegradation.
The Vadose Zone and Aquifer Modules take into account adsorption behavior of organic
chemicals by calculating a retardation factor based on a contaminant-specific partition
coefficient (Kd). To develop Kd, the modules apply a distribution coefficient normalized for
organic carbon (Koc), in conjunction with a media-specific fractional organic carbon content to
obtain the soil (or aquifer material)/water partition coefficient (Kd). The use of Koc to estimate
sorption is appropriate for organic compounds that tend to sorb preferentially on the natural
organic matter in the soil or aquifer. Although the Aquifer Module is limited to using linear Kd
values, the Vadose Zone Module can use nonlinear Freundlich sorption coefficients if these are
available for the organic compounds being modeled.
9-9
-------
Section 9.0
Vadose Zone and Aquifer Modules
Additional details on modeling organic chemical reactions within the Vadose Zone and
Aquifer Modules can be found in U.S. EPA (1996a,b; 1999e).
Organic Chemical Reaction Model
Assumptions and Inputs. Key assumptions
for the organic chemical reaction models used
in the Vadose Zone and Aquifer Modules are
summarized in the text box. Chemical-
specific inputs include
¦ Acid-catalyzed, base-
catalyzed, and neutral
hydrolysis rates, specified for
a default constant reference
temperature of 25°C;
¦ First-order aerobic (vadose
zone) and anaerobic (aquifer)
biodegradation rates; and
Key assumptions and limitations of vadose zone
and aquifer organic chemical reaction models
¦ The sorption of contaminants onto soil or aquifer
solids occurs instantaneously and is entirely
reversible.
¦ Sorption of organic compounds is linear in the
saturated zone but can be nonlinear in the vadose
zone.
¦ The model cannot explicitly consider the separate
effects of multiple degradation processes.
¦ Biodegradation is aerobic in the unsaturated zone
and anaerobic in the saturated zone.
¦ All transformation reactions are adequately
represented by first-order decay processes.
Partition coefficient normalized for soil organic carbon (Koc).
When a multichemical simulation is desired for parent and daughter compounds, the necessary
chemical-specific parameters must be repeated for all species in the decay chain.
Metal Sorption Processes. Site geochemistry and metal concentration are important
determinants of the fate and transport of metals in the subsurface. Varying geochemical
conditions result in the large variability in metal sorption behavior observed from site to site.
Metals that sorb readily at low concentrations usually show much lower sorption at higher
concentrations. To help capture this effect, the Vadose Zone and Aquifer Modules use sorption
isotherms generated by the MINTEQA2 metal speciation model (Allison et al., 1991) that are
nonlinear with respect to metal concentration. To represent nationwide variability in
geochemistry, these concentration-dependent partition coefficients have been developed for
various combinations of key geochemical parameters (MINTEQA2 master variables) known to
affect metal sorption, including the pH, iron oxide content, and organic matter content of the
subsurface environment, and the leachate organic matter content.
MINTEQA2 generates concentration-dependent effective partition coefficients (Kd
values) for various combinations of the key geochemical parameters that are sampled during the
Monte Carlo runs. Based on the selection of these parameters for each Monte Carlo realization,
the Vadose Zone and Aquifer Modules call the appropriate sorption isotherm for use during
transport modeling. The set of sorption isotherms included in the 3MRA modeling system has
two subsets of isotherms for each metal: one representing vadose zone conditions and the other
representing aquifer conditions.
In the Vadose Zone Module, metal partition coefficients are adjusted using these
isotherms to represent the general increase in Kd that is expected to occur as contaminant
9-10
-------
Section 9.0
Vadose Zone and Aquifer Modules
concentrations decrease along the transport path. This procedure is described in the background
document for the modeling of metals transport (U.S. EPA, 1996c). Because it is limited to linear
sorption processes, the Aquifer Module selects a single Kd value from the isotherm that
corresponds to the maximum ground water concentration under the source. This is appropriate
because by the time the contamination reaches the Aquifer, metal concentration is usually low
enough that a linear isotherm (i.e., a single Rvalue not dependent on metal concentration) is
appropriate.
The MINTEQA2 sorption isotherms implemented in the Vadose Zone and Aquifer
Modules are a significant improvement over those used in previous EPACMTP versions in that
they address a greater number of metals, including antimony, arsenic (+3 and +5 species),
barium, beryllium, cadmium, chromium (+3 and +6), copper, lead, mercury, nickel, selenium,
silver, thallium, vanadium, and zinc. This was made possible by improvements in MINTEQA2's
thermodynamic, iron oxide sorption, and organic matter sorption databases (U.S. EPA, 1999a).
Additional details on the MINTEQA2 modeling approach, including model formulation,
assumptions, and limitations, can be found in Allison et al. (1991) and U.S. EPA (1996c,
1999a, e).
9.3 Module Discussion
9.3.1 Strengths and Advantages
The Vadose Zone Module simulates the migration of constituents from the bottom of
land-based WMUs to the water table, and the Aquifer Module simulates the migration of
constituents in the saturated zone from the water table immediately below the WMU to the
downgradient receptor wells. Both modules have been used in regulations and have been
thoroughly tested and verified and validated as part of the integrated groundwater model
EPACMTP. Strengths of the two modules include the following:
Strengths and advantages for the Vadose Zone Module are listed below:
¦ Widely used, verified, validated, state-of-the-science approach. The Vadose
Zone Module is based on a state-of-the-science approach and is part of
EPACMTP, which has been tested, verified, and validated. EPACMTP has been
used to support regulations and has undergone various peer and public reviews,
including reviews by the SAB.
¦ Computational efficiency. The solution procedure used in the Vadose Zone
Module is computationally very efficient and stable and, therefore, ideally suited
for use in Monte Carlo frameworks.
¦ Use of nonlinear metal isotherms. The module can handle nonlinear metal
isotherms while maintaining computationally efficiency because of the solution
technique used in the module.
9-11
-------
Section 9.0
Vadose Zone and Aquifer Modules
¦ Simulation of multiple transformation products. The module can simulate the
transformation of waste constituents into multiple transformation products (up to
seven chain members).
Strengths and advantages for the Aquifer Module are listed below:
¦ Widely used, verified, validated, state-of-the-science approach. The Aquifer
Module is based on a state-of-the-science approach and is part of EPACMTP,
which has been tested, verified, and validated. EPACMTP has been used to
support regulations and underwent various peer and public reviews, including
reviews by the SAB.
¦ Computational efficiency. The module uses a special solution technique for the
boundary value problem that makes it computationally very efficient and stable.
The module is ideally suited for use in Monte Carlo framework in which
numerous simulations are required using a minimum amount of computational
time.
¦ Complete transient response at multiple well locations. The module provides
complete transient response at multiple receptor well locations.
¦ Simulation of multiple transformation products. The module can simulate the
transformation of waste constituents into multiple transformation products (up to
seven chain members).
9.3.2 Uncertainty and Limitations
Limitations and uncertainties for the Vadose Zone Module are listed below:
¦ Transient effects of the flow are not considered (i.e., year-to-year variability in
infiltration is not considered).
¦ Multiphase flow and transport are not modeled; NAPL flow and transport are not
modeled.
¦ Volatilization and vapor-phase diffusion are not modeled.
¦ Preferential flow due to fractures or heterogeneity in the vadose zone is not
considered.
¦ Clay lenses or other potential flow and transport barriers in the vadose zone are
not considered.
¦ Decay is limited to first-order reactions; lag time for decay is not considered.
9-12
-------
Section 9.0
Vadose Zone and Aquifer Modules
¦ The transport domain in the saturated zone is kept constant. Effects due to
mounding caused by infiltration from WMUs are not considered. These effects
would decrease the depth of the flow and transport domain in the vadose zone.
Limitations and uncertainties for the Aquifer Module are listed below:
¦ Transient effects of ground water flow, recharge, and infiltration are not
considered.
¦ Spatially varied recharge is not considered.
¦ Source geometry is limited to an idealized square, with two opposite sides parallel
to the flow direction.
¦ Multiphase flow and transport are not modeled. NAPL flow and transport are not
modeled.
¦ Contaminant contribution to the saturated zone via vapor-phase diffusion above
the water table is not modeled.
¦ Karst conditions are not modeled.
¦ Decay is limited to first order. Lag time for decay is not considered.
¦ The presence of different hydrogeologic zones in the flow and transport domain is
not considered.
¦ The transport domain in the saturated zone is kept constant. Effects due to
significant mounding caused by infiltration from WMUs are not considered.
¦ Domain geometry is limited to the idealized rectangular shape. Other geometries
are not considered.
¦ Only gaining streams, with axes normal to the ground water flow direction, are
permitted. Effects of streams on the flow field are not considered.
¦ Only receptor wells with small extraction rates are considered. Effects of well
extraction on the ground water flow field are not considered.
¦ There are many sources of uncertainty associated with the distribution
coefficients generated by the metal speciation model. These can be categorized as
uncertainty arising from model input parameters, uncertainty in database
equilibrium constants, and uncertainty due to application of the model.
Additional details can be found in the MINTEQA2 background documents
(Allison et al., 1991; U.S. EPA, 1991a, c).
9-13
-------
Section 9.0
Vadose Zone and Aquifer Modules
9.4 References
Allison, J.D., D.S. Brown, and K.J. Novo-Gradac. 1991. MINTEQA2/PRODEFA2
geochemical assessment model for environmental systems: Version 3.00 User Manual.
Athens, GA: U.S. Environmental Protection Agency.
Bear, J. 1972. Dynamics of Fluids in Porous Media. American Elsevier, NY.
Freeze, R.A., and J.A. Cherry. 1979. Groundwater. Prentice-Hall, NJ.
Huyakorn, P.S, and G.F.Pinder. 1983. Computational Methods in Subsurface Flow. Academic
Press, NY.
U.S. EPA (Environmental Protection Agency). 1996a. EPA 's Composite Model for Leachate
Migration with Transformation Products (EPACMTP). Background Document. Office
of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1996b. EPA's Composite Model for Leachate
Migration with Transformation Products (EPACMTP), Background Document for Finite
Source Methodology for Chemical with Transformation Products. Office of Solid Waste,
Washington, DC.
U.S. EPA (Environmental Protection Agency). 1996c. EPA 's Composite Model for Leachate
Migration with Transformation Products (EPACMTP). Background Document for
Metals Transport in the Subsurface. Office of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1997. Test and Verification of EPA's Composite
Model for Leachate Migration with Transformation Products (EPACMTP), Draft
Version. Office of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999a. Changes in theMINTEQA2Modeling
Procedure for Estimating Metal Partitioning Coefficients in Ground water for HWIR99.
Office of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999b. Incorporation of Heterogeneity into
Monte-Carlo Fate and Transport Simulations. Office of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999c. MINTEQA2/PRODEFA2, A
Geochemical Assessment Model for Environmental Systems: User Manual Supplement
for Version 4.0. Office of Solid Waste, Washington, DC
U.S. EPA (Environmental Protection Agency). 1999d. A Study to Assess the Impacts of
Fractured Media in Monte-Carlo Simulations. Office of Solid Waste, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1999e. The Vadose and Saturated Zone Modules
Extractedfrom EPACMTP for HWIR99. Office of Solid Waste, Washington, DC.
August.
9-14
-------
Section 9.0 Vadose Zone and Aquifer Modules
Van Genuchten, M. Th. 1980. A closed-form equation for predicting the hydraulic conductivity
of unsaturated soils. Soil Sci. Soc. J. 44:892-898.
9-15
-------
Section 9.0 Vadose Zone and Aquifer Modules
9-16
-------
Section 10 .0
Farm Food Chain Module
10.0 Farm Food Chain Module
10.1 Purpose and Scope
The Farm Food Chain Module predicts the accumulation of contaminants in the edible
parts of plants through the uptake of constituents from soil and the deposition of vapor-phase and
particle-bound constituents from the air. Concentrations are predicted for fruits and vegetables
that are grown above ground, as well as for root vegetables. In addition, the module predicts the
annual average contaminant concentrations in beef and milk products from cattle raised on
farms. The concentrations in produce, beef, and milk are used as inputs to the Human Exposure
Module to calculate the applied daily dose to human receptors that consume fruits and vegetables
from home gardens or consume produce, beef, or milk produced on a farm. Figure 10-1 shows
the relationship and information flow between the Farm Food Chain Module and the 3MRA
modeling system.
Key Data Inputs
• Farm area
• Plant uptake factor
• Beef biotransfer factor
Air Concentrations
Deposition Rates
Soil Concentrations
Food Item and
Soil Concentrations
Soil Concentrations
Subsurface
Concentrations
Water Column
Concentrations
Human
Exposure
Module
Vadose Zone
and Aquifer
Modules
Air Module
Watershed
Module
Land-based
Source
Modules
Surface
Water
Module
Farm
Food
Chain
Module
Figure 10-1. Information flow for the Farm Food Chain Module in the 3MRA
modeling system.
10-1
-------
Section 10 .0
Farm Food Chain Module
The modeling construct for the Farm Food Chain Module is based on recent and ongoing
research conducted by EPA ORD and presented in Methodology for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions (U.S. EPA, 1998).
The Farm Food Chain Module performs the following four functions:
1. Calculates contaminant concentrations in plants due to contaminants in air.
The Farm Food Chain Module calculates the contaminant concentration in plants
due to the deposition of particle-bound and vapor-phase contaminants onto fruits,
vegetables, and feed crops that grow above the ground.
2. Calculates contaminant concentrations in plants due to contaminants in soil.
The Farm Food Chain Module calculates the contaminant concentration in plants
due to uptake and translocation of contaminants in the soil into the edible parts of
fruits, vegetables, and feed crops that grow above the ground.
3. Calculates total contaminant concentrations in plants. The Farm Food Chain
Module sums plant concentrations across relevant mechanisms, including direct
deposition of particle-bound contaminants, vapor-phase uptake, and translocation
of contaminants from soil into the edible parts of plants.
4. Calculates contaminant concentrations in beef and milk. The Farm Food
Chain Module calculates exposures to beef and dairy cattle through ingestion of
contaminated forage, feed crops, soil, and drinking water, and the resulting beef
and milk concentrations.
For each year in the simulation, the module predicts point estimates and spatially
averaged contaminant concentrations within the area of interest (AOI). Point estimate
concentrations are used to evaluate exposures of residential home gardeners that grow and eat
fruits and/or vegetables within the AOI. The point estimates reflect the concentrations at
locations of residential receptors that are used to represent the populations in various Census
tracts throughout the AOI. The spatially averaged concentrations are used to evaluate exposures
of farmers that raise and eat their own produce, beef, or milk products. The spatial averages
reflect the concentrations at the farm boundaries delineated in the site layout that defines all of
the characteristics of the AOI.
For both farmers and residential receptors, the Farm Food Chain Module predicts a time
series of annual average concentrations of contaminants in fruits and vegetables.1 With the
exception of root vegetables, the categories of fruits and vegetables are designated as either
"exposed" or "protected." For the farm food chain modeling framework, the term "exposed"
means that the contaminant concentrations in plants are calculated based on uptake and
accumulation from aerial deposition, uptake during transpiration, and uptake from the soil. The
1 Botanically, "fruit" refers to any part that develops from the flower (i.e., reproductive parts) and includes
what are considered fruits as well as vegetables in terms of grocery items. For example, tomatoes, corn, and beans
all develop from the flower and are thus all considered fruits in botanical terms. In the context of this discussion, the
conventional grocery item terminology is used.
10-2
-------
Section 10 .0
Farm Food Chain Module
term "protected" means that the outer covering of the fruit or vegetable is not edible and serves
as a barrier to contaminant transfer from air; therefore, the contaminant concentration in plants
for protected fruits and vegetables is attributed only to the uptake of contaminants from the soil
through the root system and subsequent translocation into the edible portions of the plant. Each
of these categories is explained in further detail below.
¦ Exposed vegetables. The term "exposed vegetables" refers primarily to fruiting
plants and edible vegetative parts of plants that are exposed directly to
contaminant loadings from the air. For example, cucumbers, tomatoes, peppers,
and green beans are included in the category of exposed vegetables, although all
of these develop from flowers and are thus considered fruits in botanical terms.
¦ Protected vegetables. This category of vegetables includes plants with the edible
part protected from airborne contaminants by a nonedible covering. For example,
peas and lima beans are considered protected vegetables because their outer pod
is not eaten, so the plant load in the edible portion is exclusively due to soil-to-
plant transfer of contaminants. Aerial deposition is not included in calculating
contaminant concentrations in protected vegetables.
¦ Exposed fruits. Exposed fruits include produce commonly referred to as "fruits"
that have an edible outer skin. This category includes a number of tree fruits
(e.g., apples, pears) and various vine fruits (e.g., strawberries, grapes). The plant
concentration is a function of both the soil concentration and the concentration of
contaminants in the air.
¦ Protected fruits. Protected fruits have limited relevance in many exposure
scenarios because they include fruits with a fairly narrow range of growing
climates. Although a variety of melons can be grown throughout much of the
contiguous United States, other protected fruits, such as bananas, pineapples, and
citrus fruit, are found only in the warmest regions of the country, such as Florida.
As with protected vegetables, the concentration in the edible portion is
exclusively due to soil-to-plant transfer of contaminants.
¦ Root vegetables. This category cuts across a number of botanical categories,
including leaves, stems, and roots. The common theme for this category is that all
of these vegetables grow below the soil surface. Consequently, a host of
vegetables, such as potatoes (stem), onions (leaf), and carrots (root), are lumped
together in this category. As with protected fruits and vegetables, the
concentration in root vegetables is predicted based on the contaminant
concentrations in soil.
In addition to calculating the concentrations in produce, the Farm Food Chain Module
also generates a time series of annual average contaminant concentrations in beef and milk for
each farm within the AOI. The beef and milk concentrations are calculated for beef and dairy
cattle, respectively, and include three exposure pathways for cattle: (1) the consumption of
contaminated feed crops (i.e., forage, grain, and silage), (2) incidental ingestion of contaminated
10-3
-------
Section 10 .0
Farm Food Chain Module
soil while foraging, and (3) ingestion of contaminated drinking water from streams, ponds, or
wells on the farm.
10.2 Conceptual Approach
As shown in Figure 10-2, contaminants may be released from waste management units
(WMUs) into the air and subsequently deposit on soil and plants through wet and dry deposition.
Plants may accumulate contaminants directly from the air through the deposition of both
particle-bound and vapor-phase contaminants. In addition, this deposition may increase the
contaminant concentrations in the soil over time, and plants may take up contaminants from the
soil through the root system. Contaminants may be translocated as part of the transpiration
stream to fruits and edible vegetative parts, or they may sorb directly to the outer skin of root
vegetables. The Farm Food Chain Module performs the calculations for each of these uptake
and accumulation mechanisms. The total contaminant concentration in plants will depend both
on the category of plant being modeled (i.e., exposed versus protected) and the properties of the
contaminant of concern. For example, the uptake and accumulation of dioxin-like chemicals
from soil to edible parts of plants has been shown to be negligible because these chemicals are
strongly sorbed to the organic fraction of soil. Because the behavior of each contaminant is, to a
large degree, determined by its chemical properties, the Farm Food Chain Module uses both
empirical data and regression algorithms to predict the uptake and accumulation of contaminants
from environmental media into edible plant tissue.
Dispersion
Volatilization
Dispersion
Particulates
Vapor and Particle
Deposition to Plants
Deposition
to Soil
Wind
WMU
Translocation to
Edible Plant Parts
w
Root Uptake
Groundwater
Figure 10-2. Release, exposure, and uptake mechanisms of
contaminants in plants.
10-4
-------
Section 10 .0
Farm Food Chain Module
10.2.1 Calculate Contaminant Concentrations in Plants due to Contaminants in Air
Contaminants in the air may accumulate on the surface of plants by two mechanisms:
particle-bound deposition and vapor-phase deposition. The algorithms used to simulate these
processes and predict contaminant concentrations in exposed fruits, exposed vegetables, forage,
and silage due to contaminants in air are described below. These processes are not considered
for protected fruits, protected vegetables, or root vegetables.
The total plant load from air may be driven by either particle or vapor deposition,
depending on the contaminant of concern. For example, metals released from a land application
unit (LAU) can be deposited onto plant surfaces through particle-bound but not vapor-phase
deposition. Therefore, the Farm Food Chain Module must include both of these mechanisms in
calculating the contaminant load to plants from air. The deposition rates are calculated by the
Air Module for the entire AOI; the areal average deposition rates for farms and average point
estimates for residential receptor locations are provided by year.
Throughout this subsection, the subscript i refers to the following plant types:
¦ Exposed vegetables,
¦ Exposed fruits,
¦ Forage,
¦ Grains, and
¦ Silage.
Plant Concentrations from Deposition of Particle-Bound Contaminants. The plant
concentration due to deposition of particle-bound contaminants in the air is calculated as a
function of wet and dry deposition rates as follows:
\000x365^ParDryDep+ ParWetDepxFracAdhere'^xFraclnt1 x|l.O-exp^ipPar j
plant- dep1 Yield' x kpPar'
where
Cpiant dep1 = concentration in plant i due to particulate deposition (mg/kg DW)
1000 = units conversion factor (1,000 mg/g)
365 = units conversion factor (365 d/yr)
ParDryDep = average dry deposition rate of particulates (g/m2-d)
ParWetDep = average wet deposition rate of particulates (g/m2-d)
FracAdhere1 = fraction of wet deposition that adheres to plant i (unitless)
Fraclnt1 = interception fraction for plant i (unitless)
kpPar' = surface loss of particle-bound contaminant for plant i (1/yr)
LengthEx1 = length of plant exposure during the growing season for plant i (yr)
Yield1 = yield or standing crop biomass for plant i (kg DW/m2).
The contaminants that reach the plant through dry deposition are assumed to remain on
the plant surface until weathering occurs. In contrast, only a fraction of the contaminant from
10-5
-------
Section 10 .0
Farm Food Chain Module
wet deposition remains on the plant's surface; the rest washes off immediately. This is reflected
by the variable for the fraction of wet deposition that adheres to a plant's surface.
Not all airborne particles will contact a plant's edible surface; some will fall to the
ground, others will fall on other surfaces that will undergo weathering processes, such as wind
removal, water removal, and growth dilution, and most will end up in the soil or runoff. The
interception fraction represents the fraction of airborne contaminants that make contact with the
surface of the plant. The plant's surface can also lose contaminants due to weathering, and the
aggregate losses from the plant surface are represented by the surface loss coefficient.
The duration of exposure for the plant is often determined by the growing season and is
represented by the length of exposure of the plant to contaminants in the air. For instance, the
exposure duration for a tomato begins when the flower begins to ripen and lasts until the tomato
is harvested.
Finally, the plant biomass or yield represents the amount of standing crop assumed for a
farm or residential garden, as appropriate.
Plant Concentrations from Deposition of Vapor-Phase Contaminants. As with
particle-bound contaminants, vapor-phase deposition to plant surfaces occurs through wet and
dry deposition processes. The mechanisms for wet and dry deposition are, to a large degree,
dependent on the properties of the contaminant. Plant concentrations due to deposition of vapor-
phase contaminants can be calculated two ways. For many organic chemicals, the form of the
equation is essentially the same as the equation used to predict plant concentrations from
particle-bound contaminants, as follows:
lOOOx (VapI)ryI)ep+VapWetI)ep- FracAdhere' x 365)/ FracInt'{\-eckpVap'LengthEx ^j ^ ^ ^
plant-vap Yield' x kpVap1
where
Cpiant vap1 = concentration in plant i due to vapor deposition (mg/kg DW)
1000 = units conversion factor (1,000 mg/g)
365 = units conversion factor (365 d/yr)
VapDryDep = average dry deposition rate of vapor-phase contaminants (g/m2-yr)
VapWetDep = average wet deposition rate of vapor-phase contaminants (g/m2-d)
FracAdhere1 = fraction of wet deposition that adheres to plant i (unitless)
Fraclnt1 = interception fraction for plant i (unitless)
kpVap1 = surface loss of vapor-phase contaminant for plant i (1/yr)
LengthEx1 = length of plant exposure during the growing season for plant i (yr)
Yield1 = yield or standing crop biomass for plant i (kg DW/m2).
The variables have the same function as previously described for the particulate-phase equation
above.
For organic chemicals with a high affinity for lipid tissue, the mechanism of uptake and
accumulation of vapor-phase organics for plants is not accurately described by the term
10-6
-------
Section 10 .0
Farm Food Chain Module
"deposition." Evidence has shown that these compounds can be essentially stripped from the air
simply by coming in contact with vegetation. An alternative model for the dry deposition of
vapor-phase organic compounds is termed the "transfer" approach. This model describes the
plant concentration in terms of air-to-plant transfer rather than physical deposition on plant
surfaces. Further, EPA has demonstrated that wet deposition of vapor-phase lipophilic
compounds can be considered negligible. Consequently, the plant concentration for lipophilic
organic chemicals (functionally defined as those contaminants with an octanol-water partition
coefficient greater than or equal to 100,000) is calculated as follows:
Cairvar)nr x BTF ¦ x ECF , ¦
^ vaPur air-plant1 exposed1 (10 3"
plant_vapl 1000 x pair
where
Cpiant vapi = concentration in plant i due to vapor transfer into plant (mg/kg DW)
Cair vapor = vapor-phase concentration in air (|ig/m3)
BTFair planti = contaminant-specific air-to-plant biotransfer factor for plant i ([|ig/g
DW]/[|ig/g air])
ECFexposedi = empirical correction factor to convert the air-to-plant biotransfer
factor derived for leafy vegetation to a value for plant / (unitless)
1,000 = units conversion factor (L/m3)
Pai,. = density of air (constant at 1.19 g/L).
The dry deposition rate of vapor-phase, organic chemicals is not currently calculated by
the Air Module. Rather, it is estimated from the vapor-phase concentration in air using the
methods presented in U.S. EPA (1998), as follows:
VapDryDep=0.31536x CvAve x VapDdv (10-4)
where
VapDryDep = average dry deposition rate of vapors (g/m2-yr)
0.31536 = units conversion factor ((m/yr)/(cm/s) and (g/|ig))
CvAve = vapor-phase concentration in air (|ig/m3)
VapDdv = vapor phase dry deposition velocity - default value of 1 (cm/s).
The air-to-plant biotransfer factor describes the relationship between the contaminant
concentration in exposed plant parts and the vapor-phase contaminant concentration in air. It
may be calculated from chemical-physical properties (as it is for most organic chemicals), or it
may be derived using empirical data in the open literature or EPA sources. The empirical
correction factor (ECF) for each plant category i reflects the fact that experiments have shown
that lipophilic, persistent organics such as polycyclic aromatic hydrocarbons (PAHs) and dioxins
tend not to translocate from the outer surfaces of plants to inner plant parts. Without this
adjustment, the biotransfer factor would be inappropriately high (i.e., it would overpredict the
contaminant concentration in the inner parts of the plant). The ECF may be further adjusted to
10-7
-------
Section 10 .0
Farm Food Chain Module
account for washing and peeling of fruits and vegetables that reduce contaminant residues in the
outer skin. However, the contaminant-specific data set available to the 3MRA modeling system
does not currently include washing and peeling losses in the empirical adjustment factor for
aboveground plants. For many organic chemicals, the correction factor is simply set to 1,
indicating that the contaminant is efficiently translocated to the inner plant parts. See Volume II
of this report for further details.
10.2.2 Calculate Contaminant Concentrations in Plants due to Contaminants in Soil
Contaminant releases to soil from WMUs may occur through aerial deposition, as well as
through erosion and runoff mechanisms. Over time, the contaminant concentration in soil may
increase, with a resultant increase in the contaminant concentrations in the edible parts of plants.
Depending on the properties of a given contaminant and the physiology of exposed vegetation,
plants may take up aqueous-phase contaminants in the soil through the roots and subsequently
translocate them to edible plant parts. The plant concentration is a function of the contaminant
concentration in the root zone soil and the soil-to-plant bioconcentration factor, as follows:
The root zone soil concentration is area averaged for farms. The soil-to-plant bioconcentration
factor quantifies the potential for a contaminant in soil to be taken up by the roots and
translocated to edible plant parts. As with the air-to-plant biotransfer factor, the soil-to-plant
bioconcentration factor may be calculated from chemical-physical properties for most organic
chemicals, or derived from empirical data for metals (see Volume II for more details). Equation
10-5 is appropriate for metals for all categories of plants (e.g., exposed fruits, protected
vegetables, forage) and for organic chemicals for all categories of plants except root vegetables.
However, for certain types of chemicals, such as dioxins, studies have shown that this pathway
for uptake and accumulation of contaminants in plants is negligible; that is, the soil
concentration does not correlate with the contaminant concentration in plant tissue. For dioxin-
like chemicals, the concentration in exposed fruits and vegetables is exclusively a function of
air-to-plant transfer and particle deposition.
For root vegetables, the mechanism by which plants take up organic chemicals from the
soil and accumulate them in edible tissue is not adequately described by Equation 10-5.
Experiments suggest that contaminant concentrations found in the outer parts of the vegetable, or
skin, are due more to sorption than to passive uptake via transpiration water (U.S. EPA, 1998).
Although water-soluble contaminants may be distributed more or less uniformly throughout the
entire plant, lipophilic contaminants may be bound almost exclusively to the outer portion of the
root vegetable. The relationship between soil pore water and plant concentrations is expressed
^plant soif ^soil_RZ x soil- plant
(10-5)
where
plant concentration uptake from soil and translocation (mg/kg DW)
depth-averaged soil concentration in the root zone (|ig/g soil)
contaminant-specific soil-to-plant bioconcentration factor
([|ig/g DW]/[|ig/g soil]).
10-8
-------
Section 10 .0
Farm Food Chain Module
as a root concentration factor, which is used to predict the contaminant concentration in root
vegetables as follows:
where
r =
^RootVeg
Csoil_RZ ~~
RootCF =
ECFRootVeg
soil
The soil-to-plant bioconcentration factor (in Equation 10-5) is replaced with the root
concentration factor, which quantifies the potential for organic chemicals in soil pore water to
accumulate in root vegetables. An empirical correction factor is included to adjust for
differences between the experimental data and the application in the 3MRA modeling system.
For example, much of the experimental data on the root concentration factor is based on whole
barley roots, and a contaminant-specific correction factor is needed to adjust for the volume
differences between barley roots and bulky root vegetables. The empirical correction factor may
be further adjusted for peeling, cooking, or cleaning, which can all reduce the contaminant
concentration. However, the chemical-specific data set available to the 3MRA modeling system
does not currently include these losses in the empirical adjustment factor for root vegetables.
10.2.3 Calculate Total Contaminant Concentrations in Plants
The Farm Food Chain Module calculates the total contaminant concentration in plants for
exposed fruits, exposed vegetables, forage, and silage by summing the concentrations for all
potential exposure pathways for plants. This summation step is not needed for protected fruits
and vegetables or root vegetables, because they take up contaminants only from the soil. For all
aboveground produce, the total contaminant concentrations in plants are converted from dry
weight (DW) to wet weight (WW) by adjusting for the moisture content. These WW
concentrations (also referred to as whole weight or fresh weight) are required by the Human
Exposure Module. This conversion is not needed for root vegetables, because the concentration
is calculated in WW directly.
10.2.4 Calculate Contaminant Concentrations in Beef and Milk
The Farm Food Chain Module uses the predicted soil, plant (i.e., forage, grains, and
silage), and drinking water concentrations to predict the contaminant concentrations in beef and
milk for animals raised on beef and dairy farms within the AOI. The exposure scenario is based
on the assumption that beef and dairy cattle consume some fraction of forage, feed grain, and
silage that is grown on the farm. Furthermore, cattle are presumed to ingest surface water from
stream reaches or ponds that are delineated on the farm; where no such waterbody exists, ground
water is assumed to be the drinking water source for cattle. The incidental ingestion of soil
refers only to the surficial soil (i.e., the top 1 cm of soil) that may be ingested during normal
Qo/7 rz x RootCFx ECFRootVeg
Root_Veg ~ ^ (10"6)
^soil
contaminant concentration in root vegetables (mg/kg WW)
depth-averaged soil concentration in the root zone (|ig/g soil)
root concentration factor ([|ig/g WW plant]/[|ig/mL soil water])
empirical correction factor for root vegetables (unitless)
soil-water partition coefficient (mL/g).
10-9
-------
Section 10 .0
Farm Food Chain Module
grazing by cattle on untilled soil. All of the exposure concentrations reflect the spatial averaging
across each individual farm modeled within the AOI and a consistent temporal scale; that is, the
time series of input concentrations for plants, soil, and water are matched for each year in the
simulation.
The contaminant concentrations in beef or milk are calculated as a function of the
biotransfer of contaminants from plants, soil, and water into beef and milk as follows:
^'beef I milk feed /'
2 (c, xCi?,x/;)+(csml X CRsoll X FracBiosml ) + BTFwater x (Cwater d x IRwater) (10-7)
where
^beef/milk
contaminant concentration in beef or milk (mg/kg WW)
btf fccd =
contaminant biotransfer factor from feed into beef or milk (d/kg tissue
WW)
Q
contaminant concentration in plant type / (forage, grain, or silage) grown
on farm (mg/kg DW)
CRi
consumption rate of plant type i for beef or dairy cattle (kg plant DW/d)
f,
fraction of plant type i grown on farm (unitless)
Csoil ~
contaminant concentration in surficial soil at farm (mg/g soil)
CRSOii ~
quantity of contaminated soil eaten by beef or dairy cattle (kg soil/d)
FracBiosoil =
fraction of contaminant in soil bioavailable relative to vegetation
(unitless)
BTF
water
contaminant biotransfer factor from water into beef or milk (d/kg tissue
WW)
c d
^ water
dissolved contaminant concentration in drinking water (mg/L)
IRwater
drinking water ingestion rate for beef or dairy cattle (L/d).
Biotransfer from the soil (the second term in brackets in Equation 10-7) is adjusted by the
fraction of bioavailable contaminant in the soil to account for differences in absorption
efficiency when contaminant is ingested through plant matter versus through contaminated soil.
For some contaminants, vegetation is a more efficient vehicle for biotransfer because the
breakdown and digestion of plant matter releases a higher fraction of contaminant available for
absorption in the gut. Data on the biotransfer of dissolved contaminants in water into beef and
milk were not identified. Nevertheless, because of the potential importance of this pathway, the
biotransfer factor for water is set equal to the biotransfer factor for feed for beef and milk,
respectively.
The values for biotransfer factors, consumption rates, and fraction contaminated food
items differ for beef and dairy cattle. For example, dairy cattle tend to have a much higher water
ingestion rate to support lactation; therefore, the ingestion rate of water for dairy cattle is higher
than the value for beef cattle.
10-10
-------
Section 10 .0
Farm Food Chain Module
10.3 Module Discussion
10.3.1 Strengths and Advantages
The Farm Food Chain Module was developed to (1) predict contaminant concentrations
in produce and feed crops from multiple routes of plant uptake, and (2) estimate concentrations
in beef and milk for cattle raised on farms within the AOI. Relative to other approaches that
were considered for the 3MRA modeling system (e.g., fugacity-based models such as PlantX),
the strengths and advantages of the Farm Food Chain Module include the following:
¦ Widely used and reviewed approach. The Farm Food Chain Module was based
on the science, algorithms, and data developed by EPA's ORD to assess indirect
exposures to contaminated produce, beef, and milk. This approach has been
widely used by the EPA regions as well as other EPA program offices, and
represents a considerable investment by EPA in indirect exposure assessment.
Consequently, the approach has been reviewed in a variety of different
applications and corresponds well with the categories of produce (e.g., protected
vs. unprotected vegetables) that are needed to assess human health exposures.
¦ Simulates major plant uptake processes for contaminants in air and soil. The
contaminant concentration in plants is a function of all relevant pathways given
the type of plant and the chemical properties of the contaminant. The
concentration is summed across uptake for soil as well as uptake from the air, as
appropriate, for each contaminant/plant combination. Because some of the
WMUs release contaminants into the air, it is crucial that the model accounts for
the contaminant mass taken up directly from the air and deposited on the soil
from the air and taken up via the roots.
¦ Calculates point estimates for home gardeners and areal average estimates
for farmers. The Farm Food Chain Module can calculate the contaminant
concentrations both in plants at points specifically designated in the site layout
file (i.e., home gardeners) as well as spatially averaged across farm areas within
the AOI. The averaging functions in the module include calculating contaminant
concentrations in soil that includes both the regional and local watersheds, and air
concentrations of particulates and vapor in the same spatial area and time frame.
Therefore, the contribution to the total plant concentration from air and soil
reflects the spatial character of the plant exposure through time such that the plant
loads from soil and from air can be meaningfully added together.
¦ Data requirements and module performance consistent with the overall
design of the 3MRA modeling system. The data requirements for the Farm
Food Chain Module are relatively limited with respect to plant physiology. As a
result, a wide variety of chemical contaminants may be evaluated without
requiring inputs for individual plant species included under the categories
presented in Section 10.1 (e.g., exposed vs. protected vegetables), or for specific
compartments within the plants. Similarly, the module is computationally
efficient so that national-scale analyses involving time-varying concentration
10-11
-------
Section 10 .0
Farm Food Chain Module
profiles in air, surface water, ground water (for ingestion by cattle), and soil are
feasible in a reasonable run time. Because the system simulations may span many
years, and because the system was designed to support full Monte Carlo
implementation, it was critical that the module be able to produce results within a
time frame that would be useful to the user.
10.3.2 Uncertainty and Limitations
The technical approach underlying the Farm Food Chain Module includes a number of
assumptions, uncertainties, and limitations. Limitations include the following:
¦ System constraints limit each simulation to a single WMU and contaminant.
The 3MRA modeling system was designed to simulate a single contaminant for a
single WMU. Consequently, multiple plant loadings of a contaminant from
different WMUs (either within or outside of the AOI) are not considered. This
assumption may result in an underprediction of the contaminant concentrations in
produce, beef, and milk.
¦ Farm Food Chain Module is driven by empirical data. The Farm Food Chain
Module uses empirically derived regression algorithms, as well as empirical data
on uptake and accumulation, to predict contaminant concentrations in plants,
beef, and milk. Although other models that were evaluated have several
advantages over empirical models (for example, the compartmental model
developed by Trapp and MacFarlane, 1995), many of these models were
considered too data intensive or too computationally demanding for use in a
modeling tool intended to support national-level analyses. The reliance on
empirical data and models limits the model's ability to predict contaminant
concentrations outside of the narrow range for which data are available.
¦ Resuspension and redeposition of particle-bound contaminants are not
considered. Plant concentrations are a function of the deposition of the
contaminants that have been emitted from the WMU. The plant concentration
due to particle deposition does not include resuspension and redeposition of
contaminants bound to soil particles. Resuspension and redeposition can occur
due to tillage, wind erosion, vehicular resuspension, and rainsplash and can
increase the contaminant concentration in exposed fruits, exposed vegetables,
forage, and silage.
¦ Contaminant concentrations in beef and milk consider only the ingestion
pathway. The beef and milk calculations consider only ingestion pathways for
plant matter, surficial soil, and drinking water; exposures from the inhalation or
dermal contact are not included in the calculations. For some contaminants, this
limitation may result in an underestimate of the contaminant concentrations in
beef and milk.
10-12
-------
Section 10 .0
Farm Food Chain Module
10.4 References
Trapp, S. and J. Craig Mc Farlane, 1995. Plant Contamination: Modeling and Simulation of
Organic Chemical Processes. CRC Press, Boca Raton.
U.S. EPA (Environmental Protection Agency). 1997. Parameter Guidance Document. EPA,
National Center for Environmental Assessment, NCEA
U.S. EPA (Environmental Protection Agency). 1998. Methodology for Assessing Health Risks
Associated with Multiple Exposure Pathways to Combustor Emissions. EPA 600/R-
98/137. National Center for Environmental Assessment, Cincinnati, OH. December.
10-13
-------
Section 10 .0 Farm Food Chain Module
10-14
-------
Section 11.0
Terrestrial Food Web Module
11.0 Terrestrial Food Web Module
11.1 Purpose and Scope
The Terrestrial Food Web Module calculates the annual average contaminant
concentrations in terrestrial plants and prey (such as earthworms or small mammals) eaten by
wildlife. In addition, the module calculates spatially-averaged soil concentrations in two soil
horizons— surficial soil and root zone soil—for the receptor home ranges placed within each of
the habitats delineated within the area of interest (AOI). These concentrations are used as input
to the Ecological Exposure Module to calculate the applied dose to receptors of interest, and the
root zone soil concentration is also used by the Ecological Risk Module to predict risks to
terrestrial plants and soil communities. Figure 11-1 shows the relationship and information flow
between the Terrestrial Food Web Module and the 3MRA modeling system.
For each home range delineated within the AOI, the Terrestrial Food Web Module
predicts a time series of annual average concentrations of contaminants in soil, along with
concentrations in the specific plant and prey categories shown in Table 11-1. The Terrestrial
Food Web Module uses the same algorithms and contaminant-specific data as the Farm Food
Chain Module to calculate concentrations in plants.
Air Modu e
Air Concentrations
Deposition Rates
Watershed
Module
Soil Concentrations
Land-based
Source
Modules
Soil Concentrations
Key Data Inputs
Root zone depth
Home range area
Bioconcentration factor
Soil
Concentrations
Terrestrial
Food Web
Module
Soil, Plant, and Prey>
Concentrations
Ecological
Risk Module
Ecological
Exposure
Module
Figure 11-1. Information flow for the Terrestrial Food Web Module in the 3MRA modeling system.
11-1
-------
Section 11.0
Terrestrial Food Web Module
Table 11-1. Terrestrial Plant and Prey Categories in the Terrestrial Food Web
IVi'ivsli'iiil Phnil tries
loi'ivslriiil Pro (;M entries
Fruits, fruit/seeds (exposed fruit)
Worms
Fern(s), fungi, shoots (exposed vegetation)
Other soil invertebrates
Forbs, grasses, shrubs (forage)
Small mammals
Roots (root vegetables)
Small birds
Crops, corn (silage)
Small herpetofauna
Seeds/nuts (grains)
Herbivorous vertebrates
Omnivorous vertebrates
The parentheticals shown under the terrestrial plant categories provide the crosswalk
between categories of fruits, vegetables, and forage discussed under the Farm Food Chain
Module, and the analogous vegetation categories that are eaten by wildlife.
The Terrestrial Food Web Module uses both predictive models and empirical data to
calculate contaminant concentrations in terrestrial food items. Specifically, the Terrestrial Food
Web Module performs the following four functions:
1. Calculates contaminant concentrations in soil. The module calculates spatially
averaged soil concentrations for each home range in each habitat. The soil
concentrations reported for each site by the Land-based Source Modules and the
Watershed Module are catalogued by waste management unit (WMU) and
watershed subbasin, respectively. The Terrestrial Food Web Module determines
the overall spatial average contaminant concentration based on the proportion of
subbasins and/or WMU that overlaps the home range of wildlife species assigned
to a given habitat.
2. Calculates total contaminant concentrations in plants. The Terrestrial Food
Web Module sums plant concentrations across relevant mechanisms, including
direct deposition of particle-bound contaminants, vapor-phase uptake, and
translocation from soil into the edible parts of plants. The module uses the same
approach as described for the Farm Food Chain Module to calculate total
contaminant concentrations in plants. The only significant difference in the
respective modules is that the Terrestrial Food Web Modules defines plant
categories for consumption by wildlife rather than humans.
3. Calculates contaminant concentrations in soil invertebrates. Earthworms and
other soil invertebrates constitute a significant dietary component for many
wildlife species. Contaminant-specific soil-to-biota bioconcentration factors are
used to calculate the contaminant concentrations in soil invertebrates based on the
spatially averaged, root zone soil concentration for each wildlife species' home
range.
11-2
-------
Section 11.0
Terrestrial Food Web Module
4. Calculates contaminant concentrations in vertebrate prey categories. Small
to medium sized terrestrial vertebrates are eaten by larger predators (e.g., fox,
black bear, and red tailed hawk). Contaminant-specific soil-to-vertebrate
bioconcentration factors are used to estimate the tissue concentrations of
contaminants in vertebrate prey categories based on the spatially averaged, root
zone soil concentration for each wildlife species' home range.1 The module
reports the minimum and maximum tissue concentrations in each vertebrate prey
category; these outputs allow the Ecological Exposure Module to sample from the
spatial variability of possible prey species.
The Terrestrial Food Web Module is applied to the margin habitats in freshwater systems
and wetlands and to terrestrial habitats, such as the representative forest or grassland. Simple
terrestrial food webs were constructed for each terrestrial habitat to depict the major functional
and structural components of healthy terrestrial ecosystems. The components represent major
dietary categories for terrestrial wildlife, and represent a broad range of sizes, feeding guilds, and
taxa typical of terrestrial food webs. In addition, the terrestrial prey categories reflect available
methods and data required to estimate prey concentrations. For example, the category of "small
mammals" was developed specifically to take advantage of recent research (Sample et al., 1998)
on estimating tissue concentrations of contaminants in small mammals as a function of the soil
concentration. This research provides a significant improvement over previous approaches that
aggregated a variety of prey items into a single category of "terrestrial invertebrates."
11.2 Conceptual Approach
The Terrestrial Food Web Module was developed to allow for flexibility in calculating
contaminant concentrations in soil, plants, and prey eaten by wildlife. Despite the advantages in
such a flexible design, it was critical that the Terrestrial Food Web Module perform spatial
averaging in a manner that was both ecologically relevant and simple to implement. Therefore,
the module was designed to calculate contaminant concentrations in food items and soil in a way
that reflects the spatial scale of interest and conserves the role of various species as both predator
and prey. Most importantly, the approach takes full advantage of the information available in
the site layout file that defines the spatial character of the habitats and species' home ranges
assigned to the AOI. Indeed, the habitats and home ranges are a key feature of the 3MRA
modeling system ecological risk assessment framework, describing the spatial extent of exposure
as well as providing an important metric for reporting ecological risks.
As described in Section 3, the ecological setting for each site layout file includes one or
more habitats. Within each habitat, four home range sizes are delineated, and the receptor-
specific home ranges are assigned to the smallest one of these four home ranges. For example,
the home range of the short-tailed weasel (135,000 m2) fits into home range 3, but not home
ranges 2 or 1. Therefore, the short-tailed weasel is always assigned to home range 3. The
receptor-specific home ranges were simplified to consist of four size ranges because the current
1 Root zone, rather than surficial, soil concentrations were used to be more representative of the types of
exposures likely to be reflected in soil-to-organism bioaccumulation factors (e.g., direct ingestion of soil
invertebrates from deeper soil horizons; ingestion of prey that feed primarily on plant matter).
11-3
-------
Section 11.0
Terrestrial Food Web Module
version of the 3MRA modeling system requires all spatial information to be specified in the site
layout file a priori (i.e., home ranges cannot be created on the fly).2 The four home range areas
overlap within each habitat so that the predatory-prey relationships are maintained. Thus,
physical access by predator species to prey species is built into the site layout file by ensuring
that each home range overlaps the other home ranges. From an exposure perspective, a receptor
species eats plants, earthworms, and soil invertebrates within its home range, as well as any
mobile prey species that cross into its home range. From the perspective of the Terrestrial Food
Web Module, the contaminant concentrations in plants, earthworms, and soil invertebrates are
spatially consistent within the receptor species' home range; the concentrations in other prey
categories are spatially consistent within the prey species' home range. Consider the following
simple example.
Forest Habitat
Home Range 3
Home Range 1
Home Range 2
Home Range 4
In a hypothetical forest habitat shown in Figure 11-2, the short-tailed weasel is assigned
to home range 3. Its diet consists of soil invertebrates as well as small vertebrates, including
mammals, herpetofauna, and birds. Table 11-2 provides example concentrations of contaminant
y for soil and prey categories relevant to the weasel's diet calculated by the Terrestrial Food Web
Module during a simulation. The calculated concentrations in soil, soil invertebrates, and other
prey categories within home range 3 are internally consistent to that home range. That is, the
concentration of contaminant^ in small
mammals assigned to home range 3 is
6.7E-07 mg/kg; this concentration
includes prey species that are eaten by
the short-tailed weasel (e.g., mice,
shrews), but does not reflect the
concentration for the prey category to
which the short-tailed weasel belongs
(omnivorous vertebrates). As
described in Section 15 (Ecological
Exposure Module), the weasel will eat
any combination of small mammals,
herpetofauna, and small birds in its diet
if prey species that belong to these
categories are assigned to home ranges
that overlap the weasel's home range.
However, the weasel may only eat soil
invertebrates from within its home
range (because these organisms are
relatively sessile), and the incidental
ingestion of soil is presumed to come
exclusively from within the weasels's
home range. The concentration of
omnivorous vertebrates assigned to
home range 3 in this example is not shown in Table 11-2.
Figure 11 -2. Hypothetical forest habitat with four home
ranges shown.
2 The Terrestrial Food Web Module was designed to accept a unique home range for each wildlife species
assigned to a habitat. Although this scheme has been simplified by using only four sizes, the module is blind to the
fact that home ranges of the same size and geographical location are found in the site layout file.
11-4
-------
Section 11.0
Terrestrial Food Web Module
Table 11 -2. Example Exposure Concentrations for Contaminant y Calculated
by the Terrestrial Food Web Module
hood Item
Nome R;m»c 1
1 Ionic R;m»c 2
1 Ionic R;m»c 3
1 Ionic kiin^c 4
Soil
2.6E-07
4.9E-07
3.8E-08
3.1E-09
Small mammals
1.3E-06
1.6E-06
6.7E-07
NA
Small herpetofauna
9.4E-07
3.2E-06
NA
NA
Soil invertebrates
2.3E-06
9.1E-06
4.3E-07
1.1E-08
Small birds
2.3E-06
5.9E-06
NA
NA
NA = no receptor species in this prey category were assigned to this home range.
The methods used by the Terrestrial Food Web Module to calculate contaminant
concentrations in soil, plants, soil invertebrates (and earthworms), and vertebrate prey species
are described in the following sections. Additional detail on the data can be found in Volume II
of this report.
11.2.1 Calculate Contaminant Concentrations in Soil
The Terrestrial Food Web Module calculates contaminant concentrations in both surficial
soils and root zone soils for each home range. Because the difference between calculated
surficial and depth-averaged soil concentrations depends only on the depth of the soil horizon,
the same equations are used to calculate the spatially averaged contaminant concentration in
each home range. The surficial soil, or top layer (top 1 cm), is relevant to incidental soil
ingestion by foraging animals; thus, the surficial soil concentration is used only by the
Ecological Exposure Module to calculate the applied dose to terrestrial receptors. The root zone
soils (concentrations averaged over 5 cm) represent deeper soils that are relevant to plant uptake
through root-to-plant translocation and to direct exposures of soil fauna such as earthworms.
Consequently, the root zone soil concentrations are used by the Terrestrial Food Web Module to
predict plant and prey concentrations, and are passed to the Ecological Risk Module for use in
evaluating risks to the soil community and terrestrial plants (as receptors).
The spatially averaged soil contaminant concentrations may include a contribution from
regional watershed subbasins as well as from drainage subbasins associated with the WMU.
However, erosion and runoff within the local watershed is only relevant to certain types of
WMUs. For aerated tanks and surface impoundments, it is assumed that controls are sufficient
to prevent erosion and runoff releases within the drainage subbasin that contains the unit (i.e.,
there is no erosion or runoff directly from the aerated tank or surface impoundment). For the
land application unit, the average soil concentration in each home range may include
contributions from both the drainage subbasin and the regional watershed subbasin, depending
on the placement of the home range with respect to the WMU. In the hypothetical example in
Figure 11-2, the contaminant concentrations in soil for home range 3 and home range 4 could
include a contribution from the drainage subbasin associated with the WMU. It was assumed
that wildlife could use parts of the land application unit as habitat; however, because the landfill
and waste pile are assumed not to serve as suitable habitat for plants or prey species, these units
11-5
-------
Section 11.0
Terrestrial Food Web Module
are assumed to not contribute to direct, soil-based exposures to terrestrial receptors. The
calculation of spatially averaged contaminant concentrations in soil for each home range is a
three-step process (equations apply to both surficial soils and root zone soils). First, the
contaminant concentration in soil for the home range due to contributions from watershed
subbasins that overlap the home range is calculated as follows:
- £
where
CSoiidsbHomeRangej = annual average contaminant concentration in soil for home range j
due to contributions from watershed subbasins (mg/kg soil)
Csoildsbl = annual average contaminant concentration in watershed subbasin i
(mg/kg soil)
FracdsblHomeRangej = fraction of home range j impacted by watershed subbasin i
(unitless).
Finally, the total spatially averaged contaminant concentration for soil in the home range
is calculated by summing the contributions from the watershed subbasins and the drainage
subbasins associated with the WMU as follows:
11-6
-------
Section 11.0
Terrestrial Food Web Module
7 .AVE = C .ws . + C .dsb /-11 ^
so^ HomeRange i so'' HomeRange i s0^ HomeRange^ ' '
where
CSoiiAVEHomeRangej = annual spatially averaged contaminant concentration in soil for
home range j (mg/kg soil)
CSoiiWSHomeRangej = annual average contaminant concentration in soil for home range j
due to contributions from watershed subbasins (mg/kg soil)
CSoiidsbHomeRangej = annual average contaminant concentration in soil for home range j
due to contributions from WMU local drainage subbasins (mg/kg
soil).
11.2.2 Calculate Total Contaminant Concentrations in Plants
The Terrestrial Food Web Module calculates the total contaminant concentration in the
plant categories shown in Table 11-1 by summing the concentrations for all potential exposure
pathways for plants. Unlike the Farm Food Chain Module, the Terrestrial Food Web Module
does not distinguish between protected and exposed fruits and vegetables, because wildlife are
assumed to consume the outer surfaces of fruits and vegetables. Therefore, the exposed fruit and
vegetable algorithms from the Farm Food Chain Module are used for all plants in the Terrestrial
Food Web Module. For all aboveground plants, the total contaminant concentrations are
converted from dry weight (DW) to wet weight (WW) by adjusting for the plant moisture
content. These WW concentrations (also referred to as whole weight or fresh weight) are
required by the Ecological Exposure Module. This conversion is not needed for root vegetables,
because the concentration is calculated in WW directly. The methodology used to calculate
contaminant concentrations in plants is described fully in Section 10 (Farm Food Chain Module).
11.2.3 Calculate Contaminant Concentrations Soil Invertebrates
Contaminant concentrations in soil invertebrates are estimated as a function of the root
zone soil concentrations in each home range using contaminant-specific soil-to-tissue
bioconcentration factors3 for earthworms and other soil invertebrates. For each home range, the
tissue concentrations in earthworms and other terrestrial invertebrates are calculated as follows:
C. J™ . = C ,AVE x BCF. . (11-4)
invert HomeRange1 s HomeRange1 invert ' )
3 Bioconcentration factors (BCFs) for terrestrial prey are defined as the ratio of the contaminant
concentration in the animal to the contaminant concentration in soil; BCFs are intended to reflect relevant exposure
pathways to the study species (e.g., ingestion of contaminated soil and food).
11-7
-------
Section 11.0
Terrestrial Food Web Module
where
i AVE j
invert HomeRange
invert
annual average contaminant concentration in soil invertebrate or
earthworms for home range j (mg/kg tissue)
p AVE
1 1
soil HomeRange
annual average contaminant concentration in soil for home
range j (mg/kg soil)
BCF:
invert
contaminant-specific bioconcentration factor for soil
invertebrates (kg soil / kg tissue).
11.2.4 Calculate Contaminant Concentrations in Vertebrate Prey Categories
The Terrestrial Food Web Module calculates a range of concentrations in vertebrate prey
categories (illustrated in Table 11-1) across the four home range sizes. For example, the dietary
data for a fox assigned to a terrestrial habitat indicates that part of its diet will consist of small
mammals (e.g., rabbits, shrews, mice). The spatial linkages between the fox (predator) and
various small mammals (prey) are represented in the site layout by allowing the respective home
ranges to overlap. However, the proportion of each species consumed by the fox is unknown;
the fox may consume any combination of these animals depending on prey availability, dietary
preferences, and numerous other factors that affect prey selection. To address this uncertainty,
the Terrestrial Food Web Module estimates the contaminant concentrations for relevant prey
categories in each of the home ranges, and reports the minimum and maximum values. These
values are used by the Ecological Exposure Module to randomly select (assuming a uniform
distribution) an effective concentration for prey categories that represents the full range of
concentrations to which a predator may be exposed.
Contaminant concentrations in prey categories are estimated as a function of the root
zone soil concentrations in each home range, contaminant-specific soil-to-tissue
bioconcentration factors for each prey category, and the fraction of each prey species home
range that is contained within the habitat (i.e., the fraction that does not extend beyond the 2 km
radius of the AOI). Vertebrate prey categories considered in the module include small
mammals, small birds, small herpetofauna, large omniverous vertebrates, and large herbivorous
vertebrates. Because these prey types consist of a variety of species (e.g., a Cerulean Warbler,
Marsh Wren, and Northern Bobwhite are all considered small birds), the module calculates the
concentration of contaminants for species within each prey category and then selects the
maximum and minimum values from that range. Because minimum and maximum
concentrations are reported for each prey type, the Terrestrial Food Web Module allows the
Ecological Exposure Module to represent variability in wildlife diets. Predatory animals may
consume prey in different areas of the habitat (changes in foraging patterns) and are highly likely
to be opportunistic in their feeding habits (altering diet due to prey availability).
The concentrations in vertebrate prey species (for each home range) are calculated as
follows:
C
prey1 HomeR ange-
HomeRange-
prey
x FRAC
prey1 Horn eR ange •
(11-5)
11-8
-------
Section 11.0
Terrestrial Food Web Module
where
preyi HomeRangeJ
annual average concentration in vertebrate prey species i in home
range j (mg/kg tissue)
CSOiiAVEHomeRangej = annual average root-zone soil concentration for home range j
(mg/kg soil)
Fracpreyi HomeRangeJ = fraction of home range j that falls within the habitat for vertebrate
prey species i (unitless).
The last term in the equation, fraction of home range that falls within the habitat, is used
to prorate the contaminant concentrations calculated in vertebrate prey species. The framework
underlying the ecological exposure assessment does not assume that 100 percent of the diet
originates from the home range, since the home ranges are sometimes larger than the entire
habitat area. Consequently, the Terrestrial Food Web Module prorates the vertebrate species-
specific tissue concentrations to reflect the fact that a species may not derive all of its food from
within the AOI. Although the contaminant concentrations estimated for various vertebrate prey
species are spatially consistent within each home range, there is no requirement that
concentrations are based exclusively on food and soil from within the species' home range.
Nevertheless, the module assumes that some exposure will always occur, and sets the fraction of
home range that falls within the habitat to a minimum of 20 percent.
11.3 Module Discussion
11.3.1 Strengths and Advantages
The Terrestrial Food Web Module was developed to predict contaminant concentrations
in soil, plants, and terrestrial prey to which receptors may be exposed. The module represents a
significant step forward in addressing the spatial and temporal variability of ecological
exposures. Specific strengths and advantages of the Terrestrial Food Web Module include the
following:
Takes full advantage of functionality in the Farm Food Chain Module. By
translating the plant categories described for the Farm Food Chain Module into
relevant categories for wildlife, the contaminant concentrations predicted for
plants are consistent across both modules. As a result, many of the advantages of
the Farm Food Chain Module are also conferred on the Terrestrial Food Web
Module. For example, the calculations of contaminant concentrations in plants
reflect both the soil and air pathways of exposure, as appropriate, given the
properties of a particular chemical constituent. In addition, the Terrestrial Food
Web Module also calculates and reports spatially averaged soil concentrations for
each habitat and home range delineated at a site; this efficiency reduces the
contaminant-specific bioconcentration factor for vertebrate prey
species i (kg soil / kg tissue)
11-9
-------
Section 11.0
Terrestrial Food Web Module
overall ran time because average soil concentrations need only be calculated once
by the Terrestrial Food Web Module.
¦ Maintains spatial attributes of soil, plant, and prey concentrations calculated
during the simulation. A major strength of the Terrestrial Food Web Module is
its ability to report concentrations for soil, plants, and terrestrial prey that are
spatially consistent with the information in the site layout file. The module
calculates these concentrations for each receptor assigned to each habitat, and
reports the attributes of that concentration, including space (home range), food
category (e.g., small mammals), and time (year of the simulation). The calculated
tissue concentrations reflect the spatial boundaries for each ecological receptor
assigned to a given habitat. For example, the tissue concentration predicted for a
short-tailed weasel (which could be prey for the red-tailed hawk) is prorated by
that portion of the weasel's home range that is contained within the habitat. This
information is critical to the ecological risk assessment framework in the 3MRA
modeling system to ensure that exposures are spatially consistent with the site
layout information, and is intended to capture the range of contaminant
concentrations in soil, plants, and prey to which any particular receptor may be
exposed.
¦ Algorithms, methods, and data are consistent with numerous EPA analyses.
The Terrestrial Food Web Module, although highly flexible, was developed to
reflect current best practices in use at EPA, such as the Region VI Protocol for
Ecological Risk Screening. The module and data inputs were intentionally based
on established methods to permit application to a wide variety of chemical
contaminants. Alternative approaches, such as fugacity-type models, were
considered to be too data intensive (particularly with respect to chemical
properties), inconsistent with the performance goals for model run times, and
difficult to integrate with other model components. Consequently, the Terrestrial
Food Web Module offers significant advantages over other approaches in ease of
use for a variety of chemicals, transparency, modeling efficiency, and flexibility
in its application and future modifications.
11.3.2 Uncertainty and Limitations
The methodology developed to estimate contaminant concentrations in soil, plants, and
prey, as implemented by the Terrestrial Food Web Module, carries certain assumptions and
limitations, and acknowledges several important sources of uncertainty. The limitations and
uncertainties relevant to calculating forage and produce concentrations discussed under the Farm
Food Chain Module are also relevant to the Terrestrial Food Web Module. Thus, the following
discussion has been limited to issues exclusively related to the Terrestrial Food Web Module.
¦ The Terrestrial Food Web Module is reliant on empirical (or empirically
derived) soil-to-tissue bioconcentration factors. The lack of good quality
studies that investigate bioaccumulation in terrestrial systems contributes
considerable uncertainty. As explained in detail in Volume II of this report on
data collection, values for terrestrial BCFs reflect a broad range of data quality,
11-10
-------
Section 11.0
Terrestrial Food Web Module
with some studies reporting only percentiles off of the raw study data (e.g., a 50th
percentile from a rank order). Moreover, the bioconcentration factors for a
particular prey item vary across species, dietary preferences, seasonal resource
requirements, and climatic conditions. For instance, Sample et al. (1998)
indicated that vertebrates of varying dietary preferences (i.e., herbivores,
omnivores, insectivores) accumulate contaminants to different degrees. A default
uptake factor of 1 was used for contaminants for which BCFs could not be
identified or calculated. The default value may be unrealistically high for
contaminants that are poorly absorbed or are metabolized and excreted by
animals. Although biomagnification in the food web has been demonstrated for
very few contaminants in terrestrial systems, there is great uncertainty associated
with the BCFs derived for prey items in this analysis, particularly in the use of
default BCFs.
The Terrestrial Food Web Module does not capture the movement of
contaminants through the food chain. Because the module relies on soil-to-
tissue BCFs to predict contaminant concentrations in food items, contaminant
movement from biological compartment to biological compartment is not
simulated. For example, although the module predicts concentrations in plant
categories such as seeds and nuts, the contaminant concentrations in herbivorous
vertebrates are calculated based on soil-to-vertebrate BCFs rather than biotransfer
factors from plant matter into the tissues of herbivores. As a result, the module
does not represent contaminant movement in the food chain in the sense that the
Aquatic Food Web Module simulates movement among various biological
compartments of predator and prey species. The BCF approach, although
supported by available data, introduces uncertainty in the calculations of exposure
concentration in terrestrial food items. Because few contaminants have been
shown to accumulate in terrestrial food webs, it is likely that the BCF approach
overestimates exposures to wildlife.
The derivation methods for some prey categories rely on extrapolation of soil
BCFs from sediment studies. For invertebrates, BCFs based on sediment
exposure were adopted in the absence of data quantifying exposure via soil. This
approach introduces uncertainty because these two exposure pathways are not
equivalent. The primary literature reports bioconcentration data on various
terrestrial insects such as beetles, especially for metals; however, these data are
difficult (i.e., costly) to locate through traditional search methods because the data
are generally found as secondary assessments conducted within larger site-
specific risk analyses. The BCFs adopted from Oak Ridge work (Bechtel Jacobs,
1998) represent the best alternative, given current data limitations. Similarly, the
BCFs for herpetofauna (i.e., reptiles and amphibians) were also identified from
primary literature searches based on sediment rather than soil exposures. Because
only sediment data were identified for herpetofauna, these bioaccumulation
factors were adopted until soil-derived uptake factors can be identified; however,
there is significant uncertainty in applying sediment exposures to estimate soil
uptake values. In addition, the bioaccumulation data on herpetofauna were
gleaned almost exclusively from studies on amphibians. Application of
11-11
-------
Section 11.0
Terrestrial Food Web Module
amphibian-based BAFs and BCFs to reptiles is associated with great uncertainty,
given the physiological differences between these classes.
¦ The uptake and accumulation of contaminants within categories of plants
(e.g., exposed vegetables) is assumed to be similar. The algorithms used to
estimate biotransfer factors do not distinguish physiological differences across
various kinds of plants. For example, the category "forage" includes forbs,
grasses, fungi, shrubs, trees, and unclassified plants. Therefore, in estimating
biotransfer factors for this category, it is implicitly assumed that the physiological
differences in different plant species do not significantly affect contaminant
loadings in plant tissues. The use of empirical data on selected plant species
(typically crops) also assumes similar mechanisms of uptake and accumulation.
Adopting the methodology developed for the Farm Food Chain Module to predict
contaminant concentrations in plants that are eaten by wildlife introduces
additional uncertainty in the application of those model algorithms.
¦ A reasonable averaging depth for root zone soil concentrations is 5 cm. In
view of the multiple purposes of this soil concentration (e.g., to evaluate risks to
soil fauna and predict tissue concentrations in prey using soil-based BCFs), and
given the performance goals for science modules in the 3MRA modeling system,
5 cm was selected as a depth that is ecologically relevant (with regard to many
organisms that live in the soil) and still consistent with the design specifications
for the science modules (i.e., run time). Although the root zone depth may be
specified by the user, the selection of 5 cm represents a balancing of model
performance with ecological relevance. However, this assumption built into the
Watershed Module limits the value of information on soil concentrations, and
constitutes a source of uncertainty in its application within the Ecological
Exposure and Risk Modules. In reality, soil fauna can occupy many different soil
horizons, and plant roots (and exposures to soil contaminants) can extend well
below 5 cm. For example, some important species of soil organisms live
primarily in deeper soil horizons, while others live almost exclusively in the top
litter layer of soil. This top layer is highly enriched in organic carbon and
humics, and may serve as a significant source of exposure to certain soil species.
11.4 References
Bechtel Jacobs. 1998. Empirical Models for the Uptake of Inorganic Chemicals from Soil by
Plants. Prepared for the U.S. Department of Energy under contract DE-AC05-
980R22700.
Sample, B.E., J.J. Beauchamp, R.A. Efroymson, and G.W. Suter, II. 1998. Development and
Validation of Bioaccumulation Models for Small Mammals. Prepared for the
U.S. Department of Energy under contract DE-AC05-840R21400.
11-12
-------
Section 12.0
Aquatic Food Web Module
12.0 Aquatic Food Web Module
12.1 Purpose and Scope
The Aquatic Food Web Module calculates steady-state contaminant concentrations in
aquatic organisms (e.g., fish, benthic invertebrates, aquatic plants) consumed by human and
ecological receptors. These concentrations are used as input to the Human Exposure and
Ecological Exposure Modules to calculate applied dose to receptors of interest. Figure 12-1
shows the relationship and information flow between the Aquatic Food Web Module and the
3MRA modeling system.
Key Data Inputs
Waterbody type
Food web structure
K ,
Surface
Water
Module
Water Column and
Sediment Concentrations
Aquatic
Food Web
Module
Fish
Concentrations
Aquatic Plant and
Prey Concentrations
Human
Exposure
Module
Ecological
Exposure
Module
Figure 12-1. Information flow for the Aquatic Food Web Module in the 3MRA modeling system.
For each year in the simulation, the Aquatic Food Web Module predicts annual average
contaminant concentrations in aquatic biota in freshwater waterbodies in the area of interest
(AOI) considered capable of supporting fish (referred to as "fishable" waterbodies). The model
is flexible enough to be applied to different types of waterbodies, including stream reaches,
rivers, lakes, ponds, and permanently flooded wetlands. Simple freshwater food webs were
constructed for each type of waterbody to depict the major functional and structural components
of "healthy" freshwater ecosystems. The components of each food web represent major
categories of aquatic biota in freshwater systems: aquatic macrophytes, phytoplankton,
periphyton, zooplankton, benthic detritivores, benthic filter feeders, and various feeding guilds
of fish. Some of these concentrations are used internally to calculate concentrations in fish,
while other concentrations are reported as a time series for use in calculating exposures to
wildlife and humans, as well as to calculate ecological hazard (e.g., hazard to sediment
12-1
-------
Section 12.0
Aquatic Food Web Module
dwellers). Thus, the Aquatic Food Web Module determines which data are appropriate for use in
a given waterbody and calculates concentrations in the aquatic biota assigned to that waterbody.
Specifically, the Aquatic Food Web Module performs the following functions:
1. Selects food web appropriate for each waterbody. The Aquatic Food Web
Module matches an appropriate food web with each waterbody identified as
fishable within the AOI. Eight freshwater food webs were developed to capture
the variability in freshwater systems. They represent warmwater streams/rivers,
wetlands, ponds, and lakes; and coldwater stream s/ri vers, wetlands, ponds, and
lakes.
2. Constructs dietary matrix for food web. The Aquatic Food Web Module uses a
constrained, random prey preference sampling approach that selects preference
fractions at random between the minimum and maximum, assuming a uniform
distribution. This approach allows for the dietary composition to reflect the full
range of variability inherent in the diets of freshwater fish.
3. Calculates contaminant concentrations in food web. The Aquatic Food Web
Module calculates concentrations for the biota assigned to each freshwater food
web. The biota categories include
¦ Phytoplankton,
¦ Periphyton,
¦ Zooplankton,
¦ Aquatic plants (macrophytes),
¦ Benthic filter feeders,
¦ Benthic detritivores,
¦ Fish in various feeding guilds, and
¦ Apex predator fish.
The model will only calculate concentrations for biota that are assigned to a
particular food web and waterbody. The calculations involve mechanistic models
or the use of empirical data on bioaccumulation.
4. Reports contaminant concentrations for fish consumed by wildlife and
humans. Food webs for freshwater aquatic systems typically have a single apex
predator species (trophic level 4 [TL4]) and a number of other fish species that
occupy different feeding guilds, such as benthic feeders (e.g., catfish). These
other species are, for the purposes of exposure assessment, often grouped into the
category of trophic level 3 (TL3), indicating that they are both predator and prey
in the food web. To predict exposures for wildlife and humans that eat TL3 fish,
the model calculates an average contaminant concentration—both wholebody and
filet—for fish that fall into the category of TL3. In addition, the Aquatic Food
Web Module reports the tissue concentration for the apex predator fish in each
waterbody. The wholebody fish concentrations are used by the Ecological
Exposure Module, and the filet concentrations are used by the Human Exposure
Module.
12-2
-------
Section 12.0
Aquatic Food Web Module
12.2 Conceptual Approach
12.2.1 Select Food Web Appropriate for Each Waterbody
The 3MRA modeling system was designed to use site-based data to support a national-
level assessment strategy. Consequently, an important goal for the Aquatic Food Web Module
was to capture the variability in freshwater systems across the contiguous United States,
particularly with respect to species composition and dietary preferences. The first step in
accomplishing this was to develop a set of representative food webs. The second was to assign
each waterbody in the AOI to one of these representative food webs.
Develop Representative Freshwater Food Webs. Upon reviewing literature sources on
freshwater systems and food webs, it was apparent that the food web structure and, in many
instances, the fish species, are similar across many different freshwater habitats (Schindler et al.,
1996). There are common elements to
virtually all aquatic communities (e.g.,
periphyton, benthic detrivores, aquatic
plants). Many options to represent variability
were considered, ranging from a basic food
web consisting of three compartments, to a
complex food web that could represent
virtually any type of freshwater system.
However, the development of a single food
web—whether basic or complex—was not
consistent with the 3MRA framework goal to
use site-based information to support
national-scale assessments. Moreover,
considerable data were identified to vary the
complexity of the food web and represent
many different species of fish. To take
advantage of these data, freshwater food webs
were constructed such that the major
functional elements were represented as
simply as possible. Several useful tenets from
the literature were adopted as guidelines in
developing the aquatic food webs. They are
summarized in the text box.
The resulting freshwater food webs provide a useful framework to model contaminant
transport and fate in freshwater waterbodies, offer a reasonable representation of energy flows
typical of different habitats, and capture variability in a manner that is appropriate for the
application of the 3MRA modeling system. Figure 12-2 shows an example of a freshwater lake
food web.
The sources used to construct the freshwater food webs reflect a broad perspective,
ranging from biodiversity assessments to game fishing enthusiasts. These data sources were not
only used in constructing the food webs, but also in characterizing the fish species and in
Basic tenets in constructing aquatic food webs
¦ Predator-prey interactions should follow
common sense (i.e., larger fish eat smaller fish)
and the system should be balanced in the sense
that all prey items are connected in the food
web.
¦ Size distinctions within feeding guilds of fish
should consider the potential biomass of the
most preferred prey item and the interactions
with other components of the food web.
¦ Larger waterbodies tend to support more
functional elements, and therefore, are typically
more complex than smaller waterbodies (e.g.,
lakes are more complex than ponds).
¦ Flowing waters tend to be less complex than
still waters and have low plankton density; as a
result, zooplankton are not an important food
web component.
¦ Warmwater systems tend to support a more
diverse aquatic community than coldwater
systems; as a result, they tend to have more
functional niches and are more complex.
12-3
-------
Section 12.0
Aquatic Food Web Module
deciding which species of fish are
eaten by human receptors. These
sources offered a wide range of detail
on food webs: from site-specific
assessments to more general
constructs developed for regional
analyses. Many sources included
qualitative descriptions of aquatic
habitats, as well as indications of fish
species that are considered "typical"
for these habitats; in particular, the
fishing references provided very
useful information on the
characteristics of fish that inhabit
various freshwater systems. Many of
these texts also indicated whether the
preferred water temperature for a
given species of fish was cold water or
warm water.
The food web for each type of freshwater system reflects a number of characteristics of
the waterbody, such as water temperature, flow (i.e., flowing versus static systems), dominant
zones (e.g., pelagic versus littoral zones), and preferences of fish species for certain aquatic
systems. The fish species assigned to the eight representative food webs represent a specific
functional niche to which the species belong (e.g., feeding guilds; trophic level; size). For
example, because the zooplankton density in streams tends to be low, a fish species that
primarily feeds on zooplankton is unlikely to be assigned to stream habitats. In contrast,
piscivore-dominated lakes are characterized by large-bodied zooplankton with high grazing rates
(Schindler et al., 1996). In these lake systems, we would expect to find planktivorous species of
fish as an integral part of the food web. Thus, the concept of functional niche is particularly
important in the selection of food webs because these niches were used to inform the selection of
appropriate fish species and associated data for each habitat (e.g., lipid fraction, body weight,
dietary preferences).
The freshwater food webs contain between eight and 12 of the biota types possible in
freshwater systems. The following biota are found in freshwater food webs:
¦ Periphyton: algal species typical of freshwater systems that adhere to rocks and
detrital material; also includes small crustaceans, which are not modeled in the
Aquatic Food Web Module;
¦ Phytoplankton: primary producers in pelagic systems;
¦ Aquatic macrophytes: vascular aquatic plants (e.g., submerged, emergent);
¦ Zooplankton: various invertebrates that graze on phytoplankton;
<
-
Sa mo n ids
Smelt
Sculpin
Oligochaetes
Pontopo
Figure 12-2. Example of simplified food web for
freshwater lake (Gobas et al., 1993).
12-4
-------
Section 12.0
Aquatic Food Web Module
¦ Benthic detrivores: benthic dwellers that break down detritus in sediment (e.g.,
amphipods);
¦ Benthic filter feeders: benthic organisms that feed through a filtration
mechanism;
¦ TL3 benthivore: TL3 fish whose primary feeding preference is benthic
organisms (divided into small, medium, and large);
¦ TL3 planktivore: TL3 fish whose primary feeding preference is zooplankton;
¦ TL3 omnivore: TL3 fish who have no clear feeding preferences (divided into
small, medium, and large); and
¦ TL4 piscivore: TL4 piscivorous fish that serve as the apex predator for the
community.
Table 12-1 summarizes the food webs constructed for use in modeling contaminant
movement in the freshwater food webs. This matrix indicates which food web components are
assigned to each aquatic habitat. The presence of a prey item such as zooplankton may not
require that an obligate planktivore be assigned to the food web. In wetlands, for example,
omnivorous fish tend to feed on zooplankton, as well as on other biota (e.g., periphyton, benthos,
detritus); therefore, the biota assignments reflect the goal of accounting for significant
predator-prey interactions without imposing artificial constraints on the food web structure.
Additional details on the development of habitat-specific food webs (e.g., warmwater wetland)
are found in the 1999 background documents (U.S. EPA, 1999).
Assign Each Waterbody in the AOI to a Representative Food Web. The structure of
aquatic food web and the fish species used to parameterize the Aquatic Food Web Module were
based largely on the type of waterbody. The waterbody characteristics were developed using
information from geographical information system (GIS) data sources (e.g., National Wetlands
Inventory, or NWI), landscape data on the size of standing waterbodies, and certain conventions
used in fish ecology to identify coldwater, stenothermic fish (i.e., coldwater species with narrow
tolerance for temperature changes). For example, the threshold adopted for categorizing waters
as warm or cold is based on a maximum temperature of 25°C—the water temperature above
which coldwater, stenothermic fish cannot survive. The food web structure for warmwater
streams is typically more complex than an analogous coldwater stream; as a result, there are
frequently more functional niches in a warmwater stream than might be found in a coldwater
stream. Stenothermic fish with clear temperature preferences were generally assigned either to
warmwater or coldwater systems, but not to both. Other species that are found in both
warmwater and coldwater habitats were assigned to both categories.
12-5
-------
Section 12.0
Aquatic Food Web Module
Table 12-1. Matrix of Biota in Food Webs for Freshwater Systems in 3MRA
Biota
Coldwater habitats
Warmwater habitats
Stream
Wetland
Pond
Lake
Stream
Wetland
Pond
Lake
Periphyton
~
~
~
~
~
~
~
~
Phytoplankton
~
~
~
~
~
~
Aquatic macrophytes
~
~
~
~
~
~
~
~
Zooplankton
~
~
~
~
~
~
Benthic detrivores
~
~
~
~
~
~
~
~
Benthic filter feeders
~
~
~
~
TL3 benthivore (small)
~
~
~
TL3 benthivore (medium)
~
~
~
~
~
~
~
TL3 benthivore (large)
~
TL3 planktivore (small)
~
TL3 planktivore (medium)
~
~
TL3 planktivore (large)
~
TL3 omnivore (small)
~
~
~
~
~
TL3 omnivore (medium)
~
~
~
~
~
~
TL3 omnivore (large)
~
~
~
TL4 piscivore
~
~
~
~
~
~
~
12.2.2 Construct Dietary Matrix for Food Web
The aquatic food webs provide the framework for the Aquatic Food Web Module
simulations; the food webs identify fish and other biota presumed to be present for each
waterbody and indicate the predator-prey interactions. However, most fish are opportunistic
feeders, leading to significant variability in the dietary composition, even within feeding guilds.
There is also tremendous variability in the dietary preferences of fish associated with life stage,
region, prey density, and a host of other conditions. For many contaminants, the primary route
of exposure is through gill exchange, and therefore, the dietary preferences are not important
contributors to bioaccumulation of contaminants. However, for contaminants shown to
biomagnify with trophic level, the dietary composition may have a significant influence on fish
contaminant concentrations.
To address these variabilities in dietary composition, a random sampling algorithm was
developed to select prey preference fractions for each type of fish assigned to a food web. The
model constructs this dietary matrix for each simulation (defined as the combination of a site,
waste management unit [WMU], and contaminant) so that the predicted concentrations in fish
reflect the substantial variability in the diet. Data obtained from literature sources were
evaluated to create a database of prey preference ranges for biota in various food webs. The
Aquatic Food Web Module uses the database to (1) construct the food web-specific dietary
composition, (2) rank prey items from most preferred to least preferred, and (3) estimate the prey
12-6
-------
Section 12.0
Aquatic Food Web Module
preferences for each biota type (how much of each item is in the total diet). In this context, the
dietary composition refers to both the dietary items consumed (e.g., zooplankton, small
benthivorous fish) and the fraction of each dietary item consumed by various fish components in
the food web.
The approach to construct the dietary matrix and select prey preferences in the food web
was based on two objectives: (1) to observe the bounds as defined by the empirical data on prey
preferences, and (2) to allow variability within the bounds to be exercised. Estimating prey
preferences is accomplished using a constrained, random prey preference sampling algorithm
that selects preference fractions at random between the minimum and maximum, assuming a
uniform distribution. The algorithm maintains overall dietary preferences and allows for the
dietary composition to reflect the full range of variability inherent in the diets of freshwater fish.
The algorithm developed to solve this problem treats each dietary fraction as a "resource" to be
allocated among the prey items for a particular fish. Before any dietary fractions are assigned
for a given fish, the value of the resource remaining to be allocated is 1 (i.e., complete diet).
After all dietary fractions have been assigned (zero fractions are allowed), the value of the
resource remaining to be allocated is zero. Thus, for a given fish and prey item, the assignment
prey preference fraction must consider the minimum and maximum preference values for that
prey item, as well as the amount of resource remaining (dietary fraction yet to be assigned). The
algorithm used to perform the random sampling is described in the text box on the next page.
12.2.3 Calculate Contaminant Concentrations in Food Web
The Aquatic Food Web Module was developed to be flexible to use empirical data,
mechanistic models, or simple regression equations that use physical-chemical properties to
calculate contaminant concentrations in food web biota. The choice of method is determined by
the type of contaminant modeled, as well as by the availability of suitable empirical data on
bioaccumulation. The Aquatic Food Web Module can model the following three groups of
contaminants:1
¦ Hydrophobic, non-ionizable organics. This group includes dioxin, dioxin-like
chemicals and high-molecular-weight, highly halogenated organics that tend to
bioaccumulate in aquatic systems. Biota concentrations for these contaminants
are estimated using either a mechanistic model or empirical data on
bioaccumulation from the surface water column or the sediment. Currently, the
Aquatic Food Web Module uses a mechanistic model based on the work of Gobas
(e.g., Gobas, 1993) to predict tissue concentrations in aquatic biota. The module
uses data on metabolism when available.
1 The predictive models in the Aquatic Food Web Module can also be applied to ionizable organics.
Chemical property values such as the octanol-water partition coefficient (K,m) generated by the chemical properties
processor reflect environmental conditions (e.g., pH) that affect ionization in sediment and surface water.
12-7
-------
Section 12.0
Aquatic Food Web Module
Random Sampling Algorithm Used to Determine Aquatic Prey Preferences
The issue for the aquatic food web is to select prey preferences throughout the food web matrix
such that the observed bounds are honored (i.e., the empirical data on prey preferences), yet the
allowable variability within the bounds is exercised in a Monte Carlo sense and the diet is complete.
Expressed mathematically, the problem is:
Select Py i = j = 1 Such that
Min;i < P.. < Max.. i = 1 j = 1 M
y y g r ¦>
M P„ = 1.0 i = 1...JV
9=i
v
where
N = number of biota types that are fish
M = number of prey items
Pij = dietary fraction of the prey item for fish i for prey item j
Miny = minimum observed dietary fraction of fish i for prey item j
Maxy = maximum observed dietary fraction of fish i for prey item j.
The algorithm that was developed to solve this problem treats Py as a "resource" to be allocated
among the M prey items for a given biota type offish. Before any dietary fractions are assigned for a
given fish i, the value of the resource remaining to be allocated is 1.0 (i.e., complete diet). After all
dietary fractions have been assigned (zero fractions are allowed), the value of the resource remaining
to be allocated is 0. For a given fish i and prey item j, the assignment (Py) must consider both the Miny
and the Maxy, as well as the amount of resource remaining (dietary fraction yet to be assigned). The
assignment equation for biota type i, assuming a uniform distribution for Py, is:
with the variables defined as follows:
LB- = Maximum[Miriy, RR- - YJc=j+1 Maxik\
= 1-0 " 5jt= i Pik
UB.. = Minimum[Max.., RR- - YJc=j+1 Mmik\
where
LB = lower bound of the range
UB = upper bound of the range
RND = uniform random deviate (0-1).
12-8
-------
Section 12.0
Aquatic Food Web Module
¦ Hydrophilic, non-ionizable organics. This group includes straight-chain
aliphatics, as well as low-molecular-weight organics such as benzene, that do not
tend to bioaccumulate in aquatic systems. Biota concentrations for these
contaminants can be estimated using a regression model, a mechanistic model, or
empirical data on bioaccumulation. Currently, the Aquatic Food Web Module
uses a regression model to predict bioaccumulation factors (BAFs), unless the
contaminant is considered to be easily metabolized. For metabolizable organic
chemicals, the Aquatic Food Web Module uses empirical BAFs to predict tissue
concentrations. Tissue concentrations in aquatic plants and benthic filter feeders
are calculated based on sediment concentrations and biota sediment accumulation
factors (BSAFs).
¦ Metals and mercury. This group includes metals in various valence states as
well as mercury as methyl mercury. Biota concentrations are predicted using
empirical data on bioaccumulation. For metals, the Aquatic Food Web Module
uses median values of BAFs based on a data set that meets specific data quality
objectives (see Volume II). For mercury, tissue concentrations of mercury are
estimated using empirical BAFs based on dissolved methyl mercury
concentrations from thq Mercury Report to Congress (U.S. EPA, 1997). The
Mercury Report to Congress provides methyl mercury BAFs for fish in TL3 and
TL4. For both metals and mercury, tissue concentrations in aquatic plants and
benthic filter feeders are calculated based on sediment concentrations and BSAFs.
Depending on the contaminant of concern, the Aquatic Food Web Module requires inputs
on the characteristics of the fish species assigned to each food web (e.g., lipid fraction, body
weight, dietary preferences), as well as characteristics of the waterbody (e.g., the fraction organic
carbon in bed sediment). The Surface Water Module generates contaminant concentrations in
surface water (dissolved and total) and in sediment (dissolved in pore water and total); therefore,
the Aquatic Food Web Module is not required to predict contaminant concentrations in the
different phases in the environmental media (i.e., sorbed versus freely dissolved phases) but gets
these from the Surface Water Module. For all contaminants, the Aquatic Food Web Module
calculates tissue concentrations for benthic filter feeders, aquatic plants, and TL3 and TL4 fish.
Intermediate concentrations in food items such as zooplankton and benthic detritivores are
calculated only for hydrophobic organic chemicals and are used by the mechanistic model as
dietary inputs.
The technical approach implemented by the Aquatic Food Web Module to predict
contaminant concentrations in biota is summarized below for the three groups of contaminants.
More detailed discussions on the algorithms and data sources used by the Aquatic Food Web
Module can be found in the 1999 background documents (U.S. EPA, 1999) and in Volume II of
this report, respectively.
Hydrophobic, Non-Ionizable Organic Contaminants. For hydrophobic, non-ionizable
organic chemicals, the Aquatic Food Web Module is based on the modeling constructs
developed by Gobas et al. (1993), Thomann (e.g., Thomann et al., 1992), and a number of other
researchers (e.g., Abbott et al., 1995; Campfens and Mackay, 1997; Morrison et al., 1997; and
Zaranko et al., 1997). These models were chosen because they were developed specifically for
12-9
-------
Section 12.0
Aquatic Food Web Module
organic chemicals with significant potential to bioaccumulate in the food web, they do not have
prohibitive data requirements, they are flexible in their application to different waterbodies and
food webs, they have been peer reviewed and validated with field data, and the series of linear
equations can be solved using a flexible matrix solution technique. The form of model does not
require a system-based solution; the equations, while coupled, can be solved sequentially.
However, to accommodate the future use of more complex predator-prey relationships, which
may involve true simultaneity, a more generic, system solution was believed desirable and was
developed for the Aquatic Food Web Module. In the matrix solution, the contaminant
concentrations in fish are predicted as follows:
where
, _ [t, X Cfd t kD E(Fmc. X 9]
** (ki * kE + kM + kG) (
Cfish' = annual average whole-body concentration in fish i (mg/kg wet weight [WW])
k, = rate constant for contaminant uptake from water (L/kg-d)
Cwfd = annual average freely dissolved concentration in surface water (mg/L)
kD = rate constant for contaminant uptake from food (L/d)
FraCj = fraction of prey item j included in diet (unitless)
Cj = annual average concentration in prey item j in diet (mg/kg WW)
k2 = rate constant for contaminant elimination to water (L/d)
kE = rate constant for elimination by fecal egestion (L/d)
kM = rate constant for metabolic transformation of contaminant (L/d)
kG = rate constant for growth dilution (L/d).
Under steady-state conditions, the contaminant concentrations in periphyton,
phytoplankton, zooplankton, and aquatic macrophytes are predicted as follows, assuming that the
BCF is satisfactorily approximated by Kok:
CJ
C^d ((LipFraCj Kow) + (NonLipFraCj 0.033 Kow) + WaterFracj) (12-2)
where
Cj
LipFraCj
K,
C
ow
fd
NonLipFraCj =
WaterFrac: =
annual average concentration in prey item j (mg/kg WW)
lipid fraction in prey item j (unitless)
octanol-water partition coefficient (L/kg lipid)
annual average freely dissolved concentration in surface water (mg/L)
nonlipid organic carbon fraction in prey item j (unitless)
water fraction in prey item j (unitless).
The tissue concentrations in benthic detrivores and benthic filter feeders are also derived
assuming steady-state conditions. As described in Gobas (1993), equilibrium partitioning theory
may be used to predict concentrations in benthic organisms as follows:
12-10
-------
Section 12.0
Aquatic Food Web Module
^sediment *
Cj =
P OC
fOCsedii
sediment)
Ph
(12-3)
v
LipFraCj
where
Cj
= annual average concentration in prey item j in benthos (mg/kg WW)
c
^sediment
= annual average total concentration in sediment (mg/kg)
Poc
= density of organic carbon in sediment (kg/L)
for* -
AWVsediment
= fraction of organic carbon in sediment (unitless)
Plip
= density of lipids in benthos (kg/L)
LipFraCj
= fraction of lipid in prey item j in benthos (kg lipid/kg tissue).
Gobas points out that although more detailed models to estimate concentrations in benthos can
be derived, this model has been shown to be in better agreement with field data (e.g., see Gobas
et al., 1989; Landrum et al., 1992).
Hydrophilic, Non-Ionizable Organic Contaminants. For hydrophilic, non-ionizable
organic chemicals (defined operationally as organic chemicals with an octanol-water partition
coefficient value below 10,000), the Aquatic Food Web Module calculates concentrations in fish
using the following equation:
"fish
BAF,. x r
fd
(12-4)
where
Cgsh' = annual average concentration in fish i (mg/kg WW)
BAF; = bioaccumulation factor for fish i (L/kg tissue)
Cwfd = annual average freely dissolved concentration in surface water (mg/L).
For these types of organic chemicals, bioaccumulation is primarily a function of gill exchange
(rather than accumulation of contaminant through ingestion of contaminated food items). The
BAF values are based on empirical data when available. Otherwise, the Aquatic Food Web
Module calculates the BAF using the regression algorithm developed by Bertelsen et al. (1998):
BAFi = 10(^"'log K°w + b*h'log LipFrac> + cfi*h) + WaterFracfiJ (12-5)
12-11
-------
Section 12.0
Aquatic Food Web Module
where
BAF; = bioaccumulation factor for fish i in trophic level p (L/kg tissue)
afis,, = primary slope term (unitless)
bflsh = secondary slope term (unitless)
cflsh = empirical error term (unitless)
LipFraC; = lipid fraction in fish i (kg lipid/kg tissue)
Kow = octanol-water partition coefficient (assume L/kg lipid)
WaterFracfish' = fraction of wholebody fish i in trophic level p that is water (unitless).
The model developed by Bertelsen et al. extends previous work on the bioconcentration of
hydrophilic organics in fish presented by Veith et al. (1980); Mackay (1982); Isnard and
Lambert (1988); and others. Because gill exchange is considered to be the dominant mechanism
by which hydrophilic organics are taken up, a simpler model could be used to predict tissue
concentrations in fish. Uptake through the food web is assumed to be negligible; therefore, it is
not necessary to calculate the concentration in all of the prey items in the aquatic food web.
Metals and Mercury. The contaminant concentration in fish tissue for metals and
mercury is calculated as follows:
-------
Section 12.0
Aquatic Food Web Module
¦ Bergman, Harold L., and Elaine J. Dorward-King (eds.). 1997. Reassessment of
Metals Criteria for Aquatic Life Protection. Society of Environmental
Toxicology and Chemistry Press, Pensacola, FL. Proceedings of the Pellston
Workshop on Reassessment of Metals Criteria for Aquatic Life Protection,
February 10-14, 1996.
¦ Chapman, Peter M., Herbert E. Allen, Kathy Godtfredsen, and Michael N.
Z'Graggen. 1996. Evaluation of bioaccumulation factors in regulating metals.
Environmental Science & Technology, 30(10):448A-452A.
¦ Renner, Rebecca. 1997. Rethinking water quality standards for metals toxicity.
Environmental Science & Technology, 31(10):466A-468A.
Although uptake and accumulation are not of concern for all metals, the impact of surface
water characteristics (particularly dissolved organic carbon) on bioavailability is significant.
Several modeling approaches have been developed recently that can be used to predict
bioavailability (e.g., the Windermere Humic Aqueous Model, or WHAM), and water effects
ratios (WER) provide empirical ratios that can be used to adjust water quality criteria to account
for the mitigating effects of natural waters (see Bergman and Dorward-King, 1997, for
discussion). Moreover, as shown in Figure 12-3, the effects and accumulation of essential
metals change with concentration (i.e., bioaccumulation is nonlinear); thus, a single BAF may be
inappropriate.
k
i
Deficient
Optimal
Toxic
-C
03
CD
-C
E
c
(0
§
Lethal
Lethal
W
Metal concentration in surface water
The state of the
science on metals transport
and fate in aquatic systems
strongly suggests that the
uptake and accumulation
of essential metals (e.g.,
copper, zinc) in fish and
other aquatic organisms
are fundamentally different
than the uptake and
accumulation of
nonessential metals (e.g.,
cadmium, lead).
Moreover, laboratory
studies that report BCFs
may have limited
relevance to the behavior
of metals in the field. The
predominant metal species
in a "natural" aquatic system may be very different than the metal salt studied in the laboratory,
resulting in a very different accumulation profile. The characteristics of the aquatic system and
the presence of other cationic metals significantly influence the uptake and accumulation of a
single metal species. Although understanding of metals behavior has increased since the Aquatic
Figure 12-3. Relationship between essential metal concentration
and organism health (adapted from Chapman et al., 1996).
12-13
-------
Section 12.0
Aquatic Food Web Module
Food Web Module was developed, the module relies on empirical data and a simple model to
predict metal concentrations in aquatic biota. EPA is currently reviewing recently published
studies and data relevant to the development of alternative approaches to predict metal
concentrations in aquatic biota (e.g., Bergman et al., 1992).
12.2.4 Report Contaminant Concentrations for Fish
The Human and Ecological Exposure Modules allow receptors to consume fish from any
of the water bodies to which they have access. Both human and ecological receptors may
consume fish assigned to TL3 and TL4; however, human receptors can only eat TL3 fish that are
designated as edible by humans in the fish database. In addition, human receptors are assumed
to eat only the filet portion of the fish, whereas ecological receptors are assumed to eat the entire
fish. Therefore, the Aquatic Food Web Module performs two processing steps in generating
concentrations of contaminants in fish for use by the exposure modules.
First, the Aquatic Food Web Module calculates the filet concentration by adjusting the
wholebody concentrations of organic chemicals by the relative lipid content of filet (or muscle)
versus the wholebody of the fish. The theory supporting the bioaccumulation of organic
chemicals suggests that virtually all of the contaminant accumulates in the lipid tissue. The
adjustment factor is calculated by dividing the lipid fraction in filet by the lipid fraction in the
wholebody. For example, if the lipid fraction in filet is 3 percent and the lipid fraction in
wholebody is 10 percent, the adjustment factor is 0.3 (and the fraction for wholebody would be
0.7). Thus, by adjusting for the differences in lipid content between filet and wholebody, the
concentration in filet can be estimated from wholebody tissue concentration.
The calculation for hydrophobic organic chemicals is
CfiiJ = Cfish x FiletFrac ' (12-7)
where
Cfi|ct' = annual average concentration in filet for fish i (mg/kg WW)
Cgsh' = annual average concentration in fish i (mg/kg wet weight)
FiletFrac1 = the adjustment factor for filet (unitless).
For hydrophilic organic chemicals, the equation for calculating the BAF (Equation 12-5)
is parameterized to calculate concentrations in filet directly as described in Bertelsen et al.
(1998). For metals and mercury, the wholebody concentrations are presumed to provide a
reasonable approximation of filet concentrations given other uncertainties.
Second, the Aquatic Food Web Module calculates a waterbody-specific annual average
concentration for all TL3 fish filet presumed to be edible by humans, and all TL3 fish
(wholebody) for ecological receptors. Because information on dietary preferences for TL3 fish
is not available, this averaging implies that all TL3 fish have an equal probability of being eaten.
This implies exposure to the average filet or wholebody concentration across TL3 fish in the
waterbody for human and ecological consumption, respectively.
12-14
-------
Section 12.0
Aquatic Food Web Module
12.3 Module Discussion
12.3.1 Strengths and Advantages
The Aquatic Food Web Module was developed to predict contaminant concentrations in
aquatic biota due to long-term chemical releases into surface water. The module offers a number
of advantages relative to other approaches that were considered for the 3MRA modeling system.
Some of the major strengths of the Aquatic Food Web Module include the following:
¦ Applicable to a wide variety of chemicals. The Aquatic Food Web Module and
supporting data can be applied to wide variety of chemical contaminants ranging
from hydrophobic organics to metals. The module recognizes the type of
contaminant and uses the appropriate data and algorithms to predict tissue
concentrations. The module was designed in a modular fashion so that new
science could be incorporated to address specific chemical types. For example,
the simple approach used to predict mercury concentrations in biota could be
upgraded to include a more complex speciation model to simulate mercury
behavior in freshwater systems. The ability to handle such a broad range of
chemicals in a single module, using either empirical data on bioaccumulation or
mechanistic models, is a significant advantage to the 3MRA modeling system.
¦ Flexible enough to address any type of aquatic food web. For bioaccumulative
organic chemicals, the solution technique developed for this module is both
computationally efficient and flexible. The module will accommodate simple
food webs consisting of relatively few components or complex food webs with
well-defined predator-prey interactions. As a result, the food webs that were
defined for application to a national-scale analysis can be modified, and
additional, site-specific food webs may be developed. This flexibility ensures
that the Aquatic Food Web Module will be applicable to representative aquatic
food webs for national-scale analyses as well as user-defined food webs for
site-specific analyses.
¦ Based on peer-reviewed, validated models for organic chemicals. The
governing equations for the Aquatic Food Web Module are based on models that
have been published in peer-reviewed journals and that have been validated using
field and/or laboratory data. In particular, the algorithms used to predict the
contaminant concentrations in aquatic biota reflect the state-of-the-science for
hydrophobic organic chemicals. The module can use data on metabolic
transformation and, because the chemical properties processor can predict
properties based on environmental conditions, it may be applied to ionizable
organics as well. The level of validation of these equations and their widespread
application described in the open literature provides strong support for their use in
the 3MRA modeling system.
¦ Random sampling algorithms represent dietary variability. Dietary
preferences—both in terms of preferred food items and fraction in the diet—are a
potentially important source of variability in the chemical concentrations of
12-15
-------
Section 12.0
Aquatic Food Web Module
hydrophobic organics predicted by the Aquatic Food Web Module. Many species
of fish tend to be opportunistic feeders, and the diet may shift significantly
depending on the site conditions. To account for variability in the diet, the
module includes a random sampling algorithm that selects dietary elements from
the database, and constructs a dietary matrix of the preferences such that 100
percent of the diet is accounted for. The ability to represent this source of
variability across different aquatic food webs is a substantial improvement over
fixing the dietary composition and preferences.
12.3.2 Uncertainty and Limitations
The following uncertainties and limitations are inherent in the Aquatic Food Web
Module:
¦ The module is implemented assuming steady-state conditions. The Aquatic
Food Web Module cannot be used to evaluate the impacts from storm events, nor
can it be used to distinguish the impacts on tissue concentrations from peak
events and subsequent averaging from long-term, low-level exposures.
¦ The module relies heavily on empirical data for many contaminants. For
contaminants other than dioxin-like compounds and organics, mechanistic models
are not used to predict tissue concentrations. Hence, the Aquatic Food Web
Module estimates tissue concentrations by multiplying empirical BAFs by water
or sediment concentrations. As discussed in Volume II, these BAFs and BSAFs
are measured under conditions that may introduce uncertainty for certain
environmental settings and species.
¦ The module does not allow for separate treatment of essential metals.
Bioconcentration of essential metals is not linear, and modeling approaches are
now available to account for nonlinearity. Bioconcentration of essential metals
tends to be much greater at low concentrations than at higher concentrations
because organisms actively seek to sequester necessary nutrients. Because many
metals are regulated in biological systems, the apparent bioconcentration of
metals at low concentrations may simply result in metal accumulation at
"healthy" levels.
¦ The module currently lacks the capability to use sediment concentrations
directly in predicting tissue concentrations. The Aquatic Food Web Module
was developed primarily to use dissolved and total contaminant concentrations to
predict tissue concentrations. Although sediment concentrations are used in
predicting uptake and accumulation into benthic dwellers, the Aquatic Food Web
Module lacks the necessary algorithms to use these data directly to predict
concentrations in plants or fish. For certain chemicals (e.g., dioxins), it may be
useful to build this functionality into the module to provide greater flexibility in
data use.
12-16
-------
Section 12.0
Aquatic Food Web Module
The module has not been validated in field studies for all freshwater systems.
Much of the modeling theory on which the Aquatic Food Web Module is based is
widely accepted and has been used in numerous analyses. In particular, the
methods used to predict concentrations of organics have been validated in
coldwater lakes. However, the module has not been validated for other
freshwater aquatic habitats, nor has it been validated for application in a national-
scale analysis.
12.4 References
Abbott, Joan D., Steven W. Hinton, and Dennis L. Borton. 1995. Pilot scale validation of the
river/fish bioaccumulation modeling program for nonpolar hydrophobic organic
compounds using the model compounds 2,3,7,8-TCDD and 2,3,7,8-TDCF.
Environmental Toxicology and Chemistry, 14(11): 1999-2012.
Allen, H. E., and D. J. Hansen. 1996. The importance of trace metal speciation to water quality
criteria. Water Environment Research, 68(l):42-54. January.
Bergman, Harold L. and Elaine J. Dorward-King (eds.). 1997. Reassessment of Metals Criteria
for Aquatic Life Protection. Society of Environmental Toxicology and Chemistry Press,
Pensacola, FL. Proceedings of the Pellston Workshop on Reassessment of Metals
Criteria for Aquatic Life Protection 10-14 February 1996.
Bertelsen, Sharon L., Alex D. Hoffman, Carol A. Gallinat, Colleen M. Elonen, and John W.
Nichols. 1998. Evaluation of log KOW and tissue lipid content as predictors of chemical
partitioning to fish tissues. Environmental Toxicology and Chemistry, 17(8): 1447-1455.
Borgmann, U., and W.P. Norwood. 1999. Assessing the toxicity of lead in sediments to Hvalella
azteca: The significance of bioaccumulation and dissolved metal. Canadian Journal of
Fisheries and Aquatic Sciences: 56 (8), pp 1494-1503.
Burkhard, Lawrence P. 1998. Comparison of two models for predicting bioaccumulation of
hydrophobic organic chemicals in a Great Lakes food web. Environmental Toxicology
and Chemistry, 17(3):383-393.
Campfens, Jan, and Donald Mackay. 1997. Fugacity-based model of PCB bioaccumulation in
complex aquatic food webs. Environmental Science & Technology, 31(2):577-583.
Chapman, P.M., F. Wang, C. Janssen, G. Persoone, and H.E. Allen. 1998. Ecotoxicology of
metals in aquatic sediments: Binding and release, bioavailability, risk assessment, and
remediation. Canadian Journal of Fisheries and Aquatic Sciences: 55 (10) pp 2221-2243.
Chapman, Peter M., Herbert E. Allen, Kathy Godtfredsen, and Michael N. Z'Graggen. 1996.
Evaluation of bioaccumulation factors in regulating metals. Environmental Science &
Technology, 30(10):448A-452A.
12-17
-------
Section 12.0
Aquatic Food Web Module
Clayton, John R., Jr., Spyros P. Pavlou, and Norman F. Breitner. 1977. Polychlorinated
biphenyls in coastal marine zooplankton: bioaccumulation by equilibrium partitioning.
Environmental Science & Technology, 11(7):676-682. July.
Felley, J. D. 1992. Chapter 6: Medium—Low-Gradient Streams of the Gulf Coastal Plain. In:
Biodiversity of the Southeastern United States - Aquatic Communities, C. T. Hackney, S.
M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New York, NY. pp. 233-
269.
Garman, G. C., and L. A. Nielsen. 1992. Chapter 8: Medium-sized Rivers of the Atlantic
Coastal Plain. In: Biodiversity of the Southeastern United States - Aquatic Communities,
C. T. Hackney, S. M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New
York, NY. pp. 315-349.
Gelwick, F. P., and W. J. Matthews. 1996. Chapter 22: Trophic Relations of Stream Fishes. In:
Methods in Stream Ecology, F. R. Hauer and G. A. Lamberti (eds.). Academic Press, San
Diego, CA. pp. 475-492.
Gensemer, R.W., and R.C. Playle. 1999. The bioavailability and toxicity of aluminum in
aquatic environments. Critical Reviews in Environmental Science and Technology: 29
(4), pp 315-450.
Gerking, Shelby D. 1994. Feeding Ecology of Fish. Academic Press, Inc., San Diego, CA.
Goodyear, K.L, and S. McNeill. 1999. Bioaccumulation of heavy metals by aquatic macro-
invertebrates of different feeding guilds: A review. Science of the Total Environment:
229 (1-2), pp 1-19.
Geyer, H., G. Politzki, and D. Freitag. 1984. Prediction of ecotoxicological behavior of
chemicals: relationship between n-octanol/water partition coefficient and
bioaccumulation of organic chemicals by alga Chlorella. Chemosphere, 13(2):269-284.
Gobas, Frank A. P. C., Kathryn E. Clark, Wan Ying Shiu, and Donald Mackay. 1989.
Bioconcentration of polybrominated benzenes and biphenyls and related
superhydrophobic chemicals in fish: role of bioavailability and elimination into the feces.
Environmental Toxicology and Chemistry, 8:231-245.
Gobas, Frank A. P. C., Edmund J. McNeil, Lesley Lovett-Doust, and G. Douglas Haffner. 1991.
Bioconcentration of chlorinated aromatic hydrocarbons in aquatic macrophytes.
Environmental Science and Technology, 25(5):924-929.
Gobas, Frank A. P. C. 1993. A model for predicting the bioaccumulation of hydrophobic
organic chemicals in aquatic food-webs: application to Lake Ontario. Ecological
Modelling, 69:1-17.
12-18
-------
Section 12.0
Aquatic Food Web Module
Gobas, Frank A. P. C., Jim R. McCorquodale, and G. D. Haffner. 1993. Intestinal absorption
and biomagnification of organochlorines. Environmental Toxicology and Chemistry,
12:567-576.
Goldstein, Robert M., and Thomas P. Simon. 1998. Chapter 7: Toward a united definition of
guild structure for feeding ecology of North American freshwater fishes. In: Assessing
the Sustainability and Biological Integrity of Water Resources Using Fish Communities,
Thomas P. Simon (ed.). CRC Press, Boca Raton, FL. pp. 123-138.
Hauer, F. Richard and Gary A. Lamberti (eds.). 1996. Methods in Stream Ecology. Academic
Press, San Diego, CA.
Hauer, F. R., and V. H. Resh. 1996. Chapter 16: Benthic Macroinvertebrates. In. Methods in
Stream Ecology, F. R. Hauer and G. A. Lamberti (eds.). Academic Press, San Diego,
CA. pp. 339-369.
Hoffman, Joe D. 1992. Numerical Methods for Engineers and Scientists. McGraw-Hill, Inc.,
New York, NY.
Hollis, L., J.C. McGeer, D.G. McDonald, and C.M. Wood. 1999. Cadmium accumulation, gill
Cd binding, acclimation, and physiological effects during long-term, sublethal Cd
exposure in rainbow trout. Aquatic Toxicology: 46 (2), pp 101-119.
Hudson, R.J.M. 1998. Which aqueous species control the rates of trace metal uptake by aquatic
biota? Observations and prediction of non-equilibrium effects. Science of the Total
Environment: 219 (2-3), pp 95-115.
Hudson, Robert J. M., Steven A. Gherini, Carl J. Watras, and Donald B. Porcella. 1994.
Chapter V. 1: Modeling the biogeochemical cycle of mercury in lakes: the mercury
cycling model (MCM) and its application to the MTL study lakes. In: Mercury
Pollution: Integration and Synthesis, Carl J. Watras and John W. Huckabee (eds.). Lewis
Publishers, Boca Raton, FL. pp. 473-523.
Isnard, P., and S. Lambert. 1988. Estimating bioconcentration factors from octanol-water
partition coefficient and aqueous solubility. Chemosphere, 17(l):21-34.
Jackson, L.J. 1998. Paradigms of metal accumulation in rooted aquatic vascular plants.
Science of the Total Environment: 219 (2-3), pp 223-231.
Killgore, Jack and Jan Jeffrey Hoover. 1992. A Guide for Monitoring and Evaluating Fish
Communities in Bottomland Hardwood Wetlands. WRP Technical Note FW-EV-2.2.
U.S. Army Engineers, Waterways Experiment Station, Vicksburg, MS. January.
Kim, S. Don, H. Ma, H.E. Allen, and D.K. Cha. 1999. Influence of dissolved organic matter on
the toxicity of copper to Ceriodaphnia dubia: Effect of complexation kinetics.
Environmental Toxicology and Chemistry: 18 (11), pp 2433-2437.
12-19
-------
Section 12.0
Aquatic Food Web Module
Landrum, Peter F., Henry Lee, II, and Michael J. Lydy. 1992. Toxicokinetics in aquatic
systems: model comparisons and use in hazard assessment. Environmental Toxicology
and Chemistry, 11:1709-1725.
Lee, David S., Carter R. Gilbert, Charles H. Hocutt, Robert E. Jenkins, Don E. McAllister, and
Jay R. Stauffer, Jr. 1980. Atlas of North American Freshwater Fishes, 1980-EtSeq.
North Carolina State Museum of Natural History, Raleigh, NC.
Li, H. W., and J. L. Li. 1996. Chapter 18: Fish Community Composition. In. Methods in
Stream Ecology, F. R. Hauer and G. A. Lamberti (eds.). Academic Press, San Diego,
CA. pp. 391-406.
Lowe, R. L., and G. D. Laliberte. 1996. Chapter 13: Benthic Stream Algae: Distribution and
Structure. In: Methods in Stream Ecology, F. R. Hauer and G. A. Lamberti (eds.).
Academic Press, San Diego, CA. pp. 269-293.
Mackay, D. 1982. Correlation of bioconcentration factors. Environmental Science and
Technology, 16(5):274-278.
MacRae, R.K., D.E. Smith, N. Swoboda-Colberg, J.S. Meyer, and H.L. Bergman. 1999. Copper
binding affinity of rainbow trout (Oncorhvnchus mvkiss) and brook trout (Salvelinus
fontinalis) gills: Implications for assessing bioavailable metal. Environmental
Toxicology and Chemistry: 18 (6), pp 1180-1189.
Menzel, R. G., and C. M. Cooper. 1992. Chapter 10: Small Impoundments and Ponds. In:
Biodiversity of the Southeastern United States - Aquatic Communities, C. T. Hackney, S.
M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New York, NY. pp. 389-
420.
Mitsch, William J., and James G. Gosselink. 1993. Wetlands. 2nd Edition. Van Nostrand
Reinhold, New York, NY.
Morrison, Heather A., Frank A. P. C. Gobas, Rodica Lazar, D. Michael Whittle, and G. Douglas
Haffner. 1997. Development and verification of a benthic/pelagic food web
bioaccumulation model for PCB congeners in western Lake Erie. Environmental Science
& Technology, 31(ll):3267-3273.
Mulholland, P. J., and D. R. Lenat. 1992. Chapter 5: Streams of the Southeastern Piedmont,
Atlantic Drainage. In: Biodiversity of the Southeastern United States - Aquatic
Communities, C. T. Hackney, S. M. Adams, and W. H. Martin (eds.). John Wiley &
Sons, Inc., New York, NY. pp. 193-231.
Palmer, M. A., and D. L. Strayer. 1996. Chapter 15: Meiofauna. In: Methods in Stream
Ecology, F. R. Hauer and G. A. Lamberti (eds.). Academic Press, San Diego, CA.
pp. 315-337.
12-20
-------
Section 12.0
Aquatic Food Web Module
Rand, Gary M. (ed.). 1995. Fundamentals of Aquatic Toxicology: Effects, Environmental Fate,
and Risk Assessment. 2nd Edition. Taylor & Francis, Washington, DC.
Reinfelder, J.R., N.S. Fisher, S.N. Luoama, J.W. Nichols, and W.X. Wang. 1998. Trace element
trophic transfer in aquatic organisms: A critique of the kinetic model approach. Science
of the Total Environment: 219 (2-3), pp 117-135.
Renner, Rebecca. 1997. Rethinking water quality standards for metals toxicity. Environmental
Science & Technology, 31(10):466A-468A.
Schindler, Daniel E., Stephen R. Carpenter, Kathryn L. Cottingham, Xi He, James R. Hodgson,
James F. Kitchell, and Patricia A. Soranno. 1996. Chapter 8: Food web structure in the
littoral zone coupling to pelagic trophic cascades. In: Food Webs: Integration of
Patterns & Dynamics, Gary A. Polis and Kirk O. Winemiller (eds.). Chapman and Hall,
New York, NY. pp. 96-105.
Smith, C. Lavett. 1994. Fish Watching: An Outdoor Guide to Freshwater Fishes. Cornell
University Press, Ithaca, NY.
Smock, L. A., and E. Gilinsky. 1992. Chapter 7: Coastal Plain Blackwater Streams. In:
Biodiversity of the Southeastern United States - Aquatic Communities, C. T. Hackney, S.
M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New York, NY. pp. 271-
313.
Soballe, D. M., B. L. Kimmel, R. H. Kennedy, and R. F. Gaugush. 1992. Chapter 11:
Reservoirs. In: Biodiversity of Southeastern United States - Aquatic Communities, C. T.
Hackney, S. M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New York,
NY. pp. 421-474.
Spencer, Matthew, and Steve Beaulieu. 1997. Bioaccumulation Factors for Polycyclic Aromatic
Hydrocarbons in Fish: Methods for Dealing With Uncertainty (Unpublished).
Stephan, C. E. 1993. Derivation of Proposed Human Health and Wildlife Bioaccumulation
Factors for the Great Lakes Initiative. Environmental Research Laboratory, Office of
Research and Development, U.S. Environmental Protection Agency, Duluth, MN.
March.
Sternberg, Dick. 1996. Freshwater Gamefish of North America. Cowles Creative Publishing,
Inc., Minnetonka, Minnesota.
Thomann, Robert V., John P. Connolly, and Thomas F. Parkerton. 1992. An equilibrium model
of organic chemical accumulation in aquatic food webs with sediment interaction.
Environmental Toxicology and Chemistry, 11:615-629.
Traas, Theo P., Joan A. Stab, P. Roel Kramer, Wim P. Cofino, and Tom Aldenberg. 1996.
Modeling and risk assessment of tributyltin accumulation in the food web of a shallow
freshwater lake. Environmental Science & Technology, 30(4): 1227-1237.
12-21
-------
Section 12.0
Aquatic Food Web Module
U.S. EPA (Environmental Protection Agency). 1997. Mercury Study Report to Congress.
Volumes I - VIII. EPA 452/R-97/003. U.S. Environmental Protection Agency, Office of
Air Quality Planning and Standards and Office of Research and Development,
Washington, DC. Website at http://www.epa.gOv/ttn/uatw/l 12nmerc/mercury.html.
December.
U.S. EPA (Environmental Protection Agency). 1998a. Lake and Reservoir Bioassessment and
Biocriteria. Technical Guidance Document. EPA 841-B-98-007. U.S. Environmental
Protection Agency, Office of Wetlands, Oceans, and Watersheds, Office of Water, Office
of Science and Technology, Washington, DC. August.
U.S. EPA (Environmental Protection Agency). 1998b. Guidelines for Ecological Risk
Assessment. (Final). EPA/630/R-95/002F. U.S. Environmental Protection Agency, Risk
Assessment Forum, Washington, DC. April.
U.S. EPA (Environmental Protection Agency). 1998c. Ambient Water Quality Criteria
Derivation Methodology for Human Health. Technical Support Document. (Final
Draft). EPA/822/B-98/005. U.S. Environmental Protection Agency, Office of Science
and Technology, Washington, DC. July.
U.S. EPA (Environmental Protection Agency). 1999. Aquatic Food Web Module. Background
and Implementation for the Multimedia, Multipathway, Multireceptor Risk Assessment
(3MRA) Model for HWIR99. Office of Solid Waste, Washington DC. October.
Veith, G. D., K. J. Macek, S. R. Petrocelli, and J. Carroll. 1980. An evaluation of using
partition coefficients and water solubility to estimate bioconcentration factors for organic
chemicals in fish. In: Aquatic Toxicology, ASTMSTP 707, J. G. Eaton, P. R. Parrish, and
A. C. Hendricks (eds.). American Society for Testing and Materials, pp. 116-129.
Wallace, J. B., J. R. Webster, and R. L. Lowe. 1992. Chapter 4: High-Gradient Streams of the
Appalachians. In: Biodiversity of the Southeastern United States - Aquatic Communities,
C. T. Hackney, S. M. Adams, and W. H. Martin (eds.). John Wiley & Sons, Inc., New
York, NY. pp. 133-191.
Wetzel, Robert G. 1983. Limnology. 2nd Edition. Saunders College Publishing, Fort Worth,
TX.
Zaranko, Danuta T., Ronald W. Griffiths, and Narinder K. Kaushik. 1997. Biomagnification of
polychlorinated biphenyls through a riverine food web. Environmental Toxicology and
Chemistry, 16(7): 1463-1471.
12-22
-------
Section 13.0
Human Exposure Module
13.0 Human Exposure Module
13.1 Purpose and Scope
The Human Exposure Module calculates the applied dose to human receptors from
ingestion and inhalation of contaminated media and food. Detailed information on the Human
Exposure Module can be found in the background document (U.S. EPA, 2000). Figure 13-1
shows the relationship and information flow between the Human Exposure Module and the
Key Data Inputs
• Food intake rates
• Inhalation rate
• Body weight
Fish Concentrations
Air Concentrations
Soil Concentrations
Doses
Ground Water
Concentrations
Soil Concentrations
Food Item and
Soil Concentrations
Aquatic
Food Web
Module
Air
Module
Watershed
Module
Human Risk
Module
Human
Exposure
Module
Farm Food
Chain
Module
Land-based
Source
Modules
Vadose Zone
and Aquifer
Modules
Figure 13-1. Information flow for the Human Exposure Module
in the 3MRA modeling system.
13-1
-------
Section 13.0
Human Exposure Module
3MRA modeling system. The Human Exposure Module uses media and food concentrations
calculated in the media and food web modules to calculate applied doses. These doses are used
by the Human Risk Module to calculate risk measures.
The purpose of the Human Exposure Module is to calculate the exposure inputs needed
by the Human Risk Module to calculate receptor risk and hazard. For carcinogens, risk is
calculated from applied dose (in mg/kg-d). For noncarcinogens, hazard via the ingestion
pathway is also based on dose, whereas hazard via the inhalation pathway is based on the air
concentration to which a receptor is exposed. Thus, the Human Exposure Module has the
following functions:
1. Calculates ambient air concentrations. The Human Exposure Module
calculates areal average ambient air concentrations for farms. Ambient (outdoor)
air concentrations for residents are calculated by the Air Module and used by the
Human Exposure Module.
2. Calculates shower air concentration. The Human Exposure Module calculates
shower air concentration from ground water concentration.
3. Calculates dose from inhalation of carcinogens. The Human Exposure Module
calculates dose from inhalation of ambient (outdoor) air and shower air for
carcinogens. Noncarcinogenic risk for inhalation exposures is based directly on
air concentration, not dose; thus, the Human Exposure Module does not calculate
inhalation dose for noncarcinogens.
4. Calculates dose from ingestion of contaminants in media or food. The Human
Exposure Module calculates dose for carcinogens and noncarcinogens from
ingestion of soil, ground water, produce (fruits and vegetables), beef, milk, and
fish.
5. Calculates dose from ingestion of contaminants in breast milk. The Human
Exposure Module calculates dose to infants from ingestion of contaminated breast
milk. This route of exposure is assessed only for dioxin-like chemicals, and only
for infants.
Doses and air concentrations are calculated for each contaminant and site, as well as for
each
¦ Receptor type. The module calculates exposures for two types of human
receptors: residents and farmers. These are distinguished by how they are
located—residents at a single exposure point and farmers on an area representing
a farm. Residents are further defined by various kinds of behavior that lead to
different profiles of exposure, such as home gardening and recreational fishing.
Farmers may also be defined by behavior such as recreational fishing. Human
receptor types considered in the 3MRA modeling system are a function of the
goals of a particular analysis, and are defined by the receptor data used for the
analysis.
13-2
-------
Section 13.0
Human Exposure Module
Age cohort. The human receptors are divided into five age cohorts (infants under
1 year, children aged 1 to 5 years, children aged 6 to 11 years, children aged 12 to
19 years, and adults) that are used to determine the most appropriate exposure
factor data for various pathways, such as body weights, inhalation rates, and
consumption rates for various food products.
Exposure pathway. The Human Exposure Module considers nine potential
exposure pathways. Depending on the exposure inputs entered into the model, all
pathways could be considered for all receptor types in a given simulation.
Table 13-1 shows the pathways modeled, and indicates which receptor types are
modeled for each pathway by default. All pathways are modeled by default for
all age cohorts with two exceptions: the breast milk pathway is modeled only for
infants (and no other pathways are modeled for infants), and shower inhalation is
modeled only for children aged 12 to 19 years and adults. All of these defaults
can be changed by the user through the parameterization of the receptor-specific
inputs.
Table 13-1. Default Pathways Considered by Receptor Type
Km*|)lor Tj pe
Homo lied' l);iin
Resident Homo (>;irdciK'r Beef l';inner l);iir\ l';inner
Piilhwii.t Resident lislur (.;udener l-isher hunier lislur hunier l-isher
Air inhalation
/
/
/
/
/
/
/
/
Shower inhalation
/a
/a
/a
~a
/
/
/
/
Ground water ingestion
/a
/a
/a
~a
/
/
/
/
Soil ingestion
/
/
/
/
/
/
/
/
Produce ingestion
/
/
/
/
/
/
Beef ingestion
/
/
Milk ingestion
/
/
Fish ingestion
/
/
/
/
Breast milk ingestion
/
/
/
/
/
/
/
/
a Ground water and shower pathways are considered for residents and home gardeners only if Census data indicate
the presence of private wells in the Census block group. All farms are assumed to have a private well.
Location. The Human Exposure Module calculates exposures for residents for a
single point. For the current data set, the point is placed at the centroid of each
Census block in the area of interest (AOI). The module calculates spatially
averaged exposure for farmers at a single farm in each census block group in the
AOI that has Census data for farmers and agricultural land areas.
Year. The module calculates exposures for each year of the simulation. The
exposures predicted by the Human Exposure Module are reported as a time series
13-3
-------
Section 13.0
Human Exposure Module
of annual average applied doses or air concentrations. Thus, the equations
presented in this section are applied to input data for each year. All temporal
averaging for exposure durations exceeding 1 year is done by the Human Risk
Module.
13.2 Conceptual Approach
As shown in Table 13-1, receptors are defined by exposures through various pathways;
however, not all pathways may be completed at every location for each site, depending on the
characteristics at each site (e.g., presence of private drinking water wells or fishable
waterbodies). Pathways common to all receptors are inhalation of ambient air and incidental
ingestion of soil; residents are assumed to breathe contaminated air and ingest contaminated soil
at locations within a site at which contaminant releases are predicted. However, only a subset of
receptors have private drinking water wells. This subset is the only group of receptors that could
be exposed to contaminated ground water through direct drinking water ingestion and inhalation
associated with showering. However, the presence of ground water wells does not necessarily
mean that exposures will be estimated by the module. In some situations, the concentration of a
contaminant of concern will be zero because of site-specific characteristics. For example,
individuals may use drinking water wells and have no exposure because the wells are located
outside of the predicted contaminated ground water plume. Similarly, those receptors who use a
public water supply (whether from ground water or surface water) drink and shower with treated
water that meets all drinking water standards and, thus, are assumed to have no exposure
attributable to their water supply.
For the purpose of modeling exposure, the major differences between various receptor
types and age cohorts are the specific pathways to which the receptor is potentially exposed and
the exposure factors used in the dose calculations. Exposure factors for ingestion rates,
inhalation rates, and body weights are randomly selected for each receptor type and each age
cohort from distributions can be found in Volume II and are based on data from EPA's Exposure
I''actors Handbook (U.S. EPA, 1997).
With the exception of the shower/bathroom air concentration algorithms, the
methodologies and equations used by the Human Exposure Module to convert contaminant
concentrations in environmental media or food into human exposure values were obtained from
Methodology for Assessing Health Risks Associated with Multiple Pathways of Exposure to
Combustor Emissions (U.S. EPA, 1998), also known as the MPE methodology, and formerly
referred to as the Indirect Exposure Methodology (IEM). The MPE methodology represents the
state of the science with respect to providing reliable guidance in the proper conduct of
assessments of risks that may result from multimedia, multipathway exposures. The National
Center for Environmental Assessment (NCEA) prepared the MPE methodology as an update to
EPA's 1990 IEM document. Most of the revisions in the MPE methodology are based on
Scientific Advisory Board (SAB) and public comments on IEM. Earlier versions of this
document have undergone internal EPA and external peer review. The breast milk exposure
pathway algorithms and data were based on MPE and the Dioxin Reassessment (U.S. EPA,
1998, 2000). The shower algorithms were adapted from peer-reviewed sources (McKone, 1987
and Little, 1992).
13-4
-------
Section 13.0
Human Exposure Module
Exposures from inhalation of ambient air and shower air are expressed differently
depending on whether the exposure is being evaluated for a carcinogenic or a noncarcinogenic
effect. When inhalation exposures for carcinogenic effects are evaluated, the exposure is
expressed as an applied dose in terms of mg/kg-d. For noncarcinogenic effects, the inhalation
exposure is expressed as a concentration in mg/m3. This is because the RfC is in units of mg/m3
and must be compared to an ambient concentration without explicitly using an inhalation rate or
body weight. Ingestion exposures are estimated the same way for carcinogens as for
noncarcinogens and are expressed as an applied dose in
mg/kg-d.
13.2.1 Calculate Ambient Air Concentrations for Inhalation Exposures
The Human Exposure Module calculates annual average ambient air inhalation
concentrations. These are used to calculate dose (also in the Human Exposure Module) for
carcinogens and to predict the hazard for noncarcinogens (in the Human Risk Module).
For residents, the air concentration is simply the point estimate for the ambient air
concentration at each receptor point, which is calculated by the Air Module. For farms, the
Human Exposure Module calculates spatially averaged air concentration over the area of the
farm using air concentration data calculated by the Air Module. This concentration is the sum of
air vapor and particles (less than 10 |im).
13.2.2 Calculate Shower Air Concentrations for Inhalation Exposures
The Human Exposure Module calculates the annual average air concentration of a
contaminant to which the receptor is exposed associated with showering, including both time
spent in the shower and time spent in the bathroom immediately after the shower. This
concentration is used by the Human Exposure Module for carcinogens to predict dose, and is
used directly by the Human Risk Module to predict hazard for noncarcinogens. The primary
input is the ground water contaminant concentration calculated by the Aquifer Module.
The concentrations in shower air and bathroom air are related to each other and are
determined by solving a system of two coupled differential equations that describe the time-
varying changes in average contaminant concentration in the shower and the bathroom,
respectively, over the duration of a shower when the shower water is contaminated with
contaminant. The change in these air concentrations over time is the result of injecting
contaminated spray into an initially clean air environment rather than changes in the inputs over
time (e.g., the contaminant concentration in the water, which is constant within a given year).
The coupled differential equations are solved using a finite difference approximation. The
details of the mathematical derivation of the shower and bathroom air concentrations are
described in the U.S. EPA (2000).
The Human Exposure Module calculates the shower and bathroom air concentrations for
each of the time steps (e.g., 20 seconds) over the duration of the shower (e.g., 10 minutes) and
time spent in the bathroom after the shower (e.g., half an hour). These are then averaged to
produce a time-weighted average air concentration for shower-related exposure. This average
shower-related concentration is used in subsequent exposure and risk calculations.
13-5
-------
Section 13.0
Human Exposure Module
13.2.3 Calculate Dose from Inhalation of Carcinogens
Inhalation exposure is calculated for ambient air and shower air for carcinogens. For
noncarcinogens, the ambient and averaged shower/bathroom air concentrations described above
are used directly by the Human Risk Module.
The applied inhalation dose reflects air concentrations for a single year in the simulation.
For each realization (i.e., each time the Human Exposure Module runs), the module will read
new data on air concentrations. Although the exposure factors (e.g., body weight, inhalation
rate) will remain constant throughout the simulation, the exposure profile will change over time
as a function of changing air concentrations.
Inhalation of Contaminated Ambient Air (Carcinogens). Ambient air inhalation
exposure is calculated for all receptors that are predicted to be exposed to contaminated ambient
air from the Air Module. The following equation is used to calculate an annual average dose
from carcinogens inhaled by a human from the ambient air:
j \ ^'ambient /' ^ /i o i\
DOSC ambient = ^ C13"1)
where
Doseambient = annual average applied dose from ambient air inhalation (mg/kg-d)
Cambient = totcil annual average ambient air concentration - vapor phase and
particulate-bound (PM10) contaminant (mg/m3)
CRair = inhalation rate for age cohort (m3/d)
BW = body weight for age cohort (kg).
Inhalation of Contaminated Air in the Shower/Bathroom (Carcinogens). The
annual average dose from shower inhalation exposures is calculated for all receptors that have
wells that are predicted to be in the ground water plume contaminated by a WMU. The
governing equation is
C. x 1000 x CR ¦ x T, x Ev,
j-* _ _ _ shower air shower Jreq ^ ~
shower BW x 1440 ~
where
Doseshower
applied dose from inhalation associated with showering (mg/kg-d)
r =
^shower
average air concentration in bathroom and shower due to contaminated
shower water (mg/L)
1,000
units conversion factor (L/m3)
CRair
inhalation rate for age cohort (m3/d)
T =
shower
time exposed during showering (min/event)
Evfreq
event frequency (events/d)
1440
averaging time to convert to daily exposure (min/d).
13-6
-------
Section 13.0
Human Exposure Module
The concentration in the bathroom and shower air is calculated as described in Section 13.2.2.
13.2.4 Calculate Dose from Ingestion of Contaminants in Media or Food
The general equation for calculating annual average dose based on ingestion exposure
through a single pathway (e.g., ingestion of contaminated drinking water or ingestion of
contaminated produce) is as follows1:
Cone x CR x Frac
Dose = (13-3)
BW V 7
where
Dose = annual average dose of contaminant taken into the body (mg/kg-d)
Cone = annual average concentration of a contaminant in the medium or food item
(mg/kg or mg/L)
CR = consumption rate of the medium or food item (L/d or g/d)
Frac = fraction of the total consumption rate that is contaminated (unitless)
BW = body weight (kg).
Consumption rates of food items and body weights vary with age cohort. For some
pathways, the consumption rate is expressed per unit of body weight; in those cases, the CR/BW
terms are combined into one term, which applies to all age cohorts.
The applied dose reflects media and food concentrations for a single year in the
simulation. For each realization (i.e., each time the Human Exposure Module runs), the module
will read new data on media and food concentrations. Although the exposure factors (e.g., body
weight, food ingestion rate) will remain constant throughout the simulation, the exposure profile
will change over time as a function of changing media and food concentrations.
The following subsections provide further detail on the exposure calculation for each
pathway. Any modifications to Equation 13-3 required are noted.
Ingestion of Contaminated Soil. Dose from ingestion of contaminated soil is calculated
for all resident and farm receptors where soils are predicted to be contaminated. Soil
concentrations at residential exposure points are calculated by the Watershed Module. Spatially
averaged surficial soil contaminant concentrations for farms are calculated by the Farm Food
Chain Module.
Ingestion of Contaminated Drinking Water. Exposure from drinking water ingestion
is calculated for all farms and for residential receptors in Census block groups that report private
1 For some pathways, a unit conversion factor may be needed to convert consumption rate to units that will
cancel with the concentration units to produce mg/d. For example, if concentration is in mg/kg and consumption rate
is in g/d, a factor of 0.001 is needed to convert g to kg in the consumption rate.
13-7
-------
Section 13.0
Human Exposure Module
wells and are predicted to be contaminated by a ground water plume from a WMU. It is not
calculated for residential receptors in Census block groups that report no private wells.
Ingestion of contaminated drinking water is based only on ground water concentrations,
which are calculated by the Aquifer Module. Surface water drinking supplies are assumed to be
treated in accordance with drinking water regulations.
Ingestion of Contaminated Produce. Ingestion of contaminated produce is evaluated
for farmers and home gardeners, who are residents that grow fruits and vegetables in
contaminated air and soils. Some fraction of residential receptors are identified from Census
data as home gardeners and are assumed to grow produce, and all farm households are assumed
to grow produce. Produce concentrations are an input from the Farm Food Chain Module.
Exposures to homegrown fruits and vegetables are calculated for the following five
categories of produce:
¦ Root vegetables,
¦ Exposed vegetables,
¦ Protected vegetables,
¦ Exposed fruit, and
¦ Protected fruit.
In addition to classifying produce as a fruit or vegetable (so as to account for different
consumption rates), these categories also reflect three classifications—exposed, protected, and
root—to account for differences in concentrations due to the different mechanisms by which
contaminants can accumulate in edible plant tissues. Specifically, the terms "exposed" and
"protected" refer to whether or not the edible portion of a fruit or vegetable that grows above
ground is exposed to the atmosphere. For instance, apples are exposed fruits, and oranges are
protected fruits. Root vegetables grow below the ground where they are not exposed to the
atmosphere, so this distinction is not made for root vegetables, which are contaminated by root
uptake of contaminants from the soil. Potatoes are an example of root vegetables. Exposure to
contaminated produce is the summation of exposures from each of these five categories.
Consumption rates for produce are normalized to body weight; consequently, they are
given in g/kg-d instead of g/d. The consumption rate and body weight terms in Equation 13-3
are therefore replaced by a single body-weight-normalized consumption rate term. All produce
consumption rates are given in wet weight (WW). Therefore, the produce concentrations must
also be per WW.
Ingestion of Contaminated Beef and Milk. Beef and milk ingestion exposure is
calculated for farmers in an AOI when beef and dairy cattle are predicted to be contaminated
from the release at a WMU. Only beef farmers are assumed to ingest home-raised beef, and only
dairy farmers are assumed to ingest home-produced milk. Contaminant concentrations in beef
and milk are generated by the Farm Food Chain Module.
Consumption rates for beef and milk are normalized to body weight; consequently, they
are given in g/kg-d instead of g/d. Therefore, the consumption rate and body weight terms in
13-8
-------
Section 13.0
Human Exposure Module
Equation 13-3 are replaced by a single body-weight-normalized consumption rate term. All beef
and milk consumption rates are given in WW. Therefore, the beef and milk concentrations must
also be per WW.
Ingestion of Contaminated Fish. The recreational fisher has the same exposures as the
resident, the home gardener, and the farmer, but is also exposed through fish ingestion.
Recreational fishers may have more than one favorite fishing spot. Therefore, the fisher is
assumed to fish from up to three fishable reaches within the AOI, and fish ingestion exposure is
calculated using an average fish concentration for those fishable reaches. If fewer than three
fishable reaches occur in the AOI, then the average is calculated over all the fishable reaches that
exist. If more than three exist, three are randomly sampled out of the full set of fishable reaches.
The selection of the three fishable reaches is made twice for each site: once for residential
receptors and once for farm receptors. The fish tissue concentration includes both trophic level 3
(TL3) and trophic level 4 (TL4) fish, with each predicted to have a different fish tissue
concentration in each of the fishable stream reaches in the AOI. The fish tissue concentrations
are predicted in the Aquatic Food Web Module.
13.2.5 Calculate Dose from Ingestion of Contaminants in Breast Milk
Infant exposure occurs through the consumption of contaminated breast milk. To
calculate infant exposure through breast milk, the maternal exposure through all pathways is
summed, and then the resulting breast milk concentration is calculated as follows:
ADD x/ xf
_ mat •> am Jf
milkfat I (13-4)
where
Cmiikfat = annual average concentration in maternal milk fat (mg/kg)
ADDmat = average annual total daily dose consumed by the mother (mg/kg-d)
fam = fraction of ingested contaminant absorbed by the mother (dimensionless)
ff = proportion of contaminant that is stored in maternal fat (dimensionless)
t1/2b = biological half-life of contaminant in lactating women (days)
ffin = fraction of mother's weight that is fat (dimensionless).
Infant exposures are calculated for eight maternal receptor types: resident, home
gardener, beef farmer, dairy farmer, resident/recreational fisher, home gardener/recreational
fisher, beef farmer/recreational fisher, and dairy farmer/recreational fisher. The mother is
assumed to be an adult (as opposed to a teenager) for the purpose of calculating maternal dose in
the infant breast milk pathway.
13-9
-------
Section 13.0
Human Exposure Module
Infant breast milk exposure is then calculated as
\C -ihft x / u + C x (1 - / h )] x / x CR, ... x 0.001
j-} _ L milkfat J mbm aqueous V J mbmJ ax bmuk /1 ^
bmilk ~ rnx^ (13"5)
infant
where
Dosebmilk = annual average applied dose from breast milk ingestion (mg/kg-d)
Cmiikfat = annual average concentration in maternal milk fat (mg/kg)
fmbm = fraction of fat in breast milk (unitless)
Caqueous = annual average concentration in aqueous phase of maternal milk (mg/kg)
fai = fraction of ingested contaminant absorbed by the infant (dimensionless)
CRbmiik = ingestion rate of breast milk (mL/d)
0.001 = units conversion factor (kg/mL)
BWinfant = body weight of infant (kg).
Infant exposure to contaminants via breast milk is calculated only for dioxin-like
chemicals. The concentration in milk fat is calculated as shown in Equation 13-4. The
concentration in the aqueous phase of breast milk is set to zero because dioxin-like compounds
do not accumulate in the aqueous phase. If this pathway were extended to additional
constituents, the aqueous-phase concentration might be important. However, data are currently
inadequate to extend this pathway to other constituents.
13.3 Module Discussion
13.3.1 Strengths and Advantages
The major strengths and advantages of the Human Exposure Module include the
following:
¦ Based on peer-reviewed EPA-ORD methodology. With the exception of the
shower/bathroom air concentration models, the methods and equations used by
the Human Exposure Module to calculate exposure or applied dose for each
environmental media or food are based on Methods for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions (U.S.
EPA, 1998) also known as the MPE methodology and formerly referred to as the
Indirect Exposure Methodology (IEM). The MPE methodology represents the
state of the science with respect to providing reliable guidance in the proper
conduct of assessments of risks that may result from multimedia, multipathway
exposures. EPA's National Center for Environmental Assessment (NCEA)
prepared the MPE methodology as an update to EPA's 1990 IEM document.
Most of the revisions in the MPE methodology are based on Scientific Advisory
Board (SAB) and public comments on IEM. Earlier versions of this document
have undergone internal EPA and external peer review. The breast milk exposure
pathway algorithms and data were based on MPE (U.S. EPA, 1998) and the
Dioxin Reassessment (U.S. EPA, 2000).
13-10
-------
Section 13.0
Human Exposure Module
¦ Accounts for exposures attributable to a specific microenvironment.
Inhalation exposure to volatile chemicals from contaminated ground water during
showering is a specific pathway of concern. Therefore, a specific model that
includes volatilization of contaminants during showering and the build-up of a
contaminant's concentration in the bathroom during and after a showering activity
was incorporated into the Human Exposure Module. The shower and bathroom
algorithms were adapted from peer-reviewed sources (McKone, 1987 and Little,
1992).
¦ Accounts for spatial variability in exposure across an AOL The Human
Exposure Module carries forward all spatial information used in the 3MRA
modeling system. This provides for the analysis of exposure variability across the
AOI. By using census block centroids as the point of analysis, the module
accounts for the variability in population across the site; in addition, the number
of specific locations across the AOI can range from a few dozen to several
hundred.
¦ Accounts for different behaviors that may lead to increases in exposure.
Different behavior such as fishing, farming, and gardening are used to identify
different groups within the population that may engage in certain activities that
can lead to higher exposures. For example, a certain percentage of the population
gardens. Most gardeners eat most of the produce that they grow. If these crops
are contaminated then the gardening activity increases their exposures over the
exposures of their neighbors who might not garden.
¦ Accounts for variability in exposure due to differences in age. Within a
specific receptor type, the calculated applied dose varies due to the differences in
the exposure factors used for the different age cohorts.
¦ Accounts for inter-individual variability within a receptor-type/age cohort.
Statistical distributions were developed for intake rates and body weights for each
age cohort within each receptor type. Use of these distributions within the Monte
Carlo framework provides a way of assessing the interindividual variability due to
exposure factor variability.
13.3.2 Uncertainty and Limitations
Uncertainties and limitations associated with the Human Exposure Module include the
following:
¦ Human receptors are assumed to be stationary. It is assumed in characterizing
exposure that human receptors both reside and work at the receptor location
identified for them during site characterization (i.e., the farm area for farmers or
residential exposure area for nonfarmers). The point of exposure is, in general,
the Census block centroid for a resident or home gardener and the centroid of a
farm for farmers. This assumption may result in either an overestimate or
underestimate of exposure, because individuals may reside at the identified
13-11
-------
Section 13.0
Human Exposure Module
location within the study area but commute to work areas outside of the study
area, or commute to more or less contaminated areas within the study area.
¦ All residential receptors are assumed to be located at the Census block
centroid. The spatial point estimates of average daily doses for residential
receptors assume all receptors in a Census block are located at the centroid of the
Census block. Thus, to the extent that some receptors in fact reside or spend
appreciable time at more highly contaminated areas within the block, their
average daily doses will be underestimated. The converse is also true; that is, the
average daily dose at the centroid will overestimate dose in other areas where
concentrations are lower. If contaminant concentrations decrease approximately
linearly with distance away from the waste management unit (WMU), there is
probably little, if any, bias introduced by the centroid assumption. If
concentrations decrease nonlinearly (e.g., first-order air deposition), the centroid
assumption may overestimate the true average daily dose across the Census block
to an unknown extent. The effect of any such bias will also be influenced by the
size of the Census block—relatively larger blocks have greater potential for a
bias.
¦ Only one farm per Census block group reporting farms is modeled. The
number of farms modeled in the study AOI (which affects the farm-receptor
population possibly at risk) is the number of Census block groups that make up
the AOI that also contain farms. For example, if the AOI includes two Census
block groups, and both Census block groups contain farming land use, then two
farms will be modeled within the AOI. This could introduce a risk-conservative
bias if the AOI does not in fact contain a farm.
¦ Estimated exposures due to fish ingestion are subject to random sampling
error for both farm and residential receptors. Residential and farming fishers
are assumed to be mobile and catch fish from up to three randomly selected
fishable reaches throughout the AOI. These selected reaches may or may not
reflect actual preferred fishing locations in the AOI. There is no reason to expect
any systematic bias in estimated fish ingestion exposure.
¦ Incremental exposures are estimated. No provision is made for considering
background exposures for the purpose of generating aggregate or total risk,
hazard quotient, or margin of exposure (MOE) estimates for modeled receptors.
¦ No reduction of contaminant concentration in food items is assumed to occur
through food preparation. Washing of fruits and vegetables, and cooking of
produce, beef, and milk may reduce contaminant concentrations in foods. This is
not accounted for in the model.
13.4 References
Little, J.C. 1992. Applying the two-resistance theory to contaminant volatilization in showers.
Environmental Science and Technology. 26(7): 1341-1349.
13-12
-------
Section 13.0
Human Exposure Module
McKone, T.E. 1987. Human Exposure to Volatile Organic Compounds in Household Tap
Water; the Indoor Inhalation Pathway. Environ. Sci. Technol. 21(12): 1194 - 1201.
U.S. EPA (Environmental Protection Agency). 1997. Exposure Factors Handbook. Office of
Research and Development, U.S. Environmental Protection Agency, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1998. Methodology for Assessing Health Risks
Associated with Multiple Pathways of Exposure to Combustor Emissions. 600/R-98/137.
National Center for Environmental Assessment, Cincinnati, OH. ES-l-ES-20.
December.
U.S. EPA (Environmental Protection Agency). 2000 Background Document for the Human
Exposure and Human Risk Modules for the Multimedia, Multipathway, Multirecptor Risk
Assessment (3MRA) Model. Office of Solid Waste, Washington, DC. August.
U.S. EPA (Environmental Protection Agency). 2000. Exposure and Human Health
Reassessment of 2,3,7,8-Tetrachlordiobenzo-p-Dioxin (TCDD) and Related Compounds.
EPA/600/P-00/001Bg. National Center for Environmental Assessment, Office of
Research and Development, Washington, DC. September.
13-13
-------
Section 13.0
13-14
Human Exposure Module
-------
Section 14.0
Human Risk Module
14.0 Human Risk Module
14.1 Purpose and Scope
The Human Risk Module calculates risk measures for a given constituent, waste
concentration, and site. Detailed information on the Human Risk Module can be found in the
background document (U.S. EPA, 2000). Figure 14-1 shows the relationship and information
flow between the Human Risk Module and the 3MRA modeling system. The Human Risk
Module uses the annual average daily doses calculated in the Human Exposure Module to
calculate receptor risk statistics. These risk statistics are used by the Exit Level Processors to
determine national-level risk distributions.
Key Data Inputs
• Health benchmarks
• Population demographics
• Receptor locations
Doses
Risks
HQs
Human
Exposure
Module
Human
Risk
Module
Exit
Level
Processors I and
Figure 14-1. Information flow for the Human Risk Module
in the 3MRA modeling system.
For each constituent, waste concentration, and site, the Human Risk Module generates
risk estimates for each receptor location in the area of interest (AOI) and then calculates the
number of receptors that fall within a specified risk or hazard range to describe the distribution
of risks for the population at each site. The module also determines the timing of maximum
risks. The Human Risk Module has the following functions:
1. Calculates risk measures. The Human Risk Module calculates cancer risk,
noncancer hazard quotient (HQ), and noncancer margin of exposure (MOE) (for
breastfeeding infants only). Depending on the constituent, the Human Risk
Module calculates risk, HQ, or both. MOE is only calculated for breastfeeding
infants for dioxin-like chemicals. These calculated risk measures are specific to a
receptor type, an age cohort, an exposure pathway, a receptor location, and a
specific exposure period (identified by starting year). The Human Risk Module
14-1
-------
Section 14.0
Human Risk Module
also aggregates risks and HQs from individual exposure pathways (e.g., ground
water ingestion) to determine risk for groups of pathways (e.g., all ingestion
pathways).
2. Processes results for decision making. The Human Risk Module puts exposed
and unexposed population in the AOI into risk bins to estimate the number of
receptors that experience risk within a specified range. Each risk bin is a range of
risks or HQs. For any given exposure pathway and exposure period, the Human
Risk Module uses Census data on population for each receptor location to
determine the number of people of each receptor type and age cohort that
experience risk from the specified pathway in the specified exposure period at
risk levels within that bin. These populations are summed across receptor
locations that have risks within the same bin. For each exposure pathway or
pathway group, the Human Risk Module estimates total risk by multiplying the
population at a location by the risk for that location, and uses this to determine the
exposure period for which the total risk or HQ across all receptor types and age
cohorts is the greatest. This estimate of total risk is not intended as a final risk
measure, but is used only to identify the timing of maximum risk. The exposure
period is identified by the year in which the risk averaged over a specified
exposure period starts.
The scope of the Human Risk Module includes nine exposure pathways, four exposure
pathway groupings, four receptor types, and five age cohorts, as follows:
Exposure pathways:
¦ Air inhalation,
¦ Shower inhalation,
¦ Ground water ingestion,
¦ Soil ingestion,
¦ Fruit and vegetable ingestion,
¦ Beef ingestion,
¦ Dairy ingestion,
¦ Fish ingestion, and
¦ Breast milk ingestion (infants only).
Exposure pathway groupings:
¦ All inhalation pathways,
¦ All ingestion pathways,
¦ All ground water pathways (ground water ingestion and shower
inhalation), and
¦ All ingestion and inhalation pathways combined (if appropriate for the
constituent).
14-2
-------
Section 14.0
Human Risk Module
Receptor types:
¦ Residents,
¦ Gardeners,
¦ Farmers, and
¦ Fishers.
Age cohorts:
¦ Infants under 1 year (breastfeeding pathway only)
¦ Children aged 1 to 5 years,
¦ Children aged 6 to 11 years,
¦ Children aged 12 to 19 years, and
¦ Adults.
The module uses the above groupings in calculations, but risk results are output only for the
different exposure pathways and pathway groupings; these results are summarized across
receptor types and age cohorts.
14.2 Conceptual Approach
The Human Risk Module calculates risk or HQ by pathway, receptor type, and age
cohort; aggregates pathway risks across pathway groupings; bins population into risk or HQ
bins; estimates total population risk; and determines the timing of maximum risks. The Human
Risk Module can calculate risks and HQs for the whole AOI or within radial distance rings
within the AOI if they are defined. These functions are described in the following subsections.
14.2.1 Calculate Risk Measures
The first major function of the Human Risk Module is to calculate cancer risk and/or HQ
for a given receptor type, age cohort, exposure pathway, receptor location, and exposure period.
Cancer Risk Calculations. The governing equation used to calculate increased
incremental cancer risk over a lifetime attributable to a lifetime exposure to a contaminant at a
given dose (exposure) is
Risk = CSFxDose (14-1)
where
Risk = lifetime risk (probability units)
CSF = contaminant-specific cancer slope factor (mg/kg-d)"1
Dose = annual average daily contaminant dose (mg/kg-d), expressed as an average
daily dose over a lifetime.
This equation reflects a daily, lifetime exposure. However, the exposure duration may
vary, as may the number of days per year a receptor is actually exposed (exposure frequency).
14-3
-------
Section 14.0
Human Risk Module
Assuming that cancer risk is linearly related to exposure duration and exposure frequency,
Equation 14-1 can be modified to account for this as follows:
Risk = BxpDur^ x ExpFreq x csp x ^
LifeTime 365
where
ExpDur
Lifetime
ExpFreq
CSF
ADD
duration of exposure (yr)
average lifetime (yr)
exposure frequency (d/yr)
cancer slope factor (mg/kg-d)"1
average daily dose over specified exposure period (mg/kg-d).
Exposure duration and exposure frequency may be set to any reasonable value so long as
exposure duration is less than or equal to the lifetime value chosen and exposure frequency is
less than or equal to the maximum of 365 d/yr. The exposure duration is currently set to 9 years,
which corresponds to the recommended value for median residence time presented in the
Exposure Factors Handbook (U.S. EPA, 1997a,b,c), and to set exposure frequency to 350 d/yr.
The ADD is calculated as follows:
ExpDur
Z Dose<
ADD = — (14-3)
N v ;
where
ADD = average daily dose over entire exposure period (mg/kg-d).
t = year
ExpDur = exposure duration (yr)
Dose, = annual average daily dose in year t (mg/kg-d)
N = number of years in the exposure duration (unitless).
In addition to the straight averaging of dose over the exposure period, a child receptor
will age out of his or her initial cohort as the exposure duration progresses. For example, a child
beginning the exposure duration at age 3 will, over a 9-year exposure period, increase in age to
11 years old. The annual doses are based on exposure factors (such as body weight or intake
rate) that are age-cohort-specific. Therefore, the Human Risk Module ages each child receptor
through the appropriate age cohorts as exposure progresses. Child receptors for a specific
starting cohort are assumed to begin exposure at the midpoint of the age cohort (e.g., 3 years for
a child aged 1 to 5 years). The age of a child receptor is monitored at each year over the
exposure duration, and the dose is set in accordance with the relevant cohort as the child ages
through them.
14-4
-------
Section 14.0
Human Risk Module
The Human Risk Module calculates a time series of risks averaged over the specified
exposure duration, with each average risk representing a different starting year within the period
of calculation. Calculations continue until the exposure medium concentration drops to less than
1 percent of the peak concentration, up to a maximum of 10,000 years. If the exposure duration
were 9 years and the concentration dropped to 1 percent in year 100, the Human Risk Module
would calculate a time series of 92 annual average risks, for years 1 through 9, 2 through 10, and
so forth, up to years 92 through 100. These averages are identified by starting year (so in this
example, year 1, year 2, and so forth, through year 92), but the risk for "year 1" is the risk for the
9-year period starting in year 1.
The Human Risk Module calculates cancer risks for any given contaminant only if an
appropriate health benchmark is available. For ingestion pathways, the calculations are
performed if an oral cancer slope factor (CSF) is available; for inhalation pathways, the
calculations are performed if an inhalation CSF is available. If neither an inhalation nor an
ingestion CSF is available, then no cancer risks are calculated for the contaminant.
Noncancer HQ Calculations. The governing equation used to calculate noncancer HQs
depends on the route of exposure. For ingestion exposures, the equation is
TT„ ADD
HQ - W (14-4)
where
HQ = hazard quotient for exposure period (unitless)
ADD = average daily dose over exposure period (mg/kg-d)
RfD = reference dose (mg/kg-d).
The ADD is calculated as shown in Equation 14-3 for carcinogens.
For inhalation exposures, the equation is
HQ - w <14-5)
where
HQ = hazard quotient over exposure period (unitless)
Cavg = average air concentration over exposure period (mg/m3)
RfC = reference air concentration (mg/m3).
The average air concentration may be an ambient air concentration or a shower/bathroom
air concentration, depending on the pathway, and is calculated as
14-5
-------
Section 14.0
Human Risk Module
ExpDur- lxC
C = — - (14-6)
avg ExpDur
where
CaVg = average air concentration over exposure period (mg/m )
t = first year of exposure period
ExpDur = exposure duration (yr)
Cone, = annual average air concentration in year t (mg/m3).
HQs are calculated for a specified exposure duration. They do not vary with exposure
frequency, but are based on the assumption that the receptor has regular, chronic exposure.
Exposure duration may be set to any reasonable value less than or equal to a typical lifetime.
The exposure duration is currently set to 9 years, which corresponds to the recommended value
for median residence time presented in the Exposure Factors Handbook (U.S. EPA, 1997a,b,c).
As for risk, the Human Risk Module calculates a time series of HQs averaged over the
specified exposure duration, with each HQ representing a different starting year within the
period of calculation. Calculations continue until the exposure medium concentration drops to
less than 1 percent of the original WMU concentration, up to a maximum of 10,000 years.
The Human Risk Module calculates noncancer HQs for any given contaminant only if an
appropriate health benchmark is available. For ingestion pathways, the calculations are
performed if an oral reference dose (RfD) is available; for inhalation pathways, the calculations
are performed if an inhalation reference concentration (RfC) is available. If neither an RfD nor
an RfC is available, then no noncancer HQs are calculated for the contaminant.
Noncancer MOE Calculations. For infant exposure to breast milk, effects are typically
quantified as an MOE rather than a risk or HQ. The MOE for infants is calculated as
BM
where
Dose
MOE = (14-7)
MOE = margin of exposure (unitless)
Dose = annual average applied dose from breast milk ingestion (mg/kg-d)
BM = contaminant-specific benchmark for breast milk exposure based on
background exposure levels (mg/kg-d)
This calculation is currently performed only for 2,3,7,8-TCDD for infant exposures through the
breast milk pathway. MOE estimates are single-year averages based on 1-year average doses.
The Human Risk Module calculates an average MOE for each year in the calculation period.
14-6
-------
Section 14.0
Human Risk Module
14.2.2 Process Results for Decision Making
The Human Risk Module generates risk estimates for a variety of receptor types, age
cohorts, exposure pathways, and receptor locations for each year modeled. The module
processes and aggregates these results so that decision makers can answer a variety of questions
about potential human risks. Specifically, the Human Risk Module (1) aggregates risk measures
across pathways, (2) places population into risk bins, and (3) determines the timing of the
maximum risk or hazard.
Aggregate Risk across Pathways. The 3MRA modeling system calculates risks and
HQs for individual pathways. However, many receptors are exposed via multiple pathways, so
the 3MRA modeling system also aggregates risk across pathways, where appropriate. The
Human Risk Module performs the following aggregations:
¦ Aggregation across ingestion pathways. The Human Risk Module sums risks
and HQs across the ingestion pathways for ground water, soil, fruits and
vegetables, beef, milk, and fish.
¦ Aggregation across inhalation pathways. The Human Risk Module sums risks
and HQs across both inhalation pathways: ambient air and shower.
¦ Aggregation across all pathways. When health effects can be combined across
routes of exposure, the Human Risk Module aggregates risks for all pathways
(both ingestion and inhalation).
¦ Aggregation across ground water pathways. When health effects can be
combined across routes of exposure, the Human Risk Module aggregates risks for
both ground water pathways: ground water ingestion and shower inhalation.
Whether or not it is appropriate to aggregate risks or HQs across routes of exposure
depends on the contaminant and the health effects it causes via different routes of exposure.
Cancer and noncancer effects (risks and HQs) are never combined.
Place Population into Risk Bins. Calculated risks or HQs by pathway are a measure of
possible individual risk. The second major function of the Human Risk Module is to calculate
population risk measures. For each pathway or aggregation of pathways, the Human Risk
Module calculates the number of people in each receptor group and age cohort that experience
various risk or HQ levels across the AOI or in a distance ring. These population counts are the
basis of a series of cumulative frequency histograms, each specific to a receptor type, age cohort,
exposure pathway (or aggregation of pathways), and exposure period. Thus, the full collection
of cumulative histograms is a time series of conditional histograms for risk and/or HQ.
Although the time series of histograms appears to be annual, the histograms are actually based
on a multiyear moving average for risks and HQs, as discussed in the section on calculating risk.
The length of the averaging time depends on the specified exposure duration.
The 3MRA modeling system currently uses 1990 Census data, 1992 agricultural census
data, and 1996 fishing survey data to estimate receptor/age-cohort-specific population estimates
14-7
-------
Section 14.0
Human Risk Module
at each residential or farm location for each site, as described in Volume II of this report. These
populations are associated with census blocks. For the purpose of calculating and assigning
carcinogenic risk and noncarcinogenic hazard, all of the population in a Census block is assumed
to be located at the centroid of the block.
The 3MRA modeling system has seven bins for risk and four bins for HQ and MOE. The
bins for risk are
¦
Bin 1:
0 to 5 x 10"9
¦
Bin 2:
>5 x 10"9 to 7.5 x
10"8
¦
Bin 3:
>7.5 x 10"8 to 7.5
x 10"7
¦
Bin 4:
>7.5 x 10"7 to 2.5
x 10"(
¦
Bin 5:
>2.5 x 10"6 to 7.5
x 10"6
¦
Bin 6:
>7.5 x 10"6 to 5 x
10"5
¦
Bin 7:
>5 x 10"5.
The
HQ/MOE
bins are
¦
Bin 1:
0 to 0.05
¦
Bin 2:
> 0.05 to 0.5
¦
Bin 3:
> 0.5 to 5.0
¦
Bin 4:
> 5.0.
All bin ranges are inclusive of the upper bound.
For any exposure period, and given a receptor type, age cohort, and pathway (or pathway
aggregation), a cumulative risk histogram is generated that contains the total population
corresponding to the receptor/cohort combination, across all receptor locations in the AOI that
experiences risks falling within each risk bin range. An HQ cumulative histogram is constructed
in a similar fashion.
As an example, assume 150 adult residents reside collectively at all residential receptor
locations within an AOI. Assume that for a given contaminant and exposure pathway for a
specified exposure period starting in year 1, they fall into the bins as shown in Table 14-1. The
conditional, cumulative population for these two example exposure periods would be as shown
in Table 14-2.
Table 14-1. Example HQ Counts for Hypothetical Sites
I'opuhilion in K.icli HQ U;in<>c
Bin 1
Bin 2
liin 3
liin 4
Kxposuiv sliti liii" in...
<0.05
0.05 to 0.5
0.5 l<> 5
>5
Year 1
30
45
75
0
Year 2
15
45
60
30
Year 3
6
24
60
60
14-8
-------
Section 14.0
Human Risk Module
Table 14-2. Example Cumulative Frequency at Hypothetical Site
Percent of Population in (his IIQ k;in<>c or
Lower
Lxposiuv in...
liin 1
Bin 2
liin 3
liin 4
Year 1
20
50
100
Year 2
10
40
80
100
Year 3
4
20
60
100
Determine Timing of Maximum Risk. The third major function of the Human Risk
Module is to determine the timing of maximum risk and/or HQ across all receptor/cohort
individuals for a given exposure pathway or aggregation of pathways. The time series of risk
histograms is analyzed to determine that year in which the maximum total risk and/or HQ over
time occurs.
Specifically, the Human Risk Module first estimates the total risk/HQ for all individuals
at a site by multiplying the population at each receptor location by the calculated risk or HQ at
that location. This is then summed across all locations, receptor types, and age cohorts. The
Human Risk Module calculates this total risk for each exposure period in the time series. The
module then determines the exposure period for which this total risk is the highest.
14.3 Module Discussion
14.3.1 Strengths and Advantages
The major strengths and advantages of this module include the following:
¦ Provides coverage for key receptor populations and exposure pathways. The
Human Risk Module supports risk characterization for four types of resident
receptors and four types of farmer receptors. Together, this set of receptor
populations includes the majority of those typically considered in evaluating
multipathway exposure and risk. Further, in modeling each of these receptor
populations, the 3MRA modeling system provides coverage for key exposure
pathways related to receptor behavior (e.g., ambient air inhalation, crop
ingestion). To enhance the representativeness for risk estimates, the modeling of
specific exposure pathways is linked to known receptor activity characterized
using demographic (or other relevant) data. The 3MRA modeling system also
includes five age cohorts in modeling risk for each receptor population to reflect
age-specific differences in exposure and risk. Specifically, cohort aging is
considered in modeling risk such that the ingestion or inhalation rates used for a
given receptor are adjusted as that receptor "ages" into the next age cohort to
reflect the exposure parameters relevant for that older age group.
14-9
-------
Section 14.0
Human Risk Module
¦ GIS-based representative modeling of population-level risk. The Human Risk
Module uses a GIS-based framework to support spatially representative modeling
of risk for modeled receptors. The use of U.S. Census, Census of Agriculture and
GIRAS land use data to place residents, drinking water wells, farmers, farms, and
recreational fishers for modeling risk supports a more representative analysis of
population-level exposure which explicitly considers the spatial distribution of
these receptors across AOIs.
¦ Generation of time series of risk profiles for modeled receptor populations
provides coverage for contaminants with different temporal exposure
profiles. The Human Risk Module generates a time series of risk estimates for
each combination of receptor/pathway/location/simulation year/chemical-
endpoint that allows the temporal profile of risk over longer simulation periods
(up to 10,000 years) to be evaluated. Providing detailed tracking of temporal risk
profiles over longer modeling periods can be important in assessing multipathway
risk for pathways involving contaminants with widely varying fate and transport
profiles (e.g., an inhalation toxicant that produces risk shortly after WMU release,
versus a contaminant with low mobility that can take many years to reach an off-
site well).
¦ Use of cumulative risk histograms provides a ready means for identifying the
year of maximum risk given widely varying risk profiles for different
receptor/pathway/contaminant combinations. The Human Risk Module
produces cumulative risk histograms to characterize the distribution of risk
across specific receptor populations for a given exposure pathway/simulation
year/contaminant combination. These cumulative risk histograms can be used as
the basis for identifying the simulation year which has the maximum cumulative
risk for a given regulatory percentile of the population (e.g., the maximum
cumulative risk for the 95th percentile of beef farmers resulting from beef
ingestion). The 3MRA modeling system includes an automated procedure for
querying the entire time series of cumulative risk distributions to identify this
maximum risk year and then outputs the risk distributions for all exposure
pathways modeled for that receptor population for that year. Not only does this
procedure provide a ready means to identify the significant maximum risk year
and extract the full risk distributions generated for that year, this approach also
represents an effective means of data reduction, which is a key issue given the
large number of simulation years and receptor/pathway /location/contaminant
combinations that can be modeled. Cumulative risk distributions can be generated
for user-specified distance rings within the AOI. This provides a distance-
differentiated set of risk metrics that can be used to support decision making.
14.3.2 Uncertainty and Limitations
The following limitations or uncertainties are inherent in the Human Risk Module:
¦ Risk/HQ/MOE estimates are aggregated for certain receptor types. The four
receptor types considered by the Human Risk Module (resident, residential
14-10
-------
Section 14.0
Human Risk Module
gardener, farmer, fisher) are fewer than the number of receptor types simulated by
the Human Exposure Module in order to maintain output storage requirements at
reasonable levels. The Human Risk Module internally aggregates dairy farmers
and beef farmers into a single farmer (dairy and beef) and aggregates all of the
Human Exposure Module's receptor type-specific fishers (e.g., resident fisher)
into a single fisher receptor. Some resolution is lost by this aggregation; for
example, the risks specific to farmers who drink contaminated milk, but do not
consume contaminated beef, would not be available.
¦ Synergistic or antagonistic effects among multiple contaminants or
individual contaminant species on risk/HQ/MOE are not considered. The
Human Risk Module considers only one contaminant at a time, and the risk
associated with that contaminant is implicitly considered to be independent of the
risks posed by other contaminants.
¦ Cancer slope factors do not vary with cohort age. Age-specific differences in
exposure responses are not considered.
¦ The fraction of residents in a Census block that ingest ground water is
assumed to equal the fraction of residents in the Census block group that
have ground water wells. The Census data report the number of households
within a Census block group that are served by ground water wells. However,
this information is available only at the Census block group level, and it is not
possible to determine from Census data alone whether individual Census blocks
within a block group with wells have wells or not. The Human Risk Module
calculates at the Census block level when considering residential exposure areas.
Consequently, the actual fraction of residents on wells for any individual
residential exposure area is uncertain. The assumption is made that the fraction of
residents in an exposure area (a Census block) consuming ground water is equal
to the fraction of the population in the Census block group that are served by
ground water wells. That is not to say that all well water is contaminated. Only
those wells lying within the ground water plume from the WMU source are
potentially contaminated. To the extent that the fraction of residents on wells
differs among Census blocks within the plume and outside the plume, population
risks may be over- or underestimated.
14.4 References
U.S. EPA (Environmental Protection Agency). 1997a. Exposure Factors Handbook. Volume I-
General Factors. EPA/600/P-95/002Fa. Office of Research and Development,
Washington, DC. Website at http://www.epa.gov/mcea/exposfac.htm. August.
U.S. EPA (Environmental Protection Agency). 1997b. Exposure Factors Handbook.
Volume II-Food Ingestion Factors. EPA/600/P-95/002Fa. Office of Research and
Development, Washington, DC. Website at http://www.epa.gov/ncea/exposfac.htm.
August.
14-11
-------
Section 14.0
Human Risk Module
U.S. EPA (Environmental Protection Agency). 1997c. Exposure Factors Handbook.
Volume Ill-Activity Factors. EPA/600/P-95/002Fa. Office of Research and
Development, Washington, DC. Website at http://www.epa.gov/ncea/exposfac.htm.
August.
U.S. EPA (Environmental Protection Agency). 2000. Background Document for the Human
Exposure and Human Risk Modules for the Multimedia, Multipathway, Multireceptor
Risk Assessment (3MRA) Model for HWIR99. Office of Solid Waste, Washington, DC.
August.
14-12
-------
Section 15.0
Ecological Exposure Module
15.0 Ecological Exposure Module
15.1 Purpose and Scope
The Ecological Exposure Module calculates the applied contaminant dose (in mg/kg-d)
to ecological receptors that may be exposed via ingestion of contaminated plants, prey, and
media (i.e., soil, sediment, and surface water). The Ecological Exposure Module uses various
input contaminant concentrations from the Surface Impoundment, Surface Water, Terrestrial
Food Web, and Aquatic Food Web Modules. Detailed information on the Ecological Exposure
Module can be found in the background document (U.S. EPA, 2000). Figure 15-1 shows the
relationship and information flow between the Ecological Exposure Module and the 3MRA
modeling system.
Key Data Inputs
• Food consumption rate
• Body weight
• Dietary preferences
SI Concentrations
Water Column and
Sediment Concentrations
Applied
Doses
Soil, Plant, and Prey
Concentrations
Aquatic Plant and
Prey Concentrations
Ecological
Risk Module
Aquatic Food
Web Module
Surface
Water
Module
Terrestrial
Food Web
Module
Ecological
Exposure
Module
Surface
Impoundment
Module
Figure 15-1. Information flow for the Ecological Exposure Module
in the 3MRA modeling system.
15-1
-------
Section 15.0
Ecological Exposure Module
The methodology and equations used to calculate the applied dose to mammals and birds
assigned to habitats within the AOI are consistent with the principles and guidelines described in
the Guidelines for Ecological Risk Assessment (U.S. EPA, 1998). The basic forms of these
equations have been used by OSW and other EPA programs to predict applied doses in a variety
of ecological risk analyses, and they are similar to the exposure equations recommended by other
non-EPA risk assessors (see, for example, the Methods and Tools for Estimation of the Exposure
of Terrestrial Wildlife to Contaminants, Sample et al., 1997). The Ecological Exposure Module
performs the following three functions:
1. Constructs a dietary matrix for each receptor for each habitat in the AOL
The Ecological Exposure Module creates a diet for each ecological receptor based
on dietary preferences.
2. Calculates applied doses for animals in terrestrial habitats. Using the dietary
matrix and the media,1 plant, and prey concentrations calculated by the Terrestrial
Food Web Module, the Ecological Exposure Module calculates applied doses for
each avian and mammalian receptor species assigned to terrestrial habitats in the
AOI.
3. Calculates applied doses for animals in margin habitats (wetland or
waterbody). Using the dietary matrix and the media,1 plant, and prey
concentrations calculated by the Aquatic Food Web and Terrestrial Food Web
Modules, the Ecological Exposure Module calculates applied doses for each avian
and mammalian receptor species assigned to margin habitats in the AOI.
15.2 Conceptual Approach
The conceptual approach to predicting ecological exposures in the 3MRA modeling
system addresses several major sources of variability in ecological exposures, such as
environmental characteristics of different ecosystems, spatial resolution of contaminant
concentrations, dietary composition, and receptor-specific exposure factors. Specifically, the
approach addresses variability through (1) the development of representative habitats;
(2) selection of receptors based on ecological region; (3) the recognition of opportunistic feeding
and foraging behavior using probabilistic methods; (4) the creation of a dietary scheme specific
to region, habitat, and receptor; and (5) the application of appropriate graphical tools to capture
spatial variability in exposure.
The Ecological Exposure Module simulates exposures for birds and mammals assigned to
11 types of habitats representing waterbody margin habitats for streams/rivers, lakes, and ponds;
terrestrial habitats; and wetland margin habitats. The 3MRA modeling system exposure factor
database contains data on 57 species, including mammals (23), birds (22), and selected
herpetofauna (12). This suite of receptors covers a wide range of feeding strategies from
obligate herbivores to opportunistic feeders. Because toxicological data were considered
1 Constituent concentrations in surface impoundments may also be used to calculate exposure if the
receptor's home range overlaps the impoundment.
15-2
-------
Section 15.0
Ecological Exposure Module
insufficient to develop ingestion benchmarks for herpetofauna for the contaminants included in
this version of the 3MRA modeling system, the Ecological Exposure Module does not predict
applied doses to these receptors. However, the module was developed to calculate exposures to
herpetofauna as toxicological data become available.
For each year in the simulation, the Ecological Exposure Module predicts the applied
dose to mammals and birds assigned to waterbody margin, wetland, and terrestrial habitats
delineated within the area of interest (AOI). The model is flexible enough to be applied to any
habitat, allows a unique home range to be defined for each receptor in the AOI, and constructs a
receptor-specific diet based on information in the ecological exposure factor database. Simple
food webs, constructed to represent each type of habitat, provide the framework for creating a
matrix of dietary preferences for each simulation. That is, the diet of avian and mammalian
receptors is constructed based on the food web specific to each habitat, and the dietary
preferences can be varied with each simulation. Data on dietary preferences are used to
determine which of the 17 possible categories of food are eaten by each receptor and how much
of each food category is eaten.
The ecological exposure framework is largely based on the desire to provide an
appropriate level of resolution for national-scale analyses. The Ecological Exposure Module
performs the necessary calculations to predict contaminant exposures for birds and mammals;
the underlying framework defines the spatial scale for the exposure modeling, as well as the
habitats and receptors that are assigned to the AOI.
15.2.1 Criteria for the Ecological Exposure Module
The Ecological Exposure Module was designed to satisfy three important criteria that
respond to reviews of the earlier efforts to conduct a national-level ecological risk assessment.
Those criteria are as follows:
¦ The module must capture the wide variability in ecological systems in a manner
that is appropriate given the availability of data to characterize and evaluate
ecological exposure and risk.
¦ The module must define spatial boundaries for ecological exposures at a scale that
takes full advantage of the spatial resolution offered by the 3MRA modeling
system and is meaningful with respect to predicting ecological risks.
¦ The module must allow for the site-based assignment of ecological receptors that
reflect the major trophic elements and feeding strategies relevant to exposure, as
well as regional characteristics that influence the composition of ecological
communities.
To satisfy these criteria, the Ecological Exposure Module contains the following three
critical elements:
¦ A representative habitat scheme,
¦ Habitat-specific food webs, and
15-3
-------
Section 15.0
Ecological Exposure Module
¦ Appropriate receptor species for each habitat and food web.
Within the context of these criteria, the Ecological Exposure Module selects the media,
plant, and prey contaminant concentrations to which each receptor is exposed and calculates the
applied dose to chemical stressors across space and time. The Ecological Exposure Module was
developed to allow for flexibility in calculating exposures in habitats, home ranges, and
receptors that can be defined at virtually any scale desired by the user. For instance, a home
range can be defined as a spatially unique area for each receptor, or many receptors can share the
same home range. Regardless of how the spatial extent of exposure is defined, the module will
calculate a "spatially consistent" applied dose, allowing each receptor to ingest plants, prey, or
environmental media within its home range. Similarly, the representative habitats can be
delineated for virtually an unlimited number of sizes and shapes provided that the geographical
coordinates can be identified. Although a receptor species can be assigned to multiple habitats
within an AO I, the Ecological Exposure Module calculates a time series of applied doses to each
receptor within each habitat once and only once. That is, a given receptor may not be assigned
to multiple home ranges within a habitat.
Representative Habitat Scheme. The representative habitats shown in Table 15-1 were
designed to be general enough to encompass a broad range of ecological systems in the
conterminous 48 states and can be used to define the spatial boundaries of exposure for any site
in the 3MRA modeling system. The representative habitats capture important characteristics of a
variety of environmental settings that determine what plants and animals are likely to be present
and what exposure pathways are likely to be of interest. The representative habitat scheme
supports a level of detail commensurate with the data and models available to predict exposures
to chemical stressors. For example, there are many important ecological distinctions between
coniferous and deciduous forests; however, the fate and transport models in the 3MRA modeling
Table 15-1. Representative Habitats
for 3MRA
Terrestrial Habitats
Grassland
Shrub/scrub
Forest
Crops/pasture
Residential
Waterbody Margin Habitats
River/stream
Lake
Pond
Wetland Margin Habitats
Permanently Flooded Grassland
Permanently Flooded Shrub/scrub
Permanently Flooded Forest
15-4
-------
Section 15.0
Ecological Exposure Module
system are not capable of producing forest-specific simulations, and data on potential forest
receptors are insufficient to generate a meaningful distinction in predicted exposures and risks to
receptors found in these two types of forest habitats. Consequently, the forest habitat represents
both deciduous and coniferous forests in the 3MRA modeling system.
The type of habitat determines which species are present at a site because the habitat
provides essential resources such as food; shelter; nesting sites or materials; and appropriate sites
for behaviors such as courtship, mating, roosting, or hibernation. The criteria used to develop
the representative habitats were based on a survey of existing ecological classifications and
reflect the importance of the physical setting in terms of essential resources (Bailey, 1996;
Bourgeron and Engelking, 1994; Cowardin et al., 1979; Davis and Simon, 1995; Demarchi,
1996; Drake and Faber-Langendoen, 1997; Federal Geographic Data Committee, 1997; Kiichler,
1964; Omernik, 1987; Sawyer and Keeler-Wolf, 1995; Shafale and Weakley, 1990; USDA
Forest Service, 1994; U.S. FWS, 1998; Viereck and Elbert, 1991; Weakley et al., 1998; and
Whitney, 1985). The following subsections provide a very brief summary of the criteria and
resulting habitats that were defined to represent the wide variability in ecological systems;
additional detail is provided in U.S. EPA (1999b).
Criteria for Terrestrial and Wetland Habitats. The primary criteria for defining
terrestrial and wetland habitats are soil moisture and vegetation structure. Soil moisture, or
degree of saturation, affects soil chemistry, general vegetation structure, and habitat suitability.
Soil moisture is differentiated based on the following three categories:
¦ Terrestrial—well-aerated, nonsaturated soils;
¦ Intermittently flooded—periodically saturated or inundated but aerated for some
periods during the growing season; and
¦ Permanently flooded—saturated or inundated throughout most years.
These characteristics are quite general and do not include many of the abiotic parameters
often associated with ecological classification systems, such as latitude, climate, topography,
elevation, or soil type. Nevertheless, vegetation type is directly affected by these abiotic
parameters and is, therefore, often used as a general indicator of many abiotic characteristics.
Vegetation structure refers to the stature, spacing, and relative stem size of the dominant
vegetation. It describes the primary producers and indicates the appropriateness of the habitat
for use by major trophic levels. Generally accepted categories of vegetation structure include
grasses, herbs, shrub/scrub, forest, and cropland. Each category has dominant vegetation with
distinct height and density that, in turn, supports a distinct suite of fauna.
In addition to the general classifications cited in U.S. EPA (1999b), primary sources for
defining wetland habitats include Christensen et al. (1988), Damman and French (1987), Glaser
(1987), Gosselink and Turner (1978), Kadlec and Knight (1996), Larsen (1982), Mitsch and
Gosselink (1993), Niering (1985), Norquist (1984), Sharitz and Gibbons (1982), Verry (1997),
Windell et al. (1986), and Winter (1989).
15-5
-------
Section 15.0
Ecological Exposure Module
Criteria for Freshwater Margin Habitats. Using waterbody margin habitats in the
exposure assessment allows for the inclusion of freshwater aquatic plants and prey that are
integral parts of the food web for terrestrial receptors in habitats that include fishable
waterbodies. A brief review of the literature supports the use of two criteria, energy/flow and
size, as a simple but effective classification approach. Although aquatic classifications consist of
more complex divisions, they fundamentally include these two criteria. The energy/flow
criterion distinguishes between still water and flowing water, and the size criterion addresses the
intrinsic differences between large and small systems, such as net primary production, diversity
of habitat, and length and complexity of food webs. No commonly used size cutoffs were found
in the literature. For lakes versus ponds, the waterbody margin habitats use EPA's
Environmental Monitoring and Assessment Program (EMAP) 10-hectare cutoff for small versus
large lakes; thus, surface waterbodies with a surface area greater than 10 hectares were classified
as lakes, and those below 10 hectares were classified as ponds. For streams versus rivers, the
generally accepted stream order concept was considered, and stream order 52 was initially
proposed as a cutoff between streams (small flowing waterbodies) and rivers (large flowing
waterbodies). Based on simple mass balance calculations of probable contaminant loadings to
surface waterbodies, it was readily apparent that the predicted contaminant concentrations in
streams larger than stream order 5 would effectively be diluted below detectable levels.
Therefore, streams and rivers of order 5 and below were represented as a single habitat, and
those of order 6 and above are not included in the ecological exposure assessment. References
consulted for the development of the waterbody margin habitats include the habitat
classifications cited in U.S. EPA (1999), as well as in Davis and Simon (1995), USDA (1998),
and Caduto (1990).
Habitat-Specific Food Webs. The representative habitats are used not only to define the
spatial boundaries for site-based exposures, but also to provide the ecological basis for
development of food webs that reflect important exposure pathways to chemical stressors. The
habitat-specific food webs were designed to be simple enough to allow for parameterization in
the model, but flexible enough to capture a full range of feeding strategies and exposure
pathways. The food webs were developed based on generally accepted concepts about food
webs and natural community dynamics (Anderson, 1997; Begon and Mortimer, 1981; Caduto,
1990; Davis and Simon, 1995; Kadlec and Knight, 1996; Sample et al., 1997; Schoener, 1989;
Schoenly and Cohen, 1991; Suter, 1993; Tanner, 1978; U.S. EPA, 1993, 1994). In describing
the predator-prey interactions in the representative habitats, a number of sources were consulted
for general habitat information on receptor species, including a wide variety of field guides,
nature guides, wildlife encyclopedias, and species-specific monographs that describe the habitats
known to be frequented or used by the species. All dietary items reported as commonly eaten
were considered in constructing the food webs and, of course, in developing the ecological
exposure factors described in Volume II on data collection.
Terrestrial Habitat Food Webs. As suggested in Figure 15-2, the receptors assigned to
the terrestrial food web include primary producers (vascular plants); soil biota; and birds;
mammals, and herpetofauna across three trophic levels. Trophic level 1 (TL1) consists of
species that consume only plants (i.e., the herbivores) and that are potential prey for higher
2 The Strahler (1957) stream ordering system is used throughout this document.
15-6
-------
Section 15.0
Apex Species
Ecological Exposure Module
T3
Omnivorous Mammals
• Black bear
• Red fox
• Coyote
Omnivorous Birds
• Burrowing owl
Carnivorous Birds
• Hawks
• American kestrel
• Loggerhead shrike
Carnivorous Mammals
• Weasels
• Kit fox
Medium Omnivores
• Racoon
Carnivorous
Reptiles
Snakes
Omnivorous
Small Mammals
Shrews
• Least weasel
Mice
Carnivorous
Amphibians
Newts
• Salamanders
Frogs
Omnivorous
Reptiles
Eastern box turtle
Insectivores
Bats
Birds
Omnivorous Birds
Passerines
Ground birds
Large Herbivores
Mule deer
White-tailed deer
Small Herbivores
Rabbits
Voles
Soil Community
Other
invertebrates
Flying
invertebrates
Worms
Vascu ar Plants -
Primary Producers
Movement through the food web
of primary producer biomass
Movement through the food web
of soil community biomass
Movement through the food web
ofT1 biomass
Movement through the food web
of T2 biomass
Figure 15-2. Simple terrestrial food web showing example receptors.
trophic level species. TL1 species include small and large mammals and invertebrates. The soil
community is a unique subset within TL1 and includes invertebrate soil organisms that live in
direct contact with soil, thus reflecting a unique exposure pathway. Within this conceptual
framework, the soil community is both a source of food for certain receptors and a receptor
group evaluated by the Ecological Risk Module. The dynamics within soil communities are, in
fact, very complex and include herbivores, omnivores, and carnivores at several trophic levels
within the soil community. However, modeling this complexity within the 3MRA modeling
system is well beyond the level of resolution that can be used in a national assessment strategy.
Trophic level 2 (TL2) includes species that consume plants and/or animals (omnivores
and carnivores) and are themselves eaten by larger predators. The TL2 species include a wide
array of small- to medium-sized mammals, birds, and herpetofauna. For example, the TL2
carnivores include species of reptiles and amphibians that eat soil invertebrates, as well as
specialized feeding guilds such as insectivores. The species included in TL2 represent several
faunal classes, functional groups, and size ranges. Opportunistic feeders in TL2, such as
raccoons, increase the complexity of the web by feeding on virtually any TL1 or TL2 prey.
Trophic level 3 (T3) consists of apex species, or those that do not have any predators (other than
humans) in the habitat. Apex species include several faunal classes of receptors, such as large
15-7
-------
Section 15.0
Ecological Exposure Module
mammals (black bears) and raptors (Cooper's hawks), and tend to be among the widest ranging
receptors in the habitat.
WaterbodyWetland Margin Habitat Food Webs. As suggested by Figure 15-3, the
margin habitats for surface waterbodies (e.g., streams, lakes) include elements of both aquatic
systems and terrestrial systems. As with the terrestrial habitats, the receptors assigned to the
aquatic and wetland food webs include primary producers (vascular aquatic plants and algae);
sediment and surface water biota; and birds, mammals, and herpetofauna across three trophic
levels. These receptors may be exposed through the ingestion of aquatic biota (e.g., fish,
benthic invertebrates, and aquatic plants) and surface water, as well as through the incidental
ingestion of sediment. In addition, receptors assigned to margin habitats may be exposed
through the ingestion of soil, terrestrial plants, and prey items described in the terrestrial habitat
food web section (e.g., raccoons). For receptors in these habitats that feed strictly on aquatic
food items (e.g., muskrats and mink), the food web is assumed to be relatively simple, consisting
of the four-compartment aquatic food web directly linked to the receptor. For example, the
osprey eats fish almost exclusively and, therefore, is an apex predator at the top of this simplified
aquatic food web.
Receptors Assigned
to Aquatic Habitats Biotic Compartments in Aquatic System
Piscivores
(T2 and T3)
• Bald eagle
• Osprey
• Green heron
• Kingfisher
• Otter
• Mink
<
~
Piscivorous
Fish
i
i
<
Omnivorous
Fish
Omnivores
(T2 and T3)
• Bear
• Raccoon
• Coyote
• Amphibians
• Turtles
¦ Wading birds
j
i i
i i
i
1
Herbivorous Fish
Sediment Community
Aquatic Invertebrates
<
Herbivores
(T1)
• Beaver
• Muskrat
• Deer
• Ducks
i
J
i
Primary Producers - Aquatic Plants
Algae
Figure 15-3. Simple margin food web showing both aquatic and terrestrial components.
For omnivores and more opportunistic species in the waterbody margin habitats, the food
web is essentially the same as that for strictly terrestrial species, with aquatic prey available as
additional potential food items. Many of these species (e.g., raccoon or black bear) are equally
successful whether or not aquatic prey are available; however, these species are opportunistic
and will take advantage of any prey that are readily available.
15-8
-------
Section 15.0
Ecological Exposure Module
The receptor assignments for wetland habitats raised some unique questions with respect
to flooding regimes and the implications on habitat classification and receptors. Habitat
information for wetland species rarely indicates the degree or frequency of flooding when
describing wetland habitats. Therefore, it was difficult to differentiate between species using
intermittently flooded as opposed to permanently flooded wetland habitats. In many cases, the
food or prey items attributed to a species were used as an indicator. For example, if a source
reported that a species fishes in wetland habitats, that species was assigned to the appropriate
permanently flooded wetland habitat, which, as defined in this framework, supports fish and
other aquatic life. Because of the predominant use in the source literature of terms such as
swamps and marshes, which imply a relatively long flood duration, fewer species were assigned
to the intermittently flooded wetland habitats. Indeed, intermittently flooded wetlands are
generally less discernible from surrounding uplands than are permanently flooded wetlands and,
thus, are reported less frequently in general wildlife habitat literature. These differences are
reflected in the smaller receptor groups associated with the intermittently flooded wetlands.
Intermittently flooded wetlands are not delineated as independent habitats in the 3MRA
modeling system. Consequently, receptors assigned to intermittently flooded wetlands are
simply added to the list of receptors assigned to the terrestrial habitat if the intermittently flooded
wetland is located within the spatial boundaries of the terrestrial habitat.
Ecological Receptors. Within the framework of habitats and food webs, receptor
species were selected using a weight-of-evidence approach intended to support analyses of
various habitats across the conterminous 48 states. Most importantly, receptor species were
chosen to reflect a full range of exposures and, as a group, represent all of the faunal classes,
trophic levels, and feeding strategies that are typical of terrestrial and aquatic margin habitats.
The simple food webs created for the terrestrial and aquatic margin habitats provided the context
for receptor selection and are used to define the relationships between predators and prey. The
receptors selected to populate the food webs may be both predator and prey in a given habitat.
For example, small omnivores, such as mice or shrews, may consume a variety of plants and
animal prey items in lower trophic levels. Given their position in the food web, mice or shrews
might also be eaten by apex predators, such as the coyote or red-tailed hawk. Key determinants
for receptor selection are summarized as follows:
¦ Geographic distribution. National applicability was achieved primarily by
selecting species that are widely distributed throughout the conterminous 48 states
and then adding species to cover as many ecological regions as possible. Section
level data from Bailey's ecoregions (Bailey, 1996) were digitized and included in
the 3MRA geographic information system (GIS). Only those species documented
to occur in the section where a site was located were included in the exposure
assessment.
¦ Availability of wildlife exposure factors. The majority of the receptors selected
for the representative national data set included species for which wildlife
exposure factors were readily available. The main sources for ecological
exposure factor data were the Wildlife Exposure Factors Handbook (U.S. EPA,
1993), Methods and Tools for Estimation of the Exposure of Terrestrial Wildlife
to Contaminants (Sample et al., 1997), and the U.S. Army Corps of Engineers'
Species Profile Series (various dates and authors, e.g., Lane and Mitchell, 1997).
15-9
-------
Section 15.0
Ecological Exposure Module
¦ Faunal class. Major faunal classes—mammals, birds, reptiles, and
amphibians—generally occur throughout all trophic levels. Faunal class
distinctions are useful in selecting receptors because species within similar groups
are known to respond similarly to environmental disturbance. Although the
3MRA modeling system contains redundancy within each faunal class, exposure
factors were not available for species in all classes at all trophic levels. As a
result, certain species are underrepresented.
¦ Functional niche. Receptors experience different exposures based on dietary
composition. To ensure that all potential exposure pathways were reflected in
each habitat, receptors were selected to cover the full spectrum of dietary
preferences and feeding strategies (e.g., herbivores, carnivores, omnivores, and
insectivores).
To summarize, receptor species were chosen to reflect a full range of exposures and, as a
group, represent all of the faunal classes, trophic levels, and feeding strategies that are typical of
terrestrial and aquatic margin habitats, respectively. Table 15-2 provides a complete list of
receptor species cross-referenced by representative habitat.
Table 15-2. Representative Habitats for Receptor Species
Ko|)ivsonliili\o lliihiliils
IVnvslriiil
Speck's ^ ^ -Z. C £
\\;i lei-hod \
Mnriiin
'C
7
"Bis
2 — £
Wolliind Msirgin
¦f. 5
—
~ S.
Ill
w J. '
-------
Section 15.0
Ecological Exposure Module
Table 15-2. Representative Habitats for Receptor Species (continued)
Ko|)ivsonliili\o lliihiliils
IVnvslri
ill
\\ ;ilc'il)(i(l>
M;i riiin
Wolliind Msirgin
Speck's
(il'iisslilllds
/
cn
7.
w
'7.
is
'C
7
Z,
Xj
A
¦f.
¦f.
W
A
/
cn
Xj
Cooper's hawk
~
~
~
Coyote
~
~
~
~
~
~
Deer mouse
~
~
~
~
Eastern newt
~
~
~
~
~
Eastern cottontail rabbit
~
~
~
~
~
Eastern box turtle
~
~
~
~
~
~
~
~
~
~
~
Flatwoods salamander
~
~
~
~
~
Gopher frog
~
~
~
~
Great blue heron
~
~
~
~
~
~
Great Basin pocket mouse
~
~
Green heron
~
~
~
~
~
~
Green frog
~
~
~
~
~
~
Herring gull
~
~
~
~
~
Kit fox
~
~
Least weasel
~
~
~
Lesser scaup
~
~
~
~
Little brown bat
~
~
~
~
Loggerhead shrike
~
~
~
~
Long-tailed weasel
~
~
~
Mallard
~
~
~
~
~
~
Marsh wren
~
~
Meadow vole
~
~
~
Mink
~
~
~
~
Mule deer
~
~
~
~
~
~
Muskrat
~
~
~
Northern water snake
~
~
~
~
~
~
Northern bobwhite
~
~
~
~
Osprey
~
~
~
~
~
(continued)
15-11
-------
Section 15.0
Ecological Exposure Module
Table 15-2. Representative Habitats for Receptor Species (continued)
Ko|)ivsonliili\o lliihiliils
IVnvslriiil
Speck's ^ ^ -Z. C £
\\ ;ilc'il)(i(l>
M;i riiin
'C
7
^ H =
2 — £
Wolliind Msirgin
¦f. 5
—
~ S.
Ill
w J. '
-------
Section 15.0
Ecological Exposure Module
Table 15-3. Categories of Dietary Items for Ecological
Exposure Assessment
('iHe«iorv Diol.irv Item
Terrestrial Prey
Earthworms
Other soil invertebrates
Small mammals
Small birds
Small herpetofauna
Medium omnivores
Herbivores
Aquatic Prey
Benthic filter feeders
TL3 fish
TL4 fish
Vegetation
Aquatic plants
Fruits, fruit/seeds (single item)
Fern(s), fungi, dicot & monocot shoots
Forbs, grasses, shrubs
Roots
Crops, corn
Seeds/nuts
Environmental Media
Soil
Sediment
Surface water
The Ecological Exposure Module rank orders categories of dietary items from most
preferred to least preferred (based the maximum preference values), and the categories are
sampled starting with the most preferred category first and continuing through the ranked
categories in order until the diet is complete. The sampling approach was designed to address
the wide variability in animal diets indicated by available data and, at the same time, observe
trends in dietary preferences indicated by those data. There is no requirement that a receptor
consume every food item in the list; however, the diet is constrained by two rules: (1) any food
item with a non-zero minimum preference value must be part of the diet,3 and (2) the summation
of dietary preferences must equal 100 percent of the entire diet.
3 In general, the three or four dietary items with the highest maximum values have non-zero minimum
values. As a result, the sampling routine usually completes 100 percent of the diet quickly.
15-13
-------
Section 15.0
Ecological Exposure Module
Table 15-4 provides an example of the construction of a dietary matrix for a raccoon
assigned to a stream margin habitat. The raccoon's dietary composition is constructed based on
the prey preference data on various plants, prey, and media shown, and includes both terrestrial
and aquatic food items. Based on the preference data, the Ecological Exposure Module might
construct the dietary matrix shown in the Example Dietary Matrix column of Table 15-4.
Table 15-4. Example of Dietary Preferences for Raccoon
Minimum
Miiximiim
Kxnmplc Didiirv
l-'ooil Item
I'releiviice Viiluc
I'lVfciVllCC \ illllC
Matrix
( umuhilixc Did
(% of dicl)
(% of did)
(% of did)
(% of did)
Soil invertebrates
0
90
45
45
Fruits
25
86
26
71
Forbs
10
66
12
83
Small mammals
0
35
5
88
Benthic organisms
0
25
7
95
TL3 fish
0
23
5
100
TL4 fish
0
23
0
Small birds
0
19
0
Earthworms
0
10
0
Grain
0
10
0
Roots
0
10
0
Silage
0
10
0
Information on receptor species' dietary composition comes from a wide range of data
sources and is of two general types. Some data consist of reported quantities of certain items
eaten by particular individuals in a localized or site-specific study. These data consist of
measured stomach contents, nest or burrow contents, or counts of items observed to be eaten
during a particular time span. The principal sources for these data are the Exposure Factors
Handbook (U.S. EPA, 1993) and Sample et al. (1997). The second type of information consists
of qualitative reports of items documented to be eaten. These reports reflect a compilation of
observations and measurements for the species in general, and are reported as descriptions of the
species' potential diet. The principal sources of this type of data are the Army Corps of
Engineers' Species Profile Series, the American Society of Mammalogists' Mammalian Species
Series, and various field guides and handbooks (e.g., Willner et al., 1980).
When only qualitative data were available, the assignment of estimated dietary fractions
was based primarily on a set of decision rules implemented by a senior ecologist to maintain
consistency in interpreting qualitative descriptions. References to a diet item that implied a
single most significant component, such as "primary food source," "bulk of the diet," or
"consumes mostly," were assigned a minimum of 50 percent dietary composition. Items that are
of secondary importance but that would always make up at least some portion of the diet were
15-14
-------
Section 15.0
Ecological Exposure Module
assigned a minimum of 10 percent. Descriptions implying occasional sources of food were
given a minimum of zero and a maximum of 10 percent. A few species have relatively limited
diets and eat only one dietary item. For example, the cerulean warbler eats insects almost
exclusively. In these cases, the exclusive diet item (insects) was assigned a minimum of
95 percent and a maximum of 100 percent. The quantified dietary profiles were entered into the
exposure factor database as maximum and minimum preference values for each prey category.
15.2.3 Calculate Applied Dose for Receptors in Terrestrial Habitats
The applied dose for mammals and birds assigned to terrestrial habitats is calculated as a
function of the contaminant concentrations in soil, drinking water (e.g., streams or surface
impoundments), and contaminated plants and prey. The calculation is a summation of time-
dependent exposures to contaminated media, plants, and prey. The following example presents
the methods and equations for estimating the applied dose to the short-tailed weasel in a forest
habitat. The discussion is organized around five elements required to support the exposure
calculation:
1. Spatial boundaries that define exposure;
2. Development of the dietary matrix;
3. Contribution to exposure from drinking water;
4. Contribution to exposure from soil, plants, and terrestrial prey; and
5. Calculation of total applied dose.
Spatial Boundaries for Exposure. All of the contaminant concentrations to which a
receptor is exposed are spatially consistent; that is, each receptor may only be exposed to
contaminant concentrations that overlap with its home range. However, it is not assumed that
100 percent of the diet originates from the home range; the Ecological Exposure Module prorates
exposure based on the ratio between the area of the home range and the total habitat area. For
example, if the home range for the weasel (-135,000 m2) is 20 percent larger than the forest
habitat, the calculated exposure is prorated by a factor of 0.8 that effectively reduces the applied
dose to 80 percent of the dose that would have occurred had the weasel taken all of its food from
within the AOI.
Development of Dietary Matrix. The Ecological Exposure Module uses the dietary
preference database to construct the short-tailed weasel's diet for each iteration. The database
includes all the relevant prey categories and the range of reported dietary preferences. As
suggested in Table 15-5, the dietary matrix can vary considerably from iteration to iteration. The
values shown in the last column comprise 100 percent of the weasel's diet for one iteration.
Note that the data set follows the two constraints for the prey preference random sampling
algorithm: (1) the diet includes all items with non-zero minimum preference values, and (2) the
dietary preferences sum to 100 percent.
15-15
-------
Section 15.0
Ecological Exposure Module
Table 15-5. Example Dietary Preferences for Short-Tailed Weasel
l-'ooil Item
Minimum
I'lvl'crciUT Vsilue
(%)
Miixiiiiiini
IVcfcrcncc Ysiliie
(%)
Viiluc for
Uoiili/iilion
(%)
Small mammals
50
80
63
Small herpetofauna
0
45
0
Soil invertebrates
0
25
13
Small birds
0
25
24
Total diet
100
Exposure Contribution from Ingestion of Surface Water. The Ecological Exposure
Module estimates exposure due to surface water ingestion by calculating the total average
concentration (dissolved plus particle-bound) for all surface waters found within the home range.
This average includes any reach and/or surface impoundment that intersects the home range.
Thus, it is implicitly assumed that the receptor does not prefer one waterbody over another. For
example, if the weasel's home range overlaps two stream reaches with calculated surface water
concentrations of 2.2E-06 mg/L and 9.3E-07 mg/L, respectively, the average drinking water
concentration for contaminant^ is 1.6E-06 mg/L.
Exposure Contribution from Ingestion of Soil, Plants, and Prey. The Ecological
Exposure Module reads concentrations in plants, prey, and soil from the Terrestrial Food Web
Module. For prey, these concentrations are spatially averaged for the prey's home range. For
soil and soil invertebrate concentrations, the spatial averaging is defined by the home range of
the short-tailed weasel. That is, the weasel is assumed to consume soil and soil invertebrates
from within its home range, and the concentrations of these food items are assumed to be
homogenous within the weasel's home range. For concentrations in small vertebrates, the
weasel may take any prey that overlap its home range. Consequently, any small mammalian,
avian, or herpetofaunal species that are assigned to the habitat are potential prey for the short-
tailed weasel. This may include species within the weasel's home range, as well as species that
simply overlap the weasel home range; thus, there is exposure variability associated with the
prey selection (i.e., the home range of the prey species) due to potential differences in tissue
concentrations in small vertebrates that may be eaten. To address this variability, the Ecological
Exposure Module reads the minimum and maximum concentrations in mobile prey categories,
such as small mammals, and randomly selects a value from that range. Thus, the exposure
concentrations reflect the full range of prey combinations that could be eaten by the short-tailed
weasel. Table 15-6 provides an example exposure profile for contaminant^ for the weasel
(calculated by the Terrestrial Food Web Module and passed to the Ecological Exposure Module).
The diet includes incidental ingestion of soil, as well as the ingestion of contaminated soil
invertebrates, small mammals, birds, and herpetofauna. A single concentration for soil, plants,
and "local" food items, such as earthworms, is passed to the Ecological Exposure Module;
however, for terrestrial prey categories that are relatively mobile, the Ecological Exposure
Module receives a range of minimum and maximum prey concentrations and chooses one from
15-16
-------
Section 15.0
Ecological Exposure Module
within that range. A uniform distribution for prey concentrations is assumed to represent free
access by predators to any prey that have access to its home range.
Table 15-6. Example Exposure Concentrations to Contaminant y for Short-Tailed Weasel
hood Item
Minimum
('onccnlnilion
(m»/k»)
Miixiiiiiini
(oiHTiili'iilion
(m»/k»)
Vsilue lor lUTiilion
(m»/k»)
Soil (incidental)
3.8E-06
3.8E-06
3.8E-06
Small mammals
1.3E-06
6.7E-05
1.3E-05
Small herpetofauna
5.4E-07
5.4E-07
5.4E-07
Soil invertebrates
8.4E-06
8.4E-06
8.4E-06
Small birds
2.3E-08
7.8E-07
7.7E-07
Calculation of Total Applied Dose. After the Ecological Exposure Module constructs
the dietary matrix and selects (from the outputs of other modules) exposure concentrations for
relevant media, plants, and prey items to which the receptor has access (i.e., that overlap with the
receptor's home range), the appropriate ecological exposure factors are used to calculate the total
applied dose for each year in the simulation as shown in Equation 15-1 (all concentrations are
converted to a wet weight basis).
Dose' =
X (IRfoodCUdFracPreyfood)+(IR'foodCsolIFracSoir)+(lR
c
water water
BodyWt1
Frac
HomeRange
(15-1)
where
Dose1,,
IR1
food
c
food
FracPrey',, 3
C\
food
soil
FracSoil1
TRi
water
r
^water
BodyWt1
Frac1
HomeRange
total applied dose for receptor i (mg/kg-day)
total ingestion rate of food for receptor i (kg/day)
concentration food item j to which receptor i may be exposed
(mg/kg tissue)
dietary fraction of food item j for receptor i (unitless)
average concentration in surficial soil in home range for receptor i
(mg/kg)
dietary fraction of soil ingested by receptor i (unitless)
ingestion rate of drinking water for receptor i (L/day)
average concentration in surface water within home range for
receptor i (mg/L)
body weight of receptor i (kg)
fraction of home range for receptor i within habitat (unitless).
The applied dose reflects media and food concentrations for a single year in the
simulation. For each realization (i.e., each time the Ecological Exposure Module runs), the
15-17
-------
Section 15.0
Ecological Exposure Module
module will read new data on media, prey, and food concentrations and construct a new dietary
matrix from the dietary preference database. Although the exposure factors (e.g., body weight,
food ingestion rate) and physical description (e.g., fraction of home range in the habitat) will
remain constant throughout the simulation, the exposure profile will change over time as a
function of changing media and food concentrations, as well as the random selection of dietary
items.
15.2.4 Calculate Applied Dose for Receptors in Margin Habitats
The applied dose to mammals and birds assigned to margin habitats is calculated using
essentially the same methodology described for terrestrial habitats. Equation 15-1 is modified to
add the incidental ingestion of sediment, as appropriate, to the exposure pathways that contribute
to the total applied dose. The list of dietary preferences for margin receptors is typically longer
than that for terrestrial receptors because many species of mammals and birds that live margin
habitats consume both aquatic biota and terrestrial biota. However, there are two subtle
differences worth noting with respect to the exposure calculations for mammals and birds that
are exposed to contaminated media and aquatic biota from waterbodies in margin habitats.
1. Exposure to Aquatic Biota within a Margin Habitat. As described in Sections
8 and 12, contaminant concentrations in sediment, surface water, and aquatic
biota are calculated for each stream reach and surface waterbody within the AOI.
The Surface Water Module reports these contaminant concentrations on a reach-
specific and waterbody-specific basis. That is, even though stream reaches may
be connected, the Surface Water Module may predict different contaminant
concentrations. As a result, the exposure concentrations in sediment, surface
water, and aquatic biota must be averaged prior to being used as inputs to
Equation 15-1. The calculated concentrations in media and food are averaged for
those reaches and waterbodies that overlap the home range for any mammalian or
avian receptor assigned to a margin habitat. For example, the home range for an
osprey assigned to a stream habitat might include three fishable stream reaches.
The concentration in TL4 fish eaten by the osprey would be the average TL4 fish
concentration across all three reaches. The underlying assumption is that a
receptor may take food from any fishable reaches or waterbodies within its home
range; therefore, the exposure concentration is an effective average for the
receptor's home range.
2. Exposure to Aquatic Biota in Adjacent Margin Habitat. Margin receptors are
presumed to rely on those fishable reaches that occur within their home range as
the source of both aquatic food items and drinking water. However, a receptor's
access to fishable reaches is not strictly constrained within its own habitat. In
many instances, stream corridors are located adjacent to wetlands and, because
receptors are likely to use reaches in each of these habitats as a source of food,
adjacent margin habitats are delineated so that "reach crossover" can occur freely
(i.e., habitats and home ranges overlap reaches in each habitat). For example, a
stream margin located adjacent to a permanently flooded forest wetland would
likely be delineated such that the habitat includes a part of both waterbodies.
Consequently, the home range of a kingfisher assigned to the stream margin could
15-18
-------
Section 15.0
Ecological Exposure Module
overlap both the stream and wetland, allowing this receptor to eat fish from either
waterbody. The fish concentration to which the kingfisher is exposed would be
the average tissue concentration, by trophic level, across the wetland and stream
reaches that intersect its home range. The 3MRA modeling system models
wetlands as distinct reaches, and, in this example, a forested wetland and a stream
corridor would each be delineated and modeled separately by the system. The
ecological connection that allows the kingfisher to take fish from both the stream
and wetland is not reflected in the modeling simulation performed by the Surface
Water Module.
The Ecological Exposure Module performs the averaging functions described above
based on the data describing the habitats and home ranges for margin receptors and uses those
average concentrations as inputs in Equation 15-1.
15.3 Module Discussion
15.3.1 Strengths and Advantages
The Ecological Exposure Module was developed to predict spatially explicit exposures to
chemical contaminants for mammalian and avian receptors. Designed to address long-term,
low-level exposures to chemical contaminants, the module offers several advantages for
national-scale applications, and is flexible enough to perform site-specific analyses. The major
strengths of the module include the following:
¦ Exposure profiles calculated as total applied dose. The vast majority of
ecotoxicological data relevant to long-term exposures are reported as applied
doses. Therefore, the Ecological Exposure Module was developed to report a
time series of annual average applied doses so that the module would be broadly
applicable to a large number of chemical contaminants. Although other metrics
for exposure were considered (e.g., dietary concentrations, body burden), the
module was designed to calculate applied dose so that EPA could take full
advantage of a substantial body of data on reproductive and developmental effects
associated with chronic exposures. This approach is consistent with EPA
guidelines and the practice of ecological exposure assessment in numerous
peer-reviewed ecological risk assessments conducted by EPA.
¦ Maintains predator-prey relationships for soil, plant, and prey
concentrations in space and time. A major strength of the Ecological Exposure
Module is its ability to calculate the total applied dose for each receptor assigned
to each habitat within the AOI. For receptors in margin habitats, this is
particularly important because the diet may include food items from both the
terrestrial and aquatic food webs, as well as incidental ingestion of contaminated
sediment and surface water. The module maintains the spatial relationship for all
contaminated media and food items so that the aggregate exposure is calculated
consistently for each year of the simulation. In addition to these advantages, the
module avoids unnecessary conservatism by prorating exposure based on size of
15-19
-------
Section 15.0
Ecological Exposure Module
the receptor's home range, explicitly recognizing that feeding is not constrained
by the habitat boundaries.
¦ Dietary variability is addressed in the selection of food items. The Ecological
Exposure Module addresses this important source of variability by (1) sampling
from the minimum and maximum concentrations predicted by the Terrestrial
Food Web Module for appropriate food categories, and (2) using a random
sampling algorithm to select dietary preference fractions for each food category.
The former method explicitly recognizes that the food categories (e.g., small
mammals) are intentionally general and that the specific prey items may vary
substantially depending on feeding opportunities. The latter method, which is
similar to the sampling algorithms used in the Aquatic Food Web Module,
recognizes that the dietary proportions will be highly variable with season and
other environmental conditions. Representing this variability (rather than
assigning fixed, point estimates for diet) represents a significant improvement to
the state-of-the-science for national-scale analyses.
¦ Designed to evolve with changing assessment goals, new science, and
improved data. Given the state-of-the-science on predicting ecological exposure
concentrations, one of the primary design goals for the Ecological Exposure
Module was flexibility. The module currently predicts contaminant
concentrations for 17 food categories; however, the module could be easily
modified to include additional categories or to collapse existing categories into
fewer categories to further reduce run time or to better reflect the data for a
particular chemical constituent. Similarly, the module currently handles each
receptor home range as a unique spatial element, even though the system
aggregates these home ranges into four bin sizes. The Ecological Exposure
Module can handle either level of resolution based on how the site layout file is
constructed. Lastly, the module was designed with future modifications in mind
that address the evolving science and data for ecological risk assessment (e.g.,
calculating receptor-specific lifetime average doses rather than annual average
doses).
15.3.2 Uncertainty and Limitations
The methodology developed to characterize potential ecological exposures, as
implemented by the Ecological Exposure Module, carries certain assumptions and limitations,
and acknowledges several important sources of uncertainty.
¦ The physical description of the spatial dimensions of ecological exposure are
fixed for each site. The Ecological Exposure Module calculates the applied
doses to receptors for a single random placement of four home range sizes within
habitats delineated at each site. Although the representative habitats were
delineated using available GIS coverages to represent potential habitats at a given
site, the habitats that may actually be present at any site may be very different
from those delineated in the site layout. Despite the fact that the home ranges
15-20
-------
Section 15.0
Ecological Exposure Module
were chosen at random and retain predator-prey interactions, these are fixed in the
site layout file.
The 3MRA system is designed to model a single chemical stressor released
within the AOI for each simulation. Given the design goals for the 3MRA
modeling system (e.g., to support national-level assessment strategies of WMUs,
waste streams, and contaminants), exposure to multiple contaminants or
contaminant releases outside of the AOI are not considered. In addition,
background concentrations of contaminants are not considered in developing
exposure estimates, nor are other potential nonchemical stressors, such as habitat
fragmentation. This introduces an unquantifiable uncertainty in the prediction of
total exposure and risk.
The exposure calculations are highly dependent on reliable estimates of plant
and prey concentrations, as well as random selection of plant/prey dietary
preferences represented in the dietary matrix. Ultimately, the veracity of the
exposure estimates is a function of the data used in predicting tissue
concentrations in dietary plants and prey. As discussed in Section 11 on the
Terrestrial Food Web Module, there is significant uncertainty associated with the
use of empirical data and default factors used to predict the uptake and
accumulation of contaminants from media into biota.
The exposure profiles generated by the Ecological Exposure Module are
based on the annual average concentrations in food items and media.
Consequently, concentration spikes due to episodic events (e.g., rain storms) or
elevated WMU source releases following waste additions are not evaluated. In
addition, the annual average approach does not capture elevated exposures during
critical life stages. More specifically, the module predicts exposures only for
adult animals; intrayear contaminant exposures to juveniles, often with very
different dietary preferences, are not predicted.
Exposure doses are adjusted to account for the proportion of a species'
required home range provided at a site. However, no adjustment is made to
account for differences in species diversity in small versus large habitat patches or
in disturbed versus undisturbed areas. The 3MRA modeling system assumes that
receptor species occur in their assigned representative habitats regardless of a
site's position in the landscape. In fact, it is probably unlikely that all of the
receptor species, particularly those less adapted to human impacts and
development, would be present. Moreover, when the habitat patches at a site are
small, it is questionable whether the entire receptor group would use the habitat.
The dietary preferences for each receptor remain constant throughout each
year in the simulation. The Ecological Exposure Module constructs the dietary
preferences for each receptor based on dietary data covering one or more seasons.
Some of the seasonal variability in the diet is captured indirectly by the
hierarchical algorithm used to determine the dietary preferences. However, the
15-21
-------
Section 15.0 Ecological Exposure Module
algorithm is implemented on data across multiple seasons and, therefore, does not
necessarily reflect seasonal differences. Similarly, the estimation of ingestion
rates was based on average, gender-neutral body weights and does not account for
differences in size, season, habitat, or activity level. However, prey preferences
are represented by distributions that are intended to reflect some of the natural
variation in wildlife feeding behaviors.
15.4 References
Anderson, A.N. 1997. Using ants as bioindicators: Multiscale issues in ant community ecology.
Conservation Ecology 1:8.
Bailey, R.G. 1996. Ecosystem Geography. Washington, DC: Springer-Verlag.
Begon, M., and M. Mortimer. 1981. Population Ecology: A Unified Study of Animals and
Plants. Sunderland, MA: Sinauer Assoc. Inc.
Bourgeron, P.S., and L.D. Engelking (eds). 1994. A Preliminary Vegetation Classification of
the Western United States. Unpublished report prepared by the Western Heritage Task
Force for The Nature Conservancy. Boulder, CO.
Caduto, M.J. 1990. Pond and Brook. Hanover, NH: University Press of New England.
Christensen, N.L., R.B. Wilbur, and J.S. McLean. 1988. Soil-vegetation correlations in the
pocosins of Croatan National Forest, North Carolina. Biological Report 88(28). U.S.
Fish and Wildlife Service, Department of the Interior, Washington, DC.
Cowardin, L.M., V. Carter, F.C. Golet, and E.T. LaRoe. 1979. Classification of Wetlands and
Deepwater Habitats of the United States. FWS/OBS-79/31. Office of Biological
Services, U.S. Fish and Wildlife Service, Department of the Interior, Washington, DC.
Curry, J.P. 1994. Grassland Invertebrates: Ecology, Influence on Soil Fertility and Effects on
Plant Growth. Chapman and Hall, London.
Damman, A.W.H., and T.W. French. 1987. The ecology of peat bogs of the glaciated
Northeastern United States, a community profile. Biological Report 85(7.16). U.S. Fish
and Wildlife Service, Department of the Interior, Washington, DC.
Davis, W.S., and T.P. Simon (eds). 1995. Biological Assessment and Criteria: Tools for Water
Resource Planning and Decision Making. Boca Raton, FL: Lewis Publishers.
Demarchi, D.A. 1996. An Introduction to the Ecoregions of British Columbia. Wildlife Branch,
Ministry of Environment, Lands, and Parks, Victoria, British Columbia. Available online
at http://www.elp.gov.bc.ca/rib/wis/eco/bce.coreg.html.
15-22
-------
Section 15.0
Ecological Exposure Module
Drake, J., and D. Faber-Langendoen. 1997. An Alliance Level Classification of the Vegetation
of the Midwestern United States. A report to the University of Idaho Cooperative Fish
and Wildlife Research Unit and the National Gap Analysis Program. The Nature
Conservancy, Midwest Conservation Science Department, Minneapolis, MN.
Federal Geographic Data Committee. 1997. Vegetation Classification Standard.
U.S. Geological Survey. Reston, VA. Available online at
http://www.nbs.gov/fgdc.veg/standards/vegstd.htm.
Glaser, P H. 1987. The ecology of patterned boreal peatlands of northern Minnesota. A
community profile. Biological Report 85(7.14). U.S. Fish and Wildlife Service,
Department of the Interior, Washington, DC.
Gosselink, J.G., and R.E. Turner. 1978. The role of hydrology in freshwater wetlands. In: R.E.
Good, D.F. Whigham, and R.L. Simpson (eds). Freshwater Wetlands: Ecological
Processes and Management Potential. New York: Academic Press.
Kadlec, R.H. and R.L. Knight. 1996. Treatment Wetlands. CRC Press, Boca Raton, Florida
Kiichler, A. W. 1964. Manual to Accompany the Map Potential Natural Vegetation of the
Conterminous United States. Special Publication No. 36. American Geographical
Society, New York, NY.
Lane, John J., and Wilma A. Mitchell. 1997. Species Profile: Alligator Snapping Turtle
(Macrolemys temminckii) on Military Installations in the Southeastern United States.
Tech. Report SERDP-97-9. U.S. Army Corps of Engineers. Strategic Environmental
Research and Development Program. Vicksburg, MS. September.
Larsen, J. A. 1982. Ecology of Northern Lowland Bogs and Coniferous Forests. New York:
Academic Press.
Mitsch, W.J., and J.G. Gosselink. 1993. Wetlands. New York: Van Nostrand Reinhold.
Niering, William A. 1985. Wetlands. In. The Audubon Society Nature Guides. Alfred A.
Knopf, Inc., New York, NY.
Norquist, H.C. 1984. A comparative study of the soils and vegetation of savannas in
Mississippi. Master's thesis, Mississippi State University.
Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association
of American Geographers 77:118-125.
Sample, B.E., M.S. Alpin, R.A. Effroymson, G.W. Suter, and C.J.E.Welsh. 1997. Methods and
Tools for Estimation of the Exposure of Terrestrial Wildlife to Contaminants.
Environmental Sciences Division Publication Number 4650. Oak Ridge National
Laboratory. Oak Ridge, TN.
15-23
-------
Section 15.0
Ecological Exposure Module
Sawyer, J.O., and T. Keeler-Wolf. 1995. A Manual of California Vegetation. Sacramento, C A:
California Native Plant Society.
Schoener, T.W. 1989. Food webs from small to large. Ecology 70(6):1559-1589.
Schoenly, K., and J. Cohen. 1991. Temporal variation in food web structure: 16 empirical
cases. Ecological Monographs 61(3):267-298.
Shafale, M., and A. Weakley. 1990. Classification of the Natural Communities of North
Carolina: Third Approximation. Natural Heritage Program, Divisions of Parks and
Recreation, N.C. Department of Environment, Health, and Natural Resources, Raleigh,
NC.
Sharitz, R., and J.W. Gibbons. 1982. The Ecology of Southeastern Shrub Bogs (Pocosins) and
Carolina Bays: A Community Profile. FWS/OBS-82/04. U.S. Fish and Wildlife
Service, Department of the Interior, Washington, DC.
Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Transactions of the
American Geophysical Union 8(6):913-920.
Suter II, G.W., D.S. Vaughan, and R.H. Gardner. 1983. Risk Assessment by Analysis of
Extrapolation Error: A Demonstration for Effects of Pollutants on Fish. Environ. Toxicol.
Chem. 2:369-378.
Tanner, J. T. 1978. Guide to Study of Animal Populations. Knoxville, TN: University of
Tennessee.
USDA (U.S. Department of Agriculture) Forest Service. 1994. Ecological Subregions of the
United States. Available online at http://www.fs.fed.us/land/pubs/ecoregions.
USDA (U.S. Department of Agriculture). 1998. Stream Corridor Restoration: Principles,
Practices, and Processes. Available online at
http://www.usda.gov/stream_restoration/newtofc.htm.
U.S. EPA (Environmental Protection Agency). 1993. Wildlife Exposure Factors Handbook.
Volumes I and II. EPA/600/R-93/187. U.S. Environmental Protection Agency, Office of
Health and Environmental Assessment and Office of Research and Development,
Washington, DC. December.
U.S. EPA (Environmental Protection Agency). 1993a. A Review of Ecological Risk Assessment
Case Studies from a Risk Assessment Perspective. EPA/630/R-92/005. U.S.
Environmental Protection Agency, Risk Assessment Forum, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1994. 1:250,000 Scale Quadrangles of
Landuse/Landcover GIRAS Spatial Data in the United States. Office of Information
Resources Management (OIRM), Washington, DC. Available online at
http://www.epa.gov/ngispgm3/nsdi/projects/giras.htm.
15-24
-------
Section 15.0
Ecological Exposure Module
U.S. EPA (Environmental Protection Agency). 1998. Guidelines for Ecological Risk
Assessment. EPA/630/R-95/002F. Risk Assessment Forum. Washington, DC. April.
U.S. EPA (Environmental Protection Agency). 2000. Background Document for the Ecological
Exposure and Ecological Risk Modules for the Multimedia, Multipathway, Multireceptor
Risk Assessment (3MRA) Software System. Office of Solid Waste, Washington, DC.
August.
U.S. EPA (Environmental Protection Agency). 1999b. Data Collection for the Hazardous Waste
Identification Rule. Section 13: Ecological Receptors and Habitats. Office of Solid
Waste, Washington, DC. July.
http://www.epa.gov/epaoswer/hazwaste/id/hwirwste/risk.htm
U.S. FWS (Fish and Wildlife Service). 1998a. An Ecosystem Approach to Fish and Wildlife
Conservation. Available online at
http://bluegoose.arw.r9.fws.gov/NWRSFiles/HabitatMgmt.
Verry, E.S. 1997. Hydrologic processes of natural northern forested wetlands. In: C.C. Trettin,
M.F. Jurgensen, D.F. Grigal, M.R. Gale, and J.K. Jeglum (eds). Northern Forested
Wetlands: Ecology and Management. Boca Raton, FL: Lewis Publishers.
Viereck, L.A., and L.L. Elbert, Jr. 1991. Alaska Trees and Shrubs. Agricultural Handbook No.
410. U.S. Department of Agriculture, Forest Service, Washington, DC.
Weakley, A.S., K.D. Patterson, S. Landal, and M. Pyne. 1998. International Classification of
Ecological Communities: Terrestrial Vegetation of the Southeastern United States.
(Working draft of March 1998). The Nature Conservancy, Southeast Regional Office,
Chapel Hill, NC.
Willner, G.R., G.A. Feldhamer, E. E. Zucker, and J. A. Chapman. 1980. Ondatra zibethicus.
Pp. 1-8 in Mammalian Species, 141. American Society of Mammalogists. Available at
www. mammal society. org.
Windell, J.T., B.E. Willard, D.J. Cooper, S.Q. Foster, C.F. Knud-Hansen, L.P. Rink, and G.N.
Kiladis. 1986. An ecological characterization of Rocky Mountain montane and
subalpine wetlands. Biological Report 86(11). U.S. Fish and Wildlife Service,
Department of the Interior, Washington, DC.
Winter, T.C. 1989. Hydrologic studies of wetlands in the northern prairies. In: vanderValk
(ed). Northern Prairie Wetlands. Ames, IA: Iowa State University Press.
Whitney, Stephen. 1985. Western forests. In. The Audubon Society Nature Guides. Alfred A.
Knopf, Inc., New York, NY.
15-25
-------
Section 15.0
15-26
Ecological Exposure Module
-------
Section 16.0
Ecological Risk Module
16.0 Ecological Risk Module
16.1 Purpose and Scope
The Ecological Risk Module calculates the hazard quotient (HQ) for species of
mammals, birds, and herpetofauna and, through inference, for communities of organisms that
live in close contact with soil (including plants and soil invertebrates), sediment (benthic
invertebrates), and surface water (e.g., fish, aquatic invertebrates, algae, and aquatic plants). The
HQ is the ratio of the annual average exposure (in units of concentration or dose) to a benchmark
for ecological effects (in units of concentration or dose) and quantifies the potential for
contaminants to elicit an adverse ecological response once released into the environment. To
calculate the HQs, the Ecological Risk Module uses input concentrations from the Surface Water
and Terrestrial Food Web Modules, as well as applied doses calculated by the Ecological
Exposure Module, and compares those values to ecotoxicological benchmarks (EBs, in units of
dose) and chemical stressor concentration limits (CSCLs, in units of concentration), as
appropriate. The module calculates an HQ for every ecological receptor assigned to habitats
within the area of interest (AOI) for a given site. Detailed information on the Ecological Risk
Module can be found in the background document (U.S. EPA, 2000). Figure 16-1 shows the
relationship and information flow between the Ecological Risk Module and the 3MRA modeling
system.
Key Data Inputs
• Ecological benchmarks
• Habitat type
• Water hardness
Terrestrial
Food Web
Module
Soil
Concentrations
Ecological
Risk
Module
Exit Level
Processors
I and II
Surface
Water
Module
Water Column and
HQs
Sediment Concentrations
Ecological
Exposure
Module
Doses
Figure 16-1. Information flow for the Ecological Risk Module
in the 3MRA modeling system.
16-1
-------
Section 16.0
Ecological Risk Module
The conceptual approach to characterizing the potential for adverse ecological effects
depends on the assessment endpoints used. Assessment endpoints are explicit expressions of the
actual environmental values that are to be protected (U.S. EPA, 1998). Candidates for
assessment endpoints often include threatened or endangered species, functional attributes that
support food sources or flood control, or aesthetic values such as the existence of charismatic
species (e.g., eagles) that have special value to society (U.S. EPA, 1998). The assessment
endpoints must be defined with respect to the valued ecological entity (e.g., a particular species),
and an attribute of that entity that is to be protected (e.g., reproductive fitness). For the 3MRA
modeling system, the assessment endpoints were selected to represent multiple levels of
biological organization (e.g., populations, communities), key functional elements of natural
communities (e.g., primary producers), and biota throughout the trophic continuum. Although
the 3MRA modeling system does not restrict the selection of assessment endpoints for future
applications, the Ecological Risk Module and supporting databases were developed to evaluate
three primary assessment endpoints:
¦ Survival of species that comprise key structural and functional elements of soil,
freshwater, and benthic (sediment) communities;
¦ Reproductive fitness and survival of mammalian, avian, and herpetofaunal
wildlife populations; and
¦ Growth and survival of primary producers (e.g., plants) in terrestrial and
freshwater systems.
These endpoints are ecologically relevant to the habitat types used to represent ecological
variability in the 3MRA modeling system (see Section 15.2) and are sensitive to a broad range of
chemical stressors.
Table 16-1 describes the specific assessment endpoints addressed by the 3MRA modeling
system in terms of: (1) the significance of the ecological entity (i.e., the reason EPA wants to
protect it), (2) the ecological receptor(s) that represent that entity, (3) the characteristic(s) of the
entity that is important to protect, and (4) the measure of effect used to quantify the potential for
an adverse response. Measures of effect such as fecundity and mortality were chosen to support
the development of the EBs and CSCLs based on their relevance to the assessment endpoints.
The Ecological Risk Module performs two major functions:
1. Calculate Hazard Quotients (HQs). The Ecological Risk Module calculates
HQs for each receptor at each site according to spatial, temporal, and
environmental characteristics of the site.
2. Process the HQ Results for Decision Making. The Ecological Risk Module
tracks various attributes such as taxa and habitat type to process the HQ results
for decision making. This processing includes placing results in bins and
determining the timing of maximum risks.
16-2
-------
Table 16-1. Assessment Endpoints and Measures of Effects for the 3MRA Modeling System
Asm-smiu-iH r.ndpuiiil
r.\;ini|)k' ki'iipliii's
('h;ir;Kii'i'islii'(s)
Mi-:isuiv(s) ill' riTiii
survival and reproductive
fitness of mammalian wildlife
populations
¦ multiple trophic levels and apex predators
¦ multiple exposure pathways represented
¦ species with large foraging ranges
¦ charismatic species (e.g., eagles)
¦ sensitive species (e.g., mammals and
dioxins)
deer mouse, meadow
vole, red fox
reproductive success in
adult animals; survival
and development of
offspring
chronic/subchronic MATL for
physiological effects relevant to
reproductive success (e.g.,
gonadotoxicity; number of viable
offspring; egg production) and
development (e.g., malformations;
growth rate)
survival and reproductive
fitness of avian wildlife
populations
red-tailed hawk,
northern bobwhite
survival of amphibian
populations
¦ unique habitat niches (e.g., partially
aquatic and terrestrial)
¦ includes species that are particularly
sensitive to chemical stressors, particularly
metals
frog, newt
lethality and deformity
primarily acute LC50s for lethality
resulting from early life stage
exposures; EC20s for developmental
effects used when available
reproductive success and
survival of reptile populations
snake, turtle
reproductive success in
adult animals; survival
and development of
offspring
none identified for chemical stressors
of concern
survival of species that
comprise key structural and
functional elements of the soil
community
¦ high levels of exposure through direct
contact
¦ base of the food web in terrestrial systems
¦ vital habitat for decomposers, soil aerators
¦ essential for nutrient cycling
nematodes,
arthropods,
springtails, annelids,
mites
mortality, growth,
survival, reproductive
success
95% of soil species below low effects
concentration at 50th percentile
confidence interval; when data were
insufficient, LOEC values for
earthworms, microbes used
growth and survival of
terrestrial plants
¦ primary producers of energy
¦ food base for herbivores
¦ essential habitat for many wildife species
¦ large fraction of the earth's biomass
soy beans, alfalfa,
rye grass
plant growth and yield
10th percentile from data on LOEC
from studies on plant growth, seed
germination; when data were
insufficient, lowest LOEC value was
selected
survival of species that
comprise key structural and
functional elements of the
freshwater aquatic community
¦ high levels of exposure through direct
contact
¦ includes several species that are very
sensitive to chemical stressors (e.g.,
daphnids)
¦ important food source for wildlife that live
in waterbody margin
fish, aquatic
invertebrates
mortality, growth,
survival, reproductive
success
National Ambient Water Quality
Criteria (NAWQC) for aquatic life
(95% of aquatic species); alternate
water quality criteria developed using
abbreviated data sets
(continued)
-------
Table 16-1. (continued)
Assessment End point
Ecological Significance Example Receptors Characteristics) Measure(s) of Effect
survival of species that
comprise key structural and
functional elements of the
sediment community
¦ sediment serves as sink for persistent
contaminants, resulting in high levels of
exposure
¦ habitat for early life stages (e.g., midge
larvae)
¦ nutrient processing and decomposition
protozoa, flat worms,
ostracods
mortality, growth,
survival, reproductive
success, and community
measures such as
abundance, diversity
metals - threshold effects levels from
LOEC data associated with
community endpoints (e.g., species
abundance)
organics - water quality criteria
adjusted for sorption to organic carbon
in sediment
growth and survival of aquatic
plants and algae
¦ primary producers of energy
¦ base of food source in the aquatic system
¦ substrate for other organisms in the water
column
¦ essential habitat for developing organisms
algae and vascular
aquatic plants (e.g.,
duckweed)
growth, cell numbers,
mortality, biomass, root
length
algae - EC20 and EC50 for growth,
decreased cell numbers, reduction in
carbon fixation
vascular plants - lowest LOEC for
endpoints relevant to growth, biomass,
root number
MATC maximum allowable toxicant level
LOEC lowest observed effect concentration
LC50 lethal concentration to 50 percent of the organisms
EC50 effective concentration to elicit a response in 50 percent of the organisms
EC20 effective concentration to elicit a response in 20 percent of the organisms
s?
o
C§
cS"
s
&
>3
o
-------
Section 16.0
Ecological Risk Module
Because the assessment endpoints evaluated include both populations of wildlife species and
assemblages of species that represent communities, the calculated HQs for different receptors
represent different measures of effect at several levels of biological organization. As a result, it
is critical to understand the rationale behind the development of the EBs and CSCLs included in
the current 3MRA sample data base. Therefore, although these benchmarks are not calculated
by the Ecological Risk Module, the development of these benchmarks is discussed in this
section.
16.2 Conceptual Approach
The Ecological Risk Module quantifies the potential for adverse ecological effects by
calculating HQs for individual receptors such as raccoons, aquatic plants, or the soil community.
The HQs provide a "bright line" metric for risks to the individual organisms that represent
wildlife populations and, based on statistical inference, the risks to narrowly defined
communities (e.g., the sediment community). The Ecological Risk Module does not estimate
population-level risks, defined as the likelihood that some percentage of the individuals in a
population will sustain an adverse effect, nor does it characterize the risks to communities.
Although population-level models have long been used by ecologists to evaluate the response of
species populations to certain types of stressors, the data requirements and implications on model
run time prohibited such an approach in this version of the 3MRA modeling system. Thus, the
Ecological Risk Module uses specific measures of effect (e.g., a reduction in viable offspring) to
generate HQs that may be used to infer whether the potential for an adverse ecological response
is above levels of concern. That is, a target HQ of 1 serves as the indicator for adverse
ecological responses; a value greater than 1 indicates that the ecological response is above a
level for concern, and a value less than 1 indicates that the potential for adverse ecological
effects is at a de minimis level for the receptor for which the HQ was calculated. The
implications of this approach underscore the importance of developing EBs and CSCLs that
reflect the assessment endpoints for the modeling application, tracking the attributes of each HQ
that is calculated, and processing the HQ results in a transparent manner to support decision
making.
16.2.1 Development of EBs and CSCLs
Receptor-specific benchmarks1 (EBs and CSCLs) were developed for use in the
Ecological Risk Module. As indicated previously, these benchmarks were based on measures of
effects considered appropriate to support risk inferences for ecological receptors at various levels
of biological organization, including individual organisms, wildlife species populations, and
communities. In identifying appropriate studies from which to develop EBs and CSCLs, study
selection criteria were developed to ensure consistency in the interpretation of study results and
to satisfy data quality objectives. The study selection criteria address the desire for consistency
across EPA programs as well as within the representative national data set, the appropriateness
of the study given the assessment endpoints used in the 3MRA modeling system; and the quality
1 The term " benchmarks" is used in this discussion to refer to both EBs for mammalian and avian species
and CSCLs for selected receptors (e.g., amphibian species) and communities.
16-5
-------
Section 16.0
Ecological Risk Module
of the study with respect to endpoint selection, dose-response information, and appropriate use
of extrapolation techniques (e.g., tools for statistical inference).
The EBs represent a de minimis level of effect for mammals and birds exposed through
the ingestion of contaminated media, plants, and prey. In order of importance, the study
selection criteria considered the following:
1. Relevance of study endpoints to population-level effects;
2. Whether the study contained adequate data to demonstrate a dose-response
relationship;
3. Appropriateness of study design with respect to the exposure route (e.g., gavage
versus dietary exposure) and duration (e.g., acute versus chronic);
4. Quality of the study as determined by the use of appropriate dosing regimes,
statistical tools, etc.; and
5. Consistency with other EPA programs, such as Superfund and the Office of
Water.
The CSCLs (except those for amphibians and plants) are intended to represent a de
minimis level of effect for communities that live in close contact with the soil, sediment, or
surface water. The study selection criteria for these receptors considered the following:
1. Acceptance of the benchmark by other EPA programs (e.g., National Ambient
Water Quality Criteria);
2. Consistency with EPA guidelines on study selection for aquatic toxicity data;
3. Relevance of study to species identified as key functional elements of the
community;
4. Relevance of study endpoints to address community-level effects (e.g., growth,
survival, abundance);
5. Whether the study contained adequate data to demonstrate a dose-response
relationship; and
6. Quality of the study data with respect to the design (e.g., field versus laboratory),
and appropriate use of statistical tools to characterize effects (e.g., confidence
levels).
For amphibians, the extensive database on acute and subchronic aqueous exposures to
developing organisms was used to derive CSCLs for surface water contact at sensitive life stages
over short durations (e.g., less than 7 days). For terrestrial plants, studies on growth, seed
germination, and other relevant endpoints were used to derive CSCLs for exposure durations that
16-6
-------
Section 16.0
Ecological Risk Module
roughly correspond to a typical growing season (e.g., 3 months). The study data on vascular
aquatic plants was very limited, and short-term studies on algal growth were frequently selected
to evaluate primary producers in aquatic systems.
The following subsections provide a brief description of benchmark development for the
following receptor groups:
¦ Mammals and birds,
¦ Amphibians,
¦ Reptiles,
¦ Soil community,
¦ Terrestrial plants,
¦ Freshwater aquatic community,
¦ Sediment community, and
¦ Aquatic plants and algae.
The discussions highlight the minimum requirements for data used to support benchmark
development and the calculation methods used. To provide the context of each description, the
assessment endpoint and measure of the effect for each receptor group is presented at the
beginning of each subsection. A comprehensive description of the data collection efforts and
methodology developed to derive ecological benchmarks may be found in Volume II of this
report.
Mammals and Birds.
¦ Assessment Endpoint. survival and reproductive fitness of mammalian and avian
wildlife populations. The characteristics to be protected are: (1) the reproductive
success of adult animals, and (2) the growth and development of offspring.
¦ Measure of Effect, a de minimis thresholdfor developmental and reproductive
toxicity in mammalian and avian wildlife species. The threshold was calculated
as the geometric mean of the No Observed Adverse Effect Level (NOAEL) and
the Lowest Observed Adverse Effect Level (LOAEL), defined as the maximum
acceptable toxicant level (MATL) in the 3MRA database. Implicit in this
calculation is the assumption that the toxicological sensitivity is lognormal.
For mammals and birds, ecotoxicological data were evaluated to identify the most
appropriate studies to support EB development. Studies meeting the data quality objectives
summarized above (e.g., sufficient dose-response data), were reviewed and the study with the
most sensitive endpoint was selected to derive receptor-specific benchmarks. Once the MATL
was calculated from the NOAEL and LOAEL, the MATLs for mammals and birds were scaled
for each receptor species using the relative body weights of the test and receptor species. The
scaling equation is based on the default methodology EPA proposes for carcinogenicity
assessments and reportable quantity documents for adjusting animal data to an equivalent human
dose (U.S. EPA, 1992), and is widely used in ecological risk assessments (see, for example,
Sample et al., 1998). Research indicates that the cross-species scaling equation for mammals is
16-7
-------
Section 16.0
Ecological Risk Module
inappropriate for direct application to birds (Mineau et al., 1996). Therefore, an avian scaling
factor developed by Mineau was used to derive EBs for birds.
Amphibians.
¦ Assessment Endpoint: survival of amphibian populations. The characteristics to
be protected were the survival, growth, and development through early life stages.
¦ Measure of Effect, an acute LCS0for lethality or developmental effects resulting
from early life stage exposures.
For amphibians, the available data on toxicity were limited almost exclusively to acute
studies on lethality and, in some cases, growth and developmental effects. Amphibians appear to
be highly sensitive to a number of toxicants (e.g., trace metals) during the developmental stages
of their life cycle. After a review of several compendia presenting amphibian toxicity data (e.g.,
U.S. EPA, 1996; Power et al., 1989) as well as primary literature sources, it was determined that
the lack of standard methods on endpoints, species, and test durations would preclude the
development of a CSCL for chronic exposures. Available data typically involved aqueous
exposure during early life stages and, as a result, the only exposure pathway that could be
evaluated for amphibians was direct contact with contaminated surface waters. The CSCL for
early life stage effects was calculated as the geometric mean of acute LCS0 data (lethal
concentration to 50 percent of the organisms). A few general guidelines were followed in
selecting acute studies for developing the CSCL: (1) test duration was usually less than 15 days,
(2) only studies on lethality were included to calculate the geometric mean, and (3) exposure
occurred during early life stages (i.e., embryo, larvae, and tadpole).
Reptiles.
¦ Assessment Endpoint. survival and reproductive fitness of reptile populations.
The characteristics to be protected are: (1) the reproductive success of adult
animals, and (2) the growth and development of offspring.
¦ Measure of Effect, a de minimis thresholdfor developmental and reproductive
toxicity to reptile species. The threshold is calculated using the same approach as
that taken for mammals and birds.
Toxicity data relevant to reptile exposures could not be identified for the chemical
stressors in the representative national data set. Additional research is needed to: (1) identify
more recent data for applicability to reptile exposures and (2) determine whether methods can be
developed to extrapolate benchmarks across species.
Soil Community.
¦ Assessment Endpoint: survival of species that comprise key structural and
functional elements of the soil community. The characteristics to be protected
include mortality, growth, survival, and reproductive success of selected taxa.
16-8
-------
Section 16.0
Ecological Risk Module
¦ Measure of Effect: concentration in soil that, for 95 percent of the species, will be
below the low effects concentration at the 50th percentile confidence interval. The
CSCLs for the soil community were typically derived at a 95 percent protection
level using primarily low effects data for the endpoints of interest.
Ecotoxicological data on lowest observed effects concentrations (LOECs) were reviewed
for soil biota for a number of functional categories in soil systems (e.g., decomposers, predators)
to derive the soil community CSCLs. Thus, criteria were required both for individual study
reviews as well as to evaluate the entire data set for completeness. Criteria developed for study
selection included four categories of exposure: (1) topical application; (2) surface-soil
application, in which the soil organisms are placed onto a treated surface; (3) mixed-soil
application, in which the soil organisms are placed into a soil that was mixed with a
contaminant; and (4) food application (i.e., contaminant mixed with organic food source). The
endpoints for soil species were selected based on relevance to species populations and, in order
of preference, included reproduction, growth, mortality, population increase/decrease, sexual
development, mobility, and regeneration.
The first, and preferred, method was based on a community-level approach analogous to
that used to develop the National Ambient Water Quality Criteria (NAWQC) for the protection
of aquatic biota. Like the NAWQC, the soil community CSCLs are intended to protect the
structure and function of the community and its critical role in the terrestrial food web (e.g.,
nutrient processing). A detailed discussion of the calculations used to develop a community-
based soil CSCL are provided in Volume II of this report.2 The second method used to derive
soil CSCLs required the identification of LOECs for earthworms and microbial endpoints despite
the obvious limitations in relying on a single species to infer adverse effects to the community.
Nevertheless, earthworms have been recognized for the critical role they play in promoting soil
fertility, releasing nutrients, and providing aeration and aggregation of soil, as well as being an
important food source for higher trophic level organisms. In addition, their constant contact with
soil media and permeable epidermis makes them more susceptible to contaminant exposures.
Likewise, microbial communities play a key functional role in soil fertility, decomposition
processes, and nutrient cycling, providing nutrients in available forms to plants. Microbial
CSCLs were used only when they indicated a significantly higher sensitivity to a particular
contaminant than the corresponding earthworm toxicity data. The earthworm/microbial CSCLs
were derived using one of two approaches: (1) if more than 10 studies were identified reporting a
LOEC, the 10th percentile value was selected as the CSCL, or (2) if fewer than 10 values were
identified, the lowest LOEC was selected as the CSCL.
Terrestrial Plants.
¦ Assessment Endpoint. growth and survival of terrestrial plants. The
characteristics to be protected include the growth and yield of terrestrial plants.
2 The discussion in Volume II uses no observed effects concentrations (NOECs) rather than LOECs. The
methodology presented is entirely accurate; however, EPA has elected to use the less conservative LOECs to derive
the soil CSCLs based on a comparison between NOEC-derived CSCLs and typical background concentrations.
16-9
-------
Section 16.0
Ecological Risk Module
¦ Measure of Effect, soil concentrations related to growth, yield, seedling
emergence, and germination endpoints. The lowest observed effects data on
phytotoxicity were rank ordered, and the plant CSCL was estimated as the 10th
percentile value.
The development of CSCLs for terrestrial plants primarily included endpoints relevant to
growth and yield (e.g., seed germination, seedling emergence). Data collection and review
activities were focused on these endpoints because they are ecologically significant responses
and because the database of phytotoxicity studies provides sufficient coverage of these types of
effects (Efroymson et al., 1997). However, very few contaminants have toxicity data for a
sufficient number of terrestrial species to represent even a simple plant community, including
short-lived and long-lived plants, flowering and nonflowering plants, high seed producers, and
plants with extensive root systems (Eijsackers, 1994). Consequently, the data quality
requirements presented in Efroymson et al. (1997) regarding study preferences (e.g., field studies
were preferred over greenhouse studies) were adopted for development of soil CSCLs for plants.
The terrestrial plant CSCLs were derived using the same approach as described previously for
the earthworm/microbial CSCLs (i.e., based on the number of studies).
Freshwater Aquatic Community.
¦ Assessment Endpoint: survival of species that comprise key structural and
functional elements of the freshwater aquatic community. The characteristics to
be protected include mortality, growth, survival, and reproductive success of
selected taxa.
¦ Measure of Effect: surface water concentration that, for 95 percent of the species,
will be below the low effects concentration at 50th percentile confidence interval.
Typically, the NAWQC was chosen as the CSCL for this receptor group.
The CSCLs for the freshwater community reflect endpoints ranging from mortality to
growth and reproductive effects. As with the soil community, criteria for individual study
selection were established, as well as criteria applicable to the data set as a whole. The
minimum data requirements to derive a CSCL for the aquatic community were based on the
Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of
Aquatic Organisms and Their Uses (Stephan et al., 1985), and, for data sets that did not meet
those requirements, the Tier II guidelines proposed in the Water Quality Guidance for the Great
Lakes System and Correction; Proposed Rule (58 FR 20802). These methods require the
compilation of appropriate acute and chronic toxicity data on adverse effects to aquatic biota for
specific taxa of the freshwater community. Specifically, either the Final Chronic Value (FCV)
developed using the above guidance or the criterion continuous concentration (CCC) developed
for the Great Lakes Water Quality Initiative (GLWQI) was selected as the CSCL for the
freshwater aquatic community. If neither a CCC nor an FCV was available, a Secondary
Chronic Value (SCV) was calculated using Tier II methods developed through the GLWQI
(Stephan et al., 1985; Suter and Tsao, 1996).
16-10
-------
Section 16.0
Ecological Risk Module
Sediment Community.
¦ Assessment Endpoint: survival of species that comprise key structural and
functional elements of the sediment community. The characteristics to be
protected include mortality, growth, survival, and reproductive success of selected
taxa.
¦ Measure of Effect: two types of measures were used, one for organics (based on
the NAWQC) and one for metals (similar to the approach for earthworms and
microbes). The sediment quality criteria based on partitioning theory and the
similarity in toxicological sensitivity between water column and sediment biota
were used to derive the CSCLs for organic chemicals. Empirical data from field
studies were used to derive the CSCLs for metals.
Two methods were applied in developing the CSCLs for the benthic community (e.g.,
worms, amphipods). For metals and ionic organic chemicals, CSCLs were based on either:
(1) threshold effects levels (TELs) developed by the Florida Department of Environmental
Protection (FDEP), which are the upper limit of the range of sediment concentrations for
endpoints on survival, species diversity, and abundance; or (2) the 10th percentile effects
concentrations (ER-L) developed by National Oceanic and Atmospheric Administration's
(NOAA) National Status and Trends (NS&T) Program. The TELs were preferred over the ER-L
values because: (1) the same database was used for both the NOAA criteria and the FDEP
criteria development; (2) in most cases, the FDEP criteria were based on more appropriate
measurement endpoints; and (3) the marine TELs developed by the FDEP were found to be
analogous to TELs observed in freshwater organisms (Smith et al., 1996).
For nonionic organic chemicals, the CSCL derivation method was based on an
equilibrium partitioning relationship between sediment and surface water. This method
calculates the sediment CSCL based on the surface water FCV or SCV, assuming that the
equilibrium partitioning between sediment and the water column is a function of the fraction
organic carbon. In calculating the baseline sediment CSCL for nonionic chemicals, the fraction
organic carbon was assumed to be 1, and the organic carbon partitioning coefficients were
adopted as reported in Jones et al. (1997). However, because the HQs are intended to represent
the ecological hazard for the environmental characteristics of each site, the Ecological Risk
Module adjusts the sediment CSCL for the site-specific fraction organic carbon prior to
calculating the sediment community HQ value.
Aquatic Plants and Algae.
¦ Assessment Endpoint. growth and survival of aquatic plants and algae in
freshwater systems. The characteristics to be protected are different for aquatic
plants and algae, and include root length and biomass, and growth and cell
numbers, respectively.
¦ Measure of Effect, surface water concentrations related to gross measures of
health (e.g., biomass) for the algal community and a variety of endpoints for
aquatic plants (e.g., number of fronds, root number, plant number, root length).
16-11
-------
Section 16.0
Ecological Risk Module
For algae, the effective concentration to 20 percent of the population (EC20) was
selected as the threshold for an adverse response. For plants, the lowest LOEC
for endpoints of interest was chosen as the CSCL because of the paucity of data
and the importance of vascular plants to maintain a healthy aquatic ecosystem.
The CSCLs developed for primary producers in aquatic systems include data on both
vascular aquatic plants and algae. Algae were included in this receptor group because they have
a relatively long history of toxicity testing and have often been shown to be more sensitive than
vascular aquatic plants to chemical stressors (Klaine and Lewis, 1995). For algae, the measure
of effect was based on the EC20 and EC50 values related to growth inhibition, decreased cell
numbers, and reduction in carbon fixation as common responses measured in algal toxicity tests.
For aquatic plants such as duckweed (e.g., Lemna minor), the endpoints included LOECs for
development of fronds, biomass, root number and length, and plant number. The toxicological
data relevant to this receptor group were identified in the open literature and from data compiled
in Toxicological Benchmarks for Screening Potential Contaminants of Concern for Effects on
Aquatic Biota: 1996 Revision (Suter and Tsao, 1996).
16.2.2 Calculate Hazard Quotients
The Ecological Risk Module calculates an HQ for receptors assigned within the AOI for
a given site. The module may calculate more than one HQ for a receptor if the receptor is
assigned to more than one habitat; however, the module calculates only one HQ per habitat for
any receptor.3 In calculating an HQ, the Ecological Risk Module checks first to determine
whether an EB or CSCL (as appropriate) is available for a particular receptor. An HQ is not
reported for a receptor if no benchmark is available; however, the lack of an ecological
benchmark does not indicate the lack of ecological risk.
The Ecological Risk Module calculates HQs for each year, habitat, receptor, and site.
Annual average media concentrations and applied doses are used to represent chronic, long-term
exposures to chemical stressors released from waste management units. Although the HQ
calculation is essentially the same for any receptor, the spatial scale and environmental
characteristics relevant to the HQ calculations are handled differently for different
habitat/receptor/constituent combinations. Therefore, it is useful to organize the discussion of
HQ calculations around three basic groups of receptors:
1. Freshwater Receptors (freshwater aquatic community, sediment community,
amphibians, and aquatic plants and algae). The module calculates HQs for
these receptors based on appropriate CSCLs, determines the average
concentration across the habitat or ephemeral water body, and adjusts certain
CSCLs for environmental conditions prior to calculating the HQ.
2. Soil receptors (soil community and terrestrial plants). The module calculates
HQs for these receptors based on the depth-averaged soil concentration and the
3 There is one exception to this rule: more than one HQ can be calculated and reported for the soil
community and terrestrial plants. This exception is explained in Section 16.2.2.2.
16-12
-------
Section 16.0
Ecological Risk Module
soil CSCLs, and reports an HQ for each receptor with a unique spatial definition
(based on the receptor home ranges) in each habitat.
3. Mammals, birds, and reptiles. The module calculates HQs for mammals and
birds using the applied doses calculated by the Ecological Exposure Module and
the receptor-specific EBs. The spatial scale for each mammalian and avian HQ
calculated by the module is based on the home range of the species. As noted
earlier, insufficient data were available for the current 3MRA data set to calculate
reptile HQs.
Freshwater Receptors. By definition, freshwater habitats (i.e., waterbody and wetland
margins) include at least one fishable surface waterbody. Prior to calculating HQs for these
receptors, the Ecological Risk Module identifies which habitats are waterbody margin or
wetland, determines the number of reaches that are connected in each habitat,4 and calculates the
average surface water and sediment concentration for the habitat. For surface water, the module
calculates average concentrations for both the total (i.e., dissolved plus bound fraction) and
freely dissolved phases for each constituent.
For metals, the module adjusts the surface water CSCL to account for the effect of water
hardness on toxicity. This adjustment is made only for cationic metals for which toxicological
sensitivity has been shown to correlate with water hardness and for which a quantitative
expression of that relationship has been developed by EPA. For nonionic organics, the module
adjusts the sediment CSCL for the site-specific fraction organic carbon to ensure that the
partitioning reflects the site conditions.
Using the average surface water and sediment concentrations and the adjusted CSCLs (if
appropriate), the Ecological Risk Module calculates HQs for the aquatic community, aquatic
plants or algae, amphibians, and the benthic community as follows:
rrn = Cmediumhabitati n . n
Qreceptorhabuatl ( " )
receptor
where
HQ receptor habitat = hazard quotient for receptor in habitat i (unitless)
Cmedium habitat* = concentration in surface water or sediment in habitat i
(mg/L or mg/kg)
CSCLreceptor = chemical stressor concentration limit for receptor (mg/L or
mg/kg).
The Ecological Risk Module uses two conventions that pertain to the use of surface water
CSCLs to calculate HQs. First, the Ecological Risk Module uses the freely dissolved chemical
4 The 3MRA modeling system treats all waterbodies, including streams, lakes, ponds, and permanently
flooded wetlands, as reaches.
16-13
-------
Section 16.0
Ecological Risk Module
concentration if the corresponding CSCL also reflects freely dissolved chemical. The total water
concentration is used to calculate the HQ if no freely dissolved CSCL is available (typically,
surface water CSCLs are available only for total surface water concentrations). Second, the
module calculates HQs for amphibians in intermittently flooded wetlands, reach order 2 streams,
or any other transitory bodies of water that can serve as a breeding ground for amphibians.
Indeed, an estimated 30 percent and 50 percent of all species of caudates (salamanders) and
anurans (frogs), respectively, use temporary ponds for breeding, and many of these species
reproduce in them exclusively (Freda, 1991).
Soil Receptors. For the soil community and terrestrial plants, the Ecological Risk
Module calculates an HQ for each unique home range in each habitat delineated at the site. Four
possible unique home ranges are currently implemented in the 3MRA modeling system
(although, as indicated in Section 15, the 3MRA modeling system can process a unique home
range for every receptor). For terrestrial habitats at many sites, the two largest home range sizes
are larger than the habitat delineated within the AOI. As discussed in Section 15, these home
ranges were constrained by the size of the habitat and, therefore, have identical spatial
characteristics. In these instances, the HQs for these home ranges would be identical and reflect
the same spatial averaging within the habitat and reporting them all would be redundant.
Therefore, for most receptors, only one HQ is calculated per habitat, with the exception of plants
and soil community receptors. This exception accounts for the fact that wildlife are tightly
coupled to their environment, and adverse effects to the base of the food web may be highly
significant ecologically. Consequently, the HQs for the soil community and terrestrial plants are
calculated for unique spatial averages (defined by the home range) within each habitat. The HQs
for soil receptors are calculated as the ratio of the root zone soil concentration (depth-averaged to
5 cm) to the soil CSCL for soil biota and plants, respectively, as follows:
<16-2>
receptor
where
HQreceptor HomeRangei = hazard quotient for receptor in home range i (unitless)
CSOii HomeRangei = root zone soil concentration in home range i (mg/kg)
CSCLreceptor = chemical stressor concentration limit for receptor (mg/kg).
For any terrestrial habitat delineated within the AOI, between 1 and 4 HQs will be calculated for
the soil community and terrestrial plants, respectively, provided that the chemical database
contains CSCLs for both receptor types.
Mammals, Birds, and Reptiles. The Ecological Risk Module calculates HQs for
mammals and birds as the ratio of the applied dose from the Ecological Exposure Model to the
receptor-specific EB. Thus, Equation 16-3 has the same general form as the previous two HQ
equations; however, the applied dose is used to represent the environmental exposure through the
ingestion pathway, and the EB is used as the ecological benchmark for endpoints relevant to
population growth and survival.
16-14
-------
Section 16.0
Ecological Risk Module
AppliedDose]
receptor
receptor
habitat
(16-3)
receptor
where
AppliedDose,
FR
receptor
.receptorhabitat1
'receptor
hazard quotient for receptor in habitat i
applied dose to receptor (mg/kg-d)
ecological benchmark for receptor (mg/kg-d).
The spatial characteristics of the HQ are defined by the home range for each mammalian and
avian receptor.5 Thus, the HQ values calculated for the same receptor assigned to two different
terrestrial habitats at a site are likely to reflect the differences in soil, plant, and prey
concentrations associated with the respective home ranges.
16.2.3 Process HQ Results for Decision Making
The HQs calculated by the Ecological Risk Module provide a risk metric with essentially
a binary outcome: an HQ above 1 indicates that ecological risks are above levels of concern, and
an HQ below 1 indicates that ecological risks are below levels of concern. Naturally, this
approach implies that an HQ much greater than 1 (say, two orders of magnitude) represents a
more serious ecological threat than an HQ of 2; however, the 3MRA modeling system and data
do not quantify the probability that an adverse effect will occur. For example, an HQ of 20
calculated for a drop in egg production for the osprey does not imply that the probability of this
effect is 10 times more likely than an HQ of 2, only that the magnitude of the effect is likely to
be more serious. Similarly, because the actual dose-response relationship is not used in the HQ
calculations, the HQ of 20 does not imply that the effect would be 10 times more severe than for
osprey with an HQ of 2. The HQ approach provides limited information on the probability and
significance of ecological risks for decision-making purposes; the implications of calculated
receptor-specific HQs as to the quality of ecological systems as a whole simply cannot be
inferred. The sometimes unpredictable nature of community dynamics, as well as the presence
of other stressors (e.g., habitat alteration), are such that the potential ecosystem effects associated
with, say, a reduction in reproductive fitness of a single receptor can be difficult to predict,
particularly when the modeling simulation may run hundreds or even thousands of years.
To address this limitation for national-scale applications of the 3MRA modeling system,
the Ecological Risk Module processes the calculated HQs so that decision makers can answer a
variety of questions about potential ecological risks. Specifically, the Ecological Risk Module
(1) assigns attributes appropriate to each HQ, including the hazard bin, distance to waste
management unit (WMU), receptor taxa, and habitat type, and (2) identifies the critical year in
the simulation at which ecological risks (i.e., HQs) are collectively at the highest level.
5 The Ecological Risk Module is also designed to calculate HQs for reptiles; however, insufficient
ecotoxicity data were identified to support an EB for reptile species.
16-15
-------
Section 16.0
Ecological Risk Module
Assigning Attributes to HQs. One of the most important attributes of the HQs
calculated by the Ecological Risk Module is the hazard range (referred to as a hazard "bin") to
which the HQ is assigned. The HQs for all receptors assigned to habitats within the AOI are
placed into one of five hazard bins identified by EPA as useful to decision-makers. The HQ bins
are defined as
¦ Bin 1: <0.1
¦ Bin 2: > 0.1 to ^ 1
¦ Bin 3: >1 to ^ 10
¦ Bin 4: >10 to ^ 100
¦ Bin 5: >100.
As an example, suppose that a particular site has a total of 25 mammalian receptor
species assigned across the three habitats: 10 mammals in a forest habitat, 7 mammals in a
grassland habitat, and 8 mammals in a stream margin habitat. For a simulation that runs three
years, assume that the HQs calculated for the mammalian receptors are as shown in Table 16-2.
These HQ calculations reflect the applied doses to mammalian wildlife due to the ingestion of
contaminated media, plants, and prey. Notice that the hazard profile changes with each year in
the simulation; in this hypothetical example, the level of exposure and resulting hazard is
increasing over time.
Table 16-2. Example HQ Counts for Mammals at Hypothetical Site
Number of M;imm:il Species in Ksicli HQ K:in<>c
liin I Bin 2 Bin 3 Bin 4 Bin 5
Kxposure sliirlin<> in... <0.1 >-0.l lo--l to -< 10 >IO(o-< 100 >-100
Year 1
19
5
1
0
0
Year 2
15
7
2
1
0
Year 3
12
9
1
2
1
The Ecological Risk Module uses the HQ counts to create a cumulative frequency
histogram: the results expressed as the cumulative percentile of the total number of HQs.
Therefore, the conditional, cumulative frequency percentiles for mammalian risks (defined by
the HQ) for each year in the simulation would be as shown in Table 16-3.
Table 16-3. Example Cumulative Frequency HQ Histogram for Mammals at Hypothetical Site
Percent of M:iiiini:il Species in this HQ k;in<>e or Lower
Kxposure sl;irlin<> in...
Bin 1
liin 2
liin 3
Kin 4
Kin 5
Year 1
76
96
100
Year 2
60
88
96
100
Year 3
48
84
88
96
100
16-16
-------
Section 16.0
Ecological Risk Module
Aggregating the HQs into hazard ranges serves two important functions in the 3MRA
modeling system. First, converting the results to cumulative hazard allows the module to greatly
reduce the amount of data that it produces. Because the Ecological Risk Module calculates HQs
for every receptor in every year in the simulation, the system produces a tremendous amount of
data. Second, the use of hazard ranges recognizes that the state of ecological risk assessment
science (for national applications) lacks the data to validate the level of accuracy in risk
estimates. Accumulating the ecological hazard quotients implicitly acknowledges the limitations
in predicting ecological risks using the HQ approach. EPA developed these bin ranges to be
meaningful in terms of decision making, increasing from below levels of concern to high levels
of concern.
In addition to the hazard bin, each HQ is associated with a series of attributes that allow
ecological risks to be characterized in different ways. EPA identified five attributes considered
relevant to interpreting the potential significance for adverse ecological effects within the
context of regulatory decision making. These attributes support analyses at different scales of
organization with respect to space as well as trophic position, and include:
¦ Habitat type (e.g., forest, grassland, pond, stream, permanently flooded forest),
¦ Habitat group (i.e., terrestrial, freshwater, and wetland),
¦ Receptor group (e.g., mammals, amphibians, soil community),
¦ Trophic level (i.e., producers, TL1, TL2, TL3, top predators), and
¦ Distance to WMU (e.g., within 1 km of WMU).
The maximum HQ across the site is also reported, along with its ecological risk attributes. This
metric was added for use in pass/fail analyses that may be needed to prioritize sites for further
refinement or assessment.
In summary, for each year in the simulation, the Ecological Risk Module does all of the
necessary accounting to construct cumulative frequency histograms, expressed as a percentile of
the total HQs across all receptors or by attribute. The cumulative frequency data contain the HQ
values for each attribute of interest (e.g., receptor group, habitat type) so that the results may be
accumulated to respond to different questions (e.g., what are the risks to mammals at a site). The
cumulative frequency data on HQs across all sites serves as the input data set for the Exit Level
Processor (ELP) of the 3MRA modeling system. As described in the overview in Section 1.0,
the ELP is a system level tool used to develop cumulative frequency distributions of ecological
risks for national analyses.
Identifying the Critical Year in the Simulation. Having produced the time series of
cumulative frequencies for HQs, the Ecological Risk Module identifies the year during which the
total hazard is a maximum for each of the ecological attributes listed above (i.e., habitat type,
habitat group, receptor group, and trophic level; all HQs are specific to a distance from the
WMU). This is the critical year in the simulation (referred to as TCrit); in processing ecological
risk results for national applications, the ELP uses only the HQ results for the TCrit year for each
site included in the simulation. The Ecological Risk Module identifies the TCrit year for each of
the first four ecological attributes listed above as a function of distance to the WMU. For
example, the module produces the cumulative frequency HQ histogram for all receptor groups
(e.g., mammals, birds, aquatic community) for the entire site, and for distance intervals within
16-17
-------
Section 16.0
Ecological Risk Module
1 km of the WMU, and between 1 and 2 km of the WMU.6 In addition, the module identifies the
TCrit for the maximum HQ across all receptors assigned to the site.
The Ecological Risk Module identifies the TCrit for each attribute in three basic steps.
First, the module calculates the total HQ (by attribute) for each year in the simulation by
summing the HQs across all receptors that share that attribute. Second, the module rank orders
the time series of total HQs. Third, the module selects the Ecological Regulatory Percentile
(ERP) and identifies the TCrit year that matches the total HQ at that percentile of the rank
ordering. The Ecological Regulatory Percentile is a user-specified percentile that indicates the
level of protection required to select TCrit. For example, if the user specified 95 percent as the
Ecological Regulatory Percentile, the Ecological Risk Module would identify TCrit as the output
year when the total HQ value is greater than 95 percent of all the other total HQ values reported
for the simulation. In the current implementation of the 3MRA modeling system, the default
value for Ecological Regulatory Percentile is 100 percent; therefore, the module identifies the
maximum total HQ and reports that year as TCrit for the appropriate attribute.
16.3 Module Discussion
16.3.1 Strengths and Advantages
The Ecological Risk Module calculates HQs for all receptors assigned to habitats within
the AOI. The module offers a number advantages, including the underlying ecological
benchmarks as well as site-specific adjustments to the benchmarks that reflect the environmental
characteristics of the site. Examples of these advantages include the following:
¦ The ecological benchmarks reflect the management goals and assessment
endpoints for national-scale analyses. The ecological benchmarks in the 3MRA
modeling system were developed specifically to evaluate the ecological risks
associated with long-term, low-level contaminant releases into the environment.
The relevance of these benchmarks to national-scale applications of the 3MRA
modeling system is a major advantage of this module. Although a variety of
different benchmarks could be used by the module to calculate ecological hazard,
the benchmark database ensures that the resulting HQs will be highly relevant to
establishing exit levels that are protective of the environment. The rationale,
confidence levels, and data quality objectives are transparent for ecological
benchmarks that are incorporated into the 3MRA modeling system benchmark
database.
¦ Benchmarks are adjusted for site-specific conditions as appropriate. For
certain benchmarks and chemicals, environmental characteristics exert a strong
influence on toxicity. For example, the water quality CSCL for certain metals is
sensitive to water hardness. For each simulation, the surface water CSCL is
6 Although the Ecological Risk Module reports hazard quotients for all three distance intervals, EPA has
determined that distance to WMU is of limited value in characterizing ecological risks. The vast majority of habitats
delineated for sites of interest overlap the 1 km distance from the WMU. As a result, it would be double counting to
report these risks for both the 1 km and 1-2 km distance intervals.
16-18
-------
Section 16.0
Ecological Risk Module
adjusted for the site-specific water hardness to better represent the potential
toxicity given the specific environmental setting. Similarly, the sediment CSCL
for organic chemicals is adjusted for the site-specific organic carbon fraction in
the sediment. The ability to modify benchmarks during a simulation allows the
module to be applied to virtually any site, and adjusts the toxicity according to the
site conditions.
¦ HQs retain attributes critical to characterizing the nature of ecological risk
estimates. Each HQ calculated by the module retains all of its attributes such as
the receptor group (e.g., mammals) and habitat type (e.g., forest). In addition, the
Ecological Risk Module reports the maximum HQ calculated at each site (along
with its attributes) to allow for a rapid screening function or coarse ranking of
sites. This information on HQ attributes is critical to allow decision makers to
evaluate the results using different kinds of ecological indicators. For example, it
may be desirable to evaluate concentration levels specifically for wetlands,
because of their ecological importance. Similarly, the user may want to
understand the nature of the risk estimates with regard to the receptors most likely
at risk, or the habitats most likely at risk. Consequently, the module reports this
information to retain flexibility in the characterization of predicted ecological
risks.
¦ Binning HQ greatly reduces the volume of data. Although the module
calculates HQs for each receptor at a site, these HQs are binned into discrete
ecological hazard bins. Because these bins are developed across each of the
possible attributes (e.g., trophic level, receptor group), the binning provides a
very efficient method of storing and manipulating these data without losing any of
the relevant information. This approach offers a very practical solution to the
potential storage issues associated with storing time series of HQs for dozens of
receptors across hundreds of sites. Moreover, the use of hazard bins is a very
appropriate technique to characterize ecological risk, given the current
state-of-the-science for national-scale analyses.
16.3.2 Uncertainty and Limitations
The methodology implemented by the Ecological Risk Module to characterize the
potential for adverse ecological effects carries certain assumptions and limitations, and
acknowledges several important sources of uncertainty. A number of the limitations and
uncertainties discussed in Section 15.3 on the Ecological Exposure Module also apply to the
Ecological Risk Module. Although parts of this section are redundant with the discussion in
Section 15.3, the primary focus of the following discussion is on the implications for risk
estimation and characterization for all ecological receptors, not just those receptors for which
ingestion exposures were calculated.
¦ The HQ estimates are based exclusively on an annual time step; that is, only
annual average applied doses and media concentrations are used in
calculating the HQ. The ecological HQ estimates are based on annual averages.
This time step represents much longer-than-lifetime exposures for some receptors
16-19
-------
Section 16.0
Ecological Risk Module
and substantially less-than-lifetime for other receptors. As indicated in Section
15.3, concentration spikes due to episodic events (e.g., rain storms) or elevated
WMU source releases following waste additions are not evaluated. Even though
the measures of effect often reflect critical life stage endpoints, the use of annual
average doses and concentrations to predict risk may not adequately reflect
effects associated with elevated, short-term exposures. This "one-time-step-fits-
all" approach, although necessary to address run time issues, does limit the
applicability of the HQ estimates for certain receptors.
The HQs are not calculated at the population or community level; ecological
risks must be inferred to higher levels of biological organization. Ecosystems
are enormously complex, and our understanding of even simple community
dynamics is limited. Data on chemical stressors are seldom available above the
level of an individual organism; that is, the study endpoints focus on individual
organisms rather than processes crucial to assemblages of organisms. Although
the measures of effect are relevant to higher levels of organization, the
comparison of point estimates of simulated contaminant doses or concentrations
for individual receptors with ecological benchmarks does not support the
quantification of population or community risks. Our ability to address
community-level effects associated with contaminant releases into the
environment is even more limited; the CSCLs developed to evaluate risks to
communities are derived by statistical inference based on toxicity data for
individual organisms. This limitation in the data and technology (i.e., the
inability to run population-level models in the 3MRA modeling system)
introduces significant uncertainty in the risk estimates.
The effects of multiple stressors (chemical and nonchemical) are not
considered in developing estimates of potential ecological risk. Given the
design goals for the 3MRA modeling system (i.e., to support national-level
assessment strategies of WMUs, waste streams, and contaminants), exposure to
multiple contaminants, contaminant releases outside of the AOI, and effects
associated with multiple stressors are not considered. In addition, background
concentrations of constituents were not considered in developing exposure
estimates, nor were other potential nonchemical stressors such as habitat
fragmentation. Data availability on the antagonistic and synergistic effects
associated with multiple stressors are extremely limited at this time (with the
possible exception of narcotic contaminants in aqueous systems), and prevented
the development of a multistressor analytical approach for the universe of
constituents in the current data set. Data limitations notwithstanding, the inability
to consider multiple stressors is a limitation in our ability to interpret the risk
results generated by this module. This is a significant but unquantifiable
uncertainty inherent in the HQ calculations.
The HQ estimates reflect different endpoints at varying levels of effect. The
ecological benchmarks address a variety of receptors (e.g., soil fauna, mammals,
plants) and, because the quality and quantity of relevant data vary widely across
receptors, the HQs generated by the Ecological Risk Module represent different
16-20
-------
Section 16.0
Ecological Risk Module
levels of knowledge regarding the exposure and toxicity of chemical stressors.
The variability in supporting data suggests that the level of confidence in the
exemption criteria is dependent on the quantity and quality of available data. The
HQ methodology—the ratio of an exposure to a benchmark—is applied uniformly
across all ecological receptors. However, the data supporting the HQ calculation
vary in that they include endpoints from lethality to reproductive fitness and
address population- and community-level effects by inference. To some extent,
the HQ estimates for different receptor groups represent different risk metrics.
The interpretation of these HQ estimates is, therefore, limited by our
understanding of the potential ecological significance of the measures of effect as
well as overall confidence in the data used to support the calculations.
¦ The data used to support the development of sediment CSCLs for metals
have come under significant scrutiny because of concerns regarding the
applicability of the tests to predict sediment effects. The sediment CSCLs for
metals are ultimately based on data from the National Oceanic and Atmospheric
Administration's (NOAA) National Status and Trends (NS&T) data. The NOAA
data generally rely on observed effects to sediment biota exposed to sediment
samples. The sediment samples are contaminated with mixtures of metals and, as
a result, the observed effects cannot be definitively attributed to a specific metal.
Thus, the effects levels reported for specific metals may in fact reflect effects of a
mixture of metals. HQs based on these sediment CSCLs may overestimate the
potential for adverse effects to the sediment community.
¦ The HQs calculated for the aquatic and benthic communities are resolved at
the habitat, rather than reach level. There is some uncertainty associated with
calculating risks to aquatic life across an entire habitat (as defined within the
study area). Species of fish such as brown trout tend to use certain segments of
stream habitats; therefore, HQs at the reach level may be more appropriate.
Conversely, establishing artificial boundaries between stream reaches is contrary
to the goals of the assessment strategy, namely, to evaluate ecological risks using
the habitat as the fundamental unit. In short, the spatial scale adopted for
evaluating freshwater communities of water column and sediment biota
introduces uncertainty in HQ for these receptors. Risks may be underestimated
for stream reaches with elevated contaminant concentrations; conversely, risks
may be overestimated for a stream habitat with only a single stream reach above
levels of concern.
16.4 References
Efroymson, R.A., M.E. Will, G.W. Suter, and A.C. Wooten. 1997a. ToxicologicalBenchmarks
for Screening Contaminants of Potential Concern for Effects on Terrestrial Plants: 1997
Revision. ES/ER/TM-85/R3. Oak Ridge National Laboratory, Oak Ridge, TN.
Eijsackers, J. 1994. Ecotoxicology of Soil Organisms: Seeking the Way in a Pitch Dark
Labyrinth. Ecotoxicology of Soil Organisms. CRC Press, Inc., p. 3.
16-21
-------
Section 16.0
Ecological Risk Module
Freda, Joseph. 1991. The effects of aluminum and other metals on amphibians. Environmental
Pollution. Volume 71, pp 305-328.
Jones, D.S., G.W. Suter, II, and R.N. Hull. 1997. Toxicological Benchmarks for Screening
Contaminants of Potential Concern for Effects on Sediment-Associated Biota: 1997
Revision. ES/ER/TM-95/R4. Oak Ridge National Laboratory, Oak Ridge, TN.
Klaine, S.J., and M.A. Lewis. 1995. Algal and Plant Toxicity Testing. In: Hoffman, D.J. (ed.)
Handbook of Ecotoxicology. Lewi s Publi shers.
Mineau, P., B.T. Collins, and A. Baril. 1996. On the use of scaling factors to improve
interspecies extrapolation of acute toxicity in birds. Regul. Toxicol, and Pharmacol.
24:24-29.
Power, T., K.L. Clark, A. Harfenist, and D.B. Peakall. 1989. A Review and Evaluation of the
Amphibian Toxicological Literature. Technical Report Series No. 61. Canadian Wildlife
Service, Environment Canada, Hull, Quebec.
Sample, B. E., G.W. Suter II, R.A. Efroymson, and D.S. Jones. 1998. A Guide to the ORNL
Ecotoxicological Screening Benchmarks: Background, Development, and Application.
Revision 1.0. Oak Ridge National Laboratory, Oak Ridge, TN., ORNL/TM-13615
Smith, S.L., D.D. MacDonald, K.A. Keenleyside, C.G. Ingersoll, and L.J. Field. 1996. A
preliminary evaluation of sediment quality assessment values for freshwater ecosystems.
J. Great Lakes Res. 22(3):624-638.
Stephan, C.E., D.I. Mount, D.J. Hansen, J.H. Gentile, G.A. Chapman, and W.A. Brungs. 1985.
Guidelines for Deriving Numerical National Water Quality Criteria for the Protection of
Aquatic Organisms and Their Uses. PB85-227049. National Technical Information
Service, Springfield, VA.
Suter, G.W. II, and C.L. Tsao. 1996. Toxicological Benchmarks for Screening Potential
Contaminants of Concern for Effects on Aquatic Biota: 1996 Revision. ES/ER/TM-
96/R2. Prepared for the U.S. Department of Energy, Washington, DC.
U.S. EPA (Environmental Protection Agency). 1992. Draft Report: A Cross-Species Scaling
Factor for Carcinogen Risk Assessment Based on Equivalence of mg/kg3/4/day. Federal
Register 57 FR 24152, June 5, 1992.
U.S. EPA (Environmental Protection Agency). 1996. Amphibian Toxicity Data for Water
Quality Criteria Chemicals. EPA/600/R-96/124. National Health and Environmental
Effects Research Laboratory, Corvallis, OR.
U.S. EPA (Environmental Protection Agency). 1998. Guidelines for Ecological Risk
Assessment. EPA/630/R-95/002F. Risk Assessment Forum. Washington, DC. April.
16-22
-------
Section 16.0 Ecological Risk Module
U.S. EPA (Environmental Protection Agency). 2000 Background Document for the Ecological
Exposure and Ecological Risk Modules for the Multimedia, Multipathway, Multireceptor
Risk Assessment (3MRA) Software System. Office of Solid Waste, Washington, DC.
August.
16-23
-------
Section 16.0 Ecological Risk Module
16-24
------- |