DO NOT QUOTE OR CITE
EPA/600/AP-93/003
November 1993
External Review Draft
ADDENDUM TO THE METHODOLOGY FOR ASSESSING HEALTH RISKS
ASSOCIATED WITH INDIRECT EXPOSURE TO COMBUSTOR EMISSIONS
NOTICE
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
Exposure Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, D.C.
Printed on Recycled Paper
-------
DISCLAIMER
This document is a review draft and does not constitute U.S. Environmental
Protection Agency policy. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
November 10, 1993
-------
CONTENTS
Tables jv
Figures v
Foreword vi
Preface ; V|-j
Authors vjjj
Reviewers -....- ix
I. Introduction . l_1
II. Indirect Exposure Document Discussion by Chapter
2. Human Exposure Scenarios . 2-1
3. Air Dispersion and Deposition Modeling of Pollutant Stack Emissions ... 3-1
4. Calculating Soil Concentrations 4-1
5. Determining Exposure through the Food Chain 5-1
6. Determining Exposure from Soil Ingestion 6-1
7. Determining Exposure from Dermal Absorption via Soil 7-1
8. Dust Resuspension . . 8-1
9. Calculating Water Concentration 9-1
10. Determining Exposure from Water Ingestion 10-1
11. Determining Exposure from Fish Intake 11-1
12. Determining Exposure from Dermal Absorption from Water 12-1
13. Hazard Identification 13-1
14. Dose-Response Assessment 14-1
15. Risk Characterization . 15-1
III. Recommendations for Long Term Improvement of Multimedia Risk Modeling 111-1
IV. References R_1
Appendix A A_1
in
November 10, 1993
-------
TABLES
2-1. Example table for media levels 2-7
2-2. Example exposure parameters for an isopleth ring within the
study area 2-9
3-1. Initial list of dioxin-like compounds identified in emissions
from the incineration and combustion of anthopogenic waste .... 3-3
3-2. Initial list of organic compounds identified in emissions
from the incineration and combustion of anthropogenic wastes
(key follows) • - - - 3-4
3-3. Initial list of inorganic contaminants identified in emissions
from the incineration and combustion of anthropogenic wastes
(key follows) 3-6
3-4. Examples of precipitation scavenging coefficients (per s) in
COMPDEP 3-27
3-5. Parameters in the COMPDEP control files 3-28
3-6. Parameters in the meteorological data file for COMPDEP 3-29
3-7. Generalized particle size distribution (/vm), and proportion of
available surface area, to be used as a default in deposition
modeling if site-specific data is unavailable 3-38
11-1. Fish consumption estimates from USDA 1977-78 National
Food Consumption Survey (consumptions were recorded
for three periods; N = 36249; units = grams/day/person) 11-4
IV
November 10, 1993
-------
2-1.
9-1.
FIGURES
Simple representation of outer isopleth ring defining study
area and isopleth rings within study area
Basic structure of the water algorithm showing concentrations
considered, relationships between concentrations, and routes
of fate and dissipation .
9-2. Steady state representation for sediments in water bodies
2-5
9-4
9-10
November 10, 1993
-------
FOREWORD
The Exposure Assessment Group (EAG) within the Office of Health and
Environmental Assessment of EPA's Office of Research and Development has three main
functions: (1) to conduct exposure assessments, (2) to review assessments and related
documents, and (3) to develop guidelines for exposure assessments. The activities under
each of these functions are supported by and respond to the needs of the various EPA
program offices. In relation to the third function, EAG sponsors projects aimed at
developing or refining techniques used in exposure assessments.
In response to increasing concerns over the potential impacts of emissions from
combustion devices through indirect exposure routes, the Agency formed a
multfdisciplinary working group to review and update Agency guidance on conducting
indirect exposure assessments. The Workgroup identified the 1990 EPA report,
Methodology for Assessing Health Risks Associated with Indirect Exposure to Combustor
Emissions (EPA/600/6-90/003, January, 1990) as the best existing guidance in this area.
The Workgroup decided that this document is still generally appropriate for current needs,
although some updates would be needed. These updates have been compiled in this
Addendum to the 1990 document. Hopefully, this effort will assist assessors evaluate
combustor emissions and ensure that the most current tools and information are used.
Michael A. Callahan
Director
Exposure Assessment Group
VI
November 10, 1993
-------
PREFACE
In response to increasing concerns over the potential impacts of emissions from
combustion devices through indirect exposure routes, the Agency formed a
multidisciplinary working group to review and update Agency guidance on conducting
indirect exposure assessments. The most complete Agency document on this topic is the
1990 EPA report. Methodology for Assessing Health Risks Associated with Indirect
Exposure to Combustor Emissions (EPA/600/6-90/003, January, 1990). This document
was initially drafted in 1986 and reviewed by the Agency's Science Advisory Board in
1987. The 1990 document incorporated comments of the SAB as well as other additional
reviewers. The workgroup decided that this document is still fundamentally sound and
appropriate for current needs, although changes would be needed.
The working group was divided into three sub-groups to deal with specific technical
issues. The first subgroup focussed on modeling emissions, air dispersion, and deposition.
Another subgroup reviewed the food chain modeling, including fate and transport modeling
methodology and input parameters. The final group focussed on the remaining issues of
human exposure and contact rates.
The approach used by the working group was to review the 1990 report and to
identify and resolve issues. This addendum is formatted following this approach. After
each "Issue" is presented, this addendum then describes the workgroup's
"Conclusions/Recommendations" concerning that issue. The workgroup did, however,
decide that some sections required more attention than just issue identification and
resolution. Two sections replace the Indirect Exposure Document material currently in
Chapter 3, Air Dispersion and Deposition of Emitted Pollutants, and Section 9.2. Surface
Water, of Chapter 9. This addendum has undergone several revisions and has been
reviewed by the workgroup as well as other individuals in EPA.
As a final note,.it should be emphasized that this document was crafted as an
Addendum to the 1990 EPA report Methodology for Assessing Health Risks Associated
with Indirect Exposure to Combustor Emissions, and, as such, should be used in
combination with that report in conducting indirect exposure assessments.
VII
November 10, 1993
-------
AUTHORS
This document was developed by the Indirect Exposure Assessment Workgroup,
cochaired by Alec McBride of the Office of Solid Waste and John Schaum of the Office of
Research and Development. The principal authors are listed below:
Office of Research and Development:
Robert Ambrose
Michael Callahan
David Cleverly
Matthew Lorber
Glen Rice
John Schaum
Donna Schewede
Office of Solid Waste:
David Layland
Alec McBride
Office of Air Quality Planning and Standards:
Joe Touma
VIII
November 10, 1993
-------
REVIEWERS
This document went through a series of reviews by the Indirect Exposure
Assessment Workgroup (see below) as part of the development process. The progress of
this Work Group was monitored by an Office Director's Oversight Group consisting of
Peter Preuss (OSPRE), Sylvia Lowrance (OSW), Mark Greenwood (OPPT), John Seitz
(OAQPS), Jeffrey Dennit (OSW), and William Farland (OHEA). In addition, the final draft
was circulated for Agency-wide internal review.
INDIRECT EXPOSURE ASSESSMENT WORKGROUP
Air Modeling and Emissions
Food Chain/Fate and
Transport
Human Exposure and
Contact Rates
Joe Touma, CHAIR,
OAQPS
Glen Rice, ORD
Diane Byrne, OAQPS
Gary Victorine, Region 5
Dave Guinnep, OAQPS
Pam Blakely, Region 5
Fred Talcott, OPPE
Bob Holloway, OSW
Dave Cleverly, ORD
Sonya Sasseville, OSW
Stan Durkee,ORD
Bill Petersen, ORD
Dona Schewede, ORD
David Layland, OSW
Matt Lorber, CHAIR, ORD
Chuck Maurice, Region 5
Carol Braverman, Region 5
Rick Mattick, OPPTS
Glen Rice, ORD
Bob Ambrose, ORD
Mark Mercer, OSW
Norma Whetzel, OW
Tom Murray, OPPTS
Alec McBride, OSW
John Schaum, CHAIR,
ORD
Chuck Maurice, Region 5
Carol Braverman, Region 5
Rick Mattick, OPPTS
Glen Rice, ORD
Larry Burnes, ORD
Steve Kroner, OSW
Fred Talcott, OPPE
Dave Cleverly, ORD
Stan Durkee, ORD
Dave Jaquith, OPPTS
Don Rodier, OPPTS
Al Rubin, OW
Warren Peters, OAQPS
Dorothy Canter, OSW
Christine Dibble, OSW
ORD: Office of Research and Development
OSW: Office of Solid Waste
OAQPS: Office of Air Quality, Planning and Standards
OPPTS: Office of Prevention, Pesticides and Toxic Substances
OPPE: Office of Policy, Planning and Evaluation
OW: Office of Water
OSPRE: Office of Science, Planning and Regulatory Evaluation
OHEA: Office of Health and Environmental Assessment
OPPT: Office of Pollution Prevention an Toxics
IX
November 10, 1993
-------
-------
DRAFT - DO IMOT QUOTE OR CITE
I. Introduction
As discussed in the Preface, this "Addendum" represents the deliberations of a
working group established to review Agency guidance on conducting indirect exposure
assessments. The workgroup decided that the document Methodology for Assessing
Health Risks Associated with Indirect Exposure to Combustor Emissions (EPA, 1990a;
available through the Center for Environmental Research Information at 513-569-7562;
this document will hereafter be referred to as the Indirect Exposure Document) is largely
appropriate for current needs, although updates are needed in some areas. The purpose of
this Addendum is to detail those changes.
The most significant modifications that are being recommended to the 1990 report
are in the following areas:
• Air dispersion and deposition modeling. The workgroup recommends that Chapter
3 of the Indirect Exposure Document, Air Dispersion and Deposition of Emitted
Pollutants, be replaced with a revised Chapter 3 given in this Addendum. This
chapter describes the use of an updated version of the COMPDEP model.
• Surface water impacts. The workgroup recommends that Section 9.2 of Chapter 9,
titled Surface Water, be replaced with a revised Section 9.2 given in this
Addendum. This revised section describes a single framework for evaluating
surface water impacts to replace the three-tier approach given in the Indirect
Exposure Document.
• Incorporation of Agency risk assessment policies. This Addendum includes the
policies described in the February 26, 1992, Deputy Administrator memorandum
"Guidance on Risk Characterization for Risk Managers and Risk Assessors"
(Habicht, 1992) to identify high end and central tendency exposures be followed in
conducting indirect exposure assessments.
1-1
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
In addition to the major changes noted above, there are other recommended
changes, some of which may have significant effects on the exposure assessment results
for different constituents.
1.1. Use of the Addendum as a Screening Tool
The workgroup recognized the need for a "screening" level methodology so that
individuals assessing stack emissions can get a rapid evaluation which can simply answer
the question as to whether or not a more detailed site-specific assessment is warranted.
For example, if an incinerator is expected to be operating for only a number of months
rather than years, it may be possible that indirect impacts are negligible. Similarly, if there
are no farm sites (where animals are raised for food, e.g.) for tens to hundreds of miles
from the incinerator, it may also be possible that farm-related indirect exposures are
negligible. A screening tool could begin to more clearly define the focus of a site-specific
indirect exposure assessment.
The 1990 Indirect Exposure Document as well as this addendum were not
developed for purposes of this type of simple screening. The workgroup does recommend
that efforts should be undertaken to develop screening tools.
This addendum does, however, provide some generalized guidance on limiting the
scope of an assessment. Section 3.2 in the revised Chapter 3 of this addendum presents
a comprehensive list of contaminants to consider for indirect impacts, and then provides
generalized guidance on narrowing this list by considering, for example, feedstock
materials, emissions from surrogate incinerators, availability of health-related parameters,
estimated emission rates, potential for bioaccumulation, and so on.
While this guidance for contaminant selection can assist in narrowing the scope of
an assessment, a second objective for a screening level tool is to reduce the complexity of
the modeling procedures. Some additional thoughts on how to reduce this complexity
include:
1) Use simplified atmospheric transport models to determine an air concentrations and
depositions at a site where indirect exposures could occur, such as a farm.
I-2
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
2) Apply concentrations and depositions to a "screening scenario". This scenario could
contain all the fate, transport, and transfer algorithms that are in this Addendum and the
1990 Indirect Exposure Document. It one were to use all the algorithms in these two
documents, one would have to carefully consider all parameter selections. A host of
parameters are involved in estimating exposure media concentrations, and a second group
considers patterns of human exposure. Depending on the intended user or audience of
this "screening scenario", all parameter assignments could adhere to a "high end" or
"bounding" objective. If the final estimated risk passes one or more "screens", then there
may be technical grounds to conclude that further evaluations pertinent to those screens
are unnecessary.
1.2. Implementation of the Guidance
This guidance focusses on providing appropriate site-specific procedures for
conducting indirect exposure assessments of emissions from combustion sources. The
material also includes some information on individual parameter assumptions and other
specific factors to consider. However, a number of the assumptions needed to complete
an indirect exposure assessment may vary depending on the specific application or
program involved. Therefore, decisions are still required in applying this guidance in
different situations.
Also, given the long chain of calculations and accompanying parameter
assumptions needed to evaluate some of the indirect exposure routes, it is very important
that the exposure assessment include a detailed description of all of the uncertainties
involved and how they may affect the final results.
1.3. Future Guidance Documentation
The workgroup also discussed long term goals pertaining to guidance for indirect
exposure assessments. One is to update the 1990 IED with the information in this
addendum to produce a single methodology document. This Addendum does not update
the example calculations presented in the 1990 IED for cadmium and benzo-a-pyrene, and
this updating would occur for the next full version of the IED. A second is to produce an
I-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
accompanying parameter guidance document. While the 1990 IED and this Addendum do
contain a substantial amount of information on parameter assignment, the workgroup still
recognizes a need for additional guidance, particularly for chemical-specific
parameterization. Finally, the workgroup recognizes the need for a computer software tool
to accompany the atmospheric modeling tools, such as the COMPDEP model. Key aspects
of such a software tool are discussed in Part III of this Addendum. A parameter guidance
document would be the companion to such a software tool.
1.4. Addendum Format
The remainder of this Addendum is structured as follows. The section following
this introduction is titled, "II. Indirect Exposure Document Discussions by Chapter". This
section reviews each of the 15 chapters of the 1990 Indirect Exposure Document,
providing replacements for Chapters 3 and Section 9.2 of Chapter 9. Other than these
replacements, the section headings were numbered to exactly correspond to the 1990
document. For example, "2.4 Defining the Exposed Individual" below discusses issues
with material in Section 2.4. Defining the Exposed Individual in the 1990 Indirect Exposure
Document. The title of the third major section of this Addendum describes its purpose,
"III. Recommendations for Long-Term Improvment of Multimedia Risk Modeling".
Appendix A derives the calculation for estimating population risks for food consumption
pathways.
I-4
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
II. Indirect Exposure Document Discussions by Chapter
2. HUMAN EXPOSURE SCENARIOS
[Note: The following issue and associated conclusions/recommendations apply broadly to
all Sections of Chapter 2. ]
Issue Since the Indirect Exposure Document was published, the Agency has issued new
Exposure Guidelines (EPA, 1992a) which establish a new set of definitions and
development strategy, for exposure scenarios. Therefore, updates are needed to
achieve consistency with these Guidelines.
Conclusions/Recommendations
General Approach
A risk assessment should provide several different descriptors of risk, including
both individual and population risks. Individual risks should be presented as both central
tendency and high end estimates (see EPA, 1992a,1992c) for the population affected by
the combustor. In addition, the number of people at various exposure/risk levels should be
presented. Population risks should be expressed in several ways, also. For carcinogens,
risks may be presented as the potential number of health effect cases in the affected
population (either yearly, or over a longer period such as the projected life of the
combustor). For non-carcinogens, long-term population risk can be presented as a number
of persons whose exposure or dose exceeds the RfD, or number of persons whose
exposure might exceed an acute effect threshold (if known). For both carcinogens and
non-carcinogens, specific subgroups of the population should be highlighted, either
because they are of interest (e.g., schoolchildren), or because of potentially high
exposure/risk (e.g., subsistence farmers).
Defining the Population
Because Agency guidance on exposure/risk assessment calls for characterization of
"high end" individual exposure/risk (EPA, 1992a), it is critical that the population be
2-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR GITE
defined appropriately. Since the "high end" is defined as roughly the top ten percent of
the population in terms of exposure or risk, inclusion of large numbers of non-exposed
persons in the population can skew the high-end (and mean) value downward, while
ignoring lesser-exposed persons can have the opposite effect.
On the other hand, it is very important that the population be defined in such a way
that it is clear whether any given individual is "in" or "out" of the population being
considered. Defining the population as "anyone exposed to any chemicals from the
combustor" is not particularly useful, since for any given individual, it is unclear whether
they are "in" or "out," and therefore, it is very difficult to arrive at a population size.
Geographical boundaries and activity-related criteria can serve a useful purpose here. If
the population is defined as "all those who live or work within [some specific] boundaries,"
these are relatively straightforward criteria that can be used to define the size of the
population being evaluated.
One option for setting boundaries for the study area is to set a fixed distance from
the combustor, such as 20 km or 50 km, and draw a circle around the combustor with the
specified radius. This is not the optimum method for defining the population, however.
Although it may be quick and easy for a mathematical model to handle (an advantage), it is
crude. Not only can this procedure result in inclusion of many unexposed or negligibly
exposed persons in the study population (or exclude relevant population), but it can result
in a great deal of unnecessary work. When detailed land-use and population-activity data
are collected, it is important to avoid collecting information for wide geographical areas
where little or no exposure occurs.
A second option for setting boundaries is to focus on an area for which individual
risks may be above some specified level of concern. This has the advantage of excluding
persons outside the boundary from the study (with some important exceptions noted
below). This may save considerable time as the risk assessment is being done, since data
collected for land use, personal activities, etc. inside the boundary would all be important
to the assessment. A potential disadvantage of this method is that the "study area" may
be misinterpreted as the "area at risk." Since the boundary is not circular
(i.e., geographically arbitrary), but corresponds to a site-specific concentration/deposition
2-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
isopleth from a model, the meaning of the boundary requires explanation. The area
circumscribed by the boundary is the area for which further study will be done concerning
those who live and work, and not a "high risk zone." Although this option may involve
some additional "upfront" work to set up the analysis (relative to the first option), it should
save work later, and result in a more useful risk assessment. When defining the "study
area", one must be aware that different constituents and/or exposure pathways could
result in different "study areas". A problem which may result with this approach, is that
each constituent/pathway could result in a different "study area."
It is generally best to define the population potentially affected by the combustor
rather than those in an arbitrary circle. Defining the study population as those potentially
affected by the combustor can be done by looking at the persons living and working within
the isopleth boundary, and then adding other persons outside the boundary who are
potentially affected by water-related and other exposures.
Determining the Boundary of the Study Area for Individuals Potentially Exposed as a Result
of Deposition/Dispersion
Until more experience with indirect exposure modeling is gained, a trial and error
approach is recommended for defining the boundaries of the study area. The first step in
this approach is to run the dispersion/deposition models out to a distance of 20 to 50 km.
Then examine the risks at the outer isopleth based on the pathway and activities thought
to pose the highest potential for risk. If the risk exceeds a level of concern, then extend
the modeling out further and recheck the risks. (Note: Extending the study area beyond
50 km may require the application of long-range atmospheric transport models.)
Eventually the isopleth corresponding to the specified level of concern will be determined
and this will define the study area. The final decision on how to most efficiently conduct
this trial and error procedure is probably best determined by the modeler who can evaluate
the costs/time of running the model for various amounts of output. If risks in the outer
ring are estimated to be above the concern level during the subsequent detailed analysis
(considering all chemicals and pathways), the study area will need to be expanded.
2-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Establish Isopleth Rings Within the Outer Boundary Isopleth Ring
Several rings or just a few additional isopleth rings can now be established within
the outer boundary isopleth ring. Further analysis (as will be described below) now can
focus on activities and individuals between isopleth rings. The width between these rings
around the combustor stack depends on several things. First, the difference in
concentration/deposition values between isopleths should be meaningful. Second, the
ability of the computer software to quickly and conveniently draw isopleths may limit the
number of isopleths mapped. In general, it may be helpful to construct isopleths at a
somewhat more detailed level than necessary for the ensuing exposure analysis.
Figure 2-1 is a simple representation the outer boundary of a hypothetical study
area and the isopleth rings within the study area. The individuals who lie within the study
area as defined by the outer isopleth ring form the basic segment of the population to be
evaluated, and will generally be evaluated for all appropriate pathways. Certain individuals
who live outside the study area boundary should be added. These are discussed below.
Additional Individuals Potentially Exposed Through Intermittent Presence Within the Study
Area Boundary
Individuals who work, attend school, or otherwise spend significant amounts of
time within the study area boundary, even though they do not live within the boundary,
should also be evaluated and included in the basic population studied. These include
individuals potentially exposed by water-related routes and individuals potentially exposed
through food ingestion.
Defining the Population for Study
In summary, addition of the various segments noted above will result in a
population "potentially affected by the combustor" (hereafter called the study population).
The study population consists of:
• persons living, working, or otherwise spending significant amounts of time
2-4
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Outer study
boundary
Isopleth ring of equal
/ J concentration/deposition
/ according to atmospheric
transport modeling
Figure 2-1. Simple representation of outer isopleth ring defining
study area and isopleth rings within study area.
within the "study area boundary" defined through dispersion/deposition
modeling;
• persons potentially exposed to non-negligible amounts of chemicals (emitted
from the combustor) by water-related means; and
• persons potentially exposed to foodstuffs contaminated by "the combustor
emissions with non-negligible concentrations.
Developing Distributions of Individual Risk
The goal of this step is to assign an exposure (and risk) to the individuals within the
2-5
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
study population. This is done by considering exposure concentrations and activities. By
the combination of concentration and activities, along with other appropriate dose-
response or toxicological information, individual risk estimates can be made. The full
procedure described below can become quite complex and resource intensive. Therefore,
the assessor should attempt to reduce the scope of this effort in terms of chemicals and
pathways. One way to do this is to make bounding type assessments using maximum
exposure parameters applied at points of maximum concentrations or depositions. If these
bounding assessments demonstrate risks below a concern level, then the associated
chemicals and pathways can be safely eliminated from further study.
One can start in this process of assigning risk estimates by using isopleths of
constant concentration/deposition, and differentiating among activities (e.g., starting with
one isopleth and grouping the individuals by activity, such as farmers, residents with home
gardens, residents without home gardens, non-resident workers, etc.). The following list
of suggested steps in assigning risk estimates assumes constant concentration/deposition
isopleths will be used as a basis (if emission rates and resulting concentration/deposition
isopleths are likely to be changing over time, then these steps can be applied to time
periods during which the emissions are relatively constant and summed to get the total
exposure}:
1. First, using the methodology presented in this document, compute
contaminant levels in media of concern ( e.g., air, water, soil, groundwater,
beef, fruit/vegetables, milk, etc) for each ring. This information can be
conveniently summarized on a table such as Table 2-1, or kept in a
spreadsheet. To the extent possible the model inputs should be based on
local information. Considerable uncertainty is likely to be associated with
these estimates. See Chapter 15 for further information on how to assess
this uncertainty.
2. Next, overlay isopleths onto aerial photographs, population maps, land use
maps, etc. to identify the types of activities and numbers of people in each
2-6
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 2-1. Example table for media levels.
Air (//g/m 3)
Soil (jjg/g)
Surface water
(mg/L)
Beef (mg/kg)
Milk (mg/L)
Fish (mg/kg)
Vegetables
(mg/kg)
Others
Ring 1
Ring 2
Ring 3
Ring 4 Ring 5
ring. Several PC based models are available for generating populationdensity
maps based on Census data including the Human Exposure Model from the
Office of Air Quality Planning and Standards and the Geographic Exposure
Modeling System from the Office of Prevention, Pesticides and Toxic
Substances. Geographic Information Systems (GIS) could be used to
identify land use, complex terrain, and population centers. Some of these
systems use old Census data and it would be more accurate to use the
TIGER files from the most recent census. Satellite photographs may provide
the most recent information. Local surveys may also help locate farms and
home gardeners.
2-7
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
For each isopleth ring, estimate the number of farmers, residents with home
gardens, other residents, and other groups defined in the study population
(such as intermittently-exposed non-resident workers). Also, for each ring
locate schools, fishing areas and other areas where unusual exposure may
occur. Also estimate the number of school children (and where they go to
school), and the number of people in any other special interest group like
subsistence fishermen, developments for retired people, etc. This
information could also be kept in a spreadsheet.
For each isopleth ring and population group, determine which exposure
pathways apply. Characterize behavior parameters for each pathway,
e.g., ingestion rate, exposure duration, etc. This information can be
summarized in a table such as Table 2-2, or stored in a spreadsheet. The
level of detail used in characterizing behavior is dictated by the amount of
data which is available or could be collected. Theoretically, each person in
the ring could be surveyed and behavior identified, but this is clearly
impractical in most situations due to the size of the population and resources
available. However, even a limited (statistically-based) local survey can be
quite informative and is highly recommended. Some options in order of
increasing detail level are offered below for how to approach this problem:
a. Select the parameter values that represent best judgement of central
values for current conditions of the groups identified above.
b. Divide each group into a low, medium or high exposure level and
select the parameter values for each. The number of people in each
subgroup should also be identified.
c. Assume statistical distributions for the parameter values in each
group.
The next step is to compute exposures and compile results. The results can
2-8
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Table 2-2. Example exposure parameters for an isopleth ring within the study area.
Population size
Residence time (yr)
Soil ingestion:
RATE (mg/d)
FREQ (d/yr)
Inhalation:
RATE (m 3/d)
FREQ (hr/d)
Water ingestion:
RATE (Lid)
FREQ (d/yr)
Beef Ingestion:
RATE (g/d)
FREQ (d/yr)
percent local
Milk Ingestion:
RATE (g/d)
FREQ (d/yr)
percent local
Fish Ingestion:
RATE (g/d)
FREQ (d/yr)
percent local
Vegetable ingestion:
RATE (g/d)
FREQ (d/yr)
percent local
Gardeners
Other Residents
Farmers
School Children
2-9
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
be displayed in a variety of ways. Cumulative distributions (exposure or risk
vs number of people) are useful for identifying central and high end exposure
levels. A frequency distribution showing number of people exposed at
various exposures or risk levels is a particularly effective way to
communicate the results of the analysis. These distributions can be based
on the entire study population or individual special interest groups such as
farmers, gardeners, school children, etc. A variety of options are available
for creating these distributions. If exposure parameters can be expressed as
distributions, then a Monte Carlo simulation is probably the best approach.
This approach would involve using a frequency histogram of number of
people in each ring (based on information generated in Steps 2 and 3) to
select a location and associated media concentrations for each iteration. It
is important to separate uncertainty and variability in the assessment and to
clearly show support for all distributions selected. Further guidance on these
issues and other aspects about how to conduct Monte Carlo analysis is
provided in Chapter 15 of this document. If it is not feasible to conduct a
Monte Carlo analysis, a series of point estimates can be used to
approximate the distribution. As discussed in Step 4, local surveys could be
used to characterize consumption rates of local grown foods, frequency and
duration. For example if the surveys support generation of a low, medium
and high consumption rate for farmers and the isopleth construction
generates 5 rings, then 15 point estimates of exposure could be made and
plotted vs number of farmers. This yields an approximation of the
distribution. Since three consumption rates do not represent the full range
of variability, upper percentiles will be underestimated and lower percentiles
will be overestimated to some degree. Judgement will need to be decide if
this uncertainty is acceptable.
6. In addition to presenting the distributions of exposure, the
assessor should highlight individual exposure/risk estimates
2-10
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
for several locations {i.e. maximum deposition, place where
most people exposed, schools, etc.) and subpopulations of
interest (i.e. subsistence farmers, recreational and subsistence
anglers, nursing infants, ethnic groups or any other population
segment with unusually high exposures). These scenarios
could be charaterized using a central or range of exposure
values combined with an estimate of the number of people in
the group.
This effort should first focus on characterizing behavior describing exposure that
occurs at a person's residence. A second step is to consider exposure that occurs outside
his residence. Certain activities may involve exposure from a ring different than his
residence. For example, a child may attend school in a different ring or an adult may eat
food grown in a different ring or fish in a different ring. In later refinements, evaluation of
how these parameter values and group sizes may change in future may be desired.
Note: Issues associated with characterizing population risk are discussed in
association with Chapter 15, Risk Characterization.
2.3 Defining the Length of Time of Emissions
Issue There are no procedures suggested for addressing short term exposures from upset
or intermittent emissions.
Conclusions/Recommendations
Occassional short-term releases are generally not an issue for indirect exposure
analyses, since long-term buildup in the food chain is the principal concern for indirect
exposure. The accumulation of short-term releases can be averaged and added to the
continuous emissions if data are available to do so. However, short-term releases should
be evaluated for potential acute effects as part of the direct exposure assessment. The
highest inhalation exposures are associated with periods of highest air concentrations.
Inversions or other unusual meteorologic conditions causing stable air minimize dispersion
and can lead to unusually high air concentrations. Increases in emission rates also
2-11
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
increase air concentrations. As discussed in Chapter 3, the air model can estimate the
highest daily air concentration expected to occur during a year. The model output is
normalized by emission rate. So the highest expected emission rate can be combined with
daily high model output to estimate the highest short term exposure level. This short
term exposure level can be compared with acute exposure standards.
2.4 Defining the Exposed Individual
Issue The Indirect Exposure Document suggests that impacts from combustors are
negligible outside a 50 km radius. This is somewhat arbitrary and should be
considered on a site-specific basis.
Conclusions/Recommendations
As previously discussed, the study area should be defined by calculating risks from
the most important exposure route out to a distance where those risks are no longer of
concern. The beef pathway could be the most sensitive for dioxin, but, if emissions of
mercury are the main concern, then the fish and shellfish pathway may be the important
one.
Issue Section 2.4.4 of the Indirect Exposure Document discusses the use of mobility and
length of residency statistics to select the duration of exposure. The discussion
points out that more than half (about 60 percent) of the moves recorded by the
Census Bureau were local moves within the same county. Therefore, many
individuals who move may remain within the study area and continue to be
exposed. The discussion also points out that different populations are characterized
by different residency times in a single dwelling, e.g. farmers vs. homeowners vs.
tenants. However, the Indirect Exposure Document does not discuss the
recommendation in the Exposure Factors Handbook (EPA, 1989b) to use a central
tendency estimate for exposure duration of 9 years and a high end estimate of
30 years based on mobility statistics. More discussion is needed about whether
these estimates need revising on the basis that many of the moves may be within
the study area.
2-12
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Conclusions/Recommendations
Local census data may be used to determine appropriate values of the duration of
exposure for the local population and population subgroups of concern, particularly when
supplemented with site-specific surveys. Otherwise, the recommendation is to continue to
use the 9 year and 30 year default values. Shorter time periods may be applicable in
consideration of the expected life of the facility or duration of the burn (in the case of a
mobile incinerator). However, the assessor should consider that indirect exposures may
continue after emissions stop due to accumulation of contaminants in soil and other media.
2.5 Pathways of Human Exposure
Issue The Indirect Exposure Document does not address exposure through breast milk.
Conclusions/Recommendations
Procedures have been developed for estimating contaminant levels in human breast
milk on the basis of the contaminant intake by the mother. Such procedures have been
developed by Smith (1987) and Sullivan et.al. (1991) and are also presented in the Dioxin
Exposure Document (EPA, 1992c). The approach by Smith assumes that the
concentration in breast milk fat is the same as in maternal fat and can be calculated as:
where:
m
h
Cmilkfat
m h f j
O.693 f2
[2-1]
milkfat = concentration in maternal milk (pg/kg of milk fat)
= average maternal intake of dioxin (pg/kg of body weight/day)
= half-life of dioxin in adults (days)
= proportion of ingested dioxin that is stored jn fat
= proportion of mother's weight that is fat (kg maternal
fat/kg total body weight)
Although this procedure was developed for application to dioxin, it would be
2-13 November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
applicable to other lipophilic organics if the chemical specific inputs can be established.
This steady-state model assumes that the contaminant levels in maternal fat remain
constant. Though not described here. Smith (1987) also presents more complex
approaches which account for changes in maternal fat levels during breast feeding. The
model developed by Sullivan et al. (1991) is a variation of the models proposed by Smith
(1987). The Sullivan model considers changes in maternal fat levels and predicts chemical
concentrations in milk fat as a function of time after breast feeding begins. The model
proposed by Smith assumes that infant fat concentration at birth is zero, whereas Sullivan
assumes that the infant fat concentration at birth is equal to the mother's fat
concentration.
By way of illustration, the half-life of 2,3,7,8-TCDD in humans is estimated to be 5
to 7 years, as discussed in the Dioxin Exposure Document (EPA, 1992c). For the purpose
of this preliminary analysis, it is assumed that a 7-year half-life applies to all of the dioxin-
like compounds. Smith (1987) suggests values of 0.9 for f ., and 0.3 for f 2- Using these
assumptions and a background exposure level of 1 to 3 pg of TEQ/kg-d (derived from diet
analysis, see EPA, 1992c), the concentration in breast milk fat is predicted to be about
10 to 30 ppt of TEQ, which agrees well with the measured values.
Using the estimated contaminant concentration in breast milk, the dose to the
infant can be estimated as follows:
_ Cmilkfat
~
ED
BWinfantAT
[2-2]
where:
ADD infant
IR
milk
ED
BWSl
infant
AT
average daily dose to the infant (pg/kg/d)
ingestion rate of breast milk (kg/d)
exposure duration (yr)
body weight of infant (kg)
averaging time (yr)
fraction of fat in breast milk (dimensionless)
2-14
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
f 4 = fraction of ingested contaminant which is absorbed
(dimensionless) '
This approach assumes that the contaminant concentration in milk represents the
average over the breast feeding time period. If the dynamic models mentioned above are
used, the dose can be estimated using an integration approach to account for the changes
in concentration over time.
Smith (1987) reports that a study in Britain found that the breast milk ingestion rate
for 7 to 8 month old infants ranged from 677 to 922 mL/d and that a study in Houston
measured the mean production of lactating women to range from 723 to 751 g/d. Smith
(1987) also reports that breast milk ingestion rates remain relatively constant over time
while the infant is breast feeding, that the milk can be assumed to have a 4 percent fat
content, and that 90 percent of the ingested contaminant is absorbed. The National
Center for Health Statistics (1987) reports the following mean body weights for infants:
6-1 1 mo:
1 year:
2yr:
9.1 kg
1 1 .3 kg
13.3 kg
Using Equation [2-2] to evaluate the importance of breast feeding and assuming
that an infant breast feeds for one year, has an average weight during this period of 10 kg,
ingests 0.8 kg/d of breast milk and that the dioxin concentration in milk fat is 20 ppt of
TEQ, the ADD to the infant over this period (i.e., AT = 1 yr) is predicted to be about
60 pg of TEQ/kg-d. This value is much higher than the estimated range for background
exposure to adults (i.e., 1 to 3 pg of TEQ/kg-d). However, if a 70 yr averaging time is
used, then the LADD (Lifetime Average Daily Dose) is estimated to be 0.8 pg of TEQ/kg-d
which is near the lower end of the adult background exposure range. On a mass basis,
the cumulative dose to the infant under this scenario is about 210 ng compared to a
lifetime background dose of about 1700 to 5100 ng (suggesting that 4 to 1 2 percent of
the lifetime dose may occur as a result of breast feeding). Traditionally, EPA has used the
2-15
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
LADD as the basis for evaluating cancer risk and the ADD (i.e., the daily exposure per unit
body weight occurring during an exposure event) as the more appropriate indicator of risk
for noncancer endpoints.
The simplified procedure described above contains a number of uncertainties. A
tendency toward overestimates of the dose to the infant is caused by the assumption that
reductions do not occur in maternal fat levels during breast feeding. Sullivan et al. (1991)
estimates that the steady-state assumption may lead to overestimates of 20 percent.
Uncertainty is also introduced by the assumption that the assumed half-life rate and
partitioning factors apply to all the dioxin related compounds. Although these properties
are likely to be similar among the various congeners, some variation is expected. It is
unknown whether the net effect of these uncertainties would lead to over or under
estimates of dose. However, the simple model appears to provide reasonable predictions
of background levels found in breast milk and was judged adequate for purposes of a
preliminary analysis.
Travis et al. (1988) presented an alternative approach to estimating breast milk
contaminant levels. They proposed a biotransfer approach:
where:
*m
m
- Bml
contaminant concentration in breast milk (mg/kg)
biotransfer factor for breast milk (kg/d)
maternal intake of contaminant (mg/d)
[2-3]
They also argue that the biotransfer factor is primarily a function of the octanol-
water partition coefficient (Kow ) and developed the following geometric mean regression:
fi = 6.2*1 CT4 Kow
[2-4]
2-16
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
This regression was derived from data on 6 iipophilic compounds (log Kow range: 5.16 to
6.5), but did not include any dioxins or furans. Assuming a log Kow of 6.6 for 2378-
TCDD, a Bm of 3700 kg/d is predicted. Combing this value with a maternal intake of 10
pg/d (or 10'7 mg/d), a breast milk concentration of 37 ppt is predicted. This prediction is
about 10 times higher than what has been measured in the U.S. Thus, this approach does
not appear to work as well as the earlier approach suggested by Smith et al (1987). It is
recommended that assessors first try to apply the Smith approach and if model inputs
cannot be found then use the approach developed by Travis et al. (1988).
Issue Placenta! transfers are not addressed in the Indirect Exposure Document.
Conclusions/Recommendations
This is an area which requires research to develop a methodology.
2-17
November 10, 1993
-------
-------
DRAFT - DO NOT QUOTE OR CITE
[The following is offered as a replacement for Chapter 3 of the Indirect Exposure
Document. This alternative has been developed by David Cleverly of the Exposure
Assessment Group, and Donna Schwede and Bill Peterson of the Atmospheric Research
and Exposure Assessment Laboratory. Care has not been taken to make the units or
parameter names consistent with other chapters of the Indirect Exposure Document.
However, the units are internally consistent.]
3. AIR DISPERSION AND DEPOSITION MODELING OF THE POLLUTANT STACK
EMISSIONS
3.1. BACKGROUND AND PURPOSE
Estimation of human health risks associated with exposures to air releases of
contaminants from combustors requires estimates of the atmospheric concentrations and
annual deposition rates of the emitted materials in the areas around the combustion
facility. It has been customary for EPA to use air dispersion/deposition models to estimate
the atmospheric transport, the surface deposition flux, and the ambient air concentrations
of specific compounds attributable to smokestack emissions from an industrial combustion
source. Air dispersion models are mathematical constructs that approximate the physical
and chemical processes occurring in the atmosphere that directly influence the dispersion
of gaseous and paniculate emissions from the stack of a combustion source. These
models are computer programs encompassing a series of partial differential and algebraic
equations that calculate the dispersion and deposition of the emissions. Concentration and
deposition isopleths of the contaminants discharged from the stack are computed at
specified distances from the source. These quantities are used to estimate the magnitude
of potential exposures to the human receptor, and to estimate the potential human health
risks in populations living and working in the vicinity of the stationary combustion source.
The Agency's Guideline on Air Quality Models (EPA, 1993a) has identified
numerous air dispersion/deposition models that may be applied to the analysis of source
emissions. This chapter focuses on the COMPDEP model, which is recommended by the
Working Group as a currently available air model appropriate for use in the analysis of
3-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
indirect exposures from combustor emissions. Efforts are currently underway to create an
updated air transport model, which will be called ISC-COMPDEP. This model is a merger
of parts of the ISC2 and COMPLEX 1 model. Further model development is discussed is
Part 111 of this Addendum.
COMPDEP was first described in the Indirect Exposures Document. Recent
revisions to the computer code have been made, creating an updated version. A principal
change that was made allows the user to define up to 10 particle size categories in the
analysis of the surface deposition of contaminants that are bound to particulates in the
emissions.
The purposes of this Chapter are to: a) compile a comprehensive list of pollutants
which may potentially occur in the stack emissions from the incineration and combustion
of waste materials, b) present a hierarchical approach for determining contaminants to
evaluate in a site-specific indirect exposure assessment, c) derive representative emission
rates of the pollutants for air modeling purposes, including long-term average emissions,
and short-term emissions that may result from perturbations, upset conditions, and startup
and shutdown in operations, d) describe the COMPDEP model and associated wet and dry
deposition algorithms, and, e) to describe procedures for applying the COMPDEP model
using vapor phase and particulate phase partitioning of the emitted pollutants, and a
specific particle size distribution of particulate matter.
3.2 SELECTING TOXIC AIR POLLUTANTS
The first step in the analysis of indirect exposures is the identification and selection of
pollutants that may be emitted to the air during the combustion of anthropogenic waste
materials. This selection process should focus on chemicals that are potentially toxic to
humans, and that have a definite propensity for bioaccumulating or bioconcentrating in the
human and ecological food chains.
As the initial step in the selection process, a comprehensive list of compounds that
have been detected in combustor emissions is shown in Tables 3-1, 3-2, and 3-3. These
tables list potentially toxic organic and inorganic contaminants that may be subject to the
human health risk assessment associated with direct and indirect exposures to the
3-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-1. Initial list of dioxin-like compounds identified in emissions from the
incineration and combustion of anthropogenic wastes.
cniorinated Didenzo-p-dioxms (CDDs)
Chlorinated dibenzofurans
2,3,7,8-TCDD
1,2,3,7,8-PentaCDD
1,2,3,4,7,8-HexaCDD
1,2,3,6,7,8-HexaCDD
1,2,3,7,8,9-HexaCDD
1,2,3,4,6,7,8-HeptaCDD
1,2,3,6,7,8,9-OctaCDD
Other TetraCDDs
Other PentaCDDs
Other HexaCDDs
Other HeptaCDDs
2,3,7,8-TCDF
1,2,3,7,8-PentaCDF
2,3,4,7,8-PentaCDF
1,2,3,4,7,8-HexaCDF
1,2,3,6,7,8-HexaCDF
1,2,3,7,8,9-HexaCDF
2,3,4,6,7,8-HexaCDF
1,2,3,4,6,7,8-HeptaCDF
1,2,3,4,7,8,9-HeptaCDF
1,2,3,4,6,7,8,9-OctaCDF
Other TCDFs
Other PentaCDFs
Other HexaCDFs
Other HeptaCDFs
3-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-2. Initial list of organic compounds identified in emissions from the incineration
and combustion of anthropogenic wastes (key follows).
Organic compounds
Acetonitrile
Acrylonitrile
Anthracene
Azobenzene
Benzaldehyde
Benzene
Benz(a)anthracene
Benzo(a)pyrene
Benzo(b)fluoranthene
Benzo(e)pyrene
Benzo(g,h,i)perylene
Benzo(j)fluoranthene
Benzo{k)fluoranthene
Benzyl chloride
1,1-BiphenyI
Bis(2-chloroethoxy) methane
Bis(2-chloroethyl) ether
Bis(2-ethylhexyl) phthalate
Bis(chlorornethyl) ether
Bromochloromethane
Bromodichloromethane
Bromomethane
1 ,3-Butadiene
Butyl benzyl phthalate
Captan
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chlorocyclopentadiene
Chloroform
Chloromethane
Chrysene
Cumene
Di(2-ethylhexyl)phthalate
Dibenzofuran
Dibenzo(a,e)fluoranthene
Dibenzo(a,h)fluoranthene
1 ,2-Dibromo-3-chloropropane
Dibutyl phthalate
1 ,2-DichIorobenzene
1 ,3-Dichlorobenzene
1 ,4-Dichlorobenzene
RfDa
+;L
+
+
-
+ ;K
UR
-
-
-
-
-
-
-
.-
+ +;K
-
-
-
-
-
+ +;K
+ +;Ne
-
+ ;L
+ + +;De
+ +;De
+ +;L
+ +;L
-
+ +;L
UR
-
+;K
+ +;L
-
-
-
-
+
+ ;L
UR
-
RfCb
+;L
+ +
-
-
-
UR
-
-
-
-
-
-
-
IA
IA
-
IA
+ ;L
IA
-
-
+ + +;Ne
-
-
UR
UR
-
UR
-
UR
UR
-
+ ;Ne
-
UR
-
-
+ +;Rsp
IA
-
-
UR
Carcc
-
B1;O/I
D
B2;O/I
-
A;O/I
B2;O
B2;O
B2;O
UR
D
UR
B2;O
B2;O
D
D
B2;O/I
B2;O/I
A;O/I
D
B2;O
D
B2;l
C;O
UR
-
B2;O/I
D
D
B2;O/I
UR
B2;l
-
B2;0
D
UR
B2;l
UR
D
D
D
-
Ref.
1,2
1,2
1,2
1,2
1,2
1,2
1
1
1
1
1
1
1
1,2
1,2
1
1
2
1,2
1
1
1,2
1
1,2
1
1
1,2
1
1
1,2
1
1
1,2
1
1,2
1
1
1
1
1,2
1,2
1,2
3-4
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-2. (continued)
Organic compounds
1,1 -Dichloroethane
1 ,2-Dichloroethane
1 , 1 -Dichloroethylene
Dichloromethane
2,4-Dichlorophenol
Dimethyl phthalate
1 ,4-Dioxane
Ethyl benzene
Fluoranthene
Formaldehyde
Hexachlorobenzene
Hexachlorobutadiene
Hexachloroethane
Methyl mercury
Methylene bromide
Methyl ethyl ketone
Naphthalene
Pentachlorobenzene
Pentachlorophenol
Phenol
Polychlorinated biphenyls
Polycyclic organic matter
Pyrene
Styrene
1 ,2,4,5-Tetrachlorobenzene
1,1,1 ,2-Tetrachloroethane
1 ,1 ,2,2-Tetrachloroethane
Tetrachloroethylene
Toluene
1 ,2,4-Trichlorobenzene
1 ,1 ,1-Trichloroethane
1,1 ,2-Trichloroethane
Trichloroethylene
2,4,5-Trichlorophenol
2,4,6-Trichlorophenol
Vinyl chloride
m-Xylene
o-Xylene
p-Xylene
Xylenes
RfDa
UR
-
+ +;L
+ +;L
+ ;lm
UR
-
+ ;LK
+ ;L
+ +;L
+ +;L
+ ;L
+ +;K
+ +;IMe
+
+ ;De
UR
+ ;LK
+ +;LK
+ ;De
-
-
+ ;K
+ +;L
+ ;K
+ ;K
UR
+ +;L
+ +;LK
+ +;R
WD
+ +
UR
+ ;L
-
-
+
+
+
+ +;De
RfCb
UR
-
UR
UR
-
IA
-
+ ;De
-
-.
IA
-
UR
-
*-
+ ;De
-
-
UR
IA
-
-
-
+ +;Ne
-
-
-
-
+ +;Ne
UR
UR
UR
UR
IA
IA
-
UR
UR
UR
UR
Carcc
C;0
B2;O/I
C;O/I
B2;O/I
-
D
B2;O
D
D
B1;l
B2;O/I
C;O/I
C;0/l
-
-
D
D
D
B2;O/I
D
B2;0
UR
D
UR
-
C;0/l
C;0/l
UR
D
D
D
C;0/l
WD
UR
B2;O/I
A;O/I
-
-
-
D
Ref.
1,2
1
1,2
1
1,2
1
1
1,2
1
1
1,2
1,2
1,2
1
2
1,2
1,2
1,2
1,2
1,2
1
1
1,2
1,2
1,2
1
1
1,2
1,2
1,2
1,2
1,2
1
1,2
1,2
1,2
1,2
1,2
1,2
1,2
3-5
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Table 3-3. Initial list of inorganic contaminants identified in emissions from the
incineration and combustion of anthropogenic wastes (key follows).
Inorganic contaminants
Antimony
Arsenic, inorganic
Barium
Cadmium
Chromium (III)
Chromium (VI)
Lead + compounds
Manganese
Mercury, inorganic
Mercuric chloride
Nickel, soluble salts
Selenium + compounds
Silver
Zinc + compounds
RfDa
+
+ +
+ +;De
+ + +
+
+
+ + +
+ ;Ne
UR
UR
+ +;De
+ + +
+
+ +
RfCb
-
-
UR
UR
UR
UR
-
+ +;Rsp
UR
UR
UR
-
-
-
Carcc
-
A;l
-
B1;l
UR
A;l
B2,l
D
D
-
D
D
D
D
Ref.
1,2
1,2
1,2
1,2
1,2
1
1
1
1,2
2
1,2
1,2
1
1,2
3-6
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Key to Tables 3-2 and 3-3
a. RfD is the reference dose defined as the daily ingestion exposure of a
contaminant that is likely to be without appreciable risk during a portion of the
lifetime.
b. RfC is the reference concentration defined as the daily inhalation exposure of
a contaminant that is likely to be without appreciable risk during a portion of a
lifetime.
c. Care, is an abbreviation for carcinogen.
L = Adverse effects on the liver.
K = Adverse effects on the kidney.
De = Adverse developmental effects.
Im = Adverse effects on the immune system.
R = Adverse reproductive effects.
Ne = Adverse effects on the central nervous system.
Rsp = Adverse effects on the respiratory system.
IA = Inadequate data.
UR = Currently under review.
WD = Withdrawn
+ = Low confidence weighing of the supporting scientific information.
+ + = Medium confidence weighing of the supporting scientific information.
+ + + = High confidence weighing of the supporting scientific information.
- = No information.
Group A carcinogen: Human carcinogen (sufficient evidence of carcinogenicity in
humans).
Group B carcinogen: Probable human carcinogen (B1 - limited evidence of
carcinogenicity in humans; B2 - sufficient evidence of carcinogenicity in animals
with inadequate or lack of evidence in humans).
Group C carcinogen: Possible human carcinogen (limited evidence of
carcinogenicity in animals and inadequate or lack of human data).
Group D carcinogen: Not classifiable as to human carcinogenicity (inadequate or
no evidence).
I = Carcinogenic by the inhalation route.
O = Carcinogenic by the oral route.
Reference 1 = EPA's Integrated Risk Information System (IRIS) on-line database
as described in the Federal Register (58 FR 11490) February 25, 1993.
Reference 2 = EPA, 1989d.
3-7
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
combustor emissions. This list includes chemicals identified as most frequently detected
and measured in stack emissions from incinerators, boilers, and industrial furnaces
combusting hazardous waste, sewage sludge, municipal solid waste, biomass, automobile
tires, and medical and hospital waste (Dempsey and Oppelt, 1993; CARB, 1990a,b;
CARB, 1991; CAPCOA, 1991; EPA, 1986; EPA, 1988b). To the extent possible, chemical
speciation of the various forms of these compounds should be identified. Using these
Tables as a starting point, the assessor can narrow the number of compounds of concern
for a particular combustor using the following three steps:
STEP 1. Consider the Feed Material Being Combusted
The selection of the chemicals for site-specific analysis will largely depend on the
type of waste materials that will routinely be accepted and incinerated at the combustion
facility. If chlorine is present in the feed, then it is imperative that chlorinated dioxins and
dibenzofurans be evaluated in the exposure assessment. Even if chlorine is not specifically
identified in the waste feed, it is recommended that this class of compounds be addressed
in the assessment. It may be that only an extremely small concentration of chlorine in the
waste is sufficient to participate in the formation of CDDs/CDFs. In addition, combustion
air may contain precursor compounds and halides that may cause dioxins to be
thermalytically formed. Therefore, it is generally suggested that assessments of waste
incineration sources include dioxin-like compounds in the analysis of emissions. Products
of incomplete combustion (PICs) will form in the stack gas emissions as a consequence of
the incomplete destruction of all organic species leaving the combustion chamber.
Therefore, the polycyclic aromatic hydrocarbons (PAHs), i.e. benzo(a)pyrene, will need to
be routinely considered. Finally, to the extent a combustion facility accepts and
incinerates metal-bearing materials and wastes, the heavy metals having significant
toxicologic properties will need to be addressed in the exposure analysis, i.e., lead,
mercury and mercury compounds, chromium, cadmium and nickel.
STEP 2. Consider the Availability of Appropriate Toxicological Information
Risk assessments cannot be completed for compounds which lack verified dose-
3-8
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
response relationships for specific health endpoints. Accordingly, such compounds can be
eliminated from consideration in an exposure/risk assessment. Tables 3-2 and 3-3 contain
information which can assist in a toxicological evaluation, including availability of
Reference Doses (RfD), Reference Concentrations (RfC), a confidence rating for RfD and
RfC information, characterization of the health impact (impact to liver, kidney,
developmental effect, etc.), and the current carcinogenic rating (A through D). The
assessor should review the latest version of IRIS to confirm if new chemicals have been
added, if additional RfDs, RfCs have been verified, and if the carcinogenicity group
classification has been added or updated.
STEP 3. Narrow the List with a Quantitative Ranking Scheme
After following the first two steps above, the long list in Tables 3-1 through 3-3
may be substantially narrowed. However, it may not be necessary to conduct full
assessments on the remaining chemicals. Say, for example, a credible ranking scheme
could be developed which would rank a list of 50 chemicals (and there are over 90
chemicals listed in Tables 3.1-3.3) from most to least concern. If the detailed site-specific
assessment on the 20 chemicals of most concern indicated that chemicals 1 5 through 20
on the list resulted in negligible indirect health risk, than there would be credible grounds
to conclude that chemicals 21 through 50 do not require further indirect assessment.
Likewise, if the assessment indicated a health risk of potential consequence for chemicals
15 through 20, then evaluation of more chemicals on the list of 50 would be required.
The Risk Assessment Guidance for Superfund (RAGS; EPA, 1989c) suggests a simple
linear ranking scheme for focusing on the "most significant" chemicals. This ranking
scheme multiples a quantitative toxicity value (slope factor, e.g.) times a quantitative
concentration of the contaminant in the exposure media (soil, e.g.). For combustor
emissions, the "concentration" term should be replaced with the emission rate of the
chemical. Procedures to estimate emission rates are given in Section 3.3 below. A third
factor that should be considered is the potential for the chemical to bioaccumulate.
Factors indicating bioaccumulation potential include persistence in the environment - more
persistence translates to more bioaccumulation - and partitioning tendencies - more
3-9
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
partitioning onto soils and sediments translates to more bioaccumulation. A surrogate
factor which can be used for bioaccumulation potential is the logarithm of the octanol-
water partition coefficient, denoted as log Kow. The higher the value of the log Kow, the
greater is the possibility for partitioning onto soils and sediments. Longer persistence is
correlated to tighter sorption to soils/sediments, which is why the log Kow can capture
both persistence and partitioning tendencies appropriately.
3.3 ESTIMATING THE STACK EMISSION RATE OF THE TOXIC AIR POLLUTANTS
For the purposes of chemical selection as described above, and for air modeling and
exposure assessment, the release rate of pollutants from the combustor stack is expressed
in metric units of mass of the pollutant emitted per unit time (e.g., grams/s). The release
rate used in indirect exposure assessments should generally be reflective of the expected
average emissions from the facility over the long-term (e.g., over the engineered life of the
combustion facility). However, if acute exposures are of concern in the risk assessment,
the release rate should be reflective of maximum short-term emissions (e.g., as a result of
equipment malfunction, start-up and shut-down, or possible accidental releases). Although
mass emission rates can be estimated in any of a number of ways, the following
hierarchical order of preference is recommended:
A. For facilities that are built and operational, it is preferred that direct stack
measurements be used, using EPA recommended chemical-specific (and, wherever
possible, species or congener-specific) stack sampling, analytical, and quality control,
quality assurance protocols and procedures. Stack monitoring should provide information
on the concentrations of the pollutants (e.g., ng/dscm), the actual velocity of stack gas
release (m/s), and stack gas temperature (K). In addition, based on the characteristics and
heat value of the waste feed, the temperature of combustion, and stoichiometric
conditions, the volume of flue gases evolved per unit of time should be estimated (in
metric units of dry standard cubic meters per minute at 12% C02 or 7% O2). To calculate
a mass emission rate of the pollutant, it is necessary to multiply the estimated dry flue gas
volume (m3) per minute associated with combustion by the measured in-stack
3-10
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
concentration of the contaminant that are in units of g/dscm. (It should be noted that
concentrations reported on an equivalent oxygen or carbon dioxide basis or on a dry
volume basis must be adjusted to the actual stack gas conditions in order to determine the
actual, in-stack concentrations.) The product of this calculation (g/min) is then divided by
60 s/min to give the grams per second emission rate needed as an input into the
COMPDEP model. In cases were there are multiple stack tests from the same facility, care
should be taken to ensure that the emissions characterization reflects a wide range of
operating conditions and also accounts for any expected deterioration in facility
performance which would affect the facility's emissions over its useful life. In arriving at
the final estimate of emissions of the contaminant, the average value is considered
representative of long-term emissions, taking into consideration the average yearly
operating hours of the combustion facility.
B. For facilities that have been constructed, but not yet operational, or are in the
planning stages of development, the following approach is recommended. Reports of
stack tests which have measured the emissions of the pollutants of interest should be
collected and reviewed from facilities that are most similar in technology, design,
operation, capacity, auxiliary fuels used, waste feed types and composition, and air
pollution controls as the facility under consideration. The stack test reports should be
evaluated to determine if EPA recommended protocols, or equivalent, were used for stack
sampling and analysis that are appropriate for the specific pollutants being measured.
When combining data from test results for a number of facilities, care should be taken to
convert concentrations and stack gas parameters to a common basis (and consistent units
of measurement) that is appropriate for the facility under consideration, considering
process feed rates and other facility-specific operating conditions. For purposes of
exposure assessment and risk characterization, ranges and average values should be
developed in order to address uncertainties inherent in the emissions estimates. However,
the average emission rate of the contaminant is the value used to represent long-term, and
expected emissions from the combustor under study.
3-11
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
C. If no data exist relevant to a specific facility, then the Office of Air Quality
Planning and Standards's AP-42, Compilation of Air Pollution Emission Factors
(EPA, 1985a), can be used. Care should be taken to select emission factors which were
developed for technologies that best match the facility under consideration. The basic
limitation of these data is the fact that emission factors are not usually reflective of
specific emission control equipment. Also, information is available for only a limited
number of pollutants. ,
Specifically, the AP-42 document provides guidance on emission factors for
Municipal Waste Combustors, Sewage Sludge Incinerators, and Medical Waste
Incinerators. At this time, emissions from Hazardous Waste Incinerators are not addressed
in AP-42. Emission factors presented in AP-42 are designed for estimating emissions from
a large number of sources over a wide area. The use of AP-42 emission factors to
estimate emissions from any one facility should be done with great care. Whenever
possible, emissions measured from the facility in question should be used in preference to
the emission factor. In the absence of site specific emission test data, emissions
measured at an identical or similar facility combusting the same or similar material should
be used. AP-42 emission factors should be the last alternative because they are averages
of values determined at one or more individual facilities. The individual values which are
used to develop the average may vary considerably (for example the extreme values used
for the emission factor for mercury from municipal mass burn combustors vary by over a
factor of five from the value presented in AP-42). Emission factors for the following
compounds will be available for the following types of incineration systems:
Municipal Waste Combustors: arsenic, cadmium, chromium, mercury nickel, lead, total
tetra through octa chlorinated dibenzo-p-dioxin and furans.
Sewage Sludge Incinerators: antimony, arsenic, cobalt, beryllium, cadmium, chromium,
lead, manganese, mercury, nickel, selenium, various chlorinated dibenzo-p-dioxin and
furans.
3-12
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Medical Waste Incinerators: antimony, arsenic, beryllium, cadmium, chromium,
manganese, mercury, nickel, chlorobenzene, chlorophenol, total tetra through octa
chlorinated dibenzo-p-dioxin and furans, total PCS.
D. In the absence of suitable AP-42 emission factors, clearly documented
engineering evaluations may be used. Documentation should include copies of emission
test reports used to derive the emission estimates, any assumptions made, and the
rationale for the conclusions that are made.
3.4 INCREASES IN EMISSIONS DURING EQUIPMENT MALFUNCTIONS; STARTUP;
AND SHUTDOWN
One of the more difficult aspects in deriving pollutant emission rates from
incineration sources is the accounting for temporary increases in emissions that may occur
as a result of startup and shutdown in operations, malfunctions or perturbations in the
combustion process or changes in the removal efficiency of the air pollution control
equipment. In deriving a quantitative measure of the magnitude and duration of increased
emissions associated with these events, it is recommended that a procedure similar to the
one used to derive estimates of routine emissions be applied. In this context the following
sources of information, in order of preference, should be reviewed:
• The assessor should review stack emission testing and engineering reports
on the performance and operations of combustion technologies, furnace designs, and air
pollution control devices that are most similar in every respect to the facility under review.
Often the emissions testing is accompanied by records that may give a gross indication of
whether the system is functioning according to design specifications. Most facilities
routinely monitor carbon monoxide (CO) in the stack to provide a general indication of
combustion efficiency. Carbon monoxide inversely correlates with combustion efficiency,
and a sudden increase in concentration could imply that an upset in the quality of the burn
may have occurred. The engineering analysis in the test report may summarize these data,
and qualitatively discuss upset conditions, and the length of time upset conditions were
3-13
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
maintained. Most facilities also routinely monitor furnace temperature. Furnace
temperature is directly related to combustion efficiency and gives a qualitative measure of
whether incineration of the wastes are occurring according to the design of the
combustion chamber. Temperatures outside the prescribed envelope of routine operations
may signal a temporary malfunction or upset in the quality of the burn. This may be due
to the delivery of a highly combustible and/or wet material into the combustion chamber.
During these conditions, temporary excursions in emissions of pollutants from the stack
may occur, and this may have been recorded in the engineering test report.
• A second area of information regarding temporary increases in the emissions
during equipment malfunctions are tests designed to determine the removal efficiency of
the air pollution control devices (APCD). More specifically, some facilities may have been
required by state air pollution control agencies to provide information on the longer-term
performance of the APCD. In most cases such performance tests involve determining the
variability in the efficiency of removal of particulate matter from the combustion gases,
and/or the removal of acid gases such as sulfur dioxide and hydrogen chloride. In some
rare instances the engineering evaluation may have been undertaken to determine the
extent mercury was being reduced, or to what extent dioxins were controlled with the
APCD. These engineering analysis may be public documents, and it may be possible to
obtain copies from the appropriate state agency, or the EPA Regional office. In any case,
these data may be helpful in establishing a quantitative basis for estimating the magnitude
and duration, and frequency of occurrence of higher emissions than what is expected if
the APCD were performing according to the design specifications.
In the event these sources of information do not provide a quantitative, e.g.,
numerical, basis for estimating the magnitude and duration of increase of the pollutant
during upset conditions, then the procedure required by the State of California Air
Resources Board (CARB, 1990) in the evaluation of toxic air contaminant emissions from
waste incinerators can be applied. In this procedure, defaults for estimating emissions
representative of emergency operating conditions and poor operating conditions at the
3-14
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
facility are provided, as follows:
- Emergency operating conditions. Estimation of stack emissions of the
contaminants representative of emergency operating conditions should be based on
uncontrolled emissions over a one-hour period. For example, if dioxin is assumed to be
controlled by the APCD at a rate of 99 % relative to the uncontrolled concentration, then,
under the emergency operating conditions scenario, the stack concentration would be
assumed to be 100 times higher during a 1-hr period. The CARB has defined this
emergency condition as a sudden upset in which case the air pollution control device
(APCD) is by-passed to a waste stack, and uncontrolled emissions occur until the facility is
completely shutdown. These operating conditions include such events as: a turbine trip,
blown boiler tubes, failure in the APCD, feed system failure, fire, etc. However if the
source could demonstrate that the APCD would never be by-passed during an emergency
event, then use of the uncontrolled emissions would not realistically reflect the potential
emissions from the facility. CARB gives no guidance on an alternative approach, but does
indicate that the evaluation of potential health impacts resulting from acute exposures
should represent a range of operating conditions. These operating conditions should
reflect poor operation of the equipment and total equipment failure. For these operating
conditions the increase in emissions should be estimated resulting from reduced control
efficiency, and the duration of the event should be noted.
" Poor operating conditions. To represent stack emissions during operating
conditions that are less than normal, CARB recommends, as a default, multiplying the
average emissions (found from the source test database) by ten. This factor is
incorporated into the annual emission rate of organic and metal emissions. CARB further
recommends using a default value for the fraction of the time a facility operates under
poor conditions on an annual basis. For organic contaminants, it is assumed that the
facility operates normally 80% of the time and operates poorly for the other 20% of the
time. For metal emission estimates, CARB recommends the default assumption that the
facility operates normally 95% of the time, and under poor conditions .for the other 5% of
the time during the course of the year.
3-15
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
3.5 SHORT TERM IMPACTS DUE TO METEOROLOGIC CONDITIONS
As discussed above, short term increases in emission rates can lead to higher
atmospheric concentrations than would occur during normal operations. Even during
normal operations, fluctuations in atmospheric concentrations occur due to changing
meteorologic conditions. During times of stable air conditions, i.e. atmospheric inversions,
the dispersion of emissions is reduced and higher than normal air concentrations result. As
discussed in Chapter 2, these short term increases are unlikely to significantly affect
indirect exposures, but may be important to consider in evaluating acute effects from
inhalation. The COMPDEP model can provide the daily high concentration expected over a
one year period. Since the model assumes a constant emission rate, this high would occur
as a result of stable air conditions. In order to evaluate the maximum short term exposure
scenario, the assessor should judge whether the increases in emission rates due to upset
conditions (as discussed in section 3.4) could combine with the effects of stable air to
further increase this estimated daily high concentration from the model.
3.6 CONDENSED DESCRIPTION OF THE COMPDEP MODEL
The COMPDEP model was developed to provide estimates of the air concentrations
and wet/dry deposition fluxes of the stack emissions of contaminants from industrial
sources in varied topography (e.g, from simple to complex terrain). COMPDEP uses
standard meteorological data to produce estimates of ambient air concentration, dry
deposition and wet deposition at individual receptor sites. COMPDEP was basically
derived as a combination of the Industrial Source Complex (Short Term) model (Bowers et
al, 1979) and the COMPLEX I, an EPA recommended screening model for use in complex
terrain (EPA, 1993). To account for pollutant deposition, the concentration algorithms in
COMPLEX I were replaced with those from the Multiple Point Source Algorithm with
Terrain Adjustments Including Deposition and Sedimentation (MPTER-DS) model (Rao and
Satterfield, 1982). The MPTER-DS algorithms are based on the gradient transfer theory
described in Rao (1981) and are extensions of the traditional Gaussian plume algorithms.
The routines developed by the California Air Resources Board (CARB, 1986) were included
for determining deposition velocity. The Industrial Source Complex - Short Term (ISCST)
3-16
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
(Bowers et al, 1979) algorithms for accounting for building wake effects were included.
COMPDEP also contains algorithms for modeling wet deposition which are based on Slinn
(1984) and PEI and Cramer (1986).
In accordance with a recent review of the existing operating computer code,
COMPDEP was modified, and an updated version is currently available. In order to
eliminate any confusion, the current version has been assigned the number 93252. In this
numbering system, 93 refers to the year, the 252 refers to the day (September 9) of the
most recent modifications. COMPDEP is available to users through a electronic bulletin
board system known as the Support Center for Regulatory Air Models (SCRAM). This
electronic bulletin board is maintained by EPA's Office of Air Quality Planning and
Standards in Research Triangle Park, North Carolina through the Technology Transfer
Network. The COMPDEP model can be accessed within SCRAM under the topic, "other
models." SCRAM permits the user to download the computer code, and provides
instructions on operating the COMPDEP model and for accessing meteorological data.
User's may receive information on accessing the SCRAM by calling the systems operator
by phone at (919) 541-5384 in Durham, North Carolina during normal business hours EST.
Because of recent refinements and modifications in the COMPDEP model, EPA has
not printed a users manual. However, instructions on performing dispersion and deposition
analysis with COMPDEP are on the SCRAM system. The following subsections are a
review of the key algorithms employed within the program for estimating dispersion and
deposition of contaminant emissions.
3.6.1 Dispersion Parameters
The standard deviation of the lateral and vertical concentrations are applied in the
calculation of the ambient air concentrations of the contaminants released from the stack
of stationary industrial sources. Point estimates are used in COMPDEP to calculate the
dispersion parameters, ay and <7Z, for all terrain heights. Point estimates of ay and a z are
used by ISCST for calculating concentrations in simple terrain. These point estimates are
inserted in Equation [3-1] for calculating the ambient air concentration or the contaminant
(//g/m3).
3-17
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
where:
X
Q
K
V
"
X =
Q KV
[3-1]
2 ic us ay
ambient air concentration (/vg/m 3)
pollutant emission rate (g/s)
units conversion factor
vertical term (dimensionless)
wind speed (m/s)
standard deviation of lateral and vertical concentration distributions
(m)
crosswind distance (m)
COMPLEX I uses the point estimate for az, but uses sector averaging to calculate the
horizontal dispersion instead of calculating a ay. Equation [3-2] is used in COMPLEX I to
calculate ambient air concentrations.
where:
Q
K
V
R
A0'
Z =
Q KV
R Ad'u a
A i
[3-2]
pollutant emission rate (g/s)
units conversion factor
vertical term (dimensionless)
radial distance from source to receptor (m)
sector width in radians (dimensionless)
wind speed (m/s)
standard deviation of vertical concentration distribution (m)
3-18
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
The algorithms for calculating air concentrations were modified in COMPDEP so that
predicted concentrations match those from ISCST for receptors below plume height and
those from COMPLEX I for receptors above stack top.
3.6.2. Treatment of Terrain
COMPDEP is a merger of ISCST and COMPLEX I, and uses the method appropriate
to the elevation of the receptor being modeled. ISCST calculates an effective plume height
based on the relative difference in elevation between the stack base and the receptor. In
the ISCST component, any terrain heights above stack top are not considered in the
dispersion calculations. COMPLEX I calculates an effective height based on the relative
difference in elevation between the stack and the receptor, but applies a stability class
dependent correction factor to this adjustment. In this manner, terrain above stack top is
considered.
3.6.3. Stack-Tip Downwash
Stack-tip downwash refers to the modification of the physical stack height when
the stack gas exit velocity is less than 1.5 times the wind speed. In COMPDEP, the stack-
tip downwash and building wake effects options can be selected simultaneously. The
stack-tip downwash calculation can be used in determining the final plume height.
However, the actual stack height is used to determine the plume rise when building wake
effects are important.
3.6.4. Transitional/Final Rise
COMPDEP calculates both a final plume rise and a distance-dependent (transitional
plume rise). Users have the option of selecting which plume height will be used in
calculating concentrations. However, when building wake effects are being included, the
distance-dependent rise is always used (even if final rise is selected by the user).
3-19
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
3.6.5. Buoyancy Induced Dispersion
Users can choose to include the effects of buoyancy induced dispersion which
enhances the dispersion parameters as follows:
ze
-u
[3-3]
where:
cry,
-------
DRAFT -- DO NOT QUOTE OR CITE
3.6.7. Building Wake Effects
The method used in COMPDEP for determining building wake effects is the Huber-
Snyder method used in ISCST. Wake effects are important if the plume height, calculated
from the sum of the distance-dependent plume rise at a distance of two building heights
and the stack height (without stack-tip downwash effects), is greater than (a) 2.5 times
the building height or (b) the sum of the building height and 1.5 times the building width.
If, using these criteria, the plume is affected by the building wake, the dispersion
parameters are adjusted to account for the building effects and the distance dependent
plume rise is used in calculating concentrations at individual receptors. Because COMPLEX
I does not include building wake effects, COMPDEP does not allow the use of the ISCST
algorithms for building wake effects when modeling complex terrain receptors.
3.6.8. Dry Deposition
The dry deposition flux is calculated from the product of the air concentration (for
the fraction of the hour during which there was no precipitation) and the deposition
velocity. The code incorporated into COMPDEP is essentially the same as the original
CARB code (CARB, 1986) which is based on the work of Sehmel (1980) and Sehmel and
Hodgson (1978). The CARB algorithms represent empirical relationships for transfer
resistances as a function of particle size, density, surface roughness, and friction velocity.
In the CARB model, integrated resistances to mass transfer are computed within two
layers. In the first layer, which extends from one centimeter to one meter above the
surface, atmospheric turbulence dominates mass transfer. In the second layer, which lies
within one centimeter of the surface, the resistance to mass transfer is derived from
particle deposition measurements that were taken in a wind tunnel over various surfaces
using monodispersed particles.
The overall deposition velocity is given by Sehmel and Hodgson (1978) as:
3-21
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
vd =
1 - exp\
[3-5]
where:
vd
v
g
12
u
deposition velocity (cm/s)
gravitational settling velocity (cm/s)
atmospheric resistance integral (dimensionless)
surface resistance integral (dimensionless)
atmospheric surface layer friction velocity (cm/s)
The gravitational settling velocity is calculated as:
where:
Pp
Pa
g
V
10
0*00973
S
ID'
[3-6]
18
gravitational settling velocity (cm/s)
absolute air temperature (°K)
particle diameter (//m)
particle density (g/cm 3)
air density (g/cm 3)
acceleration of gravity (cm/s 2)
air viscosity (g/cm/s)
units conversion factor
The atmospheric resistance integral is based on surface flux profile relationships. For
neutral and stable conditions.
3-22
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
P-7]
where:
von Karman's constant
upper limit of atmospheric resistance integral (i.e., 100 cm)
lower limit of atmospheric resistance integral (i.e., 1 cm)
Monin-Obukhov length (cm)
For unstable conditions.
In = k rl (A^ + 1)(A0 - 1)1 ' * \tanAzl
[3-8]
where:
=1 ~
15 Zl
\0.25
and
15 z.
\0.25
The surface resistance integral is based on a least-squared empirical fit to the wind tunnel
data. For particles with diameters > 0.01
3-23
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
I3 = expl -378.051 + 16.498 InSc + B lnt+ - 12.804
[3-9]
where:
B =
-11.818 - 0.2863 lnt+ + 0.3226 In]-? - - 0.3385 ln( D
D
+ - 3.156xlO~13 (d uj2
and
Sc
v
D
zo
u .
10-4
Schmidt number (dimensionless)
air kinematic viscosity (cm 2/s)
Brownian diffusion coefficient (cm 2/s)
relaxation time (dimensionless)
particle diameter (pm)
surface roughness length (cm)
friction velocity (cm/s)
units conversion factor
The Brownian diffusion coefficient is expressed as:
3-24
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
where:
D
T
d.
D = (7.868xlO-10 — \(l + (°'02632\6.32 + 2.01 exp(-8.322 d )}} [3-10]
( dP>( ( dP ) )
Brownian diffusion coefficient (cm 2/s)
absolute temperature (°K)
particle diameter (//m)
Dry deposition velocities calculated by COMPDEP were compared with those
calculated by the Fugitive Dust Model (FDM) (Winges, 1990) which also uses the CARB
algorithms. There were no differences detected between the deposition velocities
calculated by the two models, so no modifications were made to the COMPDEP code. The
CARB algorithms were developed for application in flat terrain. These "flat terrain"
algorithms have not been specifically modified for handling deposition in complex terrain.
3.6.9 Wet Deposition
COMPDEP calculates the annual wet deposition flux according to a method
developed by Slinn (1984) and later modified by PEI and Cramer (1986). The wet
deposition flux is represented by Equation [3-11]:
I ' " -S -^
\/2ir a u n=i
where:
exp
exp
-0.5
Dyw =
f
[3-11]
wet deposition flux of pollutant (g/m 2/s)
fraction of the hour during which precipitation occurs
fraction of material removed per unit time in the n th particle size
category and jth precipitation intensity category (s "1)
3-25
November 10, 1993
-------
0(n)
K
Q
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-4. Examples of precipitation scavenging coefficients (per s) in COMPDEP.
Precipitation
Intensity
Heavy
Moderate
Light
Particle size category (jjM)
Less than 2
1 .46E-03
5.60E-04
2.20E-04
2 to 10
4.64E-03
8.93E-04
1 .80E-04
Greater than 10
9.69E-03
9.69E-03
9.69E-03
3.7.2. Meteorological Data Files
There are two meteorological files used by COMPDEP. The first is a file consisting
of hourly values for the wind speed, wind direction, stability class, mixing height, and
ambient air temperature, as shown in Table 3-6.
Each array contains 24 values, one for each hour. The file can be created using
either RAMMET (Catalano et al, 1987) or MPRM (Irwin and Paumier, 1990), which are
EPA meteorological preprocessors. These preprocessors use surface data (CD-144 file),
twice daily mixing heights, and on-site data . Both the CD-144 and mixing height files are
available from the National Climatic Data Center (NCDC).
The second file used by COMPDEP contains precipitation data (which is required only if
wet deposition is being calculated). Information contained in this file includes:
1. Date: described in terms of Julienne day and year
2. Hourly Precipitation data: Exact hourly quantities of hourly precipitation can be
obtained from the National Climatic Data Center TD3240 file. What is required for
COMPDEP, however, is hourly rainfall described in terms of intensity and precipitation
type. The intensity categories are: 1) none = 0.0 in/hr; 2) light = trace to 0.10 in/hr; 3)
moderate = 0.11 to 0.30 in/hr; and 4) heavy = greater than 0.30 in/hr. Precipitation type
3-27
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-5. Parameters in the COMPDEP control files.
For Each Run
Starting year
Starting Julienne day
Starting hour
Horizontal scale factor to convert user units to
kilometers
Vertical scale factor to convert user units to
meters
Pollutant half-life {s)
Test option
Terrain adjustment option
Stack-tip downwash option
Transitional plume rise option
Buoyancy induced dispersion option
Calms processing option
Dry deposition option
Wet deposition option
Building wake effects option
Anemometer height (m)
Array of wind speed profile exponents (by stability
class)
Array of terrain adjustment factors (by stability
class)
Distance limit for plume centerline above ground
(m)
Building height (m)
Building width (m)
Array of particle sizes Cum)
Surface roughness length (m)
For Each Run
Source east coordinate (user units)
Source north coordinate (user units)
Stack height (m)
Stack temperature (°K)
Stack diameter (m)
Stack gas exit velocity (m/s)
Stack ground elevation (user units)
Number of particle size categories (maximum
of 10)
Particle density (g/cm 3)
Emission rate (g/s)
Fraction of emission in each particle size
category
For Each Receptor
Receptor east coordinate (user units)
Receptor north coordinate (user units)
Receptor height above local ground (m)
Receptor ground elevation (user units)
For Each Particle Size Category
Scavenging coefficients for each precipitation
intensity (s"1)
3-28
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Table 3-6. Parameters in the meteorological data file for COMPDEP
For Each Station
For Each Day
5 digit station ID of surface data
Last 2 digits of beginning year of surface data
5 digit station ID of mixing height data
Last 2 digits of beginning year of mixing height
data
Last 2 digits of year
Month of year
Day of month
Array of Pasquill stability classes
Array of wind speed (m/s)
Array of ambient temperatures
(°K)
Array of wind directions (to
nearest 10 degrees)
Array of randomized wind
directions (to nearest
degree)
Array of mixing heights (m)
is described using a numerical indicator termed F, as: 1) steady — F = 1; 2) showers — F
= 0.50; 3) thunderstorm or squall — F = 0.25; 4) snow, ice, or none —F = 0.0.
3.8 MODEL OUTPUT
COMPDEP creates one output file that contains a verification of the model input
including options selected, source information, and receptor locations. The output of the
COMPDEP model for both surface deposition and ambient air impacts is a concentration
array of ground-level receptors in all directions around the source. It is recommended that
the minimum distance of the receptors be at 200 m from the stack, and that the maximum
distance be 50 km from the stack. It is recommended that this array include a minimum of
160 ground-level receptors consisting of 10 receptor points along each of 16 wind
3-29
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
directions every 22.5* on the polar azimuth. (Modeling experience has shown that most of
the exposures associated with stack emissions from a typical incineration process occurs
within 10 km, and that the maximum concentration is likely to occur within 1.0 km from
the source.) In specifying the radial distances of the receptors from the incinerator stack, it
is recommended that receptors be spaced every 200 m until 1 km is reached. It is
recommended that the receptors be spaced every 1 km between 1 km distance from the
source out to 5 km, and every 10 km between 10 km distance to 50 km from the source.
The printed output also includes tabulations of annual average and period (e.g. multi-year)
average values of dry deposition flux (g/m2-yr), wet deposition flux (g/m2-yr), combined
dry and wet deposition flux, and annual average ambient air concentration (//g/m3) at each
modeled receptor in all directions along the compass. In addition to the long-term average
concentration, the ISCLT portion of the COMPDEP model can produce concentrations
corresponding to hourly, daily and monthly time scales. These time-scales are: highest
concentration in 1-hr, 3-hrs, 8-hrs, and 24-hrs. These shorter time-frames are applicable to
the analysis of human health effects associated with acute exposures.
3.9 COMPDEP MODEL APPLICATIONS
3.9.1. Recommended Application of the COMPDEP Model
At the point of stack emission and during the atmospheric transport, partitioning
of the contaminants occurs between two physical phases: vapor and particle-bound
phases. Pollutants with higher vapor pressures increase the vapor phase fraction. A
vapor-to-particle ratio (V/P) needs to be established for each contaminant of interest. The
V/P ratio represents the ratio of the concentration of the contaminant in the vapor phase
to the concentration of that contaminant in the particulate phase, usually expressed in
terms of percentage with V plus P adding to 100 %, e.g., 60 %/ 40 %.
Particle-bound contaminants can be removed from the atmosphere by both wet
deposition (i.e., precipitation scavenging) and dry deposition (i.e., gravitational settling,
Brownian diffusion). The contaminants can directly deposit onto soil surfaces, plant
surfaces, in water bodies, and onto surrounding land. With respect to assessing potential
contamination of the food chain, there are two processes important to estimating above
3-30
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
ground vegetation contamination: airborne vapor phase absorption into plants, and air-
borne particle deposition onto soils and plant surfaces.
The analysis of human exposures via direct inhalation and contact with contaminated
soils and food requires certain outputs from the COMPDEP model to estimate exposure
media concentrations. At a minimum, these outputs should include:
1. An estimation of vapor-phase concentrations of the contaminants (in units of
//g/m3). The vapor phase concentration is used to estimate absorption into plants and
human exposure via direct inhalation of vapors.
2. An estimation of the wet and dry deposition flux of contaminated particles
(units of g/m2/yr). Deposition estimates are used for: plant concentrations (vegetations
for animal and human consumption), soil concentrations (soil contact exposures and crop
uptakes), and surface water concentrations (water and fish; via runoff and erosion of
watershed soils and direct depositions).
3. An estimation of the ambient air concentration of contaminants bound to
particles (units of//g/m3). This is important in estimating direct inhalation exposures to the
portion of the contaminants that is particle bound, and when added to the portion that
exists as vapor (from #1 above), will give a value relative to total inhalation exposures
(vapor-phase concentration plus the particle bound phase concentration).
It is apparent from these output requirements that air modeling must consider the
physical state of the pollutants, e.g., the portion of the emission that is vapor-phase and
the portion of the emission that is bound to particles. This ratio is likely to change from
the point of release at the stack to any point distant from the stack. In some instances, as
with semivolatile organics, the vapor fraction decreases as the gases cool and condense
onto particulates. In other instances, as with volatile elemental mercury, the contaminant
may continue to predominate as a vapor. With the current air model, it is not possible to
model changes in the physical state of contaminants (the V/P ratio) with distance from the
3-31
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
source. Instead, it is recommended that users assume that the V/P ratio of the pollutant
as it is emitted from the stack is equal to the V/P ratio of the contaminant as it may exist
in ambient air at ground-level. This assumption is recommended because: 1) for most
contaminants, the physical/chemical processes that may occur to transform this ratio in
the atmosphere from the point of release from the stack to the ground-level receptor are
largely unknown, 2) the model does not include procedures to address this phenomena,
and 3) ambient conditions are more representative of exposure conditions than stack
conditions.
In order to provide estimates of ambient air concentrations of vapors and estimates
of ambient air concentrations of particles combined with estimates of wet/dry particle
deposition flux, it is necessary that users run the COMPDEP model twice. Both model
runs should assume a "unit'emissions release rate", e.g., 1 g/s. Results from these unit
runs can easily be transformed to final outputs given assumptions on emissions in vapor
and particle forms. This will be illustrated below.
The recommended modeling procedure that will provide necessary outputs for
further exposure modeling is:
• Run 1: To estimate vapor-phase concentration of the contaminant in ambient air.
COMPDEP should be run with the wet/dry deposition switches turned to the "off"
position. This is to isolate the ambient air concentration of the contaminant in vapor-phase
from the calculation of wet and dry particle deposition flux. This inactivates a plume
depletion equation that subtracts out losses in ambient air concentration due to particle
deposition. What is left are the Gaussian dispersion algorithms.
With the "unitized" emission rate, one can reconstruct the actual predicted ambient
air concentration Oc/g/m3) of vapors by multiplying the "actual" vapor-phase emission rate
(g/s) by the "unitized" modeling result. For example, let the actual stack gas emission rate
of total (vapor plus particle components) contaminant be 1x10~5 g/s, and the V/P ratio
(expected under ambient conditions) be 60%V/40%P. Then the "actual" emission rate of
the vapor-phase portion of the contaminant is calculated to be 6x10"6 g/s (1x10~5 g/s *
0.6). If the "unitized" ambient air concentration at the ground-level receptor is estimated
3-32
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
by the COMPDEP model to be 1x10~8/yg/m3 (i.e., this concentration is predicted with a
unit emission rate of 1 g/s)> then the "actual" predicted air concentration at that receptor
can be estimated as:
(6x1 Q-6 g/s + 1 g/s) * 1x1CT8//g/m3 = 6x1Q-1Vg/m3
• Run 2: To estimate wet and dry particle deposition flux, and the ambient air
concentration of the contaminant that is particle-bound.
COMPDEP should be run with the wet/dry particle deposition switches turned to the
"on" position, and using a "unit emission rate" of 1 g/s. This second run is considered a
simulation of particle-bound contaminant only. Outputs of this run include unitized
deposition rate and unitized ambient air concentrations of particles.
Like the vapor-phase run, the "actual" deposition flux (g/m~2 -yr) and "actual"
particle-phase airborne concentrations can then be determined by multiplying the "actual"
emission rate (g/s) of the particle-bound portion of the total contaminant emissions by the
"unitized" modeling result at the ground receptor. For example, let the "actual" emission
rate of the particle-bound portion of the contaminant be 4x10"6 g/s, and the "unitized" dry
deposition flux at the ground receptor be 1x10'5 g/m2-yr. Then the "actual" predicted dry
deposition flux is 4x10'11 g/m2 (4x1 CT6 g/s -H 1 g/s * 1x10'5 g/m2-yr). Using this same
procedure, this second run provides the airborne concentration of contaminants bound to
particles (yc/g/m3).
• To estimate total ambient air concentrations of the contaminant that is vapor
phase and particle-bound to the contaminants.
Inhalation exposures will include both the concentration in vapor-phase, and the
concentration that is bound to particles in the ambient air. To make this estimate it is
necessary to add together the ambient air concentration of vapors (//g/m3) from the step 1
together with the ambient air concentration of particles (//g/m3) estimated in step 2.
3-33
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
3.9.2 Source Characteristics
After running COMPDEP with unitized vapor and particle emission rates, calculation
of the ambient air concentration and the wet/dry deposition flux of the contaminants at
points distant form the stack emission requires that the following be specified:
1. Emission rate of the vapor portion of the contaminant
2. Emission rate of the particle-bound portion of the contaminant
3. Particle size distribution of the particulate matter emissions.
The following subsections give guidance on making these estimations.
3.9.2.1. Vapor Phase/Particle Bound Phase Partitioning of Sernivolatile Organic
Compounds
In the previous section it was noted that the COMPDEP model is run twice with a
unit emission rate of 1 g/s for all the contaminants: 1) to derive the ambient air
concentration of the vapors, and 2) to derive the ambient air concentration of particles and
the estimates of wet and dry deposition flux of the particles. Reconstruction of the actual
concentrations of the contaminants requires multiplication of the unitized result by the
actual emission rate (g/s) of the contaminant assumed to be in the vapor-phase or that is
bound to particles under ambient conditions. This requires prior determination of the
partitioning of the total emission rate of the contaminant according to this same V/P ratio.
Bidleman (1988) has suggested a theoretical construct for estimating the V/P ratio
of semivolatile organic compounds in ambient air. Bidleman presents the theory that a
portion of the semivolatile organic compounds found in ambient air are freely exchangeable
between the vapor and particle phases. Bidleman defines a second portion, the
nonexchangeable fraction, as the quantity that is strongly and irreversibly adsorbed to
particulate matter, and is not at equilibrium with a corresponding vapor phase. Bidleman
cites an earlier model by Junge (1977), a theoretical model based on adsorption theory.
Junge's model mathematically describes the exchangeable fraction of the semivolatile
organic compound adsorbed to aerosol particles as a function of solute saturation vapor
3-34
November 10, 1993
-------
DRAFT --DO NOT QUOTE OR CITE
pressure and total surface area of atmospheric aerosol particles available for adsorption.
This is given by Equation [3-12].
where:
0
c
0 = c ST/(p + c ST)
[3-12]
adsorbed fraction, unitless
constant developed by Junge, 1.7 E-4 atm-cm
total surface area of atmospheric aerosols in relation to total volume
of air, cm2/cm3
solute saturation vapor pressure, atm
-1
Equation [3-1 2] requires an estimate of St, which is the total surface area of aerosol
particles in the ambient air. Bidleman (1988) provides estimates of average total surface
areas of aerosol particles relative to average total volume of air (cm2/cm3) citing a study
by Whitby (1978). In addition , Whitby estimated the average total volume of aerosol
particles per volume of air (VT = cm3 particles/cm3 of air). Whitby's (1978) calculations
varied according to the density of aerosol particles in the ambient air in different air sheds.
These were as follows (units of ST of cm2/cm3, VT of cm3/cm3):
• Clean continental background
• Average background
• Background plus local sources
• Urban
ST = 4.2 E-7
ST = 1 .5 E-6
ST = 3.5 E-6
ST = 1.1 E-5
VT = 6.5 E-1 2
VT = 3.0 E-1 1
VT = 4.3 E-1 1
VT = 7.0 E-1 1
For purposes of calculating the fraction of the semivolatile organic compound that
adsorbed (using Junge's equation) onto the surface of aerosol particles in ambient air,
Whitby's values for ST (above) are recommended. For incinerators located in urban areas,
ST corresponding to the air-shed classification "Background plus local sources" should be
appropriate. The fraction adsorbed to the surface of particles is subtracted from 1 (100
%) to derive the fraction existing in the vapor phase under ambient conditions. In this
3-35
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
manner, the vapor phase/particle-bound phase of the semivolatile organic compound is
derived. This method is most appropriate for organic compounds with low vapor
pressures, e.g., dioxins, PCBs, chlorinated benzenes, chlorinated phenols).
This approach is applied only in the context of providing a reasonable estimate of
the V/P partitioning in ambient air. As discussed earlier, the V/P ratio of the contaminant
estimated in ambient air is assumed to be equal to the V/P ratio of the contaminant at the
point of emission from the stack. Therefore the recommended procedure is to apply the
Bidleman (1988) approach to estimate the V/P ratio of the contaminant in ambient air,
then apply this same V/P ratio at the stack to partition emission the emission rate of the
contaminant between the two phases. Each phase of the contaminant, then, is modeled
independently. An area of uncertainty with respect to this simplified approach is that it
does not account for any changes in the V/P ratio from the point of release to the ground
as contaminants disperse through the atmosphere.
With regard to the more volatile compounds (e.g., volatile organics and elemental
mercury), based on the Henry's Law constant and saturated vapor pressures, one can infer
that the compounds will partition primarily into the vapor phase.
3.9.2.2. Estimation of the In-Stack Particulate Matter Particle-Size Distribution
The COMPDEP air model contains algorithms to compute the rate at which dry and
wet removal processes may deposit particulate matter emitted for the stack of the
combustor to the surface of the Earth. This requires an initial estimation of the distribution
of particulate differentiated on the basis of particle diameter before the COMPDEP program
can predict deposition flux.
The diameters of small particles comprising particulate matter in stack emissions
are usually measured in units of one millionth of a meter (micrometer, commonly called
micron), abbreviated by the letters /vm. The distribution of particulate matter by particle
diameter will differ from one combustion process to another, and is greatly dependent on
such factors as: the type of furnace and design of the combustion chamber; the
composition of the feed/fuel; the particulate matter removal efficiency and design of the air
pollution control equipment; the amount of air in excess of stoichiometric amounts that is
3-36
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
used to sustain combustion; and the temperature of combustion. Given these variables,
the particle size distribution cannot be calculated, it can only be directly measured and
inferred from prior data. Unfortunately few studies have been performed to directly
.measure particle size distributions from a variety of stationary combustion sources. Stack
measurements over a wide range of particle diameters on a site-specific basis is preferred.
These data should reflect a wide range of operating conditions, and the expected
performance of the air pollution control equipment in removing the mass of particulates
generated during combustion. In recognition of the fact that accurate and direct
measurements of particle-size distributions at a combustion facility under study often
times will be lacking, it is necessary to resort to a "generalized" assumption of the
distribution for purposes of deposition analysis using the air model. The following
procedure is recommended as a 'default' assumption of the particle size distribution to be
applied only in situations where site-specific measurements are not available. In addition
to this 'default' distribution, a procedure for apportioning the emission rate of the
contaminant that is particle-bound to the particle size distribution is given.
The generalized particle size (diameter) distribution in Table 3-7 may be used as a
default for combustion facilities equipped with either electrostatic precipitators (ESPs) or
fabric filters (FFs), because the distribution is fairly typical of particle size arrays that have
been measured at the outlet to advanced equipment designs (Buonicore and Davis, 1992).
In terms of data preference, however, site-specific particle size distributions are highly
recommended over use of the default.
When using either the default assumption or the actual distribution, the emission
rate of the particle-bound portion of the total stack emission of the contaminant is
apportioned to the particulate distribution. In high-temperature combustion processes
characteristic of hazardous waste incinerators and municipal waste combustors, the
contaminates will leave the combustion zone in the vapor phase. In other situations,
molecules leaving the furnace area will undertake chemical reactions in the combustion gas
stream to reform new molecules in the vapor phase. Once the combustion gases have
cooled after passing through a boiler, a heat economizer, or a dry/wet scrubber device,
condensation of the vapor phase chemicals may occur. The pollutants will then condense
3-37
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 3-7. Generalized particle size distribution (jjm), and proportion of available surface
area, to be used as a default in deposition modeling if site-specific data is unavailable.
Particle
Diameter
Ct/m)a
>15.0
12.5
8.1
5.5
3,6
2.0
1.1
0.7
<0.7
Particle
Radius
(fjm)
7.50
6.25
4.05
2.75
1.80
1.00
0.55
0.40
0.40
Surface
Area/
Volume
0.400
0.480
0.741
1.091
1.667
3.000
5.455
7.500
7.500
Fraction of
Total
Weight
0.128
0.105
0.104
0.073
0.103
0.105
0.082
0.076
0.224
Proportion
Available
Surface
Area
0.0512
0.0504
0.0771
0.0796
0.1717
0.3150
0.4473
0.5700
1.6800
Fraction
of Total
Surface
Area
0.0149
0.0146
0.0224
0.0231
0.0499
0.0915
0.1290
0.1656
0.4880
Sum: 3.4423
Notes: a. Geometric mean diameter in a distribution from EPA (1980).
to the surface of particles entrained in the combustion gas. By this paradigm, the
apportionment of emissions by particle diameter becomes a function of surface area of the
particle that is available for chemical adsorption. Therefore the first step to this
apportionment is to calculate the proportion of available surface area of the particles. If
particle density is held constant (i.e., 1 g/m3), then the proportion of available surface area
of aerodynamic spherical particles is the ratio of surface area to volume as follows:
(a) Assume aerodynamic spherical particles.
(b) Specific surface area of a spherical particle with a radius, r:
S = 4 nr2
(c) Volume of a spherical particle with a radius, r:
V = 4/3 /7r3(d) The ratio of surface area (S) to volume (V) is:
3-38
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
S/V = 4 OT2/ (4/3 m3)
S/V = 3/r
The following uses the particle size distribution in Table 3-7 as an example of
apportioning the emission rate of the particle-bound portion of the contaminant to the
paniculate array. This procedure can be followed for apportioning actual emissions to the
actual particle size distribution as measured at the stack. In Table 3-7, a spherical particle
having a diameter of 15 //m (column 1) has a radius of 7.5 //m (column 2). The proportion
of available surface area (assuming particle density is constant) is 0.400 (S/V = 3/7.5),
which is the value in column 3. Column 4 shows that 12.8 % (by weight) of the total
mass of particles is 15 //m. Multiplication of column 3 by column 4 gives an approximate
calculation of the relative proportion of total surface area based on the percent of particles
that are 15 //m diameter. Summation of column 4, relative proportion of surface area,
gives an estimate of the total surface area of all the particles in the particle size
distribution. In this example the sum is 3.44//m2 of total surface area available for
chemical adsorption. The last column is the fraction of total surface area represented by
the specific particle diameter in the distribution, and is calculated by dividing the relative
proportion of surface area for a given diameter (column 4) by the total surface area (3.44
//m2). In the example of the 15 //m diameter particle, the fraction of total surface area
available for adsorption is 0.0149 (0.0512 //m2/3.4423 //m2). Multiplication of the
emission rate of the contaminant that is particle-bound (//g/s) by the fraction of total
surface area for a given particle diameter gives an estimate of the emission rate of the
contaminate apportioned to that particle diameter. For example, if it is assumed that the
total emissions of 2,3,7,8-TCDD = 1 x 1Q~5//9/s, and 45% of this emission rate, or 4.5 x
10"6//m, is particle-bound, then approximately 6.7 x 10"8/yg 2,3,7,8-TCDD/s is
apportioned to the 15 //m diameter particles (4.5 E-6 //m/s x 0.0149). This procedure is
then repeated for all particle sizes in the array.
3.10 METEOROLOGICAL DATA
The Guideline on Air Quality Models (EPA, 1993a) recommends the use of five
years of meteorological data for making long-term estimates of ambient air concentrations.
3-39
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
However, it is recognized that 5 years of continuous meteorological measurements may
not be available in all cases. In those situations, it is recommended that a minimum of one
year of preprocessed meteorological data be used. It is important that the meteorological
data be representative of local conditions nearest the site. Therefore the data should have
been collected from the general vicinity of the combustor, and hourly data should be used.
Additionally, if meteorological data are not available from the general vicinity of the
combustor, then data from the nearest meteorological station should be used. This data
should be representative of site conditions both spatially and climatologically
(i.e., temporally). Meteorological data that are available from the National Weather Service
may be suitable, although on-site meteorological data is preferred. The required types of
data and appropriate measurement techniques for collecting site-specific meteorological
data are discussed in "On-Site Meteorological Program Guidance for Regulatory Modeling
Applications" (EPA, 1987a). Meteorological data includes measurements of wind speed,
wind direction, ambient temperatures, cloud cover, ceiling heights, mixing heights, sky
cover, atmospheric stability, atmospheric turbulence, hourly precipitation amounts,
intensity of precipitation events and any other data needed in the air modeling exercise.
When on-site data are used, the following information should be included: a) identification
of the meteorological station site, and who maintains the station; b) the threshold for wind
speed measurements; c) the method for collating and reducing the data; d) details on the
the number and percentage of missing values in the meteorologic record; e) if applicable,
details on the statistical methods applied to fill-in data gaps and missing values. In all
cases, the EPA Regional staff meteorologist should be consulted when selecting
meteorological data for dispersion and deposition modeling or when developing or
evaluating a meteorological monitoring program.
3-40
November 10, 1993
-------
DRAFT - DO IWOT QUOTE OR CITE
4. CALCULATING SOIL CONCENTRATIONS
4.1. Introduction
Issue Indoor soil/dust concentrations may be different than outdoor soil levels.
Conclusions/Recommendations
Contaminants from sources outside the home may be brought into the home by
airborne dust or tracking. The resulting contaminant levels may be diluted with other
indoor dusts. Contaminants which degrade outdoors due to photolysis or biological
processes, may be more persistent indoors. Exposure may result via mouthing of objects
with dust residues, dermal contact with dust residues or inhalation of suspended dusts.
Procedures for assessing such exposures are not well established. Recommendations can
be made to measure indoor dust to monitor contaminant levels. Cleaning procedures can
be implemented to reduce exposures. For assessing impacts to indoor dust, it is
recommended that the assumption be made that indoor dust originated as outdoor soil.
Therefore, soil concentrations as estimated for outdoor conditions should be used as
indoor dust concentrations. Furthermore, the soil concentration that should be used are
the ones pertinent to the 1-cm "untilled" depth.
4.2.1. Calculating Cumulative Soil Concentration
Issue The soil concentration algorithm considers deposition of particle-bound
contaminants but does not consider vapor phase depositions or diffusions.
From the Indirect Exposure Document:
Sc _ (Dyd + Dyw) [1.0 - exp(-ks Tc)J 100
Z BD ks
[4-1]
where:
4-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Sc
Dyd
Dyw
ks
Tc
100
Z
BD
soil concentration of pollutant after total time period of
deposition (pg pollutant/g soil)
yearly dry deposition rate of pollutant (g pollutant/m 2/yr)
yearly wet deposition rate of pollutant (g pollutant/m 2/yr)
soil loss constant (yr~1)
total time period over which deposition occurs (yrs)
units conversion factor
soil mixing depth (cm)
soil bulk density (g/cm 3)
(Note that the units for soil concentration in Equation [4-1] have been changed
from mg/g to //g/g, with a corresponding change in the units conversion factor.)
Conclusions/Recommendations
Volatilization loss is considered in the overall soil loss constant, with the rate
constant due to volatilization equal to an equilibrium coefficient multiplied by a gas phase
mass transfer coefficient (cm/s); see Equations (4-4) through (4-7) in the IED. Although
the derivation for this rate constant was not given in IED, it has been independently
verified as accurately representing volatilization loss (although the equation for the gas
phase mass transfer coefficient, Kt of Equation (4-6), appears to be an empirical fit to data
and is not referenced; it will be assumed to be appropriate for this application). If one
assumes that the diffusive loading is driven by the same gas phase mass transfer
coefficient (K^ as volatilization, than one can solve for the loading term as:
where:
LDIF =
Kt
C,
-va
LDIF = 0.31536 KtCm
atmospheric diffusion flux to soil (g/m2-yr)
gas phase mass transfer coefficient (cm/s; see Eq [4-6], IED)
gas phase air concentration (//g/m3)
[4-1]
4-2
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
The vapor phase air concentration that is appropriate for this formulation is the vapor
phase concentration predicted by application of the air dispersion model. Chapter 3 of this
Addendum further details application of the COMPDEP model to estimate vapor phase
concentrations.
It is recommended that this diffusive term be added to the overall soil concentration
equation as:
Sc =
(Dyd + Dyw + LDIF) [1.0 - exp(-ks Tc)J 100
Z BD ks
[4-1]
Finally, it should be noted that this formulation pertains to diffusive flux into the soil and
does not address entry of vapor-phase contaminants via wet deposition. It is also not
clear as to what extent this treatment of diffusive entry addresses the issue of "dry
deposition" of vapors onto soils. The issue of vapor-phase impacts to soils remains a topic
for future recommended research, as discussed in Part III of this. Addendum.
4.2.2 Calculating the Soil Loss Constant
Issue The soil loss constant, ks of Equation [4-2] of the Indirect Exposure Document,
does not include pollutant losses from soil erosion and surface runoff. For a small
land area within a watershed, it could be argued that the soil loss constant does not
need to consider such losses if whatever erodes or runs off in the downgradient
direction from a site of concern (i.e., a farm where indirect exposures occur) is
matched by an equal amount which erodes or runs onto it from upgradient areas.
On the other hand, for the study region as a whole, as well as individual
watersheds, losses due to soil erosion and surface runoff are important and need to
be accounted for. These losses are incorporated in Chapter 9 in the soil erosion
loss constant, kse, and the surface runoff constant, ksr (see Eq [9-4], Eq [9-6], and
Eq [9-7] of this addendum).
4-3
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
From the Indirect Exposure Document:
ks = ksl + ksg + ksv
[4-2]
where:
ks
ksl
ksg =
ksv =
so/7 foss constant due to all processes (yr~1)
loss constant due to leaching (yr'1)
loss constant due to degradation (abiotic andbiotic) (yr'1)
loss constant due to volatilization (yr'1)
Conclusions/Recommendations
It should be remembered that the soil concentration discussed in Chapter 4 of the
Indirect Exposure Document is specific to a site where exposures occur, whereas for
Chapter 9, the soil concentration is for the watershed as a whole. A watershed is likely to
have subareas that are highly eroded such that erosion is a significant loss, subareas that
are depositional where soil erosion is in fact a source term rather than a dissipation term,
and subareas where, as the Issue discussion notes above, there may be little net loss via
runoff or erosion. There will be a net loss via erosion to "average" watershed soils, hence
the consideration of these terms for soil concentration in the surface water algorithm.
An assessor can use the current formulation in Chapter 4, but the exposure site soil
concentration estimated in this way should be considered "worst case" or "hot spot",
because the loss terms of runoff and erosion will not be considered. Alternately, the soil
loss constant due to all processes, ks, can include both a soil erosion loss constant and a
surface runoff loss constant. This is consistent with the approach taken in Chapter 9 of
this addendum, and best represents average soil concentrations within a watershed:
ks - ksl + kse + ksr + ksg + ksv
[4-2]
4-4
November 10, 1993
-------
DRAFT -- DO MOT QUOTE OR CITE
where:
ks
ksl
kse =
ksr =
ksg =
ksv =
soil loss constant due to all processes (yr "1)
loss constant due to leaching (yr~1)
loss constant due to soil erosion (yr"1)
loss constant due to surface runoff (yr"1)
loss constant due to degradation (yr"1)
loss constant due to volatilization (yr"1)
Issue The loss constant due to leaching, ksl of Equation [4-3] in the Indirect Exposure
Document, does not properly account for surface runoff.
From the Indirect Exposure Document:
where:
ksl
P
I
Ev
6
Z
BD
Kd.
ksl =
P +1 -Ev
0 Z [1.0 + (BD Kds /Q)J
/oss constant due to leaching (yr'1)
average annual precipitation (cm/yr)
average annual irrigation (cm/yr)
average annual evapotranspiration (cm/yr)
soil volumetric water content (mL/cm 3)
soil depth from which leaching removal occurs (cm)
soil bulk density (g/cm 3)
soil-water partitioning coefficient (mL/g)
[4-3]
Conclusions/Recommendations
There is a need to include a runoff term, R, in the numerator of Equation [4-3] for
calculating the loss constant due to leaching, ksl, as follows:
4-5
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
ksl =
P +/-/?-£„
0 Z [1.0 + (BD Kds
[4-3]
An annual runoff amount can be estimated given isopleths of surface runoff
available in the "Water Atlas of the United States" (Gerahty, et al, 1973). The water
atlas' isopleths give annual surface water contributions, which include interflow and
ground water recharge, and they would need to be adjusted downward to reflect surface
runoff only. Another approach for estimating surface runoff is to use the Curve Number
Equation developed by the U.S. Soil Conservation Service. Isopleths of mean annual
cropland runoff corresponding to various curve numbers have been developed by Stewart
et al (1975), as reported by U.S. EPA (1985b). Curve numbers are assigned to an area
based on the soil type, land use or cover, and the hydrologic condition of the soil.
Issue Soil concentration depletion due to volatilization is modeled as a means of limiting
soil concentration. However, this mass flux never experiences rainout or washout
and subsequent redeposition. The result is an underestimate of soil concentrations
for volatile soluble compounds.
Conclusions/Recommendations
The reposition of volatilized residues of semi-volatile organic contaminants such as
dioxin is very small, and can generally be ignored. This may not be the case for more
volatile compounds, and more research is recommended. No changes to the soil
concentration algorithm to include redeposition of volatilized residues are recommended at
this time.
4.3.3 Soil Depth
Issue The concept of two mixing depths, 20 cm for tilled soils and 1 cm for untilled soils,
needs to be more fully explained.
Conclusions/Recommendations
Most literature has placed the untilled depth for atmospheric particles depositing
4-6
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
at 1 or 2 cm, with some literature on deposition of radionuclides in unfilled situations at
5 cm. The tilled mixing depth of 20 cm is used in the dioxin exposure document (EPA,
1992c) and elsewhere. An "untilled" soil concentration is one where the soil is not
disturbed, as in pasture grass or lawn settings. It is used for dermal exposures and soil
ingestion exposures. "Tilled" soil concentrations are used for the soil-to-crop algorithms
(i.e., vegetables/fruits for human consumption, or hay/silage in the beef/milk food chain
algorithm). Using the untilled soil concentration for soil-related exposures is a conservative
assumption. If instead, it is felt that dermal exposures or soil ingestion exposures occur in
conjunction with gardening or agricultural field crop production, than the tilled soil
concentration should be used.
For soluble compounds, leaching might lead to movement below 1 cm and perhaps
justify a mixing depth greater than 1 cm. However, no change to account for this is
recommended at this time.
Issue Soil/dust concentrations on hard surfaces may have an effective mixing depth of
less than 1 cm.
Conclusions/Recommendations
Deposition on to" hard surfaces may result in dust residues which have negligible
dilution with other residues. Examples may include sidewalks, slides, swings and other
play areas at schools. Exposure may result via mouthing of these objects or dermal
contact. Estimates of the ingestion of such residues or amounts transferred to skin have
not been well established. Thus, such exposures are difficult to assess quantitatively.
Recommendations can be made to measure residue levels on surfaces to monitor
contaminant levels. Cleaning procedures can be implemented to reduce exposures. If an
exposure assessment is to include contact with dust that is on hard surfaces, it is
recommended that the concentration of contaminants on that dust be assumed to be equal
to soil concentrations that are estimated for the "untilled" depth, which is in this guidance
is recommended to be 1 cm.
4-7
November 10, 1993
-------
-------
DRAFT - DO NOT QUOTE OR CITE
5. DETERMINING EXPOSURE THROUGH THE TERRESTRIAL FOOD CHAIN
5.2 Calculating Concentration of Pollutant in Plants
The approach which sums vapor transfers, particle depositions, and root uptake to
estimate total plant concentrations is an appropriate one. However, there are issues with
each pathway.
5.2.1 Plant Pollutant Concentration Due to Root Uptake
Issue The Travis and Arms approach seems to be mixing and matching various chemicals
and environmental settings to come up with an empirical transfer coefficient (Br).
Lipophilic non-pesticidal compounds are included with soluble pesticides. The
mechanisms for soil to plant transfers are different for the two classes of
compounds. Also, for the process of root uptake, the characterization of potatoes
and root vegetables as "protected" is inappropriate. The idea was that root
vegetables were protected from vapor transfers and particle depositions. The
Travis and Arms algorithm which assigns a value to Br (plant-soil bioconcentration
factor) is inappropriate for potatoes and root vegetables. The Travis and Arms
relationship is for soil to above ground plant transfers (e.g., translocation into
shoots), not soil to below ground transfers (e.g., to roots and tubers).
From the Indirect Exposure Document:
Pr, = Sc Br/
[5-2]
where:
Pr,
Sc
concentration of pollutant in ith plant group due to root uptake
(jjg pollutant /g plant tissue, dry weight fDWJ)
soil concentration of pollutant after the total period of
deposition (fjg pollutant/g soil)
5-1
November 10, 1993
-------
Br,
DRAFT - DO NOT QUOTE OR CITE
plant-soil bioconcentration factor for the ith plant group
(ffjg pollutant/g plant tissue DWJ/f/jg pollutant/g soil])
The bioconcentration factor is given as:
log Br = 1.588 - 0.578 log
[5-3]
where:
K
ow
octanol-water partition coefficient
Conclusions/Recommendations
A better approach to vegetables for which roots are the edible portions, is to
separate "above ground" and "below ground" vegetation, and state that below ground
vegetation includes edible roots such as carrots, potatoes, radishes, etc. This is the
approach taken in the dioxin exposure document (EPA, 1992c). It should be noted that
concentrations of dioxin in carrots and potatoes are among the highest found in
vegetables/fruits.
An approach developed by Briggs should be used in place of the Travis and Arms
algorithm, for soil to below ground transfers to root vegetables. The Briggs approach uses
a Root Concentration Factor (RCF) to estimate plant concentrations (Briggs, 1982).
Recommended revisions to Equations [5-2] and [5-3] of the IED are:
ScRCF
[5-2]
and:
where:
log (RCF - 0.82) = 0.77 log Kow - 1.52
[5-3]
5-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Sc
Kds =
RCF =
K
ow
concentration of pollutant in below ground plant parts due to root
uptake (JJQ pollutant/g plant tissue, fresh weight [FW])
soil concentration of pollutant (>ug pollutant/g soil)
soil-water partition coefficient (mL/g)
ratio of concentration in roots to concentration in soil pore water
([//g pollutant/g plant tissue FW]/[//g pollutant/mL pore water])
octanol-water partition coefficient (dimensionless)
The dioxin document (EPA, 1992c) uses an empirical correction factor, to reflect
the fact that the barley roots of Briggs' experiments are different than bulky below ground
vegetables such as potatoes or onions. Dioxin-like compounds will sorb to outer portions
of below ground vegetation, but inner translocation is negligible. Therefore, the
concentrations in the barley roots, for the lipophilic compounds used by Briggs in his
experiments, would be higher on a whole plant basis than concentrations that would result
from soil to root transfers for bulkier vegetation. In other words, the whole barley root
concentrations would be similar to concentrations near the skin of a potato, but not a
whole potato. The dioxin document uses an empirical reduction factor of 0.01, which was
estimated using a surface area volume to whole plant volume ratio for a carrot. A similar
fix would not be appropriate for an RCF developed for soluble compounds, which would
have transpiration stream translocation and more uniform vegetative concentrations.
For the long-term, other approaches should be reviewed for root uptake of
contaminants from soil. Also, for soil to above ground plant transfers, consideration might
be given to the Transpiration Stream factor developed by Brigg's as a simple approach for
more water soluble compounds as a possible alternative to the Travis and Arms algorithm.
The principal cause for soluble pesticides to impact above ground parts of plants from soil
7>
contamination is through the transpiration stream.
The magnitude of transfer by lipophilic compounds has been shown to be much
lower than soluble compounds. Given this, Travis' approach may be acceptable as the
transfer factor does increase with decreasing Kow, i.e., it may appropriately reflect an
observed trend for chemicals with increasing Kow. No change is therefore recommended.
5-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
and one should use the Travis and Arms Br for soil to above ground transfers for all
contaminants for which a Kow is known.
5.2.2 Plant Pollutant Concentration Due to Direct Deposition
Issue The basic framework given by Equation [5-4] is sound. However, the validity of
including an "Fw" and its default assignment of 0.02 are questioned. This
essentially zeroes out wet deposition impacts to plants.
From the Indirect Exposure Document:
1000 [Dyd + (Fw Dyw)] Ftp/ [1.0 - expf-kp Tp/JJ
Pd,=
Yp, kp
[5-4]
where:
Pdt
1000 =
Dyd =
Fw =
Dyw =
Rp, =
kp
Tp, =
Yp, =
concentration of pollutant due to direct deposition in the Ith
plant group fog pollutant/g plant tissue DWJ
units conversion factor
yearly dry deposition rate (g pollutant/m 2/yr)
fraction of wet deposition that adheres to plant surfaces
(dimensionless)
yearly wet deposition rate (g pollutant/m 2/yr)
interception fraction of the edible portion of the plant tissue for
the ith plant group (dimensionless)
plant surface loss coefficient (yr ~1)
length of the plant's exposure to deposition per harvest of the
edible portion of the ith plant group (yrsj
yield or standing crop biomass of the edible portion of the ith
plant group (kg DW/m 2)
5-4
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Conclusions/Recommendations
Setting Fw at 0.02 greatly diminishes the impact of contaminants depositing via
wet deposition. (The reference for this assignment is a memorandum which upon review
has shown not to be technically rigorous.) The dioxin exposure document (EPA, 1992c)
does not use an Fw value of 0.02 for estimating plant impacts; total deposition of
contaminant onto plants is the sum of wet and dry deposition. Several other multimedia
exposure assessments for dioxin include wet deposition in a "total" deposition term when
estimating impacts to vegetation, using the same model as the Indirect Exposure
Document. In effect, Fw is assigned a value of 1 for these dioxin exposure assessments.
However, there is one dioxin exposure assessment which uses a parameter "b", which is
the same as Fw and which was evaluated as having a value between 0.1 and 0.3
(McKone and Ryan, 1989).
A value of 1 for Fw is recommended when applying the methodology from the
Indirect Exposure Document to hydrophobic contaminants such as dioxins, PCBs, or other
organic contaminants whose log Kow exceeds 3.0. This is, at least, a conservative
assumption taken by other researchers and justified by the strong sorptive tendencies of
this class of organic chemicals to organic material. For metals and soluble organic
chemicals, those with log Kow less than 3.0, it does make intuitive sense that they would
not adhere significantly to vegetation while falling dissolved in rainwater. For these
chemicals, an Fw of 0.10 is recommended.
Issue From Equation [5-4], kp is usually described as a particle weathering factor and has
been assigned a value of 18 yr ~1, equivalent to a half-life of 14 days.
Conclusions/Recommendations
The 14 day half-life value corresponds to physical processes which remove particles
and does not consider chemical degradation. As such, a higher rate constant (i.e., a
shorter half-life) may be justified to consider chemical degradation. In the Indirect
Exposure Document, the value of 126.5 yr "1 for benzo(a)pyrene, corresponding to a 2 day
half-life, appears to be based on an atmospheric degradation half-life. A value of 14 days
is recommended for pollutants which do not degrade or degrade very slowly, such that kp
5-5
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
is based on physical processes removing particles from plants.
5.2.3 Plant Pollutant Concentrations Due to Air-to-Plant Transfer
Issue The algorithm for estimating vapor transfers to vegetations described in the IED as
Equation [5-13] is problematic and should be replaced.
From the Indirect Exposure Document (p 5-16)
= ffFv Cy) + CVJ By,-
Pa
[5-13]
where:
Pvf
Fv
Cy
Cv
Bvf
concentration of pollutant due to air-to-plant transfer in the i
plant group ffjg pollutant/g plant tissue DW)
fraction of pollutant air concentration present in the vapor
phase (dimensionless)
concentration of pollutant in air due to direct emissions
(jjg pollutant/m 3)
concentration of pollutant in air due to vaporization from
soi/ffjg pollutant/m 3)
air-to-plant biotransfer factor for the ith plant group
([ug pollutant/g plant tissue DWJ/fjjg pollutant/g air])
density of air (g/m 3)
th
Conclusions/Recommendations
First, Fv is inappropriate and should be deleted from the equation, and Cy should be
more precisely defined as, "vapor-phase concentration of pollutant in air due to direct
emissions". Recall that the air dispersion modeling includes two computer runs, one which
5-6
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
generates vapor phase air concentrations, and one which generates particle phase
concentrations and depositions. Therefore, an Fv is inappropriate for this framework
since a vapor phase concentration will already be available. An Fv would be appropriate if
air modeling assumed the entire emissions were in vapor phase and modeled air
concentrations at exposure sites. In this case, a total concentration would need to be
reduced to estimate only that amount in the vapor phase, and an Fv would be appropriate.
Second, the use of a vapor phase concentration which resulted from soil
volatilization is inappropriate. Earlier in Section 5.2.1., it was discussed that the Travis
and Arms relationship is only appropriate for soil to above ground plant transfers. It should
be assumed that the Travis and Arms empirical factor, the Brj, models all soil to above
ground transfers, which includes transpiration stream translocation, dust resuspension
followed by settling onto plants, and volatilization followed by vapor transfers. Including
Cv while also using the Travis and Arms Bi'j is double counting.
With these two changes, the plant concentration due to vapor transfers is simplified
as:
Pv, =
Cy
[5-13]
pa
Issue Serious and legitimate concern has recently been raised in the use of the Bacci
Algorithm for vapor transfers of dioxin. A recent article by in Environmental
Science and Technology (McCrady and Maggard, 1993) demonstrated that the
Bacci factor will significantly overestimate transfers of 2,3,7,8-TCDD vapor to
grass leaves.
From the Indirect Exposure Document:
5-7
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
log (Bv H) = -0.93 + 1.14 log K,
ow
[5-15]
where:
Bv
H
air-to-leaf biotransfer (fjjg/g DWJ/fjjg/gJ)
Henry's Law constant (Pa-m 3/mol)
octanol-water partition coefficient (dimensionless)
(Note: 1 atmosphere is equivalent to 9.87x10 "6 Pascals.)
Conclusions/Recommendations
First, the Bacci empirical relationship given above was published in a 1990 article
(Bacci, et al., 1990). It included the results of experimentation on 10 chemicals. A more
recent publication by Bacci and coworkers (Bacci, et al., 1992) included results from four
additional chemicals, for a total of 14 chemicals, and was also presented in a more
generalized volume/volume transfer basis rather than mass/mass transfer basis:
log Bvo, = 1.065 log Kow - log ( -£L ) - 1.654
n /
[5-15a]
where:
Bvoj = Bacci volumetric air-to-leaf biotransfer factor, unitless [(jug
contaminant/L of wet leaf)/(//g contaminant/L air)]
Kow = contaminant octanol water partition coefficient, unitless
H = contaminant Henry's Constant, atm-m3/mol.
R = ideal gas constant, 8.205 x 10~5 atm-m3/mol-deg K
T = temperature, 298.1 K
-1.654= empirical constant
Bacci, et al. (1990) showed that the volumetric transfer factor can be transformed to a
5-8
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
mass-based transfer factor by assuming that 70% of the wet leaf is water, that the leaf
density is 890 g/L, and given an air density is 1.19 g/L:
where:
B
Vol -
_ (1.19 g/L) gw/
(0.3) (890 g/L)
[5-1 5b]
mass-based air-to-leaf biotransfer factor, unitless [(//g contaminant/kg
plant dry)/(//g contaminant/kg air)]
Bacci volumetric air-to-leaf biotransfer factor, unitless [(//g
contaminant/L of wet leaf)/(/yg contaminant/L air)]
Bacci's experiments were conducted under conditions which would not account for
photodegradation of his test chemicals from the leaf surfaces. A recent study by McCrady
and Maggard (1993) which investigated the uptake and photodegradation of 2,3,7,8-
TCDD sorbed to grass foliage suggests a significant difference in experimental Bvo! for
grass plants. The authors note that the log Bvo, for 2,3,7,8-TCDD and azalea plants, using
Bacci's empirical relationship, was estimated as 8.5. The experimental log BVO| for
2,3,7,8-TCDD and grass plants reported by McCrady was 6.9 when photodegradation was
accounted for, and 7.5 in the absence of photodegradation. Since the photodegradation
experiments by McCrady best represent outdoor conditions, their work suggests that the
air-to-leaf transfer factor estimated by Bacci's algorithm may be 40 times too high for
vapor-phase transfer of 2,3,7,8-TCDD onto grass leaves.
An issue discussed by McCrady is that the theoretical time for the grass tissue to
reach a steady state is much shorter than that indicated in the Bacci experiments. Using
Bacci's results, McCrady noted that the azalea leaves theoretically take greater than 400
days to reach equilibrium, in comparison to less than 20 days to reach equilibrium for the
grass plants in their experiments. This difference is not entirely due to photodegradation.
Thus, for estimating contaminant concentrations in animal feeds such as relatively short
lived grass plants, the equilibrium Bvo| from the Bacci azalea model may overestimate the
5-9
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
contaminant concentration in grass. On the other hand, McCrady (personal
communication, J. McCrady, Corvallis Environmental Research Laboratory, EPA) suggests
that the 50-day exposure time used in Bacci's experiments may allow for considerable
diffusion into the newly formed plant surface wax. Thus, the sorbed TCDD residues may
be trapped and unable to volatilize. If so, than the relatively high volatilization and
photodegradation rates reported by McCrady may be higher than what might occur for the
longer exposure times expected in real world situations.
These arguments are being presented to demonstrate the uncertainty in choosing
either of the two reported BVO| values for estimating plant contaminant concentrations.
McCrady's results pertaining to 2,3,7,8-TCDD cannot be generalized to other compounds
or other contaminants in terms of commonly available contaminant parameters such as H
or Kow. Therefore, a McCrady framework similar to Bacci's for estimating congener-
specific Bv cannot be offered at this time. On the other hand, their work strongly suggests
that the Bacci model may be inappropriate for food chain modeling, because of their
experimental length of time, their use of an azalea leaf of high wax content, and lack of a
artificial light source simulating photodegradation.
Based on work conducted for the dioxin exposure document, the following
recommendation is made. The Bacci algorithm as more recently formulated (Bv of
Equation [5-15b] above) should be used to estimate a chemical-specific air-to-leaf transfer
factor. This algorithm is generalizable to any chemical for which a Henry's Constant, H,
and an octanol water partition coefficient, Kow, is known. However, a value arrived at
with this algorithm should be reduced by a factor of ten before use. The factor of ten
reduction is suggested based on work conducted for the dioxin exposure assessment as an
interim correction factor. Briefly, this work involved attempting to model the background
concentrations of dioxin-like compounds in beef starting with a profile of air concentrations
crafted to be typical of rural environments. This exercise demonstrated that use of the
Bacci transfer factor without empirical adjustment would significantly overestimate
concentrations, whereas use of the Bacci factor empirically reduced by a factor of 10
made predictions come in line with observations (note: the current draft of the dioxin
exposure document is as yet unreviewed; this exercise does not appear in the 1992
5-10
November 10, 1993
-------
DRAFT -- DO IMOT QUOTE OR CITE
released version of the dioxin exposure document - EPA, 1992c).
5.3 Calculating the Concentration of Pollutant in Animal Tissues
Issue The Travis and Arms biotransfer factor (Ba) is the only transfer factor in the
methodology from the Indirect Exposure Document that uses mass ingestion of
contaminants (mg/day) by animals and translates them to a tissue concentration
(mg/kg whole concentration). Other approaches depend on a concentration in
media to concentration in tissue bioconcentration/biotransfer factor.
Conclusions/Recommendations
The fact that Ba is based upon mass ingestion of contaminants doesn't invalidate
the Travis and Arms approach by itself. The approach does have the advantage that the
transfers factors are based on the commonly available Kow. The dioxin exposure
document (EPA, 1992c) uses a bioconcentration factor, which takes the ratio of the
concentration of contaminant in cattle dry matter intake (mg/kg) and translates it to a
concentration in cattle body and milk fat (mg/kg lipid based concentration). What the
dioxin document also does, is to demonstrate that the 2,3,7,8-TCDD BCF (developed by
Fries and Paustenbach) comes up with whole milk and whole beef concentrations that are
actually fairly similar to whole milk and beef concentrations that would be estimated using
the Travis and Arms approach. This exercise required assumptions on: dry matter intake
by cattle (kg/day), beef and milk fat assumptions, and so on. This gives some validation
of the Travis and Arms approach.
At this point, no changes should be made to the biotransfer factor, Ba. In addition
to the benefit of it being generalizable based on Kow, factors were also compiled for
metals in Table 5-4 of the Indirect Exposure Document.
Issue A concern is that some plants ingested by animals may not be contaminated. This
might call for an additional adjustment factor, F -^ which would be included with
Qp jjP jj of Equation [5-19], and would be defined as the fraction of the ith plant
group eaten by jth animal which is grown on impacted soil (i.e., soil impacted at
5 km, average soil impact over 50 km, etc.). The implicit assumption in the current
5-11
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
framework is that this value is 1.00. A value less than 1.00 might be called for if
cattle feed (or hog or chicken feed, etc.) were purchased by the farmer from a local
distributor who gets it from a distant location.
From the Indirect Exposure Document:
J = /"I (Qpij PfjJ + (QSj Sc)]
[5-19]
where:
AJ
dP,
QSj
Sc
BBj
concentration of pollutant in the Jtfl animal tissue group
(ug pollutant/g animal tissue DW)
quantity of ith plant group eaten by the jth animal each day
(kg plant tissue DW/day)
total concentration of pollutant in the ith plant group eaten by
thejth animal (ug pollutant/g plant tissue DW)
quantity of soil eaten by the jth animal each day (kg soil/day)
soil concentration (ug pollutant/g soil)
biotransfer factor for the jth animal tissue group (d/kg)
Conclusions/Recommendations
A factor, F^, should be added to Equation [5-19] as follows:
1 1
-------
DRAFT - DO NOT QUOTE OR CITE
Default values of 1.00 should be assumed; values less than one should be based on
information on the portion of cattle grain, silage, etc. that is transported from sites that are
not impacted by the combustor.
5.3.2. Quantity of Soil Consumed by Animals
Issue A concern here is the Indirect Exposure Document's assuming no soil ingestion by
pigs and poultry. Fries of USDA notes that pigs exhibit "rooting" behavior and
assumes a maximum soil ingestion intake of 8 percent of dry matter based on a
2 to 8 percent range noted in his earlier PCB work. Chickens could also come in
contact with soil and have soil ingestion.
Conclusions/Recommendations
A soil ingestion quantity of 8% should be assigned for pigs. No data could be
found for free ranging poultry. It seems intuitively appropriate to assign a value for .
chickens less than that for pigs, and a value comparable for that of grazing cattle.
Therefore, a value of 3% for free ranging chicken soil ingestion, the IED recommendation
for cattle soil ingestion, is made here.
5.4 Calculating Human Daily Intake
Issue Further guidance is needed on how to estimate ingestion rates of home-grown
foods.
Conclusions/Recommendations
The Indirect Exposure Document uses a two part procedure to estimate the
consumption rate of locally produced foods. The first step uses national data to estimate
total consumption rates of various food groups and then adjusts these using the fraction
homegrown. Each of these issues is discussed below.
The total consumption rates are estimated primarily on the basis of the EPA's Office
of Pesticide Program "Tolerance Assessment System" which was derived from the 1977-
78 USDA Nationwide Food Consumption Survey (USDA, 1983). The data are
conveniently summarized to provide consumption rates (normalized by body weight) for 1)
5-13
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
seven different plant groups and seven different animal groups, 2) various percentiles and
3} five different age groups. Such statistics are very useful in exposure assessments.
However, when using these data the assessor should consider the following points:
• These data represent total ingestion rates of store-bought food. Obviously, what is of
interest for a site-specific problem is the amount of food consumed which is locally grown
within the study area. Ideally, local surveys would be used to directly determine ingestion
rates of locally produced foods or (as a second choice) to estimate diet fractions needed to
adjust the total ingestion rates.
• The USDA survey was conducted over a three day period. Thus, it does not represent
long term behavior patterns which is the interest of exposure assessments used to support
analysis of chronic health effects. This problem introduces uncertainty into the estimates
of the median (50th percentile) and other percentiles. The data would be appropriate for
estimating means.
• For purposes of assessing impacts of atmospheric deposition onto plants it is important
to consider whether the edible portion is protected or not. However, even protected foods
can be impacted if they take up contaminants via the roots after deposition on to soil.
The fraction of foods which are home-produced was also based on data from the
USDA survey. Although these estimates are a potentially useful starting point to select
site-specific values, ideally they should be refined on the basis of local surveys (as
discussed above, it would be more useful to use local surveys to directly evaluate
ingestion rates of locally produced food). The Workgroup was not able to agree on further
recommendations in this area. Some of the diet fractions presented in the Indirect
Exposure Document do not appear reasonable, i.e. 50% of the lima beans consumed by
city residents are home produced. Clearly, the accuracy of these values need to be
confirmed. The Workgroup recommended further analysis of these surveys before final
recomendations are made. Additional considerations are noted below:
5-14
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
• Chapter 2 recommends developing separate exposure estimates for various segments
of the population such as farmers, home gardeners, nongardening residents, etc. These
diet fractions were presented for city, suburban and nonmetropolitan areas. Thus, it is not
clear how to apply these data to these population groups.
• The diet fraction listed for beef and dairy consumption in nonmetropolitan areas is 15%
and 3% respectively. An earlier USDA study (USDA, 1966) of 900 rural farm households
found that 44% of the beef and 40% of dairy products consumed were home-produced.
Since this study was specific to farm households, it may be more accurate for this
population segment. Similar issues would apply to the other food groups. The 1966 diet
fractions sound high, but are based on fairly old data. EPA plans to evaluate whether a
similar analysis can be made of the 1977-78 and the 1987-88 surveys.
• The USDA surveys estimated fraction home-grown by subtracting amounts of food
purchased from total ingested. Some individuals may eat locally produced foods bought
from neighbors or at farmer markets. Assessors should attempt to evaluate this practice
on the basis of local information and adjust diet fraction assumptions accordingly.
EPA is in the process of evaluating new food ingestion data as part of the update
to the Exposure Factors Handbook (EPA, 1989b) and may have a better basis for these
recommendations soon.
5-15
November 10, 1993
-------
-------
DRAFT -- DO NOT QUOTE OR CITE
6. DETERMINING EXPOSURE FROM SOIL INGESTION
6.2 Human Daily Intake
Issue There has been considerable discussion as to the appropriate soil ingestion
assumptions.
Conclusions/Recommendations
It is recommended that for children ages 1 to 6, an ingestion rate of 0.1 g/day (as
an average rate) and 0.2 g/day (as an upper estimate) and a body weight of 16 kg (see
note below) should be used, while for adults an ingestion rate of 0.05 g/day (as an
average) and 0.1 g/day (as an upper estimate) and a body weight of 70 kg should be
used. These ingestion rate recommendations are based largely on the judgement of the
Working Group. A number of soil ingestion studies (EPA, 1989b) have been conducted,
but they do not provide a strong statistical basis for evaluating upper estimates of soil
ingestion for children. These studies have been interpreted differently even within EPA.
Note that EPA, 1989b identified the range 0.1 to 0.2 g/day as representative of the mean
and recommended 0.8 g/day as an upper estimate. Virtually no reliable data are available
for evaluating adult soil ingestion. The reasons that the existing studies on children are
not well suited to deriving distributions or upper percentiles include: 1) they were
conducted over 3-4 days and therefore do not represent usual behavior and 2) they involve
substantial experimental error, i.e. the range of values seen in different children are likely
to represent uncertainty more than true variability in behavior.
Although the 16 kg body weight assumption for children is generally adequate, a
more accurate procedure is to use the average of the inverse body weights over this age
range.
These recommendations apply to children with normal mouthing tendencies. Some
children exhibit abnormal mouthing behavior or "pica" and would have much higher
ingestion rates. The associated soil ingestion rates, frequency and duration of the
behavior are not well established. Thus, no quantitative basis can be offered to evaluate
this behavior.
EPA is in the process of evaluating new soil ingestion data as part of the update to
6-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
the Exposure Factors Handbook (EPA, 1989b) and may have a better basis for
recommendations in this area soon.
6-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
7. DETERMINING EXPOSURE FROM DERMAL ABSORPTION VIA SOIL
Issue The Indirect Exposure Document does not provide guidance for quantitative
evaluation of dermal risks.
Conclusions/Recommendations
New quantitative procedures for dermal exposure have recently been published in
Dermal Exposure Assessment: Principles and Applications (EPA, 1992b), and should be
adopted here.
7-1
November 10, 1993
-------
-------
DRAFT -- DO NOT QUOTE OR CITE
8. DUST RESUSPENSION
8.4 Example Calculations
Issue The expert panel which met in September, 1992, to evaluate the August, 1992
dioxin exposure document (EPA, 1992c), recommended a different model be
adopted for evaluating suspensions of contaminated soil via wind erosion. The
model used in the dioxin document is the same one used in Chapter 8 of the
Indirect Exposure Document.
Conclusions/Recommendations
The model recommended by the expert panel was researched and it was not clear
that it offered a substantial improvement over the one used and it might even be more
data intensive. No change is recommended for the present.
'ssue The Indirect Exposure Document only addresses dust resuspension due to wind
erosion. However, "fugitive dust" generated from agricultural tilling and from
vehicle traffic on paved and unpaved roads are much larger sources of resuspended
particulate matter. Also, the I ED does not present any algorithm for estimating
dispersion of suspended dust.
Conclusions/Recommendations
The document, "Control of Open Fugitive Dust Sources" (EPA, 1988) includes
algorithms for estimating emissions from agricultural tilling and paved and unpaved roads.
A companion document, "User's Manual for the PM-10 Open Fugitive Dust Source
Computer Model Package" (EPA, 1990b), is available which implements these techniques
for use on a personal computer. The "Guideline on Air Quality Models (Revised)" (EPA,
1993a) contains information on dispersion models that can be used to estimate airborne
concentrations downwind from fugitive dust sources. Specifically, the ISC2 model may be
used for area and line sources, such as agricultural fields and roads. More simply, one can
use a box model to estimate air concentrations near the source of fugitive emissions.
Either of these two approaches will result in airborne concentrations of contaminants
bound to particles at a site downwind from where fugitive emissions occur.
8-1
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
With care, there are ways to further use these estimated particle-bound
contaminant concentrations. First, a portion of this concentration can reasonably be
assumed to be 10//m in diameter or less, and most of the fugitive emission models do
distinguish emissions that are less than and greater than 10 /;m. This portion can be
added to air concentrations which are estimated for direct, inhalation exposures. Second,
contaminants resuspended in dust can be resettled onto vegetations (above ground
fruits/vegetables, pasture grass, cattle feed) or soils assuming a velocity of deposition -
//g/m3 * m/sec = a rate of deposition in units of//g/m2-sec. Dry deposition velocities
from the COMPDEP model can be used here as one option for dry deposition. Alternately,
a velocity of dry deposition can be assumed based on an assumption of particle size and
gravitational settling. An appropriate, estimated velocity of gravitational deposition for 10
//m size particles is 0.01 m/sec (1 cm/sec), as given in Seinfeld (1986). There still would
need to be consideration of wet deposition of resuspended particles. A wet deposition
amount can be estimated as a product of the air concentration times annual rainfall times a
volumetric washout fraction for contaminants in airborne particles. This washout ratio is
defined as: (mass contaminant/volume rain) -*• (mass contaminant/volume air). Bidleman
(1988) discusses the washout ratio - saying it ranges between 103 and 106; is typically
between 105 and 106, and uses a value of 2 * 105 for his further calculations. This is
recommended as a reasonable value for first approximations if the impact of fugitive
emissions are to be considered.
In summary, wet and dry depositions of resuspended airborne particulates are given
as:
dry deposition = Vd C
pa
•wet deposition = R Wn (7
f p p
[8-1]
[8-2]
where:
'pa
contaminant air concentration estimated for resuspended particles,
8-2 November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
//g/m3 (or comparable mass/volume units)
Vd = dry deposition velocity, m/sec (0.01 m/sec)
R = annual rainfall, m/yr
Wp .= washout ratio, unitless (2* 10s)
However, depositing resuspended contaminants in dust particles should be done
carefully. For example, if one is suspending dust due to agricultural tillage, one should not
redeposit that suspended amount back onto the very same agricultural soil. That would be
double counting. In the same vein, depositing resuspended dust due to agricultural tillage
onto the field crops grown on that very same agricultural soil is also double counting.
Recall that the Travis and Arms soil to above ground plant transfer coefficient, Brj, is
expected to estimate all causes for contaminants to transfer from soils to plants growing
in the soils. Therefore, it would be invalid to resuspend dust due to agricultural tillage,
deposit it onto agricultural crops, and then also use the Travis and Arms Bri?
The only situation where resuspended impacts should be considered is when the
site of impact of such particles is physically distinct from the site of origin of the
resuspended dust. For example, dust resuspended from roadways onto soils or
vegetations would be reasonable since roadway dust is a unique source to soils or
vegetations. Another example would be the impact of agricultural tillage onto nearby
residential settings. There again, the suspended dust is a unique source to the nearby
residence.
As a final note, one should always compare the contaminant concentrations of
dispersed, resuspended contaminated dust with air concentrations estimated by
dispersion/deposition modeling from the stack to see if addition of resuspended dust adds
significantly to the reservoirs predicted to occur from the stack. In the same way, one
should compare the rate of deposition of contaminants in resuspended dust to that
estimated by the atmospheric deposition modeling. If the air concentration and deposition
rates are very small in comparison to those which are due to stack emissions, say 1 % of
concentrations and depositions due to direct emissions, they can be neglected.
8-3
November 10, 1993
-------
-------
DRAFT -- DO IMOT QUOTE OR CITE
[The following is offered as a framework for estimating impacts to water bodies from
combustors. It replaces the multiple frameworks offered in Section 9.2 of Chapter 9 of
the Indirect Exposure Document. Sections 9.1, 9.3, and 9.4 of Chapter of the Indirect
Exposure are still evaluated as being appropriate, except for issues as noted below. This
alternative has been developed by Bob Ambrose of the Athens Environmental Research
Laboratory and Matt Lorber of the Office of Health and Environmental Assessment. Care
has not been taken to make the units or parameter names consistent with other chapters
of the Indirect Exposure Document. However, the units are internally consistent.]
[Revised Section 9.2. begins here]
9.2. CALCULATING WATER CONCENTRATIONS
9.2.1 INTRODUCTION
The following framework for estimating surface water impacts from stack
emissions estimates water column as well as bed sediment concentrations. Water column
concentrations include dissolved, sorbed to suspended sediments, and total (sorbed plus
dissolved, or total contaminant divided by total water volume). This framework also
provides three concentrations for the bed sediments: dissolved in pore water, sorbed to
bed sediments, and total. The model accounts for four routes of contaminant entry into
the water body: 1) sorbed to soils eroding into the water body, 2) dissolved in runoff
water, 3) direct deposition of particle-bound contaminant; and 4) direct diffusion of vapor
phase contaminants into the water body. The model also accounts for four dissipation
processes which remove contaminants from the water column and/or bed sediment
reservoirs: 1) decay of total contaminants (sorbed + dissolved) within the water column,
2) decay of total contaminants (sorbed + dissolved) within the bed sediment, 3)
volatilization of dissolved phase out of the water column; and 4) removal of total
contaminant via "burial" from the surficial bed sediment layer. This burial rate constant is
a function of the deposition of sediments from the water column to the bed; it accounts
for the fact that much of the soil eroding into a water body annually becomes bottom
sediment rather than suspended sediment. The impact to the water body is assumed to be
9-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
uniform. This tends to be more realistic for smaller water bodies as compared to large
river systems. Key assumptions in the surface water impact algorithm are:
• Soil concentrations within a depositional area (which is distinct from a
watershed; see discussion for WAL below in Section 9.2.10) are assumed to be uniform
within the area, and can be estimated by the following key parameters representing the
area as a whole: dry and wet contaminant deposition rates, a soil dissipation rate, a soil
bulk density, and a soil mixing depth;
• The partitioning of the contaminant within the soil/water matrices - surface
soils, suspended solids in the water body, and bed sediments of the water body - can be
described by partition coefficients;
• One route of entry into the surface water body will be direct deposition. A
second route of entry is contaminant dissolved in annual surface runoff. This will be
estimated as a function of the contaminant dissolved in soil water and annual water
runoff. A third route of entry is via soil erosion. A sorbed concentration of contaminant
in soil, together with an annual soil erosion estimate, a sediment delivery ratio, and an
enrichment ratio, can be used to describe the delivery of contaminant to the water body
via soil erosion. A sediment delivery ratio serves to reduce the total potential amount of
soil erosion (i.e., the total potential equals a unit erosion rate as in kg/m 2 times a
watershed area, in m 2) reaching the water body recognizing that most of the erosion from
a watershed during a year deposits prior to reaching the water body. The enrichment ratio
recognizes the fact that soils which erode tend to be lighter in texture, more abundant in
surface area, and have higher organic carbon. All these characteristics lead to
concentrations in eroded soils which tend to be higher in concentration as compared to in-
situ soils. A fourth and final route of entry is via diffusion in the gaseous phase. The
dissolved concentration in a water body is driven toward equilibrium with the vapor phase
concentration above the water body. At equilibrium, gaseous diffusion into the water
body is matched by volatilization out of the water body. Gaseous diffusion is estimated
9-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
with a transfer rate (determined internally given user inputs) and a vapor phase air
concentration. This air concentration is specified by the user and is an output of the
atmospheric transport model.
• For the surface water solution algorithm, it is assumed that equilibrium is
maintained between contaminants within the water column and contaminants in bed
sediments. Equilibrium is established when the dissolved phase concentration in the water
column is equal to the dissolved phase concentration within the bed sediments. This
condition is imposed by the water body equations.
• A rate of contaminant "burial" in bed sediments is estimated as a function of
the rate at which sediments deposit from the water column onto the surficial sediment
layer. This burial represents a permanent sink and recognizes that a portion of the soil,
and contaminant sorbed to it, which erodes into a water body becomes bottom sediment
while the remainder becomes suspended sediment. This solution assumes that there will
be a net depositional loss, even though resuspension and redeposition of sediments is
ongoing, particularly with moving water bodies. For cases where the net deposition rate is
zero, there will be no burial loss calculated.
• Separate water column and benthic decay rate constants will allow for the
consideration of decay mechanisms which remove contaminants from the water body. If a
transformation yield coefficient is specified, the decay of one chemical is linked to the
internal loading of a second chemical.
Figure 9.1 below displays the framework for this analysis, with a listing of the ten
concentrations which are part of the solution algorithm, and the state equations which link
the concentrations.
9-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
(
\
(
X
(
decay
decay ^^
(
equilibrium reactions
kinetic loads or reactions \
diffusion
\
\ .
v |
t
sw
^%r"
-'bt"
burial
/
I
/
C
\
\ Ca
volatilization
dw
a
/
\
ivection
Egure 9.1. Basic structure of the water algorithm showing concentrations
considered, relationships between concentrations, and routes of fate and dissipation
-st
-'wtot
db
"sb
Definitions
total soil concentration mg/L
concentration dissolved in soil water mg/L
concentration sorbed to soil mg/kg
yearly dry deposition to surface water mg/yr
yearly wet deposition to surface water mg/yr
vapor phase air concentration above water fJQ/m3
total concentration in water column mg/L
total water concentration in surface
water system, including water column
plus benthic sediment (not shown in Figure 1} mg/L
dissolved phase water concentration mg/L
sorbed phase water concentration mg/kg
bottom sediment total concentration mg/L
concentration dissolved in bed sediment
pore water mg/L
concentration sorbed to bottom sediments mg/kg
9-4
November 10, 1993
-------
DRAFT -- DO IMOT QUOTE OR CITE
STATE EQUATIONS CORRESPONDING TO FIGURE 9.1
I. Soil
= Sc BD
BD
Kd
Sc Kd, BD
€„, = s
* 6S + Kds BD
Sc BD
6S + Kds BD
Sc = see Equation [9-2]
II. Surface Water System
= see Equation [9-1]
-dJdJ + (efa + Kdhs-BS)-dJdz
9-5
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Ibeitth
III. Water Column
IV. Bed Sediment
Kdbs-BS\
Note that by substituting the relationship between Cwt and Cwtot into the expression for
Cbt, we can obtain benthic concentrations as a functions of water column concentrations:
9-6
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Qbs + Kd^-BS
Ci-
where:
Sc
BD
TSS
BS
f,
water
'benth
'dw -
'db
soil concentration (//g/g)
volumetric soil water content (Lwater/L)
bed sediment porosity (Lwater/L)
soil/water partition coefficient (L/kg)
suspended sediment/surface water partition coefficient (L/kg)
bottom sediment/sediment pore water partition coefficient (L/kg)
soil bulk density (g/cm3) •
total suspended solids (mg/L)
bed sediment concentration (g/cm3)
depth of the water column (m)
depth of the upper benthic layer (m)
total depth of water body, dw + db (m)
fraction of total water body contaminant concentration that occurs in
the water column
fraction of total water body contaminant concentration that occurs in
the bed sediment
fraction of water column contaminant concentration that is dissolved
fraction of bed sediment contaminant concentration that is dissolved
9-7
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
9.2.2 CHEMICAL MASS BALANCE EQUATIONS
If Figure 9.1 is taken as a control volume for the water body, a steady-state mass
balance equation can be written that balances chemical loadings with outflow and loss:
where:
f,
water
LT
[9-1]
water
total water body concentration, including water column and benthic
sediment (mg/L)
total chemical load into water body, including deposition, runoff,
erosion, atmospheric diffusion, and internal chemical transformation
(g/yr)
average volumetric flow rate through water body (m3/yr)
total volume of water body or water body segment being considered,
including water column and benthic sediment (m3)
total first order dissipation rate constant, including water column and
benthic degradation, volatilization, and burial (yr~1)
fraction of total water body contaminant concentration that occurs in
the water column
A similar mass balance equation can be written for the watershed soils, balancing
areal deposition fluxes with chemical loss processes:
where:
So
-w
RD (1.0 -
-fo Tc)) 100 + Csh
[9-2]
average watershed soil concentration after time period of deposition
(//g pollutant/g soil)
yearly average load of pollutant to watershed on an areal basis
9-8
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
ks
Tc
2 =
BD
100 =
(g pollutant/m2-yr)
total chemical loss rate constant from soil (yr~1)
total time period over which deposition has occurred (yr)
representative watershed mixing depth to which deposited pollutant
is incorporated (cm)
representative watershed soil bulk density (g/cm3)
units conversion factor (//g-m2/g-cm2)
background "natural" soil concentration (jjg pollutant/g soil)
The major terms in Equations [9-1] and [9-2] are discussed in sections below.
9.2.3 SEDIMENT MASS BALANCE EQUATIONS
Before calculating chemical fate, a mass balance equation for sediments in the
water body must be solved. First, we begin with the water column. Referring to Figure
9.2, the soil eroding into the water body, Xw, equals the sum of the amount depositing
into the upper bed, Xd, and the advective loss from the water column, Xa. Solving for the
suspended solids concentration in the water column:
[9-3]
Vfx
WA
where:
TSS =
WAL
SD
Vfx
wdep
WA,.,
10
3 _
suspended solids concentration (mg/L)
unit soil loss, calculated in the soils section from the USLE equation
(kg/m2-yr)
watershed surface area (m2)
watershed sediment delivery ratio (unitless)
average volumetric flow rate through water body (m3/yr)
suspended solids deposition rate (m/yr)
water body surface area (m2)
units conversion factor
9-9
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
X
w
erosion
advection
TSS
(VV>
BS
X
Figure 2. Steady state representation for sediments in water bodies.
xw
x'
TSS
BS
Wdep
b
DEFINITIONS
soil erosion input from depositional area
advective loss from water body
deposition onto bottom sediment
burial below bottom sediment layer
suspended solids concentration
bottom sediments concentration
rate of deposition onto bed sediment
rate of burial
g/yr
g/yr
g/yr
g/yr
g/L
g/L
m/yr
m/yr
9-10
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Often, the user will have data on average suspended solids, but not the deposition rate. In
this case, the deposition rate can be calculated as:
Xe WAL SD 103 - Vfx TSS
WAW TSS
[9-4]
where terms are as defined above.
In the second part of the solids balance, the mass deposited to the bed, Xd, is set
equal to the mass buried, Xb. Solving for the burial rate gives:
where:
Wb
wdep
TSS
BS
10'6
w = w TSS 10'6
b dtp f.r,
DO
burial rate (m/yr)
deposition rate (m/yr)
suspended solids concentration (mg/L)
benthic solids concentration (kg/L)
conversion factor (kg/mg)
[9-5]
Finally, the benthic porosity, the volume of water per volume of benthic space, is
calculated from the benthic solids concentration and sediment density:
where:
BS
PS
benthic porosity (L/L)
benthic solids concentration (kg/L)
solids density, 2.65 kg/L
9-11
[9-6]
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
For input benthic solids concentrations between 0.5 and 1.5 kg/L, benthic porosity will
range between 0.8 and 0.4.
9.2.4. LOADS TO WATERSHED SOILS
The pollutant load term includes wetfall and dryfall fluxes, along with internal
transformation loads:
where:
Dydw
Dyww
-DIF
Lw = Dydw + Dyww + Lls + L
DIF
[9-7]
yearly average dry depositional flux of pollutant (g/m2-yr)
yearly average wet depositional flux of pollutant (g/m2-yr)
internal transformation load of pollutant per areal basis (g/m2-
yr)
atmospheric diffusion flux to soil (g/m2-yr)
Wet and dry depositional fluxes are determined by measurement or by air modeling and are
specified as input to this model. Internal loading is due to chemicals that are transformed
chemically or biologically into daughter products, which may be of interest in risk
assessment. Transformation yields can be specified for the sequential reaction:
where:
B
C
ksg(B) =
Ys(BC) =
B - C
parent compound
daughter compound (or reaction product)
first order transformation rate constant (yr~1)
reaction yield coefficient {mg of C/mg of B)
[9-8]
9-12
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
The internal loading generated by this transformation reaction is given by:
where:
Hs
ksg(B)
Ys(BC)
Sc(B)
BD
Z
10
-2 _
Ys(BC) Sc(B) BD Z 10"2
[9-9]
internal transformation load of pollutant per areal basis (g/m2-yr)
first order transformation rate constant (yr~1)
reaction yield coefficient (mg of C/mg of .B)
soil concentration of B (//g/g)
bulk density of soil, g/cm3
representative watershed mixing depth to which deposited pollutant
is incorporated (cm)
units conversion factor
This loading is added to chemical C. If the yield coefficient is set to zero, then the
compounds are treated as independent.
The load due to vapor diffusion is given as:
where:
DIF
'va
LDIF = 0.31536 KtCm
atmospheric diffusion flux to soil (g/m2-yr)
gas phase mass transfer coefficient (cm/s; see Eq [4-6], IED)
gas phase air concentration (jug/m3)
[9-10]
9.2.5 LOSS PROCESSES IN WATERSHED SOILS
The total chemical loss rate ks in equation [9-2] is due to several physical and
chemical processes:
9-13
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
where:
ks
ksl
kse
ksr
ksg
ksv
ks = ksl + kse + ksr + ksg + ksv 19-11 ]
soil loss constant due to all processes (yr~1)
soil loss constant due to leaching (yr~1)
soil loss constant due to erosion (yr~1)
soil loss constant due to runoff (yr~1)
soil loss constant due to chemical transformation/ degradation (yr"1)
soil loss constant due to volatilization (yr"1)
The degradation constant, ksg, is derived by the user. The other four constants are given
by:
ksl =
P + I - R - EV
6SZ [1.0 + (BDKds/6s)]
[9-12]
0.1 Xe SD ER I Kds BD
kse = e Is
BD Z
B, + KdsBD
[9-13]
ksr =
R 1
esZ\l + (Kd.BD/6J
[9-14]
ksv = Ke Kt
[9-15]
where:
average annual precipitation (cm/yr)
9-14
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
R
Ev
*s
Z
BD
SD
ER
Kds
*e
Ke
Kt
0.1
average annual irrigation (cm/yr)
average annual runoff (cm/yr)
average annual evapotranspiration (cm/yr)
volumetric water content (dimensionless; cm 3/cm 3)
watershed mixing zone depth (cm)
soil bulk density (g/cnn 3)
sediment delivery ratio
contaminant enrichment ratio
soil-water partition coefficient (cm 3/g)
unit soil loss (kg/m 2/yr; see Eq [9-3], IED)
equilibrium coefficient (s/cm/yr); see Eq [4-5], IED)
gas phase mass transfer coefficient (cm/s; see Eq [4-6], IED)
units conversion factor
Sc is the concentration resulting from contaminated particles depositing on and
mixing with surface soils. For strongly hydrophobic contaminants such as dioxins or
PCBs, where Kd s are very large, Sc will be essentially equal to the "total" soil
concentration, C st, as well as the sorbed concentration, C ps, and the dissolved phase
concentration, C ds/ will be vanishingly small. (See the state equations above.) However,
for a generic solution where a contaminant may not be as hydrophobic, contaminants
depositing as particles are assumed to reequilibrate in the soil/soil water system. In the
listing of state equations, the reequilibrated sorbed phase concentration, C ps, and the
dissolved phase concentration, C ds, are now used to estimate loads due to soil erosion
and loads due to surface runoff, respectively.
9.2.6 LOADS TO THE WATER BODY
Total chemical loading to the water body is comprised of five inputs:
9-15
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
where:
-Dep
-Rl
LDif
LT =
Dif
[9-16]
total contaminant load to the water body (g/yr)
deposition of particle bound contaminant (g/yr)
runoff load from impervious surfaces (g/yr)
runoff load from pervious surfaces (g/yr)
soil erosion load (g/yr)
diffusion of vapor phase contaminant (g/yr)
internal transformation load (g/yr)
The runoff and erosion loads require estimation of average contaminant
concentration in watershed soils that comprise the depositional area. This will be
discussed first.
9.2.6.1 Load due to direct deposition
The load to surface waters via direct deposition is solved as:
where:
LDep ~
Dyds =
Dyws =
[9-17]
direct deposition load (mg/yr)
representative yearly dry deposition rate of pollutant onto surface
water body (g pollutant/m 2/yr)
representative yearly wet deposition rate of pollutant onto surface
water body (g pollutant/m 2/yr)
water body area (m 2)
9-16
November 10, 1993
-------
DRAFT - DO IIMOT QUOTE OR CITE
9.2.6.2 Load due to impervious surface runoff
A fraction of the chemical wet and dry deposition in the watershed will be to
impervious surfaces. Dry deposition may accumulate and be washed off during rain
events. If the impervious surface includes gutters, the pollutant load will be transported to
surface waters, bypassing the watershed soils. The average load from such impervious
surfaces is given by:
where:
RI
yWW
'1 81
impervious surface runoff load (g/yr)
impervious watershed area receiving pollutant deposition (m2)
yearly wet deposition flux onto the watershed (g/m2/yr)
yearly dry deposition flux onto the watershed (g/m2/yr)
9.2.6.3 Load due to pervious surface runoff
Most of the chemical deposition to a watershed will be to pervious soil surfaces.
These loads are accounted for in the soil mass balance equation. During periodic runoff
events, dissolved chemical concentrations in the soil are transported to surface waters as
given by:
LR = R(WAL-WAI)
Sc BD
Kds BD
10
-2
[9-19]
where:
R
Sc
BD
0,
pervious surface runoff load (g/yr)
average annual runoff (cm/yr)
pollutant concentration in watershed soils (//g/g; Eq [9-2])
soil bulk density (g/cm3)
volumetric soil water content (cm3/cm3)
9-17
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Kds =
WAL =
WA, =
10
-2 _
soil-water partition coefficient (L/kg or cm3/g)
total watershed area receiving pollutant deposition (m2)
impervious watershed area receiving pollutant deposition (m2)
units conversion factor (g2/kg-//g)
9.2.6.4 Load due to soil erosion
During periodic erosion events, particulate chemical concentrations in the soil are
transported to surface waters as given by:
where:
Sc
BD
es
Kds
WA
L -
WA,
SD
ER
10'3
LE = Xe (WAL -
SD ER
Sc Kdv BD
s
Qs + Kds BD
to-3
[9-20]
soil erosion load (g/yr)
unit soil loss (kg/m2/yr)
pollutant concentration in watershed soils (//g/g; Eq [9-2])
soil bulk density (g/cm3)
volumetric soil water content (cm3/cm3)
soil-water partition coefficient (L/kg or cm3/g)
total watershed area receiving pollutant deposition (m2)
impervious watershed area receiving pollutant deposition (m2)
watershed sediment delivery ratio (unitless)
soil enrichment ratio {unitless)
units conversion factor (g-cm2///g-m2)
9.2.6.5 Load due to gaseous diffusion
The change in the total water concentration over time due to volatilization is given
by:
9-18
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
where:
D
'
water
H
R
Tk
10
-6 _
, tvolat n [J waterJ dw wtot
at U
caio-6V
[9-21]
total water body contaminant concentration (mg/L)
overall transfer rate (m/yr)
depth of water body (m)
fraction of total water body contaminant concentration that occurs in
the water column
fraction of water column contaminant concentration that is dissolved
vapor phase air concentration over water body Oc/g/m3)
contaminant Henry's Constant (atm-m3/mole)
universal gas constant, 8.206*10"5 atm-m3/mole-K
water body temperature (K)
units conversion factor
The right hand side of Equation [9-21] contains two terms. The first term
constitutes a first order loss rate of aqueous contaminant, and is covered in more detail in
section 9.2.8.2 below. The second term in Equation [9-21] describes diffusion of gas-
phase contaminant from the atmosphere into the water body. Because this term is
independent of water body contaminant concentration, it can be treated as an external
load. As formulated above, this term has units of mg/L-yr. It must be converted to
loading units by multiplying by the water column volume, V. Noting that V/D is equal to
the surface water area WAW, the atmospheric diffusion load is given as:
Kv CaWAw IP'6
Dif ~ HlRTt
[9-22]
where:
9-19
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
LDif -
WAW =
H
R
10
-6 _
diffusion of vapor phase contaminant (g/yr)
the overall transfer rate (m/yr)
surface water body area (m 2)
vapor phase air concentration over water body (//g/m 3)
contaminant Henry's Constant (atm-m 3/mole)
universal gas constant ( = 8.206x10 ~5 atm-m 3/mole-°K)
water body temperature (°K)
units conversion factor
9.2.6.6 Load due to internal transformation
Chemicals may be transformed chemically or biologically into daughter products,
which may be of interest in risk assessment. A transformation yield can be specified for
the sequential reaction:
where:
B
C
k(B) =
Y(BC)
Y(BQ
B ~ C
k(B)
parent compound
daughter compound (or reaction product)
first order transformation rate constant (yr"1)
reaction yield coefficient (g of C/g of B)
[9-23]
The internal loading generated by this transformation reaction is given by:
Cbt(B) Vb] Y(BQ
where:
the internal transformation load to C (g/yr)
[9-24]
9-20
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
kdb
Y(BC) =
water column transformation rate constant from B to C (yr~1)
benthic transformation rate constant from B to C (yr~1)
total water column concentration of B (mg/L)
total benthic concentration of B (mg/L)
water column volume (m3)
benthic volume (m3)
reaction yield coefficient (g of C/g of B)
This loading is added to chemical C. If the yield coefficient is set to zero, then the
compounds are treated as independent.
9.2.7 ADVECTIVE FLOW FROM THE WATER BODY
The first term in the denominator of Equation [9-1] accounts for advective flow
from the water body. It is the product of the average annual volumetric flow rate, Vfx,
and the fraction of the chemical in the water body that is present in the water column,
fwater An impacted water body derives its annual flow from its watershed, or effective
drainage area. Flow and watershed area, then, are related, and compatible values should
be specified by the user. Given the area of drainage, one way to estimate annual flow
volume is to multiply total drainage area (in length squared units) by a unit surface water
runoff (in length per time). The Water Atlas of the United States (Geraghty et al., 1 973)
provides maps with isolines of annual average surface water runoff, which is defined as all
flow contributions to surface water bodies, including direct runoff, shallow interflow, and
groundwater recharge. The range of values ranged from 5 to 40 in/yr in various parts of
the U.S.
9.2.8 CHEMICAL DISSIPATION WITHIN THE WATER BODY
The second term in the denominator of Equation [9-1] accounts for dissipation
within the water body. It is the product of the water body volume, Vt, and the total first
order dissipation rate constant, kwt. The water body volume, in units of m3, together with
the annual flow rate, in m3/yr, determines the average residence time of a pollutant
9-21
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
traveling through the water body. The residence time for Lake Erie is about 10 years, for
example, while for the larger Lake Superior, it is estimated to be 200 years. For a swiftly
moving river, on the other hand, the residence time can be on the order of hours (1 hour
= 0.0001 1 yr). Larger volumes and residence times allow the internal dissipation
processes to have a larger effect on pollutant concentration, while smaller volumes and
residence times lessen the effect. The user should take care to specify reasonable
volumes for the type of surface water body being represented. In addition, the user should
specify compatible values for related water body parameters, such as surface area, WAW.
The water body volume divided by the surface area gives the average depth, which can
vary from a fraction of a meter for small streams to a few meters for shallow reservoirs to
tens of meters for deep lakes.
The total dissipation rate constant, kwt, applies to the total water body
concentration, C^^, and includes processes affecting any of the chemical phases —
dissolved or sorbed in the water column or benthic sediments. Volatilization, water
column and benthic degradation, and burial are considered:
where:
'water
f,
benth
J
water
J b
benth
Jw.
f b
benth
[9-25]
overall total water body dissipation rate constant (yr"1)
water column degradation or transformation rate constant (yr"1)
benthic degradation or transformation rate constant (yr"1)
water column volatilization rate constant (yr"1)
benthic burial rate constant (yr"1)
fraction of total water body contaminant concentration that occurs in
the water column
fraction of total water body contaminant concentration that occurs in
the benthic sediment
These processes are described below.
9-22
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
9.2.8.1. Chemical/Biological Degradation
Contaminants can be degraded by a number of processes in the water column or in
the benthic sediment. The major chemical and biological processes include hydrolysis,
photolysis, oxidation, and biodegradation. Each of these processes have been studied and
are incorporated in a mechanistic way in advanced water quality models. In this
methodology, the overall first-order water column and benthic degradation rate constants
are user input variables, and should be based on knowledge about the contaminant
behavior in similar surface water systems,
9.2.8.2 Volatilization
Volatile chemicals can move between the water column and the overlying air, as
described by Equation [9-21]. The right hand side of this equation contains two terms.
The second term describes diffusion into the water from the atmosphere, and is treated in
Section 9.2.6.5 as an external load. The first term, (KvfwaterfdwCwtot/D), constitutes a
first order loss rate of aqueous contaminant. This term includes the quantity
fwatei-fdv\A/vtot' which is equal to the water column dissolved phase concentration Cdw and
which is subject to volatilization loss. The rate constant for volatilization from the water
column, kv, is given as:
where:
K
^yJdw
D
water column volatilization loss rate constant (yr~1 )
overall transfer rate, or conductivity (m/yr)
fraction of contaminant in the water column that is dissolved
water body depth (m)
[9-26]
The overall transfer rate, K v or conductivity, is determined by the two-layer
resistance model. The two-resistance method assumes that two "stagnant films" are
9-23
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
bounded on either side by well mixed compartments. Concentration differences serve as
the driving force for the water layer diffusion. Pressure differences drive the diffusion for
the air layer. From balance considerations, it is obvious that the same mass must pass
through both films, thus the two resistances combine in series, so that the conductivity is
the reciprocal of the total resistance:
=(RL
KL
VG
H
R TV
-i
[9-27]
where:
RL
KL
RG
KG
R
H
liquid phase resistance (year/m)
liquid phase transfer coefficient (m/year)
gas phase resistance (year/m)
gas phase transfer coefficient (m/year)
universal gas constant (atm-m 3/mole-°K)
Henry's law constant for the pollutant (atm-m 3/mole)
water body temperature (°K)
The value of K v, the conductivity, depends on the intensity of turbulence in a water
body and in the overlying atmosphere. As the Henry's Law coefficient increases, the
conductivity tends to be increasingly influenced by the intensity of turbulence in water.
As the Henry's Law coefficient decreases, the value of the conductivity tends to be
increasingly influenced by the intensity of atmospheric turbulence.
Because Henry's Law coefficient generally increases with increasing vapor pressure
of a compound and generally decreases with increasing solubility of a compound, highly
volatile low solubility compounds are most likely to exhibit mass transfer limitations in
water and relatively nonvolatile high solubility compounds are more likely to exhibit mass
transfer limitations in the air. Volatilization is usually of relatively less magnitude in lakes
and reservoirs than in rivers and streams.
The estimated volatilization rate constant is for a nominal temperature of 20 °C. It
9-24
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
is adjusted for the actual water temperature using the equation:
= K
20
where:
0
T
temperature correction factor, set to 1.026.
water body temperature (°C)
[9-28]
There have been a variety of methods proposed to compute the liquid (K L) and gas
phase (K G) transfer coefficients. The particular method that is recommended here is the
O'Connor method.
The liquid and gas film transfer coefficients computed under this option vary with the
type of water body. The type of water body is specified as one of the surface water
constants and can either be a flowing stream, river or estuary or a stagnant pond or lake.
The primary difference is that in a flowing water body, the turbulence is primarily a
function of the stream velocity, while for stagnant water bodies wind shear may dominate.
The formulations used to compute the transfer coefficients vary with the water body type,
as shown below.
9.2.8.2.1. Flowing Stream or River
For a flowing system (type 0), the transfer coefficients are controlled by flow
induced turbulence. For these systems, the liquid film transfer coefficient (K L) is
computed using the O'Connor-Dobbins (1958) formula:
' x V/2
^M ™0
[9-29]
where:
K
liquid phase transfer coefficient (m/year)
9-25
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
U
D
w
10'4 =
3.15x10 7 =
current velocity (m/s)
diffusivity of the chemical in water (cm 2/s)
water depth (m)
units conversion factor
units conversion factor
The gas transfer coefficient (K G) is assumed constant at 36500 m/yr for flowing systems.
9.2.8.2.2. Quiescent Lake or Pond
For a stagnant system (type 1), the transfer coefficients are controlled by wind
Induced turbulence. For stagnant systems, the liquid film transfer coefficient (K L) is
computed using the O'Connor (1983) equations:
P
••
°-5
[9-30]
- «* Sc-°-67(3.15xl07)
[9-31]
where:
u* = C/-5 W
D -
9-26
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
va = (1.32 + 0.009
x
105
22x1 Q-
Pw = 1 -8.8x10-5 Tw
iog(iO =
1301
998.333 + 8.1855(7;-20) + 0.00585 (fw-20)2
- 3.0233
and:
w
Sc
w
shear velocity (m/s)
drag coefficient ( = 0.0011)
wind velocity, 10m above water surface (m/s)
density of air corresponding to the water temperature (g/cm 3)
density of water corresponding to the water temperature (g/cm 3)
von Karman's constant ( = 0.4)
dimension-less viscous sublayer thickness ( = 4)
air Schmidt number (dimensionless)
water Schmidt number {dimensionless)
diffusivity of pollutant in air (cm 2/sec)
diffusivity of chemical in water (cm 2/sec)
viscosity of air corresponding to the air temperature (g/cm-s)
9-27
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
w
MW =
1 w
3.15x10 7
viscosity of water corresponding to the water temperature (g/cm-s)
dynamic viscosity of air (cm2/sec)
molecular weight
air temperature (°C)
water temperature (°C)
units conversion factor
9.2.8.3 Deposition and Burial
The benthic burial rate, Wb, is determined as a function of user input variables as
part of the sediment balance (see Section 9.2.3). This burial rate is used to determine the
mass loss of contaminant from the benthic sediment layer. As seen in Figure 9.1, the
burial loss rate is applied to the total benthic contaminant concentration, Cbt. The water
body contaminant burial loss rate is solved by equating the mass loss rate of total water
body chemical with mass loss rate of benthic chemical:
where:
'wtot
Wu
W
Vb —
db
[9-32]
total water body contaminant concentration, including water column
and benthic sediment (mg/L)
total volume of water body or water body segment being considered,
including water column and benthic sediment (m3)
first order burial rate constant for total chemical (yr~1)
total benthic contaminant concentration (mg/L)
volume of upper benthic sediment layer (m3)
depth of the upper benthic sediment layer (m)
benthic burial rate (m/yr)
From the state equations given after Figure 9.1, it is seen that the total benthic
9-28
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
contaminant concentration, Cbt, can be expressed as a function of the total water body
concentration, Cwtot. Solving for the total chemical burial rate gives:
where:
1 benth
Wu
k - f —2
Kb ~ J benth ,
[9-33]
fraction of total water body contaminant concentration that occurs in
the bed sediment
depth of the upper benthic sediment layer (m)
burial rate (m/yr)
9.2.9 PEAK STORM EVENT CONCENTRATIONS
The previous equations together describe the long-term average water body
concentrations of sediment and chemical. Short-term peak concentrations may also be of
interest in assessing risk from some chemicals. The basic mass balance equations for
determining peak storm event concentrations of sediment and chemical are:
storm
storm
103
[9-34]
where:
TSS
C
V
X
L,
storm
storm
storm
storm
storm
_ storm
storm
[9-35]
peak suspended solids concentration during storm event (mg/L)
peak chemical concentration during storm event (mg/L)
peak water body volume during storm event (m3)
total load of sediment from the watershed during the storm (kg)
total load of chemical from the watershed during the storm, including
runoff and erosion (g)
9-29
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
10
3 _
g/kg
First we calculate the water body volume during the storm assuming that the
storm runoff volume is added to the average streamflow volume:
where:
Tr
QRS
365
24
storm
365x24
CBS
yearly average stream flow (m3/yr)
duration of the storm event (hr)
total storm runoff volume (m3)
days/yr
hours/day
[9-36]
The total storm runoff volume QRS is the sum of runoff from pervious and impervious
areas of the watershed:
where:
WAL =
WA, =
Rt =
DR =
1C'2 =
QBS = (WAL-WAj DR 10"2 + WAt R, 10'2
total watershed surface area (m2)
impervious surface area on watershed (m2)
depth of total rainfall for the storm event (cm)
depth of runoff from pervious areas of the watershed (cm)
m/cm
[9-37]
The runoff depth DR is estimated by the following equation:
9-30
November 10, 1993
-------
where:
DRAFT - DO NOT QUOTE OR CITE
(Rt
0.85)
[9-38]
DR = depth of runoff from pervious areas of the watershed (cm)
Rt = depth of total rainfall for the storm event (cm)
Mt = depth of snowmelt during the storm event (cm)
S = watershed retention parameter (cm)
The watershed retention parameter S is calculated from the Soil Conservation Service
(SCS) runoff curve number input by the user according to the following equation:
S = 2.54 [(1000/CAO - 10]
[9-39]
where:
CN
runoff curve number
Next, the mass of sediment eroded from the watershed to the receiving water
during a storm event can be estimated with the Modified Universal Soil Loss Equation
(MUSCLE):
where:
y
•^
PS
%
K
LS
C
= 2.04x106
KLSCPSD
[9-40]
total load of sediment from the watershed during the storm (kg)
volume of runoff from pervious area, calculated below (km2-cm)
peak runoff from pervious area, calculated below (m3/sec)
erodibility factor (tons/acre)
topographic or slope length factor (unitless)
cover management factor (unitless)
9-31
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
P
SD
supporting practice factor (unitless)
sediment delivery ratio (unitless)
This equation is an empirical relationship and the units must be consistent with those
shown above. Selection of K, LS, C, P, and SD is discussed in Wischmeier and Smith
(1978) and EPA (1985b), and briefly reviewed in Section 9.2.10 below.
The storm runoff volume from pervious areas is calculated from the runoff depth
according to:
where:
"ps
WAL
WA,
10
'6
= (WAL-WAj) DR 10-6
volume of runoff from pervious area (km -cm)
total watershed surface area (m2)
impervious surface area on watershed (m2)
depth of runoff from pervious areas of the watershed (cm)
km2/m2
[9-41]
A trapezoidal hydrograph is assumed so that the peak runoff rate can be calculated as:
(WAL-WA}DR
10
-2
/?,+M,-0.25j 3600
[9-42]
where:
WAL =
WA, =
peak runoff from pervious area (m /sec)
total watershed surface area (m2)
impervious surface area on watershed (m2)
depth of runoff from pervious areas of the watershed (cm)
depth of total rainfall for the storm event (cm)
depth of snowmelt during the storm event (cm)
9-32
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
S
io-2 =
3600 =
watershed retention parameter (cm)
m/cm
sec/hr
Finally, the mass of chemical load to the receiving water during the storm event is
calculated:
where:
L,
storm
-sRI -
-sR
-sE
storm
+ LsR +
[9-43]
total load of chemical from the watershed during the storm, including
runoff and erosion (g)
storm runoff load from impervious surfaces (g)
storm runoff load from pervious surfaces (g)
storm erosion load from pervious surfaces (g)
Runoff from impervious surfaces includes dryfall accumulated since the previous rainfall
event and wetfall during the present rainfall event:
I D , T
7 = I yoK* ace
SRI ~ { 365
D T '
yww T
24x365J
WA,
[9-44]
where:
ydw
yww
WA,
24
storm runoff load from impervious surfaces (g)
yearly dry deposition flux onto the watershed (g/m2-yr)
yearly wet deposition flux onto the watershed (g/m2-yr)
impervious surface area on watershed (m2)
time since previous rainfall event (days)
duration of rainfall event (hours)
hours/day
9-33
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
365 =
days/yr
Runoff from pervious soil surfaces transports dissolved chemical concentrations in the soil
to the water body:
where:
-sR
"da -
WAL =
WA, =
10
-2 =
D
10
-2
[9-45]
storm runoff load from pervious surfaces (g)
dissolved chemical concentration in soil, given by the state equations
following Figure 9.1 (mg/L)
total watershed surface area (m2)
impervious surface area on watershed (m2)
depth of runoff from pervious areas of the watershed (cm)
units conversion factor (m/cm)
The mass of chemical eroded from the watershed to the receiving water during a storm
event can be calculated from the sediment loss and the particulate soil concentration:
where:
PS
X
•storm
10'3
LSE =
SD ER Cps 10-3
[9-46]
erosion load of chemical from the watershed during the storm (g)
chemical concentration sorbed to soil, given by the state equations
following Figure 9,1 (trig/kg)
total bad of sediment from the watershed during the storm (kg)
g/mg
For risk assessment purposes, the dissolved chemical concentration can be of
interest. It may be assumed, for example, that the sorbed chemical fraction is removed in
9-34
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
a water treatment plant, and that the dissolved fraction is passed through to the user. The
dissolved concentration in the river during the storm event is calculated by:
"'dstorm
= c.
storml
1 +
[9-47]
where:
^dstorm
p
storm
Kd
10-
sw
dissolved concentration of chemical during storm event (mg/L)
peak chemical concentration during storm event (mg/L)
peak suspended solids concentration during storm event (mg/L)
suspended sediment/surface water partition coefficient (L/kg)
kg/mg
9.2.10 PARAMETER GUIDANCE
The following is offered as abbreviated guidance for all user input and internally
solved terms for the water equations:
• Dydw, Dyww, Ca, Tc: Based on output from the atmospheric transport model,
the user must specify the long-term, watershed average values for three loading
parameters: dry and wet deposition of contaminants sorbed to particulates, and the vapor
phase air concentration. In addition, the user must specify the duration of the atmospheric
loading, Tc; this is set to the expected lifetime of the combustor, usually 30 years.
Together, these parameters drive the watershed soil and water body contamination. The
deposition fluxes are in units of g/m2-yr, the air concentration is in units of //g/m3, and the
time of concentration is in yr.
• ksg, Ys, Csb: The overall chemical degradation/ transformation rate constant,
yield coefficient, and background concentration must be specified for the soil. The soil
degradation rate constant, ksg, is specific to the contaminant and to environmental
conditions, such as climate, soil type, and soil moisture. In the Indirect Exposure
9-35
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Document, cadmium was assumed not to undergo soil degradation, and ksg was given a
value of 0. The degradation rate constant for benzo(a)pyrene was set at 0.803 yr"1. If
chemical C is created by a transformation reaction of chemical B in soils, the yield
coefficient Ys must be specified. One example is the methylation of inorganic mercury,
represented as chemical B, into methyl mercury, represented as chemical C, with a yield
coefficient of 1.0. If the yield coefficient is left at 0, no transformation will be calculated.
The background concentration Csb reflects the natural concentration in the soil prior to
contaminant deposition. This value will be 0.0 for most organic contaminants. Nonzero
values for the naturally-occurring metals, however, should generally be specified.
• kdw, kdb, Y: Chemical degradation/transformation rate constants and a yield
coefficient must be specified for the water column and benthic sediment. The degradation
rate constants are specific to the contaminant and to environmental conditions, such as
pH, bacterial levels, or light. If chemical C is created by a transformation reaction of
chemical B in the water column or benthic sediment, the yield coefficient Y must be
specified. If the yield coefficient is left at 0, no transformation will be calculated.
* H, Tw, Ta, u, W: Henry's Constant, H, water temperature, Tw, air temperature,
Ta, water velocity, u, and wind speed, W, are required for estimating volatile losses and
diffusive loads of contaminant into soils and the water body. Henry's constants are
known for many contaminants of interest. They are expressed here in units of atm-
m3/mole. Average water and air temperatures should be specified by the user. Values
generally range between 10 and 20 °C, and are converted internally to Kelvin units.
Water velocity can vary from essentially 0 for stagnant ponds or lakes to 1.5 m/sec for
fast-moving streams. Sustained average wind speeds may vary from close to 1 to 10
m/sec. Data may be obtained from local weather stations.
• P, I, R, EV: These are annual water balance quantities. Precipitation, P, is
available from common meteorologic references. Irrigation, I, is a site-specific parameter
and should not be zero if the watershed is comprised of irrigated farm lands. Runoff, R,
9-36
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
can be estimated using the Water Atlas of the United States (Geraghty et al., 1973). This
reference provides maps with isolines of annual average surface-water runoff, which they
define as all flow contributions to surface water bodies, including direct runoff, shallow
interflow, and ground-water recharge. The range of values shown include 5 to 15 in/yr
throughout the Midwest corn belt, 15 to 30 in/yr in the South and Northeast, 1 to 5 in/yr
in the desert Southwest, and a wide range of 10 to 40 in/yr in the far West. Since these
values are total contributions and not surface runoff, they need to be reduced to estimate
surface runoff. A reduction by 50 percent should suffice if using the Water Atlas for the R
term. Evapotranspiration, EV, can be estimated from pan evaporation data. A "potential
evapotranspiration", PEV, can be estimated by an annual pan evaporation times 0.70 (or
thereabouts), and is defined as the evapotranspiration that occurs when soil water is not
limiting. "Actual" evapotranspiration, or EV, is about half of PEV. Site specific values for
these terms should be obtained if possible.
• Z: The user must specify the mixing depth into which contaminant is
incorporated for average watershed soils. If the watershed is dominated by soils that are
not tilled, a value of 1 cm is recommended. This is a value commonly used for "non-tilled"
situations, such as undeveloped land, pasture land, or residential properties. If the
watershed is dominated by tilled agricultural land, than a value of 20 cm is recommended.
This is a value commonly used for tilled soils including tilled agricultural fields and home
gardens.
• BD, 6 s: These soil properties include the bulk density, BD, and the surface soil
water content, 0 s. BD has a relatively narrow range of about 1.2 to 1.7 g/cm 3. A value
of 1.50 g/cm 3 should suffice for most uses unless site specific data is available.
Volumetric water content, 6, can be estimated as the midpoint between a soil's field
capacity and wilting point, if a representative watershed soil can be ascertained for the
combustor being evaluated. A reasonable range for 6 is 0.10 (for very sandy soils) to 0.30
(for heavy loam/clay soils).
9-37
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
• BS, TSS, Wd , ISS: These water body properties are analogous to the bulk
density for soils noted above. The benthic sediment concentration, BS, should range
between 0.5 and 1.5 kg/L. A value of 1.0 kg/L should be reasonable for most
applications.
As described in the beginning of Section 9.2.3, either the TSS or Wdep must be
known; one can be estimated from the other (and other parameters).
TSS concentrations can vary widely between water bodies and over time. Average
values typically will range between 1 and 100 mg/L. Given similar soil erosion loads,
values will be lower for standing water bodies such as ponds or lakes as compared with
rivers and streams. The more turbulent flow in rivers will reduce the effective deposition
rate and suspend sediments to a greater degree than a relatively calm lake. If site specific
data are not available, the user should specify concentrations of 1 to 10 in ponds and
lakes, and 10 to 20 in streams and rivers.
Net solids deposition is a complex process, and depends on the size of the particle
and the shear stress near the bed. Deposition will be a fraction of the Stokes settling
velocity, which varies from 117 m/yr for fine silt (2 micron diameter) to 3000 m/yr for
medium silt (10 microns), to 73000 m/yr for fine sand (50 microns).
*
• WAL, WA,: These are the watershed surface areas for determining loading to the
water body. WAL is the total watershed surface area affected by deposition that drains to
the body of water. WA| is the impervious watershed area affected by deposition that is
guttered and that drains to the water body.
Properly assigning values to WAL and WA, is non-trivial. The areas can be quite
extensive, as the dispersion and deposition model may predict only gradual declines in
deposition as a function of distance from the stack. What is also important to consider is
the watershed hydrology in the absence of any deposition considerations. Total sediments
in a water body may have originated from watershed soils which are (or have the potential
to be) impacted as well as unimpacted by combustor depositions. If the combustor is
depositing principally on a land area which feeds a tributary of a larger river system, then
one should consider what might be termed an "effective" area. An "effective" drainage
9-38
November 10, 1993
-------
DRAFT -- DO SMOT QUOTE OR CITE
area will almost always be less than the total area of a watershed. A "watershed"
includes all the land area which contributes water to a river system. For large river
systems, this area is in the order of thousands of square miles and includes several
tributaries and smaller streams feeding into the main branch of the river. Each stream and
tributary has its own drainage area. If the deposition area can be ascertained to lie within
that drainage area, than it would be appropriate to assign WA L and WA L' based on the
drainage area size.
Another consideration for determining WAL and WA, is the location of the area
affected by deposition fallout with respect to the point where water is extracted for
drinking and fish are caught for consumption. If these points are significantly upstream in
the river system in relation to the deposition area, there is no reason to conclude that
sediments or water near where the water is extracted are impacted by the combustor. If
these withdrawal points are downgradient of the deposition area, then there is reason to
believe that sediments and water are impacted. However, if they are downgradient from
the deposition area but not at the bottom of the watershed, then sediment and water
quality further downgradient from the withdrawal points is not of concern and land
draining into these downgradient portions would not be part of the "effective drainage
area". One further possible consideration is how far upgradient in the watershed one
should go when determining the size of the effective drainage area. While sediments
introduced at the furthest points may eventually work their way down to the mouth of the
watershed, this may take geologic time and not recent historic time. Therefore, sediment
quality near a deposition area need not consider these far reaches.
For a standing water body such as a lake or a pond substantially fed by ground
water recharge, an assumption that is made using the simple framework of this
assessment is that all sediments within the lake/pond are completely mixed. Therefore,
WAL should equal all the area around the lake/pond contributing sediment, and, as in the
above discussion on river systems, a part of the land area contributing sediments to
streams or rivers which may feed the standing water body.
WAW, Vw, Db, Vfx: This set of parameters defines and characterizes the water
9-39
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
body or reach. WAW is the water body surface area, in m2. The volume of the water
column, Vw in m3, and the average flow volume, Vfx in m3/yr, determine the average
residence time of a pollutant in the water body, which can vary from hours in streams to
years in lakes. Average flows can be obtained from gaging stations, or from calculations
using watershed area and average runoff coefficients as summarized in section 9.2.7.
Flows may vary from 105 m3/yr in small streams or ponds draining less than a square
kilometer to 109 m3/yr in large rivers the size of the Potomac. Volumes for a lake or pond
can be calculated as the product of surface area and average depth. Volumes for a stream
reach can be calculated as the product of reach length, average width, and average depth.
The upper benthic sediment depth Db/ representing the portion of the bed in equilibrium
with the water column, cannot be precisely specified. Values from 0.01 to 0.05 m would
be appropriate.
• R, K, LS, P, C, SD, ER: These factors describe the erosion of soil and
contaminant from the watershed to the water body. The first five terms are multiplied
together to give the long-term average soil loss, Xe, in kg/m2-yr. These are the terms for
the Universal Soil Loss Equation, and are also described in Section 9.2.3. of the Indirect
Exposure Document. Following are brief notes on the five terms needed for the Universal
Soil Equation.
• Rainfall/erosivity index, R: The R term represents the influence of precipitation
on erosion, and is derived from data on the frequency and intensity of storms. This value
is typically derived on a storm-by-storm basis, but it has been compiled regionally for the
development of average annual values (Wischmeier and Smith, 1978). Annual values
range from < 50 for the arid western United States to >300 for the Southeast. The
value used in Indirect Exposure Document was 400.
• Soil erodibility, K: The soil erodibility factor reflects the influence of soil
properties on erosion, with values ranging from <0.05 for non-erodible sandy soils to
>0.50 for highly erodible silty soils. The value used in Indirect Exposure Document
9-40
November 10, 1993
-------
DRAFT -- DO 1MOT QUOTE OR CITE
is 0.21.
• Length-slope factor, LS: The topographic factor reflects the influence of slope
steepness and length of the field in the direction of the erosion. Steeper slopes and longer
lengths lead to higher LS values, with a range of 0.1 for slopes < 1.0 percent and lengths
< 100 ft to > 2.0 for slopes generally > 10 percent. The two key considerations for its
assignment, therefore, are the size of the field for which erosion estimates are being made,
and the slope of that field. Indirect Exposure Document uses the USLE to estimate erosion
losses off the edge of a unit area within a watershed, which might be an acre or a hectare,
for example. The LS assigned in Indirect Exposure Document is 0.179, implying flat
terrain and a small unit field. For example, a combination of slope length and slope of
200 meters and 2 percent respectively, gives an LS value of 0.20.
• Support practice factor, P: The support practice factor reflects the use of
surface conditioning, dikes, or other methods to control runoff/erosion. P can be no
greater than 1.0. However, values less than 1.0 should only be assigned when specific
practices are employed which are designed to reduce erosion. P is assigned a value of 1.0
in Indirect Exposure Document.
• Management practice factor, C: The final term in the USLE is the cover and
management practice factor, C, which primarily reflects how vegetative cover and
cropping practices, such as planting across slope rather than up and down slope,
influences erosion. C values can be no greater than 1.0, with this value appropriate for
bare soils. A C value of 1.0 is an appropriate choice for active landfills or sites of high soil
contamination (like Superfund sites) mostly devoid of vegetation. For an inactive landfill
with grass cover or any area with dense vegetative cover such as grass, a value of 0.1 or
less is appropriate. Values greater than 0.1 but less than 0.7 are appropriate for
agricultural row crops, which offer more protection than bare soil, but not as much
protection as dense vegetation. The Indirect Exposure Document assigns a value of 0.5
for C.
9-41
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
• Sediment Delivery Ratio, SD: The sediment delivery ratio for a large land area
(a watershed or part of a watershed) can be calculated based on the area of the watershed
using an approach proposed by Vanoni (1975):
where:
SD
WAL =
b =
a =
SD = a (WAL)
-b
[9-48]
watershed sediment delivery ratio (dimensionless)
watershed area receiving fallout (m 2)
empirical slope coefficient ( = -0.125)
empirical intercept coefficient (see below)
Based on various studies of sediment yields from watersheds, Vanoni concluded that the
sediment delivery ratios vary approximately with the -(1/8) power of the drainage area.
Inspection of the data presented by Vanoni suggests that the "a" parameter in the above
equation varies with the size of the watershed. Although the Agency is currently
evaluating methods for estimating sediment yields, the Vanoni equation along with the
following values for "a" are recommended in the interim:
Watershed Area
(square miles)
"a" Coefficient
< 0.1
1
10
100
1000
2.1
1.9
1.4
1.2
0.6
Note: 1 square mile is equivalent to 2.59x10 6 m 2.
9-42
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
For example, with WA L given in the Indirect Exposure Document as 1.5x10 7 m 2, SD is
estimated as 0.18.
• Enrichment Ratio, ER: Enrichment refers to the fact that erosion favors the
lighter soil particles, which have higher surface area to volume ratios and are higher in
organic matter content. Therefore, concentrations of organic contaminants, which are a
function of organic carbon content of sorbing media, would be expected to be higher in
eroded soil as compared to in-situ soil. While enrichment is best ascertained with sampling
or site-specific expertise, generally it has been assigned values in the range of 1 to 5 for
organic matter, phosphorous, and other soil-bound constituents of concern. The
enrichment ratio would be expected to be higher in sandy soils as compared to silty or
loamy soils because the finer silt particles which erode from a soil generally characterized
as sandy are more a deviation from the norm compared to silt particles which erode from a
soil generally characterized as silty or loamy. A value of three for organic contaminants
would be a reasonable first estimate.
• Kd s, Kd sw, Kd bs: These three adsorption partition coefficients describe the
partitioning of a contaminant between sorbing material, in this case soil, surface water
suspended solids and bed sediments, and water in a sorbing material/water mixture. For
organic contaminants, this partition coefficient has been estimated as a function of the
organic carbon partition coefficient and the fraction of organic carbon in the partitioning
media:
where:
Kd „ =
[9-49]
partition coefficient for pollutant j associated with sorbing material
characterized by organic carbon content, QC ,- (L/kg or cm 3/g)
sorbing material independent organic carbon partition coefficient for
contaminant j (L/kg or cm 3/g)
9-43
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
OC =
fraction organic carbon content of sorbing material (dimensionless)
In the absence of measured values, the Koc can be estimated from a chemical's
octanol water partition coefficient, Kow. Empirical equations relating Kow to Koc are
listed in Lyman, et al. (1982). Of six different equations listed in that reference, the
following derived by Karickhoff, et al. (1979) is recommended for use for chemicals with
high Kow (highly sorbed, hydrophobic contaminants such as PCBs, dioxins):
where:
log Koc = log Kow - 0.21
Koc = organic carbon partition coefficient (L/kg)
Kow = octanol-water partition coefficient (dimensionless)
[9-50]
This equation was empirically developed from a limited number of hydrophobic
contaminants (n = 10, R 2 = 1.00). It implies that Koc is very similar to Kow for strongly
sorbed compounds such as the dioxin-like compounds.
For organic contaminants, what is also then required for assignment of Kd's are the
organic carbon contents of solids and sediments of water bodies. Solids of water bodies
are generally higher than organic carbon contents of the surrounding lands. Furthermore,
organic carbon contents of suspended organic materials and solids are typically greater
than those of bottom sediments. A significant sink for strongly hydrophobic contaminants
such as dioxin-like compounds is thought to be suspended, or non-settling, organic
material. In modeling 2,3,7,8-TCDD in Lake Ontario (EPA, 1990c) using the WASP4
model, a compartment separate from suspended solids termed "non-settling organic
matter" served as a permanent sink. For the framework offered above, a single reservoir
of suspended materials onto which incoming contaminants sorb is modeled.
Foth (1978) lists the organic nitrogen content of several soil types ranging from
sand and sandy loam to clay. The range from that list is 0.0002 to 0.0024 on a fractional
basis. Assuming a carbon to nitrogen ratio of 10 (Brady, 1984; who notes that C:N ratios
9-44
November 10, 1993
-------
DRAFT - DO IMOT QUOTE OR CITE
vary from 8 to 15, with the typical range of 10 to 12), organic carbon contents of soil
generally might range from 0.002 to 0.024. A value that might be assumed to be 0.01 in
order to estimate Kd s for general purposes. The organic carbon content of bottom
sediments will be higher given arguments that erosion favors lighter textured soils with
higher organic carbon contents, and also that bottom sediments are partially comprised of
detritus materials. A range of 0.03 to 0.05, given surface soil content at 0.01, might be
reasonable. The organic carbon content of suspended materials can approach 0.20, but a
range between 0.05 and 0.10 might be more reasonable given prior assignments of 0.01
and 0.03 to 0.05.
The Kd s supplied in Indirect Exposure Document for benzo(a)pyrene and cadmium
are 1.2x10 5 L/kg and 500 L/kg, respectively.
[Revised Section 9.2. ends here]
9.3. Precipitation
Issue The procedure to estimate contaminant concentration in water of precipitation
collection systems may lead to over estimates. The procedure divides the total
deposition rate by the precipitation rate. Some of the contaminant in dry deposition
may have low solubilities and settle out of suspension.
Conclusion/Recommendation
If the collection system causes turbulent mixing of the water, particulates could be
suspended and the current approach would be appropriate. If such mixing does not occur,
the current procedure is acceptable for bounding estimates. However, if more realistic
estimates are desired, the assessor should limit the concentrations of all contaminants to
their solubility limit. For organics further refinement may be possible through the use of
equilibrium partitioning.
9.4. Groundwater
Issue The conclusion that combustor emissions do not impact groundwater was based on
an analysis of two combustors. Other combustors may have different conditions
9-45
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
which would exacerbate groundwater impacts and invalidate this conclusion.
Conclusion/Recommendation
The original examples covered a wide range of combustor types, chemicals and
geohydrologic conditions. However, site-specific conditions may be encountered which
would lead to greater groundwater impacts. In contrast to the original examples,
conditions that might enhance groundwater impacts include higher deposition rates, more
soluble compounds in the emissions and higher recharge rates. In these circumstances,
site-specific groundwater modeling may be required before this pathway can be eliminated
from consideration.
9-46
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
10. DETERMINING EXPOSURE FROM WATER INGESTION
Issue The water concentration proposed for use in estimating exposure is the total water
concentration, which includes the dissolved and sorbed forms. This was
recognized as conservative in the text.
Conclusions/Recommendations
The revised Chapter 9 does provide a model which calculates the dissolved phase
concentration (along with the total concentration and sorbed phase concentration). It is
most appropriate to use this dissolved phase concentration, termed Cdw in Chapter 9. The
basis for this recommendation is that water treatment systems filter out particulates;
hence what remains are the dissolved phase residues of the contaminant.
10-1
November 10, 1993
-------
-------
DRAFT - DO NOT QUOTE OR CITE
11. DETERMINING EXPOSURE FROM FISH INTAKE
11.2 Calculating Daily Intake from Fish
11.2.2 Bioconcentration Factor
lssue The dioxin document promotes the use of a "Biota Sediment Accumulation Factor",
or BSAF, for estimating fish lipid concentrations based on organic carbon
normalized concentrations of contaminants on bottom sediments.
Conclusions/Recommendations
The use of the BSAF is recommended for dioxin-like compounds (i.e., dioxins and
furans with dioxin-like toxicity). BSAF values for these compounds are proposed in the
dioxin exposure reassessment document (EPA, 1992c). Also, some available literature on
BSAFs for PCBs are discussed in that document. A more complete discussion on
bioaccumulation approaches for 2,3,7,8-TCDD, including the BSAF and water-column
based approaches, can be found in "Interim Report on Data and Methods for Assessment
of 2,3,7,8-Tetrachloro-p-dioxin Risks to Aquatic Life and Associated Wildlife" (EPA,
1993c).
While this recommendation is made here, it should be noted that water column
approaches for lipophilic compounds including the dioxins, are currently being used in the
Agency and used appropriately. One such factor described in EPA's Water Quality
Guidance for the Great Lakes Systems (EPA, 1993b) is termed the bioaccumulation factor,
or BAF. This factor, multiplied by the total water concentration, gives a lipid-based fish
tissue concentration. The afore-mentioned Water Quality Guidance contains a list of BAF
values for a number of pollutants. Since the aquatic impact model described in Chapter 9
does predict water column concentrations (including total, dissolved, and sorbed
concentrations), other approaches can easily be used
issue Alternate approaches for estimation of aquatic bioconcentration and
bioaccumulation may also be appropriate.
11-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Conclusions/Recommendations
For estimating bioconcentration and bioaccumulation, there is a model known as
the Food and Gill Exchange of Toxic Substances Model (FGETS). It is a product of ERL-
Athens, GA and has been validated for a number of chemicals.
11.2.3 Fish Ingestion Rate
Issue Ingestion rates specific to the water bodies impacted by the combustor are needed.
In addition to presenting the distributions of exposure, the assessor should use
point- estimate based scenarios to highlight individual exposure estimates for
several locations (i.e. maximum deposition, place where most people exposed, and
areas of special interest, such as schools) . Separate ingestion rates for freshwater
and estuarine fish and shellfish are needed, depending on the nature of the local
surface water bodies. Generally, the ingestion of marine fish and shellfish is not of
interest in indirect exposure estimates of single sources, such as combustors
because, except possibly for anadromous species such as salmon, marine life is not
expected to become contaminated to a significant extent.
Conclusion/Recommendations
The waterway impacted by the combustor should be identified and evaluations
made of how much fish could be produced and caught. Ideally these estimates are made
on the basis of local information. Table 11.1 contains ingestion rates for freshwater and
estuarine fish and shellfish and are offered as a starting point to select site-specific values.
These are based on an analysis of the results of the USDA 1977-78 National Food
Consumption Survey. When using these data the assessor should consider the following
points:
• The survey was conducted over a three day period. Thus, it does not represent long
term behavior patterns which is the interest of exposure assessments used to support
analysis of chronic health effects. This problem introduces uncertainty into the estimates
of medians (50th percentile) and other percentiles. It can provide appropriate estimates of
11-2
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
the average.
• Because most of the persons surveyed did not eat fish or shellfish during the survey
period, the 50th percentile values are zero. The mean values are more appropriate to use
as central tendency estimates of fish and shellfish consumption over a lifetime. However,
these averages are on a per capita basis, ie. averaged across all survey participants
(including fish eaters and nonfish eaters). The average fish consumed by fish eaters is
probably a more relevant estimate of central exposures. This value would be higher than
the per capita average.
• These data represent total ingestion rates of store-bought fish. Obviously, what is of
interest for a site specific survey is the amount of fish consumed from waters within the
study area. Assuming local surveys are not available EPA (1989b) recommends
approaching this problem by using judgement to estimate the number of fish meals (100 to
200 g) per year that a person may reasonably consume from the water body of concern.
By comparing these judgement based values to the national survey data the assessor can
make some evaluation of their reasonableness. EPA (1989b) also presents studies that
indicate that recreational anglers near large water bodies consume 30 g/d (as a mean) and
140 g/d (as an upper estimate). If evidence exists, that subsistence fishing occurs in the
area of interest, then even higher levels may be warranted. Wolfe and Walker (1987)
found subsistence fish ingestion rates up to 300 g/d in a study conducted in Alaska.
11-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Table 11-1. Fish consumption estimates from the USDA 1977-78 National Food Consumption Survey
(consumptions were recorded for three day periods; N = 36249; units are grams/day/person.).
Estimate
Mean
50th %
90th %
95th %
99th %
Fresh
No
Shellfish
1.64
0.00
0.00
5.29
38.00
Fresh
With
Shellfish
1.64
0.00
0.00
5.29
38.00
Estuarine
No
Shellfish
2.50
0.00
4.73
14.50
56.00
Estuarine
With
Shellfish
4.27
0.00
9.80
28.35
80.00
Marine
No
Shellfish
7.72
o.oo
28-33
48.37
93.33
Marine
With
Shellfish
8.23
0.00
30.00
51.17
97.07
Total
No
Shellfish
11.85
o.oo
42.07
66.15
128.00
Total
With
Shellfish
14.15
0.00
51.04
75.60
146.17
Estimate
Mean
50th %
90th %
95th %
99th %
Fresh +
Estuarine
No Shellfish
4.14
0.00
9.80
28.00
76.55
Fresh +
Estuarine
With Shellfish
5.92
0.00
16.53
38.00
101.33
Estuarine +
Marine
No Shellfish
10.22
0.00
37.80
56.70
112.00
Estuarine + Marine
With Shellfish
12.51
0.00
44.73
67.50
129.47
11-4
November 10, 1993
-------
DRAFT -- DO tMOT QUOTE OR CITE
12. DETERMINING EXPOSURE FROM DERMAL ABSORPTION FROM WATER
Issue New Agency guidance is available for addressing dermal absorption exposures.
Conclusions/Recommendations
This chapter should be replaced by the appropriate sections of the report, "Dermal
Exposure Assessment: Principals and Applications" (EPA, 1992b).
12-1
November 10, 1993
-------
-------
DRAFT - DO NOT QUOTE OR CITE
13. HAZARD IDENTIFICATION
Not relevant for this review.
13-1
November 10, 1993
-------
-------
DRAFT -- DO NOT QUOTE OR CITE
14. DOSE-RESPONSE ASSESSMENT
Not relevant for this review.
14-1
November 10, 1993
-------
-------
DRAFT - DO NOT QUOTE OR CITE
15. RISK CHARACTERIZATION
Section 15.2.3 Estimating Total Exposure by Each Route
Issue The Indirect Exposure Document does not fully address how to add risks across
pathways (for carcinogens vs. noncarcinogens) or how to add risks across
chemicals.
Conclusions/Recommendations
Agency guidance for adding risks across pathways are contained in Chapter 8 of
the Risk Assessment Guidance for Superfund: Volume I - Human Health Evaluation Manual
(Part A) (EPA, 1989c). The guidelines for adding risks across chemicals are included in
The Risk Assessment Guidelines of 1986 (EPA, 1987b), in the section titled "Guidelines
for the Health Assessment of Chemical Mixtures" (51 FR 34014). In general, for
carcinogens, risks should be added across chemicals. However, for systemic toxicants,
risks should be added across chemicals only when they target the same organ. Also, risks
should be added across pathways when it is reasonable to expect an individual to
experience exposure by a given set of pathways.
Issue The Indirect Exposure Methodology does not provide guidance on how to combine
the results of the indirect exposure risk assessment with the results of a direct
exposure risk analysis. Individuals will be exposed both directly through inhalation
and indirectly through oral and dermal exposures.
Conclusions/Recommendations
Although the Indirect Exposure Methodology is intended to address only indirect
exposures, in reality individuals will also receive direct inhalation exposures. The air
models discussed in Chapter 3 calculate long-term average ambient air concentrations. For
the purposes of estimating direct inhalation risk, it is important that the ambient air
concentrations of the pollutant represent the sum of both the vapor phase and the fine
particulate phase (i.e., the sorbed phases). This was discussed in the Revised Chapter 3
presented in this Addendum. Except as noted below, the combined vapor and particulate
phase concentrations should generally be used directly for estimating risk. For
15-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
carcinogens, the inhalation risk is best computed by multiplying the compound specific
Unit Risk (m3/mg) by the average (over the exposure period) concentration (mg/m3) of the
contaminant in air at the point of exposure. If Unit Risks are not available, a carcinogenic
slope factor, q " j can be used if it can be demonstrated that it is appropriate for the
inhalation pathway. This procedure involves first estimating the inhalation dose:
Ch IR ET EF ED 1000
h BW AT
[15-1]
where:
ADI h = average daily intake of the hth pollutant (mg/kg/day)
C h = ambient air concentration of the hth pollutant (jjg/m 3)
IR = inhalation rate (m 3/hr)
ET = exposure time (hours/day)
EF = exposure frequency (days/yr)
ED = exposure duration (years)
BW = body weight (kg)
AT = averaging time (days)
1000 = units conversion factor
Carcinogenic risk is calculated as follows:
where:
j.h
Individual Riskh= q*ith ADIh
inhalation carcinogenic slope factor for the hth pollutant
(mg/kg/day) "1
[15-2]
15-2
November 10, 1993
-------
DRAFT - DO IMOT QUOTE OR CITE
and the averaging time for the ADI is taken as a lifetime (e.g., 70 years). The hazard
quotient for non-carcinogens is calculated as follows:
where:
RfCh =
0.001 =
Hazard Quotient* = °h °-001
•" D >P/"t
RfCh
ambient air concentration of the hth pollutant (//g/m 3)
reference concentration of the hth pollutant (mg/m 3)
units conversion factor
[15-3]
In general, the ambient air concentration (which should include both particles and
vapors, as stated before) should not be adjusted in order to calculate the inhalation risk,
unless the particle sizes are above about 10 microns. Below this size, particles are
considered inhalable and represent an inhalation exposure. In general, only fugitive dust
emissions or resuspended dust (e.g., from mechanical disturbance of soil) would be
expected to have a significant particle component in the size range larger than 10 microns.
If significant exposures to particles much larger than 10 microns are projected to occur,
the portion of the ambient air concentration represented by particles above 10 microns
should be considered along with other oral exposures for the purpose of estimating risk.
15.3 Risk Estimation
issue The Indirect Exposure Document does not address population risk. The Agency's
risk characterization guidance recommends that population risk be considered along
with individual risk and that both types of risk descriptors be presented in Agency
risk assessments.
Conclusions/Recommendations
Two approaches can be used to estimate population risks to carcinogens:
1. Simply add the risks based on individual exposures across the population of
15-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
concern, i.e., the entire impacted area, all farmers in impacted area, school children,
etc. This is generally more conveniently estimated by multiplying the average
individual risk by the population size.
2. Base the estimate on the amount of contaminant entering the food supply per
year. This approach is probably most practical for the pathways involving ingestion
of locally produced food. This would involve estimating the annual amount of food
produced in each isopleth ring, computing the contaminant level in the food in each
ring, and adding up the products of contaminant concentration and the amount of
food produced for each ring. Then the population risk is computed as follows:
Population Risk =
FP, C,
[15-4]
where:
q* = cancer slope factor (kg-d/mg)
ED = exposure duration (yr)
BW = body weight (kg)
LT = lifetime (yr)
n = number of rings (dimensionless)
FP j = average annual food production in ring i/365 days per year (kg/d)
C = contaminant concentration in food from ring i (mg/kg)
This approach assumes that all food produced in the study area is consumed (at
any location inside or outside the study area) and that all resulting individual risks are in
the linear range of the dose-response curve. For derivation of the population .risk
equation, see Appendix A. (A similar approach could also be followed for the dairy
products ingestion and fish ingestion pathways, if dairy farming and fishing (recreational,
subsistence, commercial) are important activities in the study area.)
Either of these approaches yields the number of incremental cancer cases occurring
15-4
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
over a 70 year period. The second approach has the advantage that it accounts for
exposure which may occur as a result of exporting food outside of the study area.
However, neither approach accounts for the exposure/population risk which may occur due
to contamination of food produced outside of the study area. Although the individual risks
outside the study area are below levels of concern, the population risks could be important
if significant long-range transport occurs. This issue can be evaluated by examining how
much of the emitted contaminant is deposited in the study area. If 90 percent of the
contaminant is deposited in the study area, then, for carcinogens, approximately
90 percent of the population risk is accounted for. However, the long range transport and
impacts of incinerator emissions are not well understood and the local vs long range
impacts may not be linearly proportional to the relative amounts deposited.
In the case of non-carcinogens, population risk could be characterized in terms of
the number of individuals estimated to have a hazard quotient greater than one (1) during
any portion of their lives, i.e., persons that are exposed to a dose greater than a reference
dose (RfD).
15.4 Characterization of Uncertainty
Issue A discussion of Monte Carlo type assessments of uncertainty is included in the
Indirect Exposure Document. However, little guidance is given on how and when to
use the technique.
Conclusions/Recommendations
As discussed in Chapter 2, using local surveys and a series of point estimates,
one can develop approximations of exposure/risk distributions and estimate exposure levels
for a variety of special interest scenarios. This approach may be sufficient for many
situations. However, if more detailed information is needed on the distribution of exposure
and risk, the assessor can consider using a Monte Carlo simulation analysis. The
decisions about whether and how to conduct Monte Carlo assessments are quite
complicated and cannot be fully addressed here. However, some general guidance can be
provided.
It is critical in all uncertainty analysis and especially those based on Monte Carlo, to
15-5
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
distinguish between variability and uncertainty. Uncertainty results from lack of
knowledge about the true value of a parameter and variability is the change in a parameter
value across a population or situation. The same parameters can have elements of
uncertainty and variability. For example the contaminant concentration in soil can be
uncertain due to measurement or modeling error and spatially variable due to differences in
deposition at various locations. Distinction between these elements can be made in a
variety of ways.
In the framework of Monte Carlo simulation, the impacts of uncertainty and
variability can be systematically addressed using a "double loop" or "nested" simulation
approach. Under this approach the "outer loop" addresses the uncertain parameters, and
applies user specified probability models to make random selections for these terms. For a
given selection of the uncertain parameters, the "inner loop" allows all the population
exposure variables to vary and a distribution of population exposure or risk is obtained.
When this process is repeated for other selections of the uncertain variables a family of
curves representing potential distributions of risk in the population is obtained. These
results can then be applied, for example to estimate the fraction of the population with
exposure above a specified level and to establish error bounds on this fraction. This
approach requires developing distributions of uncertainties, in addition to population
variability and can be quite difficult. In cases where statistically based survey or
measurement data are available, statistical likelihood functions may be used to allow a
non-subjective quantification of uncertainties. However, such data are frequently
(generally?) lacking. Otherwise, it may be possible to select distributions via professional
judgement, but the assessment should clearly state that the outcome is largely dependent
on judgement used to develop the distributions.
An alternative to the double looping approach is to use sensitivity analysis to
demonstrate uncertainty and approximate error bounds. The distinction between
uncertainty and variability should be clearly shown in the final display of a distribution
generated from a Monte Carlo simulation. As an example, uncertainty bands can be
drawn around the distribution of variability.
Unless special provisions are made, all variables used in a Monte Carlo simulation
15-6
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
should be independent. In principle, it is possible to construct a multivariate probability
distribution describing the dependence between two or more variables, however this is not
commonly done due to the greater data requirements and increased complication of the
analysis. None the less, when strong dependencies between variables are present, they
may have an important impact on the conclusions of the analysis. This is particularly true
. when some individuals may have a tendency to have high end values for two or more
variables. Where dependencies can be described, a useful approach is to determine a
distribution for one and use values from this distribution to compute the second variable
(using a conditional distribution for the second variable given knowledge of the first
variable). The assessor must avoid situations where two or more values are assigned to a
single variable where it appears in different places in a risk calculation. If multiple
assignments are inadvertently made, they will inappropriately alter the shape of the
resulting probability distribution.
Inconsistencies can arise between exposure and toxicity assessments because of
fixed exposure assumptions imbedded in toxicity metrics. For example, RfCs, RfDs and
cancer slope factors have assumptions built into them about breathing rates, life
expectancy, etc.
It is often difficult to find appropriate distributions for use in Monte Carlo
assessments. Some general guidance for selecting distributions is presented below:
• Distributions derived from national data may not represent local conditions. To extent
possible use distributions or frequency histograms derived from local surveys.
When distributional data are drawn from a national or other surrogate population, it is
important that the assessor address the extent to which local conditions may differ from
the surrogate data. In addition to a qualitative statement of uncertainty, the
representativeness assumption is appropriately addressed as part of a sensitivity analysis.
• Many surveys (both local and national) addressing issues such as food consumption
were conducted over a short time period for each respondent and may not represent long
term trends (the respondents usual behavior). Generally, environmental risk assessments
for chronic health effects address risk as a function of the long term average exposures of
15-7
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
individuals, and thus are most appropriately conducted using data on long term average,
"usual" behavior. Similar problems arise when distributions of concentration data arise
from short term monitoring periods or when factors such as soil ingestion are measured
using protocols of a duration of only several days.
» Values near the tail of a distribution are very sensitive to the form of distribution
selected. Tails of ideal distributions extend to infinity. Consider using empirical
histograms when substantial data are available or truncating ideal distributions at
reasonable bounding estimates.
Issue The use of mass balance checks was not mentioned as a means of evaluating
uncertainty. A mass balance would be violated, for example, if the mass of
contaminant estimated to get into a media (for example soil) exceeds the mass
being emitted from the stack.
Conclusion/Recommendations
The COMPDEP air emissions model is designed on mass balance principles
and therefore should not lead to impossible air concentrations or deposition rates which
exceed the emission rates. Depositions are substracted out of the airborne reservoir of
particle-bound contaminants as the depositions occur. However, some of the media
transfer models, which use the output from the air models, are based on simple
transfer/partitioning approaches which do not have built in mass balances. If these models
are properly parameterized, then the possibility of meaningful mass balance violations is
minimized. This is discussed further below. However, it is generally recommended that
assessors make mass balance checks to ensure that reasonable predictions are being
made. These checks can be made in various ways. One way is to compare estimates of
the amounts of contaminant in the various media (i.e. soil, vegetation and biota) within the
study area (i.e., the area within the outer isopleth boundary, as described in Chapter 2 of
this Addendum) to the amount emitted from the stack. Comparable time frames must be
used for this type of comparison. For example, the amount of contaminant in the beef
annually produced from the study area should not exceed the amount of contaminant
emitted during the year (except in an unusual circumstance where a large cattle ranch
15-8
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
begins operation within the study area several years after incineration begins, and
accumulation in the environment to that point could lead to large withdrawals by the cattle
initially). In addition to fate, transport, and transfer mass balance transfer, one should also
be aware of mass balance violations when it comes to estimating exposures given media
concentrations. For example, the amount of impacted beef ingested by populations in the
study area should not exceed the amount of beef produced in the study area.
Because the fate and transport, and food chains algorithms, do not contain built-in
mass balance checks, mass balance violations could occur. Contaminants estimated to be
deposited onto soils partition between a sorbed phase and a dissolved phase. Transfers
occur into below ground and above ground vegetation with a biotransfer factor.
Bioconcentration factors take concentrations in one media, water concentrations for
example, and translate them to a concentration in another media, fish. In this simple
framework, mass balances are considered but are not rigorously maintained. For example,
soil to plant transfers are not explicitly modeled as a loss for soil concentration estimation.
However, there is an overall loss constant for the soil concentration algorithm, which
explicitly considers leaching, erosion, runoff, degradation, and volatilization. One could
estimate the loss via plant uptake and add that to the degradation rate constant;
otherwise, plant uptake loss is not explicitly considered.
There are other examples of this kind where exchanges between media and sinks
are not rigorously modeled. However, this particular type of theoretical mass balance
issue is not normally expected to be a fatal flaw for any contaminant evaluated. First,
dissipation from key media is modeled with loss rate constants - these key media include
soils, vegetation, and surface waters (water column and benthic sediments). Guidance
has been presented for proper parameterization of these dissipation factors in the 1990
IED as well as this Addendum. Second, all transfer and bioconcentration parameters have
been empirically developed from appropriate data sets - these key transfer factors include
soil and air to plants, contaminant intake by terrestrial animals to contaminant
concentrations in food products, and aquatic media concentrations (water or sediment) to
fish. These empirical factors are, by design and definition, approximations of observed
phenomena.
15-9
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Another type of mass balance violation could occur with improper parameterization.
For example, the aquatic impact requires the methodology user to estimate the land area
draining into an impacted water body as well as a contaminant deposition rate that
represents depositions over that entire area. If one selects a deposition rate that would
occur at the point of maximum deposition, within a few hundred meters of the stack
emission, as representing deposition over a drainage basin that is millions of square meters
!n area, and/or one estimates a drainage basin size much too large for an impacted water
body (which is defined in this methodology in terms of flow rates and volumes), one could
very easily obtain results that obviously violate mass balance considerations - that an
impacted surface water bodies receives more contaminant over a multi-year period in
runoff, erosion, and direct deposition than is emitted from the stack over that multi-year
period.
Proper parameterization of the model is the key to not violating principals of mass
balance. The following is offered as pointers to consider for mass balance concerns:
1) No parameter is trivial: All parameters used in the modeling should be presented for
others to scrutinize. Careful attention should be paid to the stack emission rate assumed
for air transport modeling, as every subsequent media concentration is proportionally
related to it. All fate, transport, and transfer parameters specific to the contaminant are
the second most important model parameters. Insure that all parameters are in the proper
units. If an air concentration is presented to the assessor in units of //g/m3, and the food
chain models require its units to instead be in pg/m3, then an oversight in converting will
result in all media concentrations related to air concentrations (i.e., plants, animal food
products, and water and fish of the surface water body) to be overestimated by a factor of
a million.
2) Do reality checks with estimated media concentrations: This is, of course, easier said
than done. It means that assessors should evaluate their predicted concentrations with
observed concentrations in the literature. If predicted concentrations exceed observations
by an order of magnitude, for example, one should very carefully evaluate model input
15-10
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
parameters for mistakes.
While this addendum cannot offer a comprehensive tabular summary of media
concentrations for key contaminants, the following is offered. As part of EPA's
reassessment of dioxin-like compounds, exposure media concentrations were compiled for
these compounds. In addition to a comprehensive survey of the literature, a principal
purpose of this compilation was to estimate background exposures to dioxin-like
compounds. Dioxin-like compounds are known to be ubiquitous in the environment, and
concentrations are present where there is no immediate source of release. Literature
reports of exposure media concentrations which were evaluated as best representing
background conditions - not directly associated with a contaminated site or downwind of a
incinerator known to be emitting dioxin-like compounds - were culled out and reviewed.
Average concentrations of dioxin toxic equivalents (TEQ) were estimated for these
exposure media:
Air
Soil
Fish
Water
Milk
Beef/veal
0.095 pg/m3
8.0 ng/kg (ppt)
1.2 ng/kg (ppt whole fish of various lipid contents)
0.0056 pg/L (ppq)
0.07 ng/L (ppt whole milk at 3.5% fat)
0.48 ng/kg (ppt whole product assuming 19% fat)
After conducting an assessment for dioxin TEQs, assessors should compare their predicted
exposure media with the above. Since the above values were selected to represent
background levels, it may be possible that these levels are exceeded near a source.
However, if predicted concentrations are significantly higher than these, then a careful
review of input parameters and intermediate exposure media should occur. A first check
would be on air concentrations predicted by the air transport model for dioxin TEQs. The
above concentration at 0.095 pg/m3 was an average for urban/suburban areas in the
United States. An evaluation of available air concentrations in rural areas where no source
was immediately nearby indicated that air concentrations in these areas are about 5 times
15-11
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
lower than urban concentrations, or at about 0.020 pg/m3. If an air prediction of TEQs is
1.0 pg/m3, which is about an order of magnitude higher than average urban conditions and
50 times higher than rural conditions, one should carefully evaluate emission assumptions
and all other air transport modeling parameters. If the air prediction of TEQs appears more
in bounds, such as 0.01 pg/m3, but the beef concentration is 5.0 ppt, which is an order of
magnitude higher than 0.48 ppt, than again the assessor has to review his food chain
input parameters for errors. Any other evaluations of model performance, such as further
comparisons with literature data or sensitivity analyses, will assist in addressing concerns
about mass balance violations.
15-12
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
III. RECOMMENDATIONS FOR LONG-TERM IMPROVEMENT OF MULTIMEDIA RISK
MODELING
1. Air Dispersion and Deposition
Although ISC-COMPDEP will be based on the EPA's ISC2 and COMPLEX I
regulatory dispersion models, its application will be in a different context than required by
the Clean Air Act. Therefore, the subgroup recommends that ISC-COMPDEP be peer
reviewed for indirect exposure analysis.
The subgroup recommends longer term development (1 to 2 years) of the next
generation of air dispersion models. This would involve various projects, most importantly:
1. the development of the modeling capability to make reliable predictions of the
wet and dry deposition of chemicals that exist in the atmosphere in the vapor
phase. The LD!F term to revised Equation [4-1] of this Addendum is recommended
for diffusive entry of vapor-phase contaminants into soil. This could answer the
need for an algorithm for dry deposition of vapors into soils. Currently, there is no
approach for wet deposition of vapors;
2. the development of a more sophisticated model for use in complex terrain;
3. the evaluation of algorithms for wet deposition of particles.
In the development of these models, consideration should be given to integrating
the modules that estimate bioaccumulation in food chain, surface runoff in the receiving
waters and other components of the indirect exposure methodology. This would combine
all modeling efforts into one assessment tool.
It is important that air deposition modeling be kept current with the developments
in the state-of-the-science. The short-terrn recommendations are not sufficient for
developing the state-of-the-science deposition model. Therefore, the subgroup provides
the following long-term recommendations.
111-1
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
1. There is a need to develop a suitable deposition algorithm for gases. Once
developed and tested, such an algorithm can be put into the ISC2 model and a
future version of ISC-COMPDEP.
2. The dry deposition algorithm in ISC2 was not intended for complex terrain
applications. In order to determine if the algorithm is applicable to complex terrain,
additional testing and, if necessary, modifications are needed.
3. It is necessary to develop and evaluate suitable wet deposition algorithms.
4. For national applications of the ISC-COMPDEP model, a species-dependent
chemical data base is needed. For example, the dry deposition algorithm requires
site specific data on particle sizes and densities and gas deposition algorithms
require data on chemical properties. Such data are not readily available.
5. For national applications of ISC-COMPDEP, a workbook example guidance
document to illustrate all the steps needed for indirect exposure analysis would be
very useful.
6. If refined concentration estimates are desired for receptors in complex terrain,
then there is a need to further update the ISC-COMPDEP model. COMPLEX I is a
screening level model, and research by ORD/AREAL has identified a more refined
model (i.e., CTDMPLUS). It is a major effort to modify this model to include a
deposition algorithm.
7. Climatological models can be used to provide annual average estimates with
lower costs than the models mentioned above which require on-site hourly data.
However, research is needed for developing a climatological model for complex
terrain that uses the CTDMPLUS state-of-the-science formulation, a suitable dry and
MI-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
wet deposition algorithm for this model.
2. Indirect Exposure Analysis - General
The procedures described in this Addendum and the Indirect Exposure Document
are not currently compiled into a single integrated model. The COMPDEP model is an
independent computer code. There is no computer tool, however, which takes the air
concentrations and depositions from COMPDEP, and using algorithms of this Addendum
and the Indirect Exposure Document, estimates exposure media concentrations and
subsequent exposures. Such a model should have several software features.
First, it should be interactive. This could be accomplished by writing an indirect
exposure methodology shell that would call various data input, process, and post-process
modules. Options for such a shell include standard languages such as C+ +, or Windows.
Second, it should be integrated. Calculated concentrations and fluxes in individual media
modules should be transferred to other media modules via common blocks or arguments.
The user should also have the opportunity to directly specify loadings to individual media.
Third, it should be modular. Users should be able to substitute alternate modules for
individual media quickly and easily. This would allow specialized applications (e.g., for
mercury) as well as evolutionary improvement of the overall system with time. Finally, it
should be linked to a statistical driver. This would facilitate such operations as automated
sensitivity analysis, first-order error analysis, and Monte Carlo analysis. These operations
should encourage users to explore what controls the uncertainty in the overall risk
estimates.
Having such a software tool is the first step in the long term improvement of
multimedia risk assessment. The proper parameterization of each application will always
be a major issue. Supporting national databases should be made easily accessible to the
user. For example, the interactive program DBAPE (Data Base and Parameter Estimation
for Soils) is available to all users, distributed through the Center for Exposure Assessment
Modeling. Eventually, such programs should be accessible through the multimedia shell, or
through an accompanying geographical information system.
1-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Some multimedia applications inevitably will have to be specialized for particular
chemicals or locations. A linked set of more sophisticated mass balance and chemical
speciation models should be available for these special applications, for testing and
improving components of the nationwide multimedia model, and for parameterizing
regional or nationwide applications.
Because mercury is presently an important and difficult contaminant, the
development of state-of-the-art compartment mass balance models for mercury would be
useful. Recommendations for such a model are being considered, but are considered
beyond the scope of this document.
3. Animal Tissue Biotransfer Factors and Food Consumption Rates
The mass intake to whole body concentration approach used in the Indirect
Exposure Document may not be the optimum approach for the determination of pollutant
concentration in animal tissue. A concentration to concentration approach for the
bioconcentration transfer factor should be considered for the long run. Also, recognizing
the importance of the animal intake assumptions (fraction of the intakes in grain, forages,
and silage, etc.) the assumptions made for the current draft should be reevaluated for
accuracy.
1-4
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
IV. REFERENCES
Bacci, E., D. Calamari, C. Gaggi, M. Vighi, 1990. Bioconcentration of Organic
Chemical Vapors in Plant Leaves: Experimental Measurements and
Correlation. Environ. Sci. Technol. 24: 885-889.
Bacci, E., M.J. Cerejeira, C. Gaggi, G. Chemello, D. Calamari, M. Vighi, 1992.
Chlorinated Dioxins: Volatilization from Soils and Bioconcentration in Plant Leaves.
Bulletin of Environmental Contamination and Toxicology 48(3):401-408.
Bidleman, T.F., 1988. Atmospheric Processes. Wet and Dry Deposition of Organic
Compounds are Controlled by Their Vapor-Particle Partitioning. Envion. Sci.
Techol., 22:4, pp 361-367.
Bowers, J.F., J.R. Bjorklund and C.S. Cheney, 1979. Industrial Source Complex (ISC)
Dispersion Model User's Guide. Volume I. EPA/450/4-79/030. U.S. Environmental
Protection Agency, Research Triangle Park, NC.
Brady, N.C., 1984. The Nature and Properties of Soils. Ninth Edition. New York,
NY: Macmillan.
Briggs, G.G., R.H. Bromilow, and A.A. Evans, 1982. Relationships between lipophilicity
and root uptake and translocation of non-ionized chemicals by barley. Pesticide
Science. 13:495-504.
Buonicore, A.J., W.T. Davis, eds. 1992. Air Pollution Engineering Manual. Air and Waste
Management Association, Van Nostrand Reinhold, New York, N.Y.
California Air Resources Board (CARB), 1986. Subroutines for calculating dry deposition
velocities using Sehmel's curves. Prepared by Bart Croes, California Air Resources
Board.
California Air Resources Board (CARB), 1990a. Health Risk Assessment Guidelines for
Nonhazardous Waste Incinerators. Prepared by the Stationary Source Division of
the Air Resources Board, and the California Department of Health Services.
California Air Resources Board (CARB), 1990b. Technical Support Document to Proposed
Dioxins Control Measures for Medical Waste Incinerators. May 25, 1990.
California Air Resources Board (CARB), 1991. Air Pollution Control at Resource Recovery
Facilities. 1991 update.
R-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
Catalano, J.A., D.B. Turner, and J.H. Novak, 1987. User's Guide for RAM-Second
Edition. EPA/600/8-87/046. U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Committee of the California Air Pollution Control Officers Association (CAPCOA), 1991.
Air toxics "hot spots" program. Risk assessment guidelines.
Dempsey, C.R. and E.T. Oppelt, 1993. Incineration of hazardous waste: a critical review
update. Air and Waste. 43:25-73.
Foth, H.D, 1978. Fundamentals of Soil Science. 6th ed. New York: John Wiley and
Sons.
Geraghty, J.J., D.W. Miller, F. Vander Leeden, and F.L. Troise, 1973. Water Atlas of the
U.S. Port Washington, NY: Water Information Center.
Karickhoff, S.W., D.S. Brown, and T.A. Scott, 1979. Sorption of hydrophobic pollutants
on natural sediments. Water Research 13: 241-248.
Habicht, F.H., 1992. Guidance on Risk Characterization for Risk Managers and Risk
Assessors. Memorandum from the Deputy Administrator to the Assistant
Administrators and Regional Administrators dated February 26, 1992.
U.S. Environmental Protection Agency, Washington, D.C.
Irwin, J.S., and J.O. Paumier, 1990. Meteorological Processor for Regulatory Models
(MPRM 1.2) User's Guide (Revised). EPA/600/3-88-043, U.S. Environmental
Protection Agency, Research Triangle Park, NC.
Lyman, W.J., W.F. Reehl, and D.H. Rosenblatt, 1982. Handbook of Chemical Property
Estimation Methods. New York: Mcgraw-Hill.
McCrady, J.K., and S.P. Maggard, 1993. Uptake and photodegradation of 2,3,7,8-
tetrachlorodibenzo-p-dioxin sorbed to grass foliage. Env. Sci. Technol 27:343-350.
McKone, T.E., and P.B. Ryan, 1989. Human exposures to chemicals through the food
chain: an uncertainty analysis. Environmental Science and Technology.
23:1154-1163.
O'Conner, D.J. and W.E. Dobbins, 1958. Mechanism of Reaeration in Natural Streams.
ASCE Transactions, pp. 641-684. Paper No. 2934.
O'Conner, D.J., 1983. Wind effects on gas-liquid transfer coefficients. Journal of
Environmental Engineering. 109(9):731 -752.
R-2
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
PEI Associates, Inc. and H.E. Cramer Company, Inc., 1986. Air Quality Modeling Analysis
of Municipal Waste Combustors. U.S. Environmental Protection Agency, Research
Triangle Park, N.C.
Rao, K.S., 1981. Analytical Solutions of a Gradient-Transfer Model for Plume Deposition
and Sedimentation. NOAA Technical Memorandum ERL ARL-109.
Rao, K.S., and L. Satterfield, 1982. MPTER-DS: The MPTER Model Including Deposition
and Sedimentation. EPA/600/8-82/024. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Sehmel, G.A., 1980. Particle and gas dry deposition: A review. Atmospheric
Environment 14:983-1011.
Sehmel, G.A., and W.H. Hodgson, 1978. A Model for Predicting Dry Deposition of
Particles and Gases to Environmental Surfaces. PNL-SA-6721. Battelle, Pacific
Northwest Laboratory, Richland, WA.
Seinfeld, J.H., 1986. Atmospheric Chemistry and Phvsics of Air Pollution. John Wiley
and Sons, NY).
Slinn, W.G.N., 1984. Precipitation scavenging. In Atmospheric Science and Power
Production., ed. Darryl Randerson. DOE/TIC-27601. U.S. Department of Energy.
Smith, A.M., 1987. Infant exposure assessment for breast milk dioxins and furans derived
from waste incineration emissions. Risk Analysis. 7(3):347-353.
Stewart, B.A., D.A. Woolhiser, W.H. Wischmeier, J.H. Caro, and M.H. Frere, 1975.
Control of Water Pollution From Croplands, Vol. I. U.S. Environmental Protection
Agency, Washington, D.C. EPA-600/2-75-026a.
Sullivan, M. J., S. R. Custance, and C. J. Miller, 1991. Infant exposure to dioxin in
mother's milk resulting from maternal ingestion of contaminated fish.
Chemosphere. 23(8-10): 1387-1396.
Travis, C.C., H.A. Hattemer-Frey, and A.A. Arms, 1988. Relationship between Dietary
Intake of Organic Chemicals and Their Concentrations in Human Adipose Tissue and
Breast Milk. Arch. Environ. Contam. Toxicol. 17:473-478.
U.S. Department of Agriculture. 1983. Food Consumption: Households in the U.S.,
Seasons and Year 1977-1978. Consumer Nutrition Division, Hyatsville, MD. NFCS
1977-78. Report No. H-6. USDA Nationwide Food Consumption Survey .
R-3
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
U.S. Department of Agriculture. 1966. Household Food Comsumption survey, 1965-1966.
Report 12. Food Consumption of Households in the U.S. - Season and Year 1965-
1966.
U.S. Environmental Protection Agency, 1980. Environmental Assessment of a Waste-to-
Energy Process. Braintree Municipal Incinerator. Office of Research and
Development, Washington, DC., EPA-600/7-80-149.
U.S. Environmental Protection Agency, 1985a. Compilation of Air Pollutant Emission
Factors. Fourth Edition. Office of Air Quality Planning and Standards, Research
Triangle Park, N.C. AP-42.
U.S. Environmental Protection Agency, 1985b. Water Quality Assessment: A Screening
Procedure for Toxic and Conventional Pollutants in Surface and Ground
Water - Part I (Revised). Office of Research and Development, Athens, GA.
EPA/600/6-85/002a.
U.S. Environmental Protection Agency, 1986. Characterization of stack emissions from
municipal refuse-to-energy systems. Atomspheric Sciences Research Laboratory,
Research Triangle Park, N.C.
U.S. Environmental Protection Agency, 1987a. On-site Meteorological Program
Guidance for Regulatory Modeling Applications. Office of Air Quality
Planning and Standards, Research Triangle Park, N.C.
U.S. Environmental Protection Agency, 1987b. The Risk Assessment Guidelines of 1986.
Office of Health and Environmental Assessment, Washington D.C.
EPA/600/8-87/045.
U.S. Environmental Protection Agency, 1987c. Municipal waste combustion study.
Emission data base for municipal waste combustors. Office of Air Quality Planning
and Standards, Research Triangle Park, N.C. EPA/530-SW-87-021b.
U.S. Environmental Protection Agency, 1988. Control of Open Fugitive Dust Sources.
Office of Air Quality Planning and Standards, Research Triangle Park, N.C.
EPA-450/3-88-008.
U.S. Environmental Protection Agency, 1988b. Hospital waste combustion study: data
gathering phase. Office of Air Quality Planning and Standards, Research Triangle
Park, N.C. EPA-450/3-88-017.
U.S. Environmental Protection Agency, 1989a. Development of Risk Assessment
Methodology for Land Application and Distribution and Marketing of Municipal
Sludge. Office of Health and Environmental Assessment, Washington, D.C.
EPA/600/6-89/001.
R-4
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
U.S. Environmental Protection Agency, 1989b. Exposure Factors Handbook. Office of
Health and Environmental Assessment, Exposure Assessment Group, Washington,
D.C. EPA/600/8-89/043.
U.S. Environmental Protection Agency, 1989c. Risk Assessment Guidance for Superfund:
Volume I: Human Health Evaluation Manual (Part A). Office of Emergency and
Remedial Response, Washington, D.C. EPA/540/1-89/002.
U.S. Environmental Protection Agency, 1989d. Health Effects Assessment Summary
Tables, Fourth Quarter FY 1989. Office of Solid Waste and Emergency Response,
Washington, DC. OERR 9200 6-303-894.
U.S. Environmental Protection Agency, 1990a. Methodology for Assessing Health Risks
Associated with Indirect Exposure to Combustor Emissions, Interim Final. Office of
Health and Environmental Assessment, Washington, D.C. EPA/600/6-90/003.
U.S. Environmental Protection Agency, 1990b. User's Manual for the PM-10 Open
Fugitive Dust Source Computer Model Package. Office of Air Quality Planning and
Standards, Research Triangle Park, N.C. EPA-450/3-90-010.
U.S. Environmental Protection Agency, 1990c. Lake Ontario TCDD Bioaccumulation Study
Final Report. Cooperative study including US EPA, New York State Department of
Environmental Conservation, New York State Department of Health, and Occidental
Chemical Corporation. May 1990.
U.S. Environmental Protection Agency, 1991. Guidance for Risk Assessment. Risk
Assessment Council. Office of Regulatory Management and Evaluation,
Washington, D.C.
U.S. Environmental Protection Agency, 1992a. Guidelines for Exposure Assessment.
Federal Register Publication on May 29, 1992 (57 FR 22888). EPA/600-Z-92/001.
U.S. Environmental Protection Agency, 1992b. Dermal Exposure Assessment: Principles
and Applications. Office of Health and Environmental Assessment, Washington,
D.C. EPA/600/891/011B.
U.S. Environmental Protection Agency, 1992c. Estimating Exposure to Dioxin-Like
Compounds. Draft Report. Office of Research and Development, Washington, D.C.
EPA/600/6-88/005B.
U.S. Environmental Protection Agency, 1992d. Workbook of Screening Techniques for
Assessing Impacts of Toxic Air Pollutants. EPA-454/R-92-024. Office of Air
Quality Planning and Standards, Research Triangle Park, NC. December, 1992.
R-5
November 10, 1993,
-------
DRAFT - DO NOT QUOTE OR CITE
U.S. Environmental Protection Agency, 1993a. Guideline on Air Quality Models (Revised).
EPA/450/2-78/072R. Office of Air Quality Planning and Standards, Research
Triangle Park, NC.
U.S. Environmental Protection Agency, 1993b. Federal Register Notice. Water Quality
Guidance for the Great Lakes System and Correction; Proposed Rules. April 16.
58(72)20802-21047.
it
U.S. Environmental Protection Agency, 1993d. Interim Report on Data and Methods for
Assessment of 2,3,7,8-Tetrachloro-p-dioxin Risks to Aquatic Life and Associated
Wildlife. Office of Research and Development, Duluth, MN. EPA/600/R-93/055.
Vanoni, V.A., editor 1975. Sedimentation Engineering. American Society of Civil
Engineers, New York, NY. pp. 460-463.
Wolfe, RJ and RJ Walker. 1987. Subsistence Economics in Alaska: Productivity,
Geography and Developmental Impacts. Arctic Anthropology 24(2):56-81.
Winges, K.D., 1990. User's Guide to the Fugitive Dust Model (FDM) (revised). Volume 1:
User's Instructions. EPA-910/9-88-202R. U.S. Environmental Protection Agency,
Region 10, Seattle, WA.
Wischmeier, W.H. and D.D. Smith. 1978. Predicting Rainfall Erosion Losses -A
Guide to Conservation Planning. Agriculture Handbook No. 537. U.S. Dept
of Agriculture, Washington, D.C.
Whitby, K. T. 1978. The physical characteristics of sulfur aerosols. Atmospheric
Environment. 12:135-159.
R-6
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
Appendix A: Approaches to Estimating Population Risk from Food Production
1. By definition:
Population Risk = £ IndividualRisks
[A-1]
2. Assuming all individual risks are in the linear dose-response range:
where:
q'
d
n
Population Risk = q * £ d/
1 = 1
= n q * davg
cancer slope factor (kg-day/mg)
dose (mg/kg-day)
population size (dimerisionless)
[A-2]
3. For example, the total population risk from dioxin in each type of food supply in the
U.S. is:
where:
q*
nus
USavg
Population Risk = q * nus dUSavg
cancer slope factor (kg-day/mg)
U.S. population size (dimensionless)
U.S. average dose (mg/kg-day)
[A-3]
A-1
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
4. Expanding the dose term:
where:
q*
nus =
USavg =
IR
ED
BW
LT
Population Risk = Q' "us CuS^IRUSavg ED
cancer slope factor (kg-day/mg)
U.S. population size (dimensionless)
U.S. average concentration of dioxin in the food (mg/g)
ingestion rate of the food (g/d)
exposure duration (yr)
body weight (kg)
lifetime (yr)
[A-4]
5. Assuming that all the food produced is consumed, then:
where:
FP
us -
n
us -
IR
USavg
FPUS = nUS IR USavg
U.S. production of the food (g/d)
U.S. population size (dimensionless)
U.S. average ingestion rate of the food (g/d)
6. Substituting FP us into Equation [C-4] produces:
[A-5]
Population Risk =
[A-6]
where:
A-2
November 10, 1993
-------
DRAFT -- DO NOT QUOTE OR CITE
q* = cancer slope factor (kg-day/mg)
c usavg = u-s- average concentration of dioxin in the food (mg/g)
FP us = U.S. production of the food (g/d)
ED = exposure duration (yr)
BW = body weight (kg)
LT = lifetime (yr)
7. Assume one is interested in risks created by dioxin contaminated food produced in area
X but potentially consumed anywhere in the U.S. The concentration term then becomes:
= mass dioxin contaminated food produced within area X
total U.S. food production
_CXFPX
[A-7]
FP,
where:
USavg
X
FP
FP
x -
us =
us
U.S. average concentration of dioxin in the food (mg/g)
concentration of dioxin in the food from area X (mg/g)
production of the food in area X (g/d)
U.S. production of the food (g/d)
8. Substituting for C USavg into Equation [C-6]:
Population Risk = Q* E° °x FP*
[A-8]
where:
q*
ED
cx
FP
X
cancer slope factor (kg-day/mg)
exposure duration (yr)
concentration of dioxin in the food from area X (mg/g)
production of the food in area X (g/d)
A-3
November 10, 1993
-------
DRAFT - DO NOT QUOTE OR CITE
BW
LT
body weight (kg)
lifetime (yr)
9. Assume area X is divided into 5 rings with differing amounts of the food production
and dioxin levels in the food:
Population Risk = %' ED J^FP-, C,
DVV LI
[A-9]
where:
q* = cancer slope factor (kg-day/mg)
ED = exposure duration (yr)
BW = body weight (kg)
LT = lifetime (yr)
FP j = production of the food in the ith ring of area X (g/d)
C j = concentration of dioxin in the food from the ith ring of area X (mg/g)
A-4 November 10, 1993
•{JU.S. GOVERNMENT PRINTING OFFICE; 1993 - 550-064/80001
------- |