EPA 530-R-23-004
Beneficial Use Evaluation:
Flue Gas Desulfurization Gypsum as
an Agricultural Amendment
March 2023
Prepared By:
United States Environmental Protection Agency
Office of Land and Emergency Management
Office of Resource Conservation and Recovery
U.S. Department of Agriculture
Agricultural Research Service
and
RTI International
EPA Contract No. EP-W-15-005
** rnA ynited states _ USDA Agricultural Research Service
Environmental Protection — . °
Agency ^9^1 U.S. DEPARTMENT OF AGRICULTURE
-------
Disclaimer
This document was prepared by the United States Environmental Protection Agency (EPA) Office
of Resource Conservation and Recovery (ORCR) and the U.S. Department of Agriculture (USDA)
Agricultural Research Service (ARS). Any opinions, findings, conclusions, or recommendations do
not change or substitute for any statutory or regulatory provisions. This document does not impose
legally binding requirements, nor does it confer legal rights, impose legal obligations, or implement
any statutory or regulatory provisions. Mention of trade names or commercial products is not
intended to constitute an endorsement or recommendation for use.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
Executive Summary
The United States (U.S.) Environmental Protection Agency (hereafter "EPA" or "the Agency")
Sustainable Materials Management (SMM) Program supports the productive and sustainable use of
resources throughout all stages of their lifecycles, from resource acquisition through disposal. The
SMM Program seeks to avoid or minimize adverse impacts to the environment while accounting
for economic efficiency and social considerations. The beneficial use of nonhazardous industrial
materials presents a significant opportunity to advance the goals of the SMM Program by providing
economic benefits, preserving natural resources, and avoiding negative environmental impacts
associated with acquisition and processing of virgin raw materials. Therefore, EPA supports the
beneficial use of these industrial materials when done in a manner that is protective of human
health and the environment.
State, tribal and territorial regulatory bodies often make the determination whether or not to allow
a given beneficial use within their jurisdiction. Although requests for such determinations have
increased over time, it has been reported that insufficient information about the potential impacts
to human health and the environment from these uses has been a major barrier to making decisions
about proposed beneficial uses. To help address this barrier, EPA developed two documents:
Methodology for Evaluating the Beneficial Use of Industrial Non-Hazardous Secondary Materials
and Beneficial Use Compendium: A Collection of Resources and Tools to Support Beneficial Use
Evaluation ("the Methodology" and "the Compendium," respectively). These documents provide
an analytical framework that can be used to evaluate the potential for adverse environmental
impacts from a wide range of industrial materials and their proposed beneficial uses, as well as a
list of existing resources and tools that can assist with these evaluations.
The primary purpose of this document is to demonstrate how the analytical framework from the
Methodology and Compendium can be applied to a real-world beneficial use scenario, specifically
the use of flue gas desulfurization (FGD) gypsum as an agricultural amendment. FGD gypsum is a
type of coal combustion residual (CCR) generated from the pollution control technologies designed
to reduce sulfur gas emissions from electric utilities. FGD gypsum can substitute for mined gypsum,
which is a mineral that occurs naturally in sedimentary rock formations, because both materials
are composed primarily of calcium sulfate. FGD gypsum has been shown to offer a range of benefits
when applied to fields, such as a providing key nutrients to crops and limiting phosphorus runoff
to nearby water bodies. Yet there is also potential for higher levels of some trace contaminants in
FGD gypsum as a result of the industrial process that generates this material, which warranted
further evaluation to ensure that application of this industrial material will not harm human health
or the environment.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
As detailed in this document, EPA sequentially applied each step of the analytical framework,
culminating in a national-scale probabilistic model of potential environmental fate and transport.
No concerns were identified for the vast majority of modeled application scenarios. Some limited
potential for risk was identified from release of selenium to surface water when FGD gypsum is
applied on across every available field at the highest rates and frequencies. Yet even in this extreme
and unlikely scenario, identified risks can be mitigated through minor limits on application
practices. Based on these results, the beneficial use of FGD gypsum can provide meaningful benefits
to agricultural fields while remaining protective of human health and the environment.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
Table of Contents
1. Introduction 1-1
1.1. Background 1-1
1.2. Document Organization 1-2
2. Planning and Scoping 2-1
2.1. Background 2-1
2.2. Flue Gas Desulfurization Gypsum 2-1
2.3. Agricultural Uses of Gypsum 2-3
2.4. Conceptual Model 2-6
3. Existing Evaluations 3-1
3.1. Identification of Existing Evaluations 3-1
3.2. Review of Data Quality in Existing Evaluations 3-1
3.3. Application of Findings from Existing Evaluations 3-6
3.4. Review of Available Literature 3-7
4. Comparison with Analogous Product 4-1
4.1. Comparison Approach 4-1
4.2. Comparison for Releases to Soil 4-3
4.3. Comparison of Releases to Water 4-9
4.4. Comparison of Releases to Air 4-13
4.5. Summary of Comparisons 4-14
5. Screening Analysis 5-1
5.1. Data Preparation 5-1
5.2. Screening Results 5-4
5.3. Summary 5-11
6. Risk Modeling 6-1
6.1. Model Inputs 6-1
6.2. Model Design 6-6
6.3. Model Results 6-10
6.4. Summary 6-13
7. Uncertainty and Sensitivity Analyses 7-1
7.1. Uncertainty Analyses 7-1
7.2. Sensitivity Analyses 7-22
7.3. Summary 7-26
8. Final Summary and Conclusions 8-1
8.1. Evaluation Summary 8-1
8.2. Conclusions 8-3
9. References 9-1
Appendix A. Constituent Data
Appendix B. Benchmarks
Appendix C. Use Characterization
Appendix D. Screening Analysis
Appendix E. Probabilistic Modeling
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
List of Tables
Table 3-1. Constituents Retained for Comparison with Analogous Product 3-7
Table 4-1. Comparison of Washed and Unwashed FGD Gypsum Bulk Content 4-4
Table 4-2. Relative Mass Loss from Unwashed Gypsum 4-5
Table 4-3. Comparison of FGD and Mined Gypsum Bulk Content 4-7
Table 4-4. Comparison of Washed/Unwashed FGD Gypsum Median Leachate Concentrations 4-10
Table 4-5. Percent Difference Between Washed/Unwashed FGD Gypsum 90th Percentile Leachate
Concentrations 4-11
Table 4-6. Constituents Retained for Screening Analysis 4-15
Table 5-1. Aluminum and Iron Solubility in Surface Water 5-4
Table 5-2. Preliminary Screening Results for Soil Pathways 5-5
Table 5-3. Final Screening Results for Soil Pathways 5-6
Table 5-4. Preliminary Screening Results for Ground Water Pathways 5-7
Table 5-5. Final Screening Results for Ground Water Pathways 5-8
Table 5-6. Final Screening Results for Surface Water Pathways 5-9
Table 5-7. Final Screening Results for Air Pathways (Mercury Only) 5-10
Table 5-8. Constituents Retained for Risk Modeling 5-11
Table 6-1. Summary of Constituent Data in the Probabilistic Analysis 6-2
Table 6-2. Summary of Exposure and Toxicity Data in the Probabilistic Analysis 6-4
Table 6-3. Summary of FGD Gypsum Use Data in the Probabilistic Analysis 6-5
Table 6-4. Summary of Environmental Data in the Probabilistic Analysis 6-6
Table 6-5. National Risk Results for Soil Pathways 6-11
Table 6-6. National Results for Ground Water Pathways 6-12
Table 6-7. National Risk Results for Surface Water Pathways 6-13
Table 6-8. Constituents Retained for Uncertainty and Sensitivity Analyses 6-14
Table 7-1. Comparison of Data Collected by EPA and from Other Sources 7-3
Table 7-2. Comparison of Model Results for Hexavalent and Trivalent Chromium 7-19
Table 7-3. Comparison of FGD Gypsum and Surface Soil Concentrations 7-20
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
List of Figures
Figure 2-1: Diagram of generic coal combustion processes 2-2
Figure 2-2: Diagram of a generic flue gas desulfurization scrubber 2-3
Figure 2-3: Conceptual model for FGD used in agricultural applications 2-7
Figure 4-1: Categorization of trace elements based on partitioning in flue gas (adapted from
Clarke and Sloss, 1992) 4-8
Figure 6-1: Aerial view of conceptual model for ground water plumes 6-7
Figure 6-2: Map of drainage areas within a sample HUC10 boundary 6-9
Figure 7-1: Locations of coal-fired plants without scrubbers and 2017 SO2 NAAQS exceedances 7-6
Figure 7-2: Percentage of agricultural land irrigated in each use area 7-9
Figure 7-3: Percentage of agricultural land with tile drains in each use area 7-10
Figure 7-4: Comparison of lead leached from washed and unwashed samples 7-12
Figure 7-5: Occurrence of impaired waterways for selenium (top) and mercury (bottom) 7-21
Figure 7-6: Relationship between bulk and leachable content 7-23
Figure 7-7: Geographic variability of modeled risks by individual HUC4 7-25
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
Acronym List
ACAA
American Coal Ash Association
AD
Anderson-Darling
AERMOD
American Meteorological Society/EPA Regulatory Model
ASTSWMO
Association of State and Territorial Solid Waste Management Officials
ATSDR
Agency for Toxic Substances and Disease Registry
BCE
Before the common era
BFI
Base flow index
CCR
Coal combustion residuals
DAF
Dilution-attenuation factor
DBP
Disinfection byproducts
EIA
Energy Information Agency
EPA
Environmental Protection Agency
EPACMTP
EPA's Composite Model for Leachate Migration with Transformation Products
EPRI
Electric Power Research Institute
FGD
Flue gas desulfurization
HELP
Hydrologic Evaluation of Landfill Performance model
HQ.
Hazard quotient
HUC
Hydrologic unit code
IWEM
Industrial Waste Evaluation Model
KS
Kolmogorov-Smirnov
LAU
Land application unit
LEAF
Leaching Environmental Assessment Framework
L/S
Liquid-to-solid ratio
MIDEQ
Michigan Department of Environmental Quality
MINTEQA2
Metal Speciation Equilibrium Model for Surface and Groundwater
ND
Non-detect
NHD
National Hydrography Dataset
OLEM
Office of Land and Emergency Management
ORNL
Oak Ridge National Laboratory
QA/QC
Quality assurance/quality control
SSM
Sustainable Materials Management
SPLC
Synthetic Precipitation Leaching Procedure
TCLP
Toxicity Characteristic Leaching Procedure
USDA
United States Department of Agriculture
USGS
United States Geological Survey
WSR
Wilcoxon signed-rank
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Front Matter
-------
1. Introduction
The United States Environmental Protection Agency ("EPA" or "the Agency") Sustainable
Materials Management (SMM) Program aims to minimize the negative environmental impacts of
materials through the sustainable use/reuse of resources throughout the product lifecycle, from
resource acquisition through ultimate disposal. When done in a responsible manner, the beneficial
use of secondary materials can advance these goals. Beneficial use involves the substitution of these
secondary materials, either as generated or following additional processing, for some or all of the
virgin, raw materials in a natural or commercial product (an "analogous product") in a way that
provides a functional benefit, meets product specifications, and does not pose concern to human
health or the environment.
Coal combustion residuals (CCRs) are the byproducts of coal combustion that are captured from
plant effluent and flue gases prior to discharge to the environment. Once generated, CCRs may be
either disposed of or beneficially used. Flue gas desulfurization (FGD) gypsum is one type of
gypsum that is generated by the pollution control technologies intended to reduce sulfur emissions
from plant stacks. One use that has been proposed for FGD gypsum is as an agricultural amendment
for fields, which would replace the naturally occuring gypsum that would otherwise have to be
mined.
1.1. Background
A survey of state beneficial use programs conducted by the Association of State and Territorial
Solid Waste Management Officials in 2006 found that, although the number of requests for
determinations is increasing, "insufficient information to determine human or ecological impacts
of use rather than disposal" has been a major barrier for states when reviewing proposed beneficial
uses (ASTSWMO, 2007). To help address this barrier, the EPA Office of Land and Emergency
Management (OLEM) developed two documents to provide a framework that can be used to ensure
that evaluations are conducted in a manner that is clear, consistent and comprehensive:
¦ Methodology for Evaluating Beneficial Uses of Industrial Non-Hazardous Secondary Materials
(U.S. EPA, 2016a)
¦ Beneficial Use Compendium: A Collection of Resources and Tools to Support Beneficial Use
Evaluations (U.S. EPA, 2016b)
EPA applied both documents to the evaluation of FGD gypsum in agricultural applications. EPA
partnered with the U.S. Department of Agriculture (USDA) - Agricultural Research Service to
ensure that all the data and assumptions relied upon in this evaluation accurately reflect current
agricultural practices. The remainder of this document details the step-wise evaluation. The scope
of the evaluation was limited to FGD gypsum generated in the United States through forced
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 1: Introduction
-------
oxidation scrubbers following particulate collection. This evaluation does not address products that
contain additional additives that may alter either the composition of or releases from FGD gypsum.
1.2. Document Organization
This beneficial use evaluation is divided into seven main sections and five appendices. The main
text provides a step-wise summary of the analyses performed, model results and conclusions. At
the end of each discrete analysis, a summary of the release pathways and constituents retained for
further evaluation is provided. Appendices provide more detailed discussion of the data and models
that underpin the analyses discussed in the main text. The remainder of this section provides a
brief summary of the contents in each section and appendix.
¦ Section 2 (Planning and Scoping): This section aims to identify the questions that will be
answered by the evaluation and the types of information required to answer them. The
information and conceptual model defined in this section formed the basis for all subsequent
data collection efforts.
¦ Section 3 (Existing Evaluations): This step consists of a literature review to identify any existing
evaluations that are of sufficient quality to rely upon in the beneficial use evaluation. The
purpose of this step is to avoid duplication of effort by building on previous works.
¦ Section 4 (Comparison with Analogous Product): This step consists of a comparison between the
beneficial use and an analogous product made with virgin materials. The objective is to
determine whether the potential for adverse impacts from the beneficial use is comparable to
or lower than from an analogous product.
¦ Section 5 (Screening Analysis): This step characterizes the potential for adverse impacts from
the beneficial use through a comparison with screening benchmarks. The objective is to
identify individual constituents or entire exposure pathways that can be eliminated from
further consideration with a high degree of confidence prior to more intensive modeling.
¦ Section 6 (Risk Modeling): This step consists of a refined, quantitative and qualitative
characterization of the potential for adverse impacts from the beneficial use. The objective is
to reduce remaining uncertainties enough to permit well-substantiated conclusions about the
proposed use
¦ Section 7 (Uncertainty and Sensitivity Analysis): This step of consists of a review of major
uncertainties associated with the model and identification of any sensitive model inputs that
might drive identified risks. The goal is to discuss the key findings from the main analysis while
considering the potential effects of uncertainties to reach a final set of conclusions about the
proposed beneficial use.
¦ Section 8 (Final Characterization): This is the final phase for beneficial use evaluations
conducted using this methodology. The objective is to integrate key findings, assumptions,
limitations and uncertainties identified throughout the evaluation into final conclusions about
the potential impacts to human health and the environment associated with the beneficial use.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 1: Introduction
-------
¦ Appendix A (Constituent Data): This appendix provides a summary of the collection and
management of raw data drawn from the available literature and considered in the beneficial
use evaluation of FGD gypsum in agricultural applications.
¦ Appendix B (Benchmarks): This appendix describes the approach used to identify benchmarks
used in this beneficial use evaluation to estimate the potential for adverse impacts to human
and ecological receptors.
¦ Appendix C (Use Characterization): This appendix provides a summary of the collection and
management of data used to define how FGD gypsum may be used in agricultural applications.
¦ Appendix D (Screening Analysis): This appendix provides a summary of the model inputs used
to conduct the air pathway screening.
¦ Appendix E (Probabilistic Modeling): This appendix provides a summary of the data
management and modeling used to model receptor exposures on a national scale.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 1: Introduction
-------
2. Planning and Scoping
This section defines the scope of this beneficial use evaluation and details the conceptual model
for the different environmental releases and associated exposures that may occur. This information
helps to define the questions that the beneficial use evaluation will aim to answer and ensures that
the objectives of the evaluation are well-defined, realistic, and form a sound basis for subsequent
beneficial use determinations. The following subsections detail the information that forms the
basis for all subsequent data collection and analytical efforts.
2.1. Background
Calcium sulfate dihydrate [CaS04*2H20], more commonly known as gypsum, occurs naturally in
sedimentary rock formations across the globe. The utility of gypsum has been recognized for
centuries, with the oldest known use as a building material dating back as early as 6,000 BCE in
Anatolia and 3,700 BCE in Egypt (U.S. DOI, 2001). Naturally occurring gypsum remains a
commodity of great economic value, with large quantities extracted each year from mines and
quarries (hereafter referred to as "mined gypsum"). In 2014, the United States produced an
estimated 17.1 million tons of mined gypsum from 17 states (U.S. DOI, 2015). Once extracted,
mined gypsum may be further ground into a fine powder and heated at high temperatures to drive
off the majority of chemically bound water. The resulting powder is used in a number of
commercial products, such as cement and wallboard.
In recent decades, industries have explored the potential use of byproduct gypsum, which is
generated as the byproduct of various industrial processes, as a substitute for mined gypsum.
Because these synthetic gypsums are also composed primarily of calcium sulfate, it is sometimes
possible to substitute them for mined gypsum. This is evidenced by the fact that synthetic gypsums
currently account for approximately half of the of gypsum use in the United States (U.S. DOI,
2015). Yet, even though the composition of these byproduct gypsums is predominately calcium
sulfate, there is the potential for higher levels of some trace contaminants introduced by the
industrial processes. These contaminants may be released into the environment when synthetic
gypsum is used in place of mined gypsum. Therefore, further evaluation is warranted to determine
whether the use of synthetic gypsum is an appropriate beneficial use.
2.2. Flue Gas Desulfurization Gypsum
The largest source of synthetic gypsum in the United States is FGD gypsum, a CCR generated at
coal-fired electric utilities by the pollution control technologies intended to reduce sulfur
emissions from plant stacks. These utilities may employ any number of different pollution control
devices to remove sulfur (often referred to as "flue gas desulfurization units" or "scrubbers"). These
devices differ in how they remove sulfur gases, but all generate some form of FGD waste that can
range from a dry powder to a wet sludge. FGD gypsum is a specific subset of the wet sludges.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
, 2-1
Section 2: Planning and Scoping
-------
Figure 2-1 illustrates a generalized layout of a coal-fired plant and the collection points for fly ash
and FGD gypsum.
Flue Gas
Particulate
Collection
Device
Boiler
Stack
*1
Figure 2-1: Diagram of generic coal combustion processes.
The generation of FGD gypsum begins with the removal of as much fly ash from, the flue gas as
practicable, which is accomplished with collection devices, such as electrostatic precipitators, bag
houses or cyclones. The performance of different removal systems varies, but current technologies
have achieved greater than 99% reduction of the initial particulate mass. The remaining flue gas is
sprayed with a wet limestone-based reagent, which reacts with and entrains the sulfur dioxide,
reducing the amount that can escape into the atmosphere. In the presence of excess oxygen, the
chemical reaction between limestone and sulfur dioxide produces gypsum. To ensure that the
majority of the resulting sludge is gypsum, a power plant may pump air into the chamber during
the reaction in a process called "forced oxidation" (EPRI, 2008a). This process is illustrated in
Figure 2-2.
The American Coal Ash Association (ACAA) tracks the quantities of different CCRs generated and
beneficially used through voluntary annual surveys.1 According to these surveys, approximately
23 million tons of FGD gypsum were generated in 2019. Of these, 13 million tons were beneficially
used (ACAA, 2021). EPA previously conducted an evaluation of the largest single use of FGD
gypsum, as a raw material for wallboard, and found it to be an appropriate use (U.S. EPA, 2014a).
1) In 2020, the ACAA survey response rate was equivalent to 55% of the total U.S. coal-fired electric generation capacity.
This estimated response rate is based on a ratio of the generating capacity of the individual plants reporting and the total
coal-fired generation capacity reported by the Energy Information Administration (EIA) in 2020 (available online at:
). Reported beneficial use rates were extrapolated for the entire industry sector using the 2020
survey data, historical ACAA survey data, EIA data, and other miscellaneous data sources.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
jr a 2-2
Section 2: Planning and Scoping
-------
This document details the beneficial use evaluation conducted for the different uses of FGD
gypsum in agriculture. In 2018, approximately 1 million tons of FGD gypsum were used in various
agricultural applications. This represents the fourth largest use of FGD gypsum listed in the survey,
but the single largest use of any CCR in agriculture.
clean gas
gypsum
Figure 2-2: Diagram of a generic flue gas desulfurization scrubber.
After generation, FGD gypsum may be washed to reduce impurities, such as soluble salts, and to
meet market specifications for products such as wallboard (Henkels and Gaynor, 1996). Yet
washing also creates a new waste stream that must be managed appropriately, so there is incentive
to avoid the practice if it is not required. EPA did not identify any existing legal requirements or
industry standards that specify the use of washed FGD gypsum in agriculture. Thus, this beneficial
use evaluation considers use of both washed and unwashed FGD gypsum. Washing is not known
to be a common practice for mined gypsum and so it is not considered in this evaluation.
2.3. Agricultural Applications of Gypsum
The following subsections describe the objectives of agricultural applications of gypsum considered
in this evaluation. These applications were selected based on a review of the available literature
and current recommendations from state extension services. It is also important to note that
inclusion of the subsequently described uses in this beneficial use evaluation does not necessarily
reflect the widespread adoption of these applications at the time of the evaluation. Although the
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 2: Planning and Scoping
-------
described uses have demonstrated the potential to achieve desired benefits, a particular use may
still be uncommon due to high upfront costs or other barriers to application.
2.3.1. Nutrient Amendment
Calcium and sulfur are essential nutrients for plant growth and development. Calcium is
incorporated in plant cell walls, where it acts as a cement between adjacent cells and is involved
in cell elongation of root tip growth. Sulfur is incorporated by plants as a component of amino
acids (e.g., methionine, cysteine) that are essential to the structure of proteins and involved in
many enzyme activities and other functions (Jones, 1982). Insufficient amounts of these elements
can inhibit plant development and decrease crop yields. Different crops will have varying
sensitivities to the level of calcium or sulfur in the soil. Some crops, such as peanuts, tend to be
more sensitive to calcium levels (Bledsoe et al., 1949; Cox et al., 1976; Hallock and Garren, 1968;
Howe et al., 2012), while other crops, such as alfalfa and soybean, tend to be more sensitive to
sulfur levels (Chen et al., 2005). Soil deficiencies are considered distinct from other physiological
conditions (e.g., disease) or environmental conditions (e.g., heat stress) that might prevent plants
from drawing nutrients from the soil. Calcium and sulfur are two of the primary components of
gypsum. Therefore, this material can provide a concentrated source of both elements. While there
are other fertilizers available on the market, the higher solubility of gypsum compared to other
sources makes it an attractive source of nutrients (OSU-E, 2011).
2.3.2. Soluble Phosphorus
Some soils contain phosphorus in excess of that needed by crops as a result of current and historic
application of livestock wastes and, to a lesser degree, chemical fertilizer. High excess phosphorus
in surface soils can result in releases to nearby water bodies, directly through overland runoff or
indirectly through subsurface tile drainage. Because algae in freshwater bodies are commonly
limited by phosphorus concentrations, loading from fields may result in the eutrophication of
downstream waters.
Gypsum has been shown to effectively reduce soluble phosphorus in soils with high phosphorus
(Anderson et al., 1995; Stout et al., 1999; Dao, 1999; Norton, 2008; Torbert et al., 2005; Watts and
Torbert, 2009; Endale et al., 2014; Torbert and Watts, 2014; Watts and Torbert, 2016; King et al.,
2016). It has been suggested that this results from the formation of insoluble complexes between
calcium and phosphate (e.g., hydroxyapatite, fluorapatite) (Lindsay, 1979; Brauer et al., 2005), but
the exact mechanism is not yet fully understood. Under a broad range of manure loading, pH and
redox conditions, gypsum has been demonstrated to reduce water-soluble phosphorus. Reductions
of water-soluble phosphorus at or above 50% have been reported. While the greatest reductions
in dissolved phosphorus occurred immediately after application, gypsum has also been reported to
reduce phosphorus concentrations in succeeding runoff events regardless of timing, suggesting that
the effect is persistent over a growing season. Improvement of water quality by reducing
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 2: Planning and Scoping
-------
phosphorus runoff with gypsum is an established practice for which the National Resources
Conservation Service of the USDA has developed standard conservation practices (USDA, 2015a).
2.3.3. Aluminum Toxicity
When soil pH drops below 5.2, aluminum ions (Al3+) readily dissolve into water present within the
pore spaces of the soil. At lower pH, the dissolved Al3+ concentrations may reach levels that are
phytotoxic to the growing roots of crops. Some plant species have adapted to high dissolved Al3+,
but many commercial crops and most vegetables are sensitive to high Al3+ concentrations.
Gypsum has been shown to reduce the adverse effects of subsoil Al3+ (Reeve and Sumner, 1972;
Pavan et al., 1984; Oates and Caldwell, 1985; Shainberg et al., 1989; Sumner et al., 1986, 1990). It
can reduce or eliminate the negative effects of Al3+ in the subsoil on plant rooting because the
dissolved sulfate (SO4) penetrates to a soil depth where it reacts with Al3+ to form a complex ion
that readily leaches from the soil. At the same time, Ca2+ replaces any exchangeable Al3+ still bound
on the surfaces of clay and organic matter. For greatest effect, application would also involve
mixing liming products with the subsoil layer to adjust the pH and ensure any remaining excess
Al3+ has been removed from solution.
2.3.4. Sodic Soils
Most clay particles have a negative electrical charge on the external surface, which will repel other
clay particles. However, positive cations present near the clay surface can attract (or "flocculate")
clay particles toward one another. The resulting aggregation of particles helps to form the structure
of soil pores that allows infiltration (Horn et al., 1995). Divalent cations (e.g., Ca2+, Mg2+) have the
greatest ability to flocculate clays. While monovalent ions (e.g., K+, Na+) also attract negatively
charged clay particles, they allow a greater number of water molecules to surround each clay
particle. The greater contact with water results in increased suspension and dispersal of smaller
clay particles. Movement of these smaller suspended particles can clog the soil pore spaces and
prevent water from infiltrating through the soil column. Reduced infiltration limits the availability
of water to crop roots (USDA, 1954). Sodic soils are a particular type of dispersed soil that occurs
when the soil contains high levels of exchangeable sodium (Na+) relative to the levels of divalent
cations.
Soil amendments like gypsum that contain divalent cations (e.g., Ca2+) can ameliorate sodic soils
by displacing Na+ from the surface of clay particles (Suarez, 2001). Studies have shown that, when
gypsum is incorporated into topsoil of sodic soils, the added Ca2+ promotes aggregation of clays if
non-saline irrigation water is applied to leach Na+ out of the soil (CSU-E, 2007; Suarez, 2001). This
increases the percolation of rain and irrigation water, as well as the retention of water, which
supports plant growth. Remediation of sodic soils with gypsum is a well-developed practice, and
the USDA has included it in Standard Conservation Practice (USDA, 2010).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 2: Planning and Scoping
-------
2.3.5. Infiltration
The concentration of ions in rainwater is many-fold lower than in soil pore water. In certain soils
that contain higher levels of clay, rainfall can deplete soluble Ca2+ from the topsoil. If the depletion
of Ca2+ from the soil results in an excess of monovalent cations, then the topsoil may disperse in a
similar way as sodic soils. However, poor infiltration is distinct from sodic soils in that the soil has
depleted calcium, rather than excessive sodium levels, and is more likely to occur primarily near
the soil surface. The reduced infiltration limits the availability of water to crop roots (US Salinity
Laboratory, 1954).
Soil amendments like gypsum can provide a fresh supply of Ca2+ to the soil, which supports clay
aggregation and improves water infiltration (Norton, 1995; Norton and Dontsova, 1998; Zhang et
al., 1998). Improved infiltration can be economically important in some soil series, and gypsum
has been marketed in the United States based on this benefit. Utilization of gypsum to improve
water infiltration is a well-developed practice that the USDA has included in the Standard
Conservation Practice (USDA, 2015a). Similar benefits for non-sodic, heavy-textured soils has not
yet been demonstrated in the literature, although some research on this application has been
published in recent years (EPRI, 2006; OSU-E, 2011).
2.4. Conceptual Model
The Supplemental Report to Congress on Remaining Wastes from Fossil Fuel Combustion
Technical Background Document: Beneficial Use of Fossil Fuel Combustion Wastes (U.S. EPA,
1998) identified the following types of releases to the surrounding environment that may occur
from CCR products: 1) generation of dust, 2) emanation to air, 3) leaching to ground and surface
water, and 4) decay of naturally occurring radionuclides. Because this evaluation addresses the
beneficial use of CCRs, these findings are considered applicable to the current evaluation of FGD
gypsum used in agriculture. Therefore, each of these release routes was included in the conceptual
model for further consideration.
EPA developed a conceptual model with available information to help organize and visualize the
different media that may come in contact with the FGD gypsum, the types of releases that may
occur, and the receptors that may be exposed. This conceptual model formed the basis for all
subsequent data collection and modeling efforts. Every use of gypsum considered in this evaluation
involved applications directly on the ground surface or mixed into with surface soils. As a result,
regardless of the specific use, the gypsum will be exposed to the same media, and the routes through
which chemical constituents can be released into the environment will be fundamentally the same.
Therefore, a single conceptual model was used to represent all of the uses. The exposure pathways
that may be present once the FGD gypsum has been applied to the land are depicted in Figure 2-3
and discussed in the following subsections. It is assumed that FGD gypsum will be treated as a
valuable commodity prior to application and managed in a way that will minimize releases during
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 2: Planning and Scoping
-------
transport and storage. Therefore, these stages of the product lifecycle were not included in the
conceptual model.
SOURCE
RELEASE
MECHANISMS
MEDIA
EXPOSURE
ROUTES
RECEPTORS
Soil
Applicator)
Mot ing with
Soil
L
n
^ Infiltration
Ground
Water
Overland
Runoff
Surface
Water
Volatilization
C Direct Contact
Ingestion of J
Biota
Terrestrial Ecological
Receptors
Soil Ingestion
Beef & Mifc
Ingestion
Produce
Ingestion
Sediment
1
Direct Exposure |
Resident Farmer or
Adult'Chiid Resident
Surface
Water
C Direct Contact ^
Ingestion of J
Qinta
c
Benthc Ecological
Receptors
Direct Contact
Aquatic Ecological
Receptors
Fish Ingestion
Adult/Child
Recreational Fisher
Ground Water
Ingestion
Dermal
Contact
Inhalation Of
Vapor
Resident Farmer or
Adult Child Resident
c
Direct Contact
Aquatic Ecological
Receptors
Fish Ingestion
Adult'Chiid
Recreational Fisher
Sediment
C Direct Contact ^
Ingestion of
Biota
Benthic Ecological
Receptors
Surface
Water
1
1
c
Inhalation
Resident Farmer or
Adufe Child Resident
Direct Contact
Aquatic Ecological
Receptors
Fish Ingestion
Adult Child
Recreational Fisher
Sediment
C Direct Contact —¦
Ingestion of
Biota "
Benthic Ecological
Receptors
Figure 2-3: Conceptual model for FGD used in agricultural applications.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 2: Planning and Scoping
2-7
-------
2.4.1. Mixing with Soil
Any chemical constituents present within the gypsum will be mixed with surface soils during
application. Human and ecological receptors may be exposed directly to these constituents through
the incidental ingestion of gypsum present in the fields or tracked into nearby homes. The
constituents may also accumulate in crops grown on the gypsum-amended soil and in livestock
that are fed these crops or that graze around the fields. Human receptors may be exposed to these
constituents through the consumption of the resulting produce, meat and dairy.
2.4.2. Infiltration
Soluble constituents may leach into the water from precipitation and irrigation water that comes
in contact with the gypsum. The fraction of the resulting leachate that is not evaporated or taken
up by plants will percolate through the soil and mix with underlying ground water. Transport of
the dissolved constituents will be driven by the advective flow of ground water until the water is
either extracted from a well for consumption or discharged into a water body. The constituents
that discharge to water bodies will mix with the surface water and either remain dissolved or settle
out to the underlying sediments. Human receptors can be exposed through ingestion of and dermal
contact with water drawn from private wells, as well as inhalation of any constituents that
volatilize from this water. Ecological receptors may be exposed through direct contact and
ingestion of the constituents in surface water and sediment. These constituents may accumulate in
the tissue of fish and other aquatic receptors. Human receptors may then be exposed through the
elevated concentration in biota caught from the water body.
2.4.3. Overland Runoff
Soluble constituents may leach into the water from precipitation and irrigation water that comes
in contact with the gypsum. Some fraction of the runoff will flow overland and carry dissolved
constituents and suspended particulates into downgradient water bodies. The constituents
entering the water body will mix with the surface water and either remain dissolved or settle out
into the underlying sediment. Ecological receptors may be exposed through direct contact and
ingestion of the constituents in surface water and sediment. These constituents may accumulate in
the tissue of fish and other aquatic receptors. Human receptors may then be exposed through the
elevated concentration in biota caught from the water body.
While some of the suspended particulates may settle out onto downgradient soils, the resulting soil
concentrations will always be lower than at the site of application. Because it is possible for
sensitive human and ecological receptors to be present near the farm, these lower downgradient
exposures were not retained for further evaluation. This evaluation also did not consider exposures
from surface water used as a source of potable water. Surface water is assumed to be routed through
a municipal water treatment facility prior to consumption, reducing the levels of any chemical
constituents, suspended solids, pathogens, and other contaminants to below applicable standards
prior to distribution. In addition, this evaluation did not consider exposures to constituents that
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
2-8
Section 2: Planning and Scoping
-------
may occur during swimming or other recreational activities near a water body. For human
receptors, it is assumed that these exposures are infrequent and small in comparison to similar
exposures from bathing with ground water.
2.4.4. Volatilization
Constituents with high vapor pressure may volatilize from gypsum under standard ambient
conditions. Once entrained in the air, prevailing wind currents will drive constituent transport
until the vapors are sequestered by fine particulates or water droplets and deposited into nearby
water bodies. Ecological receptors can be exposed to constituents through direct contact with and
ingestion of surface water and sediment. Human receptors may be exposed to these elevated
concentrations through the consumption of fish and other biota. Human receptors may also inhale
the vapors suspended in the air.
Some of the volatilized mass may also settle out onto nearby soil through a combination of dry and
wet deposition. However, the resulting soil concentrations will inevitably be lower than at the
initial site of application. Therefore, EPA did not explicitly evaluate exposures from deposition to
downgradient soils because higher soil exposures are possible around the site of application.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
2-9
Section 2: Planning and Scoping
-------
3. Existing Evaluations
EPA conducted a search of the publicly available literature to identify any evaluations that had
previously drawn conclusions relevant to the potential for adverse impacts associated with the use
of FGD gypsum in agriculture. The purpose of this review was to avoid duplicating previous
analyses that are sufficient to demonstrate whether individual constituents or entire exposure
pathways pose concern to human health or the environment. The remainder of this section
summarizes the identification and review of these existing evaluations.
3.1. Identification of Existing Evaluations
EPA first reviewed all of the available literature and assembled those sources that appeared to
contain information on the constituent concentrations present in or released from FGD and mined
gypsum. A number of relevant literature sources, in particular grey literature, had already been
obtained through previous EPA or USDA investigations. Thus, EPA began with a review of the
references cited in these studies. After exhausting the references in those and subsequently
collected sources, EPA queried Environmental Sciences and Pollution Management, EBSCO
HOST, PubMed, Science Direct, Web of Science, and JSTOR for the key words "gypsum," "flue gas
desulfurization gypsum," "FGD gypsum," "mined gypsum," "natural gypsum," and "synthetic
gypsum." Although some literature sources used other terms, such as "coal gypsum" or "FGD
products," these sources tended to be older and more ambiguous about whether the analyzed
materials fit the definition of FGD gypsum. Because capturing available information on the
composition and behavior of gypsum was a primary goal of the literature search, search terms
related to the specific beneficial use were not used. The literature search resulted in a total of 121
unique sources, of which 70 were determined to contain potentially relevant information based
on preliminary review of abstracts and tables. Further review of this subset identified two sources
that had conducted analyses with the available data that might be used to draw conclusions about
FGD gypsum:
¦ Roper et al., 2013: Analysis of Naturally-Occurring Radionuclides in Coal Combustion Fly
Ash, Gypsum, and Scrubber Residues.
¦ U.S. EPA, 2014b: Final Human and Ecological Risk Assessment of Coal Combustion Residuals.
3.2. Review of Data Quality in Existing Evaluations
EPA reviewed existing evaluations identified in the literature according to the recommendations
in Summary of General Assessment Factors for Evaluating the Quality of Scientific and Technical
Information (U.S. EPA, 2003a). The focus of this review was to determine whether the quality of
these historical evaluations was sufficient to form a defensible basis for conclusions about FGD
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 3: Existing Evaluations
-------
gypsum in agriculture. The review determined whether the existing evaluations pertain to or can
be used to draw conclusions about FGD gypsum (i.e., applicability and utility), clearly and
sufficiently explain the data and assumptions relied upon (i.e., clarity and completeness), use
analytical methods that are both reasonable and relevant to the intended application of the data
(i.e., soundness), considered potential key sources of variability and uncertainty (i.e. variability
and uncertainty), and had undergone sufficient external review (i.e., evaluation and review). A
summary of this review for both of the existing evaluations identified during the literature review
is provided in the following subsections.
3.2.1. Roper et al. (2013)
Roper et al. (2013) measured the activity of naturally occurring radionuclides in the uranium,
thorium and potassium decay chains from samples of FGD gypsum collected across the United
States. The study found through a direct comparison that both typical and high-end activities in
FGD gypsum were similar to or lower than those reported from extensive sampling of European
mined gypsum. Based on this comparison, the author concluded that levels of naturally occurring
radionuclides in FGD gypsum are lower than those in mined gypsum.
Applicability and Utility:
Roper et al. (2013) explicitly measured activities in FGD gypsum collected from coal-fired utilities
in the United States. As a result, these data are directly applicable to the current beneficial use
evaluation. However, the study compared the FGD gypsum to mined gypsum samples collected
from around Europe (El Afifi et al., 2006; Trevisi et al., 2012). Therefore, EPA reviewed the
literature for supplementary information to determine whether gypsum mined in Europe could
differ in composition from gypsum mined in North America.
EPA identified several studies that measured the activity of mined gypsum and wallboard from the
United States (LRL, 1962; LLL, 1977; Zikovsky and Kennedy, 1992; SFDTET, 2009). Both the
average and high-end values reported in these studies are similar to those measured in European
samples, and so there is minimal concern that the use of European mined gypsum will substantially
skew the comparison results. Therefore, the findings of Roper et al. (2013) are considered fully
applicable to the current beneficial use evaluation.
Clarity and Completeness:
All of the methods and instruments used to assemble the data relied upon in both Roper et al.
(2013) and the supplementary evaluations are well documented. The data relied upon are either
presented in the text or documented through reference to other publicly available literature
sources. Therefore, the existing evaluation is considered clear and complete.
Soundness:
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 3: Existing Evaluations
-------
The methods employed to collect and analyze the samples in Roper et al. (2013) and the
supplementary evaluations conform with standard laboratory methods. Roper et al. (2013) ensured
data quality through use of standards from the National Institute of Standards and Technology to
calibrate equipment. Although some of the instruments used in older, supplementary studies are
not the most current, the principal difference is sensitivity and ability to detect constituents at low
concentrations. Use of less sensitive instruments may result in higher detection limits, but this is
not a major source of uncertainty when the majority of samples have detected values. Therefore,
the data used in Roper et al. (2013) and supplemental evaluations are considered sound.
Variability and Uncertainty:
The 20 FGD gypsum samples analyzed in Roper et al. (2013) are the same samples collected and
analyzed for inorganic constituents in the two EPA characterizations relied upon in the current
beneficial use evaluation (U.S. EPA, 2008, 2009). These samples reflect a range of coal types,
pollution control technologies, and wash status found across the United States. As a result, there is
reasonable confidence that these samples are representative of the variability in FGD gypsum. This
conclusion is further corroborated by an uncertainty analysis discussed in Section 7 (Uncertainty
and Sensitivity Analyses) that found that average and high-end concentrations of inorganic
constituents from this set of samples are consistently either comparable to or higher than those
from other data sources. Therefore, the potential to underestimate concentrations with these data
is considered minimal.
Roper et al. (2013) compared the FGD gypsum data to summary statistics from over 500 mined
gypsum samples collected across Europe. The supplemental sources provided data on an additional
38 samples. Therefore, there is reasonable confidence that the variability of North American mined
gypsum has generally been captured by the available data. Although some of the mined gypsum
data sources are one or more decades old, this is not a major source of uncertainty because
geological background levels are unlikely to shift dramatically over time without significant
anthropogenic disturbance.
Evaluation and Review:
Roper et al. (2013) and all of the supplemental evaluations have been published in peer-reviewed
journals. There are no known conflicts of interest for the authors of Roper et al. (2013) that might
diminish their capacity to provide an impartial, technically sound, and objective analysis. EPA did
not review the backgrounds of each author in supplemental evaluations, as these studies were used
only as sources of raw data. While there may be the potential for bias in sample collection in a
particular study, the large number of samples available from across Europe makes it unlikely that
this would impact the conclusions in Roper et al. (2013). Therefore, these evaluations are
considered to have undergone sufficient evaluation and review.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 3: Existing Evaluations
-------
3.2.2. U.S. EPA (2014b)
U.S. EPA (2014b) evaluated the potential risks to human health and the environment associated
with the range of known disposal practices for different CCRs, including FGD gypsum. As part of
this assessment, EPA conducted a screening analysis on uncontrolled releases of windblown dust
from landfills at the point of highest air concentration. Constituent concentrations in the dust were
estimated using high-end values drawn from all different CCR types (e.g., fly ash, FGD gypsum).
This analysis found that the risks from inhalation of chemical constituents present in the dust are
below levels of concern, but potential risks are possible from inhalation of fine particulate matter
with a diameter less than 2.5 |_im (PM2.5). These small particulates pose greater health risks because
of the potential to be inhaled more deeply into the lungs. Based on this screening-level evaluation,
EPA concluded that uncovered CCR landfills have the potential to result in PM2.5 concentrations
nearly twice the relevant national ambient air quality standards (NAAQS).
Applicability and Utility:
The same inhalation exposure pathway exists for FGD gypsum used in agriculture; however, there
is a high degree of confidence that the releases from agriculture will be less frequent and of lower
magnitude than those from uncovered landfills. High-end constituent concentrations in FGD
gypsum are often lower than those modeled in U.S. EPA (2014b) due to consideration of fly ash in
that evaluation. The potential for generation of dust is also much lower from agricultural practices.
Dust generation by wind or mechanical disturbance will be highest during application, which will
be infrequent and relatively short in duration. During the remainder of the year, releases will be
at a much lower magnitude because the gypsum is not piled high above the ground surface and the
presence of plant cover will further limit wind erosion. Because the releases of inorganic elements
from FGD gypsum used in agriculture will be lower than those from CCR placed in uncovered
landfills, EPA determined that the findings of U.S. EPA (2014b) can be extrapolated to draw
conclusions about the inorganic constituents in FGD gypsum.
U.S. EPA (2014b) identified the potential for adverse impacts associated with particulate matter.
EPA reviewed the literature for supplementary information to determine whether the use of FGD
gypsum in agriculture may result in similar risks. Studies have found that the distribution of
particle sizes in FGD gypsum is dominated by particles with a diameter greater than 10 |_im. Under
the controlled conditions in the power plant, the gypsum particles tend to precipitate fairly
uniformly with respect to particle size and shape. In addition, the smallest particles tend to be
washed out during dewatering (Henkles and Gaynor, 1996; U.S. DOE, 2005). This results in a
distribution skewed towards particle sizes larger than those modeled in U.S. EPA (2014b), with
median and low-end (10th percentile) particle diameters closer to 50 |_im and 20 |_im, respectively.
In contrast, several samples of fly ash had median particle diameters approaching 10 |_im and all
had low-end diameters less than 10 |_im. Because potential releases of fine respirable particulates
from FGD gypsum used in agriculture will be lower than those from CCR placed in uncovered
Beneficial Use Evaluation of FGD Gypsum in Agriculture
3-4
Section 3: Existing Evaluations
-------
landfills, EPA determined that the findings of U.S. EPA (2014b) can be extrapolated to draw
conclusions about the inorganic constituents in FGD gypsum.
Clarity and Completeness:
All of the methods used to assemble and analyze the data relied upon in U.S. EPA (2014b) are well
documented. All data and assumptions used in the screening model are summarized in the text.
Therefore, the information contained in U.S. EPA (2014b) is considered clear and complete for the
intended use.
Soundness:
U.S. EPA (2014b) relied on data assembled from a number of sources to conduct the evaluation.
These samples were collected and analyzed through various methods, all of which are validated
consistent with current standards. The principal difference between the instruments is the relative
sensitivity, which determines the ability to detect constituents at low concentrations. This is not a
major concern, because the evaluation relied on upper percentile concentrations that reflect
detected values. Therefore, all of the data used in this evaluation are considered sound.
The American Meteorological Society/EPA Regulatory Model (AERMOD) was used in U.S. EPA
(2014b) to screen potential exposures from inhalation. AERMOD is a regulatory steady-state plume
model that estimates the amount of atmospheric dispersion and deposition during windblown
transport. This model has undergone validation and been found to provide reasonable estimates of
downwind air concentrations and deposition rates (U.S. EPA, 2003b). Therefore, the use of this
model in the evaluation is considered sound.
Variability and Uncertainty:
The data used in U.S. EPA (2014b) reflect the range of coal types and pollution control technologies
found across the United States. There is a high degree of confidence that these data provide a good
estimate of the variability of these secondary materials. In addition, the bulk content (i.e., mg/kg)
of inorganic constituents and small particulates used in U.S. EPA (2014b) include all types of CCR,
including fly ash. Measured levels of inorganic constituents and small particulates are frequently
lower in FGD gypsum than fly ash. Therefore, EPA concluded that application of the data from
U.S. EPA (2014b) adequately address variability and uncertainty by providing an upper bound on
potential exposures.
Evaluation and Review:
The database used in U.S. EPA (2014b) contains data collected over a series of regulatory activities
between 1998 and 2010. These data were either collected and analyzed by EPA or were provided
by States, public advocacy groups, or regulated facilities. Many of the samples provided from
outside parties were originally collected as part of regulatory compliance activities. Given the large
overall number of samples and the reliance on high-end concentrations to draw conclusions, there
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 3: Existing Evaluations
-------
is minimal concern that the data relied upon is biased in a way that would underestimate
exposures.
The draft of U.S. EPA (2014b) (Draft Human and Ecological Risk Assessment of Coal Combustion
Residuals; U.S. EPA, 2009a) was submitted for both peer review and public comment as part of a
proposed Agency rulemaking. In response to those comments, EPA revised the risk assessment and
replied to the peer and substantive public comments in two separate documents (U.S.
EPA 2014c,d). Because the full extent of the data and analyses have been subject to review and
comment by independent experts and the public, this evaluation is considered to have undergone
sufficient evaluation and review.
3.3. Application of Findings from Existing Evaluations
After the existing evaluations were identified and determined to be of adequate quality, the
findings were considered in light of all supporting information to reach conclusions about FGD
gypsum used in agriculture. If the available information was adequate to demonstrate that the
potential for adverse impacts is comparable to or lower than from an analogous product, or at or
below relevant regulatory and health-based benchmarks, then no further evaluation is warranted
for that constituent or exposure pathway. A summary of the conclusions is provided in the
following subsections.
3.3.1. Roper et al. (2013)
Roper et al. (2013) found that the activity of naturally occurring radionuclides in FGD gypsum is
comparable to or lower than in European mined gypsum. Supplemental information shows that
North American mined gypsum is comparable to European gypsum, so there is little concern that
this comparison overestimates mined gypsum activity. Based on the review of all available
information, EPA concluded that the activity of naturally occurring radionuclides in FGD gypsum
is comparable to or lower than that in mined gypsum. Therefore, direct exposure to radiation from
FGD gypsum applied to the soil was not retained for further evaluation.
3.3.2. U.S. EPA (2014b)
U.S. EPA (2014b) found that windblown dust from uncovered landfills does not pose concern from
inorganic constituents, but might for small particulates. Knowledge of how the gypsum will be
used is sufficient to demonstrate that the magnitude of releases of FGD gypsum will be far lower
when used in agricultural applications. Supplemental information shows that the many inorganic
constituents and small particulates are present at lower levels in FGD gypsum than in the fly ash
modeled in U.S. EPA (2014b). Given the relatively low exceedance of NAAQS identified for
uncovered landfills, the potential reductions in releases associated with use of FGD gypsum in
agriculture will be sufficient to reduce the potential exposures to below levels of concern, provided
that the gypsum is managed as a valuable product. Therefore, EPA concluded that exposures to
Beneficial Use Evaluation of FGD Gypsum in Agriculture
3~6
Section 3: Existing Evaluations
-------
inorganics and small particulates from windblown dust do not pose a concern to human health.
Therefore, these constituents were not retained for further evaluation. However, constituents that
may volatilize (i.e., mercury) were retained for the air pathway.
3.4. Review of Available Literature
At this stage of the beneficial use evaluation, EPA reviewed all of the remaining literature sources
assembled and identified the constituents that could be carried forward for quantitative evaluation
for each exposure pathway based on the availability of constituent data needed to characterize
releases and toxicological data needed to characterize the risks from exposure:
¦ EPA assembled a database of all the available constituent data. This database includes
information on the identity and concentration of the various constituents that may be present
in and released from both FGD and mined gypsums. A more detailed discussion about the
development and management of this database can be found in Appendix A (Constituent Data).
¦ EPA identified toxicity values for human health and ecological receptors according to a
hierarchy of data sources. A more detailed discussion about the selection of these values can be
found in Appendix B (Benchmarks).
A total of 23 unique constituents were identified that had not been addressed by existing
evaluations and that had sufficient constituent data and toxicity values to characterize the risks
from potential exposures. Constituents that may be present in FGD gypsum, but that could not
undergo a quantitative evaluation are discussed further in Section 7 (Uncertainty and Sensitivity
Analyses). The list of constituents that were carried forward for quantitative analysis is presented
in Table 3-1.
Table 3-1. Constituents Retained for Comparison with Analogous Product
Constituent
CASRN
Human Health
Ecological
Soil
Ground
Water
Surface
Water
Air
Soil
Surface
Water
Sediment
Aluminum
7429-90-5
X
X
X
—
—
X
--
Antimony
7440-36-0
X
X
--
--
X
X
X
Arsenic
7440-38-2
X
X
X
--
X
X
X
Barium
7440-39-3
X
X
X
--
X
X
X
Beryllium
7440-41 -7
X
X
X
--
X
X
--
Boron
7440-42-8
X
X
--
--
X
X
--
Cadmium
7440-43-9
X
X
X
--
X
X
X
Chloride
16887-00-6
--
--
--
--
—
X
--
Chromium
7440-47-3
X
X
X
--
X
X
X
Cobalt
7440-48-4
X
X
--
--
X
X
X
Copper
7440-50-8
X
X
--
--
X
X
X
Iron
7439-89-6
X
X
X
--
--
X
--
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 3: Existing Evaluations
-------
Table 3-1. Constituents Retained for Comparison with Analogous Product
Constituent
CASRN
Human Health
Ecological
Soil
Ground
Water
Surface
Water
Air
Soil
Surface
Water
Sediment
Lead
7439-92-1
X
X
—
--
X
X
X
Magnesium
7439-95-4
—
—
—
—
—
X
--
Manganese
7439-96-5
X
X
X
—
X
X
X
Mercury
7439-97-6
X
X
X
X
X
X
X
Molybdenum
7439-98-7
X
X
X
--
X
X
—
Nickel
7440-02-0
X
X
X
—
X
X
X
Selenium
7782-49-2
X
X
X
--
X
X
X
Strontium
7440-24-6
X
X
X
--
X
X
--
Thallium
7440-28-0
X
X
X
—
X
X
—
Vanadium
7440-62-2
X
X
X
—
X
X
X
Zinc
7440-66-6
X
X
X
--
X
X
X
X - Retained for further evaluation
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 3: Existing Evaluations
-------
4. Comparison with Analogous Product
FGD gypsum has been shown to function as a replacement for mined gypsum in some agricultural
applications. However, this secondary material may contain higher concentrations of inorganic
constituents, which accumulate in the limestone spray along with sulfur dioxide gas. To determine
if the beneficial use of this secondary material may result in higher releases of and subsequent
exposures to these constituents, EPA conducted a comparison of FGD and mined gypsums with
the data available. This section details the data relied upon, the approaches used to qualitatively
compare the materials, the other lines of evidence considered, and the Agency's interpretation of
results.
4.1. Comparison Approach
This section describes the primary statistical approach used to compare FGD and mined gypsum.
All of the data used in these comparisons were drawn from the gypsum database discussed in
Appendix A (Constituent Data). When mined gypsum data were not available to conduct a
quantitative comparison, the constituents were automatically retained for further evaluation.
Additional quantitative and qualitative lines of evidence were also considered when available;
these pathway-specific considerations are discussed in the subsections dedicated to that type of
release.
4.1.1. Handling of Non-detect Data
The comparison of some constituents was complicated by a large number of non-detects in one or
both of the datasets. These non-detect values were not always the lowest values reported for a
given constituent due to the variable detection limits found across different studies. To best address
each comparison based on the amount of detected data, EPA binned the constituents for each
release pathway into one of three groups:
¦ Group 1: Where non-detects account for less than 20% of both datasets, there was reasonable
confidence that the number of non-detects would not interfere with the conclusions of the
statistical tests. Selected statistical tests were conducted with non-detects set to the reported
detection limit based on the requirements of the statistical tests.
¦ Group 2: Where non-detects account for 20% to 50% of either dataset, EPA used bootstrapping
to fill data gaps prior to comparison. This involved fitting the detected data to a gamma,
lognormal or Weibull distribution and selecting the distribution with the best agreement based
on log-likelihood statistics. The selected distribution was then randomly sampled 1,000 times
for each non-detect value observed in the dataset at values below each of the corresponding
detection limits, as this was the highest value that may be present. All the sampled values were
then arranged in order from smallest to largest and the median of consecutive sets of 1,000
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
random numbers was calculated. Each non-detect value was then replaced with one of these
median values. The resulting distribution was used in the comparison with all values treated
as detections. Similar methods have previously been described by Frey and Zhao (2004) and
Zhao and Frey (2004).
¦ Group 3: Where non-detects account for greater than 50% of either dataset, a statistical test
was not conducted because it is unlikely that a statistical test would provide a reliable estimate
of comparability. Because bootstrapping and other available methods used to fill the data gaps
rely on the detected data, there is too much uncertainty introduced by the fitted distribution
when non-detects represent a majority of the dataset. Therefore, statistical comparisons were
not conducted for these constituents. Instead, EPA weighed other available lines of evidence.
When these lines of evidence involved a comparison between datasets, EPA set all non-detect
data to half the reported detection limit according to the recommendations in Risk Assessment
Guidance for Superfund (RAGS) Part A (U.S. EPA, 1989) and EPA Region 3 Guidance on
Handling Chemical Concentration Data near the Detection Limit in Risk Assessments (U.S.
EPA, 1991).
4.1.2. Separation of Washed and Unwashed FGD Gypsum
EPA separated the available FGD gypsum data into three sets based on wash status: washed,
unwashed and unknown. The Agency compared washed and unwashed data to determine whether
there was a substantial difference between these two materials that might skew the results of the
evaluation. Because unknown data could not be reliably sorted into one of these two categories, it
was included in this comparison as a separate category. When washed and unwashed samples were
found to be substantially different, the three datasets were kept separate and unknown data were
excluded from the evaluation. When washed and unwashed data were found to be comparable,
EPA combined all three sets of data into a single dataset. When both washed and unwashed
measurements were available for a sample in this single dataset, the data were averaged to avoid
biasing subsequent analyses toward those with multiple measurements.
4.1.3. Statistical Tests
Distributions for environmental data are often positively skewed, with a longer tail in the direction
of higher concentrations (U.S EPA, 2006). The parametric distributions that best describe the
concentrations present in and released from gypsum are likely to differ among constituents.
Nonparametric tests were selected because these tests avoid assumptions about both the parametric
form of distributions and the exact values of non-detect samples (U.S. EPA, 2010). EPA used one
or more of the following statistical tests, depending on the amount of data available:
¦ The Kolmogorov-Smirnov (KS) test compares continuous distributions to test the hypothesis
that the distributions are the same. This test can detect differences anywhere along the range
of the data, although it is most sensitive to differences around the median (Darling, 1957).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
¦ The Anderson-Darling (AD) test is similar to the KS test, but is more sensitive to differences
at the tails of the distributions (Engmann and Cousineau, 2011).
¦ The Wilcoxon Signed-Ranks (WSR) test is analogous to the parametric paired t-test. This test
compares the similarity of distribution medians (Hollander et al., 2013).
When sufficient data were available, EPA used both the KS and AD tests to compare samples
because these tests compare entire distributions, rather than just medians. Both tests were used
because neither was more robust for the purposes of this evaluation. In addition, agreement
between these two tests provides greater certainty in the results obtained. Yet the KS and AD tests
both require a sufficient sample size to compute the percentiles used in these tests. If the sample
size was too small, then too many percentiles would require interpolation and the comparison
would depend too heavily on the interpolation method used. Therefore, when datasets were judged
to be too small to support these tests, EPA used the WSR test instead.
For all tests, the calculated p-value was compared to a confidence level representing the acceptable
likelihood of incorrectly rejecting the null hypothesis (i.e., distributions are the same). When the
p-value is lower than the specified confidence level, the null hypothesis is rejected in favor of the
alternative hypothesis (i.e., distributions are different). For this evaluation, a confidence level of
90% (a = 0.10) was selected because of the two-tailed test. This results in around a 5% chance of
concluding that concentrations in FGD gypsum are either higher or lower than mined gypsum
when the two distributions are actually the same. When p-values fell below this value, EPA
inspected the distributions to determine which had the higher concentrations. Where releases of
mined gypsum were found to be significantly higher than those of FGD gypsum, the constituent
was removed from further consideration. When more than one test was used, the results were
compared for agreement. And, because the potential for error cannot be entirely eliminated, EPA
considered other quantitative and qualitative lines of evidence to corroborate the statistical results.
4.2. Comparison for Releases to Soil
EPA reviewed the data available in the constituent database to determine how best to compare the
exposures that may result from mixing gypsum with surface soil. The exposures pathways from the
conceptual model considered in the comparison were direct contact with and ingestion of soil, as
well as ingestion of produce and animal products raised on the soil. The magnitudes of these
exposures are directly proportional to the bulk content of constituents in the soil column. The
factor driving accumulation of these constituents in the soil is the bulk content in the gypsum.
Thus, EPA concluded that a comparison of bulk content in gypsum would provide a suitable
surrogate for exposures that may result from releases to soil.
4.2.1. Comparison of Washed and Unwashed FGD Gypsum
EPA identified seven sample pairs in the available data that had been collected from the same
facilities both before and after washing. These samples provide the most direct comparison of the
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 4: Comparison with Analogous Product
-------
changes that can result from washing. In addition, these samples represent a range of coal sources
and pollution control technologies (U.S. EPA, 2009). Due to the small sample size, EPA used the
WSR test to compare the bulk content of constituents in washed and unwashed FGD gypsum. The
results of this comparison are presented in Table 4-1, with instances of higher unwashed
concentrations highlighted.
Table 4-1. Comparison of Washed and Unwashed FGD Gypsum Bulk Content
Constituent
Unwashed
Washed
WSR
p-value
Result
Detection
Frequency
Median
Value
(mg/kg)
Detection
Frequency
Median
Value
(mg/kg)
Group 1
Aluminum
7/7
959
7/7
1,836
0.297
Comparable
Antimony
7/7
1.6
7/7
1.9
0.866
Comparable
Arsenic
7/7
3.5
7/7
2.3
0.176
Comparable
Barium
7/7
10.0
7/7
14.0
0.462
Comparable
Cadmium
7/7
0.30
7/7
0.40
0.834
Comparable
Chloride
7/7
1,639
7/7
275
0.016
Unwashed Higher
Chromium
7/7
9.1
7/7
7.8
0.938
Comparable
Cobalt
7/7
2.3
7/7
2.6
0.578
Comparable
Iron
7/7
1,610
7/7
1,583
0.578
Comparable
Lead
7/7
0.90
7/7
1.6
0.681
Comparable
Mercury
7/7
0.54
7/7
0.49
0.295
Comparable
Molybdenum
7/7
3.1
7/7
3.7
0.375
Comparable
Selenium
7/7
4.9
7/7
4.5
0.208
Comparable
Strontium
7/7
289
7/7
281
0.529
Comparable
Thallium
7/7
0.60
7/7
0.60
0.423
Comparable
Group 3
Beryllium
0/0
--
0/0
--
--
--
Boron
0/0
--
0/0
--
--
--
Copper
0/0
--
0/0
--
--
--
Manganese
0/0
--
0/0
--
--
--
Nickel
0/0
--
0/0
--
--
--
Vanadium
0/0
--
0/0
--
--
--
Zinc
0/0
--
0/0
--
--
--
Chloride is the only constituent in this dataset that exhibited both a large and consistent difference
between washed and unwashed samples. These results are supported by the fact that a primary
goal of washing is to reduce the amount of soluble salts, such as chlorides (Gustin and Ladwig,
2010). There is no indication that current washing practices substantially decrease the bulk content
of other constituents. In fact, comparison of individual sample pairs in the constituent database
shows that measured concentrations in washed samples can be higher than those in corresponding
unwashed samples. These increases are not isolated to certain sample pairs or constituents, making
them unlikely to be the result of sampling or analytical error. Instead, this indicates that losses
Beneficial Use Evaluation of FGD Gypsum in Agriculture
4 ~ 4
Section 4: Comparison with Analogous Product
-------
from washing are minor enough to be masked by a combination of sample heterogeneity, matrix
interference, and other sources of measurement uncertainty. Because further comparison of
washed and unwashed samples collected from different sources and analyzed in different
laboratories would only compound this uncertainty, EPA did not conduct a direct comparison with
the full dataset.
As a secondary line of evidence, EPA calculated the constituent mass lost during washing as a
percentage of the bulk content for all available samples of unwashed FGD gypsum. EPA compared
both the 90th and 50th percentile values for both variables to determine if mass lost to leaching
represented a substantial and consistent fraction of the initial mass present. Relative mass loss was
calculated using LEAF Method 1316, which measures leaching (mg/L) as a function of the liquid-
to-solid (L/S) ratio, multiplied by the relevant L/S ratio (L/kg). EPA selected samples at a L/S ratio
of 2:1 based on recommended wash ratios of between 1.5:1 and 2.5:1 to remove high solute
concentrations (Genck et al., 2008). The results of this comparison are presented in Table 4-2, with
instances of high and consistent losses highlighted. Reported values are rounded to the nearest
whole percent. This comparison does not rely on statistical tests and so samples were not divided
into separate groups.
Table 4-2. Relative Mass Loss from Unwashed Gypsum
Constituent
Detection Frequency
Percent Mass Lost
Unwashed Bulk
Content
Method 1316
Mass Loss
90th Percentile
50th Percentile
Aluminum
21 / 21
4/11
0%
0%
Antimony
21 / 21
5/11
0%
1%
Arsenic
27/29
3/11
0%
0%
Barium
21 / 21
11/11
0%
1%
Beryllium
7/10
0/11
5%
14%
Boron
11 / 11
11 / 11
65%
33%
Cadmium
21 / 21
6/11
4%
3%
Chloride
18/18
11 / 11
100%
100%
Chromium
21 / 21
9/11
0%
0%
Cobalt
20/21
6/11
3%
1%
Copper
14/16
10/11
2%
2%
Iron
28/28
4/11
0%
0%
Lead
20/21
6/11
1%
1%
Manganese
10/10
11 / 11
40%
34%
Mercury
35/35
7/11
0%
0%
Molybdenum
21 / 21
11/11
15%
9%
Nickel
10/10
11/11
39%
5%
Selenium
29/29
11/11
11%
11%
Strontium
21 / 21
11/11
2%
1%
Thallium
19/19
9/11
2%
7%
Vanadium
10/10
10/11
2%
1%
Zinc
10/10
11/11
5%
5%
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
The results of this comparison indicate high and consistent mass loss for chloride relative to the
bulk content of these constituents, which agrees well with the chloride results presented in
Table 4-1. Similarly, high and consistent loss was observed for boron and manganese. Based on the
available lines of evidence, EPA concluded that there is a clear difference between the bulk content
of boron, chloride and manganese in washed and unwashed gypsum. For these three constituents,
washed and unwashed data were kept separate in subsequent analyses.
The results of this comparison indicate moderate mass loss for molybdenum, nickel and selenium.
However, these losses were not consistent between the 90th and 50th percentiles. The results also
disagree with the results presented in Table 4-1, which did not identify any differences between
washed and unwashed samples for molybdenum and selenium. A review of the raw data found
that the upper percentile losses for these constituents were driven by a single high measurement,
which may skew results. In addition, the unwashed sample with the highest molybdenum loss had
a corresponding washed sample with a higher measured bulk content. Therefore, it is likely that
differences for these three constituents fall within the bounds of measurement uncertainty and
cannot be reliably identified. Based on these lines of evidence, EPA concluded that it would be
inappropriate to base recommendations for the management of FGD gypsum on these differences.
Thus, all available data were combined into a single set for these constituents.
For the remaining constituents, there is no evidence that current washing practices substantially
reduce bulk content. These results agree well with those presented in Table 4-1. While there are
isolated reports in the literature of larger reductions for some constituents, such as mercury, the
same studies note this behavior is unusual (Gustin and Ladwig, 2010). However, because such losses
cannot be reliably predicted, EPA concluded it would be inappropriate to base recommendations
for the management of FGD gypsum on these differences. Thus, all available data were combined
into a single dataset.
4.2.2. Comparison of Mined and FGD Gypsum
EPA assembled all the available data for mined and FGD gypsum for a direct statistical comparison.
Because of the greater number of samples available, EPA relied on both the KS and AD tests. This
comparison will tend to underpredict the relative constituent mass applied from mined gypsum.
The purity of this material can be as low as 66% and is frequently less than that of FGD gypsum,
which is consistently at or above 95% (Henkels and Gaynor, 1996; OSU-E, 2011). Somewhat higher
mined gypsum application rates would be needed to achieve the same calcium sulfate loading onto
agricultural fields, which will result in higher mass loading of chemical constituents from mined
gypsum than assumed by a direct comparison of the materials. EPA directly compared mined and
FGD gypsum because it was difficult to incorporate variable mass loading in the comparisons and
there is an added degree of confidence when concentrations in mined gypsum are found to be
comparable or higher than FGD gypsum. Results of this comparison are presented in Table 4-3,
Beneficial Use Evaluation of FGD Gypsum in Agriculture
4-6
Section 4: Comparison with Analogous Product
-------
with instances of higher FGD concentrations highlighted. Because some p-values are extremely
small, the reported values are truncated for values < 0.001 for ease of presentation.
Table 4-3. Comparison of FGD and Mined Gypsum Bulk Content
Washed
Status
FGD
Mined
KS
p-value
AD
p-value
Constituent
Detection
Frequency
Median
(mg/kg)
Detection
Frequency
Median
(mg/kg)
Result
Group 1
Aluminum
--
53/53
380
15/15
1,516
0.015
0.032
Mined Higher
Antimony
--
50/54
0.60
15/17
0.44
0.475
0.475
Comparable
Barium
--
55/55
12.0
17/17
12.0
0.644
0.604
Comparable
Boron
Unwashed
11 / 11
51.0
14/14
8.9
0.001
< 0.001
FGD Higher
Washed
20/20
8.6
14/14
8.9
0.818
0.830
Comparable
Cadmium
--
50/55
0.14
13/16
0.05
0.001
0.006
FGD Higher
Chromium
--
57/60
2.9
17/18
1.5
0.042
0.003
FGD Higher
Iron
--
65/65
1,000
18/18
1,133
0.481
0.152
Comparable
Manganese
Unwashed
10/10
27.0
16/16
28.0
0.417
0.688
Comparable
Washed
20/20
7.5
16/16
28.0
0.002
0.005
Mined Higher
Mercury
--
94/96
0.34
17/19
0.002
< 0.001
< 0.001
FGD Higher
Molybdenum
--
52/56
0.95
16/18
0.77
0.143
0.213
Comparable
Nickel
--
41 / 47
1.3
16/18
1.6
0.526
0.379
Comparable
Strontium
--
50/50
161
15/15
1,140
< 0.001
< 0.001
Mined Higher
Vanadium
--
38/41
2.0
16/17
3.0
0.554
0.358
Comparable
Zinc*
--
43/44
7.0
18/18
5.9
0.121
0.043
Mined Higher
Group 2
Arsenic
--
56/69
2.8
11 / 16
1.6
< 0.001
< 0.001
FGD Higher
Beryllium
--
23/41
0.04
8/14
0.01
0.038
0.026
FGD Higher
Cobalt
--
40/55
0.45
16/18
0.62
0.265
0.243
Comparable
Copper
--
38/50
1.5
17/18
1.6
0.532
0.206
Comparable
Lead
--
45/53
1.1
14/18
1.7
0.608
0.398
Comparable
Selenium
--
64/69
5.6
11/15
0.21
< 0.001
< 0.001
FGD Higher
Thallium
--
43/45
0.02
12/15
0.01
0.080
0.010
FGD Higher
Group 3
Chloride
Unwashed
18/18
833
0/0
--
--
--
--
Washed
14/14
219
0/0
--
--
--
--
(ND): The reported value is the same as half the detection limit.
* Though the median FGD gypsum concentration is higher, the distribution for mined gypsum has a tail with the highest concentrations.
This explains the difference between the results of the KS and AD tests.
The results of this comparison indicate that there is a potential for higher concentrations of arsenic,
beryllium, boron (unwashed), cadmium, chromium, mercury, selenium and thallium in FGD
gypsum. Many of these constituents have higher volatility and, as a result, are more likely to pass
through particulate control technologies and become entrained in the limestone slurry. Previous
studies have shown that an appreciable fraction of boron, mercury, selenium and the halogen
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
group (e.g., bromide, chloride, fluoride, iodide) can be released from flue gas stacks as vapor (Cheng
et al., 2009). Other studies have found that a much smaller fraction (around 1%) of arsenic can also
be released as vapor (Meij and Alderliesten, 1989). These findings are further corroborated by the
larger relative differences seen between median FGD and mined gypsum concentrations for the
more volatile constituents (i.e., boron, mercury, selenium) compared to the less volatile
constituents (i.e., arsenic, beryllium, cadmium, chromium, thallium). Figure 4-1 presents a list of
constituents found in coal ranked by volatility, with the constituents that may become
concentrated in FGD gypsum based on this comparison and reports in the literature highlighted in
yellow.
.£
ra
o
>
?
8
®
Figure 4-1: Categorization of trace elements based on partitioning in flue gas (adapted from
Clarke and Sloss, 1992)
The more semi-volatile constituents may be present in the vapor phase initially but will condense
and/or nucleate out of the flue gas as very fine particles (i.e., below 1 jam in diameter) due to large
drops in temperature following combustion. Particulate control devices are the least effective at
removal of these fine particulates, so there is the potential for the entrainment of these particulates
in the limestone slurry to contribute additional constituent mass to the gypsum. However,
substantial enrichment was not observed for all of the semi-volatile constituents, including lead
and antimony. Because metals with similar volatility are expected to partition similarly across air
pollution control systems (U.S. EPA, 1996a), the entrainment of small particulates alone does not
explain differences observed for arsenic, beryllium, cadmium, chromium or thallium.
Studies have shown that the slurry sprayed into the scrubber can account for over 90% of the mass
in FGD gypsum for many semivolatile constituents, such as arsenic, cadmium, lead and zinc
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Section 4: Comparison with Analogous Product
Class III
Volatilized and emitted fully
in the vapor phase - Not
enriched in the fly ash
Class II
Enriched in the fly ash and
depleted in the bottom ash
Class I
Equally distributed between
bottom ash and fly ash
-------
(Gutberlet, 1984; Gutberlet et al., 1985). Because much of the constituent mass is geogenic in
origin, it is reasonable that, even for the constituents found to be statistically different, there is
considerable overlap with distributions for mined gypsum. The highest measured concentrations
of chromium, cadmium and zinc were found in mined gypsum. Thus, it is possible that some of
the observed differences are driven more by natural variations in the parent minerals, rather than
contributions from the combustion of coal.
Based on the results of this comparison, there is a high degree of confidence that the bulk content
of boron, mercury and selenium can be higher in FGD gypsum. Although mined gypsum data were
not available for any of the halogens, there is also a high degree of confidence these constituents
will be higher in FGD gypsum based on the tendency for halogens to concentrate in the vapor
phase. There is uncertainty whether coal combustion results in higher levels of arsenic, beryllium,
cadmium, chromium or thallium in FGD gypsum, but these constituents were retained for further
evaluation out of an abundance of caution.
4.3. Comparison of Releases to Water
EPA reviewed all the data available in the constituent database to determine how best to compare
the exposures that may result from releases through leaching. The exposure routes from the
conceptual model considered in the comparison were ingestion of impacted ground water and fish
caught from nearby surface water bodies. Due to the trace levels of most constituents in leachate,
it is assumed that releases from both FGD and mined gypsums will behave the same once mixed
with environmental media, where ambient conditions will dictate fate and transport. Thus, EPA
believes that a comparison of leachate concentrations provides a suitable surrogate for exposures
that may result from releases to ground or surface water.
4.3.1. Comparison of Washed and Unwashed FGD Gypsum
Based on the comparison discussed in Section 4.2.1 for bulk content, EPA determined that the
seven sample pairs would also provide the best comparison of washed and unwashed leaching
behavior. These leachate samples were analyzed with EPA Method 1313 and provide data over the
entire pH range of interest. The available data were pooled into a single distribution because a
similar amount of data are available for each sample across the pH range. Because of the relatively
small number of source samples (n = 7), EPA used the WSR test for this comparison. The results of
this comparison are presented in Table 4-4, with instances of higher unwashed concentrations
highlighted. Because some p-values are extremely small, the reported values are truncated for
values below 0.001 for ease of presentation.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 4: Comparison with Analogous Product
-------
Table 4-4. Comparison of Washed/Unwashed FGD Gypsum Median Leachate Concentrations
Unwashed
Washed
WSR
p-value
Constituent
Detection
Frequency
Median
(M9/L)
Detection
Frequency
Median
(M9/L)
Result
Group 1
Aluminum
24/24
473
21 / 22
546
0.921
Comparable
Barium
24/24
98
23/23
80
0.015
Unwashed Higher
Boron
24/24
5,296
19/23
264
< 0.001
Unwashed Higher
Manganese
23/23
1,413
21/23
274
0.001
Unwashed Higher
Nickel
24/24
125
23/23
46
0.001
Unwashed Higher
Selenium
24/24
171
23/23
42
0.004
Unwashed Higher
Strontium
24/24
775
23/23
547
0.307
Comparable
Vanadium
20/24
24
19/23
9.0
0.110
Comparable
Zinc
24/24
212
23/23
188
0.282
Comparable
Group 2
Chromium
18/23
10
17/23
15
0.173
Comparable
Copper
13/24
7
14/23
10
0.617
Comparable
Iron*
17/24
343
12/23
44
0.124
Comparable
Molybdenum
16/24
13
13/23
8
0.036
Unwashed Higher
Group 3
Antimony
3/24
5.6
0/23
5.6
--
--
Arsenic
7/24
6.4(ND)
4/24
6.4(ND)
--
--
Beryllium
0/24
6.4(ND)
0/23
6.4(ND)
--
--
Cadmium
13/23
3.0
8/23
1.7
--
--
Chloride
15/17
237,596
5/17
4,130(ND)
--
--
Cobalt
17/24
10.0
7/24
41 (ND)
--
--
Lead
5/24
2.3 (ND)
0/23
2.3 (ND)
--
--
Mercury
7/24
0.0036(ND)
10/23
0.0036
--
--
Thallium
15/24
7.5
0/23
5.1
--
--
(ND): The reported value is the detection limit.
* Though the median unwashed concentration is much higher, the distribution for washed gypsum has a tail with the highest
concentrations.
The results of this comparison indicate that barium, boron, manganese, molybdenum, nickel and
selenium may be released from unwashed gypsum at higher rates than washed gypsum. Statistical
comparisons were not conducted for antimony, arsenic, beryllium, cadmium, chlorine, cobalt,
lead, mercury or thallium because of the high proportion of non-detect samples in these datasets.
However, it is notable that cadmium, chloride, cobalt and thallium all fall into Group 3 because of
a decrease in the detection frequency after washing. This is a strong indication that substantial
differences also exist for these constituents.
As a secondary line of evidence, EPA conducted a comparison of the 90th percentile leachate
concentrations from all samples. These values were chosen because the extreme values are the
most likely to shift as a result of changes in leaching behavior and most likely to reflect detected
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
4-10
Section 4: Comparison with Analogous Product
-------
values for Group 3 constituents. To ensure a direct comparison, the percentiles were calculated
from the raw data with non-detects set to half the detection limit interpolated between 0.2 pH
intervals for a total of 16 comparison points. The percent differences were then averaged across a
pH range of 5 to 8, except where both datasets were entirely non-detects. In these cases, the
resulting 0% difference was excluded to avoid skewing the calculated percentages. If the average
percent difference was greater than 44%, then washed and unwashed samples were judged to be
substantially different and not the result of measurement uncertainty. This cutoff represents the
maximum difference expected to occur between multiple measurements of a single sample based
on the repeatability of EPA Method 1313 observed during inter-laboratory validation (U.S. EPA,
2012a). The results of this comparison are presented in Table 4-5, with values rounded to the
nearest whole percent. Negative values reflect an average concentration increase measured for
washed samples.
Table 4-5. Percent Difference Between Washed/Unwashed FGD Gypsum 90th
Percentile Leachate Concentrations
Constituent
90th Percentile
Detection Frequency
Percent
Difference
Result
Unwashed
Washed
Aluminum
16/16
16/16
- 42%
Comparable
Antimony
9/16
3/16
57%
Unwashed Higher
Arsenic
16/16
16/16
21%
Comparable
Barium
16/16
16/16
13%
Comparable
Beryllium
0/16
0/16
--
--
Boron
16/16
16/16
97%
Unwashed Higher
Cadmium
16/16
10/16
66%
Unwashed Higher
Chloride
16/16
16/16
95%
Unwashed Higher
Chromium
16/16
16/16
- 18%
Comparable
Cobalt
16/16
8/16
81%
Unwashed Higher
Copper
16/16
16/16
- 24%
Comparable
Iron
16/16
16/16
- 24%
Comparable
Lead
16/16
0/16
79%
Unwashed Higher
Manganese
16/16
16/16
92%
Unwashed Higher
Mercury
16/16
13/16
- 2%
Comparable
Molybdenum
16/16
16/16
54%
Unwashed Higher
Nickel
16/16
16/16
57%
Unwashed Higher
Selenium
16/16
16/16
58%
Unwashed Higher
Strontium
16/16
16/16
- 8%
Comparable
Thallium
16/16
0/16
73%
Unwashed Higher
Vanadium
16/16
16/16
1%
Comparable
Zinc
16/16
16/16
47%
Unwashed Higher
* Value not presented because all data were non-detects.
The results of this comparison indicate that antimony, boron, cadmium, chloride, cobalt, lead,
manganese, molybdenum, nickel, selenium, thallium and zinc can be released from unwashed
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
gypsum at higher rates than washed gypsum. These results agree well with those presented in
Table 4-4 and are consistent with the decrease in detection frequency observed for constituents in
Group 3 after washing. Based on these lines of evidence, washed and unwashed leachate data for
these 11 constituents were kept separate in all subsequent analyses.
The results for barium disagreed between the two comparisons. The results presented in Table 4-4
indicate that differences between washed and unwashed barium are statistically significant;
however, the results in Table 4-5 indicate that the magnitude of these differences are well within
the range of measurement variability. Because the observed differences for barium are so small in
magnitude, EPA concluded that any differences that do exist will not substantively change the
results of the evaluation. Thus, EPA combined washed and unwashed data for this constituent.
The results for zinc also disagreed between the two comparisons. The results presented in
Table 4-5 indicate that the magnitude of differences for zinc were somewhat higher than the
typical range of measurement variability; however, the results in Table 4-4 indicate that these
differences are not statistically significant. This indicates that there is high variability in both the
washed and unwashed samples across the pH range, but no consistent shift in leachate
concentration. Because the observed differences for zinc were not significant, EPA combined
washed and unwashed data for this constituent.
For all the remaining constituents, there is no evidence that current washing practices substantially
alter leaching behavior. Although a comparison could not be conducted for beryllium, all the LEAF
data available for this comparison are non-detects. Therefore, EPA concluded that it was most
appropriate to combine all available leachate data into a single dataset for the remaining 12
constituents.
4.3.2. Comparison of Mined and FGD Gypsum
After reviewing the leachate data for FGD and mined gypsum, EPA determined that insufficient
information was available to conduct a reliable comparison of these materials. Method 1313 data
are available for only two mined gypsum samples, resulting in a total of eight data points over the
relevant pH range. Due to the small number of samples, no statements can be made about the
representativeness of these data. EPA considered merging the Method 1313 data with Synthetic
Precipitation Leaching Procedure (SPLP) and Toxicity Characteristic Leaching Procedure (TCLP)
data into a single distribution for this comparison, but ultimately determined that this would
increase the overall uncertainty. Because SPLP and TCLP measure the concentration released at a
single pH, no information is provided about broader leaching behavior. As a result, it is not possible
to determine whether the measured concentrations are limited by solubility or available content.
In addition, the pH values of the available single-pH data tend to be clustered at or above a pH of
7. Combining the leachate data from these different methods may bias the overall distribution and
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 4: Comparison with Analogous Product
-------
lead to erroneous conclusions about whether the two materials are comparable. Because of these
uncertainties, EPA chose to retain all constituents with FGD gypsum data for further evaluation.
4.4. Comparison of Releases to Air
EPA reviewed all the data available in the constituent database to determine how best to compare
the exposures that could result from volatilization of mercury from the gypsum. The exposure
routes for mercury vapor from the conceptual model considered in the comparison were inhalation
of ambient air, contact with surface water and ingestion of fish caught from the water bodies. Due
to the trace levels of mercury in the air, it is assumed that releases from both FGD and mined
gypsums will behave the same once mixed with environmental media where ambient conditions
will dictate fate and transport.
EPA identified multiple studies that measured emission rates from FGD gypsum (Pekney et al.
2009; Gustin and Ladwig, 2010; Cheng et al. 2012; Briggs et al., 2014). However, there is strong
evidence in the literature that mixing gypsum with agricultural soil will alter emission rates.
Mercury present in the soil column is typically associated with organic content, forming complexes
with sulfur-containing functional groups (e.g., thiol and disulfide) (Meili, 1991; Yin et al., 1997;
Xia et al., 1999; Skyllberg et al., 2006; Oswald et al., 2014). It has been shown that mercury applied
to the soil will form these complexes within days or weeks of application (Hintleman et al., 2002;
Harris et al., 2007). Given the high sulfur content of gypsum and organic content of agricultural
soils, it is reasonable to expect that mercury in the applied gypsum will behave similarly to
unamended soils soon after application. This conclusion is corroborated by studies that have shown
that, after controlling for bulk mercury concentration, the emission rates from FGD gypsum mixed
with soil are far closer to those from the unamended soil than those from the original FGD gypsum
(Cheng et al., 2012; Briggs et al., 2014). Studies on the mercury emission rates from natural soils
have found a strong and consistent relationship between the mercury content of soils and the
resulting emission rates (Eckley et al., 2011; 2015; 2016). Thus, EPA determined that comparison
of bulk mercury concentrations in the gypsum would provide a suitable surrogate for exposures
that result from volatilization of mercury to ambient air.
4.4.1. Comparison of Washed and Unwashed FGD Gypsum
EPA determined that comparison of the bulk mercury concentration in washed and unwashed
FGD gypsum was the most appropriate comparison for emission of mercury to ambient air. The
comparison of bulk content discussed in Section 4.2.1 previously demonstrated that mercury
concentrations in washed and unwashed FGD gypsum are comparable. Therefore, EPA did not
conduct further comparisons for these materials and combined available washed and unwashed
emission data into a single dataset.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 4: Comparison with Analogous Product
-------
4.4.2. Comparison of Mined and FGD Gypsum
EPA determined that comparison of the bulk mercury concentration in mined and FGD gypsum
was the most appropriate comparison for emission of mercury to ambient air. The comparison
discussed in Section 4.2.2 previously demonstrated that bulk content of mercury in FGD gypsum
can be higher than in mined gypsum. These results agree with analyses in the Agency's 2014 Coal
Combustion Residual Beneficial Use Evaluation that previously demonstrated the potential for
higher mercury concentrations in wallboard made with FGD gypsum (U.S. EPA, 2014a).
Therefore, EPA did not conduct further comparisons and retained mercury vapor for further
evaluation.
4.5. Summary of Comparisons
EPA first compared washed and unwashed FGD gypsum to determine whether any differences
exist between these materials that could have a substantial impact on the results and conclusions
of this evaluation. When washed and unwashed gypsum were found to be substantially different,
the data for these samples were kept separate for all future analyses. Otherwise, all available data
were combined into a single dataset. For bulk content, the comparisons found that substantial
differences exist for boron, chloride, and manganese. For leachate, the comparisons found that
substantial differences exist for antimony, boron, cadmium, chloride, cobalt, lead, manganese,
molybdenum, nickel, selenium and thallium. For volatile mercury emissions, no differences were
identified.
Differences in leaching behavior of washed and unwashed samples were observed for constituents
that are highly soluble over some or all of the relevant pH range, resulting in the quick depletion
of leachable mass and a sharp drop in leachate concentration. Differences in the bulk content were
observed for some, but not all, of these highly soluble constituents. There are two main reasons for
this discrepancy. First, the solubility of constituents can vary by orders of magnitude across the pH
range. While a given constituent may exhibit highly soluble behavior across part of the pH range,
it might not at the prevailing pH during washing. Second, even when a constituent is highly
soluble over the entire pH range, the constituent mass that is available to leach may be only a small
fraction of the total mass present. The remainder may be so tightly adsorbed or complexed that it
is unable to dissolve under the specified environmental conditions. Although it is likely that there
are some differences for these remaining constituents between washed and unwashed samples,
they are so minor that they cannot be reliably differentiated from noise. Therefore, EPA concluded
that these differences would not provide a sound basis for recommendations about the appropriate
use of FGD gypsum.
After separating washed and unwashed data for the relevant constituents, EPA compared mined
and FGD gypsum to determine which constituents may be present in and released from FGD
gypsum at higher levels. For bulk content, the comparison indicated that levels of arsenic,
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
4-14
Section 4: Comparison with Analogous Product
-------
beryllium, boron (unwashed only), cadmium, chromium, mercury, selenium and thallium may be
higher in FGD gypsum. For leachate, insufficient data for mined gypsum precluded a comparison,
and so all constituents with sufficient data for FGD gypsum were retained. For volatile mercury
emissions, the comparison indicated that releases of mercury can be higher from FGD gypsum.
Because the potential exists for higher releases of these constituents, EPA retained each of them
for a screening analysis.
The comparison of mined and FGD gypsum found many constituents to be present at comparable
levels in these materials. This makes sense, given that the limestone used in FGD gypsum has been
shown to account for a majority of the mass in FGD gypsum for some constituents. Both limestone
and mined gypsum are excavated from the earth with minimal processing that would further
concentrate inorganic constituents. The greatest differences between FGD and mined gypsum
were identified for the constituents most likely to volatilize at the high temperature present during
coal combustion (e.g., mercury, selenium). These are the constituents most likely to pass through
particulate control devices and be captured in the limestone spray. Table 4-6 provides a summary
of the constituents carried forward for each impacted medium.
Table 4-6. Constituents Retained for Screening Analysis
Constituent
CASRN
Human Health
Ecological
Soil
Ground
Water
Fish
Air
Soil
Surface
Water
Sediment
Aluminum
7429-90-5
—
X
X
—
—
X
—
Antimony
7440-36-0
—
X
—
—
—
X
X
Arsenic
7440-38-2
X
X
X
—
X
X
X
Barium
7440-39-3
—
X
X
—
—
X
X
Beryllium
7440-41-7
X
X
X
—
X
X
—
Boron
7440-42-8
X
X
—
—
X
X
—
Cadmium
7440-43-9
X
X
X
—
X
X
X
Chloride
16887-00-6
—
—
—
—
—
X
—
Chromium
7440-47-3
X
X
X
—
X
X
X
Cobalt
7440-48-4
—
X
—
—
—
X
X
Copper
7440-50-8
—
X
—
—
—
X
X
Iron
7439-89-6
—
X
X
—
—
X
—
Lead
7439-92-1
—
X
—
—
—
X
X
Manganese
7439-96-5
—
X
X
—
—
X
X
Mercury
7439-97-6
X
X
X
X
X
X
X
Molybdenum
7439-98-7
—
X
X
—
—
X
—
Nickel
7440-02-0
—
X
X
—
—
X
X
Selenium
7782-49-2
X
X
X
—
X
X
X
Strontium
7440-24-6
—
X
X
—
—
X
—
Thallium
7440-28-0
X
X
X
—
X
X
—
Vanadium
7440-62-2
—
X
X
—
—
X
X
Zinc
7440-66-6
—
X
X
—
—
X
X
x - Retained for further evaluation
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 4: Comparison with Analogous Product
-------
5. Screening Analysis
A screening analysis is a streamlined approach that reduces the complexity of the modeled system
through a combination of high-end data and simplifying assumptions, which ensure that exposure
estimates that may range anywhere from a reasonable upper bound to unrealistically extreme to
ensure that risks are not underestimated. If a potential exposure is found to be below levels of
concern based on this screening, it can be eliminated from further consideration with a high degree
of confidence. The screening for FGD gypsum considered each exposure pathway carried forward
from Section 4 (Comparison with Analogous Product). This section provides a summary of the
approach used to estimate exposures and the results of comparison with relevant benchmarks.
5.1. Data Preparation
All of the data used in this screening analysis were drawn from the gypsum database discussed in
Appendix A (Constituent Data). This subsection details additional steps taken to prepare the
constituent data for the screening to ensure that the calculated exposures reflect anywhere from a
high-end to an upper-bound estimate of what may result from FGD gypsum. Additional pathway-
specific considerations are discussed in subsequent sections dedicated to each medium.
5.1.1. Non-Detect Data
Non-detect measurements in the dataset represent constituent concentrations below the level that
an analytical methodology can differentiate from background noise. These measurements do not
provide definitive evidence that a constituent is or is not present but do indicate that constituents
are not present at concentrations any higher than the detection limit. Thus, eliminating non-
detects outright may unduly bias the remaining, truncated data set toward the higher, detected
values. Non-detect values were replaced with half of the reported detection limit according to the
recommendations in Risk Assessment Guidance for Superfund (RAGS) Part A (U.S. EPA, 1989)
and EPA Region 3 Guidance on Handling Chemical Concentration Data near the Detection Limit
in Risk Assessments (U.S. EPA, 1991).
5.1.2. Available Content
Available content (also commonly referred to as "leachable content" or "soluble content") is the
total constituent mass that can leach from a material over time. The remaining constituent mass
may be tightly bound in poorly soluble mineral phases, such as alumina-silicate. Most laboratory
leachate tests measure the constituent mass that can be released into a fixed amount of water, but
do not provide a direct measurement of the total mass available to be released over time. Instead,
the available content was estimated with Method 1313 data as the highest concentration released
over the entire pH range (in mg/L) multiplied by the L/S ratio of that sample (in L/kg). This is
considered a reasonable estimate because the highly acidic pH will dissolve iron hydroxides and
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 5: Screening Analysis
-------
other complexations that may initially limit the release of constituents, and the high L/S ratio will
ensure that all of the trace constituent mass can be dissolved (U.S. EPA, 2014e). The resulting
available content (in mg/kg) should be less than or equal to the total constituent mass, but for some
highly soluble constituents, the calculated available content may be slightly higher than the
measured bulk content as a result of measurement uncertainty. In these instances, the available
content was set to the measured total constituent mass. Sufficient data were not available to
calculate the available content of beryllium, boron, copper, manganese, nickel, vanadium and zinc.
Therefore, EPA made the assumption that the entire bulk content was leachable, which will tend
to overestimate releases of these constituents.
Available content could not be calculated for samples without measured leachate concentrations
over the full pH range (e.g., Methods 1311 and 1312). Therefore, to make the best use of all
available data, EPA calculated the fraction of the total bulk content that is leachable for every
Method 1313 sample by dividing the available content by the bulk content. For each constituent,
the leachable fractions were assembled into a distribution that was applied to all samples to
estimate the available content.
5.1.3. Available Content-Limited Behavior
When the solubility of a constituent in water is greater than the total mass available to be released
from FGD gypsum, this can result in the rapid release of all the constituent mass present. Because
the total mass that can be released in a given year is limited by the application rate, the dissolved
concentration is strongly dependent on the amount of water present. Laboratory tests typically
specify the ratio of water and solids. The L/S ratio used in a particular test can differ from what
occurs in the field because of the amount of rainfall. Therefore, the measured concentration must
be adjusted to ensure that releases are not underestimated (U.S. EPA, 2014e). For constituents
found to exhibit leaching behavior limited by the available content, the measured leachate
concentrations were adjusted based on Equation 5-1:
r -n c - rc ^(LSM)(r)(n)
(5-1) CF - (CM) (LS )(p)(d)
Where:
Cp = Field leachate concentration [mg/m3]
CM = Measured leachate concentration [mg/m3]
LSm = Measured L/S ratio [Unitless; 10 for Method 1313, 20 for Methods 1311 and 1312]
LS = Saturated soil L/S ratio [Unitless; 0.5]
r = FGD gypsum application rate [kg/m2 ¦ yr]
n = Years of application [yr]
p = Soil density [1,400 kg/m3]
d = Soil mixing depth [0.2 m]
EPA determined that the constituents identified in Section 4 (Comparison with Analogous Product)
with different washed and unwashed leaching behavior are those that exhibit behavior limited by
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 5: Screening Analysis
-------
available content (i.e., antimony, beryllium, boron, cadmium, chloride, cobalt, lead, manganese,
molybdenum, nickel, selenium and thallium). This is because the loss of some fraction of the
available content during washing limits the amount that can be released in subsequent leaching
events. In contrast, leaching that is limited by the solubility of a constituent is anticipated to
continue at approximately the same concentration until the available content is depleted. For this
screening, EPA assumed that these 12 constituents exhibited available content-limited behavior in
every sample over the pH range relevant to agricultural soils (i.e., 5 to 8).
Because this screening combined high-end values for bulk content, leachate concentration and
leachable fraction that had been calculated independently from available data, a low L/S ratio
might result in an unrealistic scenario where the available content of these 12 constituents does
not deplete within a year. Therefore, EPA adjusted the leachate concentrations based on an
assumed 100 years of application. While this will result in higher leachate concentrations than will
actually occur in the field, it will not result in a dramatic overestimation. Any concentration higher
than that needed to deplete available content will result in faster depletion and the same
annualized concentration. It is possible that this adjustment could push dissolved concentrations
above solubility limits; however, past studies have found that similar adjustments provided a
reasonable estimate of field leaching (U.S. EPA, 2014e).
5.1.4. Depletion of Constituents
Due to the relatively low annual application rates identified for FGD gypsum, there is the potential
that even constituents constrained by solubility limits will be depleted from the soil by runoff and
infiltration prior to the next round of application. This can result in periods when no constituent
mass remains to be released. All of the fate and transport models considered for this beneficial use
evaluation require a leachate concentration that is constant throughout the year. Therefore, when
constituents were found to deplete before the next round of application, an annualized leachate
concentration was calculated. EPA first identified the minimum amount of water required to
deplete all of the available constituent mass applied. If the amount of infiltration or runoff was
greater than this minimum amount, the leachate concentration was multiplied by the ratio of the
two values to estimate the average dissolved concentration over the course of the year. If the
amount of infiltration or runoff was less than this minimum amount, then the measured leachate
concentration was used without any additional adjustment.
5.1.5. Solubility Limits
In fresh surface waters, it is unlikely that dissolved constituent concentrations will exceed the
solubility limits for the common solid phases of these elements. It is assumed that concentrations
any higher than these solubility limits will precipitate out of solution as a solid. EPA calculated
solubility limits for aluminum and iron because these are two major constituents found in FGD
gypsum that are the most likely to exceed respective solubility limits. The geochemical speciation
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 5: Screening Analysis
-------
model MINTEQA2 was used to estimate the solubility limits for both constituents (U.S. EPA,
2001). This model assumes typical values for concentrations of dissolved ions and organic matter
in freshwater bodies to identify the dominant solid phase for constituents and the corresponding
solubility limits as a function of water pH. The highest limit for each constituent over the pH range
was used as an upper bound on surface water concentrations in this screening.
The most soluble solid phases for each constituent that limits dissolved concentrations are
aluminum hydroxide [Al(OH)3] and iron carbonate [FeCCh], While iron(II) is likely to oxidize to
the far less soluble iron(III) in oxygenated surface water, EPA considered the possibility of
reducing conditions to ensure that the estimated upper bound did not underestimate potential
water concentrations. Table 5-1 presents the results of the MINTEQA2 modeling, with the upper
bound of solubility limits highlighted. For both constituents, the highest solubility limit was
identified between a pH of 5.5 to 6. These upper bounds were used as limits on the dissolved
concentration in the surface water. Further discussion of this modeling effort can be found in
Appendix D (Screening Analysis).
Table 5-1. Aluminum and Iron Solubility in Surface Water
pH Range
Aluminum Solubility
(M9/L)
Iron Solubility
(M9/L)
5.5 to 6.0
11
No Limit Found
6.0 to 6.5
1
5,100,000
6.5 to 7.0
0.7
250,000
7.0 to 7.5
0.9
25,000
7.5 to 8.0
2
2,900
8.0 to 8.5
7
320
5.2. Screening Results
A single scenario was applied to all exposure pathways associated with each environmental
medium. In this scenario, FGD gypsum is applied to an agricultural field at 3 tons/acre, which
reflects one of the highest annual rates identified from the literature and summarized in Appendix
C (Use Characterization). The field covers the entirety of a 1,728,000-m2 (427-acre) watershed that
drains into an adjacent lake with a volume of 144,317 m3 (117 acre-ft). These dimensions are based
on a real-world watershed included in the Agency's Food Quality Protection Act Index Reservoir
Screening Tool (FIRST), which was determined to be a high risk for surface water contamination.
Applications occur over a 100-year timeframe. It is assumed the FGD gypsum is well distributed
within the top 20 cm (8 in) of the soil column based on standard tilling depths (U.S. EPA, 2005).
Although the same scenario was applied to each pathway, there was no attempt to account for
mass balance between the pathways, to ensure that each pathway result would reflect the high-
end estimate of potential exposures. Assumptions specific to a particular pathway are discussed in
the relevant subsection. The calculated exposure concentrations were compared directly to the
lowest of the relevant benchmarks identified for each receptor in Appendix B (Benchmarks).
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 5: Screening Analysis
-------
5.2.1. Soil Pathways
As a preliminary screen for soil pathways, constituents were applied at the 90th percentile bulk
content and allowed to accumulate in the soil column without losses. Exposures to soil were
assumed to occur within the field boundary. Although the resulting mixture of soil and FGD
gypsum may be transported to downgradient soils by wind or overland runoff, the resulting
downgradient concentrations will inevitably be lower than at the point of application.
To estimate the FGD gypsum that may be transported as a solid from the point of application to
downgradient sediments through both wind and overland runoff, a 100-year soil concentration was
calculated assuming no loss of constituent mass after 100 years of mixing within the top 20 cm of
soil and then divided by a single dilution and attenuation factor (DAF) of 10. This CCR-specific
DAF was previously used in Human and Ecological Risk Assessment of Coal Combustion Residuals
(U.S. EPA, 2014b) and represents unmitigated transport of CCRs from uncovered, above-grade
landfills through wind and overland runoff. A DAF of 10, lower than the smallest value identified
for any constituent, was applied to all constituents for ease of calculation and to ensure that
sediment concentrations were not underestimated.
The calculated soil concentrations were compared to available benchmarks. For human receptors,
these benchmarks addressed ingestion of soil, consumption of crops grown on the field, and
consumption of beef and milk from cattle both fed on crops grown from the field and allowed to
graze. For ecological receptors, these benchmarks addressed ingestion and direct contact with soil
and sediment. Table 5-2 presents the results of this preliminary round of screening.
Table 5-2. Preliminary Screening Results for Soil Pathways
Constituent
Wash
Human
Ecological
Status
Soil
Produce*
Beef
Milk
Soil
Sediment
Arsenic
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Screen Out
Screen Out
Beryllium
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Screen Out
--
Boron
Unwashed
Screen Out
Screen Out
Screen Out
Screen Out
Retain
Screen Out
Cadmium
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Screen Out
Screen Out
Chromium
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Retain
Screen Out
Mercury
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Retain
Screen Out
Selenium
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Retain
Screen Out
Thallium
Combined
Screen Out
Screen Out
Retain
Retain
Screen Out
--
-- No screening benchmark identified for comparison.
* Same results for all individual crop categories.
The results of the preliminary screening identified potential concerns for human and ecological
receptors. Concentrations of thallium were found to be above benchmarks for human receptors.
Concentrations of boron (unwashed), chromium, mercury and selenium were found to be above
benchmarks for ecological receptors. Therefore, these five constituents were retained for a second,
refined round of screening.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 5: Screening Analysis
-------
This final round of screening used the same inputs as the preliminary round, but also accounted
for moderate losses of constituent mass over the 100 years of application. EPA first calculated the
soluble content by multiplying the 90th percentile bulk content by the 50th percentile available
fraction to retain more constituent mass in the soil. Losses from leaching were estimated for each
constituent at the 50th percentile leachate concentration. The infiltration depth was set to 5 cm/yr
(2 in/yr) and the overland runoff depth was set to 17 cm/yr (6.7 in/yr) for a total annual water
depth of 23 cm/yr (8.7 in/yr). The infiltration depth was drawn from the default low-end values
modeled with the Hydrologic Evaluation of Landfill Performance (HELP) model for climate
stations within the evaluation area (U.S. EPA, 1994). The annual overland runoff depth represents
the annual amount of overland runoff from the 1,728,000-m2 (427-acre) watershed required to
maintain a constant water level in the reservoir. Any soluble mass that remained at the end of a
year was summed with the insoluble mass to calculate accumulation. The calculated media
concentrations were compared directly to the same screening benchmarks. The results of the final
round of screening are presented in Table 5-3. For selenium and thallium, which were found to
have comparable washed and unwashed bulk content, but different leaching behavior, the results
are now presented separately.
Table 5-3. Final Screening Results for Soil Pathways.
Constituent
Wash
Human
Ecological
Status
Beef
Milk
Soil
Boron
Unwashed
—
—
Screen Out
Chromium
Combined
--
--
Retain
Mercury
Combined
--
--
Retain
Selenium
Unwashed
--
--
Retain
Washed
--
--
Retain
Thallium
Unwashed
Retain
Retain
--
Washed
Retain
Retain
--
-- Screened out in the preliminary screening
These results identified potential concerns associated with thallium (both washed and unwashed)
for human receptors and with chromium, mercury and selenium (both washed and unwashed) for
ecological receptors. Mercury and thallium in particular often had low available fractions that
result in the majority of constituent mass remaining in the soil, regardless of the magnitude of
leachate concentrations. In contrast, boron was modeled with high available content and
frequently exhibits available content-controlled behavior. Thus, it is reasonable that this
constituent would deplete from the soil column relatively quickly. Based on these results,
chromium, mercury, selenium and thallium were carried forward for further evaluation in Section
6 (Risk Modeling).
5.2.2. Ground Water Pathways
As a preliminary screen for ground water pathways, EPA assumed that receptors were exposed to
leachate as it was released from the fields with no dilution or attenuation. Annualized leachate
Beneficial Use Evaluation of FGD Gypsum in Agriculture
5—0
Section 5: Screening Analysis
-------
concentrations were calculated using the 90th percentiles for leachate concentration, leachable
fraction, and bulk content. An infiltration rate of 5 cm/yr (2 in/yr) was selected as a floor based on
a lower bound on values calculated with the HELP model for each soil type at climate stations
within the evaluation area (U.S. EPA, 1994). If the leachable mass of a constituent was depleted by
release of the 90th percentile leachate concentration into this amount of infiltration, then the
leachate concentration was scaled to reflect the average leachate concentration over the year.
Otherwise the 90th percentile leachate concentration was used directly for comparison. These
leachate concentrations were compared to available benchmarks for ingestion, dermal contact, and
inhalation of vapor. The results of this preliminary screen are presented in Table 5-4.
Table 5-4. Preliminary Screening Results for Ground Water Pathways
Constituent
Wash
Human Receptors
Status
Ingestion
Dermal
Inhalation
Aluminum
Combined
Screen Out
—
—
Antimony
Unwashed
Retain
Screen Out
—
Washed
Screen Out
Screen Out
—
Arsenic
Combined
Retain
Screen Out
—
Barium
Combined
Screen Out
—
—
Beryllium
Combined
Screen Out
Screen Out
—
Boron
Unwashed
Screen Out
—
—
Washed
Screen Out
—
—
Cadmium
Unwashed
Screen Out
Screen Out
—
Washed
Screen Out
Screen Out
—
Chloride
Unwashed
Screen Out
—
—
Washed
Screen Out
—
—
Chromium
Combined
Retain
Screen Out
—
Cobalt
Unwashed
Retain
—
—
Washed
Screen Out
—
—
Copper
Combined
Screen Out
Screen Out
—
Iron
Combined
Screen Out
—
—
Lead
Unwashed
Screen Out
—
—
Washed
Screen Out
—
—
Manganese
Unwashed
Retained
Screen Out
—
Washed
Screen Out
Screen Out
—
Mercury
Combined
Screen Out
Screen Out
Screen Out
Molybdenum
Unwashed
Screen Out
—
—
Washed
Screen Out
—
—
Nickel
Unwashed
Screen Out
Screen Out
—
Washed
Screen Out
Screen Out
—
Selenium
Unwashed
Retain
Screen Out
—
Washed
Retain
Screen Out
—
Strontium
Combined
Screen Out
—
—
Thallium
Unwashed
Retain
Screen Out
—
Washed
Retain
Screen Out
—
Vanadium
Combined
Screen Out
Screen Out
—
Zinc
Combined
Screen Out
Screen Out
--
-- No benchmark or complete exposure pathway identified for comparison.
The results of this first round of screening identified the potential for concern associated with the
concentrations of antimony (unwashed), arsenic, chromium, cobalt (unwashed), manganese
(unwashed), selenium (both washed and unwashed) and thallium (both washed and unwashed) for
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 5: Screening Analysis
-------
human receptors. These seven constituents were carried forward for a second, refined round of
screening that accounted for some of the dilution and attenuation that may occur in the
environment between the point of release (i.e., fields) and the point of exposures (i.e., private
wells). To estimate well concentrations, EPA used the land application module in the Industrial
Waste Evaluation Model (IWEM; U.S. EPA, 2015) with the following inputs:
¦ Soil Type is the type of subsurface materials in the unsaturated zone immediately below the
field and in the saturated zone below the water table. Both soil types were set to "unknown"
for this screen. For the unsaturated zone, this results in a probabilistic sampling of the
different soil types associated with the selected geographic location. For the saturated zone,
this selection provides values representative of the average aquifer characteristics across the
United States.
¦ Infiltration Rate is the amount of precipitation that percolates into the field in a given year.
The annualized constituent mass flux is greatest when the infiltration rate is equal to the rate
required to deplete a given constituent. Thus, different infiltration rates were chosen for each
constituent. However, if the calculated rate was less than the floor of 5 cm/yr, this floor was
used instead. In addition, a ceiling was selected through trial and error to prevent flooding of
the field based on the other inputs used in the model. If the calculated rate was greater than
the ceiling of 29 cm/yr, this value was used instead.
¦ Climate Stations are facilities with instruments that measure local atmospheric conditions.
These stations provide local precipitation data and determine the infiltration rate into soils
surrounding the field. Of the 102 stations available in the HELP model, EPA selected Grand
Junction, Colorado, as the climate station to limit infiltration outside the field boundary and
minimize the amount of dilution that may occur in the water table.
¦ Distance to Receptor is the shortest straight-line distance to the closest drinking water well.
The nearest well was set 75 m (250 ft) from the edge of the field. This distance was selected
based on recommendations for minimum offset distances from large contamination sources
(U.S. EPA, 2002a). This value was selected to reflect a large and continuous source.
IWEM outputs a single concentration for each constituent that represents the 90th percentile from
10,000 model runs. These concentrations were compared to benchmarks for the ingestion of
ground water. The results of the final round of screening are presented in Table 5-5.
Table 5-5. Final Screening Results for Ground Water Pathways
Constituent
Wash Status
Human
Ingestion
Antimony
Unwashed
Retain
Arsenic
Combined
Retain
Chromium
Combined
Retain
Cobalt
Unwashed
Screen Out
Selenium
Unwashed
Screen Out
Washed
Screen Out
Thallium
Unwashed
Retain
Washed
Retain
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 5: Screening Analysis
-------
This final round of screening identified potential concerns for human receptors. Concentrations of
antimony (unwashed), arsenic, chromium and thallium (both washed and unwashed) were found
above benchmarks for human consumption of ground water. Based on these results, these four
constituents were retained for further evaluation in Section 6 (Risk Modeling).
5.2.3. Surface Water Pathways
A single round of screening was conducted for surface water pathways because it would be difficult
to ensure that results would still reflect a reasonable high-end scenario after altering the empirical
conceptual model. In this scenario, EPA assumed that all the water in the reservoir originated from
overland runoff, with a minimum value set to 17 cm/yr (6.7 in/yr) to represent the annual amount
of overland runoff from the watershed needed to maintain a constant water level in the reservoir.
Contributions from ground water were not considered because that would only reduce
concentrations in the water body due to dilution in the subsurface. If the leachable mass of a
constituent was depleted by release of the 90th percentile leachate concentration into this amount
of infiltration, then the leachate concentration was scaled to reflect the average leachate
concentration over the year. Otherwise the 90th percentile leachate concentration was used
directly for comparison. The calculated water concentrations were compared to benchmarks for
the ingestion of fish caught from the water body by human receptors and both direct contact with
and ingestion of surface water and sediment by aquatic ecological receptors. The results of the
screening are presented in Table 5-6.
Table 5-6. Final Screening Results for Surface Water Pathways
Constituent
Wash
Human
Ecological
Status
Fish Ingestion
Surface Water
Sediment
Aluminum
Combined
Screen Out
Screen Out
—
Antimony
Unwashed
Screen Out
Screen Out
Retain
Washed
Screen Out
Screen Out
Retain
Arsenic
Combined
Retain
Screen Out
Screen Out
Barium
Combined
Screen Out
Screen Out
Screen Out
Beryllium
Combined
Screen Out
Screen Out
—
Boron
Unwashed
Screen Out
Screen Out
—
Washed
Screen Out
Screen Out
—
Cadmium
Unwashed
Retain
Retain
Retain
Washed
Screen Out
Screen Out
Screen Out
Chloride
Unwashed
Screen Out
Screen Out
—
Washed
Screen Out
Screen Out
—
Chromium
Combined
Screen Out
Retain
Retain
Cobalt
Unwashed
—
Screen Out
Screen Out
Washed
—
Screen Out
Screen Out
Copper
Combined
—
Screen Out
Screen Out
Iron
Combined
Screen Out
Retain
—
Lead
Unwashed
—
Retain
Retain
Washed
—
Screen Out
Retain
Manganese
Unwashed
Screen Out
Retain
Retain
Washed
Screen Out
Screen Out
Screen Out
Mercury
Combined
Retain
Screen Out
Retain
Molybdenum
Unwashed
Screen Out
Screen Out
—
Washed
Screen Out
Screen Out
--
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 5: Screening Analysis
-------
Table 5-6. Final Screening Results for Surface Water Pathways
Constituent
Wash
Human
Ecological
Status
Fish Ingestion
Surface Water
Sediment
Nickel
Unwashed
Screen Out
Screen Out
Retain
Washed
Screen Out
Screen Out
Retain
Selenium
Unwashed
Retain
Retain
Screen Out
Washed
Retain
Retain
Screen Out
Strontium
Combined
Screen Out
Screen Out
—
Thallium
Unwashed
Retain
Screen Out
—
Washed
Retain
Screen Out
—
Vanadium
Combined
Screen Out
Screen Out
—
Zinc
Combined
Screen Out
Screen Out
Retain
— Benchmark value not available
The surface water screening identified potential concerns to both human and ecological receptors.
Concentrations of arsenic, cadmium (unwashed), mercury, selenium (both washed and unwashed)
and thallium (both washed and unwashed) were above benchmarks for human consumption of
fish. Concentrations of cadmium (unwashed), chromium, iron, lead (unwashed), manganese
(unwashed), mercury and selenium (both washed and unwashed) were above benchmarks for
ecological exposure to surface water. Concentrations of antimony (both washed and unwashed),
cadmium (unwashed), chromium, lead (both washed and unwashed), mercury, nickel (both
washed and unwashed) and zinc were above benchmarks for ecological exposure to sediment.
Therefore, all of these constituents were retained for further evaluation in Section 6 (Risk
Modeling).
5.2.4. Air Pathways
As a screen for mercury in air (other constituents do not volatize under ambient conditions), EPA
assumed that the entire constituent mass applied to the soil each year volatilizes prior to the next
round of application. Calculated with the 90th percentile bulk content, this resulted in a
continuous emission rate of 77 ng/m2-hr. This emission rate was input into AERMOD to estimate
air dispersion and deposition. Further information about the inputs to the model can be found in
Appendix D (Screening Analysis). The maximum ambient air concentration was compared directly
to the benchmark for inhalation. The maximum deposition rate of vapor onto soil was used to
calculate a dissolved mercury concentration based on equations outlined in Human Health Risk
Assessment Protocol for Hazardous Combustion Facilities (U.S. EPA, 2005). The calculated surface
water concentrations were compared to benchmarks for human ingestion of fish and ecological
exposure to both surface water and sediment. Table 5-7 presents the results of this final round of
screening.
Table 5-7. Final Screening Results for Air Pathways (Mercury Only
Constituent
Wash
Human
Ecological
Status
Inhalation
Fish Ingestion
Surface Water
Sediment
Mercury
Combined
Screen Out
Screen Out
Screen Out
Screen Out
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
5-10
Section 5: Screening Analysis
-------
These results indicate that potential exposures from volatilization of mercury fall over an order of
magnitude below levels of concern for all exposure routes. Because no concerns were identified in
this scenario even with these high-end assumptions, EPA did not further refine the emission
estimates. Therefore, this pathway was not carried forward for further evaluation.
5.3. Summary
EPA applied the constituent concentration data assembled in Appendix A (Constituent Data) to
provide a point estimate of exposures that falls somewhere between the high-end and worst-case
of possible exposures to each media. The concentrations modeled in environmental media were
compared directly to benchmarks identified in Appendix B (Benchmarks), which were developed
to protect human and the environment. Where higher concentrations than these benchmarks
were identified, the screening scenario was refined to the extent possible to reflect fate and
transport that will occur in the environment. Constituents still above relevant benchmarks were
retained for further evaluation in Section 6 (Risk Modeling). The screening results for each medium
are summarized in Table 5-8.
Table 5-8. Constituents Retained for Risk Modeling
Constituent
CASRN
Human Health
Ecological
Soil
Ground
Water
Surface
Water
Air
Soil
Surface
Water
Sediment
Aluminum
7429-90-5
—
—
--
--
—
--
--
Antimony
7440-36-0
—
X
--
--
—
--
X
Arsenic
7440-38-2
—
X
X
--
—
--
--
Barium
7440-39-3
—
—
--
--
—
--
--
Beryllium
7440-41-7
—
—
--
--
—
--
--
Boron
7440-42-8
—
—
--
--
—
--
--
Cadmium
7440-43-9
—
—
X
--
—
X
X
Chloride
16887-00-6
—
—
--
--
—
--
--
Chromium
7440-47-3
—
X
--
--
X
X
X
Cobalt
7440-48-4
—
—
--
--
—
--
--
Copper
7440-50-8
—
—
--
--
—
--
--
Iron
7439-89-6
—
—
--
--
—
X
--
Lead
7439-92-1
—
—
--
--
—
X
X
Manganese
7439-96-5
—
—
--
--
—
X
X
Mercury
7439-97-6
—
—
X
--
X
--
X
Molybdenum
7439-98-7
—
—
--
--
—
--
--
Nickel
7440-02-0
—
—
--
--
—
--
X
Selenium
7782-49-2
—
—
X
--
X
X
--
Strontium
7440-24-6
—
—
—
--
—
—
—
Thallium
7440-28-0
X
X
X
—
—
--
--
Vanadium
7440-62-2
—
—
--
--
—
--
--
Zinc
7440-66-6
-
-
—
—
-
—
X
x - Retained for further evaluation.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 5: Screening Analysis
-------
6. Risk Modeling
This step of the beneficial use evaluation consists of a national-scale evaluation designed to capture
the variability in constituent properties, environmental conditions and receptor characteristics
that may impact the fate and transport of constituents released from FGD gypsum during use. The
modeled results are intended to provide a best estimate of the long-term (i.e., chronic) risks that
may result from use of FGD gypsum in agriculture. The full-scale model considered each exposure
pathway and associated constituents carried forward from Section 5 (Screening Analysis). The
remainder of this section describes the handling of the available data, the model design and the
model results.
6.1. Model Inputs
Where data availability permitted, EPA compiled data for each model input into distributions that
could be probabilistically sampled. Multiple distributions were created for some model inputs
based on location to better capture any relevant geographic correlations (e.g., soil type, field size,
precipitation rate). The spatial resolution at which the data were aggregated were selected to best
capture the variability of data, while also minimizing the computational intensity necessary for
the probabilistic model results to converge (i.e., independent runs of the model will return the
equivalent results). Local-scale distributions were compiled at either the 10 or 12-digit hydrologic
unit codes (HUC10, 12).2 Regional-scale distributions were compiled at different scales wider than
a HUC10 (e.g., HUC8, state-wide). National-scale distributions were compiled for the entire
country. Further details about data collection and preparation are discussed in the relevant
appendices discussed.
6.1.1. Constituent Data
A detailed discussion of the efforts to identify, review and assemble the constituent data used in
this beneficial use evaluation is provided in Appendix A (Constituent Data). This subsection details
the data management to define empirical distributions and prepare the data for use in fate and
transport models. A summary of the model inputs is presented in Table 6-1. The majority of the
data available in the literature were blinded, meaning the reported constituent concentrations
could not be linked to a specific utility. As a result, it was not possible for EPA to link constituent
concentrations present in and released from FGD gypsum with specific geographic areas where it
might be applied. Instead, EPA aggregated all of the available data into national-scale distributions
and applied the same distributions to all agricultural fields.
2) HUCs map the areal extent of surface water drainage across the United States with a hierarchical system of nested
hydrologic units at different spatial scales that range from region (HUC2) to sub-watershed (HUC12). The size of
the drainage area is indicated by the number of digits, with larger numbers representing smaller areas.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 6: Risk Modeling
-------
Table 6-1. Summary of Constituent Data in the Probabilistic Ana
ysis
Type
Spatial Resolution
Variability
Appendix
Bulk Content
National : Country-Wide
Distribution
Available Content
National : Country-Wide
Distribution
Appendix A
pH-Dependent Leachate Concentration
National : Country-Wide
Distribution
Non-Detect Data
Non-detect measurements in the dataset represent constituent concentrations below the level that
an analytical methodology can differentiate from background noise and do not provide definitive
evidence that a constituent is or is not present. However, non-detect measurements can provide
useful information because it is known that the constituent is not present at concentrations any
higher than the detection limit. Eliminating non-detect values entirely may unduly bias the
remaining, truncated data set toward the higher, detected values. Instead, non-detect values were
replaced with half of the reported detection limit according to the recommendations in Risk
Assessment Guidance for Superfund (RAGS) Part A (U.S. EPA, 1989) and EPA Region 3 Guidance
on Handling Chemical Concentration Data near the Detection Limit in Risk Assessments (U.S.
EPA, 1991).
Washed/Unwashed Data
When constituent concentrations present in or released from washed and unwashed samples were
found to be comparable (See: Section 4: Comparison with Analogous Product), all the available
data were combined into a single empirical distribution regardless of wash status. If data for both
washed and unwashed versions of the same sample were available, concentrations were averaged
to avoid bias towards a particular sample source. This was not possible for leachate samples when
the final pH of washed and unwashed samples were different. These samples were determined to
not be duplicative because they capture the leaching behavior of the sample at different portions
of the relevant pH range.
Leachate pH
The available leachate data are compiled from tests at a single pH (i.e., TCLP, SPLP) and at multiple
pH (i.e., LEAF Method 1313). While it is possible to interpolate among the data from Method 1313
to estimate leachate concentrations at any given pH within the relevant range, interpolation is not
possible for single pH tests. A consequence of interpolation is that probabilistically sampling
distributions of leachate data based on a specific pH assigned to each model run would heavily bias
against the selection of single pH data. Instead, EPA divided the available leachate data into 6 bins
of 0.5 pH increments. If a value fell on the cusp of two bins, it was placed in both. EPA incorporated
the Method 1313 data into these bins after interpolation at 0.25 pH increments to ensure even
coverage of the pH range while not overwhelming single-point data. This coverage is important
because dramatic shifts in leachate concentration can occur over a small pH range. Although
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 6: Risk Modeling
-------
Method 1313 data in the distribution will be sampled more frequently, this reflects the better
coverage of sample variability within each pH bin. The specific pH assigned to each model run was
then used to select the pH bin sampled.
Data Sampling
EPA designed a framework to sequentially pull data for each model run. This framework is meant
to make the best use of all available data while maintaining real-world connections that already
exist within the dataset. First, the wash status for the particular model run was assigned. Both
washed and unwashed FGD gypsum were modeled with the same frequency. The bulk content of
each constituent in FGD gypsum was then sampled from the relevant empirical distributions. Next,
leachate concentrations and available fraction were sampled. If the FGD gypsum associated with
the bulk content also had data on the available fraction and/or leachate concentration, these data
were assigned by default for every model run associated with that sample. If data for one or both
of these variables were not available for the selected FGD gypsum sample, values were sampled
probabilistically from a distribution of all available data.
Available Content
To determine whether a constituent exhibits leaching behavior limited by the available content
over a specific range of pH values, EPA compared the leachate concentration measured at each
interpolated pH point to the maximum pH concentration measured for that sample. If the two
values fell within 44% (See Section 5: Screening Analysis), then the constituent is labeled as
available content-limited. Otherwise, the constituent was labeled as solubility limited. Available
content-limited leachate concentrations were adjusted based on Equation 5-1 to reflect differences
between the L/S ratios used in laboratory tests and those present in the field.
Because the full-scale model independently sampled values for bulk content, leachate and available
fraction, a low L/S ratio might result in an unrealistic scenario where the available content does
not deplete within the year. Therefore, to ensure that exposures were not underestimated, EPA
adjusted the leachate concentrations based on 100 years of application. While this will result in
higher leachate concentrations than will actually occur in the field, it will not result in a dramatic
overestimation. Any concentration higher than that needed to deplete the leachable content will
result in faster depletion and the same annual average concentration. It is possible that this
adjustment could push the dissolved concentration above solubility limits; however, past studies
have found that similar adjustments provided a reasonable estimate of field leaching (U.S. EPA,
2014e).
Annualized Leachate Concentration
It is assumed that FGD gypsum is applied only once in any given year. Due to the relatively low
annual application rates identified for FGD gypsum, there is a potential that constituents will be
depleted from the soil by runoff and infiltration prior to the next round of application, even for
constituents with leaching behavior limited by solubility. This can result in parts of the year when
Beneficial Use Evaluation of FGD Gypsum in Agriculture
6-3
Section 6: Risk Modeling
-------
no constituent mass remains to be released. The fate and transport models considered for this
beneficial use evaluation require a fixed leachate concentration provided in annual time steps (U.S.
EPA, 2003d,e). Therefore, when constituents were found to deplete before the next application,
an annualized leachate concentration was calculated. To calculate this concentration, EPA first
calculated the amount of rainfall that contributes to the ground and surface water pathways
through infiltration and runoff in a single year. If the amount of water required to deplete a
constituent was less than this amount, the leachate concentration was multiplied by the ratio of
the two values to estimate the fraction of water that would be free of that constituent. If the ratio
was greater than one, then the leachate concentration was used without any additional adjustment.
6.1.2. Exposure Factors
A detailed discussion of the data used to characterize the rate at which receptors are exposed to
environmental media and the resulting likelihood of adverse health effects is provided in Appendix
B (Benchmarks). These data include information about receptor physiology, mobility, dietary
habits, and susceptibility. A summary of the model inputs for exposure and toxicity is presented in
Table 6-2. The available data were often based on national surveys and it was not possible for EPA
to link receptor characteristics to specific geographic areas. Thus, EPA aggregated all of the
available data into national-scale distributions. In instances where a full distribution could not be
developed, constant values intended to capture reasonable high-end exposures were used instead.
This approach makes the best use of available data and ensures that the potential exposures are not
underestimated.
Table 6-2. Summary of Exposure and Toxicity Data in the Probabilistic Analysis
Type
Spatial Resolution
Variability
Appendix
Exposure Averaging Time
National : Country-Wide
Constant
Exposure Frequency
National : Country-Wide
Constant
Fraction of Media Contaminated
National : Country-Wide
Constant
Fraction of Fish Consumed from Tropic Levels
National: Country-Wide
Constant
Bioconcentration and Biotransformation Factors
National: Country-Wide
Constant
Ecological Benchmarks
National: Country-Wide
Constant
Appendix B
Human Toxicity Values
National: Country-Wide
Constant
Cattle Ingestion Rate of Soil and Crops
National: Country-Wide
Constant
Human Ingestion Rate of Fish
National: Country-Wide
Constant
Human Ingestion Rate of Water, Beef, and Milk
National: Country-Wide
Distribution
Body Weight
National: Country-Wide
Distribution
Exposure Duration
National: Country-Wide
Distribution
Beneficial Use Evaluation of FGD Gypsum in Agriculture
6-4
Section 6: Risk Modeling
-------
6.1.3. Extent of FGD Gypsum Use
A detailed discussion of the data used to characterize the rate and extent to which FGD gypsum
may be used for different purposes is provided in Appendix C (Use Characterization). EPA first
defined the maximum geographic area over which FGD gypsum might be applied based on
economic feasibility. A maximum area was defined for each potential use utilizing the available
data on farmer willingness to pay; the relative locations of coal-fired utilities that generate FGD
gypsum and quarries that mine natural gypsum; and the costs associated with the production,
transport, and application of both types of gypsum. EPA then further defined the boundaries of
agricultural fields within the economic feasibility zone where FGD gypsum might provide specific
benefits. This required an initial delineation of agricultural fields that was accomplished with a
combination of satellite imagery and survey data collected from both USDA and EPA. Given that
crop patterns change over time as fields are left fallow and later resewn, data from 2010 to 2015
were used to capture total acreage that may be farmed. EPA used available data on soil conditions
and crop types in those fields to determine which are likely to benefit from application of FGD
gypsum. A summary of the relevant model inputs for gypsum use is provided in Table 6-3.
Table 6-3. Summary of FGD Gypsum
Jse Data in the Probabilistic Analysis
Type
Spatial Resolution
Variability
Appendix
Total Field Area
Local: HUC12
Constant
Distance to Surface Water
Regional: HUC8
Distribution
Distance to Drinking Water Wells
Regional : State-Wide
Distribution
Appendix C
Percent Field Area with Gypsum Application
National : Country-Wide
Distribution
Years of Application
National : Country-Wide
Distribution
6.1.4. Environmental Data
A detailed discussion of the data that characterize the properties of environmental media that
impact constituent fate and transport is provided in Appendix E (Probabilistic Modeling). EPA
used the locations of fields for each potential use to identify data on soil type, hydrogeologic
environment and climate. A summary of relevant model inputs for environmental characteristics
is provided in Table 6-4. Site-based data on soil type and distance to receptors were drawn based
on prevalence within the boundaries of fields to capture local variability. When field data were
not available, regional data were collected by assigning fields to the nearest reference point in
national databases, such as the HELP model climate stations and U.S. Geological Survey (USGS)
hydrological regions. When environmental parameters could not be linked based on location, EPA
sampled from national-scale distributions.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 6: Risk Modeling
-------
Table 6-4. Summary of Environmental Data in the Probabilistic Analysis
Type
Spatial Resolution
Variability
Appendix
Water Body Geometry
Local : Headwater - Mainstem - HUC12
Constant
Annual Average Water Flow
Local : Headwater - Mainstem - HUC12
Constant
Base Flow Index
Local : Headwater - Mainstem - HUC12
Constant
Stream Annual Flow Mixing Volume
Local : Headwater - Mainstem - HUC12
Constant
Climate Center
Local : HUC10
Constant
Soil Composition
Local : HUC10
Distribution
Soil pH
Local : HUC10
Distribution
Appendix E
Hydrogeological Environment
Local : HUC10
Distribution
Total Suspended Solids
Regional : HUC2
Distribution
Bed Sediment Particle Concentration
National : Country-Wide
Constant
Bed Sediment Porosity
National : Country-Wide
Constant
Depth of Upper Benthic Layer
National : Country-Wide
Constant
Saturated and Unsaturated Soil Kd Values
National : Country-Wide
Distribution
Bed and Suspended Sediment Kd Values
National : Country-Wide
Distribution
6.2. Model Design
For each model run, EPA first used the partitioning module of the land application unit (LAU)
model to determine what fraction of annual precipitation infiltrates to ground water or runs off
overland directly to surface water (U.S. EPA, 2003f). The calculated depth of precipitation was
used together with the constituent data (e.g., bulk content, available content) and soil properties
(e.g., pH) to calculate an annualized leachate concentration. The calculated concentrations were
used in both ground and surface water models to conserve mass among pathways. The following
subsections summarize how fate and transport was modeled for different environmental media
that may be impacted by FGD gypsum. A more detailed discussion of how the data were derived
is provided in Appendix E (Probabilistic Modeling).
6.2.1. Soil Pathways
To estimate accumulation in surface soil, EPA performed a simple mass balance on the soil column.
Annual additions were calculated based on the FGD gypsum application rate and bulk content
assigned to each model run. Annual losses were calculated based on the combined rate of runoff
and infiltration, the available fraction and the leachate concentration for each model run. Each
year, the insoluble fraction of the bulk content accumulated in the soil without any losses. The
mass loss from leachate was subtracted from the accumulated soluble mass. Any soluble mass
remaining at the end of the year was added to the insoluble mass to calculate accumulation. In
every model iteration, the soil concentration was recorded over a 200-year time interval starting
at the first year of application. The recorded concentrations were averaged over the subset of the
time interval relevant to the modeled receptor, centered on the year of highest concentration.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 6: Risk Modeling
-------
6.2.2. Ground Water Pathways
The EPA Composite Model for Leachate Migration with Transformation Products (EPACMTP;
U.S. EPA, 2003c,d,e) was used to model fate and transport through the subsurface environment
and estimate concentrations at a specified downgradient point (i.e., private well or water body).
The source area for each model run was defined as the total area of cropland within the relevant
area with FGD gypsum applied. Figure 6-2 depicts an aerial view of the conceptual model.
-—I Surface
Ground water wells were treated as discrete points in the landscape, while surface water bodies
were treated as straight lines with lengths set equal to the longest National Hydrography Dataset
Plus (NHDPlus) flowline located in the modeled catchment or HUC12. The downgradient distance
to nearest receptor from the field boundary was drawn from empirical distributions aggregated at
the state level for ground water wells and at the HUC8 level for surface water bodies to minimize
computational intensity while still capturing spatial variability. Once a distance was selected, a
random number generator was used to offset the well or water body centroid randomly within the
plume width, as estimated before each model run based on predicted dispersion coefficients. Any
model runs that placed the well or water body entirely outside the modeled plume boundary were
omitted. This ensured that the model results effectively captured highly exposed receptors.
For ground water, the well was assumed to be screened at a discrete point beneath the water table.
In each model run, this point was allowed to vary to either a depth of 10 m below the water table
or to the bottom of the aquifer, whichever was shallower. For surface water, the concentration
along the width of the plume that intersected with the water body was used to calculate a mass
flux from ground water to surface water. Because the focus of this evaluation is the risks related to
application of FGD gypsum, the water discharging to the remainder of the water body length was
assumed to not contribute additional constituent mass. For headwaters, the centroid of the water
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 6: Risk Modeling
-------
body was allowed to vary within the plume in each model run. For higher order streams, the water
body intersected the entire plume width.
In each model iteration, the ground water concentration at the point of exposure was averaged
over the selected exposure interval. For drinking water wells, this time interval was recorded
around the peak concentration. For ground water discharge to surface water, concentrations were
recorded at both the final year of FGD gypsum application (maximum of 100 years) and the peak
ground water concentration at any point in the future. These two timeframes were chosen to
reflect near-term exposures, which are assumed to occur concurrently with overland runoff, and
far-term exposures, which may take several hundreds or thousands of years to occur. In cases
where the ground water concentration was found to still be increasing after 10,000 years,
EPACMTP stopped modeling and reported the ground water concentration at that time as the
peak. This value was held constant over the relevant time interval.
6.2.3. Surface Water Pathways
EPA grouped water bodies into two sets for this evaluation based on the relationship of Strahler
stream order and hydrologic unit code (HUC).3 The first set consists of first- and second-order
streams that are almost entirely contained within the boundaries of individual HUC12, referred to
in this evaluation as "headwater streams." EPA used the concentrations modeled at each headwater
outfall to estimate ecological exposure to surface water and sediment. The majority of land initially
drains to these streams and so they collectively provide extensive habitat for wildlife. Although
ecological receptors will also be present in higher-order streams, EPA focused on these smaller
streams in part to manage computational intensity. The second set consisted of streams at or above
third order that flow across HUC 12 boundaries, referred to in this evaluation as "mainstem
streams." EPA first modeled the cumulative mass loading to each HUC 12 outfall until the stream
order reached sixth order or greater. EPA then calculated the resulting constituent concentrations
at each HUC10 outfall, which were used to estimate human exposure to fish. Although fish can be
present in smaller streams, these water bodies are unlikely to support a population that could
sustain fishing rates that correspond with the high-end ingestion rates that were considered in this
evaluation.
The drainage area upgradient of each outfall is the total land area that contributes runoff through
that discrete point in the landscape. For headwater streams, the drainage area was defined as the
sum of all NHDPlus catchments between the point of origin and the outfall to a higher-order water
bodies. For mainstem streams, the drainage area was defined as the total land area within the
HUC12 boundary. The fraction of each drainage area covered in cropland was recorded and held
3) Strahler stream order is used to define stream size based on a hierarchy of the tributaries. Initial streams without
any upstream tributaries are first-order. Each time two streams of the same order intersect, the number increases.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
6-8
Section 6: Risk Modeling
-------
constant.4 This value was multiplied by the fraction of cropland with gypsum applied. Little data
are available to estimate and so this fraction was allowed to vary in each run based on a flat
distribution that ranged from none to all of the cropland. The concentration in runoff was
calculated by multiplying the leachate concentration from the field with the fraction of the
drainage area with FGD gypsum applied. This approach accounts for the mixing of runoff from the
entire drainage area once it enters the water body. Figure 6-3 presents an example of a HUC10 and
all of the outfalls located within it.
Legend
HUC 10 Boundary
• HUCioOutfal
| | Headwater Drainage Area
e Headwater Outfall
| Field Boundaries
I Surface Water Body
Figure 6-2: Map of drainage areas within a sample HUC10 boundary.
The contributions from overland runoff and ground water discharge to any outfall were related
through the base flow index (BFI). This value reported in the NHDPlus dataset represents the
cumulative fraction of water flow at a given point that originates from base flow (i.e., ground water
discharge) compared to other sources (i.e., runoff). The modeled concentrations in ground water
and overland runoff were weighted based on the BFI to approximate the contributions from each
and obtain a concentration at each headwater or HUC12 outfall.
Larger water bodies flow through multiple HUC12s. Therefore, the flow rate at any given point
may include drainage from one or more upgradient HUC 12s. Because there can be a great deal of
4) As described in Appendix C (Use Characterization), the total cropland area represents the cumulative land area
used to grow crops between the years 2010 and 2015. This total area will not be in active use in any given year, as
some fraction will inevitably be left fallow. However, it represents a best estimate of the total area over which FGD
gypsum may be applied over time.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
S-S
Section 6: Risk Modeling
-------
variability among different HUC12s, it is not realistic to assign the entire concentration for these
water bodies based on the characteristics of a single HUC12. Therefore, EPA mapped out the flow
path of each stream through successive HUC12 outfalls and subtracted the annual average flow
rate reported for each HUC12 outfall from those immediately downstream to obtain contributions
from individual HUC12 to the overall flow. In each model run, the inputs for every HUC12 in the
flow path were sampled independently from relevant distributions to calculate the constituent
mass load to the local stream segment. The mass loading contributed by each successive HUC12
was summed along the flow path. At each HUC10 outfall, the cumulative mass loading and the
annual flow rate were used to calculate the surface water concentration. Concentrations were
calculated for each successive HUC10 until the stream either terminated (e.g., outfall to ocean) or
intersected with a stream of sixth-order or more. This limit was selected both to limit the
computational intensity from extremely long travel distances and the fact that the large drainage
areas for these higher-order streams makes BFI measurements less reliable.
In each model iteration for both headwater and mainstem streams, the surface water concentration
at the relevant outfall was recorded at the year of peak concentration. Concentrations were
recorded at both the final year of FGD gypsum application (maximum of 100 years) and the peak
ground water concentration at any point in the future. These two timeframes were chosen to
reflect near-term exposures, which are assumed to occur concurrently with overland runoff, and
far-term exposures, which may take several hundreds or thousands of years to occur. In cases
where the ground water concentration was found to still be increasing after 10,000 years,
EPACMTP stopped modeling and reported the ground water concentration at that time as the
peak.
6.3. Model Results
The concentrations modeled in this beneficial use evaluation are intended to account for potential
sources of variability associated with FGD gypsum, environmental media, and exposed receptors.
In total, the fate and transport models were run one hundred times within every HUC10, resulting
in up to two million individual model runs across the country for a single use. This subsection
summarizes the model results for use to limit phosphorus runoff. This use was selected because it
results in the highest annual mass loading to the environment. Because the model is not site-
specific, the combination of model inputs inevitably results in some combinations that are outside
of what will realistically occur in the field. As a result, chronic exposures above levels of concern
indicate that further evaluation is required to determine if risks are driven by a particular subset
of modeled scenarios. Therefore, constituents found above levels of concern were carried forward
to Section 7 (Uncertainty and Sensitivity Analyses).
The modeled concentrations in soil, ground water and surface water were used together with the
long-term exposure and toxicity data discussed in Appendix B (Benchmarks) to calculate the
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
6-10
Section 6: Risk Modeling
-------
likelihood that adverse health effects will occur. These effects can be divided into two broad types:
carcinogenic and noncarcinogenic. Carcinogenic effects are those that ultimately result in the
occurrence of cancer somewhere in the individual. The likelihood of carcinogenic effects is
expressed as the increased lifetime probability of cancer that results from an incremental change
in exposure. A risk of 1x10 5 was selected as the point at which further evaluation was warranted
for human receptors. Noncarcinogenic effects are those that result in adverse health effects other
than cancer. The likelihood of noncarcinogenic effects is expressed as a ratio of the exposure level
and the level below which no adverse effects are known or anticipated to occur. This ratio is known
as a hazard quotient (HQ). An HQ > 1 was selected as the point at which further evaluation was
warranted for human and ecological receptors.
6.3.1. Soil Pathways
The full-scale model results for soil pathways are presented in Table 6-5. The reported results
reflect the most sensitive receptors for noncarcinogens (i.e., children 1 to 5 years) and the most
mobile and/or toxic valence states. None of the constituents carried forward to this stage of the
evaluation for this exposure pathway had an identified carcinogenic endpoint. Values that exceed
the risk criteria (i.e., HQ_> 1) are shown in bold.
Table 6-5. National Risk Results for Soil Pathways
Human
Ecological
Constituent
Wash Status
Beef Ingestion
Milk Ingestion
Soil
90th
50th
90th
50th
90th
50th
Noncancer Hazard Quotient
Chromium
Combined
—
—
—
—
1.3
0.19
Mercury
Combined
—
—
—
—
0.65
0.13
Selenium
Unwashed
—
—
—
—
0.88
0.03
Washed
—
—
—
—
0.86
0.08
Thallium
Unwashed
1.4
0.03
1.3
0.03
—
—
Washed
1.7
0.04
1.5
0.04
—
— Screened out in a previous step.
The model results show potential concerns associated with thallium (both washed and unwashed)
for human receptors and chromium for ecological receptors. It is notable that slightly higher risks
for thallium and selenium were sometimes identified for washed samples compared to unwashed
samples. This occurred because it was not possible to differentiate between the bulk content of
washed and unwashed FGD gypsum as a result of measurement uncertainty. However, differences
were identified for washed and unwashed leachate of these two constituents. The use of a single
distribution for bulk content and separate distributions for leachate resulted in slightly higher
estimates of accumulation in the soil due to reduced leaching from washed samples. Because the
differences identified between washed and unwashed soil results are small and primarily an artifact
of data limitations, the two values can be considered effectively the same. Based on these results,
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 6: Risk Modeling
-------
EPA carried chromium and thallium forward for further evaluation in Section 7 (Uncertainty and
Sensitivity Analyses).
6.3.2. Ground Water Pathways
The full-scale model results for ground water pathways are presented in Table 6-6. The reported
results reflect the most sensitive receptors for carcinogens (i.e., adults) and for noncarcinogens (i.e.,
children 1 to 5 years) and the most mobile and/or toxic valence states. Values that exceed the risk
criteria (i.e., risk > lxlO"5 or HQ_> 1) are shown in bold. Because some modeled values are extremely
small, reported values are truncated below an HQ_< 0.01 and risks < lxlO"7 for ease of presentation.
Table 6-6. National Results for Ground Water Pathways
Constituent
Wash Status
Human
Drinking Water Ingestion
90th 50th
Cancer Risk
Arsenic
Combined
1.1x10"6
< 1.0x10~7
Chromium
Combined
< i.ox-icr7
< 1.0x10~7
Noncancer Hazard Quotient
Antimony
Unwashed
< 0.01
< 0.01
Arsenic
Combined
0.03
< 0.01
Chromium
Combined
< 0.01
< 0.01
Thallium
Unwashed
0.05
< 0.01
Washed
These results indicate that all risks from potential exposures to ground water fall well below levels
of concern. Because no concerns were identified for this pathway, even at high-end exposures,
EPA did not retain any constituents for further evaluation. Given that the use with the highest
mass loading to ground water did not pose concern, EPA did not model the remaining uses.
6.3.3. Surface Water Pathways
The results of the full-scale modeling for surface water pathways are presented in Table 6-7. The
reported results reflect the most sensitive receptors for carcinogens (i.e., adult recreational fishers
> 21 years) and for noncarcinogens (i.e., children 1 to 5 years). The results also reflect the most
mobile and/or toxic valence states for each constituent. Values that exceed the risk criteria (i.e.,
risk > lxlO"5 or HQ_> 1) are shown in bold. Reported values reflect exposures from combined runoff
and ground water discharge. All risks from peak ground water discharge only fall below levels of
concern. Because some modeled values are extremely small, reported values are truncated below
an HQ_< 0.01 and risks < lxlO"7 for ease of presentation.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
6-12
Section 6: Risk Modeling
-------
Table 6-7. National Risk Results for Surface Water Pathways
Human
Ecological
Constituent
Wash Status
Fish Ingestion
Surface Water
Sediment
90th
50th
90th
50th
90th
50th
Cancer Risk
Arsenic
Combined
i.ox-icr7
—
—
—
—
Noncancer Hazard Quotient
Antimony
Unwashed
—
—
—
—
< 0.01
< 0.01
Washed
—
—
—
—
< 0.01
< 0.01
Arsenic
Combined
< 0.01
< 0.01
—
—
—
Cadmium
Unwashed
< 0.01
< 0.01
0.03
< 0.01
0.1
< 0.01
Chromium
Combined
—
—
0.07
< 0.01
< 0.01
< 0.01
Iron
Combined
—
—
0.02
< 0.01
—
Lead
Unwashed
—
—
0.02
< 0.01
0.01
< 0.01
Washed
—
—
—
—
0.01
< 0.01
Manganese
Unwashed
—
—
0.2
0.01
0.03
< 0.01
Mercury
Combined
0.8
0.03
—
—
0.05
< 0.01
Nickel
Unwashed
—
—
—
—
0.01
< 0.01
Washed
—
—
—
—
0.01
< 0.01
Selenium
Unwashed
1.1
0.04
1.3
0.09
—
—
Washed
1.0
0.04
1.3
0.09
—
—
Thallium
Unwashed
0.2
< 0.01
—
—
—
—
Washed
0.04
< 0.01
—
—
—
—
Zinc
Combined
0.2
— Previously screened out.
These results identified potential concerns to both human and ecological receptors. Concentrations
of selenium (both washed and unwashed) were found to be at or above benchmarks for human
consumption of fish and for ecological exposure to surface water. Therefore, EPA carried selenium
forward for further evaluation in Section 7 (Uncertainty and Sensitivity Analyses).
6.4. Summary
EPA refined the screening analysis discussed in Section 5 (Screening Analysis) to incorporate
sources of variability and provide a best estimate of exposures that could result from use of FGD
gypsum in agriculture at a national scale. The concentrations modeled in each environmental
medium were used to probabilistically calculate risks to human and ecological receptors. Where
risks were identified above levels of concern, constituents were retained for further evaluation in
Section 7 (Uncertainty and Sensitivity Analyses). The results of the full-scale model are
summarized in Table 6-8.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 6: Risk Modeling
-------
Table 6-8. Constituents Retained for Uncertainty and Sensitivity Analyses
Constituent
CASRN
Human Health
Ecological
Soil
Ground
Water
Fish
Ingestion
Soil
Surface
Water
Sediment
Antimony
7440-36-0
—
—
—
—
—
—
Arsenic
7440-38-2
—
—
—
—
—
—
Cadmium
7440-43-9
—
—
—
—
—
—
Chromium
7440-47-3
—
—
—
X
—
—
Iron
7439-89-6
—
—
—
—
—
—
Lead
7439-92-1
—
—
—
—
—
—
Manganese
7439-96-5
—
—
—
—
—
—
Mercury
7439-97-6
—
—
—
—
—
—
Nickel
7440-02-0
—
—
—
—
—
—
Selenium
7782-49-2
—
—
X
—
X
—
Thallium
7440-28-0
X
—
—
—
—
—
Zinc
7440-66-6
—
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 6: Risk Modeling
-------
7. Uncertainty and Sensitivity Analyses
This step consists of a review of the uncertainties associated with this beneficial use evaluation and
identification of any sensitive model inputs that might drive the identified risks. In any evaluation,
there will always be some sources of uncertainty. Characterization of uncertainties associated with
the data and modeling approach used in the evaluation can provide a better understanding of the
potential impacts on the analytical results and conclusions. The identification of sensitive inputs
can help define measures that may be targeted to effectively mitigate the identified risks. This
discussion focuses primarily on constituents and exposure pathways previously found to pose
potential risk in Section 6 (Risk Modeling).
7.1. Uncertainty Analyses
Uncertainty exists to some degree in any evaluation, and it may bias model results higher or lower
than actual values. It is important to understand both the direction and magnitude of uncertainties
present in the evaluation. The direction of an uncertainty is the tendency for it to push a predicted
value higher or lower than the true value, while the magnitude of an uncertainty is the extent to
which it may push the predicted value away from the true value. There are three primary causes
of uncertainty:
¦ Variability is the extent to which characteristics of environmental systems are heterogeneous.
Uncertainty is introduced if the distributions used as inputs for the models do not fully capture
the extent of real-world variability. Although variability can be better captured by collecting
additional data, it cannot be eliminated and must be treated explicitly in the assessment.
¦ Data uncertainty is a description of the imperfection in knowledge of the true value of a
particular model input. This uncertainty is generally reducible through additional research and
information-gathering.
¦ Model error occurs because models and their mathematical expressions are simplifications of
reality that are used to approximate real-world conditions, processes and their relationships.
These assumptions are sometimes necessary to solve complex mathematical equations or to fill
gaps in available knowledge. However, the simplification of complex systems may misrepresent
real-world conditions to an unknown degree.
Potential sources of uncertainty were mitigated to the extent practicable prior to running the full-
scale model. For example, uncertainties about the exact distribution of certain model inputs were
addressed through point values or distributions intended to reasonably bound the true range while
remaining protective. However, it is still useful to characterize the remaining uncertainties to
understand whether and how analytical results might change if these uncertainties could be fully
addressed in the model. The following text details the current understanding of the magnitude and
direction of major uncertainties identified for this beneficial use evaluation, grouped by topic area.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
7.1.1. Material Characterization
As part of this evaluation, EPA reviewed the available literature and assembled data on constituent
concentrations present in or released from FGD and mined gypsum. Appendix A (Constituent
Data) details the collection and review of this type of data. When individual data points or entire
studies were found to introduce unacceptable levels of uncertainty into the evaluation, these data
were removed prior to any quantitative analysis. The following text discusses the potential sources
of uncertainty identified in the remaining data.
Additional Treatment:
Pelletization is a treatment that involves tumbling gypsum with a binding agent, resulting in more
uniformly sized pellets. Known examples of binding agents include sodium lignosulfonate and
black liquor (U.S. EPA, 2012b).5 The advertised benefit of pelletized gypsum is that the material is
more uniform in size, which produces less dust and is easier to both transport and spread. Pelletized
mined gypsum is already available on the market (Chen et al, 2014; EPRI, 2008b; 2012; 2013; Kost
et al., 2014; U.S. EPA, 2012b). EPA did not identify any references in the literature for pelletized
FGD gypsum. However, this may be a result of a burgeoning market for this secondary material.
There were not enough data available to determine whether the pelletization process contributes
additional constituent mass to the mined gypsum or alters the leaching behavior of the gypsum.
Therefore, EPA treated pelletized gypsum as a separate material from untreated gypsum that fell
outside the scope of this evaluation. Further evaluation of pelletized gypsum may be warranted if
the same treatment is applied to FGD gypsum.
Bulk Characterization Data:
A number of the studies relied upon to characterize FGD gypsum blinded the source of the samples.
In some cases, this information was unknown even to the authors. As a result, despite attempts to
reduce bias through data management, there remains the potential for some uneven weighting of
the gypsum dataset toward certain regions of the country. Regardless, there is confidence that the
full range of coal characteristics have been captured in the available dataset. Samples collected by
EPA reflect a range of coal types, pollution control technologies, and wash status found across the
United States (U.S EPA, 2008; 2009b). Table 7-1 compares the data assembled by EPA with those
from all other literature sources for the constituents that were collected by EPA and were found
to have comparable washed and unwashed bulk content.
5) Black liquor is a secondary material generated by the kraft pulping process. This liquid contains a mixture of
pulping residues (e.g., lignin, hemicellulose) and inorganic compounds (e.g., sodium hydroxide).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
Table 7-1. Comparison of Data Collected by EPA and from Other Sources
Constituent
U.S. EPA (2008a; 2009b)
All Other Literature
Detection
Frequency
50th
Percentile
90th
Percentile
Max
Detection
Frequency
50th
Percentile
90th
Percentile
Max
Antimony
13/13
1.8
5.5
8.2
37/39
0.33
8.3
23.9
Arsenic
13 / 13
2.9
5.5
10
41 / 54
2.8
6.3
11
Barium
13 / 13
27.6
55.8
67
40/40
10
49.3
81.8
Cadmium
13 / 13
0.30
0.50
0.58
37/40
0.11
0.47
1.9
Chromium
13 / 13
7.7
13.4
14.9
42/45
3.6
7.4
15.0
Cobalt
13 / 13
2.5
3.4
4.3
27/40
0.25
0.66
3.1
Lead
13 / 13
2.4
3.8
6.5
32/40
1.0
2.0
8.3
Mercury
13 / 13
0.40
1.3
3.1
79/81
0.30
1.0
2.3
Molybdenum
13 / 13
3.7
6.3
12
38/41
0.7
2.5
6.2
Selenium
13 / 13
11.5
34.4
46
49/55
5.5
19.6
32
Strontium
13 / 13
177
383
530
37/37
154
338
405
Thallium
13 / 13
0.60
1.1
2.3
28/30
0.01
0.10
2.8
This comparison shows that the concentrations measured by EPA tend to be somewhat higher
than the remaining dataset. However, there is considerable overlap in the range reported by both
sets, with the major exceptions of cobalt and thallium. Although maximum concentrations are
similar for both of these constituents, both the median and high-end concentrations are an order
of magnitude different. It is unlikely that the differences result from analytical error in the EPA
data, as Agency quality assurance and quality control (QA/QC) protocols were followed and this
type of error would be expected to propagate to additional constituents. Instead, it is more likely
that the samples collected by EPA reflect a wider swath of the FGD gypsum generated in the
United States. EPA aimed to capture different coal types and pollution control technologies with
these samples, while other studies focused on samples that are commercially available now. It is
possible that higher concentrations in EPA samples reflect FGD gypsum that is not currently
marketed for use, but that may be in the future. Based on these considerations, EPA concluded
that it was appropriate to combine all the available data in the current evaluation. While some
uncertainty remains about the exact shape of the distributions, the amount of overlap provides
confidence that high-end concentrations have been adequately captured. The general agreement
between EPA data and other sources also provides confidence that pH-dependent leachate data
drawn from U.S. EPA (2008a; 2009b) also adequately capture high-end leaching behavior.
Constituents Without Characterization Data:
There are several constituents for which human or ecological benchmarks were identified, but for
which sufficient bulk content or leachate data were not available to reliably characterize potential
exposures. The full-scale results presented in Section 6 (Risk Modeling) indicate that constituents
most likely to pose environmental concerns are those that volatilize in the flue gas and concentrate
in FGD gypsum. The only other constituents that are known to be particularly volatile are
members of the halogen group. Of these elements, one or more relevant benchmarks were
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
identified for bromide, fluoride and iodide. Therefore, EPA focused on these constituents for
further consideration. The following discussion relies on all available sources of information to fill
data gaps and, as a result, includes a greater amount of uncertainty than the main evaluation.
Therefore, the concentrations estimated in this section should not be used outside of the context
discussed in this document:
¦ Bromide: EPA identified an ecological soil benchmark at 10 mg/kg (ORNL, 1997). Leachate
data, but no usable bulk content data, were found for this constituent. EPA instead used the
relationship between chlorine and bromide in coal to estimate a high-end concentration. The
typical ratio of chlorine/bromide in coal is 0.02 (U.S. DOI, 2012). Multiplying the 90th
percentile chlorine concentration in unwashed FGD gypsum by this ratio yields a bromide
concentration of 52 mg/kg. Accumulation in the soil over 100 years under the screening
scenario discussed in Section 5 (Screening Analysis) with no losses would result in a soil
concentration of 13 mg/kg. However, this is not realistic because bromide is highly soluble.
Accumulation with losses set at the 50th percentile unwashed leachate concentration of
160 ug/L results in depletion of the bromide added to the soil each year. Therefore, bromide
is unlikely to drive environmental concerns for soil.
¦ Fluoride: EPA identified an ecological surface water benchmark of 2,700 |_ig/L (MIDEQ, 2007)
and an MCL of 4,000 |_ig/L. Bulk content data, but no usable leachate data, were found for this
constituent. The comparison of washed and unwashed samples revealed that concentrations
of some washed samples were measured at higher levels than corresponding unwashed
samples. This indicates that losses during washing are within the range of measurement
uncertainty and so the two types of samples were combined. The 90th percentile
concentration of combined samples is 1,350 mg/kg. Assuming complete washout of fluoride
mass under the screening scenario discussed in Section 5 (Screening Analysis), the resulting
water concentration would be approximately 5,300 jag/L, which is almost twice the identified
ecological benchmark. However, fluoride in FGD gypsum is typically associated with fluorite
(CaF2) and so releases are controlled by the solubility of this mineral (Alvarez-Ayuso and
Querol, 2007). Thus, leaching of fluoride is not anticipated to sustain such high concentration.
Furthermore, combined with the order-of-magnitude or more decrease in concentration
observed for other constituents between screening and full-scale analyses, indicates fluoride
is unlikely to drive environmental concerns for ground or surface waters.
EPA also identified an ecological soil benchmark of 200 mg/kg (ORNL, 1997) and a human
health screening value of 3,100 mg/kg for fluoride. Bulk content data, but no usable leachate
data, were found for this constituent. The Agency identified one study from Spain that
reported a single sample of FGD gypsum with a fluoride leachate concentration in deionized
water mixed at a L/S ratio of 10:1 resulted in release of around 20% of the bulk constituent
mass present (Alvarez-Ayuso et al., 2006), which supports the conclusion that fluoride will
exhibit solubility-limited behavior. The leachate of other constituents in this sample generally
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
fell at or below median values in the FGD database, making it unlikely these data will
substantially overestimate potential losses. Accumulation in the soil over 100 years under the
screening scenario discussed in Section 5 (Screening Analysis) with no losses results in a
concentration of 325 mg/kg. Accumulation with leachate set to 20% of applied mass results in
a peak concentration of 150 mg/kg, which is below the identified benchmark. Therefore,
fluoride is unlikely to drive environmental concerns for soil.
¦ Iodide: EPA identified an ecological benchmark for iodide in soil of 4 mg/kg (ORNL, 1997).
However, no usable bulk content or leachate data were found for this constituent and no other
means to estimate values was identified. Therefore, no further evaluation for this constituent
is possible.
Available data indicates that bromide and other halogens have low toxicity (WHO, 2009). Indeed,
some are essential nutrients. However, it has been documented in laboratory and field studies that
higher levels of halogens in surface water can increase formation of disinfection byproducts (DBPs)
during water treatment (Luong et al., 1982; Heller-Grossman et al., 1993; Pourmoghaddas et al.,
1993; Cowman and Singer 1996; Chang et al., 2001; U.S EPA, 2002b; Duong et al., 2003; Liang and
Singer 2003; Ates et al., 2007; McTigue et al., 2014; Regli et al., 2015). Bromate (BrO 3) can form
when ozone reacts directly with bromide. Hypobromite (BrO) can form when chloride reacts with
bromide, which can then react with organic matter to form a range of brominated and mixed
chloro-bromo trihalomethanes. Although MCLs have been promulgated for total trihalomethanes
and other DBPs, there are currently no surface water benchmarks for halogens that address
subsequent formation of DBPs. There is also insufficient information available to reliably estimate
the extent to which DBPs may form. As a result, this may result in an underestimation of potential
risk, but the magnitude of this uncertainty is unknown. However, available leachate data
demonstrate that washing can reduce releases of both chloride and bromide by an order of
magnitude or more. Washing is sufficient to reduce leaching of chloride over an order of
magnitude and bromide to below detection limits in the most samples. Therefore, if the formation
of DBPs is a concern in a given area, then washing the FGD gypsum is an effective method to
substantially reduce releases of halogens to the environment.
7.1.2. Farming Practices
EPA reviewed the available literature to assemble data on where and how gypsum might be applied
across the country. Appendix C (Use Characterization) details the collection and review of this
type of data. There was little information available to define how gypsum is currently used in many
regions of the country. There is also the potential for practices to change over time as barriers are
removed. Therefore, EPA aimed to define the maximum extent that FGD gypsum might be used
without consideration of limits, such as regional availability of the material. This allowed EPA to
evaluate each of the different uses, but likely overestimated the area over which FGD gypsum will
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
actually be applied. The following text discusses the sources of uncertainty associated with where
and how FGD gypsum is applied.
Extent of Use:
This beneficial use evaluation defined the maximum range that FGD gypsum might be used based
on the location of coal-fired utilities that the Energy Information Administration (EIA) database
reported as generating this secondary material (EIA, 2017). This range represents a snapshot in
time and will be subject to change as older plants close and new plants open. There is no reliable
way to forecast where a new plant will open and the type of pollution control technology it would
install. However, it is possible to review the current landscape to determine if any existing facili ties
may retrofit and install forced oxidation scrubbers with limestone. EPA found that 293 of the 597
power plants operating in 2017 already produced FGD gypsum or have some other form of sulfur
dioxide control. Figure 7-1 provides a comparison of the relative locations of the current economic
feasibility zone for FGD gypsum, coal-fired plants without sulfur dioxide controls, and non-
attainment areas for sulfur dioxide to determine where new sulfur dioxide controls are most likely
to occur in the future.
Based on this map, the vast majority of non-attainment areas for sulfur dioxide are located within
the current economic feasibility zone. The facilities located outside the modeled zone tend to have
smaller generating capacities, which make them less likely to be a major future source of sulfur
dioxide. Installation of a scrubber system onto one the facilities located near the non-attainment
areas may extend the feasibility zone slightly. However, further expansion would be limited by
the larger number of gypsum mines in surrounding areas. Based on this analysis, there is little
concern that the extent that potential FGD gypsum use was underestimated. Instead, it is far more
likely that the feasibility zone overestimates the area that gypsum will actually be used.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
7-6
Section 7: Uncertainty and Sensitivity Analyses
-------
Extent of Application:
EPA estimated the total extent of agricultural land in the United States with data from satellite
imagery and USDA surveys collected between 2010 and 2015. Multiple years of data were used to
capture the periodic rotation of land use. EPA did not identify any reliable way to estimate how
much of this total land will be utilized in any given year and so EPA assumed that application of
FGD gypsum could range anywhere from 0 to 100% of the field area. In each model run, the extent
of application was independently varied. This allowed consideration of variable mass loadings from
different regions to the watershed outfalls, though some application still occurred in nearly every
region in a given model run. This assumption ensures that the evaluation does not underestimate
risks from releases that have the potential to compound from different parts of a watershed. EPA
did not identify any data that could be used to further refine the model.
It is considered unlikely for a number of reasons that FGD gypsum will be applied every year at
high-end rates over all the cropland in a watershed. First, application over the full area may simply
not be needed. Beneficial use of FGD gypsum application are only needed for specific purposes
that rarely extend to the full area. For example, the highest rate of FGD gypsum applications
modeled was for the reduction of phosphorus in runoff, but this use would only be applied on
individual fields where the potential for excessive phosphorus in runoff had been identified (i.e.,
high soil test phosphorus or use of manures as fertilizers). Also, in any given year, the actual extent
of farmed land will be less than the total possible because some fields will be left fallow or
transitioned to crops that do not require FGD amendments as a result of either agronomic or
economic conditions. Second, it is possible that the benefits of application will extend beyond a
single year. Finally, there may also physical limitations to how much FGD gypsum can be applied
in a given area based on generation rates. The ACAA estimates that nearly 18 million tons of FGD
gypsum were generated in 2020, but that nearly two-thirds of that were already diverted to
wallboard production or other uses (ACAA, 2021). Even if all of the remaining FGD gypsum were
directed to agricultural fields, that would only allow application of 3 tons/acre on less than 1% of
the nearly 193 million acres of modeled cropland. Altogether, this is expected to result in an
overestimation of potential risks. The magnitude of this uncertainty is generally expected to be
large and to be even larger for higher-order streams evaluated for human exposure through fish
ingestion. This is because the total land area that feeds into these streams is larger, sometimes
spanning across multiple states. As the contributing land area increases, it becomes progressively
less likely that a majority of the land area would have FGD gypsum applied at high-end rates in a
given year.
Duration and Frequency of Application:
EPA selected 100 years as a reasonable upper bound on the duration of application. This value has
been used in previous evaluations of agricultural amendments (U.S. EPA, 1992a,b). In each model
run, the number of years was varied between 1 and 100 based on a flat distribution. It is unknown
how much FGD gypsum will be generated or otherwise available for use that far into the future.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
Even if the use of this secondary material is still common at that point, it is unlikely that it will be
applied every year over that maximum duration. A combination of fallow periods needed for soil
health and economic drivers that rotate crops or take fields out of production will periodically halt
applications for one or more years at a time. It is also possible that application of FGD gypsum will
not be as frequent in a given area because the benefits provided are sustained for more than a single
year, further reducing need for annual application. Based on these considerations, the modeled
duration and frequency of application is likely to result in an overestimation of potential risks. The
magnitude of this uncertainty is expected to be large.
Tilling Practices:
Conservation tilling is a general term for a varied set of practices that minimize disturbance to the
soil during farming. It is estimated that nearly half of all farms in the United States currently
implement some form of conservation tilling (USDA, 2014a). Potential benefits include reduced
soil erosion, reduced nutrient runoff, and increased retention of both moisture and organic matter
in the soil (USDA, 2015b). However, changes in tillage system management often change over time
depending on land ownership/management and crops being grown. Cropping systems that
maintain "no-tillage" operations over many years are relatively rare. Most conservation tillage
systems instead involve some limited tillage operations to prepare a seed bed for planting or to
distribute fertilizers and other amendments into the soil profile. Also, some crops, such as peanuts
and potatoes grow underground and require soil disturbance in order to harvest the crop. Based
on consultation with USDA staff, EPA assumed that soil tillage intervals between 5 and 20 years
would occur as part of normal farm production. Thus, long-term exposures were assumed to be
better reflected by concentrations mixed in the soil column.
In the absence of tilling, insoluble constituent mass will accumulate in the topmost soil at higher
rates than modeled in this evaluation. However, the lowest soil benchmarks identified for several
constituents (e.g., selenium) are protective of plants. Higher temporary accumulation in the soil
surface is unlikely to impact uptake across the root zone. Potential risks for other receptors in tilled
soil were only identified after nearly 100 years of consecutive application. A shorter period without
tilling is unlikely to result in exposures appreciably higher than those modeled in this evaluation.
Furthermore, it is unlikely that the applied gypsum will remain at the soil surface for 20 years or
more. Other additions to the soil, such as plant residue and manure, will further dilute and limit
exposures to the topmost soil. Therefore, EPA concluded that the magnitude of this uncertainty is
small.
7.1.3. Water Budget
EPA used precipitation data from climate stations together with regional soil properties to model
infiltration rates across the country. Any water not lost to infiltration or evapotranspiration was
assumed to run off into nearby water bodies. Appendix E (Probabilistic Modeling) details the
methodology used to estimate this mass balance. The Agency is aware of other potential sources
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
and sinks for water, but it was not possible to incorporate each one quantitatively with the data
available. The following text discusses the methodology used to model water flow through the
environment and potential sources of uncertainty associated with the available data.
Irrigation:
In some areas of the country, precipitation is not sufficient in volume and/or frequency to meet
crop needs. In these areas, rain water may be supplemented through irrigation from nearby surface
or ground water. EPA did not identify a reliable means to estimate the additional volume of water
that might be applied on an annual basis. This amount is likely to vary each year based on rainfall,
irrigation water availability, and the type of crops grown. The rate and time (day and year) of
application will influence how much of the irrigation water evaporates, infiltrates, or runs off. To
better understand the effect this uncertainty might have on calculated risks, EPA multiplied the
total cropland in each county by the percent irrigation reported in the 2012 Census of Agriculture
(USDA, 2014b) and then calculated the fraction of total cropland irrigated in each use zone.
Figure 7-2 presents data on the prevalence of irrigation as a fraction of total field area both in each
county and for each use. Percentages were calculated on a county basis and so do not align exactly
with the T lUCs considered in this evaluation.
Phosphorus
Ca or S Nutrient
Infiltration
Sodic Soils
Aluminum Toxicity
10.2%
11.7%
11.2%
13.2%
18.3%
Figure 7-2: Percentage of agricultural land irrigated in each use area
The greatest density of irrigated fields occurs in the western United States, in areas that fall largely
outside of the economic feasibility zone for FGD gypsum. Given the arid environment in these
locations, it is likely that irrigation rates are more closely tied to plant requirements with an aim
to minimize losses to runoff or infiltration. Higher density irrigation on the east coast occurs in
areas in Florida and along the Mississippi river that already receive a substantial amount of
Beneficial Use Evaluation of FGD Gypsum in Agriculture
7-9
Section 7: Uncertainty and Sensitivity Analyses
-------
precipitation. In these areas, the available constituent mass applied each year was often depleted
over the course of the year by precipitation alone. Therefore, any further infiltration or runoff
from irrigation is expected to only further dilute concentrations in the environment. As a result,
the effects of this uncertainty on calculated risks are anticipated to be minimal.
Tile Drains:
Subsurface drainage tiles are used in areas of flat terrain and poorly drained soil to drain away high
ground water tables and prevent the inundation of fields. Tile drains have been used since the early
1900s, primarily in the Midwestern United States. There remains a fair amount of uncertainty
about the exact location and spacing of tile drains in the United States (Williams et al., 2015).
However, even if the location of these tile drains were well known, some are quite old and may
have become so clogged with sediment over time that the capacity to transmit water is greatly
diminished. To better understand the effect this uncertainty may have on calculated risks, EPA
estimated prevalence of tile drains based on data reported by the World Resources Institute (WRI,
2007). Figure 7-3 presents data on the prevalence of tile drains as a fraction of total field area in
each county and in each use zone. Percentages were calculated on a county basis and so do not
align exactly with the HUCs modeled in this evaluation.
Phosphorus
Ca or S Nutrient
Infiltration
Sodic Soils
Aluminum Toxicity
4.0%
Percent Tile Drains
o
1-20
21-40
41 - 60
61 - 80
¦ 81 -100
Figure 7-3: Percentage of agricultural land with tile drains in each use area
EPA used the base flow index (BIT) to estimate the relative fraction of surface water flow that
originates from ground water and overland flow (or near surface discharge). The USGS calculates
BFI with the approach proposed by the British Institute of Hydrology (Institute of Hydrology,
1980). The method uses measured flow minimums to estimate the annual volume of base flow to
water bodies and calculates a ratio of the base flow to the total flow volume for a given year based
on multiple years of data. Therefore, to the extent that the existing tile drains still divert infiltration
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
^ a 7-10
Section 7: Uncertainty and Sensitivity Analyses
-------
directly into water bodies, the effects should already be reflected in this ratio. As a result, the
effects of this uncertainty on calculated risks are anticipated to be minimal.
Precipitation Data:
The weather data used in this risk assessment was collected for a period from 1961 through 1990.
Therefore, some uncertainty is introduced because any changes in weather patterns that have
occurred since 1990 are not reflected in this data set. The National Climate Assessment Report
documents region-specific changes in rainfall, temperature and episodic rainfall events over recent
decades (Melillo et al., 2014). In general, this report identified a trend towards greater amounts of
rainfall that are more concentrated in discrete events, particularly in the Northeast and Midwest.
More intense storms could result in a larger fraction of the precipitation directed to overland runoff
than predicted in this evaluation because storm events would be more likely to exceed the rate
that water can infiltrate into the soil. This might increase or decrease concentrations in water
bodies in different circumstances. For example, concentrations may increase from the greater
constituent mass that would flow directly into the water body, but concentrations may be balanced
out by greater runoff from the remainder of the watershed or decrease from higher total flow from
contributed by upstream watersheds. Thus, the overall effects of this uncertainty are unknown.
7.1.4. Fate and Transport
EPA used data from the LEAF test methods to estimate the initial release of constituent mass from
FGD gypsum. EPA then used a combination of EPACMTP and other models to simulate the
subsequent movement of these constituents through the environment. Appendix E (Probabilistic
Modeling) details the methodology used to model fate and transport. The following text discusses
potential sources of uncertainty associated with the data and models used to estimate the fate and
transport of constituents in this evaluation.
Leaching Behavior:
EPA made an initial determination about constituent leaching behavior based on a comparison of
mean washed and unwashed concentrations measured with LEAF Method 1313 across the relevant
pH range, as discussed in Section 4 (Comparison with Analogous Material). A second, more refined
analysis of leaching behavior was based on measured concentrations at each pH, as discussed in
Section 6 (Risk Modeling). The agreement between the approaches is generally good, with the
exceptions of antimony and lead. A comparison of washed and unwashed samples in Section 4
indicated that these two constituents are availability-limited over the relevant pH range, while the
comparison in Section 6 indicated the constituents are solubility-limited. Figure 7-4 presents the
pH-dependent leaching behavior of two sets of samples for lead. These two sample pairs were
chosen because the unwashed data were detected over the majority of the pH range. WAU/WAW
and TAU/TAW are the sample IDs for unwashed/washed sample pairs.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
1000
S 100
o
4-»
ro
ai
— .
\
vrt-
v
A
\
/
Detection Limit
Unwashed Sample WAU
¦ Washed Sample WAW
Unwashed Sample TAU
¦ Washed Sample TAW
7
PH
10 11 12 13
Figure 7-4: Comparison of lead leached from washed and unwashed samples.
At a highly acidic pH values, lead appears to be solubility limited because washed and unwashed
concentrations are nearly identical. Yet, as the pH increases (i.e., becomes more basic), differences
between washed and unwashed samples become apparent. After washing, samples measured above
a pH of 3 are all non-detect, which indicates washout is occurring. It is likely that this discrepancy
is caused by the presence of different lead compounds within the gypsum. The first is more soluble
and readily washes out around a neutral pH during washing. The second is only soluble at a highly
acidic pH and is retained during washing. Because the method used to determine leaching behavior
in Section 6 is based on a comparison with the highest measured leachate concentration, it appears
that the constituent is solubility-controlled over the full pH range. It is more likely that, under
typical environmental conditions, a small fraction of the constituent mass would quickly wash out,
followed by solubility-controlled leaching at a far lower rate for any remaining leachable mass.
Calculating the available content based on the maximum concentration over the relevant pH
range, rather than the full pH range, results in a 90th percentile available content closer to 5% of
the total mass. This is far lower than 100% of the total mass used in this evaluation. Thus, the
current evaluation overestimates exposures to these two constituents due to leaching. However,
because neither antimony nor lead were found to be risk drivers in this evaluation, the magnitude
of this uncertainty is considered negligible.
Field Distribution:
Due to model limitations, EPA had to assume that all the farm fields with FGD gypsum applied in
a given watershed formed a continuous parcel of land and, thus, a single source of leachate. In
reality, fields can be dispersed widely and non-continuously across the landscape. The greater the
distance between individual fields, the greater the opportunity for dilution and attenuation in the
environment before a release reaches downgradient wells or water bodies. Additionally, some farm
fields may be located downgradient of or entirely outside the flow path of some private wells,
limiting the impact to some water supplies. This is likely to result in an overestimation of risk for
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
groundwater and surface water to some degree; however, the magnitude of this uncertainty is
unknown.
Temporal Variability of Releases:
This beneficial use evaluation focused on the potential for adverse impacts associated with longer-
term exposures, which are based on environmental concentrations averaged over the course of a
year. This approach is believed to reasonably reflect exposures that may result from groundwater
pathways, such as discharge to surface water, because transport is a gradual and continuous process.
However, runoff events are intermittent throughout the year. As a result, there is potential for
leaching of most soluble constituents at higher concentrations following individual runoff events.
This might result in shorter periods of higher concentration in nearby streams than modeled in
this evaluation.
EPA did not identify sufficient models or time-dependent data to support estimates of shorter-
term exposures on a national scale. Such modeling would require information on both the specific
time of application at different fields across a watershed, the specific location of the fields relative
to the water bodies, and the relative timing, intensity, and duration of individual runoff events.
The current model instead assumes that mixing of precipitation and FGD gypsum is uniform, that
contact between the two occurs long enough to achieve near equilibrium concentrations in the
runoff, and that there are no losses of dissolved constituent mass as runoff flows to nearby water
bodies. However, there are a number of reasons why these assumptions may not always hold.
FGD gypsum is unlikely to be applied to all modeled fields at the same time or even in the same
year. Thus, releases to runoff from different parts of the watershed can occur at different times,
resulting in greater dilution of runoff from individual fields. Applications will not occur if the
ground is already saturated due to difficulty operating spreading equipment on water-logged soil.
As a result, some of the initially released mass would first infiltrate to the subsurface and smaller
precipitation events may not exceed the initial abstraction at all (i.e., water diverted to infiltration,
evaporation, or other pathway prior to runoff). Once runoff begins, the duration of contact
between the flowing water and the soil will decrease. As a result, runoff may not always have
enough sustained contact with the FGD gypsum or mixing with the intermingled water for
leachate to achieve equilibrium concentrations. Even after release, there is potential for sorption
and other interactions between runoff and the soil matrix that could limit immediate transport to
surface water and may further promote infiltration to groundwater. Finally, not all runoff will
reach the water body at the same time or same location. Travel times from the furthest point of a
HUC10 watershed to a higher-order stream can take multiple days. Combined, all of these
considerations will considerably limit shorter-term concentrations in water bodies.
There is also some potential for the methods used to estimate the magnitude of selenium in leachate
to overestimate short-term concentrations. Batch leaching tests, such as EPA Method 1313 and
1316, measure dissolved concentrations under equilibrium conditions. As a result, these tests do
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
not provide any information about factors that would affect the rate of dissolution prior to reaching
equilibrium. For example, the selenium captured in FGD gypsum may associate with calcium. It
has been shown that shown that selenate can substitute for sulfate in the gypsum structure
(Fernandez-Gonzalez et al., 2006). Because this selenium is incorporated within FGD gypsum,
rather than sorbed to the surface, releases would be limited by the rate at which the bulk gypsum
dissolves. Therefore, it is unlikely all of this selenium would be released until all of the applied
FGD gypsum had also been depleted from the fields.
Some studies have analyzed runoff from fields treated with FGD gypsum (Torbert and Watts, 2014;
Watts and Torbert, 2017; Schomberg et al., 2018; Torbert et al., 2018). Washed FGD gypsum was
applied to field plots and the runoff generated by simulated rainstorms was collected at 10-minute
intervals. The studies reported that cadmium was not detected in any samples (<2 |_ig/L), manganese
was detected only in initial runoff events (65 to 290 jag/L), and selenium was detected only in initial
runoff events at low concentrations (5 jag/L). The reported values for cadmium and manganese
align well with median leachate measured for washed FGD gypsum over the relevant pH range
(0.85 |_ig/L and 65 |_ig/L, respectively), but the values for selenium are considerably lower than the
median washed leachate (45 jag/L). These studies provide some confirmation that measured
leachate concentrations can provide a reasonable estimate of runoff concentrations and may, in
some cases, overestimate these releases. However, it is difficult to draw broader conclusions from
the studies because of the lack of data on leaching potential of the FGD gypsum prior to application,
different soil types, and other environmental conditions evaluated. Therefore, the magnitude of
this uncertainty is not known.
Water Body Size:
To estimate ecological exposure to surface water and sediment, EPA modeled concentrations at
the outfall of 1st and 2nd order streams ("headwater streams") to any higher-order streams. Stream
order is based on Strahler number, which assigns an order of 1 to initial headwater streams and
increases each time two streams of the same order intersect, and was used as a metric for relative
stream size and flow. The rationale for this approach is that a majority of runoff first flows through
headwater streams, which provides a best estimate of immediate releases prior to further mixing
and dilution during flow through successive watersheds. The cumulative land area that drains to
these streams provides an extensive amount of habitat for wildlife, though is possible that some of
the smaller streams are too small or ephemeral to sustain a complex ecological community. It is
not known whether or to what extent this approach may overestimate risk; however, it is unlikely
to underestimate risk.
To estimate human exposure to fish, EPA modeled concentrations at successive HUC10 outfalls
until the stream order reached 6th order or above, as well as any HUC12 outfalls that discharge
directly into high-order streams. The rationale for this approach is that:
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
¦ Streams below 3rd order are unlikely to support a fish population that could sustain fishing at
the rates that correspond with the ingestion rates modeled in this evaluation;
¦ That data used to characterize some variables, such as BFI, may become less reliable in high-
order streams due to long travel times and large cumulative drainage areas;
¦ The probabilistic modeling of surface water concentrations over the great distances covered
by high-order streams was prohibitively time and resource intensive; and
¦ The addition of HUC12 outfalls captures releases from land area that would have otherwise
been omitted due to the dominance of high-order streams, such as the Mississippi River, in
certain regions of the country.
The current evaluation may overestimate risks to some degree by excluding high-order streams. It
is anticipated that concentrations will generally decrease as stream order increases because the
total drainage area contains proportionally less agricultural land. Yet, streams between 3rd and 5th
orders represent nearly 90% of the flowlines above 2nd order (U.S. EPA, 2008b). Therefore, the
exclusion of even higher-order streams may not have a substantial impact on overall risks.
Water Body Type:
As part of this evaluation, EPA modeled the transport and accumulation of constituent mass in
surface water as it flowed through multiple HUC10 watersheds. This evaluation relied on the
National Hydrography Dataset (NHD) to define the direction and magnitude of flow in each
segment of the water body. However, there were not sufficient data to model every water body
within the defined use zones. In particular, EPA was not able to model "terminal water bodies,"
which are those with an NHD flowline that has a terminal flag (i.e., unidirectional flow over the
ground surface stops). For example, fish ponds and other relatively small and static (i.e., lentic)
water bodies. Because there is no flow path through these water bodies, there is no information
that could be used to estimate the associated volume or turnover rate. Modeling these water bodies
would require a number of additional assumptions that would introduce a significant amount of
uncertainty into the evaluation. It is possible that the exclusion of these water bodies
underestimates potential risks because the relatively small volume combined with longer hydraulic
residence time could result in longer exposures to higher concentrations. In addition, EPA
developed a separate water quality benchmark for selenium in these lentic water bodies to account
for the effects of prolonged exposure (U.S. EPA, 2016c). Therefore, further evaluation may be
warranted before FGD gypsum is applied in the vicinity of these types of water bodies.
7.1.5. Exposures
EPA used the constituent concentrations modeled in each medium together with available data on
receptor characteristics, behavior and sensitivity to estimate potential exposure and resulting risks.
Appendix B (Benchmarks) details the data and approach to develop benchmarks used to calculate
risk. The Agency is aware of other potential receptors and types of exposures beyond those
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
evaluated, but could not quantitatively account for them with available data. The following text
discusses the potential sources of uncertainty associated with the data and methodology used to
calculate exposure and subsequent risk.
Fish Ingestion Rates:
Results presented in Section 6 (Risk Modeling) reflect modeled risks for recreational fishers. These
receptors were selected because they are more likely to consume fish caught from a single water
body. Therefore, these individuals and their families represent a sensitive subpopulation that is
more likely to be exposed through fish ingestion (U.S. EPA, 2011). Subsistence fishers are another
sensitive subpopulation that could be exposed at levels higher than the general population. This
subpopulation is not well defined or characterized and may include a diverse range of rural and
urban receptors that rely on fresh-caught fish as a major portion of their overall diet. Given the far
greater uncertainty associated with these subsistence fishers, EPA considered these receptor as part
of an uncertainty analysis to provide further context for results. The primary difference between
recreational and subsistence fishers in this evaluation is the rate of fish ingestion, with subsistence
fishers consuming about three times more fish. However, there are uncertainties associated with
the ingestion rates used for both recreational and subsistence populations.
The full-scale model relied on a fixed, high-end ingestion rate to characterize fish ingestion for
each age cohort due to a lack of data that would allow for a broader characterization of these
subpopulations. This can overestimate exposures because a single value does not reflect the full
variability of the modeled population. This uncertainty will be greater in areas where diets may
vary throughout the year based on seasonal access to fish and the availability of other protein
sources, such as wild game. The model also assumed that all the fish consumed were caught from
a single affected waterway. This could overestimate exposures to the extent that the diet also
incorporates fish sourced from beyond local waters. As a result, the data used to characterize fish
exposure is expected to overestimate potential risks. However, the magnitude of this uncertainty
is not known.
Constituents Without Benchmarks:
There are some constituents for which human or ecological benchmarks were not identified. It
was not possible to quantitatively evaluate these constituents in either the screening or full-scale
modeling. For other constituents, toxicity values were identified for some, but not all, relevant
exposure pathways. In these cases, the potential risks to receptors in these media could not be fully
quantified. The absence of a toxicity value is not necessarily equivalent to the absence of risk.
Constituents may pose pathway-specific risks or may influence the fate and transport or toxicity
of another constituent, resulting in an underestimation or overestimation of risk. The magnitude
of this potential underestimation is unknown.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
Aggregate Exposures:
Aggregate exposure is the combined exposure to a single constituent through multiple exposure
pathways. Aggregate exposures may be simultaneous or sequential, but all occur within the critical
window for the health effect. This beneficial use evaluation considered potential risks to human
health from individual constituents and individual pathways. It is possible that individuals could
be exposed simultaneously through ingestion of ground water, soil, produce, livestock and fish.
However, it is highly unlikely that receptors would be exposed to high-end concentrations through
every route. Even if receptors are exposed to high-end concentrations through every pathway, the
constituents found to accumulate in each media are different. Therefore, the overall magnitude of
this uncertainty is considered minimal.
Cumulative Exposure:
Cumulative exposure is the combined exposure to multiple stressors that produce the same health
effect. These different stressors may interact with one another in antagonistic or synergistic ways
that serve to mitigate or exacerbate potential health effects. The extent of these interactions may
change based on the level of the stressors present and the order of exposure. The toxicity values
used in the current evaluation do not account for these types of interactions. Where the potential
for simultaneous exposure to multiple constituents exists, current EPA policy is to assume that the
risks resulting from these exposures are additive (U.S. EPA, 2000).
The only constituents carried forward to the full-scale evaluation that share a similar health
endpoint are arsenic and chromium in ground water (cancer) and mercury and selenium in fish
(neurological). Arsenic and chromium are both far below levels of concern in all media and so the
uncertainty associated with this endpoint is minimal. However, mercury and selenium were both
identified near levels of concern for fish ingestion. There is no relationship between the levels of
mercury and selenium in FGD gypsum, so it is unlikely that receptors would be exposed to high-
end concentrations of both constituents at the same time. Furthermore, numerous investigations
have found selenium can mitigate the toxicity of mercury (HHS, 2003). Recent studies have
proposed several mechanisms for detoxification, such as mercury sequestration in metabolically
inert compounds, formation of selenium-based antioxidants, demethylation of methylmercury, or
replenishment of selenium-containing enzymes needed for metabolism (Bjorklund et al., 2017;
Ralston and Raymond, 2018). Despite this evidence from the literature, it is not possible to quantify
whether and to what extent selenium will reduce mercury toxicity in all circumstances. However,
it is unlikely that cumulative exposure will compound the risk of the two constituents. Therefore,
the uncertainty associated with these endpoints is considered minimal.
Sulfate does not share any known health endpoints with selenium; however, available research
indicates the presence of dissolved sulfate can reduce the bioavailability and toxicity of selenium
(Banuelos et al., 1990; Banuelos & Mayland, 2000; Bell et al., 1992; Brix et al., 2001; Chaney et al.,
2014; Hopper & Parker, 1999; Qin et al., 2013; and Yang, 1995). This is attributed to the fact that
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
selenium and sulfur have similar ionic structures and so the two elements can be transported by
the same membrane proteins. As a result, sulfate can compete with selenium for cell uptake. The
potential risks from selenium identified in this evaluation are driven by bioaccumulation in the
food chain. Thus, competition between sulfate and selenium for uptake by algae and other lower
trophic level organisms in surface water (e.g., Daphnia) would also result in larger reductions in
uptake by fish and other higher-order receptors. Sulfate is a primary component of FGD gypsum
(CaS04) and leaches at concentrations at or above 1,000 mg/L from both washed and unwashed
samples. Brix et al. (2001) studied selenium in aquatic environments and found that sulfate strongly
inhibited selenium uptake even at much lower concentrations. Yet there have also been
documented cases of selenium accumulation in wildlife around waters with sulfate concentrations
as high as 2,000 to 100,000 mg/L (Birkner, 1978; Skorupa, 1998). The differences among these
studies is likely associated with the oxidation state of selenium, with sulfate competing with most
effectively with selenate (Se+6, VI) (Ogle et al., 1988). EPA was not able to quantitatively evaluate
transport of sulfate ions through the subsurface or the impact on selenium uptake at various
concentrations and different environmental conditions due to methodological challenges and a
lack of relevant data (U.S. EPA, 2016c). Because any competition from sulfate would reduce
exposure to selenium, the inability to quantify the effects of sulfate will result in an overestimation
of risk. However, the magnitude of this uncertainty is unknown.
Selenium Speciation
The oxidation state of selenium can impact the mobility of this constituent in the groundwater.
The most common forms of selenium found dissolved in groundwater under the standard range of
environmental conditions are selenite (Se+4, IV) and selenate (Se+6, VI), with the latter as the
more mobile form. Available information indicates that the dominant form of selenium expected
in bulk FGD gypsum is selenate (EPRI, 2011). Therefore, all selenium applied and leached in the
full-scale model was assumed to be present as selenate.
Previous modeling found the difference between high-end surface water concentrations resulting
from groundwater transport for the two selenium species was over a factor of 100 due to differences
in retention onto subsurface soils (U.S. EPA, 2014b). Given the magnitude of the difference relative
to the modeled risks for selenate, EPA did not separately model the transport of selenite. Once
released into the environment, the dominant oxidation state will be controlled by local pH and
redox conditions that can be influenced by plant and microbial activity. This evaluation could not
consider how these types of site-specific factors may affect the oxidation state of selenium during
transport through the subsurface. However, to the extent that some fraction of the selenium is
either initially present as selenite or converted to this state after application, the full-scale model
has the potential to result in an overestimation of risk for this constituent. The magnitude of this
uncertainty is unknown.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
Chromium Speciation:
Results presented in Section 6 (Risk Modeling) reflect the most toxic species of each constituent.
For chromium, the hexavalent (Cr+6, VI) species is both more toxic and more mobile in the
environment than the trivalent (Cr+3, III) species. EPA did not identify much data on how much,
if any, chromium (VI) is initially present in FGD gypsum. Torbert et al. (2018) analyzed runoff
from a samples off FGD gypsum and found measurable chromium (VI). Out an abundance of
caution, the Agency initially assumed that all chromium was present in the hexavalent state. To
understand the extent to which effects may vary, EPA recalculated risks with the reference dose
for chromium (III). Table 7-2 compares these results of the modeled risks for different chromium
species.
Table 7-2. Comparison of Model Results for Hexavalent and Trivalent Chromium
Constituent
Wash Status
Phosphorus
Runoff
90th Percentile HQ
. ,.ljt Nutrient
Infiltration . .
Amendment
Sodic
Soils
Aluminum
Toxicity
Chromium (VI)
Combined
1.3
1.1
0.57
0.75
0.74
Chromium (III)
Combined
0.02
0.01
Although the current evaluation assumed all of applied chromium was hexavalent, the model used
empirical leachate data to estimate losses from the soil. The speciation of chromium in the leachate
is not known. However, chromium (III) is less mobile and so could result in higher estimates of
accumulation in the soil. If even a minor fraction of the applied chromium is trivalent, then it
would eliminate potential long-term risks because the risks for chromium (III) are two orders of
magnitude lower than those for chromium (VI). In addition, reduction of chromium (VI) has been
shown to be energetically favorable and unlikely to reverse in high-organic, aerobic soils (Brose
and James, 2010; HHS, 2012). Even if all the chromium present in FGD gypsum is hexavalent at
the time of application, no single application at the rates considered in this evaluation would pose
short-term risk. Therefore, EPA concludes that long-term risks from chromium in soil will also be
below levels of concern.
Background Soil Concentrations:
Background concentrations are the constituent levels found in environmental media that have not
been impacted by releases from the waste. Background concentrations may originate from natural
or anthropogenic sources. The current evaluation assumed that background concentrations in each
medium (e.g., soil, ground water) were negligible. The modeled exposures are based solely on
releases from applied FGD gypsum. This approach was selected because background can be highly
variable, even over small areas, and so it is not possible to reliably characterize contributions from
background without robust, local data. The following text discusses the potential sources of
uncertainty associated with background.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
To better understand how long-term application of FGD gypsum might add to exposures from soil,
EPA compared the modeled concentration of thallium accumulated from FGD gypsum application
with measured background surface soil concentrations from across the United States (U.S. DOI,
2013). The results of this comparison are presented in Table 7-3. Although this type of data is
useful for qualitative comparisons, the data cannot be used to reliably estimate total soil
concentration that might result from application of FGD gypsum. Soil concentration are highly
site-specific and can vary considerably over small areas. Therefore, a nation-wide or even a state-
wide dataset may not provide accurate estimates of total concentrations. In particular, agricultural
fields that have been heavily engineered may have higher concentrations of some elements than
other undisturbed soil. Therefore, the concentrations in this table should not be extrapolated
outside of this limited context of this discussion.
Table 7-3. Comparison of FGD Gypsum and Surface Soil Concentrations
Constituent
Percentile FGD Gypsum
Accumulation (mg/kg)
50th 90th
Percentile Background
Surface Soil (mg/kg)
10th 50th 90th
Thallium
0.001 0.05
0.2 0.4 0.7
Long-term accumulation of thallium from FGD gypsum is lower than the majority of background
surface soil concentrations, even at high-end concentrations. That would mean that more than
90% of existing surface soils pose higher risks than those modeled in this evaluation. The majority
of thallium in background soil may not be bioavailable in some areas, but this is unlikely to be true
across the entire country. Instead, it is more likely that this evaluation overestimated exposures to
beef and milk through the combination of high-end data and assumptions intended to protect
human health. There is greater uncertainty associated with the beef and milk pathways because
both require modeling accumulation in the soil, followed by sequential uptake into plants and
cattle prior to ingestion by human receptors. These multiple levels of accumulation compound the
uncertainties.
It is not possible to substantially refine risk estimates with available data. However, the fact that
high-end (i.e., 90th percentile) thallium accumulation in soil from FGD gypsum application is
lower than the low-end (i.e., 10th percentile) of existing background concentrations indicates that
contributions to existing exposures is minor. The majority of thallium was found to remain in the
soil, rather than be released into infiltration or runoff, so it is not likely that the magnitude of
potential thallium accumulation was underestimated. Therefore, given the low exceedance of the
health-based criteria identified in the full-scale model, EPA concludes that all risks from thallium
in soil are below levels of concern.
Background Water Concentrations:
Mercury and selenium are the two modeled constituents found at or near levels of concern in
surface water. FGD gypsum is not the only source of these contaminants in the environment. Other
point and non-point sources from either natural or anthropogenic sources can also contribute to
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
a 7-20
Section 7: Uncertainty and Sensitivity Analyses
-------
levels in surface water. Both constituents are causes of contamination reported by states under the
Clean Water Act Section 303(d). This list of impaired or threatened waterways is compiled by
states, which have primary responsibility to notify the public of chemical contamination that may
present a public health hazard. Figure 7-5 presents two maps of waterways within the economic
feasibility zone that were reported as impaired for selenium and mercury for any reason during
the most recent round of reporting in 2016. Impaired waters are shown in red. Blank areas
represent regions outside the feasibility zone or where state data were not available in a form that
could be readily mapped.
Figure 7-5: Occurrence of impaired waterways for selenium (top) and mercury (bottom)
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Section 7: Uncertainty and Sensitivity Analyses
-------
These figures are provided for illustrative purposes only. The available data are not complete and
recommendations based on this current list will gradually become outdated as sources of
contamination are addressed and updated. Proximity to an impaired waterbody does not
necessarily mean use of FGD gypsum is inappropriate. The majority of modeled scenarios fell far
below levels of concern and are not likely to represent a substantial source of either mercury or
selenium. However, existing (background) sources of contamination can still be an important
consideration when determining where application of FGD gypsum may be appropriate.
7.2. Sensitivity Analyses
The purpose of these sensitivity analyses is to review the results of the full-scale model reported
in Section 6 (Risk Analysis) and identify any sensitive model inputs that could be used to limit
releases and reduce modeled risks to below levels of concern. Based on the uncertainty analyses,
EPA previously determined that all risks from soil pathways fall below levels of concern and so
the following discussion focuses on the remaining risks identified for releases to surface water. The
model found that risks for this pathway fell below levels of concern for a majority of model runs;
risks are instead driven by high-end application scenarios. Therefore, it is likely that modeled risks
can be mitigated with only minor limits on applications. Such limits can inform best management
practices for application of FGD gypsum; however, the limits identified in this evaluation are
intended to be informative and not prescriptive. States and others knowledge of actual application
practices and local environmental conditions should make the decision about the appropriateness
of any limits on use.
7.2.1. Constituent Concentrations
Identified risks from FGD gypsum might be managed through limits on the concentrations allowed
in FGD gypsum. However, the total mass of a constituent is not always a reliable indication of how
much can readily leach out. Therefore, EPA plotted the total and leachable contents of each sample
for which both were available to better understand whether limits based on total content could
reliably reduce leachable content. Figure 7-6 presents the results of this comparison. The graph
contains data for washed and unwashed samples and are intended to present general relationships,
rather than representative distributions. Consideration of washed and unwashed data separately
did not substantially alter the relationship.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
50
Selenium
o
00.••••¦'
2
10 20 30 40 50
Bulk Content (mg/kg)
Figure 7-6: Relationship between bulk and teachable content
There is a clear relationship between the amount of selenium in FGD gypsum and the amount
available to be leached out. In many cases, almost all of the mass in the FGD gypsum is leachable,
though it may be released gradually over time. Based on this relationship, limits on selenium
concentration may be one method to control releases. Therefore, EPA filtered the full-scale model,
controlling for bulk selenium concentration in the applied FGD gypsum, to better understand the
potential effect of such limits on national risks. This review indicates that application of FGD
gypsum containing less than 25 mg/kg selenium would not pose any concerns to human health or
the environment when applied at agronomically relevant rates. This identified concentration
corresponds to the 90th percentile of all model runs. As a result, the vast majority of FGD gypsum
generated in the United States would not warrant any limits when applied in agronomically
relevant rates.
7.2.2. Application Rate
There is a clear and direct relationship between the mass of FGD gypsum applied and the amount
of selenium that can be released from a watershed. Thus, identified risks may be managed through
limits on the rate at which the FGD gypsum is applied to fields. EPA filtered the full-scale model
results, controlling for application rate, to better understand the potential effect of such limits on
national risks. This review indicates that an average application around 1 ton/acre would not pose
any concerns to human health or the environment, even if widely applied across a watershed.
This identified rate is greater than the time-averaged, high-end rates for several uses: 1.7 tons/acre
every 2 years for nutrient application, 10 tons/acre every 10 years for sodic soils, and 11 tons/acre
every 10 years for aluminum toxicity. Although use for sodic soils and aluminum toxicity tend to
have higher individual-year application rates, the total area of over which FGD gypsum is expected
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Section 7: Uncertainty and Sensitivity Analyses
DO
E
o
u
_0)
-Q
TO
ro
40
30
20
10
-------
to be applied be far smaller. This is because these two uses are expected to target to specific problem
areas, rather than entire fields or watersheds. Based on this sensitivity analysis, it is expected that
these three uses of FGD gypsum would not warrant any limits when applied in agronomically
relevant rates. Use to prevent phosphorus runoff and improve drainage might pose concerns where
FGD gypsum is annually applied across a watershed. In these instances, limiting applications to an
average of 1 ton/acre can ensure that these uses do not pose concern to either human health or the
environment.
7.2.3. Application Area
There is a clear and direct relationship between the proportion of a watershed over which FGD
gypsum is applied and the amount of selenium that can be released. Thus, identified risks may be
managed through limits placed on the area of application. To better understand the potential effect
of such limits on national risks, EPA filtered the full-scale model results to control for application
area as a proportion of the total watershed. EPA controlled for the proportion because the size of
each watershed can vary considerably. The same field area in a larger watershed would provide
greater opportunity for mixing and dilution with precipitation that falls outside the fields. Thus,
the proportion of the watershed with FGD gypsum applied provides a more consistent frame of
reference for comparison. This review found that application on 40% or less of the drainage area
for a headwater stream posed no concerns to human or ecological receptors. For comparison, the
50th percentile of modeled headwater drainage areas is about 1,300 acres, while the 90th percentile
is about 5,800 acres.
The identified proportion of 40% corresponds to the 90th percentile of all model runs. This is both
because the area dedicated to agricultural fields in many watersheds is already less than 40% of the
total land area and because the model allowed the field area with FGD gypsum applied in each run
to vary probabilistically. The maximum proportion in any individual model run was 93% of the
entire watershed. Based on this sensitivity analysis, it is anticipated the use of FGD gypsum will
pose no concerns in many regions of the country. Although it may ultimately not be practical to
implement limits based on application area because there can be many land owners within a single
watershed that would need to coordinate, the identified limit of 40% can still provide one means
to understand where widespread application of FGD gypsum might warrant further review.
7.2.4. Regional Variation
Potential risks associated with the use of FGD gypsum may differ across the country as a result of
local conditions (e.g., precipitation rate, amount of farmland). Thus, a single set of management
standards may not be equally appropriate for each region. To understand how modeled risks vary
based on geography, EPA aggregated model results for ecological exposure to selenium at a HUC4
level. As part of this analysis, EPA included all fields within the economic feasibility zone. Figure
7-7 depicts how the 90th percentile of modeled long-term risks vary geographically. Each shaded
region represents an individual HUC4, some of which extend outside the borders of the country.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 7: Uncertainty and Sensitivity Analyses
-------
Regions that are not fully shaded reflect those that fall partially outside the economic feasibility
zone modeled in this evaluation.
K
Figure 7-7: Geographic variability of modeled risks by individual HUC4.
Hazard Quotient
] Not Modeled
0-0 09
^ 0 10-0 49
0 50-099
^ 1.0-4.9
5 0 - 9 9
~ 10 0-249
[] 25.0 - 49 9
| 500 - 999
¦ 100-250
These results show that risks vary across the country. These findings generally align with previous
sensitivity analysis on application area. Risks are lowest in the South and along both coasts where
there is less cumulative field area for application of FGD gypsum. Risks increase somewhat in the
Midwest where there is more agricultural land. Risks are more variable further west. Pockets of
lower risk are attributed to the limited number of fields in these regions, which result in few model
runs for those areas, combined with an arid environment where there is effectively no modeled
infiltration or runoff. Pockets of disproportionately high risk are attributed to regions of low
precipitation that could result in extremely high modeled concentrations for constituents like
selenium with high solubility. The magnitude of risk in these parts of the country are expected be
exaggerated, but caution may still be warranted because these regions tend to be those with greater
tendency toward existing issues with water bodies impaired by selenium. Although modeled risks
can be disproportionately high in the west, it is clear that the small number fields and associated
model runs did not inappropriately skew high-end risks on a national scale. Based on this
sensitivity analysis, it is anticipated the use of FGD gypsum will pose no concerns in many regions
of the country.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
7.3. Summary
EPA applied the best information available to characterize the most likely management and release
scenarios with an aim to minimize the influence of uncertainties on a national scale. As a result,
while there is potential for management practices and associated releases at individual farm fields
to diverge from those modeled, it is expected that the broader potential for releases is adequately
captured within the probabilistic model. Yet sources of uncertainty inevitably remain that may
individually underestimate or overestimate risks to some degree. In this section, EPA analyzed
these cumulative effect of these uncertainties to better understand the magnitude of each and the
potential to impact the model results reported in Section 6 (Risk Modeling).
EPA identified two instances where uncertainties resulted in an overestimation of risk substantial
enough to impact the conclusions drawn from the full-scale model. Both instances involved
accumulation of constituent mass in soil. EPA determined that the full-scale model overestimated
risks from chromium and thallium accumulation in soil:
¦ The full-scale model identified potential risks to ecological receptors from direct contact with
and ingestion of chromium in soil. This analysis assumed that all of the applied chromium mass
was present in the hexavalent oxidation state. This assumption was made because EPA did not
identify any data on the speciation of chromium or any other constituent in FGD gypsum.
However, based on a review of the literature, EPA concluded that hexavalent chromium
applied to agricultural soils will tend to convert to the trivalent state over time. Therefore, it is
unlikely that hexavalent chromium will accumulate to the concentrations modeled.
¦ The full-scale model identified potential risks to human health from ingestion of thallium that
had accumulated in both milk and beef from livestock. However, a comparison of modeled
thallium concentrations with measurements of surface soils across the country found the
accumulated mass to be a small fraction of the existing mass in background soil. Were this
accurate, over 90% of background soils would already pose greater risk. The modeled risks are
more likely to be driven by the compounding uncertainty of successive accumulation from
gypsum-amended soil into the forage or feed and then into the livestock.
Based on these considerations and fact that both constituents had low exceedances of health-based
criteria after nearly 100 years of application, EPA concluded that the accumulation of chromium
and thallium in soil pose no short- or long-term risks to either human or ecological receptors. With
elimination of these two constituents, all potential risks from exposure to soil fall below levels of
concern. Therefore, EPA concludes that accumulation of FGD gypsum in agricultural soils does
not warrant further evaluation.
EPA identified a number of uncertainties for releases to surface water that have the potential to
either overestimate or underestimate risk, but the data available to characterize these uncertainties
are often limited. A qualitative review of these uncertainties found many are likely to be relatively
Beneficial Use Evaluation of FGD Gypsum in Agriculture
7-26
Section 7: Uncertainty and Sensitivity Analyses
-------
minor or have no effect on conclusions (e.g., overestimation of risk for constituents found to pose
no concern). The single greatest uncertainty identified how widely or frequently this material will
applied in the absence of any restrictions. Therefore, EPA assumed that gypsum could be applied
as frequently as every year on all agricultural fields where it might provide a benefit. As a result,
modeled results will overestimate risk to the extent that FGD gypsum is not spread as widely or as
frequently due to economic or other practical considerations. Yet it cannot be ruled out that such
widespread applications are possible. Therefore, EPA determined that it was most appropriate and
protective to draw conclusions about the full range of potential uses based on the results of the
full-scale model.
EPA conducted sensitivity analyses to review the results of the full-scale model reported in Section
6 (Risk Analysis) and identify any sensitive model inputs that could be used to reduce the modeled
risks to below levels of concern. These analyses confirm that modeled risks are driven by a small
subset of model runs that reflect wide applications at high concentrations, rates, and frequencies.
However, the vast majority of modeled application scenarios pose no concerns to human health or
the environment. Indeed, there are a number of uses and regions of the country for there is likely
no potential for concern at all. In instances where FGD gypsum might be applied at the highest
rates and frequencies over a majority of a given watershed, it is possible to mitigate the potential
for risk with minor limits on the either application rate and area or the concentration of selenium
in the FGD gypsum. Based on these results, there is a high degree of confidence in the principal
finding of the full-scale analysis that application of FGD gypsum will not pose any concerns in the
majority of application scenarios.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 7: Uncertainty and Sensitivity Analyses
-------
8. Final Summary and Conclusions
The Methodology for Evaluating Beneficial Uses of Industrial Non-Hazardous Secondary Materials
(U.S. EPA, 2016a) and the Beneficial Use Compendium: A Collection of Resources and Tools to
Support Beneficial Use Evaluations (U.S. EPA, 2016b) provide both an analytical framework and
resources that can aid states, tribes, local governments and others with evaluations of the potential
environmental impacts associated with the beneficial use of industrial materials. This current
evaluation provides an example of these resources applied to an unencapsulated beneficial use:
FGD gypsum used as a replacement for mined gypsum on agricultural fields. This section provides
a summary of the full evaluation and the conclusions that can be drawn based on a quantitative
and qualitative review of all available information.
8.1. Evaluation Summary
EPA sequentially applied each step of the analytical framework, culminating is a national-scale
probabilistic model of environmental fate and transport. Following completion of each step, EPA
reviewed the findings and identified any individual constituents or exposure pathways that posed
no concerns. These constituents and pathways were removed from further consideration before
proceeding on to the next step to help streamline subsequent steps. The following text provides a
summary of each step of this evaluation and the associated findings. The purpose of this summary
is to highlight how each step contributed to the characterization of potential environmental
impacts that might result from use of this industrial material. The first section of the document
provides a general introduction to the evaluation. The subsequent sections are summarized below:
Section 2 (Planning and Scoping): Prior to any quantitative analysis, EPA reviewed all available
information about FGD gypsum use and composition to define the scope of the evaluation. Every
use of gypsum considered in this evaluation is either for application directly on the ground surface
or mixed together with surficial soils. As a result, every use may result in the same types of releases
to the environment. Thus, a single conceptual model was used to represent all uses. This conceptual
model formed the basis for all subsequent data collection and modeling efforts.
Section 3 (Existing Evaluations): EPA reviewed the available literature and identified two sources
with information relevant to the current evaluation. A review of the data quality in both found
these sources to be an appropriate basis for conclusions about FGD gypsum used in agricultural
applications. Based on the information provided by these existing evaluations, EPA concluded that
potential exposures from windblown dust fall below levels of concern and that potential exposures
from radiation are comparable to those from mined gypsum. Therefore, these pathways were not
carried forward for further evaluation.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 8: Final Conclusions
-------
Section 4 (Comparison with Analogous Product): FGD gypsum was compared to mined gypsum
to determine whether there is potential for greater releases from FGD gypsum than from analogous
materials that would otherwise be used. Based on these comparisons, EPA concluded that many of
the constituents in FGD gypsum are comparable to those in mined gypsum. This is because a major
source of constituent mass in FGD gypsum is the limestone used in wet scrubbers. Limestone is a
naturally occurring mineral similar to gypsum and so it is reasonable that the bulk content of many
elements would be similar. The constituents found to be higher in FGD gypsum tend to be those
that are most volatile during coal combustion. This allows the constituents to pass through
particulate control devices and enter the wet scrubber, where they are captured along with sulfur
dioxide. For releases to leachate, a comparison could not be conducted due to a lack of leachate
data for mined gypsum. Therefore, all constituents were carried forward for this pathway. For
releases to air, the comparison found the potential for greater volatilization of mercury from FGD
gypsum. Therefore, this pathway was also retained for further evaluation.
Section 5 (Screening Analysis): EPA conducted a streamlined analysis that used a combination of
high-end waste characterization data and simplifying assumptions to capture an upper bound of
possible exposures. Exposures found to be below levels of concern based on this screening were
eliminated from further consideration with a high degree of confidence. Based on the results of
this analysis, all exposures from releases to air were found to be below levels of concern. Various
constituents were retained for further evaluation of exposures from releases to soil (i.e., chromium,
mercury, selenium, thallium); ground water (i.e., antimony, arsenic, chromium, thallium), surface
water (i.e., arsenic, cadmium, chromium, iron, lead, mercury, manganese, selenium, thallium) and
sediment (i.e., antimony, cadmium, chromium, lead, manganese, mercury, nickel, zinc).
Section 6 (Risk Modeling): EPA conducted a more refined, full-scale analysis to better incorporate
the variability of constituent characteristics, environmental setting and receptor behavior that can
impact constituent release, transport and exposure. The probabilistic results provide a best estimate
of risks that may result from the use of FGD gypsum in agriculture. No concerns were identified
from releases to ground water or sediment. High-end risks (i.e., 90th percentile) were identified
for chromium for ecological receptors in soil, selenium for ecological receptors in surface water
and human ingestion of fish, and thallium for human ingestion of beef and milk. Corresponding
median scenarios (i.e., 50th percentile) are all one or more orders of magnitude below levels of
concern for each of these exposure pathways.
Section 7 (Uncertainty and Sensitivity Analyses): EPA reviewed the major sources of uncertainty
known to be associated with both the data and modeling approach used in this evaluation. The
purpose of this review was to determine whether the magnitude of these uncertainties is great
enough to alter the conclusions that would otherwise be drawn from the full-scale model. EPA did
identify two instances where the full-scale model is known to overestimate risk to a degree that
could affect the conclusions drawn from the full-scale model. The available data and assumptions
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 8: Final Conclusions
-------
used to model chromium and thallium accumulation in the soil are expected to substantially
overestimate risk for these constituents. Based on this review, EPA concluded that accumulation
of chromium in soil and thallium in beef/milk pose no concerns. Therefore, the only remaining
risks identified in the full-scale analysis are from selenium accumulation in surface water.
EPA conducted additional sensitivity analyses to better understand what model inputs drive these
remaining potential risks and might inform any limits on applications that could ensure they do
not occur. These analyses show that modeled risks are driven by a relatively small subset of model
runs that reflect wide applications at the highest rates and frequencies. However, the vast majority
of modeled application scenarios pose no concerns to human health or the environment. In
instances where applications are expected to be applied at the highest rates and frequencies over a
majority of a given watershed, it is possible to mitigate the potential risks identified in this
evaluation with minor limits on application practices.
8.2. Conclusions
Based on available data and modeling documented in this evaluation, EPA reached the following
conclusions about the use of FGD gypsum as an agricultural amendment:
¦ The limestone slurry used in wet scrubbers is a major source of constituent mass to FGD gypsum.
Both limestone and mined gypsum are naturally occuring and, as a result, many constituents
are present at comparable levels in both FGD and mined gypsum. The constituents found at
higher levels in FGD gypsum tend to be those that are most volatile during coal combustion.
These volatile constituents are able pass more easily through particulate control devices and are
instead captured in wet scrubbers along with sulfur dioxide. Little leachate data were identified
for mined gypsum and so similar comparisons could not be made for that release pathway. It is
possible the leaching behavior of some constituents may also be comparable between the two
materials, particularly where the majority of the constituent mass originates from the limestone
slurry. However, the constituents with the highest modeled risks were also those that are most
volatile and most highly concentrated in FGD gypsum. Therefore, it is unlikely that this lack of
data would alter the conclusions of this evaluation.
¦ Washing of FGD gypsum was found to reliably reduce the bulk content of boron, chloride and
manganese. Data on the bulk content (i.e., mg/kg) for bromide was not available for comparison;
however, it is expected that similar reductions would be found for this constituent. Washing
was found to reliably reduce leachate concentrations (i.e., mg/L) of antimony, bromide, boron,
cadmium, chloride, cobalt, lead, manganese, molybdenum, nickel, selenium and thallium. The
magnitude of reduction varies among constituents. For antimony, bromide, lead, and thallium,
washing reduced leachate concentrations in all samples to below detection limits. Reasons for
the absence of reduction in other constituents may vary and include that a given constituent
has low solubility under the conditions present during washing, a substantial fraction of the
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 8: Final Conclusions
-------
overall constituent mass is present in a relatively recalcitrant form, and/or that a real reduction
exists but the magnitude is too small to reliably distinguish from natural sample variability.
¦ Application of FGD gypsum did not result in accumulation of constituent mass in soil, crops,
livestock, air, or ground water at levels that pose concern for human health or the environment
under any of the application scenarios evaluated.
¦ High-end risks from selenium were identified for ecological receptors in smaller headwater
streams and for human fishers who consume high quantities of fish caught from larger mainstem
streams. These risks were found to be only slightly above relevant health benchmarks. This
indicates that the vast majority of modeled application scenarios for FGD gypsum will pose no
concerns to human health or the environment. In areas where FGD gypsum could be applied
both widely and at high rates, modeled risks can be mitigated through minor limits on the
application practices, such as those identified in sensitivity analyses.
¦ This evaluation could not consider impacts to more smaller and more static water bodies, such
as farm fish ponds, due to a lack of information about the associated locations, volumes, drainage
areas, or turnover rates. There is an unknown potential for greater accumulation of constituent
mass in these water bodies due to the greater residence time of water in these systems. Thus,
further consideration may be warranted prior to substantial applications in drainage area of such
water bodies.
¦ This evaluation focused on potential environmental impacts unique to FGD gypsum. However,
risks can also result from the mismanagement of other types of gypsum. For example, the high
concentrations of sulfate in gypsum may pose risk to cattle that are allowed to graze in fields
too soon after application. USDA has developed guidelines to address the risks that are shared
among all types of gypsum in Conservation Practice Standard: Amending Soil Properties with
Gypsum Products {USDA, 2015b). These guidelines identify agronomically relevant application
rates intended to ensure the management of agricultural amendments remains protective of
human health and the environment.
It has previously been established that there are a number of compelling benefits associated with
the use of FGD gypsum in agriculture, such as a providing key nutrients to crops and limiting
phosphorus runoff to nearby water bodies. These uses may also provide benefits outside of the
fields, such as helping to reduce greenhouse gas emissions from mining, diverting these waste from
CCR landfills, and providing cost saving to farmers. Based on these model results, application of
FGD gypsum to fields at the agronomically relevant rates considered in this evaluation can provide
benefits while remaining protective of human health and the environment.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 8: Final Conclusions
-------
9. References
ACAA (American Coal Ash Association). 2021. "2020 Coal Combustion Product (CCP) Production
and Use Survey Report." November.
Ahuja, L.R., J.D. Ross and O.R. Lehman. 1981. "A Theoretical Analysis of Interflow of Water
through Surface Soil Horizons with Implications for Movement of Chemicals in Field Runoff."
Water Resources Research. 17:65-72.
Alvarez-Ayuso, E., and X. Querol. 2007. "Stabilization of FGD gypsum for its Disposal in Landfills
Using Amorphous Aluminum Oxide as a Fluoride Retention Additive." Chemosphere. 69:295-
302.
Alvarez-Ayuso, E., X. Querol, and A. Tomas. 2006. "Environmental Impact of a Coal Combustion-
Desulphurization Plant: Abatement Capacity of Desulphurization Process and Environmental
Characterization of Combustion By-Products." Chemosphere 65:2009-2017.
Anderson, D.L., O.H. Tuovinen, A. Faber and I. Ostrokowski. 1995. "Use of Soil Amendments to
Reduce Soluble Phosphorus in Dairy Soils." Ecological Engineering. 5:229-246.
ASTSWMO (Association of State and Territorial Solid Waste Management Officials). 2007.
"ASTSWMO 2006 Beneficial Use Survey Report." Washington, DC. November.
Ates, N., M. Kitis and U. Yetis. 2007. "Formation of Chlorination By-Products in Waters with Low
SUVA-Correlations with SUVA and Differential UV Spectroscopy." Water Research
41(18):4139-4148.
Banuelos, G.S., D.W. Meek and G.J. Hoffman. 1990. "The Influence of Selenium, Salinity, and
Boron on Selenium Uptake in Wild Mustard." Plant and Soil. 127:201-206.
Banuelos, G.S., and H. F. Mayland. 2000. "Absorption and Distribution of Selenium in Animals
Consuming Canola Grown for Selenium Phytoremediation." Ecotoxicology and
Environmental Safety. 46:322-328.
Bell, P.F., D.R. Parker, and A.L. Page. 1992. "Division S-4 - Soil Fertility and Plant Nutrition." Soil
Science Society of America Journal. 56:1818-1824.
Birkner, J.H. 1978. "Selenium in Aquatic Organisms from Seleniferous habitats." PhD Thesis.
Colorado State University, Fort Collins, CO.
Bjorklund, G., M. Dadar, J. Mutter and J. Aaseth. 2017. "The Toxicology of Mercury: Current
Research and Emerging Trends." Environmental Research. 159:545-554.
Bledsoe, R.W., C.L. Comar and H.C. Harris. 1949. "Absorption of Radioactive Calcium by the
Peanut Fruit." Science. 109:329-330.
Brauer, D., G.E. Aiken, D.H. Pote, S.J. Livingston, L.D. Norton, T.R. Way and J.H. Edwards. 2005.
"Amendments Effects on Soil Test Phosphorus." Journal of Environmental Quality. 34:1682-
1686.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 9: References
-------
Briggs, C.W., R. Fine, M. Markee and M.S. Gustin. 2014. "Investigation of the Potential for
Mercury Release from Flue Gas Desulfurization Solids Applied as an Agricultural
Amendment." Journal of Environmental Quality. 43(l):253-62.
Brix, K.V., J.S. Volosin, W.J. Adams, R.J. Reash, R.C. Carlton and D.O. Mclntyre. 2001. "Effects of
Sulfate on the Acute Toxicity of Selenate to Freshwater Organisms." Environmental
Toxicology and Chemistry. 5:1037-1045.
Brose, D.A. and B.R. James. 2010. "Oxidation-Reduction Transformations of Chromium in Aerobic
Soils and the Role of Electron-Shuttling Quinones." Environmental Science and Technology
44:9438-9444.
Chaney, R.L., R.D. Reeves, LA. Baklanov, T. Centofanti, C.L. Broadhurst, A.J.M. Baker, J.S. Angle,
A. van der Ent and R.J. Roseberg. 2014. "Phytoremediation and Phytomining: Using Plants to
Remediate Contaminated or Mineralized Environments." Chapter 15, pp. 365-391. In Plant
Ecology and Evolution in Harsh Environments. R. Rajakaruna, R.S. Boyd and T. Harris (Eds.)
Nova Science Publishers, NY.
Chang, E.E., Y.P. Lin and P.C. Chiang. 2001. "Effects of Bromide on the Formation of THMs and
HAAs." Chemosphere 43(8):1029-1034.
Chen, L., D. Kost, Y. Tian, X. Guo, D. Watts, D. Norton, R.P. Wolkowski and W.A. Dick. 2014.
"Effects of Gypsum on Trace Metals in Soil and Earthworms." Journal of Environmental
Quality. 43:263-272.
Chen, L., W.A. Dick and S. Nelson. 2005. "Soil Fertility: Flue Gas Desulfurization Products as
Sulfur Sources for Alfalfa and Soybean." Agronomy Journal. 97:265-271.
Cheng, C., P. Hack, P. Chu, Y. Chang, T. Lin, C. Ko, P. Chiang, C. He, Y. Lai and W. Pan. 2009.
"Partitioning of Mercury, Arsenic, Selenium, Boron, and Chloride in a Full-Scale Coal
Combustion Process Equipped with selective catalytic reduction, electrostatic precipitation,
and Flue Gas Desulfurization Systems." Energy and Fuels. 23(10):4805-4816.
Cheng, C., Y. Chang, K.R. Sistani, Y. Wang, W. Lu, C. Lin, J. Dong, C. Hu and W. Pan. 2012.
"Mercury Emission and Plant Uptake of Trace Elements during Early Stage of Soil Amendment
Using Flue Gas Desulfurization Materials." Journal of Air and Waste Management Association.
62(2):139-150.
Clarke, L.B. and L.L. Sloss. 1992. "Trace Elements - Emissions from Coal Combustion and
Gasification." IEACR/49. IEA Coal Research: London, UK.
CSU-E. (Colorado State University-Extension). 2012. "Managing Sodic Soils." Fact Sheet No. 0504.
Prepared by J.G. Davis, R.M. Waskom and T.A. Bauder. Fort Collins, CO.
Cowman, G.A. and P.C. Singer. 1996. "Effect of Bromide Ion on Haloacetic Acid Speciation
Resulting from Chlorination and Chloramination of Aquatic Humic Substances."
Environmental Science and Technology 30(1): 16-24.
Cox, F.R., G.A. Sullivan and C.K. Martin. 1976. "Effect of Calcium and Irrigation Treatments on
Peanut Yield, Grade and Seed Quality." Peanut Science 3:81-85.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 9: References
-------
Dao, T.H. 1999. "Coamendments to Modify Phosphorus Extractability and Nitrogen/Phosphorus
Ratio in Feedlot Manure and Composted Manure." Journal of Environmental Quality 28:1114-
1121.
Darling, D.A. 1957. "The Kolmogorov-Smirnov, Cramer-von Mises Tests." The Annals of
Mathematical Statistics. 28:823-838.
Duong, H.A., M. Berg, M.H. Hoang, H.V. Pham, H. Gallard, W. Giger and U. von Gunten. 2003.
"Trihalomethane Formation by Chlorination of Ammonium- and Bromide-Containing
Groundwater in Water Supplies of Hanoi, Vietnam." Water Research. 37(13):3242-52.
Eckley, C.S., M. Gustin, M.B Miller and F. Marsik. 2011. "Scaling Non-Point-Source Mercury
Emissions from Two Active Industrial Gold Mines: Influential Variables and Annual Emission
Estimates." Environmental Science and Technology. 45(2):392-399.
Eckley, C.S., P. Blanchard, D. McLennan, R. Mintz and M. Sekela. 2015. "Soil-Air Mercury Flux
Near a Large Industrial Emission Source Before and After Closure (Flin Flon, Manitoba,
Canada)." Environmental Science and Technology. 49(16):9750-9757.
Eckley, C.S., M.T. Tate, C. Lin, M. Gustin, S. Dent, C. Eagles-Smith, M.A. Lutz, K.P. Wickland, B.
Wang, J.E. Gray, G.C. Edwards, D.P. Krabbenhoft and D.B. Smith. 2016. "Surface-Air Mercury
Fluxes across Western North America: A Synthesis of Spatial Trends and Controlling
Variables." Science of the Total Environment. 568:651-665.
EIA (Energy Information Administration). 2017. "Electricity: Analysis and Projections." Form
EIA-923. Washington, DC.
El Afifi, E.M., M.A. Hilal, S.M. Khalifa and H.F. Aly. 2006. "Evaluation of U, Th, K and Emanated
Radon in Some NORM and TENORM Samples." Radiation Measurement. 41(5): 627-633.
Endale, D.M., H.H. Schomberg, D.S. Fisher, D.H. Franklin and M.B. Jenkins. 2014. "Flue Gas
Desulfurization Gypsum: Implication for Runoff and Nutrient Losses Associated with Broiler
Litter Use on Pastures on Ultisols." Journal of Environmental Quality 43:281-289.
Engmann, S., and D. Cousineau. 2011. "Comparing Distributions: the Two-Sample Anderson-
Darling Test as an Alternative to the Kolmogorov-Smirnov Test." Journal of Applied
Quantitative Methods. 6(3): 1-17.
EPRI (Electric Power Research Institute). 2006. "A Review of Agricultural and Other Land
Application Uses of Flue Gas Desulfurization Products." Document No. 1010385. Prepared by
W. Dick, D. Kost and N. Nakano of Ohio State University. Palo Alto, CA.
EPRI. 2008a. "Impact of Air Emissions Controls on Coal Combustion Products." Document No.
1015544. Prepared by G. Blythe and J. Paradis of URS Corporation for EPRI and Ameren
Corporation. Palo Alto, CA. October.
EPRI. 2008b. "Flue Gas Desulfurization Gypsum Agricultural Network: 2008 Progress Report."
Prepared by W. Dick and D. Kost of Ohio State University. Palo Alto, CA. December.
EPRI. 2011. "Selenium Speciation and Management in Wet FGD Systems." Document No.
1022160. Prepared by K. Searcy of Trimeric Corporation, G. Blythe and J. Paradis of URS
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 9: References
-------
Corporation, and D. Wallschager of Trent University for EPRI and the U.S. DOE. Palo Alto,
CA. December.
EPRI 2012. "Flue Gas Desulfurization Gypsum Agricultural Network: Ohio Sites 1 (Mixed Hay)
and 2 (Corn)." Prepared by W. Dick and D. Kost of Ohio State University and D. Smeal of New
Mexico State University. Palo Alto, CA. October.
EPRI. 2013. "Flue Gas Desulfurization Gypsum Agricultural Network: Indiana Kingman Research
Station (Corn and Soybeans)." Prepared by W. Dick and D. Kost of Ohio State University and
L.D. Norton of USDA. Palo Alto, CA. October.
Fernandez-Gonzalez, A., A. Andara, J.M. Alia, M. Prieto. 2006. "Miscibility in the CaS04d2H20-
CaSe04d2H20 System: Implications for the Crystallization and Dehydration Behaviour."
Chemical Geology. 225:256-265.
Freeze, R.A. and J.A. Cherry. 1979. "Groundwater." Prentice-Hall: Englewood Cliffs, NJ.
Frey, H.C., and Y. Zhao. 2004. "Quantification of Variability and Uncertainty for Air Toxic
Emission Inventories with Censored Emission Factor Data." Environmental Science and
Technology. 38(22):6094-6100.
Genck, W.J., D.S. Dickey, F.A. Baczek, D.C. Bedell, K. Brown, W. Chen, D.E. Ellis, P. Harriott, W.
Li, J.K. McGillicuddy, T.P McNulty, J.Y. Oldshue, F. Schoenbrunn, J.C. Smith, D.C. Taylor and
D.R. Wells. 2008. "Perry's Chemical Engineers' Handbook 8th Edition." McGraw-Hill. New
York, NY.
Gustin, M. and K. Ladwig. 2010. "Supporting information: Laboratory Investigation of Hg Release
from Flue Gas Desulfurization Products." Environmental Science and Technology 44:4012-
4018.
Gutberlet, H. 1984. Measurement of heavy metal removal by a Flue Gas Desulfurization Plant
Working by the Lime Scrubbing Method. Research Report No. ENV-492-D(B), Commission of
the European Communities, Luxembourg, 112 pp.
Gutberlet, H., A. Dieckmann, and S. Neuhaus. 1985. Measurement of heavy metal separation in a
wet-limestone scrubber flue gas desulfurization system. Proceedings of VGB Conference,
Kraftwerk und Umwelt 1985, pp. 95-104.
Hallock, D.L. and K.H. Garren. 1968. "Pod Breakdown, Yield, and Grade of Virginia Type Peanuts
as Affected by Ca, Mg, and K Sulfates." Agronomy Journal. 60:253-257.
Harris, R.C., J.W. Rudd, M. Amyot, C.L. Babiarz, K.G. Beaty, P.J. Blanchfield, R.A. Bodaly, B.A.
Branfireun, C.C. Gilmour, J.A. Graydon, A. Heyes, H. Hintelmann, J.P. Hurley, C.A. Kelly,
D.P. Krabbenhoft, S.E. Lindberg, R.P. Mason, M.J. Paterson, C.L. Podemski, A. Robinson, K.A.
Sandilands, G.R. Southworth, V.L. St. Louis and M.T. Tate. 2007. "Whole-Ecosystem Study
Shows Rapid Fish-Mercury Response to Changes in Mercury Deposition." Proceedings of the
National Academy of Sciences of the United States of America 104(42):16586-16591.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 9: References
-------
Heller-Grossman, L., J. Manka and M. Rebhun. 1993. "Formation and Distribution of Haloacetic
Acids, THM, and TOX in Chlorination of Bromide-rich Lake Water." Water Research 27(8):
1323-1331.
Henkels, P.J. and J.C. Gaynor. 1996. "Characterizing Synthetic Gypsum for Wallboard
Manufacture." Spring National Meeting of the American Chemical Society, Symposium on
Conversion of FOD Residues and Utility Fly Ash to Marketable Products, March 24-28. New
Orleans, LA.
HHS (United States Department of Health and Human Services). 2003. "Toxicological Profile for
Selenium." Prepared by the Agency for Toxic Substances and Disease Registry. Atlanta, GA.
September.
HHS. 2012. "Toxicological Profile for Chromium." Prepared by the Agency for Toxic Substances
and Disease Registry. Atlanta, GA. September.
Hintelmann, H., R. Harris, A. Heyes, J.P. Hurley, C.A. Kelly, D.P. Krabbenhoft, S. Lindberg,
J.W.M. Rudd, K.J. Scott and V.L. St. Louis. 2002. "Reactivity and Mobility of New and Old
Mercury Deposition in a Boreal Forest Ecosystem during the First Year of the METAALICUS
Study." Environmental Science and Technology 36:5034-5040.
Hollander, M., D.A. Wolfe, and E. Chicken. 2013. "Nonparametric Statistical Methods, 3rd
Edition." John Wiley and Sons, Inc.: New York, NY.
Hopper, J.L. and D.R. Parker. 1999. "Plant Availability of Selenite and Selenate as Influenced by
the Competing ions Phosphate and Sulfate." Plant Soil. 210:199-207.
Horn R., R. Baumgartl, R. Kayser, and S. Baasch. 1995. "Effect of Aggregate Strength on strength
and Stress Distribution In Structured Soils." In Soil Structure: Its Development and Function.
Edited by K.H. Hartge and B.A. Stewart. CRC-Lewis Publishers: Boca Raton, FL.
Howe, J.A., R.J. Florence, G. Harris, E. van Santen, J.P. Beasley, J.P. Bostick and K.B. Balkcom.
2012. "Effect of Cultivar, Irrigation, and Soil Calcium on Runner Peanut Response to Gypsum."
Agronomy Journal. 104:1312-1320.
Institute of Hydrology. 1980. "Low Flow Studies Report No.l." Wallingford, Oxon. January.
Jones, U.S. 1982. "Fertilizers and Soil Fertility, Second Edition." Reston Publishing Company:
Reston, VA.
King, K.W., M.R. Williams, W.A. Dick and G.A. LaBarge. 2016. "Decreasing Phosphorus Loss in
Tile-Drained Landscapes." Journal of Environmental Quality. 45:1722-1730.
Kost, D., L. Chen, X. Guo, Y. Tian, K. Ladwig and W. A. Dick. 2014. "Effects of Flue Gas
Desulfurization and Mined Gypsums on Soil Properties and on Hay and Corn Growth in
Eastern Ohio." Journal of Environmental Quality 43(1):312-21.
Liang, L. and P. C. Singer. 2003. "Factors Influencing the Formation and Relative Distribution of
Haloacetic Acids and Trihalomethanes in Drinking Water." Environmental Science and
Technology. 37(13):2920-2928.
Lindsay, W.L. 1979. "Chemical Equilibria in Soils." John Wiley & Sons: New York, NY.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 9: References
-------
LLL (Lawrence Livermore Laboratory). 1977. "The Radium-226 Content of Agricultural Gypsums."
UCRL-78720. Prepared by C.L. Lindeken and D.G. Coles for the U.S. Energy Research and
Development Administration under Contract No. W-7405-Eng-48. Livermore, CA. February.
LRL (Lawrence Radiation Laboratory). 1962. "Portland Cement for a Low-Background Counting
Facility." UCRL-10475. Prepared by H.A. Wollenberg and A.R. Smith for the U.S. Energy
Research and Development Administration under Contract No. W-7405-Eng-48. Berkeley,
CA. October.
Luong, T.V., C.J. Peters, and R. Perry. 1982. "Influence of Bromide and Ammonia upon the
Formation of Trihalomethanes under Water-Treatment Conditions." Environmental Science
and Technology. 16(8):473-479.
McTigue, N.E., D.A. Cornwell, K. Graf and R. Brown. 2014. "Occurrence and Consequences of
Increased Bromide in Drinking Water Sources." Journal of the American Water Works
Association. 106(11):492-508.
Meij, R., and P.T. Alderliesten. 1989. The emission of inorganic trace compounds at coal-fired
power plants equipped with wet flue-gas desulphurization plants. Proceedings of the 8th
World Clean Air Congress, The Hague, The Netherlands, 11-15 Sept. 1989. Elsevier:
Amsterdam, The Netherlands, vol. IV, pp. 303-8.
Meili, M. 1991. "The Coupling Of Mercury and Organic Material in the Biogeochemical Cycle -
Towards a Mechanistic Model for the Boreal Forest Zone." Water, Air, and Soil Pollution.
56:333-347.
Melillo, J., M.T. Richmond and G.W. Yohe (eds.). 2014. Climate Change Impacts in the United
States: The Third National Climate Assessment. U.S. Global Change Research Program.
MIDEQ_ (Michigan Department of Environmental Quality). 2007. "Aquatic Life Fact Sheet for
Fluoride." February.
Norton L.D. 1995. "Mineralogy of High Calcium/Sulfur-Containing Coal Combustion By-Products
and their Effect on Soil Surface Sealing." In Agriculture Utilization of Urban and Industrial
Byproducts. Edited by D.L. Karlen, R.J. Wright and W.O. Kemper. American Society of
Agronomy Special Publication 58. American Society of Agronomy: Madison, WI.
Norton, L.D. 2008. "Gypsum Soil Amendment as a Management Practice in Conservation Tillage
to Improve Water Quality." Journal of Soil and Water Conservation 63(2):46A-48A.
Norton, L.D., and K. Dontsova. 1998. "Use of Soil Amendments to Prevent Soil Surface Sealing and
Control Erosion." Advances in GeoEcology. 31:581-587.
Oates, K.M. and A.G. Caldwell. 1985. "Use of By-Product Gypsum to Alleviate Soil Acidity." Soil
Science Society of America Journal. 49:915-918.
Ogle, R.S., K.J. Maler, P. Kiffney, M.J. Williams, A. Brasher, L.A. Melton and A.W. Knight. 1988.
"Bioaccumulation of Selenium in Aquatic Ecosystems." Lake and Reservoir Management.
4(2):165-173.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Section 9: References
-------
ORNL (Oak Ridge National Laboratory). 1997. Toxicological Benchmarks for Screening
Contaminants of Potential Concern for Effects on Terrestrial Plants: 1997 Revision. ES/ER/TM-
85/R3. Prepared by R.A. Efroymson, M.E. Will, G.W. Suter, and A.C. Wooten of ORNL for
U.S. DOE under Contract No. DE-AC05-840R21400. Oak Ridge, TN. November.
OSU-E (Ohio State University-Extension). 2011. "Gypsum as an Agricultural Amendment: General
Use Guidelines." Bulletin 945. Prepared by L.C. Chen and W.A. Dick.
Oswald, C.J., A. Heyes and B.A. Branfireun. 2014. "Fate and Transport of Ambient Mercury and
Applied Mercury Isotope in Terrestrial Upland Soils: Insights from the METAALICUS
Watershed." Environmental Science and Technology 48(2): 1023-1031.
Pavan, M.A., F.T. Bingham and P.F Pratt. 1984. "Redistribution of Exchangeable Calcium,
Magnesium and Aluminum Following Lime of Gypsum Applications to a Brazilian Oxisol." Soil
Science Society of America Journal. 48:33-38.
Pekney, N.J., D. Martello, K. Schroeder and E. Granite. 2009. "Environmental Chamber
Measurements of Mercury Flux from Coal Utilization By-Products." Fuel. 88:890-897.
Pourmoghaddas, H., A.A. Stevens, R.N. Kinman, R.C. Dressman, L.A. Moore and J.C. Ireland. 1993.
"Effect of Bromide Ion on Formation of HAA5 During Chlorination." Journal of the American
Water Works Association. 85(l):82-87.
Qin, H., J. Zhu , L. Liang, M. Wang, H. Su. 2013. "The Bioavailability of Selenium and Risk
Assessment for Human Selenium Poisoning in High-Se Areas, China." Environment
International. 52:66-74
Ralston, N.V.C. and L.J Raymond. 2018. "Mercury's Neurotoxicity is Characterized by its
Disruption of Selenium Biochemistry." Biochimica et Biophysica Acta — General Subjects
1862(11):2405-2416.
Reeve, N.G. and M.E Sumner. 1972. "Amelioration of Subsoil Acidity in Natal Oxisols by Leaching
Surface Applied Amendments." Agrochemophysica 4:1-6.
Regli, S., J. Chen, M. Messner, M.S. Elovitz, F.J. Letkiewicz, R.A. Pegram, T.J. Pepping, S.D.
Richardson and J.M. Wright. 2015. "Estimating Potential Increased Bladder Cancer Risk Due
to Increased Bromide Concentrations in Sources of Disinfected Drinking Waters."
Environmental Science and Technology. 49:13094-13102.
Roper, A.R, M.G. Stabin, R.C. Delapp and D.S. Kosson. 2013. "Analysis of Naturally-Occurring
Radionuclides in Coal Combustion Fly Ash, Gypsum, and Scrubber Residue Samples." Health
Physics Society. 104(3):264-269.
Schomberg, H.H., D.M. Endale, M.B. Jenkins, R.L. Chaney and D.H. Franklin. 2018. "Metals in
Soil and Runoff from a Piedmont Hay Field Amended with Broiler Litter and Flue Gas
Desulfurization Gypsum." Journal of Environmental Quality. 47:326-335.
Shainberg, I., M.E. Sumner, W.P. Miller, M.P. W. Farina, M.A. Pavan and M.Y. Fey. 1989. "Use of
Gypsum On Soils: A Review." In Advances in Soil Science, Volume 9. Springer-Verlag Inc.
New York, NY.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Section 9: References
-------
Sharpley, A.N. 1985. "Depth of Surface Soil-Runoff Interaction as Affected by Rainfall, Soil Slope,
and Management." Soil Science Society of America. 49(4):1010-1015.
Skorupa, J.P. 1998. Selenium Poisoning of Fish and Wildlife in Nature: Lessons from Twelve Real-
World Experiences." In Environmental Chemistry of Selenium. Edited by W.T. Frankenberger
Jr. and R.A. Engberg. Published by Marcel Dekker, Inc.: New York, NY.
Skyllberg, U., P.R. Bloom, J. Qian, C.M. Lin and W.F. Bleam. 2006. "Complexation of Mercury (II)
in soil organic matter: EXAFS evidence for linear two-coordination with Reduced Sulfur
Groups." Environmental Science and Technology 40:4174-4180.
SFDTET (State and Federal Drywall Technical Evaluation Team). 2009. "Technical Evaluation of
'EPA's Analysis of Florida Drywall Samples and Review of Analytical Results from Florida
Department of Health,' from the EPA's National Air and Radiation Environmental Health
Laboratory (NAREL)." Prepared for J. Recht of the U.S. Consumer Product Safety Comission
and A.M. Viamonte Ros of the Florida Department of Health. August.
Stout, W.L., A.N. Sharpley and H.B. Pionke. 1999. Reducing soil phosphorus solubility with Coal
Combustion By-Products." Journal of Environmental Quality. 27:111-118.
Suarez, D.L. 2001. "Sodic Soil Reclamation: Modelling and Field Study." Australian Journal of Soil
Research 39:1225-1246.
Sumner, M.E., H. Shahandeh, J. Bouton and J. Hammel. 1986. "Amelioration of an Acid Soil Profile
through Deep Liming and Surface Application Of Gypsum." Soil Science Society of America
Journal 50:1254-1258.
Sumner, M.E., D.E. Radcliffe, M. McCray, E. Carter and R.L. Clark. 1990. "Gypsum as an
Ameliorant for Subsoil Hardpans." Soil Technology. 3:253-258.
Torbert, H.A., K.W. King and R.D. Harmel. 2005. "Impact of Soil Amendments on Reducing
Phosphorus Losses from Runoff in Sod." Journal of Environmental Quality. 34:1415-1421.
Torbert, H.A. and D.B. Watts. 2014. "Impact of Flue Gas Desulfurization Gypsum Application on
Water Quality in a Coastal Plain Soil." Journal of Environmental Quality. 43:273-80.
Torbert, H.A., D.B. Watts and R.L. Chaney. 2018. "Impact of Flue Gas Desulfurization Gypsum
and Manure Application on Transfer of Potentially Toxic Elements to Plants, Soil, and Runoff."
Journal of Environmental Quality. 47:865-872.
Trevisi, R., S. Risica, M. D'Alessandro, D. Paradiso and C. Nuccetelli. 2012. "Natural Radioactivity
in Building Materials in the European Union: A Database and an Estimate of Radiological
Significance." Journal of Environmental Radioactivity 105:11-20.
USDA (United States Department of Agriculture). 1954. "Diagnosis and Improvement of Saline and
Alkali Soils." Agriculture Handbook 60. Prepared by L.E. Allison, L. Bernstein, C.A. Bower,
J.W. Brown, M. Fireman, J.T. Hatcher, H.E. Hayward, G.A. Pearson, R.C. Reeve, L.A. Richards
and L.V. Wilcox of the U.S. Salinity Laboratory. Edited by L.A. Richards of the Agricultural
Research Service. Washington, DC. February.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
9-8
Section 9: References
-------
USDA. 2010. "Natural Resources Conservation Service Conservation Practice Standard: Salinity
and Sodic Soil Management." Code 610. Prepared by the USDA National Agricultural Statistics
Service. October.
USDA. 2014a. "2012 Census of Agriculture Highlights: Conservation." ACH12-6. Prepared by the
USDA National Agricultural Statistics Service. July.
USDA. 2014b. "2012 Census of Agriculture: United States Summary and State Data Volume 1:
Geographic Area Series Part 51." AC-12-A-51. Prepared by the USDA National Agricultural
Statistics Service. May.
USDA. 2015a. "Amending Soil Properties with Gypsum Products Code 333." CPS 333-1. Prepared
by the USDA Natural Resources Conservation Service. March.
USDA. 2015b. "Conservation Practice Adoption Rates Vary Widely by Crop and Region."
Economic Information Bulletin 147. Prepared by T. Wade, R. Claassen and S. Wallender of the
USDA Economic Research Service. December.
U.S. DOE (U.S. Department of Energy). 2005. "Fate of Mercury in Synthetic Gypsum Used for
Wallboard Production." Prepared by J. Marshall of USG Corporation and G.M. Blythe and M.
Richardson for URS Corporation for the DOE National Energy Technology Laboratory under
Contract No. DE-FC26-04NT42080. Pittsburgh, PA. April.
U.S. DOI (United Stated Department of the Interior). 2001. "Gypsum." Prepared by D.W. Olson of
the United States Geological Survey. In the U.S. Geological Survey Minerals Yearbook.
U.S. DOI. 2012. "Mercury and Halogens in Coal - Their Role in Determining Mercury Emissions
from Coal Combustion." Fact Sheet 2012-312. Prepared by A. Kolker of the United States
Geological Survey; J.C. Quick and C.L Senior of the Utah State Geological Survey, and H.E.
Belkin of ADA Environmental Solutions, Inc. Reston, VA. November.
U.S. DOI. 2013. "Geochemical and Mineralogical Data for Soils of the Conterminous United
States." USGS Data Series 801. Prepared by D.B. Smith, W.F. Cannon, L.G. Woodruff, F. Solano,
J.E. Kilburn and D.L. Frey of the United States Geological Survey. Reston, VA.
U.S. DOI. 2015. "Minerals Commodity Summaries 2015." Prepared by the United States Geological
Survey. Reston, VA. January.
U.S. EPA. 1989. "Risk Assessment Guidance for Superfund. Volume I: Human Health Evaluation
Manual (Part A) Interim Final." EPA/540/1-89/002. Prepared by the EPA Office of Emergency
and Remedial Response. Washington, DC. December.
U.S. EPA. 1991. "EPA Region 3 Guidance on Handling Chemical Concentration Data Near the
Detection Limit in Risk Assessments." EPA/903/8-91/001. Prepared by the EPA Region 3.
Philadelphia, PA.
U.S. EPA. 1992a. "Potential Uses of Phosphogypsum and Associated Risks: Background
Information Document." EPA 402-R92-002. Prepared by S. Cohen and Associates, Inc. for the
Office of Air and Radiation under Contract No. 68D90170. Washington D.C., May.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
9-9
Section 9: References
-------
U.S. EPA. 1992b. Technical Support Document for Land Application of Sewage Sludge. Volumes I
and II. EPA 822/R-93-001a,b. Prepared by Eastern Research Group for the EPA Office of Water
under Contract No. 68D90170. Washington D.C., November.
U.S. EPA. 1994. "The Hydrological Evaluation of Landfill Performance (HELP) Model:
Engineering Documentation for Version 3." EPA/600/R-94/168a. Prepared by P.R. Schroeder,
C.M. Lloyd, and P.A. Zappi of the U.S. Army Corp of Engineers and N.M. Aziz of Clemson
University for the EPA Risk Reduction Engineering Laboratory under Interagency Agreement
No. DW21931425. Cincinnati, OH. September.
U.S. EPA. 1996a. "Draft Technical Support Document for HWC MACT Standards Volume VII:
Miscellaneous Technical Issues." Prepared by the EPA Office of Solid Waste. Washington, DC.
February.
U.S. EPA. 1998. "Supplemental Report to Congress on Remaining Wastes from Fossil Fuel
Combustion Technical Background Document: Beneficial Use of Fossil Fuel Combustion (FCC)
Wastes." Prepared by the EPA Office of Solid Waste. Washington, DC. October.
U.S. EPA. 1999. MINTEQA2/PRODEFA2, A Geochemical Assessment Model for Environmental
Systems: User Manual Supplement for Version 4.0. Prepared by HydroGeoLogic, Inc. and
Allison Geoscience Consultants, Inc. for the Office of Research and Development. Athens, GA.
September.
U.S. EPA. 2000. "Supplementary Guidance for Conducting Health Risk Assessment of Chemical
Mixtures." EPA/630/R-00/002. Prepared by the EPA Risk Assessment Forum, Washington, DC.
August. Washington, DC. August.
U.S. EPA. 2001. Revisions in Input Data and Modeling Procedures for Using MINTEQA2 in
Estimating Metal Partition Coefficients. Prepared by Allison Geoscience Consultants, Inc. for
the EPA Office of Solid Waste. Washington, DC.
U.S. EPA. 2002a. "Drinking Water from Household Wells." EPA 816-K-02-003. Prepared by the
EPA Office of Water. Washington, DC. January.
U.S. EPA. 2002b. "The Occurrence of Disinfection By-Products (DBPs) of Health Concern in
Drinking Water: Results of a Nationwide DBP Occurrence Study." EPA/600/R-02/068.
Prepared by H.S. Weinberg of the University of North Carolina at Chapel Hill, S.W. Krasner
of the Metropolitan Water District of Southern California, and S.D. Richardson and A.D.
Thurston Jr. of the EPA Office of Research and Development. Athens, GA. September.
U.S. EPA. 2003a. "Summary of General Assessment Factors for Evaluating the Quality of Scientific
and Technical Information." EPA 100/B-03/001. Prepared by the EPA Science Policy Council.
Washington, DC. June.
U.S. EPA. 2003b. AERMOD: Latest Features and Evaluation Results. EPA-454/R-03-003. Prepared
by the EPA Office of Air Quality Planning and Standards. Research Triangle Park, NC. June.
U.S. EPA. 2003c. "EPA's Composite Model for Leachate Migration with Transformation Products
(EPACMTP). Parameters/Data Background Document." EPA53-R-03-003. Prepared by
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
9-10
Section 9: References
-------
HydroGeoLogic, Inc. and RMC, Inc. for the EPA Office of Solid Waste under Contract Number
68-W-01-004. Washington, DC. April.
U.S. EPA. 2003d. "EPA's Composite Model for Leachate Migration with Transformation Products
(EPACMTP). Technical Background Document." EPA53-R-03-002. Prepared by
HydroGeoLogic, Inc. and RMC, Inc. for the EPA Office of Solid Waste under Contract Number
68-W-01-004. Washington, DC. April.
U.S. EPA. 2003e. "Multimedia, Multipathway, and Multireceptor Risk Assessment (3MRA)
Modeling System. Volume I: Modeling System and Science. SAB Review Draft." EPA530-D-
03-001a. Prepared by the EPA Office of Research and Development and Office of Solid Waste.
July.
U.S. EPA. 2004. "Draft Aquatic Life Water Quality Criteria for Selenium - 2004." EPA-822-D-04-
001. Prepared by the EPA Office of Water. Washington, DC. November.
U.S. EPA. 2005. "Human Health Risk Assessment Protocol for Hazardous Waste Combustion
Facilities." EPA/530/R-05/006. Prepared by the EPA Office of Solid Waste and Emergency
Response. Washington, DC. September.
U.S. EPA. 2006. "Data Quality Assessment: Statistical Methods for Practitioners." EPA QA/G-9S.
Prepared by the EPA Office of Environmental Information. Washington, DC. February.
U.S. EPA. 2007. "Ecological Soil Screening Levels for Selenium. Interim Final." OSWER Directive
9285.7-72. Prepared by the EPA Office of Solid Waste and Emergency Response. Washington,
DC.
U.S. EPA. 2008a. Characterization of Coal Combustion Residues from Electric Utilities Using Wet
Scrubbers for Multi-Pollutant Control. EPA/600/R-08/077. Prepared by Vanderbilt University;
ARCADIS; Turner Technology, LLC and Energy Research Centre of Netherland for Office of
Research and Development under Contract No. EP-C-09-027. Research Triangle Park, NC.
July.
U.S. EPA. 2008b. "Strahler Stream Order and Strahler Calculator Values in NHDPlus." Prepared
by S.M. Pierson, B.J. Rosenbaum, L.D. McKay and T.G. Dewald for the EPA Office of Water.
September.
U.S. EPA. 2009a. "Draft Human and Ecological Risk Assessment of Coal Combustion Residuals."
Prepared for the EPA Office of Solid Waste and Emergency Response under Contract No.
Contract EP-W2-09-004. Washington, DC. August.
U.S. EPA. 2009b. Characterization of Coal Combustion Residues from Electric Utilities - Leaching
and Characterization Data. EPA-600/R-09/151. Prepared by D. Kosson, F. Sanchez and R.
Delapp of Vanderbilt University; P. Kariher of ARCADIS; L.H. Turner of Turner Technology,
LLC; and P. Seignette of the Energy Research Centre of Netherland for Office of Research and
Development under Contract No. EP-C-09-027. Research Triangle Park, NC. December.
U.S. EPA. 2010. "ProUCL Version 4.00.02 User Guide." EPA/600/R-07/038. Prepared by A. Singh,
R. Maichle and N. Armbya of Lockheed Martin Environmental Services for the EPA Office of
Research and Development and Region 4. Washington, DC.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
9-11
Section 9: References
-------
U.S. EPA. 2011. "Exposure Factors Handbook: 2011 Edition." EPA/600/R-090/052F. Prepared by
the EPA Office of Research and Development. Washington, DC. September.
U.S. EPA. 2012a. "Interlaboratory Validation of the Leaching Environmental Assessment
Framework (LEAF) Method 1313 and Method 1316." EPA 600/R-12/623. Prepared by A.C.
Garrabrants, D.S. Kosson and R. DeLapp of Vanderbilt University; L. Stefanski of North
Carolina State University; P.F.A.B. Seignette of the Energy Research Centre of the Netherlands,
H.A. van der Sloot of Hans vander Sloot Consultancy, P. Kariher of ARCADIS-US, Inc., and
M. Baldwin of U.S. EPA for the EPA Office of Research and Development under Contract EP-
C-09-27. Washington, DC. September.
U.S. EPA. 2012b. "Mined Gypsum LEAF Methods 1313 and 1316 Report: Evaluation of Metal
Leaching from Contaminated Soils, FGD, and other Coal Combustion Byproducts in Reuse
Scenarios." Prepared by Pegasus Technical Services, Inc. for the EPA Office of Research and
Development. Cincinnati, OH. July.
U.S. EPA. 2014a. "Coal Combustion Residual Beneficial Use Evaluation: Fly Ash Concrete and FGD
Gypsum Wallboard." EPA530-R-14-001. Prepared by the EPA Office of Solid Waste and
Emergency Response. Washington, DC. February.
U.S. EPA. 2014b. "Final Human and Ecological Risk Assessment of Coal Combustion Residuals."
RIN: 2050-AE81. Prepared by the EPA Office of Solid Waste and Emergency Response.
Washington, DC. December.
U.S. EPA. 2014c. " Response to External Peer Review Comments: Human and Ecological Risk
Assessment for Coal Combustion Residuals." Regulation Identifier Number 2050-AE81.
Prepared by the EPA Office of Solid Waste and Emergency Response. Washington, DC.
December.
U.S. EPA. 2014d. " Response to Public Review Comments: Human and Ecological Risk Assessment
for Coal Combustion Residuals." Regulation Identifier Number 2050-AE81. Prepared by the
EPA Office of Solid Waste and Emergency Response. Washington, DC. December.
U.S. EPA. 2014e. "Leaching Test Relationships, Laboratory-to-Field Comparisons and
Recommendations for Leaching Evaluation using the Leaching Environmental Assessment
Framework." EPA 600/R-14/061. Prepared by D.S. Kosson and A.C. Garrabrants of Vanderbilt
University; H.A. van der Sloot of Hans van der Sloot Consultancy and P.F.A.B. Seignette of the
Energy Research Center of the Netherlands for the EPA Office of Research and Development
under Contract No. EP-C-09-027. Research Triangle Park, NC. September.
U.S. EPA. 2015. "Industrial Waste Management Evaluation Model (IWEM) Version 3.1: Technical
Background Document." EPA530-R-15-004. Prepared by the EPA Office of Solid Waste and
Emergency Response, Washington, DC. June.
U.S. EPA. 2016a. "Methodology for Evaluating the Beneficial Use of Industrial Non-Hazardous
Secondary Materials." EPA 530-R-16-011. Prepared by the EPA Office of Land and Emergency
Management. Washington, DC. April.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
9-12
Section 9: References
-------
U.S. EPA. 2016b. "Beneficial Use Compendium: A Collection of Resources and Tools to Support
Beneficial Use Evaluations." EPA 530-R-16-009. Office of Land and Emergency Management.
Washington, DC. June.
U.S. EPA. 2016c. "Aquatic Life Ambient Water Quality Criterion for Selenium - Freshwater." EPA
EPA 822-R-16-006. Office of Water. Washington, DC. June.
Watts, D.B, and H.A. Torbert. 2009. "Impact of Gypsum Applied to Buffer Strips on Reducing
Soluble Phosphorus in Surface Water Runoff." Journal of Environmental Quality 38:1511-
1517.
Watts, D.B. and H.A. Torbert. 2016. "Influence of Flue Gas Desulfurization Gypsum on Reducing
Soluble Phosphorus in Successive Runoff Events from a Coastal Plains Bermudagrass Pasture."
Journal of Environmental Quality. 45:1071-1079.
Watts, D.B. and H.A. Torbert. 2017. "Three Annual Flue Gas Desulfurization Gypsum Applications
on Macronutrient and Micronutrient Losses in Runoff From Bermudagrass Fertilized With
Poultry Litter." Soil Science. 182(1): 18-27.
WHO (World Health Organization). 2009. "Bromide in Drinking-Water: Background Document
for Development of WHO Guidelines for Drinking-Water Quality." WHO/HSE/WSH/09.01/6.
Geneva, Switzerland.
Williams, M.R., K.W. King, M.L., Macrae, W. Ford, C. Van Esbroeck, R.I. Brunke, M.C. English
and S.L. Schiff. 2015. "Uncertainty in Nutrient Loads from Tile-Drained Landscapes: Effect of
Sampling Frequency, Calculation Algorithm and Compositing Strategy." Journal of Hydrology.
530:306-316.
WRI (World Resources Institute). 2007. "Assessing U.S. Farm Drainage: Can GIS Lead to Better
Estimates of Subsurface Drainage Extent?" Washington, D.C. August.
Xia, K., U.L. Skyllberg, W.F. Bleam, P.R. Bloom, E.A. Nater and P.A. Helmke. 1999. "X-Ray
Absorption Spectroscopic Evidence for the Complexation of Hg (II) by Reduced Sulfur in Soil
Humic Substances." Environmental Science and Technology 33:257-61.
Yang G., Y. Xia. 1995. "Studies on Human Dietary Requirements and Safe Range of Dietary
Intakes of Selenium in China and their Application in the Prevention of Related Endemic
Diseases." Biomedical and Environmental Sciences. 8:187-201.
Yin, Y., H.E. Allen, C.P. Huang, D.L. Sparks and P.F. Sanders. 1997. "Kinetics of Mercury (II)
Adsorption and Desorption on Soil." Environmental Science and Technology 31:496-503.
Zhang, X.C., W.P. Miller, M.A. Nearing and L.D. Norton. 1998. "Effects of Surface Treatment on
Surface Sealing, Runoff, and Interrill Erosion." Transactions of the American Society of
Agricultural Engineers 41:989-994.
Zhao, Y. and H.C. Frey. 2004. "Quantification of Variability and Uncertainty for Censored Data
Sets and Application to Air Toxic Emission Factors." Risk Analysis 24(4):1019-1034.
Zikovsky, L. and G. Kennedy. 1992. "Radioactivity of Building Materials Available in Canada."
Health Physics. 63(4):449-452.
Beneficial Use Evaluation of FGD Gypsum in Aqriculture
9-13
Section 9: References
-------
Appendix A: Constituent Data
This appendix provides a summary of the collection and management of raw data drawn from the
available literature and considered in the beneficial use evaluation of flue gas desulfurization (FGD)
gypsum in agricultural applications. The rationale for excluding any literature sources that were
considered, but not retained, is discussed. The raw data for those literature sources that were relied
upon in the current evaluation can be found in the constituent database.
¦ Attachment A-1: provides a summary of communications between EPA and authors to obtain
unreported data or clarify analytical methodology.
¦ Attachment A-2: provides a summary of bulk and leachate concentration data used in the
evaluation following the data quality review described in this appendix.
A.1 Data Collection
USDA and EPA reviewed the available literature and assembled those that appeared to contain
information on the constituent concentrations present in or released from FGD and mined gypsum.
A number of relevant literature sources, in particular grey literature, had already been obtained
through previous EPA or USDA investigations. Thus, EPA began with a review of the references
cited in these and all subsequently collected studies. EPA also queried Environmental Sciences and
Pollution Management, EBSCO HOST, PubMed, Science Direct, Web of Science, and JSTOR for
the key words: "gypsum," "flue gas desulfurization gypsum," "FGD gypsum," "mined gypsum,"
"natural gypsum," and "synthetic gypsum." Although some literature sources used other terms,
such as "coal gypsum" or "FGD products," these studies tended to be older and ambiguous about
whether the materials analyzed fit the current definition of FGD gypsum. Because capturing
available information on the composition and behavior of gypsum was a primary goal of the
literature search, search terms related to the specific beneficial use were not used. The literature
search resulted in a total of 199 unique sources, of which 75 were determined to contained
potentially relevant information based on preliminary review of abstracts and tables.
A.2 Data Quality Review
EPA reviewed all of the literature sources assembled to ensure that the data drawn from each were
of sufficient quality to form the basis for conclusions about the beneficial use of FGD gypsum. The
following subsections detail how the Agency applied the data quality assessment factors outlined
in A Summary of General Assessment Factors for Evaluation the Quality of Scientific and Technical
Information (U.S. EPA, 2003). When it was determined that data from a particular literature source
were not germane, then those data were removed from the database entirely. When individual
data points or entire studies were found to introduce an unacceptable level of uncertainty into the
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
evaluation, these data were filtered out from the dataset prior to analysis. However, these data
were retained in the database for reference.
A.2.1 Clarity and Completeness
Clarity and completeness are the degree to which a literature source transparently documents all
assumptions, methods, quality assurance protocols, results, and other key information. An
evaluation that is both clear and complete provides enough detail that an outside party with access
to the proper resources can replicate the analyses. During the review of the assembled literature,
EPA found that some authors chose to present summary statistics instead of full datasets. Others
did not specify information about the gypsum that, while not the focus of that particular literature
source, was of importance to the current beneficial use evaluation. EPA made an effort to reach
out to the authors and obtain the missing information. Those who responded did not always have
answers to the questions, either because that information was never collected or because the
authors no longer had access to the raw data. Although some questions remained unanswered, no
data were eliminated because of insufficient clarity or completeness. Instead, the data are presented
with the relevant field marked as "unknown" for transparency. A summary of the effort to contact
authors and the additional information obtained are provided in Attachment A-1.
A.2.2 Evaluation and Review
Evaluation and review is the extent to which a literature source has undergone independent
verification, validation and peer review. An independent review is one conducted by objective
technical experts who were not associated with the generation of the work under review either
directly through substantial contribution to its development, or indirectly through significant
consultation during the development of the work. Independent review is intended to identify any
errors or bias present in how data are collected, handled or interpreted and also to ensure that the
findings are accurate, reliable and unbiased.
The majority of literature sources assembled were drawn from independently peer-reviewed
journals; however, a number of grey literature sources were also identified. Some of these sources
were data submitted directly to EPA by states and other parties. These data were made available
to the public and a panel of independent peer reviewers for comment as part of the development
of Human and Ecological Risk Assessment of Coal Combustion Wastes (U.S. EPA, 2014). Some of
the sources reviewed were grey literature written by EPA and USDA. While these studies were
not all submitted for an independent external review, the data were all collected according to
detailed sampling and analysis plans that specify relevant QA/QC procedures.
In several cases, data from grey literature sources were later published in peer-reviewed journals.
Comparison of the data reported in the two sources identified occasional differences in values
between the grey and published sources. EPA reached out to the authors to determine the cause
of these apparent discrepancies. Only one author responded to clarify that the grey literature in
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
EPRI (2008) contained preliminary data that was later updated. In most other cases, data from the
grey literature sources were used because they were more complete and any discrepancies
identified were minor. In a few cases, individual data points from the later peer-reviewed sources
were included because they were not reported in the earlier grey literature. A summary of the
studies removed from the database due to duplication of data is presented in Table A-1.
Table A-1: Duplicate Sources Filtered from Database
Grey Literature Citation
Journal Citation
Gypsum
Type
Media Analyzed
EPRI (2013a), EPRI (2013b)
Norton (2011)
FGD, Mined
Crop, Leachate, Soil
EPRI (2011b)
DeSutter et al. (2011)
FGD
Crop, Soil
EPRI (2011c)
DeSutter et al. (2014)
FGD, Mined
Crop, Soil
EPRI (2012a)
Kost et al. (2014)
FGD, Mined
Crop, Gypsum, Soil
EPRI (2008), EPRI(2012a), EPRI (2013a), and EPRI (2013b)
Chen et al. (2014)
FGD, Mined
Gypsum
A.2.3 Soundness
Soundness is the extent to which the methods employed by a literature source are reasonable and
consistent with the intended application of the data. This means that any methods used to collect
and measure data have demonstrated the technical ability to reliably and repeatedly achieve
desired levels of accuracy and precision, and that any methods used to analyze and interpret data,
such as equations, models and simplifying assumptions, are adequately justified and rooted in
accepted scientific principles.
Sample Collection Methods:
Some studies did not report the approach used to collect solid soil and gypsum samples. In cases
where the samples were provided by the facilities, the author may not have this information. As a
result, it is sometimes unclear whether the data represent individual grab samples or composite
samples collected over an unspecified area or time interval. Both collection methods are valid and
reflect the material under evaluation, but provide somewhat different information. Individual grab
samples have the potential to capture isolated "hot spots," while composite samples provide a more
typical estimate of the concentrations present either spatially or temporally. The available dataset
is known to contain a combination of both types of samples. Combining data points obtained from
different collection methods treats them as identical, which will introduce some uncertainty into
the evaluation.
This evaluation focuses primarily on chronic exposures that are best captured by data that provide
representative values for each source. While individual grab samples may over- or underestimate
these values on a case-by-case basis, these samples still reflect actual concentrations in the gypsum.
A large fraction of the available data for many constituents are grab samples. Eliminating these
samples would greatly reduce the available data and might omit high-end values that ensure the
evaluation remains protective. In addition, a comparison of grab samples in the database from the
same sources indicates that there is not a high degree of variability among the gypsum produced
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
at a facility at a given time. This may be attributed to the substantial mixing of the slurry that
occurs during gypsum production and handling. Therefore, the uncertainty introduced into the
evaluation is relatively small and no samples were filtered out due to the sample collection methods
used.
Sample Preparation Methods:
EPA reviewed the methods used to prepare samples for analysis reported by each source to
determine whether the resulting data were of sufficient quality to incorporate in this evaluation.
The purpose of the preparation methods is to convert constituent mass into a soluble form that can
be measured by standard analytical instruments. The majority of studies report digestion by a
combination of heat and one or more of the following acids: hydrochloric acid, hydrofluoric acid,
nitric acid and perchloric acid. Different acids are effective at breaking down different compounds,
such as organic matter, iron oxides and silicates. As a result, a combination of acids is often used to
maximize dissolution.
FGD gypsum typically composed of 95% or more calcium sulfate, which is a mineral that dissolves
readily in water, and less than 1% silica (Henkels and Gaynor, 1995). Therefore, it is unlikely there
is a substantial amount of constituent mass retained in recalcitrant minerals. Thus, EPA concluded
that combining data with different acids is unlikely to introduce substantial uncertainty into the
evaluation. No samples were filtered out due to the sample preparation methods used.
Detection Limits:
A detection limit is the lowest level at which an analytical instrument can reliably differentiate
actual constituent concentrations from background noise. When a constituent is not detected
above this limit, the analytical results are typically reported as less than the detection limit because
the potential still exists for the constituent to be present at a lower level. The detection limit varies
among studies because of differences in the methods used to prepare samples, the sensitivity of
analytical instruments, and interference from solid media or other chemical constituents.
Non-detect (or "left-censored") data are typically the lowest values in a dataset. Elimination of
these non-detect data may bias the remaining dataset high. Therefore, EPA incorporated all of
these data into the constituent database at the detection limit and flagged the values as non-detect
in a separate column. However, non-detect values were not always the lowest values reported for
some constituents and, in some cases, were the highest. High detection limits introduce a great
amount of uncertainty into the evaluation and can bias the overall dataset high. Therefore, EPA
filtered out any non-detect values that were greater than the 90th percentile of the remaining,
detected data. A summary of the data filtered out from the database prior to analysis due to high
detection limits is presented in Table A-2.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Table A-2: High Detection Limits Filtered from Database
Source
Gypsum Type
Constituent
Reported
Detection Limit
Reported
Units
FGD
Beryllium
3.1
mg/kg
Bryant et al. (2012)
FGD
Cadmium
3.1
mg/kg
FGD
Lead
3.1
mg/kg
FGD
Thallium
2.5
mg/kg
Chen et al. (2008)
FGD
Arsenic
11
mg/kg
FGD
Cadmium
1
mg/kg
FGD
Lead
5
mg/kg
DeSutter and Cihacek (2009)
Mined
Arsenic
2.6
ug/g
Mined
Selenium
1.2
ug/g
OSU-E (2005)
Mined
Cadmium
0.48
ppm
Mined
Selenium
1.45
ppm
FGD
Cadmium
1
ug/g
FGD
Cadmium
1
ug/g
EERC (2007)
FGD
Cadmium
1
ug/g
FGD
Lead
3
ug/g
FGD
Lead
3
ug/g
FGD
Lead
3
ug/g
FGD
Lead
3
ug/g
EPRI (2008)
FGD
Thallium
1.44
ug/g
EPRI (2011a)
Mined
Beryllium
0.1
mg/kg
Mined
Beryllium
0.1
mg/kg
Mined
Arsenic
4.21
mg/kg
EPRI (2012b)
Mined
Beryllium
0.18
mg/kg
FGD
Beryllium
0.18
mg/kg
Mined
Selenium
4.86
mg/kg
EPRI (2013b)
FGD
Thallium
1.44
mg/kg
FGD
Thallium
1.44
mg/kg
FGD
Beryllium
0.16
mg/kg
Schomberg et al. (2018)
FGD
Beryllium
0.13
mg/kg
FGD
Thallium
1.3
mg/kg
FGD
Thallium
1.6
mg/kg
Mined
Cadmium
0.9
mg/kg
Yost et al. (2010)
FGD
Lead
3.4
mg/kg
FGD
Lead
3.5
mg/kg
A.2.4 Applicability and Utility
Applicability and utility is the extent to which the findings of a literature source are relevant for
the intended use. This means that the purpose, design and findings of the data can support a similar
set of conclusions when applied to the conceptual model for the beneficial use. EPA reviewed each
of the studies collected to ensure that the data contained were representative of the materials used
and environmental conditions relevant to the current evaluation.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Country of Origin:
When reviewing the available literature, EPA did not initially screen based on the country in
which the gypsum was generated, as this data might be useful later for comparisons. Thus, some
data were collected from countries outside the United States. Differences in the composition of the
coal burnt and the pollution control technologies used in these countries may result in a trace
element composition that does not reflect gypsum generated in the United States. Thus, to ensure
data relied upon was representative, EPA filtered out data on gypsums generated outside North
America from the constituent database. An exception was made where measurements were not
directly related to composition of the gypsum, such as crop uptake from gypsum amended soils.
Even if the gypsum used in the literature source had higher levels of trace elements than may be
found in the North America, the literature source can still provide useful information about the
tendency of plants to accumulate constituents mass based on the concentrations present in
surrounding soils. A summary of the data removed from the database due to the country of origin
is presented in Table A-3.
Table A-3: Foreign Gypsum Filtered from Database
Source
Countries
Gypsum
Type
Media
Alvarez-Ayuso et al. (2006)
Spain
FGD
Bulk Concentration
Alvarez-Ayuso and Querol (2007)
Spain
FGD
Bulk Concentration
Alvarez-Ayuso et al. (2011)
Spain
FGD
Bulk Concentration, Leachate
Amezketa et al. (2005)
Spain
FGD, Mined
Bulk Concentration
Berland et al. (2003)
Germany, Japan, United Kingdom
FGD
Bulk Concentration
Ralloetal. (2010)
Spain
FGD
Bulk Concentration
Stergarsek et al. (2008)
Slovenia
FGD
Bulk Concentration
Yodthongdee et al. (2013)
Thailand
FGD
Bulk Concentration
Pelletized Samples:
Some studies measured constituent concentrations present in and released from mined gypsum
that had been pelletized. Pelletization is a process that coats the gypsum with a binding solution,
such as a cellulose-based polymer (U.S. EPA, 2012). The resulting product is marketed as "easier to
transport and apply" and "providing more efficient delivery of nutrients." There was not enough
information to determine whether pelletization may contribute additional constituent mass or
alter the leaching behavior of mined gypsum. Therefore, EPA filtered out data for the mined
gypsum samples that had been pelletized. Samples of FGD gypsum were typically collected directly
from utilities and so none had undergone additional processing. A summary of the data filtered out
prior to quantitative analysis presented in Table A-4.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Table A-4: Pelletized Gypsum Filtered from Database
Source
Gypsum Type
Media
U.S. EPA (2012)
Mined
Bulk Concentration, Leachate
Chen et al. (2014)
Mined
Bulk Concentration
EPRI (2008)
Mined
Bulk Concentration
EPRI (2012a)
Mined
Bulk Concentration, Leachate
EPRI (2013a)
Mined
Bulk Concentration, Leachate
EPRI (2013b)
Mined
Bulk Concentration, Leachate
Kost et al. (2014)
Mined
Bulk Concentration
Materials Other Than Gypsum:
Not all of the studies identified from the literature focused exclusively on gypsum. Several also
contained information on other FGD materials or other coal combustion residuals. Because these
non-gypsum materials are not the focus of this evaluation, these data were not even considered for
the database. However, other studies appeared to measure gypsum, but further inspection revealed
that the studies erroneously labeled the material as gypsum or had mixed gypsum with other
materials (e.g., chicken litter) prior to analysis. Because these measurements cannot be used to
characterize raw gypsum, these data were also removed from the constituent database. A summary
of the non-gypsum materials removed from the database is presented in Table A-4.
Table A-4: Non-Gypsum Material Filtered from Database
Source
Material Analyzed
Alvarez-Ayuso and Querol (2007)
FGD gypsum treated with aluminum oxide
Chen et al. (2008)
"FGD Product"
Chen et al. (2009)
FGD-CaSOs
EPRI (2008)
FGD containing fly ash
U.S. EPA (2009)
FGD containing fly ash
Analytes and Reported Units:
Environmental contamination was not the primary focus of every literature source and, as a result,
many studies reported data on some analytes that were not germane to the current evaluation.
Other studies reported data on relevant analytes, but in units that could not be reliably converted
into a useable form with the information provided (e.g., mass percent). These data could not be
incorporated into a quantitative evaluation and were removed from the constituent database. Table
A-5 presents a summary of the data removed from the database because of unusable analytes or
units.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Table A-5: Unsuitable Analytes and Units Filtered from Database
Source
Gypsum Type
Analyte
Units
EERC (2007)
FGD
Calcium oxide, Magnesium oxide, Phosphorus
pentoxide, Potassium oxide, Sodium oxide
%
EPRI (2008)
FGD, Mined
Lime test index, Total Neutralizing Potential
various
EPRI (2012a)
FGD, Mined
Lime test index, Total Neutralizing Potential
various
EPRI (2012b)
FGD, Mined
Total Neutralizing Potential
various
EPRI (2013a)
FGD, Mined
Total Neutralizing Potential
various
EPRI (2013b)
FGD, Mined
Calcium carbonate, Lime test index, Nitrogen,
Phosphorus
various
Kost et al. (2014)
FGD, Mined
Total Neutralizing Potential
various
Stergarsek et al. (2008)
FGD
Arsenic, Bromine, Mercury, Selenium
mg/ton coal
Stout et al. (1999)
FGD
CaCC>3 equivalent
g/kg
Wallboard Samples:
Some data included in the evaluation for total content were gypsum that had been processed into
wallboard. Certain steps in the wallboard production process, specifically washing to remove
impurities and heated drying, have the potential to alter the constituent composition of the
gypsum. Other steps in the production process are unlikely to alter constituent concentrations
within the bulk gypsum (EERC, 2003). While additives, such as starch and vermiculite, may be
mixed in during these processes, these typically account for less than a half percent of the total
product mass and are not anticipated to appreciably alter overall constituent concentrations.
Heated drying and calcination processes expose FGD gypsum to temperatures above 128°C (262°F).
These elevated temperatures accelerate the release of mercury, but the fraction that will volatilize
depends on a number of factors that include the specific temperature and drying time
(NETL, 2008). Because less energy-intensive dewatering processes are available that do not result
in elevated releases to the surrounding air; such as hydrocyclones, centrifuges and belt presses
(Genck et al., 2008); EPA assumes that gypsum intended for agricultural applications will not be
exposed such high temperatures after generation. To determine if inclusion of these wallboard data
in the dataset may have biased constituent distributions lower, EPA compared the bulk content
distributions with and without these data. EPA found that both the median and high-end
concentrations were unchanged after removal of these data. Therefore, the inclusion of these data
are unlikely skew model results in this instance.
Mercury Emission Sample Collection:
During the review of the assembled data, EPA identified differences among the mercury emission
data that could not be resolved with the information reported in the literature. These discrepancies
arose primarily from the disparate methods used to sample for mercury. Some studies measured
the total mercury captured by filters over a specified timeframe from which a time-averaged
emission rate was calculated, while others used a real-time analyzer to measure the actual emission
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
rate at specified time intervals. Even when samples were collected with similar methods, each
literature source reported values averaged over different timeframes. It was not possible to
transform and aggregate all of the available data in a meaningful way and, because mercury
emissions can be highly variable over time, presenting the data without these caveats in the
database runs the risk of being misleading. Therefore, because these data were not relied upon in
the evaluation and to avoid confusion from the presentation of these data without proper context,
EPA chose to remove these data from the database. A summary of the mercury emission data
removed from the database is presented in Table A-6.
Table A-6: Mercury Emission Data Filtered from Database
Source
Media
Briggs et al. (2014)
Unamended Soil, Mixed Soil and FGD/Mined Gypsum
Cheng et al. (2012)
FGD Gypsum
Gustin and Ladwig (2010)
FGD Gypsum
Pekney et al. (2009)
FGD Gypsum
Shock et al. (2009)
FGD Gypsum, Mined Gypsum
Wang et al. (2013)
Mixed Soil and FGD Gypsum
Xin et al. (2006)
FGD Gypsum
Sampling Depth:
When reviewing the soil data used to calculate BCFs, EPA identified some soil samples that were
collected from depths outside of the typical crop root zone. Crops are unlikely to be exposed to
concentrations this deep and so use of these data to calculate BCFs was considered inappropriate.
Samples that included depths much greater than 20 cm were excluded. A summary of the data
filtered out prior to quantitative analysis presented in Table A-7.
Table A-7: Soil Depth
Filtered from Database
Source
Media
Depth
DeSutter et al. (2014)
Unamended Soil, FGD Amended Soil, and
Mined Gypsum Amended Soil
15 - 30 cm
EPRI (2011c)
Unamended Soil, FGD Amended Soil, and
Mined Gypsum Amended Soil
15 - 30 cm
EPRI (2012a)
Unamended Soil, FGD Amended Soil, and
Mined Gypsum Amended Soil
15 - 30 cm
Lysimeter Data:
Lysimeters are sampling devices placed under the soil to collect leachate samples that reflect the
mixture of gypsum and soil. In theory, such samples can empirically demonstrate how interactions
with the soil affect the leaching behavior of gypsum to provide a more accurate estimate of field
leaching. However, in practice, available samples introduce too much uncertainty to incorporate
into this evaluation:
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
¦ The sources identified did not separately test the leaching behavior of the gypsum applied to
the soil. Therefore, it is not possible to determine how leachate concentrations changed as a
result of mixing.
¦ Each of the sources reflect only a few soils amended with single type of gypsum. It is not
possible to make statements about the representativeness of these samples without additional
information.
¦ The studies only measured leaching over the course of about a year. This could underestimate
long-term leaching because constituent mass has not had enough time to fully migrate
through the soil column.
¦ For multiple constituents, all or nearly all of the lysimeter samples collected among these
literature sources were non-detect (i.e., aluminum, antimony, arsenic, beryllium, cadmium,
cobalt, copper, iron, lead, selenium, thallium). Non-detect data has the potential to yield
useful information, but the detection limits were often near or above the highest leachate
concentration measured from unmixed FGD gypsum (i.e., antimony, arsenic, lead, thallium).
¦ The available leachate data from unmixed FGD gypsum demonstrates there is a high potential
for depletion of available constituent mass within a year. Because the studies do not specify
the amount of precipitation and the infiltration captured in the lysimeter, the magnitude of
this dilution cannot be estimated.
Based on these considerations, any conclusions drawn with this data would require substantial
caveats. Because these data could not be incorporated into the evaluation, they were filtered out
of the constituent database when characterizing leaching behavior. A summary of the data filtered
out prior to quantitative analysis presented in Table A-8.
Table A-8: Lysimeter Data Filtered from Database
Source
Media
Briggsetal. (2014)
Unamended Soil, Mixed Soil and FGD/Mined Gypsum
EPRI (2012a)
Mixed Soil and FGD/Mined Gypsum
EPRI (2013a)
Unamended Soil, Mixed Soil and FGD/Mined Gypsum
EPRI (2013b)
Unamended Soil, Mixed Soil and FGD/Mined Gypsum
Wang (2012)
Unamended Soil, Mixed Soil and FGD Gypsum
Environmental Factors:
EPA reviewed the assembled literature to determine whether each source reported sufficient
information about the environmental factors that control releases into each media. Information on
these factors allows EPA to appropriately apply the data to corresponding field conditions. EPA
identified the key properties through a review of the literature. Some have been well established
in the literature, while others are suspected and are the topic of continued investigation.
The soil pH and composition (e.g., clay), as well as the type of plant grown, were identified as the
key factors most likely to influence plant uptake of inorganics (ORNL, 1984; U.S. EPA, 1999).
When available, this information was assembled in the database along with the reported soil and
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ -jq
Appendix A: Constituent Data
-------
plant concentrations. All studies reported information on the type of plant grown, but some did
not report soil pH and/or composition. EPA did not remove any samples from the database when
this information was not available because it was still known that the samples represent conditions
under which the crops can be grown. There was no indication that any of the studies took steps to
alter the soil in ways that would result in unrealistic conditions. Therefore, no data were removed
from the database due to missing information about soil pH or composition.
The liquid to solid (L/S) ratio and the equilibrium pH of leachate samples were identified as two
key factors most likely to influence the leaching behavior of inorganics (U.S EPA, 2010). When
available, this information was assembled in the database along with the leachate concentrations.
The L/S ratio is specified by most leaching tests. Therefore, even when this information was not
explicitly reported, it could be easily inferred from the test used. Although many tests also specify
the initial pH of the eluent, some do not specify a final, equilibrium pH of the solution. This is
because these extraction tests are designed to assess the behavior of materials exposed to specific
inputs (e.g., acid rain). The final equilibrium pH can vary dramatically from the initial pH based
on the chemistry of the material tested. Leachate concentrations can vary by orders of magnitude
with incremental changes in pH. Thus, if the final pH of a sample was unknown, the samples
introduce too much uncertainty and were entirely removed from the database. If the final pH was
recorded but fell outside the relevant pH range of between 5 and 8, then the data were retained in
the database but filtered out prior to quantitative analysis. Table A-9 presents a summary of the
data removed because of unknown or unsuitable environmental conditions.
Table A-9: Incomplete Environmental
Data Filtered from Database
Source
Gypsum Type
Media
Missing Information
Bryant et al. (2012)
FGD
Leachate
No final pH
Pasini and Walker (2012)
FGD
Leachate
No final pH
Xin et al. (2006)
FGD
Leachate
No final pH
A.2.5 Variability and Uncertainty
Variability and uncertainty are the extent to which a literature source effectively characterizes,
either quantitatively or qualitatively, these two factors in the information relied upon and in the
procedures, measures, methods or models utilized. Proper characterization of the major sources of
variability and uncertainty provides greater confidence that the data can form the basis for sound
conclusions in the beneficial use evaluation.
Each individual literature source provides raw data on only a few samples and so is unlikely to
fully capture the variability of constituent concentrations present in and released from gypsum.
To address this fact, this beneficial use evaluation aggregated data from all of the available sources
found to otherwise be of sufficient quality. Because more data ensures better characterization of
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
the gypsum, there is no reason to exclude any individual sample because it alone does not fully
capture variability.
Each literature source provides raw data on constituent concentrations present in or released from
gypsum. The methods used to measure these concentrations were found to be sound, so there is
minimal uncertainty associated with the detected values reported in any of the literature sources.
Therefore, no data were eliminated as a result of uncertainties about the specific values reported.
A.3 Data Management
The data found to be of sufficient quality were incorporated into the database as reported in each
literature source. Once all of the data were assembled, additional management steps were taken
prior to use of the data in any quantitative analyses to ensure the dataset was not biased towards a
single samples or source. To do this, EPA identified and combined replicate and duplicate samples.
Replicate samples are two or more measurements of the same sample, often collected for QA/QC
purposes. These replicate samples are typically averaged into a single data point to obtain a
representative value for the sample. Some literature sources only report the mean of the replicates
and the corresponding standard deviation. EPA did not make an attempt to reach out and obtain
values for each individual replicate from these sources because the same average value would have
been calculated for this evaluation prior to use in any quantitative analyses.
Duplicate samples are two or more field samples intended to represent the same source, which are
collected and analyzed in an identical manner. For a number of reasons, such as heterogeneity of
the source material and sensitivity of the analytical equipment, values measured for these samples
may not be identical. This evaluation averaged duplicate samples to obtain a more representative
value for that source. EPA chose to treat any samples collected from the same facility as duplicates,
regardless of whether the samples were collected as part of separate studies or at different times.
This was done to avoid biasing the overall data set towards facilities that had been more heavily
sampled. Each literature source had a unique approach to labeling samples that sometimes made it
difficult to identify the facility associated with a sample. There was often enough other information
presented, such as the geographic location of the source, to determine whether samples were
duplicates. However, it is possible some that some duplicate sample remain.
To prepare duplicate and replicate samples for quantitative analysis, EPA first averaged all
replicates for a single sample and then all duplicates from a single source. Where duplicates and
replicates were a mixture of detect and non-detect values, the non-detect values were set to the
detection limit. Because the constituent was detected in other duplicated or replicates, it was
assumed that the detection limit would provide the best, conservative estimate of the true value.
The resulting averaged value was flagged as detected for the quantitative analysis.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
A.4 Summary of Available Data
Table A-10 presents a summary of the available data for FGD gypsum bulk concentration and
leachate available to characterize each of the release pathways, after filtering for data quality and
managing replicate and duplicate samples. Both bulk concentration and leachate data were
necessary to provide realistic estimates of potential exposures. Constituents with insufficient data
to characterize bulk concentration (e.g., bromide, tin), leachate (e.g., fluoride, lithium, titanium),
or both (e.g., silver) were not retained for full evaluation. The uncertainties introduced due to this
lack of data are discussed as further in Section 7 (Risk Characterization). Constituents known to
be macronutrients for plants and animals (i.e., calcium, carbon, magnesium, nitrogen, potassium,
sulfur/sulfate) were not retained for evaluation and are not discussed further in this document.
Contributions of these nutrients from applied FGD gypsum should be factored into relevant
nutrient management plans. A summary of the concentration data used in the evaluation is
presented in Attachment A-2.
Table A-10: Summary of Filtered Constituent Data
Constituent
CASRN
Bulk Concentration
Detected / Total Samples
Leachate
Detected / Total Samples
Unwashed
Washed
Unknown
Unwashed
Washed
Unknown
Aluminum
7429-90-5
21 / 21
29/29
11/11
37/39
37/38
4/7
Antimony
7440-36-0
21 / 21
29/31
8/10
8/39
11/39
7/7
Arsenic
7440-38-2
27/29
31 / 33
6/16
11/46
5/53
0/10
Barium
7440-39-3
21 / 21
31 / 31
11/11
46/46
53/53
10/10
Beryllium
7440-41-7
7/10
13/21
3/10
1 / 39
5/39
2/7
Bismuth
7440-69-9
0/0
0/2
0/0
0/0
0/0
0/0
Bromide
7726-95-6
0/0
0/0
0/0
5/19
0/23
0/0
Boron
7440-42-8
11/11
19/20
10/12
39/39
31 / 39
4/7
Cadmium
7440-43-9
21 / 21
29/31
8/16
24/45
18/53
3/10
Calcium
7440-70-2
20/20
29/29
12/12
39/39
39/39
7/7
Carbon
7440-44-0
11/11
9/9
0/0
0/0
0/0
0/0
Chloride
16887-00-6
18/18
14/14
0/1
22/22
10/23
0/0
Chromium
7440-47-3
21 / 21
31 / 31
13 / 16
30/45
32/53
4/10
Cobalt
7440-48-4
20/21
24/31
4/11
25/39
18/39
4/7
Copper
7440-50-8
14/16
18/22
7/13
24/39
25/39
4/7
Fluoride
16984-48-8
5/11
4/9
0/0
0/0
0/0
0/0
Iron
7439-89-6
28/28
29/29
16/16
27/39
24/39
6/7
Lead
7439-92-1
20/21
25/32
8/16
7/46
13/53
4/10
Lithium
7439-93-2
10/10
19/19
7/9
0/0
0/0
3/3
Magnesium
7439-95-4
21 / 21
29/29
13 / 13
39/39
39/39
7/7
Manganese
7439-96-5
10/10
20/20
12/13
37/38
37/39
7/7
Mercury
7439-97-6
35/35
51 / 51
20/22
30/52
41 / 57
7/8
Molybdenum
7439-98-7
21 / 21
30/31
9/12
24/39
26/39
7/7
Nickel
7440-02-0
10/10
21 / 22
11/16
37/39
34/39
6/7
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Table A-10: Summary of Filtered Constituent Data
Constituent
CASRN
Bulk Concentration
Detected / Total Samples
Leachate
Detected / Total Samples
Unwashed
Washed
Unknown
Unwashed
Washed
Unknown
Potassium
7440-09-7
20/21
27/29
11/12
2/2
10/10
4/7
Selenium
7782-49-2
29/29
32/32
11/16
43/46
41 / 53
4/10
Silicon
7440-21-3
21 / 21
29/29
8/8
0/0
0/0
3/3
Silver
7440-22-4
0/1
0/2
0/2
0/7
0/14
0/3
Sulfate
14808-79-8
0/0
0/0
1 / 1
22/22
23 /23
0/0
Sodium
7440-23-5
20/21
22/29
6/10
2/2
10/10
7/7
Strontium
7440-24-6
21 / 21
29/29
8/8
39/39
39/39
7/7
Tin
7440-31-5
0/0
0/0
0/0
0/36
1 / 29
0/0
Titanium
7440-32-6
3/11
3/9
0/0
0/0
0/0
0/0
Thallium
7440-28-0
19/19
28/30
4/8
18/39
7/39
6/7
Uranium
7440-61-1
0/0
2/2
0/0
0/0
0/0
0/0
Vanadium
7440-62-2
10/10
21 / 22
8/10
31 / 39
26/39
3/7
Zinc
7440-66-6
10/10
22/22
12/13
39/39
39/39
6/7
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
A.5 References
Al-Abed, S. R., G. Jegadeesan, K. G. Scheckel and T. Tolaymat. 2008. "Speciation, Characterization,
and Mobility of As, Se, and Hg in Flue Gas Desulphurization Residues." Environmental Science
and Technology. 42(5):1693-1698.
Alva, A. K., O. Prakash and S. Paramasivam. 1998a. "Transport of Nitrogen Forms in a Sandy
Entisol with Coal Combustion By-Product Gypsum Amendment." Journal of Environmental
Science and Health. A33(6):1023-1039.
Alva, A. K., O. Prakash and S. Paramasivam. 1998b. "Flue-Gas Desulfurization Gypsum Effects on
Leaching of Magnesium and Potassium from a Candler Fine Sand." Communications in Soil
Science and Plant Analysis. 29(3-4):459-466.
Alvarez-Ayuso, E. and X. Querol (2007). "Stabilization of FGD Gypsum for its Disposal in Landfills
Using Amorphous Aluminium Oxide as a Fluoride Retention Additive." Chemosphere. 69:295-
302.
Alvarez-Ayuso, E., X. Querol and A. Tomas. 2006. "Environmental Impact of a Coal Combustion-
Desulphurisation Plant: Abatement Capacity of Desulphurisation Process and Environmental
Characterisation of Combustion By-Products." Chemosphere. 65:2009-2017.
Alvarez-Ayuso, E., X. Querol and A. Tomas. 2008a. "Implications of Moisture Content
Determination in the Environmental Characterization of FGD Gypsum for its Disposal in
Landfills." Journal of Hazardous Materials. 153(l-2):544-550.
Alvarez-Ayuso, E., X. Querol, J. C. Ballesteros and A. Gimenez. 2008b. "Risk Minimisation of FGD
Gypsum Leachates by Incorporation of Aluminium Sulphate." Science of the Total
Environment. 406(l-2):69-75.
Alvarez-Ayuso, E., A. Gimenez and J.C. Ballesteros. 2011. "Fluoride Accumulation by Plants
Grown in Acid Soils Amended with Flue Gas Desulphurisation Gypsum." Journal of Hazardous
Materials. 192:1659-1666.
Amezketa, E., R. Aragiies and R. Gazol. 2005. "Efficiency of Sulfuric Acid, Mined Gypsum, and
Two Gypsum By-Products in Soil Crusting Prevention and Sodic Soil Reclamation." Agronomy
Journal. 97(3):983-989.
Briggs, C.W., R. Fine, M. Markee and M.S. Gustin. 2014. "Investigation of the Potential for
Mercury Release from Flue Gas Desulfurization Solids Applied as an Agricultural
Amendment." Journal of Environmental Quality. 43(l):253-262.
Bryant, R.B., A.R. Buda, P.J. Kleinman, C.D. Church, L.S. Saporito, G.J. Folmar, S. Bose and A.L.
Allen. 2012. "Using Flue Gas Desulfurization Gypsum to Remove Dissolved Phosphorus from
Agricultural Drainage Waters." Journal of Environmental Quality. 41(3):664-671.
Chen, L., D. Kost and W.A. Dick. 2008. "Flue Gas Desulfurization Products as Sulfur Sources for
Corn." Soil Science Society of America Journal. 72(5):1464-1470.
Chen, L., C. Ramsier, J. Bigham, B. Slater, D. Kost, Y.B. Lee and W.A. Dick. 2009. "Oxidation of
FGD-CaS03 and Effect on Soil Chemical Properties when Applied to the Soil Surface." Fuel.
88(7):1167-1172.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Chen, L., D. Kost, Y. Tian, X. Guo, D. Watts, D. Norton, R.P. Wolkowski and W.A. Dick. 2014.
"Effects of Gypsum on Trace Metals in Soils and Earthworms." Journal of Environmental
Quality. 43(l):263-272.
Cheng, C., Y. Chang, K.R. Sistani, Y. Wang, W. Lu, C. Lin, J. Dong, C. Hu and W. Pan. 2012.
"Mercury Emission and Plant Uptake of Trace Elements during Early Stage of Soil Amendment
Using Flue Gas Desulfurization Materials." Journal of the Air and Waste Management
Association. 62(2):139-150.
Clark, R.B., K.D. Ritcheyand V.C. Baligar. 2001. "Benefits and Constraints for Use of FGD Products
on Agricultural Land." Fuel. 80(6):821-828.
DeSutter, T.M. and L.J. Cihacek. 2009. "Potential Agricultural Uses of Flue Gas Desulfurization
Gypsum in the Northern Great Plains " Agronomy Journal. 101(4):817-825.
Desutter, T.M., J. Lukach and L.J. Cihacek. 2011. "Sulfur Fertilization of Canola (Brassica napus)
with Flue Gas Desulfurization Gypsum: An Assessment Literature source." Communications in
Soil Science and Plant Analysis. 42(20):2537-2547.
DeSutter, T.M., L.J. Cihacek and S. Rahman. 2014. "Application of Flue Gas Desulfurization
Gypsum and its Impact on Wheat Grain and Soil Chemistry." Journal of Environmental
Quality. 43(1):303-311.
EERC (Energy and Environmental Research Center). 2003. "Review of Handling and Use of FGD
Material." 2003-EERC-04-04. Prepared by T.D. Berland, B.A. Pflughoeft-Hassett, K.E. Dockter,
D.J. Hassett and L.V. Heebink under U.S. DOE Cooperative Agreement No. DE-FC26-
98FT40321. April.
EERC. 2007. "Mercury and Air Toxic Element Impacts of Coal Combustion By-Product Dispoal
and Utilization." 2007-EERC-10-03. Prepared by D.J. Hassett, L.V. Heebink, D.F. Plughoeft-
Hassett, T.D. Buckley and E.J. Zacher of the EERC; M. Xin and M.S. Gustin of the University
of Nevada, Reno; and R. Jung for the U.S. DOE under Cooperative Agreement No. DE-FC26-
02NT41727. October.
EPRI (Electric Power Research Institute). 2008. Flue Gas Desulfurization Gypsum Agricultural
Network: 2008 Progress Report." Prepared by W. Dick and D. Kost of Ohio State University.
Palo Alto, CA. December.
EPRI. 2011a. "Composition and Leaching of FGD Gypsum and Mined Gypsum." Prepared by B.
Hensel and B. Hennings of Natural Resource Technology, L. Dayson and S. Whitacre of Ohio
State University, P. Kariher of Arcadis and M. Gustin of University of Nevada, Reno. Palo Alto,
CA. November.
EPRI. 2011b. "Flue Gas Desulfurization Gypsum Agricultural Network: North Dakota Sites 3, 4,
and 5 (Canola)." Prepared by W. Dick and D. Kost of Ohio State University and T. DeSutter
and L. Cihacek of North Dakota State University. Palo Alto, CA. November.
EPRI. 2011c. "Flue Gas Desulfurization Gypsum Agricultural Network: North Dakota Sites 1 and
2 (Wheat)." Prepared by W. Dick and D. Kost of Ohio State University and T. DeSutter and L.
Cihacek of North Dakota State University. Palo Alto, CA. December.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
EPRI. 2012a. "Flue Gas Desulfurization Gypsum Agricultural Network: New Mexico Sites 1
(Alfalfa) and 2 (Sodic Soils)." Prepared by W. Dick and D. Kost of Ohio State University. Palo
Alto, CA. September.
EPRI 2012b. "Flue Gas Desulfurization Gypsum Agricultural Network: Ohio Sites 1 (Mixed Hay)
and 2 (Corn)." Prepared by W. Dick and D. Kost of Ohio State University and D. Smeal of New
Mexico State University. Palo Alto, CA. October.
EPRI. 2013a. "Flue Gas Desulfurization Gypsum Agricultural Network: Wisconsin Arlington
Research Station Fields 295 and 27 (Alfalfa)." Prepared by W. Dick and D. Kost of Ohio State
University, L.D. Norton of USDA, and R.P. Wolkowski of the University of Wisonsin-Madison.
Palo Alto, CA. May.
EPRI. 2013b. "Flue Gas Desulfurization Gypsum Agricultural Network: Indiana Kingman Research
Station (Corn and Soybeans)." Prepared by W. Dick and D. Kost of Ohio State University and
L.D. Norton of USDA. Palo Alto, CA. October.
FIPR (Florida Institute of Phosphate Research). 1995. Literature Review on Gypsum as a Calcium
and Sulfur Source for Crops and Soils in the Southeastern Unites States. Publication No. 01-
118-118. Prepared by M.E. Sumner. Bartow, FL. August.
Genck, W.J., D.S. Dickey, F.A. Baczek, D.C. Bedell, K. Brown, W. Chen, D.E. Ellis, P. Harriott, W.
Li, J.K. McGillicuddy, T.P McNulty, J.Y. Oldshue, F. Schoenbrunn, J.C. Smith, D.C. Taylor and
D.R. Wells. 2008. "Perry's Chemical Engineers' Handbook 8th Edition." McGraw-Hill. New
York, NY.
Grichar, W.J., B.A. Besler and K.D. Brewer. 2002. "Comparison of Agricultural and Power Plant
By-Product Gypsum for South Texas Peanut Production." Texas Journal of Agriculture and
Natural Resources. 15:44-50.
Gustin, M. and K. Ladwig (2010). "Supporting Information: Laboratory investigation of Hg release
from flue gas desulfurization products." Environmental Science and Technology. 44:4012-
4018.
Henkels, P.J. and J.C. Gaynor. 1996. "Characterizing Synthetic Gypsum for Wallboard
Manufacture." Presented at the Spring National Meeting of the American Chemical Society,
March 24-28. New Orleans, LA.
Kairies, C.L., K.T. Schroeder and C.R. Cardone. 2006. "Mercury in Gypsum Produced from Flue
Gas Desulfurization." Fuel. 85(17-18):2530-2536.
Kost, D., L. Chen, X. Guo, Y. Tian, K. Ladwig and W. A. Dick (2014). "Effects of Flue Gas
Desulfurization and Mined Gypsums on Soil Properties and on Hay and Corn Growth in
Eastern Ohio." Journal of Environmental Quality. 43(1):312-321.
Lee, J., K. Cho, L. Cheng, T.C. Keener, G. Jegadeesan and S.R. Al-Abed. 2009. "Investigation of a
Mercury Speciation Technique for Flue Gas Desulfurization Materials." Journal of the Air and
Waste Management Association. 59(8):972-979.
NETL (National Energy Technology Laboratory). 2005. "Fate of Mercury in Synthetic Gypsum
Used for Wallboard Production: Topical Report, Task 1 Wallboard Plant Test Results."
Prepared J. Marshall of USG Corporation and G.M. Blythe and M. Richardson of URS
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Corporation for the DOE NETL under Cooperative Agreement Number DE-FC26-04NT42080.
Pittsburgh, Pennsylvania. April.
NETL. 2008. "Fate of Mercury in Synthetic Gypsum Used for Wallboard Production: Final Report."
Prepared J. Marshall of USG Corporation and G.M. Blythe and M. Richardson of URS
Corporation for the DOE National Energy Technology Laboratory under Cooperative
Agreement Number DE-FC26-04NT42080. Pittsburgh, Pennsylvania. April.
Norton, L. D. (2011). "Environmental Evaluation of Flue Gas Desulfurization Gypsum as a BMP
for Erosion Control." ISELE Paper Number 11069. Prepared at the International Symposium
on Erosion and Landscape Evolution, September 18-21. Anchorage, AK.
ORNL (Oak Ridge National Laboratory). 1984. "A Review and Analysis of Parameters for Assessing
Transport of Environmentally Released Radionuclides through Agriculture." ORNL-5786.
Prepared by C.F. Baes III, R.D. Sharp, A.L. Sjoreen and R.W. Shor Office for the EPA Office of
Air and Radiation under Interagency Agreement AD-89-F-2-A106. Oak Ridge, TN. September.
OSU-E (Ohio State University-Extension). 2005. "Gypsum for Agricultural Use in Ohio—Sources
and Quality of Available Products." ANR-20-05. Prepared by K. Dontsova, Y.B. Lee, B.L. Slater
and J.M. Bigham for the OSU School of Natural Resources. Columbus, Ohio.
OSU-E. 2011. "Gypsum as an Agricultural Amendment: General Use Guidelines." Bulletin 945.
Prepared by L.C. Chen and W.A. Dick.
Pasini, R. and H. W. Walker. 2012. "Estimating Constituent Release from FGD Gypsum Under
Different Management Scenarios." Fuel. 95:190-196.
Pekney, N.J., D. Martello, K. Schroeder and E. Granite. 2009. "Environmental Chamber
Measurements of Mercury Flux from Coal Utilization By-Products." Fuel. 88:890-897.
Rallo, M., M.A. Lopez-Anton, R. Perry and M.M. Maroto-Valer. 2010. "Mercury Speciation in
Gypsums Produced from Flue Gas Desulfurization by Temperature Programmed
Decomposition." Fuel. 89(8):2157-2159.
Rhoton, F.E., D.S. McChesney and H.H. Schomberg. 2011. "Erodibility of a Sodic Soil Amended
With FGD Gypsum." Presented at the 2011 World of Coal Ash Conference, May 9-12. Denver,
CO.
Sheng, J., A. Adeli, J.P. Brooks, M.R. McLaughlin and J. Read. 2014. "Effects of Bedding Materials
in Applied Poultry Litter and Immobilizing Agents on Runoff Water, Soil Properties, and
Bermudagrass Growth." Journal of Environmental Quality. 43(l):290-296.
Shock, S.S., J.J. Noggle, N. Bloom and L.J. Yost. 2009. "Evaluation of Potential for Mercury
Volatilization from Natural and FGD Gypsum Products Using Flux-Chamber Tests."
Environmental Science and Technology. 43(7):2282-2287.
Schomberg, H.H., D.M. Endale, M.B. Jenkins, R.L. Chaney and D.H. Franklin. 2018. "Metals in
Soil and Runoff from a Piedmont Hay Field Amended with Broiler Litter and Flue Gas
Desulfurization Gypsum." Journal of Environmental Quality. 47:326-335.
Stergarsek, A., M. Horvat, J. Kotnik, J. Tratnik, P. Frkal, D. Kocman, R. Jacimovic, V. Fajon, M.
Ponikvar, I. Hrastel, J. Lenart, B. Debeljak and M. Cujez. 2008. "The Role of Flue Gas
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ -|g
Appendix A: Constituent Data
-------
Desulphurisation in Mercury Speciation and Distribution in a Lignite Burning Power Plant."
Fuel. 87:3504-3512.
Stout, W.L., A.N. Sharpley, W.J. Gburek and H.B. Pionke. 1999. "Reducing Phosphorus Export
from Croplands with FBC Fly Ash and FGD Gypsum." Fuel. 78(2): 175-178.
Tubail, K., L. Chen, F. C. Michel, H. M. Keener, J. F. Rigot, M. Klingman, D. Kost and W. A. Dick
2008. "Gypsum Additions Reduce Ammonia Nitrogen Losses During Composting of Dairy
Manure and Biosolids." Compost Science and Utilization. 16(4):285-293.
U.S. EPA (Environmental Protection Agency). 1999. "Estimating Risk from Contaminants
Contained in Agricultural Fertilizers (Draft Report)." Prepared by RTI International for the
EPA Office of Solid Waste under Contract Number 68-W-98-0085. Washington, DC. August.
US EPA. 2003. "Summary of General Assessment Factors for Evaluating the Quality of Scientific
and Technical Information." EPA 100/B-03/001. Prepared by the EPA Science Policy Council.
Washington, DC. June.
U.S. EPA. 2009. "Evaluating the Thermal Stability of Mercury and Other Metals in Coal
Combustion Residues Used in the Production of Cement Clinker, Asphalt, and Wallboard."
Prepared by the EPA Office of Research and Development. Research Triangle Park, NC.
December.
U.S. EPA. 2010. "Background Information for the Leaching Environmental Assessment
Framework (LEAF) Test Methods." EPA/600/R-10/170. Prepared by A.C. Garrabrants, D.S.
Kosson and F. Sanchez of Vanderbilt University; H.A. van der Sloot of Van der Sloot
Consultancy; and O. Hjelmar of DHI for the EPA Office of Research and Development under
Contract Number EP-C-09-027. Research Triangle Park, NC. November.
U.S. EPA. 2012. "Mined Gypsum LEAF Methods 1313 and 1316 Report: Evaluation of Metal
Leaching from Contaminated Soils, FGD, and other Coal Combustion Byproducts in Reuse
Scenarios." Prepared by Pegasus Technical Services, Inc. for the EPA Office of Research and
Development. Cincinnati, OH. July.
U.S. EPA. 2014b. "Final Human and Ecological Risk Assessment of Coal Combustion Residuals."
RIN: 2050-AE81. Prepared by the EPA Office of Solid Waste and Emergency Response.
Washington, DC. December.
Wang, K. 2012. "Mercury Transportation in Soil Using Gypsum from Flue Gas Desulphurization
Unit in Coal-Fired Power Plant." Thesis. Western Kentucky University.
Wang, K., W. Orndorff, Y. Cao and W. Pan. 2013. "Mercury Transportation in Soil Using Gypsum
from Flue Gas Desulfurization Unit in Coal-Fired Power Plant." Journal of Environmental
Sciences. 25(9):1858-1864.
Xin, M., M. S. Gustin and K. Ladwig (2006). "Laboratory Literature Source of Air-Water-Coal
Combustion Product (Fly Ash and FGD Solid) Mercury Exchange." Fuel. 85(16):2260-2267.
Yodthongdee, Y., P. Sooksamiti, J. Jakmunee and S. Lapanantnoppakhun. 2013. "Uptake of
Nutrients in Vegetables Grown on FGD-Gypsum-Amended Soils." Oriental Journal of
Chemistry. 29(3):1027-1032.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ -jg
Appendix A: Constituent Data
-------
Yost, L.J., S.S. Schock, S.E. Holm, Y.W. Lowney and J.J. Noggle. 2010. "Lack of Complete Exposure
Pathways for Metals in Natural and FGD Gypsum." Human and Ecological Risk Assessment.
16:317-339.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Attachment A-1. Summary of Communications
Date
Contact Name I Affiliation and Title
4/10/15 Ardeshir Adeli USDA, Soil Scientist
Question/Request: Were the gypsum samples in the Sheng et al. (2014) literature source washed or unwashed?
Summary of Response: A full load truck of FGD gypsum was delivered from the power plant either from Georgia
or Alabama; contact was unsure of the process related to the FGD gypsum.
4/10/15 Dr. Souhail Al-Abed U.S. EPA, Research Chemist
Question/Request: Were the gypsum samples in the Al-Abed et al. (2008) literature source washed or unwashed?
Summary of Response: The gypsum was not washed.
4/10/15 Ashok Alva U.S. EPA, Research Soil Scientist
Question/Request: Were the gypsum samples in the Alva et al. (1998a,b) literature source washed or unwashed?
Summary of Response: The FGD gypsum used in the literature source was a byproduct from scrubber from a
power plant near Tampa, FL. The sample was collected from the bulk storage landfill near the plant and there were
no notes on what was done to the FGD gypsum before being sent to storage. The samples underwent no
pretreatment before being used in the experiment.
_ , .. . Institute de Recursos Naturales y Agrobiologfa de Salamanca,
4/10/15 Esther Alvarez-Ayuso , . ' M M
_ Research Associate
Question/Request: Were the gypsum samples in the following studies washed or unwashed?
¦ Alvarez-Ayuso et al. (2006)
¦ Alvarez-Ayuso and Querol (2007)
¦ Alvarez-Ayuso et al. (2008a)
¦ Alvarez-Ayuso et al. (2008b)
¦ Alvarez-Ayuso et al. (2011)
Summary of Response: No response received.
... . Institute Nacional Del Carbon, Professor Chemical Processes in
4/10/15 M. Antonia Lopez-Anton _ , _ .
Energy and Environment
Question/Request: Were the gypsum samples in the Rallo et al. (2010) literature source washed or unwashed?
Summary of Response: The gypsums used in the literature source were taken directly from the combustion plant
as obtained from the wet desulfurization system, unwashed.
4/10/15 Virupax Baligar USDA, Research Soil Scientist
Question/Request: Were the gypsum samples in the Clark et al. (2001) literature source washed or unwashed?
Summary of Response: The FGD was unwashed, as the materials used had been collected at the laboratory. They
did not go to the bottled gypsum one gets from a chemical distributor, theyjust used FGD materials.
4/10/15 Dr. Candace Kairies-Beatty Winona State University, Assistant Professor
Question/Request: Were the gypsum samples in the Kairies et al. (2006) literature source washed or unwashed?
Summary of Response: No response received.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
4/9/15 Dr. Ray Bryant USDA, Research Soil Scientist
Question/Request: Were the gypsum samples in the Bryant et al. (2012) literature source washed or unwashed?
Summary of Response: The partner for the experiment was Constellation Energy (now Raven Power) plant in
Brandon Shores, MD. The plant came on line in 2010 with modern technology. They use a forced oxidation wet
scrubber system after the removal of fly ash. The ditch filter was constructed in 2007 prior to their plant coming on
line. Constellation Energy provided the FGD gypsum from a plant that used a scrubbing process similar to what
they were building at the time, so it should fit this same description. The samples were used as it was delivered and
were not washed.
4/10/15 Liming Chen The Ohio State University, Research Associate
Question/Request: In Chen et al. (2008) data in Table 2 overlap with data in Chen et al. (2009), but the sulfur
values differ between the two sources, is there an explanation?
In Chen et al. (2014) data in Table 3 overlap with data in EPRI (2013), but some values for barium and molybdenum
differ between the two sources, is there an explanation?
Were the gypsum samples in the following studies washed or unwashed?
¦ Chen et al. (2008)
¦ Chen et al. (2009)
¦ Chen et al. (2014)
¦ OSU-E (2011)
¦ Tubail et al. (2008)
Summary of Response: No response received.
4/10/15 Chin-Min Cheng The Ohio State University, Senior Research Associate
Question/Request: Requested measurement data behind Figures 2-5 and Table 3 from Cheng et al. (2012).
Additional requests:
1) Table 1 in the 2012 article indicates that the data for the AFO-gypsum is from your 2009 article (Table 8,1
assume) but the data do not appear to match up between the two sources. Could you please explain and indicate
which source would be the best to use?
2) Figures 2 through 6 and Table 3 in the 2012 article appear to be based on measurements of air, water, soil, and
crops that would be of great interest to our EPA client for incorporation in their database. Would you be able to
supply the actual individual measurements behind these exhibits in a spreadsheet or other tabular format?
3) Were the gypsum samples in the literature source washed or unwashed?
Summary of Response: Attached the data for the Figures 2-6 and provided the data for Table 3 later. In response
to the first question, the data shown in the 2012 paper were from a separate analysis. The statement is not
accurate. They actually took another aliquot of AFO-Gypsum sample and analyzed the chemical composition with
other two FGD samples before the experiment, Cheng attached the data in the spreadsheet. The plant did not use
centrifuge in its process when the literature source was carried out. The gravity belt thickener received gypsum
slurry directly from the underflow of hydroclone. The sample was collected from the disposal end. So, the sample
was unwashed.
4/10/15 Dr. Thomas DeSutter North Dakota State University, Associate Professor
Question/Request: DeSutter et al. (2014) and EPRI (2011 d) overlap, but the Table 3 data in DeSutter et al. (2014)
do not seem to match any in EPRI (2011 d), would like to verify that they are truly different.
Were the gypsum samples in the DeSutter and Cihacek (2009) literature source washed or unwashed?
Summary of Response: Dr. DeSutter attached two documents. On page 305 of the article you will find: "Seed
mass was corrected to 13.5% moisture before reporting." The EPRI article was not corrected for this. They did not
clean the samples with water or any other solvent, the FGD was received from Muscatine in 55 gal steel drums.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
2/05/15 Dr. Warren A. Dick Ohio State University, Professor of Soil Science
Question/Request: There are discrepancies in the pre-treatment data between the 2008 progress report (Tables 4-
3 and 4-4) and the 2011 North Dakota Sites 1 and 2 (Wheat) report (Table 2-1). Specifically, while the Mehlich 3
data match, the 2011 pre-treatment data for EPA method 3051a do not exactly match those of the 2008 tables, for
both the Gary and Wayne sites. Another concern with the EPA3051a data is that the second row (constituents As-
V) appears to have the exact same data between the Gary and Wayne sites for all scenarios.
Summary of Response: Two Excel files were provided that contained the corrected data. Table 4.3 was correct
Table 4.4 was provided with corrected data. Data were transferred properly to Table 4.4 from computer files.
Concerning Table 2-1 (also sent as an attachment), as far as OSU could ascertain the response is as follows. The
EPA 3051a numbers in the table, not in parentheses, are from Tom DeSutter analyses except for Hg. The EPA 3051a
numbers in parentheses were from the same sites and the same sample, but were analyzed by STAR Laboratory at
OSU. For the most part, the values compare very well. The OSU data (in parentheses) are also the same data as in
Table 4.3 and in corrected Table 4.4. The Hg values from STAR Laboratory for Sites 1 and 2 were reversed and this
has been corrected in the revised Table 2-1.
2/17/15 Dr. Warren A. Dick Ohio State University, Professor of Soil Science
Question/Request:
Issue 1. What are the site names associated with the "Site 1" and "Site 2" designations in EPRI 2011 Table 2-1?
Comparing the data values in EPRI 2011 Table 2-1 to those in EPRI 2008 Tables 4-3 and 4-4, as well as EPRI 2011
Tables 3-4 and 3-5, the site names for the designations are not consistent, as follows:
¦ EPRI 2008 site 1 = Wayne
¦ EPRI 2011 site 1= Gary
¦ EPRI 2008 site 2 = Gary
¦ EPRI 2011 site 2 = Wayne
¦ EPRI 2008 Table 4-3 (Gary) replicates data found in EPRI 2011 Tables 3-4 and 2-1, but matches Site 1 data in
Table 3-4 and Site 2 data in Table 2-1
¦ EPRI 2008 Table 4-4 (Wayne) replicates data found in EPRI 2011 tables 3-5 and 2-1, but matches site 2 data in
Table 3-5 and site 1 data in Table 2-1
Issue 2. As shown below, there are a few data discrepancies between the 2008 and 2011 reports; which is the
correct value in each case?
¦ EPRI 2008 Table 4-3: EPA3051a value for Check 0 Sr is 26.98, but the value from 2011 Table 3-4 EPA 3051a is
30.0. Is this rounding?
¦ EPRI 2008 Table 4-4: EPA3051 a value for Check 0 P is 311.5, but the value from 2011 Table 3-5 EPA 3051a is
31.5. 311.5 is more in line with the other measurements - is this the correct value?"
¦ EPRI 2008 Table 4-4: EPA3051a value for Check 0 S is 269, but the value from 2011 Table 3-5 EPA 3051a is
6269. We suspect that 269 is correct for both because of the zero application rate. Is this correct?
Summary of Response: A file was provided with responses. It was stated that Gary and Wayne were used as site
descriptors. The researchers did not recall how the descriptors were changed to Site 1 and Site 2 and then there
was a mix-up.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
2/20/15 Dr. Warren A. Dick Ohio State University, Professor of Soil Science
Question/Request: Tables 2-5 and 2-6 in EPRI (2012c) - FGD gypsum Ag Network NMexico 1 Alfalfa and 2 Sodic
Soils [1025355].pdf are missing headers for the second set of results (see attached Word file). Could you supply
[headers] or let us know if they should be the same as in Table 2-7.
Also, there are all zeros in a few tables here and there for non-detects and apparent detections that appear to be
truncations (e.g., "<0.000" or "0.000").
What is the best way to deal with these? (Some may be other authors - we can contact them)
Summary of Response: Dr. Warren provided two tables submitted to EPRI and that included all of the column
headings for Table 2-5 and Table 2-6. The numbering of the tables, themselves, is slightly different due to changes
made by EPRI when they reformatted the report for publication. The last page of the attached file showed the
detection limits that were mostly used in the reports. These detection limits were followed as much as possible in
their reports to give some consistency among the reports as the day-to-day detection limits did vary in a very
slight way. Trying to come up with values of samples being analyzed on different days near these detection limits
causes problems in reporting and so the Table A-2 (or in some reports Table B-2) was used. If there are any
questions about detection limits, refer to the Table A-2. This would include the question about the Mn value in
Table A-10 from the 2013 EPRI Report where Mn should be reported as <0.001 mg/L.
The EPRI Table 2011 b is not the work of Ohio State University. Dr. Warren believes that the data are from leaching
measurements done for Ken Ladwig. Dr. Warren also reported that Soils Data 2009 2011 FGD Gypsum Watkinsville
Data for EPA.xIsx, worksheet RSP0911 Soil Rufus gkg, is not their work. He believes that these data are from
USDA-ARS in Watkinsville, GA. The other detection limits or values listed are also from other reports that were not
from Ohio State as they did not do any analyses of samples from Watkinsville.
4/13/15 Dr. Warren A. Dick Ohio State University, Professor of Soil Science
Question/Request: Kost et al. (2014) and EPRI (2012b) overlap, but Table 1 data in Kost et al. (2014) do not all
match those in EPRI (2012b) Table 3-1, is there an explanation? Were the gypsum samples in the Kost et al. (2014)
literature source washed or unwashed?
Summary of Response: No response received.
5/28/15 Dr. Warren A. Dick Ohio State University, Professor of Soil Science
Question/Request: Specifically, in Ohio State Extension Bulletin 945 (Table 1-3), Arsenic in mined gypsum is
reported as 462 ppm. However, the primary data source cited (Dontsova et al 2005, Table 4) reports Arsenic in
natural gypsum as < 0.52 ppm. Was there some type of conversion performed that was not documented or was it
recorded incorrectly?
Summary of Response: Table 1 -3 in the OSU Extension Bulletin is not properly created. The data in Table 1 -3 are
not from Dontsova et al, 2005. The FGD gypsum data came from a 2005 Agronomy Journal publication (attached
publication). The source of this gypsum is given in the publication. The arsenic number seems extremely high but
unfortunately there are no samples archived from that work. Dr. Dick assumed that all of the As numbers from the
Agronomy Journal article were originally in ug/kg and somehow they got converted to mg/kg without moving the
decimal. If that is the case, the values would be 0.119, 0.363 and 0.462 mg/kg. These values are much more in line
with anything they have ever analyzed. The FGD gypsums values in Table 1-3 seem to be a composite of values
from various samples, but he only had values associated with this particular table (i.e., Table 1 -3).
4/10/15 Dinku Endale USDA, Agricultural Engineer
Question/Request: Were the gypsum samples in the literature source washed or unwashed?
Summary of Response: The FGD gypsum used came directly from Duke Energy's Marshall Steam Station, in
Terrell, North Carolina, mid-March 2009, and stock piled under cover on site. The samples were applied directly to
the plots each year from this pile. Table 1 of the paper gives some nutrient content values. They were unsure of
washing during the manufacturing process and suggested contacting Duke Energy to try and find out. Their
understanding was that the FGD gypsum was high grade.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
4/10/15 James Grichar Texas A&M University, Senior Research Scientist
Question/Request: Table 3 units for yield are given as are these the correct units?
Were the gypsum samples in the Grichar et al. (2002) literature source washed or unwashed?
Summary of Response: No, should be lbs/acre. The gypsum was unwashed.
_ University of Nevada, Reno, Prof., Environmental and Resource
4/10/15 Dr. MaeGustin *
Sciences
Multiple questions posed via a single email.
Question/Request 1: Requested the data behind the figures in Gustin and Ladwig (2010).
Were the gypsum samples in the Gustin and Ladwig (2010) literature source washed or unwashed?
Summary of Response 1: Provided an Excel spreadsheet with the mercury data for Figure 1 but not for other
figures. Gustin stated that she has all the raw data files, but cannot just turn them over, because they were paid for
by EPRI and she will need to get verification from them before she can provide them. The raw data files also do not
have the same labels as in the paper. Thus, this will take time for her to organize. If RTI needs to use the graphs in
the paper then they can get copyright permission from EST. She also stated that it would also take her several
hours to get the raw data in a format that RTI could interpret given the plant labels in the paper are different than
those in the raw data files, and she did not have time to do this.
Table 1 of Gustin and Ladwig (2010) states specifically whether the gypsum was washed or not.
Question/Request 2: For Briggs et al. (2014), are the zero values in Table 1 for Methylmercury in Alabama FGD
gypsum and Ohio FGD gypsum correct? What was the detection limit that was used?
Were the gypsum samples in the Briggs et al. (2014) literature source washed or unwashed?
Summary of Response 2: The values were 0 based on our detection limit. "Methylmercury concentrations in FGD
gypsum materials were 1.1 ± 0.1 ng/g for Indiana FGD gypsum and below detection for the OH and AL FGD
gypsum and for the GYP. Clarification email received on 4/13/2015 stating that the Indiana and Ohio FGD gypsums
were washed, and the Alabama gypsum was unwashed.
Question/Request 3: Were the gypsum samples in the Xin et al. (2006) literature source washed or unwashed?
Summary of Response 3: No response received.
4/10/15 Loreal Heebink Energy and Environmental Research Center, Research Chemist
Question/Request: Were the gypsum samples in the EERC (2003) literature source washed or unwashed? Were
the gypsum samples in the EERC (2007) literature source washed or unwashed?
Summary of Response: In regards to EERC (2003), the results shown in "Review of handling and use of FGD
material" were taken from literature and the author does not remember any of the references specifying whether
the material was washed. No response received for the EERC (2007) question.
4/10/15 Dr. Milena Horvat Jozef Stefan Institute, Department Head, Environmental Sciences
Question/Request: Were the gypsum samples in the Stergarsek et al. (2008) literature source washed or
unwashed?
Summary of Response: No response received.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
2/05/15 Ken Ladwig EPRI
Question/Request: When we have questions about data in some of the EPRI FGD/Ag Network reports should I
contact you, or are you comfortable with us going to the report authors?
Were the gypsum samples in the EPRI studies washed or unwashed?
Summary of Response: Permission provided to go directly to Warren Dick of Ohio State University.
EPRI (2008): This was preliminary data for the sites which eventually had full reports. You should refer to the full
reports rather than these data.
¦ EPRI (2011c): Unwashed
¦ EPRI (2011 d): Unwashed
¦ EPRI (2012b): Washed
¦ EPRI (2012c): Unwashed
¦ EPRI (2013): Washed
¦ EPRI (2013f): Washed
6/30/15 Ken Ladwig EPRI
Question/Request: Requested raw data behind Table 2 and Figure 3 in Gustin and Ladwig (2010) for Plants B, C,
and G.
Summary of Response: Attached six Excel spreadsheets with the raw data for Plants B, C, and G.
4/10/15 Somchai Chiang Mai University, Prof., Department of Chemistry
Lapanantnoppakhun
Question/Request: Were the gypsum samples in the Yodthongdee et al. (2013) literature source washed or
unwashed?
Summary of Response: Used unwashed gypsum for the literature source.
4/10/15 Dr. Joo-Youp Lee University of Cincinnati, Chemical Engineering Asst. Professor
Question/Request: In Lee et al. (2009), what were the detection limits for the BDL values in Table 2?
Summary of Response: No response received.
4/10/15 Daniel McChesney USDA, Soil Scientist
Question/Request: Were the gypsum samples in the Rhoton et al. (2011) literature source washed or unwashed?
Summary of Response: No response received.
4/10/15 Charles Miller U.S. DOE, National Energy Technology Laboratory
Question/Request: Were the gypsum samples in the NETL (2005) literature source washed or unwashed?
Summary of Response: Could not answer the question with certainty. The gypsum used for the testing was taken
out of USG's inventory of feedstock at the wallboard plant, and that the wallboard met USG's acceptance criteria.
However, those acceptance criteria include a specification on chloride concentrations in the feedstock, requires
efficient washing of the gypsum to meet wallboard feedstock specifications.
For example, typical specs are around 100 ppm chloride max and 10% free moisture. If the cake is unwashed, this
10% moisture would be FGD liquor, so the FGD liquor would be limited to only 1000 ppm chloride to meet that
spec. Mr. Miller felt certain that the FGD system that provided the gypsum for Wallboard Test 1 operates with a
considerably higher chloride concentration than 1000 ppm. Therefore, he was relatively certain this gypsum was
washed as it was dewatered.
4/10/15 Dr. Darrell Norton Purdue University, Professor (Emeritus)
Question/Request: Norton (2011) and EPRI (2013f) overlap, but there are data discrepancies between the two
data sources, is there an explanation?
Summary of Response: No response received.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Date
Contact Name
Affiliation and Title
4/10/15 Natalie Pekney U.S. DOE National Energy Technology Laboratory
Question/Request: Were the gypsum samples in the Pekney et al. (2009) literature source washed or unwashed?
Summary of Response: Pekney was unable to obtain the information. To avoid bias she was not informed of the
source, or specific power plant, from which the samples came.
4/10/15 Dr. Andrew Sharpley University of Arkansas, Professor
Question/Request: Were the gypsum samples in the Stout et al. (1999) literature source washed or unwashed?
Summary of Response: Did not know the answer to the question, other than they did not wash it and it would
have had to have been at the point of generation. Unable to find out any more information because the senior
author of this research and person who designed the literature source and got the material, passed away in 2001,
and the last two authors retired about 14 years ago and have had no ongoing contact with the research
community.
4/10/15 Allen Torbert USDA, Supervisory Soil Scientist
Question/Request: Verify the detection limits for the "ND" results in Table 2 for Gypsum, B, Cu, Mn of the Torbert
and Watts (2014) literature source.
Were the gypsum samples in the Torbert and Watts (2014) literature source washed or unwashed?
Summary of Response: The detection limits used were B <50 mg/kg; Cu < 0.8 mg/kg; and Mn < 2 mg/kg. The
gypsum used in this literature source was washed.
4/10/15 Kelin Wang Louisiana State University, Graduate Research Assistant
Question/Request: For Wang (2012), verify that for Table 27 control value given as 0 ppb; Table 28 control value
given as 0 ug. For Wang et al. (2013), provide the detection limits for Table 6.
Summary of Response: They are correct. She did not detect the mercury in control chamber. They are below the
detection limit. The detection limit is 0.1 ppb. The detection limit of table 27 and 28 is 0.1 ppb.
4/10/15 Dr. Harold Walker Stony Brook University, Director, Civil Engineering Program
Question/Request: Were the gypsum samples in the Pasini and Walker (2012) literature source washed or
unwashed?
Summary of Response: They did not perform any washing procedure prior to use in leaching experiments and are
not aware of the utility washing the material either. It was their understanding that the material they received was
consistent with the material being disposed of in the landfill.
2/12/15 Dr. Lisa Yost Environ (formerly at Exponent)
Question/Request: We are collecting compositional data on natural and FGD (synthetic) gypsum for the U.S. EPA.
One of our data sources is your 2010 article published in Human and Ecological Risk Assessment, "Lack of
complete exposure pathways for metals in natural and FGD gypsum". In Tables 1 and 2 you provide compositional
data for natural and synthetic gypsum, respectively. However the results are presented as statistics (e.g., means,
min, max) of multiple samples. Would it be possible to obtain the individual sample results for the data in Tables 1
and 2? If so, a spreadsheet format would be ideal, although we can use any tabular format. Along with the
constituent concentrations, we also are interested in the detection limits for any non-detects and any sample
identification information (e.g., location, site name, sample ID, plant, process, dates) you can provide so we can
properly distinguish the samples in the database.
We also had a question on the analytical methods. What procedure (e.g., U.S. EPA Method 3051, or 3051A) was
used to digest/solubilize the sample prior to sample analysis?
Summary of Response: Dr. Yost replied that she was the author and would like to help but she needs her client's
permission which is complicated by the fact that the client is no longer at that the previous job. However, she
stated that she would give it a try. (On 3/19/2015, Dr. Yost suggested a call. RTI and EPA suggested dates and
times for a call but received no response).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Attachment A-2. Concentration Summary
The tables below provide summary statistics on the bulk and leachate concentrations in FGD
gypsum. The summary statistics presented below reflect the raw data after the filtering discussed
in this appendix. All of the data is weighted equally and does not account for the prevalence of
different environmental conditions captured in the full-scale model, such as variations in soil pH.
Summary of Filtered Bulk Concentration Data
Constituent
Bulk Concentration (mg/kg)
Min
Median
Mean
St. Dev
Geo Mean
Max
All Samples
Aluminum
61
380
1,232
2,287
502
12,700
Antimony
0.06
0.60
2.6
4.2
0.76
24
Arsenic
0.19
2.8
3.2
2.4
2.2
11
Barium
0.74
12
22
21
13
82
Beryllium
0.01
0.03
0.05
0.10
0.03
0.60
Boron
0.76
10
33
67
13
387
Cadmium
0.01
0.13
0.24
0.31
0.13
1.9
Chloride
34
480
687
621
427
2,616
Chromium
0.10
2.9
4.4
3.8
3.0
15
Cobalt
0.04
0.31
0.80
1.0
0.40
4.3
Copper
0.001
1.3
1.5
1.0
0.95
4.1
Iron
222
1,000
1,161
852
952
5,881
Lead
0.002
1.0
1.5
1.5
1.0
8.3
Magnesium
70
1,322
1612
1,566
1,001
7,430
Manganese
0.03
8.75
22
33
8.9
161
Mercury
0.01
0.34
0.46
0.50
0.30
3.1
Molybdenum
0.11
0.95
1.6
2.0
1.0
12
Nickel
0.09
1.3
1.6
1.4
1.3
6.8
Selenium
0.73
5.4
9.4
9.5
6.1
46
Strontium
71
161
197
104
174
531
Thallium
0.002
0.02
0.30
0.58
0.05
2.8
Vanadium
0.15
1.9
3.4
4.8
2.2
30
Zinc
1.6
6.2
9.1
6.8
7.0
29
Unwashed Only
Aluminum
142
959
2,080
3,028
1,035
12,700
Antimony
0.08
0.66
2.4
3.0
0.95
9.1
Arsenic
0.95
3.0
3.6
2.3
3.1
11
Barium
2.4
27
33
25
22
82
Beryllium
0.01
0.04
0.06
0.05
0.03
0.13
Boron
9.4
51
100
108
64
387
Cadmium
0.02
0.22
0.31
0.27
0.21
1.2
Chloride
81
833
1,209
1,177
790
4,816
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Summary of Filtered Bulk Concentration Data
Constituent
Bulk Concentration (mg/kg)
Min
Median
Mean
St. Dev
Geo Mean
Max
Chromium
1.0
5.4
6.4
4.5
4.9
17
Cobalt
0.22
1.1
1.4
1.2
0.98
4.4
Copper
0.38
1.93
1.9
0.96
1.6
3.5
Iron
635
1,512
1,711
1,006
1,540
5,881
Lead
0.63
1.2
1.6
0.84
1.4
3.3
Magnesium
201
1,927
2,677
2,227
1,777
7,430
Manganese
8.7
27
49
54
28
161
Mercury
0.01
0.41
0.58
0.53
0.38
2.3
Molybdenum
0.54
1.8
2.6
2.7
1.8
12
Nickel
0.84
2.2
2.1
1.2
1.9
5.0
Selenium
0.73
6.6
11
9.1
7.4
33
Strontium
97
172
215
117
192
534
Thallium
0.01
0.28
0.48
0.60
0.14
2.3
Vanadium
1.6
3.2
6.3
8.5
3.9
30
Zinc
1.8
9.0
10
7.1
7.5
23
Washed Only
Aluminum
61
256
1,289
2.777
397
11,600
Antimony
0.06
0.5
2.3
4.8
0.58
24
Arsenic
0.27
3.1
3.5
2.4
2.7
10
Barium
0.74
9.3
15
16
9.7
53
Beryllium
0.01
0.03
0.05
0.13
0.02
0.6
Boron
2.2
8.6
11
9.0
CO
CO
44
Cadmium
0.01
0.12
0.17
0.15
0.10
0.47
Chloride
15
219
271
328
127
1,255
Chromium
0.60
2.5
4.3
4.6
2.9
20
Cobalt
0.07
0.23
0.91
1.3
0.39
4.2
Copper
0.001
1.0
1.2
0.94
0.71
4.1
Iron
277
808
979
510
853
2,114
Lead
0.002
1.0
1.9
2.6
0.91
12
Magnesium
58
989
1,143
909
800
4,134
Manganese
0.97
7.5
14.8
19
7.9
79
Mercury
0.007
0.36
0.50
0.52
0.28
3.1
Molybdenum
0.15
0.97
1.7
19
0.93
8.1
Nickel
0.35
1.2
1.2
0.44
1.1
2.2
Selenium
2.2
5.03
10
11
6.6
46
Strontium
88
173
212
108
189
527
Thallium
0.002
0.02
0.30
0.57
0.05
2.8
Vanadium
0.67
1.6
2.4
2.3
1.8
10
Zinc
2.1
7.0
9.0
6.0
7.3
27
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^
Appendix A: Constituent Data
-------
Summary of Filtered Leachate Concentration Data
Constituent
Leachate Concentration (Mg/L)
Min
Median
Mean
St. Dev
Geo Mean
Max
All Samples
Aluminum
2.0
220
723
1,352
237
10,282
Antimony
0.02
2.8
3.7
5.3
1.9
30
Arsenic
2.5
3.2
8.1
23
3.9
197
Barium
1.9
76
88
81
65
565
Beryllium
0.23
3.2
2.6
1.3
2.0
7.0
Boron
3.9
417
2,826
5,222
447
22,396
Cadmium
0.01
0.85
3.9
8.2
1.1
55
Chloride
2065
20,229
91,515
110,872
24,766
344,368
Chromium
0.05
9.00
10.9
16
5.1
158
Cobalt
0.37
7.00
9.5
13
5.1
69
Copper
0.50
3.50
11
17
5.5
95
Iron
0.43
4.80
2,268
5,117
36
27,320
Lead
0.07
1.2
7.9
12
1.9
31
Magnesium
290
13,890
62,672
102,178
13,405
525,800
Manganese
1.7
273
1,710
3615
253
23,659
Mercury
0.0004
0.01
0.10
0.34
0.01
3.3
Molybdenum
0.36
6.2
20
36
8.1
170
Nickel
0.05
45
81
97
20
434
Selenium
2.9
67
213
418
77
2,478
Strontium
114
539
914
764
668
3,100
Thallium
0.01
2.6
4.5
5.8
1.7
34
Vanadium
0.04
8.0
31
94
5.5
657
Zinc
1.0
165
222
275
96
1,641
Unwashed Only
Aluminum
5.0
385
716
926
293
3,790
Antimony
0.16
2.8
3.6
2.8
2.9
17
Arsenic
2.5
3.2
14
35
5.2
197
Barium
22
88
109
82
91
445
Beryllium
0.31
3.2
3.1
0.60
2.9
3.2
Boron
58
2,214
5,282
6,691
1,922
22,396
Cadmium
0.05
1.9
7.0
12
2.6
55
Chloride
10,822
143,264
165,538
114,436
105,966
344,368
Chromium
0.16
9.0
13
23
7.3
158
Cobalt
2.1
9.0
13
15
7.3
58
Copper
1.6
8.0
15
18
8.7
92
Iron
1.5
443
2,946
4,929
120
23,590
Lead
0.08
1.2
6.4
11
2.1
31
Magnesium
1,148
5,785
104,169
121,151
44,344
525,800
Manganese
1.7
1,206
3,254
4,899
920
23,659
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix A: Constituent Data
-------
Summary of Filtered Leachate Concentration Data
Constituent
Leachate Concentration (Mg/L)
Min
Median
Mean
St. Dev
Geo Mean
Max
Mercury
0.001
0.01
0.14
0.48
0.01
3.3
Molybdenum
3.8
12
28
45
12
170
Nickel
0.05
98
124
111
66
434
Selenium
13
189
291
430
159
2,478
Strontium
114
596
924
710
684
2,772
Thallium
0.01
2.6
6.9
7.4
3.9
34
Vanadium
1.4
13
53
135
13
657
Zinc
3.4
215
329
355
205
1,641
Washed Only
Aluminum
2.0
171
847
1,769
251
10,282
Antimony
0.02
2.8
2.2
1.4
1.3
7.0
Arsenic
2.5
3.2
4.1
5.8
3.2
44
Barium
8.9
60
73
79
54
565
Beryllium
0.31
3.2
2.6
1.4
1.9
7.0
Boron
3.9
209
840
2,002
156
7,900
Cadmium
0.01
0.85
1.7
1.8
0.77
7.0
Chloride
2,065
2,065
20,711
36,941
6,166
116,409
Chromium
0.05
9.0
11
8.9
5.0
34
Cobalt
2.1
2.1
6.7
11
4.2
69
Copper
0.67
3.5
9.7
16
5.0
95
Iron
1.4
1.6
1,991
5,667
16
27,320
Lead
0.07
1.2
9.0
13
1.8
31
Magnesium
300
4,397
31,396
70,825
5,565
353,800
Manganese
1.7
81
493
900
89
4,902
Mercury
0.0004
0.02
0.06
0.15
0.01
0.83
Molybdenum
0.36
3.8
15
25
6.5
92
Nickel
0.05
37
52
66
11
310
Selenium
2.9
45
181
435
56
2,064
Strontium
290
516
945
851
680
3,100
Thallium
0.01
2.6
2.6
2.8
0.93
13
Vanadium
0.04
7.0
14
21
3.1
69
Zinc
3.8
150
154
123
77
492
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix A: Constituent Data
-------
Appendix B. Benchmarks
This appendix describes the approach used to select the chronic benchmarks used in this beneficial
use evaluation to estimate the potential for adverse impacts to human and ecological receptors. An
adverse effect is any abnormal, harmful, or undesirable change that results from exposure to a
chemical constituent or other stressor. It is important be aware that the benchmarks considered
relevant and appropriate for this evaluation may not be the same as those other beneficial use
evaluations. In some cases, other appropriate benchmarks may be available or have already been
defined by state or federal regulatory bodies based on the intended use.
B.1 Human Health Benchmarks
Adverse health effects for human receptors are divided into two main categories: carcinogenic and
noncarcinogenic. Carcinogenic effects are those that result in the development of cancer
somewhere in the organism. Noncarcinogenic effects are those that result in any adverse health
effect other than cancer. Some stressors may result in both carcinogenic and noncarcinogenic
effects, depending on the route through which the receptor is exposed and the magnitude of the
exposure. EPA used Equation B.1 and Equation B.2 to calculate noncancer hazard quotient and
cancer risk, respectively. These equations do not include any additional factors required to
harmonize units.
C ia ¦ EF ¦ IR ¦ ABS
(B.1)
HQ
v BW ¦ RfD
C ia ¦ CSF ¦ EF ¦ ED ¦ IR ¦ ABS
(B.2)
Risk —zr~r
AT-BW
Where:
ABS
- Absorption factor (%)
AT
- Averaging time (yr)
BW
- Body weight (kg)
CSF
- Cancer slope factor (mg/kg-day)-1
C
ia - Concentration in a given media (mg/kg or mg/L)
ED
- Exposure duration (yr)
EF
- Exposure frequency (days/yr)
IR
- Intake rate (mg/day or L/day)
RfD
- Reference dose (mg/kg-day)
The equations presented above are generalized, and can be further refined to address indirect
exposures to contaminated media by substituting in Equations B.3 through B.7 for specific
variables in Equations B.1 and B.2. These equations can be applied as presented to calculate risks
from a specified constituent concentration (Section 6: Risk Modeling), or rearranged to calculate
the concentration that corresponds to a specified risk (Section 5: Screening Analysis). The
Beneficial Use Evaluation of FGD Gypsum in Agriculture g ^
Appendix B: Benchmarks
-------
following subsections describe the values used in this evaluation for each of the variables listed in
these equations.
(B.3)
(B.4)
(B.5)
(B. 6)
(B. 7)
Where:
BCF
C
^-Fish Cwat r[(Ft3 ' BCFT3) + (FT4 ¦ BCFT4)]
Cpro uc BCFPro uc ¦ CSoil(l — MAF)
-B f
^Milk
BCF
B f
BCF
Milk
(C ¦ Q ¦ Osoil + (Csoil) (I BCF; ¦ Q( ¦ f()
(C ¦ Q ¦ Osoil + (Csoil) (I BCF; ¦ Qi ¦
i= Grain,Forag ,Silag
i= Grain,Forag ,Silag
^¦Skin ' ^Wat r(Kp)(EV)(tgv nt)(SA)
Bioconcentration factor (unitless)
ia - Concentration in media (mg/kg for fish, produce, beef) (mg/L for water, milk)
Cskin ' IR ~ Uptake rate through skin (mg/day)
EV - Event Frequency (events/day)
f - Fraction of Media Contaminated (unitless)
FT3 - Fraction of Fish Ingested from Trophic Level 3 (unitless)
FT4 - Fraction of Fish Ingested from Trophic Level 4 (unitless)
Kp - Dermal Permeability Coefficient (cm/hr)
MAF - Moisture Adjustment Factor (%)
Q - Ingestion Rate by Cow (kg/day)
SA - Skin Area (cm2)
t£v nt ~ Duration of Individual Exposure Event (hr/events)
B.1.2 Target Hazard Quotient and Risk
Target hazard quotient and target risk are unitless numbers that represent the estimated likelihood
that a non-carcinogenic or carcinogenic adverse effect will occur. Target hazard quotients,
calculated for non-carcinogenic constituents, are the ratio of the constituent concentration to
which a receptor may be exposed and the concentration below which no adverse effects are known
or anticipated to occur. For the screening analysis, the target hazard quotient was set to 1.0 based
on the recommendations of Risk Assessment Guidance for Superfund (US EPA, 1989). Target risks
are established for carcinogenic constituents. Unlike approaches for assessing some non-
carcinogenic constituents, this approach assumes that there is some risk of cancer at any level of
exposure. Any increase in exposure to a carcinogen translates to some increased probability of
developing cancer. The current evaluation considered cancer risks within the 1x10 4 and lxlO"6
risk range. From this range, the specific target risk of lxlO 5 was selected based on the US EPA
Office of Resource Conservation and Recovery's presumptive listing benchmark (59 FR 66075).
This level is equivalent to one additional incidence of cancer for every 100,000 individuals exposed
to a given carcinogen.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g ^
Appendix B: Benchmarks
-------
B.1.2 Toxicity Values
Human health benchmarks are based on specific adverse effects that may occur. Reference doses
(RfDs) and reference concentrations (RfCs) are used to evaluate noncancer effects from oral and
inhalation exposures, respectively. RfDs and RfCs are estimates of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without appreciable risk of
deleterious noncancer effects. However, an average lifetime exposure above the RfD (or RfC) does
not imply that an adverse health effect will necessarily occur. Oral cancer slope factors (CSFs) are
used to evaluate carcinogenic effects from oral exposures. The CSF is an upperbound estimate
(approximating a 95% confidence limit) of the increased human cancer risk from a lifetime of
exposure.
EPA identified toxicity values according to the hierarchy established in the 2003 Office of Solid
Waste and Emergency Response Directive 9285.7-53, which encourages prioritization of toxicity
values from sources that are current, transparent and publicly available, and that have been peer
reviewed (U.S. EPA, 2003). Accordingly, a three-tiered approach was followed to use higher
priority data sources based on availability. Values in lower tiers may not be calculated in the same
way as RfDs, RfCs and CSFs, but are treated as equivalent for the purposes of this evaluation.
Tier I
Integrated Risk Information System (IRIS) contains RfDs and RfCs for chronic noncarcinogenic
health effects, and oral CSFs for carcinogenic effects. IRIS is considered the highest quality science-
based, developed to support EPA regulatory activities. IRIS assessments have been peer-reviewed
and represent Agency-wide consensus.
Tier II
Provisional Peer-Reviewed Toxicity Values (PPRTVs) are derived by the Superfund Program after
a review of the relevant scientific literature using the methods, sources of data and guidance for
value derivation used by the EPA IRIS Program. All provisional peer-reviewed toxicity values
receive internal review by EPA scientists and external peer review by independent scientific
experts. However, PPRTVs do not reflect Agency-wide consensus, because PPRTVs are developed
specifically for the Superfund Program. PPTRVs include cancer and noncancer values for both oral
and inhalation exposure and are treated as equivalent to RfDs, RfCs and CSFs.
Tier III
Agency for Toxic Substances and Disease Registry (ATSDR) Chronic Minimal Risk Levels (MRLs)
are substance-specific health guidance levels for noncarcinogenic effects. An MRL is intended to
be an estimate of the daily human exposure to a hazardous substance that is likely to be without
appreciable risk of adverse noncancer health effects over a specified duration of exposure. MRLs
are derived for oral and inhalation routes of exposure in a manner similar to RfDs and RfCs,
respectively. MRLs have undergone both internal and external peer review.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
New Jersey Department of Environmental Protection (NJDEP, 2009) provides an oral cancer
benchmark for chromium (VI) for their soil cleanup program based on information from a study
by the National Toxicity Program (NTP, 2008).
For lead, EPA currently has no consensus on the development of an RfD or CSF because of the
difficulty associated with identifying an effect threshold needed to develop these benchmarks.
Therefore, the maximum contaminant level (MCL) for drinking water was used as an alternative
method of approximating human health risk.
Under this evaluation, only oral benchmarks were collected for all constituents except for
elemental mercury which is volatile and thus was evaluated for vapor inhalation as described in
Appendix D (Screening Analysis). The chronic oral human health toxicity values used in this
evaluation are summarized in Table B-1. Values were last reviewed in January 2019.
Table B-1. Chronic Oral Human Health Toxicity Values
Constituent
CASRN
Value
Target Organ
Type
Citation
Cancer (mg/kg-day)"1
Arsenic
7440-38-2
1.50E+00
Cancer
IRIS
U.S. EPA (1995a)
Chromium (VI)
18540-29-9
5.00E-01
Cancer
NJDEP
NJDEP (2009)
oncancer (mg/kg-day)
Aluminum
7429-90-5
1E+00
Neurological
PPRTV
U.S. EPA (2006a)
Antimony
7440-36-0
4E-04
Hematological
IRIS
U.S. EPA (1987)
Arsenic
7440-38-2
3E-04
Dermal, Cardiovascular
IRIS
U.S. EPA (1991a)
Barium
7440-39-3
2E-01
Kidney
IRIS
U.S. EPA (2005a)
Beryllium
7440-41-7
2E-03
Gastrointestinal
IRIS
U.S. EPA (1998a)
Boron
7440-42-8
2E-01
Developmental
IRIS
U.S. EPA (2004)
Cadmium
7440-43-9
1E-03 (Food)
5E-04 (Water)
Kidney
IRIS
U.S. EPA (1989)
Chromium (III)
16065-83-1
1.5E+00
No system effect in lab study.
Respiratory, Immunological
IRIS
U.S. EPA (1998b)
Chromium (VI)
18540-29-9
3E-03
No system effect in lab study.
Respiratory, Gastrointestinal,
Immunological, Hematological,
Reproductive, Developmental
IRIS
U.S. EPA (1998c)
Cobalt
7440-48-4
3E-04
Thyroid
PPRTV
U.S. EPA (2008)
Copper
7440-50-8
1E-02
Gastrointestinal
ATSDR
ATSDR (2004)
Iron
7439-89-6
7E-01
Gastrointestinal
PPRTV
U.S. EPA (2006)
Manganese
7439-96-5
1.4E-01
Nervous System / CNS Effects
IRIS
U.S. EPA (1995b)
Mercury (II)
7487-94-7
3E-04
Immunological, Urinary
IRIS
U.S. EPA (1995c)
Mercury (Methyl)
22967-92-6
1E-04
Nervous System, Developmental
IRIS
U.S. EPA (2001)
Molybdenum
7439-98-7
5E-03
Urinary
IRIS
U.S. EPA (1992a)
Nickel
7440-02-0
2E-02
Body Weight, Cardiovascular, Liver
IRIS
U.S. EPA (1991b)
Selenium
7782-49-2
5E-03
Dermal, Hematological, Nervous System
IRIS
U.S. EPA (1991c)
Strontium
7440-24-6
6E-01
Bone/Teeth, Musculoskeletal
IRIS
U.S. EPA (1992b)
Thallium
7440-28-0
1E-05
Hair follicular atrophy
PPRTV
U.S. EPA (2012)
Vanadium
7440-62-2
5E-03
Dermal
IRIS
U.S. EPA (1988)
Zinc
7440-66-6
3E-01
Immunological, Hematological
IRIS
U.S. EPA (2005b)
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
B.1.2 Exposure Factors
Exposure factors are data that quantify human behavior patterns (e.g., ingestion rates of drinking
water and fish) and physiological characteristics (e.g., body weight) that affect an individual's
exposure to environmental contaminants. These data can be used to construct realistic assumptions
concerning the magnitude of exposure to and subsequent intake of a contaminant in the
environment. The exposure factors data also enables EPA to differentiate the exposures of
individuals of different ages. The derivation of human exposure factors used in both the screening
and probabilistic analyses are described below.
The Agency relied primarily on the Exposure Factors Handbook (EFH) (U.S. EPA, 2011) and the
Child-Specific Exposure Factors Handbook (CSEFH) (U.S. EPA, 2008). Where sufficient data were
available, the percentiles and corresponding data points obtained from these two sources were used
to develop a cumulative distribution in order to capture variability within the U.S. population.
Otherwise, EPA relied on point values selected based on the recommendations of the EFH, CSEFH
or established Agency guidance (U.S. EPA, 1991; 2014).
There has been considerable effort across the Agency to improve the accuracy and consistency of
childhood exposure assessments. In the Guidance on Selecting Age Groups for Monitoring and
Assessing Childhood Exposures to Environmental Contaminants (U.S. EPA, 2005c), EPA
recommended specific age cohorts (i.e., groups) intended to better capture the large variability in
physiological and behavioral characteristics of child receptors during different stages of
development. Narrower age cohorts were identified where rapid developmental changes occur,
while broader age groups were identified where the rate of development decreases. These age
groupings and the supporting rationale for their selection have been subjected to internal and
external scientific peer review. In total, receptors were divided into the eight distinct age cohort
groupings recommended by U.S. EPA (2005c). The general methodology for collecting human
exposure data for the probabilistic analysis relied on the EPA data from the 2011 EFH or 2008
CSEFH in one of three ways:
1. When the available data were adequate (as for most input variables), nonparametric
approaches were used to fit distributions to the cumulative distribution (percentiles) of the
data using @Risk software (available at www.palisade.com/risk/). Fitting nonparametric
distributions removed parameter uncertainty associated with the fitting of specific
parametric distributions (U.S. EPA, 2000).
2. When the available data were not adequate to support the statistical fitting for a specific age
cohort, the data fit to the closest age cohort available was used instead.
3. When available data were not adequate for either of the above methods, variables were fixed
at values recommended in the EFH or CSEFH or according to established EPA policy.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
This section describes how the various distributions of exposure factor data were collected and
processed for use in the probabilistic analysis. Probabilistic exposure analyses involve sampling
values from a distribution with the same characteristics as the data using the values to estimate
risk. For most variables for which distributions were developed, EPA exposure factor data were
analyzed to fit nonparametric models. Steps in the development of distributions included preparing
data, fitting models, assessing fit, and preparing parameters to characterize distributional
uncertainty in the model inputs.
For many exposure factors, the data from the EFH and CSEFH include sample sizes and estimates
of the following parameters for specific receptor types and age groups: mean, standard deviation,
standard error, and percentiles corresponding to some subset of the following probabilities: 0.01,
0.02, 0.05, 0.10, 0.15, 0.25, 0.50, 0.75, 0.85, 0.90, 0.95, 0.98, and 0.99. These percentile data, where
available, were used as a basis for fitting distributions. Although in no case are all of these
percentiles provided for a single factor, seven or more are typically available. Therefore, using the
percentiles provided a better representation of the available information than fitting distributions
simply based on the method of moments (i.e., selecting models that agree with the data mean and
standard deviation).
Neither the EFH nor the CSEFH makes use of the standardized age cohorts recommended in U.S.
EPA (2005c). Different exposure factors are reported for different age categories based on the
information available in the scientific literature. Therefore, to obtain the percentiles for fitting the
eight standardized age cohorts used for the revised risk assessment, each cohort-specific value for
a given exposure factor was assigned to one of the cohorts. When multiple cohorts were fit into a
single cohort, the percentiles from the EFH or CSEFH were averaged within each cohort (e.g., data
on 6 year olds and 9 year olds were averaged for the 6 to <11 age cohort). If sample sizes were
available, weighted averages were used, with weights proportional to sample sizes. If sample sizes
were not available, equal weights were assumed (i.e., the percentiles were simply averaged).
Nonparametric distributions were used to characterize the data. The nonparametric approach fits
an optimal smooth curve to the cumulative distribution (percentiles) of the data. The best
nonparametric fit is selected as the one that minimized the distance between the smooth curve
and the empirical curve generated by the percentiles of the data. The maximum and minimum are
used to specify the range of the simulated values. Depending on the data set, there could be more
than one distribution (parametric or nonparametric) that could be considered a good fit for the
data. Selecting an incorrect exposure distribution model may bias the risk assessment results,
producing incorrect conclusions. Therefore, the application of goodness-of-fit statistics was
required to select between competing distributions and to reduce model uncertainty. One
goodness-of-fit statistic available was the root mean squared error, defined as the root of the
average of the squared differences between the predicted percentile and the observed percentile.
The other goodness-of-fit statistic available was the Chi-square based comparison of the empirical
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
cumulative distribution (derived from the cumulative data) or the nonparametric cumulative
distribution. Graphical displays were also used to evaluate the appropriateness of the selected
distribution. A plot of the observed percentiles (from the cumulative data) vs. the nonparametric
cumulative distribution was created. In some cases, exposure distributions are highly skewed, and
there is a probability, although small, that a combination of extreme values might be selected from
the tails of the distributions. The resulting distributions are discussed in the following subsections,
with the specific percentiles presented in accompanying tables. Highlighted values are those
selected for use in the screening analysis.
Drinking Water Ingestion:
Drinking water intake data were obtained from Table 3-19 of the CSEFH and Table 3-38 of the
EFH, for children and adults, respectively. Weighted averages of percentiles and means were
calculated for the 0 to < 1 year infant (based on birth to < 1 month, 1 to < 3 months, 3 to < 6 months
and 6 to < 12 months), the 16 to < 21 years age group (based on 16 to < 18 years and 18 to < 21
years) and adults (based on 20 to 44, 45 to 64, and 64 to 74 years), using the number of observations
in each sub-cohort as weights. Table B-2 presents the water ingestion data used each age cohort.
Table B-2. Drinking Water Consumption Rate Data (mL/kg-day)
Age Group
(Years)
N
Mean
Percentile
P10
P25
P50
P75
P90
P95
P99
Infant™
948
71
7
25
66
104
140
164
217
1 to < 2
880
27
4
9
20
36
56
75
109
2 to < 3
879
26
4
9
21
36
52
62
121
3 to < 6
3,703
24
3
8
19
33
49
65
97
6 to < 11
1,439
17
3
6
13
23
35
45
72
11 to < 16
911
13
2
5
10
17
26
34
54
16 to < 21
700
13
2
5
10
17
27
34
61
Adult
7,616
16
2
6
12
22
34
42
64
Source: Table 3-19, CSEFH for child cohorts and Table 3-38, EFH (U.S. EPA, 2011) for adults.
1) Weighted average based on sub-cohorts presented in the CSEFH (U.S. EPA, 2008).
Drinking water consumption rate data for infants (birth to < 1 month, 1 to < 3 months, 3 to < 6
months and 6 to < 12 months) are available from Table 3-19 of the CSEFH. Table B-3 presents the
water ingestion data used for infants. As drinking water concentrations are provided by the model
as annual averages, the infant consumption rates were averaged to estimate a 0 to < 1 year infant.
The data were weighted by sample size because the small sample sizes did not meet minimum
requirements as described in the Third Report on Nutrition Monitoring in the United States
(IBNMRR, 1995) for numerous percentiles in numerous sub-cohorts.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
Table B-3. Drinking Water Consumption Rates for Infant Sub-cohorts (mL/kg-day)
Age Group
(Months)
N
Mean
Percentile
P10
P25
P50
P75
P90
P95
P99
Birth to < 1
37
137
11
65
138
197
235
238
263
1 to < 3
108
119
12
71
107
151
228
285
345
3 to < 6
269
80
7
27
77
118
148
173
222
6 to < 12
534
53
5
12
47
81
112
129
186
Weighted Average
948
71
7
25
66
104
140
164
217
Source: Table 3-19, CSEFH<^S. EPA, 2008).
Beef Ingestion Rate:
Consumption rates for beef are presented in Table B-4. These data are for consumption of
homegrown beef. Table 13-33 of the EFH provides data (in g WW/kg-d) for farming households
by age groups (6-11 years and 12-19 years) and for adult farmers (i.e., households who farm). Data
for ages 6-11 years were used for the 1 to <2, 2 to <3, 3 to < 6, and 6 to < 11 years age groups.
Data for ages 12-19 years were used for the 11 to <16 and 16 to <21 years age groups.
Beef consumption rate data were adjusted to account for post-cooking losses. A mean net post-
cooking loss of 29.7 percent accounts for losses from cutting, shrinkage, excess fat, bones, scraps,
and juices. This value was obtained from Table 13-69 of the EFH. Values shown in Table B-4 are
before these adjustments.
Table B-4. Beef Consumption Data (g WW/kg-d)
Age Group
(Years)
N
Mean
Percentile
P01
P05
P10
P25
P50
P75
P90
P95
P99
1 to < 2
38
3.77
0.35
0.66
0.75
1.32
2.11
4.43
11.4
12.5
13.3
2 to < 3
38
3.77
0.35
0.66
0.75
1.32
2.11
4.43
11.4
12.5
13.3
3 to < 6
38
3.77
0.35
0.66
0.75
1.32
2.11
4.43
11.4
12.5
13.3
6 to <11
38
3.77
0.35
0.66
0.75
1.32
2.11
4.43
11.4
12.5
13.3
11 to <16
41
1.72
0.38
0.48
0.51
0.9
1.51
2.44
3.53
3.57
4.28
16 to < 21
41
1.72
0.38
0.48
0.51
0.90
1.51
2.44
3.53
3.57
4.28
Adult
182
2.63
0.27
0.39
0.59
0.90
1.64
3.25
5.39
7.51
11.3
Sources: Table 13-33 EFH (U.S. EPA, 2011).
Milk Ingestion Rate:
Consumption rates for milk are presented in Table B-5. Table 13-25 of the EFH provides data (in
g WW/kg-d) for adult farmers. Data for children consuming homegrown milk are not available.
Therefore, we used data for general population from EFH Table 11-3. These data were provided
for ages 1-2, 3-5, 6-12, and 13-19 years. Data for ages 1-2 years were used for the 1 to <2 and 2 to
<3 years age groups. Data for ages 3-5 were used for the 3 to <6 years age group. Data for ages 6-
12 years were used for the 6 to < 11 years age group. Data for ages 13-19 years were used for the 11
to <16 and 16 to <21 years age groups.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix B: Benchmarks
-------
Table B-5. Milk Consumption Data (g WW/kg-d)
Age Group
(Years)
N
Mean
Percentile
P01
P05
P10
P25
P50
P75
P90
P95
P99
1 to < 2
1,052
43.2
1.0
5.7
10.7
20.3
39.1
59.4
84.1
94.7
141.2
2 to < 3
1,052
43.2
1.0
5.7
10.7
20.3
39.1
59.4
84.1
94.7
141.2
3 to < 6
978
24.0
0.9
4.5
8.3
13.6
20.7
32.0
41.9
51.1
68.2
6 to < 11
2,256
12.9
0.5
1.5
2.6
5.6
10.8
17.8
26
31.8
42.9
11 to < 16
3,450
5.5
0.1
0.4
0.6
1.6
4.0
7.6
12.3
16.4
24.9
16 to < 21
3,450
5.5
0.1
0.4
0.6
1.6
4.0
7.6
12.3
16.4
24.9
Adult
63
17.1
0.4
0.74
3.18
9.06
12.1
20.4
34.9
44.0
80.1
Fish Consumption:
Fish consumption data were obtained from Table E-3 of U.S. EPA (2015), based on the data
presented in Table 10-1 of the EFH. Values were selected for consumer-only ingestion rates for
uncooked finfish (excludes shellfish because of focus on fresh water). From the available data,
mean concentrations were used to represent recreational fishers, while the 95th percentile rates
were used to represent subsistence fishers. Table B-6 presents fish consumption rate data used to
prepare Monte Carlo simulations.
Table B-6. Fish Intake Rates for All Ages (g/kg-day)
Age Group
(Years)
Recreational
Fisher
Subsistence Fisher
1 to < 2
1.60
4.90
2 to < 3
1.60
4.90
3 to < 6
1.30
3.60
6 to < 11
1.10
2.90
11 to < 16
0.660
1.70
16 to < 21
0.660
1.70
Adult
0.665
2.05
Sources: Table E-3, U.S. EPA (2015)
Exposure Duration
Exposure durations for residents were determined using data on residential occupancy from Tables
16-109 and 16-113of the EFH. The data represent the total time a person (both male and female)
is expected to live at a single location, based on age. For adult residents, data reported for farm
residents were used to capture highly exposed receptors within the population. For child residents,
data reported for the 3-year age group were used to represent infants to < 6 years. Data reported
for ages 6 and 9 were averaged to represent 6 to < 11 years. Data on ages 12 and 15 were averaged
to represent 11 to < 16 years. Data reported for age 18 were used to represent 16 to < 21 years.
Table B-7 presents exposure duration data rounded to the nearest whole year. The source for the
recommended EFH distribution was a Monte Carlo simulation that estimated a probability
distribution for residential occupancy period based on the probability of moving and dying.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g g
Appendix B: Benchmarks
-------
Table B-7. Human Exposure Durations (ED) for All Ages (year)
Age Group
(Years)
Mean
Percentile
P25
P50
P75
P90
P95
P99
Infant
7
3
5
8
13
17
22
1 to < 2
7
3
5
8
13
17
22
2 to < 3
7
3
5
8
13
17
22
3 to < 6
7
3
5
8
13
17
22
6 to < 11
8
5
8
11
16
18
22
11 to < 16
9
5
9
13
16
18
23
16 to < 21
8
4
7
11
16
19
23
Adult
N/A
2
10
27
48
58
N/A
N/A - Not Available
Sources: Children: Table 16-109, EFH (U.S. EPA, 2011)
Adult farmer: Table 16-113, EFH (U.S. EPA, 2011)
Body Weight:
Weighted averages of percentiles and means were calculated for the infant age group (based on
birth to < 1 month, 1 to < 3 months, 3 to < 6 months and 6 to < 12 months) and adults (based on
male and female data). Table B-8 presents body weight data. Body weight data were obtained from
Table 8-3 of the EFH where data were presented by age for males and females combined.
Table B-8. Body Weight Data (kg)
Age Group
(Years)
N
Mean
Percentile
P05
P10
P15
P25
P50
P75
P85
P90
P95
Infant
1,858
7.8
6
6.4
6.7
7.1
7.8
8.6
9.0
9.3
9.7
1 to < 2
1,176
11.4
8.9
9.3
9.7
10.3
11.3
12.4
13
13.4
14
2 to < 3
1,144
13.8
10.9
11.5
11.9
12.4
13.6
14.9
15.8
16.3
17.1
3 to < 6
2,318
18.6
13.5
14.4
14.9
15.8
17.8
20.3
22
23.6
26.2
6 to < 11
3,593
31.8
19.7
21.3
22.3
24.4
29.3
36.8
42.1
45.6
52.5
11 to < 16
5,297
56.8
34
37.2
40.6
45
54.2
65
73
79.3
88.8
16 to < 21
4,851
71.6
48.2
52
54.5
58.4
67.6
80.6
90.8
97.7
108
Adult
12,504
71.4
52.9
56.0
58.2
61.7
69.3
78.5
84.9
89.8
97.6
Sources: Table 8-3 EFH (U.S. EPA, 2011) for children and adults.
Fixed Parameters
Certain parameters were fixed either because the available data were not adequate to generate a
full distribution or because only a single, high-end value was necessary to screen out the associated
exposure pathway. Table B-9 lists the parameters along with the value selected and source.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g
Appendix B: Benchmarks
-------
Table B-9. Human Ex
josure Factor Data Used in Modeling: Constants
Averaging Time
AT
All Ages
yr
U.S. EPA (2014)
Exposure Frequency
EF
All Ages
d/yr
U.S. EPA (2014)
Event Frequency
EV
All Ages
event/day
U.S. EPA (2014)
Event Duration
tEvent
All Ages
hr/event
U.S. EPA (2014)
Skin Surface Area
SA
All Child
Cohorts
cm2
U.S. EPA (2014)
Soil + Dust
Ingestion Rate
CRsoii
All Child
Cohorts
mg/day
U.S. EPA (2014)
Protected Fruit
Ingestion Rate
CRpFruit
Age 1 to <2
gtwwj/kg-day
90th Percentile from U.S.
EPA (2011), Table 13-62
Exposed Fruit
Ingestion Rate
CREFruit
Age 1 to <2
gtwwj/kg-day
90th Percentile from U.S.
EPA (2011), Table 13-61
Protected Vegetable
Ingestion Rate
CRpVeg
Age 1 to <2
gtwwj/kg-day
90th Percentile from U.S.
EPA (2011), Table 13-64
Exposed Vegetable
Ingestion Rate
CREVeg
Age 1 to <2
gtwwj/kg-day
90th Percentile from U.S.
EPA (2011), Table 13-63
Root Vegetable
Ingestion Rate
CRRVeg
Age 1 to <2
gtwwj/kg-day
90th Percentile from U.S.
EPA (2011), Table 13-65
Arsenic Absorption
Factor (Soil)
ABS(soil)
All Ages
%
U.S. EPA (2012)
Arsenic Absorption
Factor (Other Media)
ABS(other)
All Ages
%
U.S. EPA (1991)
Beef Preparation Loss*
LoSSprep
All Ages
%
U.S. EPA (2011), Table 13-69
Trophic Level 3 Fish
Consumed
FT3
All Ages
36
0/
U.S. EPA (2015)
Trophic Level 4 Fish
Consumed
Ft4
All Ages
64
70
* Beef preparation losses applied because measured ingestion rates reflect based on foods as brought into the household and not
in the form in which they are consumed.
B.1.3 Produce and Animal Product Exposure Factors
Chemical-specific factors were used to estimate the degree to which inorganic constituents may
accumulate in different plants and animals, as well as the resulting human exposures from
consumption of produce and animal products (i.e., beef, milk, fish). EPA reviewed the available
literature to assemble values.
Produce
Bioconcentration factors (BCFs) are used to estimate the magnitude of accumulation into produce.
Where possible, EPA relied on data from field studies because of the potential for greenhouse pot
studies to overpredict uptake (U.S. EPA, 1992c). However, pot studies were used when field study
data were not available. In instances where both the soil and crop data were non-detect, the data
were filtered out to avoid introducing excessive uncertainty. The individual data points used to
calculate BCFs were drawn from three sources:
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 11
Appendix B: Benchmarks
-------
¦ The Gypsum Constituent Database [Appendix A (Constituent Data)],
¦ Technical Support Document for Land Application of Sewage Sludge (U.S. EPA, 1992c),
¦ Estimating Risk from Contaminants Contained in Agricultural Fertilizers (U.S. EPA, 1999).
EPA divided the available data into different categories of plants (e.g., protected fruit, exposed
vegetable) to better capture the variability in produce consumed by both humans and livestock.
Individual plant species were mapped to plant categories according to the methodology in outlined
in U.S. EPA (1999). In some cases, some plants were mapped to multiple categories (e.g., corn to
both grain and protected vegetable). When multiple data points were available for a given plant
species, the values were averaged to prevent biasing the BCF toward those with more data. BCFs
were then calculated for each plant category as the geometric mean of the values for each species
in line with the recommended approach to calculate aquatic BCFs (U.S. EPA, 2016a). When no
data were available from the three data sources, EPA used BCFs previously calculated in ORNL
(1984). Due to the effort involved in compiling these data, values were only assembled in this
document for the constituents carried forward for the screening analysis.
Animal Products
BCFs are used to estimate the amount of constituent mass that may accumulate from the water
body (from dissolved and dissolved/suspended concentrations, respectively) into fish tissue.
Biotransfer factors (BTFs) were used to estimate the amount of constituent mass that may
accumulate from bulk soil and plant matter into beef and milk. Calculated BCFs and BTFs were
drawn from the available literature:
¦ Primary literature: These are generally papers that may either focus on a single chemical (i.e.,
USFWS, 1989; Kumada et al., 1973; Lemly, 1985; Murphy et al., 1978) or contain data on
multiple chemicals (i.e., Barrows et al., 1980; U.S. EPA, 1993).
¦ EPA databases/publications: These included ECOTOX (U.S. EPA, 2018) and the Mercury
Report to Congress (U.S. EPA, 1997a).
¦ Other government databases/publications: These included Oak Ridge National Laboratory
(ORNL), ATSDR and the Hazardous Substances Data Bank.
When sufficient data were available, separate BCFs were used for different fish trophic levels.
Where data for only one trophic level of fish were available, those data were used as a surrogate
for the other trophic level. Where data were only available for whole fish, those were used as a
surrogate for filet. Whole fish values from the correct trophic level were preferred as surrogates to
filet values from the other trophic level. So, given a TL3 whole fish value and a TL4 filet value, the
TL3 whole value would be preferred to the TL4 filet as a surrogate for TL3 filet.
Table B-10 lists all of the chemical-specific parameters collected for this analysis, presented in
alphabetical order based on the name of the constituent. In some instances, adequate data was not
available on one or more of the parameters for a given constituent. In these cases, the analysis
Beneficial Use Evaluation of FGD Gypsum in Agriculture g ^
Appendix B: Benchmarks
-------
could not quantitatively consider exposures through the associated pathway. When the reference
is a compilation, the original paper from which the value was drawn is also listed, if available. BCFs
were calculated using data maintained in the EPA's FGD Gypsum Database. Using the soil to crop
linkage table in the database, BCF values were calculated by dividing the crop concentration by
the soil concentration reported in the literature.
Table B-10. Bioconcentration and Biotransfer Factors
Parameter
Value
Reference
Comment
Aluminum
BCFts
3.6E+01
ECOTOX Cleveland, et al. (1991)
T4 whole fish was used for T3 filet
BCFt4
3.6E+01
ECOTOX Cleveland, et al. (1991)
T4 whole fish was used for T4 filet
Antimony
BCFts
0
Barrows et al., 1980
T3 whole fish (sunfish) was used for T3 filet.
BCFt4
0
Barrows et al., 1980
T3 whole fish (sunfish) was used for T4 filet.
Arsenic
BTFbeef
2.0E-03
ORNL (1984)
—
BTF milk
6.0E-05
ORNL (1984)
—
BCFts
4.0E+00
Barrows et al., 1980
T3 whole fish (sunfish) used for T3 filet.
BCFt4
4.0E+00
Barrows et al., 1980
T3 whole fish (sunfish) was used for T4 filet.
BCFExfruit
6.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFExveg
2.3E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFForage
6.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFGrain
2.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFpro fruit
6.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFproveg
1.2E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFRoot
1.2E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFsilage
6.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
Barium
BCFts
1.3E+02
ATSDR Hope, 1996
T3 whole fish (sunfish) was used for T3 filet.
BCFt4
1.3E+02
ATSDR Hope, 1996
T3 whole fish (sunfish) was used for T4 filet.
Beryllium
BTFbeef
1.0E-03
ORNL (1984)
—
BTF milk
9.0E-07
ORNL (1984)
—
BCFts
1.9E+01
Barrows et al., 1980
T3 whole fish (sunfish) was used for T3 filet.
BCFt4
1.9E+01
Barrows et al., 1980
T3 whole fish (sunfish) was used for T4 filet.
BCFExfruit
1.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFExveg
1.0E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFForage
4.6E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFGrain
1.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFpro fruit
1.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFproveg
1.5E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFRoot
1.0E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFsilage
4.6E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 13
Appendix B: Benchmarks
-------
Table B-10. Bioconcentration and Biotransfer Factors
Parameter
Value
Reference
Comment
Boron
BTFbeef
8.0E-04
ORNL (1984)
—
BTF milk
1.5E-03
ORNL (1984)
—
BCFts
—
—
—
BCFt4
—
—
—
BCFExfruit
2.0E+00
ORNL (1984)
—
BCFExveg
4.0E+00
ORNL (1984)
—
BCFForage
4.2E+00
Gypsum Database, U.S. EPA, 1992; 1999
Geometric mean of multiple crop types.
BCFGrain
6.1E-01
Gypsum Database, U.S. EPA, 1992; 1999
Geometric mean of multiple crop types.
BCFpro fruit
2.0E+00
ORNL (1984)
—
BCFproveg
2.0E+00
ORNL (1984)
—
BCFRoot
4.0E+00
ORNL (1984)
—
BCFsilage
4.2E+00
Gypsum Database, U.S. EPA, 1992; 1999
Geometric mean of multiple crop types.
Cadmium
BTFbeef
5.5E-04
ORNL (1984)
—
BTF milk
1.0E-03
ORNL (1984)
—
BCFts
2.7E+02
Kumada et al., 1972
T3 whole fish (rainbow trout) used for T3 and T4 filet.
Geomean of multiple values.
BCFt4
2.7E+02
Kumada et al., 1972
T3 whole fish (rainbow tr
Geometric mean of mult
out) was used for T4 filet,
pie values.
BCFExfruit
5.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFExveg
5.5E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFForage
2.0E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFGrain
8.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFpro fruit
5.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFproveg
7.2E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFRoot
1.3E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
BCFsilage
2.1E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of mult
pie crop types.
Chloride
BCFts
—
—
—
BCFt4
—
—
—
Chromium
BTFbeef
5.5E-03
ORNL (1984)
—
BTF milk
1.5E-03
ORNL (1984)
—
BCFts
6.0E-01
U.S. EPA, 1993
T4 filet was used for T3 filet.
BCFt4
6.0E-01
U.S. EPA, 1993 derived from Buhler et
al.,1977 and Calamari et al., 1982
—
BCFExfruit
3.3E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFExveg
8.4E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFForage
2.8E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFGrain
2.1E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFpro fruit
3.3E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFproveg
3.3E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFRoot
8.1E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFsilage
2.8E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
Cobalt
BCFts
—
—
—
BCFt4
—
—
—
Beneficial Use Evaluation of FGD Gypsum in Agriculture g ^
Appendix B: Benchmarks
-------
Table B-10. Bioconcentration and Biotransfer Factors
Parameter
Value
Reference
Comment
Copper
BCFts
0
U.S. EPA (1993)
—
BCFt4
0
U.S. EPA (1993)
—
Iron
BCFts
1.9E+01
ECOTOXPreez et al„ 1993
Species is banded bream
BCFt4
1.9E+01
HSDB Nakamoto and Hassler, 1992
T3 filet (banded bream) was used for T4 filet
Lead
BCFts
4.6E+01
U.S. EPA (1993)
T3 whole fish (bluegiII) was used for T3 filet
BCFt4
4.6E+01
U.S. EPA (1993)
T3 whole fish (bluegiII) was used for T4 filet
Manganese
BCFts
4.0E-01
ECOTOX Litzke and Hubel, 1993
Species was common carp
BCFt4
2.0E-01
ECOTOX Litzke and Hubel, 1993
Species was rainbow trout
Mercury
BTFbeef
6.0E-03
Calculated from U.S. EPA, 1997a
Converted from dry to fresh weight assuming a 70
percent moisture content in beef (US. EPA, 2005d)
BTF milk
2.6E-03
Calculated from U.S. EPA, 1997a
Converted from dry to fresh weight assuming a 87
percent moisture content in milk (US. EPA, 2005d)
BCFts
1.6E+06
U.S. EPA, 1997a
Methyl mercury
BCFt4
6.8E+06
U.S. EPA, 1997a
Methyl mercury
BCFExfruit
3.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFExveg
1.4E-01
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFForage
4.6E-01
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFGrain
6.6E-02
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFpro fruit
3.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFproveg
2.1E-02
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFRoot
1.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
BCFsilage
4.6E-01
Gypsum Database, U.S. EPA, 1992
1999
Divalent mercury. Geomean of multiple crop types.
Molybdenum
BCFts
4.0E+00
USFWS, 1989
T4 filet (rainbow trout and steelhead trout) was used
for T3 filet
BCFt4
4.0E+00
USFWS, 1989
Geometric mean of multiple values. Species were
rainbow trout and steelhead trout.
Nickel
BCFts
8.0E-01
U.S. EPA, 1993 derived from Calamari et
al„ 1982
T4 filet was used as a surrogate for T3 filet
BCFt4
8.0E-01
U.S. EPA, 1993 derived from Calamari et
al„ 1982
—
Selenium
BTFbeef
1.5E-02
ORNL (1984)
—
BTFmilk
5.9E-03
ORNL (1984)
—
BCFts
4.9E+02
Lemly, 1985
Species were threadfin shad and blueback herring. *
BCFt4
1.7E+03
Lemly, 1985
Species were threadfin shad and blueback herring. *
BCFExfruit
2.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFExveg
1.2E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFForage
1.6E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFGrain
3.0E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFpro fruit
2.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFproveg
1.6E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFRoot
3.4E-02
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFsilage
1.6E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 15
Appendix B: Benchmarks
-------
Table B-10. Bioconcentration and Biotransfer Factors
Parameter
Value
Reference
Comment
Strontium
BCFts
9.5E+00
ECOTOX Aleksanyan et al., 1978
Species was common carp
BCFt4
9.5E+00
ECOTOX Aleksanyan et al., 1978
T3 filet (common carp) was used for T4 filet
Thallium
BTFbeef
4.0E-02
ORNL (1984)
—
BTF milk
2.0E-03
ORNL (1984)
—
BCFts
3.4E+01
Barrows et al., 1980
T3 whole fish (sunfish) was used for T3 filet
BCFt4
1.3E+02
U.S. EPA, 1993 derived from Zitko et al.,
1975
—
BCFExfruit
4.0E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFExveg
4.0E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFForage
6.5E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFGrain
4.0E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFpro fruit
4.0E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFproveg
4.0E-04
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFRoot
4.0E-03
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
BCFsilage
6.5E-01
Gypsum Database, U.S. EPA, 1992
1999
Geometric mean of multiple crop types.
Vanadium
BCFts
3.2E+01
ECOTOX Bell et al., 1980
T4 whole fish was used for T3 filet
BCFt4
3.2E+01
ECOTOX Bell et al., 1980
T4 whole fish was used for T4 filet
Zinc
BCFts
3.5E+02
Murphy et al., 1978
T3 whole fish (bluegiII) was used for T3 filet. Geomean
of multiple values.
BCFt4
3.5E+02
Murphy et al., 1978
T3 whole fish (bluegiII) was used for T4 filet. Geomean
of multiple values
* In Lemly's paper on selenium, the BCFs are reported in L/g, but based on other data in the paper, the units are clearly actually L/kg.
- ECOTOX Wright, 1977 means the value was obtained from ECOTOX, which cites Wright (1977).
- ATSDR Hope, 1996" means the value was obtained from ATSDR, which cites Hope (1996).
Table B-11 presents additional factors used to calculate the accumulation of constituents in plants
and animal. Values were all drawn from the EPA guidance documents, with the exception of the
fraction of soil and forage consumed by cows raised for beef and milk. These values were set to 50
percent to reflect that, under pasturing conditions, cattle would not be allowed to graze in fields.
The growing season is assumed to be approximately half the year.
Table B-11. Plant and Animal Exposure Factor Data Used in Modeling: Constants
Variable
Value
Units
Citation
Exposed Fruit
85
Moisture Adjustment Factor
(MAF)
Exposed Vegetable
92
Protected Fruit
90
%
U.S. EPA (1997b)
Protected Vegetable
80
Root Vegetable
87
Beneficial Use Evaluation of FGD Gypsum in Agriculture g
Appendix B: Benchmarks
-------
Table B-11. Plant and Animal Exposure Factor Data Used in Modeling: Constants
Variable
Value
Units
Citation
Forage
50
Fraction of Media
Contaminated
(f)
Grain
100
%
Assumption
Silage
100
Soil
50
Forage
8.8
Beef Cattle Ingestion Rate
Grain
0.47
kg
-------
constituent. Eco-SSLs are derived based on toxicity reference values (TRVs), which represent a
receptor-class (e.g., bird, mammal) level estimate of the soil concentration at which no adverse
effects are observed from chronic exposure. Where data were available, TRVs were calculated for
four classes of terrestrial receptors: plants, invertebrates, birds and mammals. For birds and
mammals, EPA selected several different species representing different trophic levels and dietary
habits, and selected the most protective (i.e., lowest) value as the Eco-SSL for that group for each
constituent. A generic food-chain model was used to estimate the relationship between the
concentration of the contaminant in soil and the resulting dose for the receptor.
Tier II
United States Department of Energy (DOE) reports developed by Oak Ridge National Laboratory
(ORNL, 1997a,b) that calculate benchmarks for soil invertebrates and microbial processes. These
values represent Effects Range-Low (ER-L), which are calculated as the lower 10th percentile of
available lowest observed effects concentration (LOEC) data from laboratory data or field surveys.
When the available data was based on a lethal concentration 50% (LC50) or other endpoint that
includes a 50% or greater reduction in survivorship, the value was divided by a factor of five.
Table B-12. Ecological Benchmarks for Soil (Terrestrial Receptors)
Constituent
CASRN
Value
(lYig/kg dryweiqht)
Receptor
Sources
Arsenic
7440-38-2
18
Plants
U.S. EPA (2005e)
Beryllium
7440-41-7
10
Plants
ORNL (1997b)
Boron
7440-42-8
0.5
Birds
ORNL (1997b)
Cadmium
7440-43-9
0.36
Mammals
U.S. EPA (2005f)
Chromium (III)
16065-83-1
26
Birds
U.S. EPA (2008c)
Chromium (VI)
18540-29-9
0.4
Invertebrates
ORNL (1997a)
Mercury
7487-94-7
0.1
Invertebrates
ORNL (1997a)
Selenium
7446-08-4
0.52
Plants
U.S. EPA (2007a)
Thallium
7440-61-1
1.0
Mammals
ORNL (1997a)
B.2.2 Surface Water
Surface water benchmarks were selected to protect animals in water bodies that may be exposed
through direct contact with surface water or through ingestion of other biota that live in the water.
EPA chose aquatic criteria appropriate for species living in the freshwater bodies because coastal
waters were not modeled in this risk assessment. The hierarchy is as follows, with the selected
values presented in Table B-13.
Tier I
EPA National Recommended Surface Water Quality Criteria provide chronic benchmarks based
on Criterion Continuous Concentration (CCC). These values are estimates of the highest
concentration of a chemical to which an aquatic community can be exposed indefinitely without
resulting in an unacceptable effect. Values are only developed when sufficient data are available,
with at least eight LC50s and three CVs. First, a Final Acute Value (FAV) is calculated, which
Beneficial Use Evaluation of FGD Gypsum in Agriculture g
Appendix B: Benchmarks
-------
represents the fifth percentile of the distribution of 48- to 96-hour LC50 values or equivalent
median EC50 values. CCCs are the FAVs divided by the Final Acute-Chronic Ratio (FACR), which
is the geometric mean of quotients of at least three LC50/CV.
Tier II
The Great Lakes Initiative Clearinghouse (GLI, 2013) database contains chronic values compiled
from states and tribes from around the Great Lakes. When the minimum data requirements for
calculating a CCC were not met, Secondary Continuous Concentrations (SCCs) were calculated in
the same manner as CCCs with an adjustment factor applied based on the quantity of data available.
In instances where multiple values were available from different sources, EPA selected values
based on consideration of which were highest data quality, most recent, and lowest. Individual
sources for the values in the database are listed after the citation to the GLI database.
Table B-13. Ecological Benchmarks for Surface Water (Freshwater Community)
Aluminum
7429-90-5
87
U.S. EPA (1988b)
Antimony
7440-36-0
190
OHEPA (2006a)
Arsenic
7440-38-2
150
U.S. EPA (1996)
Barium
7440-39-3
220
OHEPA (2006b)
Beryllium
7440-41-7
11
NYDEC (1984)
Boron
7440-42-8
7,200
MIDEQ (2011a)
Cadmium
7440-43-9
0.72
U.S. EPA (2016b)
Chloride
16887-00-6
230,000
U.S. EPA (1986)
Chromium (III)
16065-83-1
74
U.S. EPA (1996)
Chromium (VI)
18540-29-9
11
Cobalt
7440-48-4
19
INDEM (1999a)
Copper
7440-50-8
9
U.S. EPA (2007b)
Iron
7439-89-6
1,000
U.S. EPA (1986)
Lead
7439-92-1
2.5
U.S. EPA (1985)
Manganese
7439-96-5
93
WIDNR (2005)
Mercury (total)
7439-97-6
0.77
U.S. EPA (1996)
Molybdenum
7439-98-7
800
INDEM (1998)
Nickel
7440-02-0
52
U.S. EPA (1995d)
Selenium
7782-49-2
1.5 (lentic)
3.1 (lotic)
U.S. EPA (2016c)
Strontium
7440-24-6
5,300
OHEPA (2006c)
Thallium
7440-28-0
6
INDEM (1999b)
Vanadium
7440-62-2
27
MIDEQ (2011b)
Zinc
7440-66-6
120
U.S. EPA (1996)
B.2.3 Sediment
Sediment benchmarks were selected to protect invertebrates that may be exposed to sediment
through direct contact with sediment or through ingestion of other biota that live in the sediment.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g
Appendix B: Benchmarks
-------
EPA chose aquatic criteria appropriate for species living in the freshwater bodies because coastal
waters were not modeled in this risk assessment. The hierarchy is summarized below, with the
identified values presented in Table B-14.
Tier I
The Florida Department of Environmental Protection (FLDEP) developed Threshold Effects
Concentrations (TECs) that identify concentrations below which harmful effects on sediment-
dwelling organisms are unlikely to be observed (FLDEP, 2003). TECs were derived by taking the
geometric mean of Effects Range-Low (ER-L) and Threshold Effects Level (TEL) data from various
sources. ER-L are calculated as the lower 10th percentile of available lowest observed effects
concentration (LOEC) data from laboratory data or field surveys. TELs are calculated as the
geometric mean of the 15th percentile of the effects level data set and the 50th percentile of the
no effects level data set. At least three separate sources were required to develop a TEC.
Tier II
When TECs could not be calculated, EPA identified individual values from the available literature:
¦ ER-Ls were drawn from the National Oceanic and Atmospheric Administrations (NOAA,
1991) and the Ontario Ministry of Environment and Energy (OMEE, 1993). ER-Ls are
calculated as the lower 10th percentile in the distribution of biological effects data from
matching biological and chemical laboratory data or field surveys.
¦ No Observed Effects Concentrations (NOECs) were drawn from Washington Department of
Ecology (WDOE, 2013). No Observed Effects Concentration represent the highest
concentration at which no effects were identified in laboratory studies.
Table B-14. Ecological Benchmarks for Sediment
Constituent
CASRN
Value
(mq/kq drvweiaht)
Receptor
Source
Antimony
7440-36-0
2
Invertebrates
NOAA (1991)
Arsenic
7440-38-2
9.8
Invertebrates
FLDEP (2003)
Barium
7440-39-3
20
Invertebrates
FLDEP (2003)
Cadmium
7440-43-9
1
Invertebrates
FLDEP (2003)
Chromium
7440-47-3
43.4
Invertebrates
FLDEP (2003)
Cobalt
7440-48-4
50
Invertebrates
OMEE (1993)
Copper
7440-50-8
31.6
Invertebrates
FLDEP (2003)
Iron
7439-89-6
20,000
Invertebrates
OMEE (1993)
Lead
7439-92-1
35.8
Invertebrates
FLDEP (2003)
Manganese
7439-96-5
460
Invertebrates
OMEE (1993)
Mercury
7487-94-7
0.18
Invertebrates
FLDEP (2003)
Nickel
7440-02-0
22.7
Invertebrates
FLDEP (2003)
Selenium
7782-49-2
11
Invertebrates
WDOE (2013)
Zinc
7440-66-6
121
Invertebrates
FLDEP (2003)
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 20
Appendix B: Benchmarks
-------
B.3 References
Aleksanyan, O.M., N.R. Kosinova and T.M. Krylova. 1978. "Effects of Strontium-90 and Metaphos
on Cyprinus carpio." Radiobiologiia.
ATSDR (Agency for Toxic Substances and Disease Registry). 2004. "Toxicological Profile for
Copper." Atlanta, GA. September.
Barrows, M.E., S.R. Petrocelli, K.J. Macek, and J.J. Carroll. 1980. "Chapter 24: Bioconcentration
and Elimination of Selected Water Pollutants by Bluegill Sunfish (Lepomis macrochirus)." In:
Dynamics, Exposure and Hazard Assessment of Toxic Chemicals. Edited by R. Haque. Ann
Arbor Science Publishers Inc., Ann Arbor, MI.
Bell, M.V., K.F. Kelly and J.R. Sargent. 1980. "The Uptake of Orthovanadate into Various Organs
of the Common Eel, Anguilla anguilla, Maintained in Fresh Water." Science of the Total
Environment. 16(2):99-108.
Buhler, D.R., R.M. Stokes and R.S. Caldwell. 1977. "Tissue Accumulation and Enzymatic Effects
of Hexavalent Chromium in Rainbow Trout (Salmo gairdneri)." Journal of the Fisheries
Research Board of Canada. 34(1):9-18.
Calamari, D., G.F. Gaggino and G. Pacchetti. 1982. "Toxicokinetics of Low Levels of Cd, Cr, Ni and
Their Mixture in Long-Term Treatment on Salmo gairdneri Rich." Chemosphere. ll(l):59-70.
Cleveland, L., D.R. Buckler and W.G. Brumbaugh. 1991. "Residue Dynamics and Effects of
Aluminum on Growth and Mortality in Brook Trout." Environmental Toxicology and
Chemistry. 10(2):243-248.
FLDEP (Florida Department of Environmental Protection). 2003. "Development and Evaluation of
Numerical Sediment Quality Assessment Guidelines for Florida Inland Waters-Technical
Report." Prepared by the USGS and MacDonald Environmental Sciences Ltd. Tallahassee, FL.
January.
Hope, B., C. Loy, and P. Miller. 1996. "Uptake and Trophic Transfer of Barium in a Terrestrial
Ecosystem." Bulletin of Environmental Contamination and Toxicology. 56:683-689.
IBNMRR (Interagency Board for Nutrition Monitoring and Related Research). 1995. "Third Report
on Nutrition Monitoring in the United States: Volumes 1 and 2." U.S. Government Printing
Office, Washington, DC.
INDEM (Indiana Department of Environmental Management). 1998. "Aquatic Life Fact Sheet for
Molybdenum." October.
INDEM. 1999a. "Aquatic Life Fact Sheet for Cobalt." March.
INDEM. 1999b. "Aquatic Life Fact Sheet for Thallium." March.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 21
Appendix B: Benchmarks
-------
Kumada, H., S. Kimura, M. Yokote, and Y. Matida. 1972. Acute and chronic toxicity, uptake, and
retention of cadmium in freshwater organisms. Bulletin of Freshwater Fisheries Research
Laboratory 22(2): 157-165.
Lemly, A.D. 1985. "Toxicology of Selenium in a Freshwater Reservoir: Implications for
Environmental Hazard Evaluation and Safety." Ecotoxicology and Environmental Safety
10:314-338.
Litzke, J. and K. Hubel. 1993. "Aquarium Experiments with Rainbow Trout (Oncorhynchus mykiss
Walbaum) and Carp (Cyprinus carpio L.) to Examine the Accumulation of Radionuclides."
Archiv fur Hydrobiologie. 129(1): 109-119.
MIDEQ, (Michigan Department of Environmental Quality). 2011a. "Aquatic Life Fact Sheet for
Boron." September.
MIDEQ. 2011b. "Aquatic Life Fact Sheet for Vanadium." March.
Murphy, B.R., G.J. Atchison, and A.W. Mcintosh. 1978. "Cadmium and Zinc in Muscle of Bluegill
(Lepomis macrochirus) and Largemouth Bass (Micropterus salmoides) from an Industrially
Contaminated Lake." Environmental Pollution. 17:253-257.
Nakamoto, R.J. and T.J. Hassler, 1992. "Selenium and Other Trace Elements in Bluegills from
Agricultural Return Flows in the San Joaquin Valley, California." Archives of Environmental
Contamination and Toxicology. 22:88-98.
NJDEP (New Jersey Department of Environmental Protection). 2009. "Derivation of Ingestion-
Based Soil Remediation Criterion for Cr+6 Based on the NTP Chronic Bioassay Data for Sodium
Dichromate Dihydrate." Prepared by A. Stern for the NJDEP Chromium Workgroup. Trenton,
NJ. April.
NOAA (National Oceanic and Atmospheric Administration). 1991. "The Potential for Biological
Effects of Sediment-Sorbed Contaminants Tested in the National Status and Trends Program."
Technical Memorandum NOS OMA 52. Prepared by E.R. Long and L.G. Morgan. Washington,
DC. August.
NTP (National Toxicology Program) 2008. "NTP Technical Report on the Toxicology and
Carcinogenesis Studies of Sodium Dichromate Dihydrate (CAS NO. 7789-12-0) in F344/N Rats
and B6C3F1 Mice (Drinking Water Studies)." NTP TR 546. Research Triangle Park, NC. July.
NYDEC (New York Department of Environmental Conservation). 1984. "Aquatic Life Fact Sheet
for Beryllium)." July.
OHEPA. (Ohio Environmental Protection Agency). 2006a. "Aquatic Life Fact Sheet for
Antimony." March.
OHEPA. 2006b. "Aquatic Life Fact Sheet for Barium." March.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 22
Appendix B: Benchmarks
-------
OHEPA. 2006c. "Aquatic Life Fact Sheet for Strontium." March.
OMEE (Ontario Ministry of Environment and Energy). 1993. "Guidelines for the Protection and
Management of Aquatic Sediment Quality in Ontario." Prepared by D. Persaud, R. Jaagumagi
and A. Hayton. August.
ORNL (Oak Ridge National Laboratory). 1984. "A Review and Analysis of Parameters for Assessing
Transport of Environmentally Released Radionuclides through Agriculture." ORNL-5786.
Prepared by C.F. Baes III, R.D. Sharp, A.L. Sjoreen and R.W. Shor Office for the EPA Office of
Air and Radiation under Interagency Agreement AD-89-F-2-A106. Oak Ridge, TN. September.
ORNL. 1997a. "Toxicological Benchmarks for Contaminants of Potential Concern for Effects on
Soil and Litter Invertebrates and Heterotrophic Process: 1997 Revision." ES/ER/TM-126/R2.
Prepared by R.A. Efroymson, M.E. Will and G.W. Suter II for the U.S. DOE. Oak Ridge, TN.
November.
ORNL. 1997b. "Toxicological benchmarks for Screening Potential Contaminants of Concern for
Effects on Terrestrial Plants: 1997 Revision." ES/ER/TM-85/R3. Prepared by R.A. Efroymson,
M.E. Will, G. W. Suter II and A. C. Wooten for the U.S. DOE. Oak Ridge, TN. November.
du Preez, H.H., E. van Rensburg and J.H.J van Vuren. 1993. "Preliminary Laboratory Investigation
of the Bioconcentration of Zinc and Iron in Selected Tissues of the Banded Tilapia, Tilapia
sparrmanii (Cichlidae)." Bulletin of Environmental Contamination and Toxicology. 50:674-
681.
U.S. EPA (Environmental Protection Agency). 1985. "Ambient Water Quality Criteria for Lead -
1984." EPA 440/5-84-027. Prepared by the EPA Office of Water. Washington, DC. January.
U.S. EPA. 1986. "Quality Criteria for Water 1986." EPA 440/5-86-001. Prepared by the EPA Office
of Water. Washington, DC. May.
U.S. EPA. 1987. "Chemical Assessment Summary: Antimony; CASRN 7440-36-0." Prepared by the
EPA Office of Research and Development. Washington, DC. January.
U.S. EPA. 1988a. "Chemical Assessment Summary. Vanadium pentoxide; CASRN 1314-62-1."
Prepared by the Office of Research and Development. Washington, DC. June.
U.S. EPA. 1988b. "Ambient Water Quality Criteria for Aluminum - 1988." EPA 440/5-86-008.
Prepared by the EPA Office of Water. Washington, DC. August.
U.S. EPA. 1989a. "Risk Assessment Guidance for Superfund. Volume I: Human Health Evaluation
Manual (Part A) Interim Final." EPA/540/1-89/002. Prepared by the EPA Office of Emergency
and Remedial Response. Washington, DC. December.
U.S. EPA. 1989b. "Chemical Assessment Summary: Cadmium; CASRN 7440-43-9." Prepared by
the EPA Office of Research and Development. Washington, DC. October.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 23
Appendix B: Benchmarks
-------
U.S. EPA. 1991a. "Chemical Assessment Summary: Nickel, Soluble Salts (CASRN Various)."
Prepared by the EPA Office of Research and Development. Washington, DC. December.
U.S. EPA. 1991b. "Chemical Assessment Summary: Selenium and Compounds; CASRN 7782-49-
2." Prepared by the EPA Office of Research and Development. Washington, DC. June.
U.S. EPA. 1991c. "Risk Assessment Guidance for Superfund (RAGS) Volume I: Human Health
Evaluation Manual Supplemental Guidance "Standard Default Exposure Factors" (Interim
Final)." OSWER Directive 9285.6-03. Prepared by the EPA Office of Solid Waste and
Emergency Response. Washington, DC. March.
U.S. EPA. 1992a. "Chemical Assessment Summary: Molybdenum; CASRN 7439-98-7." Prepared
by the EPA Office of Research and Development. Washington, DC. November.
U.S. EPA. 1992b. "Chemical Assessment Summary: Strontium; CASRN 7440-24-6." Prepared by
the Office of Research and Development. Washington, DC. October.
U.S. EPA. 1992c. "Technical Support Document for Land Application of Sewage Sludge:
Volume 1." EPA 822/R-93-001a. Prepared by the Eastern Research Group for the EPA Office
of Water. Washington, DC. November.
U.S. EPA. 1993. "Derivation of Proposed Human Health and Wildlife Bioaccumulation Factors for
the Great Lakes Initiative (Draft). EPA 905-R-93-103. Prepared by the EPA Office of Research
and Development. Duluth, MN. March.
U.S. EPA. 1995a. "Chemical Assessment Summary: Arsenic; CASRN 7440-38-2." Prepared by the
EPA Office of Research and Development. Washington, DC. June.
U.S. EPA. 1995b. "Chemical Assessment Summary: Manganese; CASRN 7439-96-5." Prepared by
the EPA Office of Research and Development. Washington, DC. November.
U.S. EPA. 1995c. "Chemical Assessment Summary: Mercury Chloride (Hg,C12); CASRN 7487-94-
7." Prepared by the EPA Office of Research and Development. Washington, DC. May.
U.S. EPA. 1996. "1995 Updates: Water Quality Criteria Documents for the Protection of Aquatic
Life in Ambient Water." EPA-820-B-96-001. Prepared by the EPA Office of Water.
Washington, DC. September.
U.S. EPA. 1997a. "Mercury Study Report to Congress. Volume III - Fate and Transport of Mercury
in the Environment." EPA 452/R-97/005. Prepared by the EPA Office of Air Quality Planning
and Standards and Office of Research and Development. Washington, DC. December.
U.S. EPA. 1997b. "The Parameter Guidance Document. A Companion Document to the
Methodology for Assessing Health Risks Associated with Multiple Pathways Exposure to
Combustor Emissions. Draft." Prepared by the EPA Office of Research and Development.
Cincinnati, OH. March.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 24
Appendix B: Benchmarks
-------
U.S. EPA. 1998a. "Toxicological Review of Beryllium and Compounds (CAS No. 7440-41-7)."
Prepared by the EPA Office of Research and Development. Washington, DC. April.
U.S. EPA. 1998b. "Toxicological Review of Trivalent Chromium (CAS No. 16065-83-1)." Prepared
by the EPA Office of Research and Development. Washington, DC. August.
U.S. EPA. 1998c. "Toxicological Review of Hexavalent Chromium (CASRN 18540-29-9)." Prepared
by the EPA Office of Research and Development. Washington, DC. August.
U.S. EPA. 1999. "Estimating Risk from Contaminants Contained in Agricultural Fertilizers (Draft
Report)." Prepared by RTI International for the EPA Office of Solid Waste under Contract
Number 68-W-98-0085. Washington, DC. August.
U.S. EPA. 2000. "Options for Development of Parametric Probability Distributions for Exposure
Factors." EPA.600/R-00/058. Prepared by the EPA Office of Research and Development.
Washington, DC. July.
U.S. EPA. 2001. "Chemical Assessment Summary: Methylmercury (MeHg); CASRN 22967-92-6."
Prepared by the EPA Office of Research and Development. Washington, DC. July.
U.S. EPA. 2003. "Human Health Toxicity Values in Superfund Risk Assessments." OSWER
Directive 9285.7-53. Prepared by the EPA Office of Solid Waste and Emergency Response,
Washington, DC. December.
U.S. EPA. 2004. "Toxicological Review of Boron and Compounds (CAS No. 7440-42-8)." EPA
635/04/052. Prepared by the EPA Office of Research and Development. Washington, DC.
Washington, DC. June.
U.S. EPA. 2005a. "Toxicological Review of Barium and Compounds (CAS No. 7440-39-3)."
EPA/635/R-05/001. Prepared by the EPA Office of Research and Development. Washington,
DC. Washington, DC. June.
U.S. EPA. 2005b. "Toxicological Review of Zinc and Compounds (CASRN 7440-66-6)."
EPA/635/R-05/002. Prepared by the EPA Office of Research and Development. Washington,
DC. July.
U.S. EPA. 2005c. "Guidance on Selecting Age Groups for Monitoring and Assessing Childhood
Exposures to Environmental Contaminants." Prepared by the EPA Risk Assessment Forum,
Washington, DC. November.
U.S. EPA. 2005d. "Human Health Risk Assessment Protocol for Hazardous Waste Combustion
Facilities: Final." EPA-530-R-05-006. Prepared by the EPA Office of Solid Waste and
Emergency Response. Washington, DC. September.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 25
Appendix B: Benchmarks
-------
U.S. EPA. 2005e. "Ecological Soil Screening Levels for Arsenic: Interim Final." OSWER Directive
9285.7-62. Prepared by the EPA Office of Solid Waste and Emergency Response. Washington,
DC. February.
U.S. EPA. 2005f. "Ecological Soil Screening Levels for Cadmium: Interim Final." OSWER Directive
9285.7-65. Prepared by the EPA Office of Solid Waste and Emergency Response. Washington,
DC. March.
U.S. EPA. 2006a. "Provisional Peer Reviewed Toxicity Values for Aluminum (CASRN 7429-90-
5)." Prepared by the EPA Office of Research and Development. Cincinnati, OH. October.
U.S. EPA. 2006b. "Provisional Peer Reviewed Toxicity Values for Iron and Compounds (CASRN
7439-89-6)." Prepared by the EPA Office of Research and Development. Cincinnati, OH.
September.
U.S. EPA. 2007a. "Ecological Soil Screening Levels for Selenium: Interim Final." OSWER Directive
9285.7-72. Prepared by the Office of Solid Waste and Emergency Response. Washington, DC.
July.
U.S. EPA. 2007b. "Aquatic Life Ambient Freshwater Quality Criteria - Copper 2007 Revision."
EPA-822-R-07-001. Prepared by the EPA Office of Water. Washington, DC. February.
U.S. EPA. 2008a. "Provisional Peer Reviewed Toxicity Values for Cobalt (CASRN 7440-48-4)."
Prepared by the EPA Office of Research and Development. Cincinnati, OH. August.
U.S. EPA. 2008b. "Child-Specific Exposure Factors Handbook." EPA-600/R-06-096F. Prepared by
the EPA Office of Research and Development. Washington, DC. September.
U.S. EPA. 2008c. "Ecological Soil Screening Levels for Chromium: Interim Final.: OSWER
Directive 9285.7-66. Prepared by the EPA Office of Solid Waste and Emergency Response.
Washington, DC. April.
U.S. EPA. 2011. "Exposure Factors Handbook: 2011 Edition." EPA/600/R-090/052F. Prepared by
the EPA Office of Research and Development. Washington, DC. September.
U.S. EPA. 2012a. "Provisional Peer Reviewed Toxicity Values for Thallium Soluble Salts [Metallic
Thallium (7440-28-0), Thallium (I) acetate (563-68-8), Thallium (I) carbonate (6533-73-9),
Thallium (I) chloride (7791-12-0), Thallium (I) nitrate (10102-45-1), and Thallium (I) sulfate
(7446-18-6)]." Prepared by the EPA Office of Research and Development. Cincinnati, OH.
November.
U.S. EPA. 2012b. "Recommendations for Default Value for Relative Bioavailability of Arsenic in
Soil." OSWER 9200.1-113. Prepared by the EPA Office of Solid Waste and Emergency
Response. Washington, DC. December.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 25
Appendix B: Benchmarks
-------
U.S. EPA. 2014. "Human Health Evaluation Manual, Supplemental Guidance: Update of Standard
Default Exposure Factors." OSWER Directive 9200.1-120. Prepared by the EPA Office of Solid
Waste and Emergency Response. Washington, DC. February.
U.S. EPA. 2015. "Environmental Assessment for the Effluent Limitations Guidelines and Standards
for the Steam Electric Power Generating Point Source Category." EPA-821-R-15-006.
Prepared by the EPA Office of Water. Washington, DC. September.
U.S. EPA. 2016a. "Development of National Bioaccumulation Factors: Supplemental Information
for EPA's 2015 Human Health Criteria Update." EPA 822-R-16-001. Prepared by the EPA
Office of Water. Washington, DC. January.
U.S. EPA. 2016b. "Aquatic Life Ambient Water Quality Criteria for Cadmium - 2016." EPA-820-
R-16-002. Prepared by the EPA Office of Water. Washington, DC. March.
U.S. EPA. 2016c. Aquatic Life Ambient Water Quality Criteria for Selenium - Freshwater 2016.
EPA-822-R-16-006. Prepared by the EPA Office of Water. Washington, DC. June.
U.S. EPA. 2018. "ECOTOX User Guide: ECOTOXicology Knowledgebase System User Guide -
Version 5.0." Prepared by the EPA Office of Research and Development. Duluth, MN. June.
USFWS (United States Fish and Wildlife Service). 1989. "Molybdenum Hazards to Fish, Wildlife,
and Invertebrates: A Synoptic Review." Biological Report 85(1.19). Contaminant Hazard
Reviews Report No. 19. Laurel, MD. August.
WDOE (Washington State Department of Ecology). 2013. "Sediment Management Standards."
Publication 13-09-055. Olympia, WA. September.
WIDNR (Wisconsin Department of Natural Resources). 2005. "Aquatic Life Fact Sheet for
Manganese." April.
Zitko, V. 1975. "Toxicity and Pollution Potential of Thallium." The Science of the Total
Environment. 4:185-192.
Beneficial Use Evaluation of FGD Gypsum in Agriculture g 27
Appendix B: Benchmarks
-------
Appendix C. Use Characterization
This appendix describes the approach used to characterize how and where FGD gypsum might be
applied across the continental United States. This information was used in the evaluation to
characterize the environmental conditions that may be present where this secondary material is
applied, the proximity to nearby receptors, and the rate that constituents may be released into
surrounding soil, ground water, surface water and air. In any given year, the extent of land used
for crops and the rate of gypsum application will change based on a combination of agronomic and
economic considerations. Therefore, this evaluation aims to capture the maximum extent of
cropland that is available based on both past and present use.
C.1 Application Rate and Frequency
To estimate the total mass of gypsum that may be applied to agricultural land, EPA reviewed peer
reviewed literature, government and industry reports, and state extension services. These sources
provided a mixture of current practices, recommended rates, and theoretical ranges that could
provide benefits. As a result, the rates reported sometimes varied considerably among different
sources. EPA considered all the sources that reported benefits, though not all sources identified
measurable benefits from the same application rates. Because of uncertainty about how practices
may evolve if use of FGD gypsum becomes more widespread, EPA considered both demonstrated
and theoretical rates. However, as a general criterion for all uses, the USDA National Resources
Conservation Service stipulates that annual applications should not exceed 5 tons/acre (USDA,
2015a). Therefore, this was set as an upper bound on average annual applications.
C.1.1 Reduction in Soluble Phosphorus
Identified literature sources that address applications to reduce soluble phosphorus include Stout
et al. (2000), Brauer et al. (2005), Watts and Torbert (2009), OSU-E (2011), Endale et al. (2014),
Torbert and Watts (2014), Adeli et al. (2015) and USDA (2015a). Across the various soil types and
local conditions studied, application rates ranged from as low as 0.5 tons/acre to as high as 4.5
tons/acre (OSU-E, 2011). Application frequency was not discussed widely in the literature for this
use. USDA staff indicated that, for applications with manure, annual application was a reasonable
assumption. For application on soils containing residual high phosphorus, application every other
or third year may be a more reasonable assumption (Dick, 2015; Torbert, 2015).
Table C-1 presents a summary of the application rates and frequencies for this use of FGD gypsum.
To obtain reasonable bounds on application rates, EPA separately grouped the moderate and
highest values reported in each of the available source and calculated an average for each. Because
USDA (2015a) placed a lower bound on applications of 1 ton/acre, EPA treated this application
rate as the low end. Few studies reported rates lower than this and those that did were similar.
EPA assumed annual applications for all locations under the assumption of manure application.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
Table C-1. Application Rates and Frequency for Phosphorus Runoff
Modeled Frequency
of Application
Mass Applied
(tons/acre)
Low
Moderate
High
1 Year
1.0
1.5
3.0
C.1.2 Nutrient Source
Identified literature sources that address potential application rates for calcium or sulfur nutrient
amendments include FIPR (1995), Grichar et al. (2002), UW-E, (2003), Sumner (2007), Chen et al.
(2008) and OSU-E (2011). Across the various crops and soils studied, application rates as a sulfur
source range from as low as 0.03 tons/acre (FIPR, 1995) to as high as 0.27 tons/acre (OSU-E, 2011).
Application rates as a calcium source tend to be greater, ranging from as low as 0.04 tons/acre
(FIPR, 1995) to 2 tons/acre (Chen et al., 2008). However, the highest application rates associated
with peanuts and tomatoes are unlikely to be applied every year. State Extension Services from
outside the southeast state that peanuts and tomatoes are grown in rotation to control disease and
pests (UM/UW-E, 1991; TAMU-E, 2015; UC-IPM, 2013; APC, 2015). APC (2015) states that
farmers are often successful using a two- or three-year rotation with either soybeans, cotton or
corn. UC-IPM (2013) suggests growing tomatoes in a two- or three-year rotation.
Table C-2 presents a summary of the application rates and frequencies for this use of FGD gypsum.
To obtain reasonable bounds on application rates, EPA separately grouped the lowest, moderate
and highest values for both calcium and sulfur reported in each available source and calculated an
average value for all three. The application frequency, particularly the higher calcium applications,
is anticipated to be every two years at most.
Table C-2. Application Rates and Frequency for Nutrient Amendment
Modeled Frequency
of Application
Mass Applied
(tons/acre)
Low
Moderate
High
2 Year
0.2
0.8
1.7
C.1.3 Sodic Soils
The identified literature sources that address use in sodic soil include KSU-E (1992), PNE (2007),
ASCE (2012), CSU-E, (2012) and USDA (2015a). Several sources provided equations to calculate
necessary applications rates as a function of soil cation exchange capacity, initial and target sodium
adsorption ratio, bulk density, and/or soil depth. Some sources applied these equations to realistic
soil conditions and provided recommended application rates. These values ranged from a low of 1
ton/acre to a high of 10 tons/acre (PNE, 2007; OSU-E, 2011). Application frequency was not
discussed widely in the literature for this use. USDA staff indicated that at the higher rates reported
in the literature, applications may occur every 10 to 20 years (Chaney, 2016).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
Table C-3 presents a summary of the application rates and frequency modeled for this use of FGD
gypsum. Because of the relatively few numerical rates reported in the literature, EPA selected low,
moderate and high rates based on those reported in OSU-E (2011). EPA assumed that applications
would occur, on average, every 10 years.
Table C-3. Application Rates and Frequency for Sodic Soils
Modeled Frequency
of Application
Mass Applied
(tons/acre)
Low
Moderate
High
10 Years
1.0
5.0
10.0
C.1.4 Improve Infiltration
Literature sources that address potential application rates to improve infiltration include Ben-Hur
et al. (1992), FIPR (1995), UC-E (1997), Yu et al. (2003), Amezketa et al. (2005), OSU-E (2011),
Buckley and Wolkowski (2014) and USDA (2015a). Application rates reported in the literature
ranged from as low as 0.01 tons/acre (FIPR, 1995) to as high as 4.5 tons/acre (OSU-E, 2011). For
application frequency, multiple sources reported that applications should occur on an annual basis
until the problem is remedied. Applications may then continue on a more intermittent basis as
necessary afterwards (FIPR, 1995; UC-E, 1997; USDA, 2015a).
Table C-4 presents a summary of the application rates and frequency modeled for this use of FGD
gypsum based on available data. To obtain reasonable bounds on application rates, EPA separately
grouped the lowest, moderate and highest values reported in each of the available source and
calculated an average for each. The range of values agree well with those recommended in USDA
(2015a). Applications are assumed to occur on an annual basis for the full duration of application.
Table C-4. Application Rates and Frequencies
Modeled Frequency
of Application
Mass Applied
(tons/acre)
Low
Moderate
High
1 Year
0.25
0.75
2.0
C.1.5 Aluminum Toxicity
Literature sources that address application rates to address subsoil aluminum toxicity as a source
of calcium or sulfur include Feldhake and Ritchey (1996), Miller and Sumner (1997), Toma et al.
(1999), Farina et al. (2000a,b), Ritchey and Snuffer (2002), Chen et al. (2005) and Caires et al.
(2011). Application rates reported in the literature ranged from as low as 0.4 tons/acre (Ritchey
and Snuffer, 2002) to as high as 15.6 tons/acre (Toma et al., 1999). For application frequency, Caires
et al. (2011) found that ".. .about 10% of Ca from gypsum was still adsorbed in the upper 10 cm of
soil several years post application." Both Farina et al. (2000a) and Miller and Sumner (1997)
characterized effects as lasting 10 years. When Toma et al. (1999) studied the longevity of effects
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix C: Use Characterization
-------
from two earlier studies they found gypsum applications still effective after around 15 years.
Limestone may also be applied along with the FGD gypsum to counteract the displacement of
magnesium and potassium.
Table C-5 presents a summary of the application rates and frequency modeled for this use of FGD
gypsum. To obtain reasonable bounds on application rates, EPA separately grouped the lowest and
highest values reported in each of the available source and calculated an average for each. For this
use, few sources reported a moderate application rate. Therefore, EPA used the "normal"
application rate reported by OSU-E (2011) based on a review of the literature. These applications
on an annual basis with applications spread out between with the same amount each time. Based
on the frequencies reported by different sources, EPA assumed that applications would occur, on
average, every 10 years.
Table C-5. Application Rates and Frequencies
Frequency of
Application
Mass Applied
(tons/acre)
Low
Moderate
High
10 Years
1.5
3.0
11.0
C.2 Field Boundaries
The size and location of agricultural fields provide important information about where and how
much gypsum may be applied across the landscape. To estimate field boundaries, EPA initially
relied on the Common Land Unit (CLU) and Cropland Data Layer (CDL) datasets:
¦ A CLU is the smallest unit of land associated with USDA management programs that has a
permanent, contiguous boundary, a common land cover and land management, and both a
common owner and producer. CLU boundaries are delineated from relatively stable features
such as fence lines, roads, and/or waterways. The most recent CLU data is from 2008. The
USDA has since restricted access to subsequent CLU polygons following of the enactment of
The Food, Conservation, and Energy Act of 2008 (Public Law 110-234). The available shapefile
does not contain any data about land use and so the shapefile alone is not sufficient for the
purpose of defining field boundaries.
¦ The Cropland Data Layer (CDL) blends extensive field data and satellite information to produce
a detailed raster map aggregated into the following ten generalized groups: corn, cotton, rice,
soybeans, wheat, vegetables and ground fruit, orchards and vineyards, other grains, other row
crops, and other crops. These data have been produced annually since 2008. For this evaluation,
EPA used five years' worth of data from between 2010 and 2015 to capture the maximum
extent of cropland. The available raster data is generated at a lower resolution than the CLU
polygons, which adds greater uncertainty about exact borders. Therefore, the raster data alone
is not sufficient for the purpose of defining field boundaries.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix C: Use Characterization
-------
Figure C-1 presents the overlap of CLU and CDL data across the country. Areas covered in grey
are those where CLU polygons are available (CLU data is not available for every state). The area
covered in green are the total extent of cropland predicted by CDL raster data.
Figure C-1: Comparison of Coverage for CLU and CDL Datasets
For the purposes of this analysis, fields were divided in two categories based on the available data:
those with and without CLU data. CLUs are provided on a county-by-county basis within each
state. In some states, not all counties were delineated. In total, there were four states with no CLU
data (AL, DE, FL and NM), 22 states for which all counties had CLU data, and 21 states that
possessed a mixture of counties with and without CLU data. No field boundaries were developed
for Alaska, California or Hawaii because these states fell outside the economic feasibility zone. EPA
used similar approaches to delineate boundaries for fields in areas with and without CLU data. For
areas with CLU data, the CDL raster data was overlain on top of the CLU polygons and used along
with supplementary datasets to predict which CLU polygons were most likely to be cropland. The
following data sources were used together with CDL and CLU data to refine the field boundaries:
¦ County boundary polygons from the U.S. Census cartographic, boundary files. This layer was
used so that data processing could be conducted and aggregated on a county-by-county basis.
¦ High-resolution National Hydrography Dataset (NLID) Plus dataset. These flowlines and water
body polygons were used to identify the location of streams and lakes where cropland is
unlikely.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
¦ TIGER roadways by state (U.S. DOC, 2013). The roadway lines were used to identify areas
likely to be roads, shoulders and ditches where cropland is unlikely.
¦ National Land Cover Dataset (NLCD) (USGS, 2011). These layers were used to identify areas
of the landscape covered by impervious surfaces (e.g., building, parking lot) and forests where
cropland is unlikely. Raster files were extracted by counties and converted to polygons.
Through trial and error paired with visual inspection against satellite imagery, EPA identified the
combination of variables that best captured the extent of cropland. Once, these variables had been
identified, EPA applied the same approach to areas without CLU data. In these areas, the CDL
raster file was converted directly into field polygons and used along with supplementary datasets
to predict areas most likely to be cropland. The datasets were managed as described below:
¦ County borders were used as a hard boundary on field area because CLU data is reported on a
county-by-county basis. When field polygons crossed country lines, fields were split along that
line and assigned to the county it is located. This process also allowed processing of field data
in more manageable units (county rather than state).
¦ Polygons with a total area less than four acres were removed from the dataset. A review of the
polygons overlain on satellite imagery found these areas most likely to reflect noise in the CDL
datasets or small features (e.g., buildings) in the CLU dataset that would skew estimates of field
size lower.
¦ CLU polygon with less than 50% overlap with the CDL raster was deleted. This value was
selected based on trial and error to eliminate issues, such as bleed over of raster data from
adjacent polygons. In these instances, the raster area may be continuous and larger than four
acres, but only present around the periphery of a polygon.
¦ NHD flowlines were overlain on the polygons. If the flowline intersected with a CLU polygon,
the CLU was assumed to be a water body along with any associated buffer areas and the entire
polygon was removed. If the flowline intersected with a non-CLU polygon, the line was treated
as a natural barrier between fields and simply subtracted out.
¦ Fields with both an area-to-perimeter ratio less than 60 and overlap with the CDL raster greater
than 50% were merged together with adjacent polygons when the polygon was surrounded by
potential cropland on more than one side. Visual inspection against satellite imagery found
that these areas likely reflect terracing and other practices intended to prevent erosion.
¦ The shape and size of each remaining polygon was used to identify remaining areas that are
unlikely to be cropland. EPA removed isolated polygons when the compactness, calculated as
4tt (Pe^r6aer > was < 0.25 or the area-to-perimeter ratio was < 30 for CLU polygons and <15
for CDL polygons. Different values were used for CLU and CDL fields because of the blockier
polygons formed by the CDL data. These metrics were used together to identify long and
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
narrow polygons more likely to reflect drainage ditches, buffer areas and other managed areas
where crops are not grown.
¦ NHD water body polygons were used to capture portions of fields that contain open water.
Portions of a polygon that overlapped with these areas were removed.
¦ TIGER roadway lines buffered on either side by 10 meters were used to capture roadway width,
shoulders, rights of way and/or drainage ditches. Portions of a polygon that overlapped with
these buffer areas were removed.
¦ NLCD polygons for impervious surfaces and forests were used to capture areas that are covered
by either forests or impervious surfaces (e.g., buildings, sidewalks, parking lots) that are clearly
not used as cropland. Portions of a polygon that overlapped with these areas were removed.
After applying these steps, EPA recalculated the area of the remaining polygons and removed those
that had been reduced to less than four acres. The remaining polygon area was assumed to be
entirely cropland. EPA conducted QA/QC on randomly selected counties from areas with and
without CLU data to ensure that each data file had been correctly extracted, converted to polygons,
and applied. Visual inspection of the fields overlain on satellite imagery was used to ensure that
the resulting fields aligned with the apparent land use. Based on this review, EPA believes that the
field boundaries provides reasonable estimate of field location and area.
This process resulted in over five million individual fields across 47 states (except CA, HI and AK)
and 2,893 counties (out of 3,219). Not all of this land will be in active use in any given year as a
result of economic incentives and crop rotation. EPA did not identify any data that could be used
to reliably set a fraction of this land expected to be in active in any given year. Therefore, EPA
relied on the delineated fields to define the maximum extent of cropland for this evaluation. In
each model run, the fraction of the total cropland with FGD gypsum applied in any given year was
allowed to vary anywhere from 0 to 100% based on a flat distribution.
C.3 Extent of Use
To delineate the geographic area over which FGD gypsum may be applied, EPA first defined the
maximum area that it might be economical to apply the secondary material. This approach assumed
that the compounding costs of purchase, transportation and application were the primary factors
that determined whether FGD gypsum will be used. These costs do not consider whether any
individual utility has the ability to meet the demands of the market in the surrounding area. Over
a third FGD gypsum currently generated is diverted towards wallboard production (ACAA, 2018).
Demand may exceed generating capacity if agricultural uses become widespread, resulting in
smaller distribution areas around some utilities than considered in this evaluation. Impacts from
imported sources of gypsum were assumed to be negligible.
Farmer willingness to pay for gypsum was estimated in 2011 to be between $20 and $25 per ton
(OSU-E, 2011). Accounting for inflation, this range becomes $21.59 to $26.99 in 2015 dollars. Based
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
on this calculation, it was assumed that the maximum farmers would be willing to pay for gypsum
was $27 per ton. To account for the potential subsidies from USDA and other sources, EPA assumed
farmers might be compensated for up to half the cost of gypsum-related costs, resulting in a total
allowable cost of $54 per ton. EPA used this value to draw a distance around each source of gypsum
that corresponds to this maximum cost.
To calculate this maximum distance, EPA summed the cumulative costs of purchase, transport and
application. EPA first identified the sources of FGD and mined gypsum. Data on active electric
utilities listed as generating FGD gypsum were drawn from the U.S. Energy Information
Administration. A total of 115 utilities that generate FGD gypsum were identified. The average
cost of purchase for FGD gypsum was estimated as $4.05 per ton (EIA, 2014). Data on active mines
and quarries that produce mined gypsum were identified using data from Mine Safety and Health
Administration. A total of 61 mines and quarries that produce mined gypsum were identified. The
average cost of mined gypsum was estimated as $9.00 per ton (USGS, 2015). The cost of transport
by truck was set as $0.19 per ton-mile (U.S. DOT, 2016). The transport distance was calculated as
the closest straight-line distance from the source to the boundary of each county. The cost of
application was based on the average field size for each county, calculated from data from the 2012
USDA Census of Agriculture based on total acreage and number of farms (USDA, 2012). The range
of application rates identified for each use were considered to identify the furthest distance gypsum
may be economical. The cost of application was estimated to be as the same as the cost of spreading
lime at $4.39 per acre (Bongiovanni and Lowenberg-Deboer, 2000). Based on the field size and
distance to closest gypsum source for each county boundary, a cumulative cost was calculated for
both mined and FGD gypsum. If the cost of either material was below $54 per ton for a county, its
use was considered economical. If both FGD and mined gypsum were economical for a given
county, it was assumed farmers would choose the more affordable source.
Figure C-1 presents the maximum economic feasibility zone for FGD gypsum by county. Because
no sources of FGD gypsum were identified in Alaska or Hawaii, these states were not considered
in the analysis. A total of 87 out of 3,108 counties in the continental United States did not have
sufficient information in the agricultural census and could not be assigned. Therefore, if these
counties fell adjacent to one for which FGD gypsum was the most economical, then it was assumed
that FGD gypsum would also be used in that county. However, if it was surrounded on all sides by
an area without FGD gypsum, then it was assumed mined gypsum or no gypsum was the more
economical choice.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
No gypsum
FGD Gypsum
Mined Gypsum
FGD Gypsum Source
Mined Gypsum Source
600
Mies
Figure C-1: Economic Feasibility Zones for FGD Gypsum
This economic feasibility zone was used as an initial boundary on the geographic extent that FGD
gypsum might be applied. Because this boundary is based purely on economic feasibility, it does
not indicate whether cropland is present or whether application of gypsum would provide a benefit
in that area. Therefore, EPA used field boundaries together with soil and crop data to further refine
the boundaries for each use. The data used to define the boundaries are described in the following
subsections. These boundaries are intended to capture the widest range that this secondary
material might conceivably provide a benefit and should not be construed to mean that FGD
gypsum will be used over the entire areas shown.
C.3.1 Reduce Soluble Phosphorus
EPA reviewed the methods available to estimate the net amount of soluble phosphorus present in
different regions of the country and concluded that the Nutrient Use Geographic Information
System (NuGIS) developed by the International Plant Nutrition Institute provided the most
current estimate (IPNI, 2012a,b). NuGIS used information from the USDA Census of Agriculture
(USDA, 2012) to estimate the rate of phosphorus input, biological fixation, and removal by crops
to obtain an annual net balance of phosphorus. Positive balances mean that there is more soluble
phosphorus present than needed for crop production. A positive balance does not mean that
nutrient runoff is or will become an issue, but it provides an indication of the areas where gypsum
application would be most likely. The smallest relevant geographic unit available in NuGIS is a
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix C: Use Characterization
-------
HUC8. Therefore, EPA defined the use area as any HUC8 (and all HUC10 contained therein) with
a net phosphorus greater than zero based on the most recent year data was available for that HUC8.
Figure C-2 depicts the geographic area over which FGD gypsum was evaluated to limit runoff of
soluble phosphorus.
Figure C-2: Use Zone to Reduce Soluble Phosphorus
C.3.2 Nutrient Amendment
EPA consulted with USDA to identify the crops that exhibit a particular sensitivity to deficiencies
of either calcium or sulfur in the soil (Boem et al., 2007; Yencho et aL, 2008; DeSutter et. aL, 2011).
Deficiencies in the soil are treated as distinct from deficiencies that result from physiological
conditions or environmental stress. Based on consultation with USDA, the following crops were
identified as those most likely to benefit from the application of FGD gypsum as a nutrient
amendment (Chaney, 2016):
¦ Calcium sensitive: broccoli, cabbage, peanut, potato and tomato
¦ Sulfur sensitive: alfalfa, canola/rapeseed, cauliflower, mustard/kale greens, radish, sugar beet
and turnip
EPA used the 2012 United States Census of Agriculture to identify counties in which these crops
were grown. If more than 100 acres of any of the listed crops were grown in a county in the 2002,
2007 or 2012 census, then that county was included (USDA, 2002; 2007; 2012). This threshold was
used to determine where widespread application of gypsum most likely drive risks would be most
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 10
Appendix C: Use Characterization
-------
likely to occur. Because this beneficial use evaluation was organized around individual HUG, any
HUC10 that overlapped with one of the identified counties was included. This could result in an
overestimation of the area where gypsum is likely to be applied. Figure C-3 depicts the geographic
area over which FGD gypsum was evaluated for use as a nutrient amendment.
Figure C-3: Use Zone for Nutrient Amendment
C.3.3 Improve Infiltration
EPA used soil characterization data from the National Cooperative Soil Survey (NCSS) Soil
Characterization Database (NCSS, 2016). This database includes site- and depth-specific chemical
and physical soil characteristics. To determine regions that may be susceptible to surface crusting
and reduced infiltration from calcium depletion, EPA used the measured cation exchange capacity
(CEC) and percent of soil exchange sites saturated with basic cations (e.g., Ca+2, Na ). EPA
calculated the fraction of the total exchange sites occupied by Ca+2 (Ca0/o) with data from Table
"CEC_and_Bases" according following hierarchy, based on available data:
¦ Cao^ was calculated as the Mehlich extractable Ca+2 (CaNH4) divided by total base saturation,
measured at a pH of 7 (basesa ).
¦ If a sample had no measured value for basesa , then CaNI) was divided by base saturation,
measured at a pH of 8.2 (baseca ).
¦ If a sample had no measured value for either basesa or baseca , then CaNH was divided by
total cation exchange capacity (CECNH).
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ -j -j
Appendix C: Use Characterization
-------
¦ If a sample had no measured value for basesa , baseca or CECNH4, then CaNH^ was divided by
the sum of major extractable bases (CaNH4 + MgNH4 + Na.NH + KNH4).
USDA (2015b) indicates that a base saturation equal to 70% is the bottom of the balanced range.
EPA therefore filtered for samples with Ca0/o < 70%. If Ca0/o was found to be below 70% in one or
more soil sample within agricultural fields in a given HUG 10, that HUG 10 was included in the
evaluation. There are a substantial number 1IUC10 for which no data are available. EPA also
retained these HUC10 out of an abundance of caution. Figure C-4 depicts the geographic area over
which FGD gypsum was evaluated for use to improve infiltration.
150 300 450
Mi les
Figure C-4: Use Zone for Improved Infiltration
C.3.4 Ameliorate Sodic Soils
There is general agreement in the literature that a sodium adsorption ratio (SAR) above 13 results
in sodic conditions harmful to plants, although levels below 13 have also been found to be harmful
(ASCE, 2012). The Colorado State Extension recommends a final SAR below 10 after treatment
with gypsum (CSU-E, 2012). EPA used data from the Soil Survey Geographic Database (USDA,
2016a) to calculate an average SAR over the top 36 inches of the soil column, intended to reflect
the possible root zone. If a Soil Survey Geographic Database (SSURGO) map unit with a SAR
greater than 10 overlapped with any of the agricultural field in a HUC10, the entire HUC10 was
retained for further evaluation. Figure C-5 depicts the geographic area over which FGD gypsum
was evaluated for sodic soils.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 12
Appendix C: Use Characterization
-------
Figure C-5: Use Zone for Sodic Soils
C.3.5 Subsoil Aluminum Toxicity
The Natural Resource Conservation Service uses a taxonomic hierarchy to classify all soils in the
U.S. into 11 soil orders that are subdivided into over 24,000 individual soil series. Davis (2016)
advised that soils associated with subsoil A1 toxicity ".. .would probably rest in the Ultisol or Oxisol
soil order, those soils with sesquic and kaolinitic minerology and low pH subsoils." According to
the NRCS Soil Series Extent Mapping Tool, Ultisols are found in the FGDG economic feasibility
zone, but Oxisols are not (USDA, 2016b). EPA identified areas of potential use with data from the
SSURGO, a digital soil survey that is the most detailed level of soil geographic data developed by
the National Cooperative Soil Survey (USDA, 2016a). EPA used the SSURGO database to identify
units in which at least 75% of the soils were Ultisols. If one of these map units overlapped with
any of the agricultural field in a HUC10, the entire HUC10 was retained for further evaluation.
Figure C-6 depicts the geographic area over which FGD gypsum might be used to ameliorate
aluminum toxicity.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 13
Appendix C: Use Characterization
-------
Figure C-6: Use Zone for Subsurface Aluminum Toxicity
C.4 Field Properties
EPA used the delineated field boundaries together with other databases to define the model inputs
used to characterize the environmental media within and around the fields. The following text
describes how the field boundaries were generally used to select data and additional criteria used
to ensure the inputs were relevant and appropriate. Further discussion of how the assembled data
were used in the full-scale models is discussed in Appendix E (Probabilistic Modeling).
C.4.1 Distance to Receptors
Distance to receptor is a key factor in the evaluation of constituent fate and transport in ground
water because greater distances allow for more dilution and attenuation prior to exposure. EPA
used the National Hydrography Dataset Plus (NHDplus) flowlines. No publicly-available data is
available for actual well locations. Instead, EPA used synthetic population data which estimates
the most likely location of households in a given area.
Synthetic households and residents were placed to match the population distribution estimated by
the LandScan USA 90-meter gridded population data set (Bhaduri et al., 2007), which distributes
the US population across a grid of 90-meter square cells using a combination of satellite imagery
and other geographic data layers, which include 2000 Census boundaries. The number of
households is constrained by the population reported in a given 2000 Census block. Once the
correct number of households is generated for a 90-meter grid cell, they are placed randomly
within that 90-meter area. As a result, the synthetic population provides estimates of household
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix C: Use Characterization
-------
locations at a finer resolution than is available from the 2000 Census data alone (Wheaton et al.,
2009; Grefenstette et al., 2013). Although these household locations may not coincide with actual
houses, the locations provide a representative distribution of likely home locations.
LandScan does not provide any information on which houses obtain water from private wells.
Instead, EPA relied on data collected as part of the 1990 Census. For each census block, the
percentage was calculated by dividing the total number of households that rely on drilled or dug
wells by the total number of households. Each synthetic household location was then linked with
the corresponding 1990 Census block group boundary so that the calculated percentages were
transferred to each household. A SQL query was used to sum the number of households in each
1990 census block group and to calculate the total number of households on well in that block.
The query then randomly selected this number of households on private within each block group
and flagged each residence for use in the evaluation.
Figure C-7 provides an example of the relative location of fields used in the modeling, streams and
synthetic population households. The households are coded by the source of drinking water. As
can be seen in this figure, the placement of synthetic households tends to be concentrated near
roadways, impervious surfaces and other indicators of human activity. In addition, in rural areas
that tend to have the highest concentration of agricultural fields, the majority of homes are
reported as relying on private wells as a source of drinking water. Therefore, there is minimal
additional uncertainty introduced through the use of well data from the 1990 Census data.
Figure C-7. Synthetic population locations by water source in the vicinity of agricultural fields.
This evaluation considered potential risks to highly exposed individuals, which are hypothetical
receptors that reflect a upper bound on realistic exposures that might occur within the exposed
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 15
Appendix C: Use Characterization
-------
population. To best capture these receptors, A GIS algorithm was applied to develop a distribution
of distances from the each agricultural field to the nearest receptor (i.e., water body, household).
Values were aggregated at the state level because it is believed that population mobility will not
have a major impact on the overall distribution at this scale. Because the fate and transport is not
necessarily limited by state boundaries, EPA allowed the nearest receptor to be located within an
adjacent state for fields near these boundaries.
To develop distributions for nearest water body, both fields and NHD flowlines were projected to
an equidistant projection to preserve distance. EPA calculated a straight-line distance from each
field boundary to the nearest NHDplus flowline with the Near command in ArcGIS using the
closest linear feature option. EPA aggregated distances from any HUC12 catchment that
intersected the state border. The distribution of distances for headwater and mainstem streams
were similar and so a single distribution was complied that included all water bodies, regardless of
stream size. During review of the distributions for each state, a maximum distance was set at 1,609
meters (1 mile), chosen as the approximate 95th percentile of all measured distances. All distances
greater than this maximum were capped at that value. The purpose of this maximum distance is to
limit model run times and compounding uncertainty from longer distance model runs. A number
of fields were found to be directly adjacent to water bodies, which may be a result of real-world
conditions or because buffer areas were not identified when delineating field areas. This boundary
condition has the potential to return anomalous results because there is no space for the infiltrating
leachate to mix with the water table before discharge into the water body. Therefore, a setback
distance of one meter was assigned to each of these fields. Figure C-8 presents an example
distributions for distance to nearest water body.
100%
90%
80%
70%
60%
_0J
'¦P
5 50%
u
i—
V
40%
30%
20%
10%
0%
0 200 400 600 800 1000 1200 1400 1600 1800
Distance to Nearest Surface Water Body [m]
Figure C-8. Cumulative probability distribution of distance from the edge of
agricultural fields to the nearest surface water body.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 16
Appendix C: Use Characterization
-------
To calculate the distance to nearest well, both fields and the synthetic households were projected
to an equidistant projection to preserve distance. EPA calculated a straight-line distance from each
field boundary to the nearest household with the Near command in ArcGIS using the closest linear
feature option. EPA aggregated distances from any residence within two miles of a state border. A
maximum distance of 3,219 meters (2 miles) was established based on the 95th percentile of all
distances to nearest households. All distances greater than this maximum were capped at that
value. The purpose of this maximum distance is to limit model run times and compounding
uncertainty from longer distance model runs. Because the placement of synthetic households is
randomized, some were located within agricultural fields. In these instances, a default distance of
15 meters was used instead based on a minimum setback distance recommended to protect water
supplies from agricultural runoff (U.S. EPA, 2002). Figure C-9 presents an example distributions for
distance to nearest residence.
Distance to Nearest Household [m]
Figure C-9. Cumulative probability distributions for the distance from the edge of
agricultural fields to the nearest ground water well.
C.4.1 Soil Properties
Soil properties are key factors in the evaluation of constituent fate and transport in ground water
because they determine the extent to which contaminant can be released from and migrate
through the soil. EPA used the Soil Survey Geographic (SSURGO) database to identify relevant soil
pH, soil texture, bulk density and other inputs from within field boundaries. In instances where
the soil pH reported within a field fell outside of the range of 5 to 8 considered in this evaluation,
these values were filtered out of the ultimate distribution because it is unlikely that the soil would
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 17
Appendix C: Use Characterization
-------
support commercial agriculture without application of other soil amendments, such as lime, to first
adjust the pH. Filtering of these pH values did not affect data collected for any other variable.
Further discussion on how the data from the SSURGO database were processed and incorporated
into the model is discussed in Appendix E (Probabilistic Modeling). Figure C-10 presents a summary
of the prevalence of different soil pH and textures used in the model runs based on all fields
considered across the country. These distributions would vary for different uses of FGD gypsum.
25%
20%
>
u
c
OJ
3 15%
CT
0)
0)
> 10%
4->
_ro
a>
oc
5%
0%
5-5.5 5.5-6 6-6.5 6.5-7 7-7.5 7
Q
-
-------
C.7 References
ACAA (American Coal Ash Association). 2018. "2017 Coal Combustion Product (CCP) Production
& Use Survey Report." November.
APC (American Peanut Council). 2015. "The Peanut Industry: Growing & Harvesting." Available
at: www.peanutsusa.com/about-peanuts/the-peanut-industry3/ll-growing-harvesting.html.
Last accessed 1/10/16.
ASCE (American Society of Civil Engineers). 2012. "Agricultural Salinity Assessment and
Management, Second Edition. Manuals of Practice 71." W.W. Wallender and K.K. Tanji (ed.).
American Society of Civil Engineering. Reston, VA.
Adeli, A., J. Sheng, J.N. Jenkins and G. Feng. 2015. "Composting and Gypsum Amendment of
Broiler Litter to Reduce Nutrient Leaching Loss." Journal of Environmental Quality. 44:676-
683
Amezketa, E., R. Aragiies and R. Gazol. 2005. "Efficiency of Sulfuric Acid, Mined Gypsum, and
Two Gypsum By-Products in Soil Crusting Prevention and Sodic Soil Reclamation." Agronomy
Journal. 97:983-989.
Ben-Hur, M., R. Stern, A.J. van der Merwe and I. Shainberg. 1992. "Slope and Gypsum Effects on
Infiltration and Erodibility of Dispersive and Nondispersive Soils." Soil Science Society of
America Journal. 56:1571-1576.
Bhaduri, B., E. Bright, P. Coleman, and M. Urban. "LandScan USA: a High-Resolution Geospatial
and Temporal Modeling Approach for Population Distribution and Dynamics." Geojournal.
69(1/2):103-117.
Boem, F.H.G., P. Prystupa and G. Ferrais. 2007. "Seed Number and Yield Determination in Sulfur
Deficient Soybean Crops." Journal of Plant Nutrition. 30:93-104.
Bongiovanni, R. and and J. Lowenberg-DeBoer. 2000. "Economics of Variable Rate Lime in
Indiana." Precision Agriculture. 2(1): 55-70.
Brauer, D., G.R. Aiken, D.H. Pote, S.J. Livingston, L.D. Norton, T.R. Way and J.H Edwards. 2005.
"Amendment Effects on Soil Test Phosphorus." Journal of Environmental Quality. 34:1682-
1686.
Buckley, M.E. and R.P. Wolkowski. 2014. "In-Season Effect of Flue Gas Desulfurization Gypsum
on Soil Physical Properties." Journal of Environmental Quality. 43:322-327.
Caires, E.F., F.J. Garbuio, S. Churka and H.A.W. Joris. 2011. "Use of Gypsum for Crop Grain
Production under a Subtropical No-Till Cropping System." Agronomy Journal. 103:1804-1814.
Chaney, R. 2016. Personal Communication on June 7, 2016.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 19
Appendix C: Use Characterization
-------
Chen, L., W.A. Dick and S. Nelson Jr. 2005. "Soil Fertility: Flue Gas Desulfurization Products as
Sulfur Sources for Alfalfa and Soybean." American Society of Agronomy. 97:265-271.
Chen, L., D. Kost and W.A. Dick. 2008. "Flue Gas Desulfurization Products as Sulfur Sources for
Corn." Soil Science Society of America Journal. 72(5):1464-1470.
CSU-E. (Colorado State University-Extension). 2012. "Managing Sodic Soils." Fact Sheet No. 0504.
Prepared by J.G. Davis, R.M. Waskom and T.A. Bauder. Fort Collins, CO.
Davis, S. 2016. Personal Communication on March 31, 2016.
DeSutter, T.M., J. Lukach and L.J. Cihacek. 2011. "Sulfur Fertilization of Canola (Brassica napus)
with Flue Gas Desulfurization Gypsum: An Assessment Study." Communications in Soil
Science and Plant Analysis. 42:2537-2547.
Dick, W. 2015. Personal Communication on June 17, 2015.
EIA (Energy Information Administration). 2014. "Annual Electric Utility Data - EIA-906/920/923
Data File." Available online at: https://www.eia.gov/electricitv/data/eia923/. Last accessed
1/10/16.
Endale, D.M., H.H. Schomerg, D.S. Fisher, D.H. Franklin and M.B. Jenkins. 2014. "Flue Gas
Desulfurization Gypsum: Implication for Runoff and Nutrient Losses Associated with Broiler
Litter Use on Pastures on Ultisols." Journal of Environmental Quality. 43:281-289.
Farina, M.P.W., P. Channon and G.R. Thibaud. 2000a. "A Comparison of Strategies for
Ameliorating Subsoil Acidity: I. Long-Term Growth Effects." Soil Science Society of America
Journal. 64:646-651.
Farina, M.P.W., P. Channon and G.R. Thibaud. 2000b. "A Comparison of Strategies for
Ameliorating Subsoil Acidity: II. Long-Term Soil Effects." Soil Science Society of America
Journal. 64:652-658.
Feldhake, C.M. and K.D. Ritchey. 1996. "Flue Gas Desulfurization Gypsum Improves Orchardgrass
Root Density and Water Extraction in an Acid Subsoil." Plant and Soil. 178:273-281.
FIPR (Florida Institute of Phosphate Research). 1995. Literature Review on Gypsum as a Calcium
and Sulfur Source for Crops and Soils in the Southeastern Unites States. Publication No. 01-
118-118. Prepared by M.E. Sumner. Bartow, FL. August.
Grefenstette, J.J., S.T. Brown, R. Rosenfeld, J. Depasse, N.T.B. Stone, P.C. Cooley, W.D. Wheaton,
A. Fyshe, D.D. Galloway, A. Sriram, H. Guclu, T. Abraham and D.S. Burke. 2013. "FRED (A
Framework for Reconstructing Epidemic Dynamics): An Open-Source Software System for
Modeling Infectious Diseases and Control Strategies Using Census-Based Populations." BMC
Public Health. 13:940.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 20
Appendix C: Use Characterization
-------
Grichar, W.J., B.A. Besler and K.D. Brewer. 2002. "Comparison of Agricultural and Power Plant
By-Product Gypsum for South Texas Peanut Production." Texas Journal of Agriculture and
National Resources. 15:44-50.
IPNI (International Plant Nutrition Institute). 2012a. "NUGIS: A Nutrient Use Geographic
Information System for the U.S." Prepared by P.E. Fixen, R. Williams and Q.B. Rund.
IPNI. 2012b. "NuGIS: Nutrient Use Geographic Information System." Available online at:
www.ipni.net/nugis. Last accessed: 6/26/2015.
KSU-E (Kansas State University-Extension). 1992. "Management of Saline and Sodic Soils." MF-
1022. Prepared by R.E. Lamond and D.A. Whitney. May.
Miller, W.P. and M.E. Sumner. 1997. "Agricultural and Industrial Uses of By-Product Gypsums."
In Agricultural Uses of By-Products and Wastes. Edited by J.E. Rechcigl and H.C. MacKinnon.
American Chemical Society: Washington, DC.
NCSS (National Cooperative Soil Survey). 2016. "National Cooperative Soil Survey
Characterization Database." Available online at: ncsslabdatamart.sc.egov.usda.gov/. Last
accessed on 1/10/16.
OSU-E (Ohio State University-Extension). 2011. "Gypsum as an Agricultural Amendment: General
Use Guidelines." Bulletin 945. Prepared by L.C. Chen and W.A. Dick.
PNE (Pacific Northwest Extension). 2007. "Managing Salt-Affected Soils for Crop Production."
PNW 601-E. Prepared by D.A. Horneck, J.W. Ellsworth, B.G. Hopkins, D.M. Sullivan and R.G.
Stevens. November.
Ritchey, K.D. and J.D. Snuffer. 2002. "Limestone, Gypsum and Magnesium Oxide Influence
Restoration of an Abandoned Appalachian Pasture." Agronomy Journal. 94:830-839.
Stout, W.L., A.N. Sharpley and J. Landa. 2000. "Effectiveness of Coal Combustion By-Products in
Controlling Phosphorus Export from Soils." Journal of Environmental Quality. 29:1239-1244.
Sumner, M.E. 2007. "Soil Chemical Responses to FGD Gypsum and Their Impact on Crop Yields."
[PowerPoint Slides]. Presented at the Workshop on Agricultural and Industrial Uses of FGD
Gypsum in Atlanta, GA.
Toma, M., M.E. Sumner, G. Weeks and M. Saigusa. 1999. "Long-term Effects of Gypsum on Crop
Yield and Subsoil Chemical Properties." Soil Science of America Journal. 39:891-895.
Torbert, H.A. 2015. Personal Communication on June 28, 2015.
Torbert, H.A. and D.B. Watts. 2014. "Impact of Flue Gas Desulfurization Gypsum Application on
Water Quality in a Coastal Plain Soil." Journal of Environmental Quality. 43:273-280.
TAMU-E (Texas A&M University-Extension). 2015. "Vegetable Rotations, Successions and
Intercropping." Prepared by R.E. Roberts.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix C: Use Characterization
-------
UC-E (University of California-Extension). 1997. "Can Gypsum Improve Water Penetration?"
Publication IG8-97. Prepared by B. Peacock.
UC-IPM (University of California Integrated Pest Management Program) 2013. "University of
California Pest Management Guidelines." Available online at: www.ipm.ucdavis.edu/PMG/
r783900611.html. Last accessed on 6/26/2015.
UM/UW-E (University of Minnesota and University of Wisconsin-Extension) 1991. "Peanut."
Prepared by D.H. Putnam, T.M. Teynor and E.A. Oelke of the University of Minnesota and
E.S. Oplinger, K.A. Kelling and J.D. Doll of the University of Wisconsin-Madison.
USDA (United Stated Department of Agriculture). 2002. "2002 Census of Agriculture." Available
online at: agcensus.mannlib.Cornell.edu/AgCensus/censusParts.do?vear=2002. Last accessed on
1/10/16.
USDA. 2007. "2007 Census of Agriculture." Available online at: www.agcensus.usda.gov/
Publications/2007/. Las Accessed on 1/10/16.
USDA. 2012. "2012 Census of Agriculture." Available online at: www.agcensus.usda.gov/
Publications/2012/. Las Accessed on 1/10/16.
USDA. 2014. "2012 Census of Agriculture." AC-12-A-51. Prepared by the USDA National
Agricultural Statistics Service, Washington, DC. May.
USDA. 2015a. "Conservation Practice Standard: Amending Soil Properties with Gypsum
Products." Standard 333-CPS-l. Prepared by the USDA Natural Resources Conservation
Service. Washington, DC. June.
USDA. 2015b. "Published Crop-Specific Data Layer." National Agricultural Statistics Service.
Washington, DC. Available online at: nassgeodata.gmu.edu/CropScape/. Last accessed on
1/10/16.
USDA. 2016a. "Soil Survey Geographic (SSURGO) Database." Natural Resources Conservation
Service. Available online at: sdmdataaccess.sc.egov.usda.gov. Last accessed on 11/1/16.
USDA. 2016b. "Web Soil Survey." Natural Resources Conservation Service. Available online at:
websoilsurvev.sc.egov.usda. gov/. Last accessed on 11/1/16.
U.S. DOC (United States Department of Commerce). 2013. "TIGER/Line Shapefiles (machine
readable data files)." Prepared by the U.S. Census Bureau. Available online at:
www.census.gov/geo/maps-data/data/tiger-lirie.html
USGS (United States Geological Survey). 2011. "NLCD 2011 Land Cover (2011 Edition)." Available
online at: www.mrlc.gov/
USGS. 2015. "Mineral Commodity Summaries 2015." Reston, VA. January.
U.S. DOT (United States Department of Transportation). 2016. "2016 National Transportation
Statistics." Available online at: www.bts.gov/publications/national transportation statistics.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 22
Appendix C: Use Characterization
-------
U.S. EPA (United States Environmental Protection Agency). 2002. "Drinking Water from
Household Wells." EPA 816-K-02-003. Prepared by the EPA Office of Water. Washington,
DC. January.
UW-E (University Wisconsin-Extension). 2003. "Using Recycled Wallboard for Crop Production."
A3782. Prepared by R.P. Wolkowski for WasteCap Wisconsin, Inc. under a grant from the U.S.
EPA. Madison, WI. May.
Watts, D.B. and A. Torbert. 2009. "Impact of Gypsum Applied to Grass Buffer Strips on Reducing
Soluble P in Surface Water Runoff." Journal of Environmental Quality. 38:1511-1517.
Wheaton, W.D., J.C. Cajka, B.M. Chasteen D.K. Wagener, P.C. Cooley, L. Ganapathi, D.J.
Roberts and J.L. Allpress. 2009. "Synthesized Population Databases: A US Geospatial Database
for Agent-Based Models." Methods Report RTI Press. 10:905.
Yencho, G.C., P.H McCord, K.G. Haynes and S.B.R. Sterrett. 2008. "Internal Heat Necrosis of
Potato - A Review." American Journal of Potato Research. 85:69-76.
Yu, J., T. Lei, I. Shainberg, A.I. Mamedov and G.J. Levy. 2003. "Infiltration and Erosion in Soils
Treated with Dry PAM and Gypsum." Soil Science Society of American Journal. 67:630-636.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 23
Appendix C: Use Characterization
-------
Appendix D. Screening Analysis
Modeling was performed to estimate ambient air mercury concentrations (for the inhalation
pathway) and deposition rates (for the fish ingestion pathway) to evaluate potential exposures
resulting from the volatilization of mercury from FGD gypsum applied to agricultural fields.
D.1 Volatilization Rate
For screening purposes, the mercury volatilization rate was set to 102.4 ng/m2-hr. This value was
calculated as the constant volatilization rate needed to deplete the 90th percentile bulk mercury
concentration before the next annual application. This mercury emission rate was combined with
air dispersion and deposition modeling results to estimate maximum, off-field ambient air
concentrations and deposition rates used to characterize potential exposures posed to highly
exposed individuals living in close proximity to the field.
D.2 Air Dispersion and Deposition Modeling
EPA conducted dispersion modeling to estimate ambient air concentrations and total combined,
wet and dry vapor depositions rates for mercury. Modeling was performed with the American
Meteorological Society/Environmental Protection Agency Regulatory Model (AERMOD; U.S.
EPA, 2011, 2015) to produce results for the inhalation screen and to develop deposition rates to
evaluate surface water impacts and estimate fish ingestion exposures.
Modeling was performed for a representative 427-acre farm scenario using five years of
meteorological data. This representative FGD application scenario was formulated based on the
U.S. EPA's Office of Pesticide Programs (OPP) FIRST model. This pesticide exposure assessment
scenario was developed to characterize human exposures resulting from the ingestion of drinking
water obtained from an index reservoir (U.S. EPA, 2008). The reservoir simulated by OPP
modeling is an actual small drinking water reservoir located in Shipman, Illinois. Shipman City
Lake is 13 acres in area, 9 feet deep, has a mean hydraulic residence time of 6 months, a watershed
area of 427 acres, and a drainage area to capacity ratio (volume of water in the lake) of
approximately 12. Under the current FGD gypsum evaluation, it is assumed that FGD gypsum is
applied to the watershed (representing a crop field) for 100 years. Modeling was performed using
meteorological and land use data from three representative National Weather Service (NWS)
stations located in areas where FGD gypsum use is economically feasible. These locations, which
represent a range of climatic conditions, included Chicago O'Hare, Charleston, S.C., and Seattle.
AERMOD modeling was conducting based on a unitized emission rate (e.g., 1 mg/m2-s). The
resulting air concentrations and deposition rates are called unitized air concentrations (e.g., |ig/m3
per unit emission rate of 1 mg/m2-s) and unitized deposition rates (e.g., g/m2-yr per unit emission
rate of 1 mg/m2-s). These are multiplied by the elemental mercury emission rate, along with
Beneficial Use Evaluation of FGD Gypsum in Agriculture p ^
Appendix D: Screening Analysis
-------
appropriate conversion factors, to calculate the vapor-phase air concentrations and deposition
rates. Modeling results for the three representative meteorological locations are presented in
Table D-1. Considering both air concentration and total deposition outputs, Chicago O'Hare
outputs were identified as the most conservative for use as input to the screening. As shown below
in Figure D-1, Chicago O'Hare is also the NWS station associated with meteorological location
where Shipman City Lake is located.
Table D-1. AERMOD Maximum Unitized Air Concentrations and Depositions Rates
Unitized Annual Average AERMOD Output
National Weather Station
Chicago
O'Hare, IL
Charleston,
S.C.
Seattle, OR
Air concentration from vapor phase
ug/m3 per mg/(s-m2)
3.1E+04
3.1E+04
2.7E+04
Dry deposition from vapor phase
[(g/m2/yr) per (mg/(s-m2))]
1.0
1.6
1.2
Wet deposition from vapor phase
[(g/m2/yr) per (mg/(s-m2))]
1.4E-04
1.8E-04
2.5E-04
Total (wet and dry) deposition from vapor phase
[(g/m2/yr) per (mg/(s-m2))]
1.0
1.6
1.2
Modeling was performed for elemental mercury using the constituent specific inputs shown in
Table D-2. These parameters include diffusivity in air (Da) for the pollutant being modeled (cm2/s),
diffusivity in water (Dw) for the pollutant being modeled (cm2/s), the cuticular resistance to uptake
by lipids (rcl) for individual leaves (s/cm), and Henry's law constant (Pa-m3/mol).These parameters
are used by AERMOD in estimating dry and wet depositions for gaseous pollutants. Chronic
exposures to elemental mercury (assumed to bioaccumulate as methyl mercury for fish ingestion
exposures) were evaluated using outputs averaged over one year.
Seattle
Salem
Bismark
•Burlington Portland
Billings]
Minneapolis
Hartford
Muskegeon
Winnemucca
Grand Island1
iSan Francisco
Boulder
Fresno
Huntington I ~
[La? Vegas
JK5s«ngeles
[A[buquerque
Phoenix
Shrevport
New Orleans
Houston1
Figure D-1. Meteorological stations and regions.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix D: Screening Analysis
-------
Table D-2. Elemental Mercury Modeling Parameters
Parameter
Value
Reference
Diffusivity in air (Da) (cm2/s)
1.1E-02
U.S. EPA (2005b)
Diffusivity in water (Dw) (cm2/s)
3.0E-05
U.S. EPA (2005b)
Cuticular resistance to uptake by lipids (rcl)
for individual leaves (s/cm)
1.0E+05
ANL (2002)
Henry's Law constant (Pa m3/mol)
7.2E+02
U.S. EPA (2005b)
Air concentrations and deposition rates were evaluated at a range of distances included in previous
EPA analyses and models, such as the Multimedia, Multiple Exposure Pathway, Multiple Receptor
Risk Assessment (3MRA) modeling system (U.S. EPA, 2003), the solvent wipes risk assessment
(U.S. EPA, 2009), the CCR screening level analysis (U.S. EPA, 2014). The specific distances (0 m,
25 m, 50 m, 75 m, 150 m, 300 m, 500 m, 1,000 m, and 2,000 m) were selected to ensure complete
coverage in the air and deposition estimates, particularly near the source of the emissions.
To capture national variability in the estimates, the air dispersion and dispersion modeling
considered representative meteorological locations where FGD gypsum will likely be applied. The
representative locations were selected from the 41 U.S. EPA Office of Water land application
meteorological regions and locations. Because it was not feasible to run AERMOD (to obtain results
for longer averaging times) for all of the 41 meteorological stations, EPA performed AERMOD
modeling using meteorological and land use data for three locations where FGD gypsum could be
applied. AERMOD outputs were generated for the range of receptor distances (0, 25 m, 50 m, 75
m, 150 m, 300 m, 500 m, 1,000 m, and 2,000 m) and included unitized annual average vapor air
concentrations and yearly average wet and dry vapor deposition rates for elemental mercury. EPA
then used these AERMOD outputs to estimate location specific ambient air concentrations and
deposition rates for the chronic screening.
D.3 Air and Surface Water Impacts
The maximum ambient air concentrations were estimated using the constant emission rate of
102.4 ng/m2-hr and the maximum, off-farm AERMOD vapor air concentration. The resulting
maximum ambient air concentration was compared to the elemental mercury reference
concentration, resulting in a screening ratio orders of magnitude below levels of concern.
Surface water impacts from wet and dry deposition of mercury vapor were also estimated using
the constant assumed mercury volatilization rate and AERMOD deposition rates for elemental
mercury. The dissolved mercury concentration was calculated using surface water equations
presented in Chapter 5 and Appendix B of U.S. EPA's Human Health Risk Assessment Protocol
(HHRAP) for Hazardous Combustion Facilities (U.S. EPA, 2005b). The maximum air concentration
and total unitized wet and dry deposition rates of mercury from the vapor phase associated with
the Chicago O'Hare NWS station presented in Table D-1 were used with HHRAP equations to
Beneficial Use Evaluation of FGD Gypsum in Agriculture p ^
Appendix D: Screening Analysis
-------
estimate the dissolved fraction of total mercury in the water body resulting from deposition and
vapor phase diffusion from the air to a quiescent water body.
HHRAP Equations B-4-7, B-4-8, B-4-12 through B-4-24 (excluding B-4-14) and Equation 5-36c
were used to estimate mass loadings and losses through the air-water interface and equilibrium
mass partitioning between sediment, suspended and settled, and the dissolved phases. The index
reservoir specific data was combined with constituent-specific information presented in Table D-
3 for dissolved mercury species, as inputs to the HHRAP equations. HHRAP default values were
assumed in lieu of unknown site-specific data including the 85/15 apportionment assumption
between divalent and methyl-mercury species. The dissolved surface water concentration was
compared to the recreational and subsistence fish ingestion screening-level benchmarks and was
found to be below a level of concern with screening ratios orders of magnitude below levels of
concern.
Table D-3. Mercury Surface Water Modeling Parameters
Parameter
Value
Reference
Diffusivity in air (Da) (cm2/s) for Divalent Mercury
5.2E-02
U.S. EPA (2014)
Diffusivity in water (Dw) (cm2/s) for Divalent Mercury
1.8E-05
U.S. EPA (2014)
Henry's law constant (atm-m3/mol) for Divalent Mercury
7.1E-10
U.S. EPA (2014)
Suspended sediments partitioning Coefficient (Log Kd)
(L/g) for Divalent Mercury
5.3
U.S. EPA (2005a), Table 5
Bed sediments partitioning Coefficient (Log Kd) (L/g)
for Divalent Mercury
4.9
U.S. EPA (2005a), Table 4
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix D: Screening Analysis
-------
D.4 References
ANL (Argonne National Laboratory). 2002. "Deposition Parameterizations for the Industrial
Source Complex (ISC3) Model." ANL/ER/TR-01/003. Prepared by M.I. Wesely, P.V. Doskey
and J.D. Shannon for DOE and EPA under Contract No. W-31-109-Eng-38. Argonne, IL. June.
U.S. EPA (Environmental Protection Agency). 2003. "Multimedia, Multipathway, and
Multireceptor Risk Assessment (3MRA) Modeling System. Volume I: Modeling System and
Science." EPA530-D-03-001a. Prepared by the EPA Office of Research and Development and
Office of Solid Waste. July.
U.S. EPA. 2005a. "Partition Coefficients for Metals in Surface Water Soil and Waste." EPA-600-R-
05-074. Prepared by J.D. Allison and T.L. Allison of HydroGeoLogic, Inc. and Allison
Geoscience Consultants, Inc. for the EPA Office of Research and Development under Contract
68-C6-0020. Washington, DC. July.
U.S. EPA. 2005b. "Human Health Risk Assessment Protocol (HHRAP) for Hazardous Waste
Combustion Facilities." EPA-530-R-05-006. Prepared by the EPA Office of Solid Waste and
Emergency Response. Washington, DC. September.
U.S. EPA. 2008. "FIRST: A Screening Model to Estimate Pesticide Concentrations in Drinking
Water, Version 1.1.1." Prepared by the EPA Office of Pesticide Programs. Washington, DC.
March.
U.S. EPA. 2009. "Risk-Based Mass Loading Limits for Solvents in Disposed Wipes and Laundry
Sludges Managed in Municipal Landfills." Prepared by RTI International for the EPA Office of
Resource Conservation and Recovery under Contract EP-W-09-004. Washington, DC. October.
U.S. EPA. 2011. "Addendum: User's Guide for the AMS/EPA Regulatory Model - AERMOD." EPA-
454/B-03-001. Prepared by the EPA Office of Air and Radiation. Research Triangle Park, NC.
February.
U.S. EPA. 2014. "Human and Ecological Risk Assessment of Coal Combustion Wastes." Regulation
Identifier Number: 2050-AE81. Prepared by the EPA Office of Solid Waste and Emergency
Response. Washington, DC. December.
U.S. EPA. 2015. "AERMOD Implementation Guide." Prepared by the EPA Office of Air Quality
Planning and Standards. Research Triangle Park, NC. August.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix D: Screening Analysis
-------
Appendix E: Probabilistic Analysis
The probabilistic analysis conducted for this evaluation considered risks that result from releases
to soil, ground water and surface water. Modeling these pathways required calculating the flux of
constituent mass to overland runoff and subsurface infiltration, as well as modeling the fate and
transport each constituent within aquifers and surface water bodies. This appendix discusses the
models and equations used to model concentrations in each media, as well as the data used in each.
E.1 Data Sources
The scale of data assembly, analysis and application in this evaluation was conducted at the
watershed and sub-watershed scales, unless otherwise noted. Watersheds in the United States have
been delineated by the U.S. Geological Survey (USGS) using a national standard hierarchical
system based on surface hydrologic features and are classified into six types of hydrologic units.
Watersheds and sub-watersheds are assigned the hydrologic unit code (HUC) of 10 and 12,
respectively, corresponding to the number of digits in their unique identifiers. HUC 10 and HUC 12
will be used frequently hereafter to refer to these watershed and sub-watershed scales.
E.I .1 Soil Type
The primary data source for soil properties was the Soil Survey Geographic (SSURGO) database.
SSURGO is a repository of nationwide soil properties collected by the National Cooperative Soil
Survey over the last century (USDA, 2017). SSURGO data were collected at scales ranging from
1:12,000 to 1:63,360 and are linked to map unit polygons ranging between 1 and 10 acres. These
map units provide the finest spatial resolution and span most of the conterminous United States.
Soil attributes linked to these map unit polygons are stored within a relational database broken out
by soil component and soil horizon. Each map unit contains data on the prevalence of each
component and horizon within the map unit. Data extracted from SSURGO were evaluated and
used at the HUC 10 scale. EPA extracted two types of data from SSURGO:
1. Measured (i.e., numerical) data are those that can be weighted by soil horizon and component.
These data include pH, percent organic matter and percent silt. Measured data were extracted
for the top 20 cm of the soil column (i.e., root zone) and weighted by the thickness of each
horizon present to obtain a representative value for each soil component. Component values
were then weighted by the relative prevalence of each component to obtain a representative
value for the entire map unit.
2. Categorical data (i.e. non-numerical data) are those that can be characterized by a dominant
type. These data include soil texture and hydrogeological group. First, the characteristic of the
dominant horizon within each component was identified and assigned. Then, the
characteristic of the dominant component was identified, and was assigned to the map unit.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
Table E-1. Soil Texture Crosswalk
All tabular data (weighted or dominant)
were joined to a 30 m gridded (raster)
version of the SSURGO data, since joining to
the polygon version presented multiple
processing and display problems due to the
very large number of map unit polygons
(~36 million) in the continental United
States. Producing a raster version of
SSURGO soils data (using the ArcGIS
Lookup command) enabled tabulations of
soil parameters over several polygon
features (HUC10, county).
Soil texture data provide data on bulk
density, saturated water content, saturated
hydraulic conductivity, and van Genuchten
soil moisture retention parameters (alpha
and beta) correlated on a national scale
based on the work of Carsel and Parrish
(1988) and Carsel et al. (1988). For measured
data, GIS software was used to extract soil
parameter grids within the boundaries of
modeled agricultural fields. Using the
extracted data, EPA calculated mean measured soil parameter values by HUC10. These average
values were used as model inputs for all agricultural fields within a given HUC10. For categorical
soil parameters, a distribution of values was created, which allowed for probabilistic sampling.
Table E-1 shows the crosswalk used to assign the SSURGO detailed soil textures to basic Soil
Conservation Service (SCS) textures, and then to the EPACMTP megatextures. SSURGO soils are
classified into 21 texture classes, which map to 12 SCS textures. EPACMTP uses three soil
megatextures to represent the variability of hydrologic soil properties, so each SSURGO soil texture
was crosswalked to the EPACMTP megatexture with the most similar hydrogeologic properties.
E.1.2 Hydrogeological Environment
Each HUC10 modeled in this analysis was assigned one or more hydrogeologic environment(s)
from EPA's Hydrogeologic Database (HGDB) to characterize four subsurface parameters required
by EPACMTP: depth to ground water, aquifer thickness, hydraulic gradient, and saturated
hydraulic conductivity. The HGDB was developed by the American Petroleum Institute (Newell
et al., 1989; 1990) to specify correlated empirical probability distributions of these four parameters
for the 12 distinct hydrogeologic environments described in Newell et al. (1990). To assign the
HGDB distributions to the HUClOs modeled in this assessment, EPA first developed a national
geographic coverage of the 12 hydrogeologic environments, and then used GIS software to overlay
Detailed SSURGO
Soil Texture
Basic SCS
EPACMTP
Texture
Soil
Megatexture
Loamy Sand
Loamy Coarse Sand
Loamy Sand
Loamy Fine Sand
Loamy Very Fine Sand
Sand
Coarse Sand
Sand
Sandy Loam
Fine Sand
Very Fine Sand
Sandy Loam
Coarse Sandy Loam
Sandy Loam
Fine Sandy Loam
Very Fine Sandy Loam
Silt Loam
Silt Loam
Silt
Silt
Loam
Loam
Silt Loam
Sandy Clay Loam
Sandy Clay Loam
Clay Loam
Clay Loam
Silty Clay Loam
Silty Clay Loam
Sandy Clay
Sandy Clay
Silty Clay
Silty Clay
Silty Clay
Loam
Clay
Clay
SCS = Soil Conservation Service
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
the HUC10 locations and dimensions and assign the hydrogeologic environments to each HUC10.
Of the 12 environments defined, only 9 intersected the distribution of HUClOs used in this
assessment.
EPACMTP uses the HGDB for national and regional analyses. Therefore, it was necessary to assign
each HUC10 to one or more hydrogeologic environments corresponding to the HGDB data set.
Given the national scale of the risk assessment, only national data sets were used to delineate
hydrogeologic environments, defined by an approximate 1:7,500,000 map scale. The following
individual map layers were combined using GIS software to develop a single map layer for
assigning the 12 hydrogeologic environments across the United States:
¦ Shallowest principal aquifers from Principal Aquifers of the Conterminous United States,
Hawaii, Puerto Rico, and the U.S. Virgin Islands [USGS map file: aquifrp025]. 1:2,500,000 map
scale, was used as the base layer in the assessment and to delineate several of the 12
hydrogeologic environments.
¦ Alluvial and glacial aquifers from Aquifers of Alluvial and Glacial Origin [USGS map file:
alvaqfp025]. 1:2,500,000 map scale, was used to represent alluvial and glacial aquifers for the 22
states north of the southernmost line of glaciation. Note that the alluvial aquifers in this
coverage are identical to those in the Hunt (1979) surficial geology layer below.
¦ Surficial geology of the conterminous United States was taken from:
Surficial Geology of the Conterminous United States [map file: geol75m], 1:7,500,000 map
scale, provided by Hunt (1979), these data were used to characterize shallow soil lithology
and alluvial aquifers.
The Surficial Deposits and Materials in the Eastern and Central United States (East of 102
degrees West Longitude) [map file: sfgeoep020]. 1:1,000,00 map scale, includes the line of
maximum glacial advance and represents surficial materials that accumulated or formed
during the past two million years, including residual soils, alluvium, and glacial deposits.
¦ Karst aquifers from Engineering Aspects ofKarst [map file: karst0p075], l:7,500,000-map scale,
showing karst and pseudokarst (i.e., karst-like terrain produced by processes other than the
dissolution of rocks) across the United States.
¦ Bedrock geology from Generalized Geologic Map of the United States [map file: geolgyp075],
1:7,500,00 map scale, showing the bedrock geology at or near land surface (i.e., beneath surficial
soils, alluvium and glacial deposits).
¦ STATSGO soils, 1:250,000 map scale, from the digital map and attribute data for soils.
As described below, these data layers were used to develop a national hydrogeologic environment
layer in GIS for assigning an aquifer type to a point or area of interest. To create the hydrogeologic
environment layer, each individual data layer described above was obtained as a GIS shapefile and
processed, as needed, to ensure that coordinate systems matched and the layers could be overlain.
Table E-2 describes how the polygons comprising the 9 applicable hydrogeologic environments
were developed in the GIS using these layers.
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
Table E-2. GIS Procedures for Assembling National Coverage of Hydrogeologic Environments.
Hydrogeologic
Environment
Description
Metamorphic and
igneous rock
Select polygons where ROCKNAME = "igneous and metamorphic-rock aquifers" from the
principal aquifer layer [aquifrp025] AND "igneous" or "metamorphic" polygons from the
bedrock geology layer [geolgyp075] AND polygons from the Hunt (1979) surficial geology
layer [geol75m] derived from or directly overlaying igneous or metamorphic rock.
Bedded
sedimentary rock
Select polygons where the ROCK NAME = "sandstone and carbonate-rock aquifers" OR
AQNAME was "Other rocks" from the principal aquifer layer [aquifrp025] AND
"sedimentary" polygons from the bedrock geology layer [geolgyp075].
Till over
sedimentary rock
Select polygons that had a surficial geology [geol75m] = "mg: deposits of mountain
glaciers", "tg: till or ground moraine" "ts: ice-laid deposits, like tg but mostly sand and silt",
OR "ts/K,T: thin ice-laid deposits, like ts but thin and discontinuous..AND overlaying
"sedimentary" polygons from the bedrock geology layer [geolgyp075].
4) Sand and gravel
Select polygons where AQ NAME = "other rocks" OR "high plains aquifer" from the
principal aquifer layer [aquifrp025] AND sand and gravel related names in SURFICIALG
from the Hunt (1979) surficial geology layer [geol75m] (SURFICIALG contained many
different types of sand and gravel deposits).
Alluvial valleys,
basins and fans
Select polygons where AQ NAME = "Unconsolidated sand and gravel aquifers" from the
principal aquifer layer [aquifrp025] name OR surficial geology type (SURFICIALG) was either
"fg: fan gravels" or "fs: fan sands" in the Hunt (1979) surficial geology layer [geol75m].
River valleys and
6) floodplains with
overbank deposits
Select polygons where SURFICIALG = "al: floodplain and alluvium gravel terraces" in Hunt
(1979) surficial geology layer [geol75m] AND STATSGO soils with < 50% sand AND a low
permeability (< 0.0147 inches per hour).
River valleys and
7) floodplains without
overbank deposits
Select polygons where SURFICIALG = "al: floodplain and alluvium gravel terraces" in Hunt
(1979) surficial geology layer [geol75m] AND STATSGO soils that do NOT have < 50% sand
AND a low permeability (< 0.0147 inches per hour).
8) Outwash
Select polygons where ORIGIN AGE = "glaciofluvial (outwash) deposits" in the Surficial
Deposits and Materials layer [sfgeomean020] AND where SURFICIALG = anything but "ts:
ice-laid deposits, like tg but mostly sand and silt" in the Hunt (1979) surficial geology layer
[geol75m].
Till and till over
outwash
Select polygons from Hunt (1979) surficial geology layer [geol75,] that were not already
classified hydrogeologic environment 8 AND where SURFICIALG = "w: gravel, sand and clay
deposited by glacial streams adjacent to or downstream from temporary ice fronts" OR "ts:
ice-laid deposits, like tg but mostly sand and silt" OR "tg: till, or ground moraine".
One or more of the nine hydrogeological environments were assigned to each HUC10 based on
overlap of the environments and the HUC10 boundary. When a HUC10 spanned more than one
environment, the hydrogeological environment was varied probabilistically based on relative
percentage when constructing the database of field properties to simulate for that HUC10. Once
hydrogeologic environments were assigned, a preprocessing run of EPACMTP was conducted to
construct a set of randomly generated but correlated hydrogeologic parameters (depth to ground
water, saturated hydraulic conductivity, aquifer thickness, hydraulic gradient) for each occurrence
of the hydrogeologic environments in the source data files. Missing values in the HGDB data set
were filled using correlations during EPACMTP execution, as described in U.S. EPA (1997).
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
E.I .3 Precipitation Data
Forty-one meteorological stations were chosen to represent the nine general climate regions of the
continental United States. Figure E-2 shows these stations and their boundaries. The approach used
to define the forty-one regions included the following three main steps:
1. Identify contiguous areas that have similar environments, as defined by Bailey regions (Bailey
et al., 1994). Bailey's ecoregions and subregions are used to associate coverage areas with
meteorological stations. This hierarchical classification scheme is based primarily on rainfall
regimes; subregions are delineated by elevation and other factor relevant to ecology.
2. Select one meteorological station to represent each contiguous area. Station locations were
selected based on considerations of the following factors:
• Major National Weather Service (NWS) stations were selected because these stations are
expected to have high-quality equipment that is kept in good repair and is suitably sited.
• Number of years of surface-level meteorological data available (minimum of five years).
More years of data provide a more realistic long-term estimate.
• Aimed for locations that are central within each region. All other factors being equal, a
central location is expected to be most representative of the larger, contiguous region
because it has the smallest average distance to all points within that region.
3. Identify the boundaries of the area to be represented by each meteorological station. Thiessen
polygons, which are created by a geographic information systems (GIS) procedure that assigns
every point on a map to the closest station, were used as the first step in drawing the
boundaries. Meteorological boundaries were adjusted to fall along the Bailey boundaries.
Figure E-1. Meteorological stations and regions
All available daily precipitation data associated with a HUC10 and soil characteristics for the
farmland within the region were used by the Land Application Unit (LAU) model. This model
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ g
Appendix E: Probabilistic Analysis
-------
retunecl annual average runoff and infiltration rates over the specified timeframe. Long-term
infiltration and runoff rates were used to calculate concentrations in the leachate flowing to
ground and surface water. Estimates of runoff and infiltration rates always began with the first
year of available precipitation data and proceeded chronologically. If the model duration exceeded
the number of years of precipitation data, the available meteorological data was repeated from the
first year as many times as needed.
E.1.4 Surface Water Location and Navigation
This section describes the data used to define the location of each individual surface water body
and the relationship between these water bodies (i.e. flow rate and direction). This work relied on
data from the Stream-Catchment database (StreamCat; Hill et al., 2016), the enhanced National
Hydrography Dataset (NHDPlus; McKay et al., 2017), and the USGS Watershed Boundary Dataset
(USGS, 2013). Using these data sources, EPA accumulated a number of catchment-level data
attributes, including the navigation relationships for each NHDPlus catchment at the HUC12
levels and above. Assembly and management of such large amounts of data required use of
automated routines performed with GIS software. The following text described how these
processes were implemented, how the quality control (QC) review was conducted to ensure that
the data were assembled properly, and how identified issues were addressed.
Surface water pathways were evaluated at multiple scales: catchment, HUC12, and HUC10.
Catchments are typically smaller than HUC12s, which are always smaller than HUClOs. HUClOs
can easily be derived from HUC12s due to the nested structure of these data. Figure E-3 presents
the conceptual model of flow from the initial catchments to the outfall of a HUC10.
Figure E-2: Conceptual Model for a HUC10 and Associated Outfalls
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
Headwaters: Headwaters were defined using NHDPlus catchments associated with Strahler stream
orders 1 and 2. Although headwater streams have been defined elsewhere as order 3 and below,
EPA limited this evaluation to those that fall entirely within the boundaries of a single HUC12.
Exposures were calculated at the outfall from a headwater to any stream greater than order 2. All
of the catchments upstream of that outfall were merged to obtain total drainage area and other
stream properties. In total, modeling was conducted on this evaluation included a total of 178,506
separate headwater streams.
In some cases the stream order classification was anomalous. For example, in rare cases, headwater
streams flowed long distances before terminating or intersecting a larger stream. This results in a
catchment area that spans multiple HUC12s. Although these are real streams, a catchment area
that extends beyond the bounds of a single HUC12 conflicted with the automated process used to
aggregate headwater data. To address this conflict, EPA removed catchments with a drainage area
greater than half of the area of the HUC12 of origin. To identify these large catchments, EPA used
the "totdasqkm" field (i.e., total drainage area) in the NHDPlus table "plusflowlinevaa" (value-
added attributes; McKay et al., 2017). EPA mapped each catchment to a HUC12 based on the
location of its centroid and compared the catchment's total drainage area with that of the
corresponding HUC12. This was encountered most frequently in particularly dry or flat areas, such
as southwestern deserts and the Everglades, and were often areas outside the scope this evaluation.
Headwaters were reviewed for quality control (QC) through visual spot checks conducted across
the country (10 checks performed in randomly selected states over an area approximately the size
of a HUC4). One check confirmed that the headwater catchments were associated with only Order
1 and 2 stream segments. In instances that an Order 1 or 2 stream was omitted, the catchment area
was compared with the HUC12 boundary to check for the issues described above. Another check
ensured that drainage areas were calculated correctly by manually summing the area of each
catchment along the headwater stream and comparing it to the total area calculated by the
automated routine. During the QC process, it was found that the routine did not combine all the
catchments along some Order 1 streams. EPA could not determine the frequency that this occurred
because it would have required visual inspection of every stream. However, visual inspection
conducted across multiple states as part of this QC effort indicates that it is not common (i.e., <1%
of headwaters). Each of the defined headwaters was modeled as a separate headwater.
Mainstem Navigation: Mainstem streams refers to the primary route of flow through a drainage
system that contains multiple stream segments. Navigation refers to the tracking of surface water
flows through each individual stream segments, beginning with headwater streams and continuing
downstream until reaching either a coast or a stream of Strahler order 6 or higher, which was used
to denote large rivers in this evaluation. This threshold for large rivers resulted from an analysis of
watershed contributions in terms of area and characteristics. Streams of order 6 began to exhibit
trends that did not follow the same pattern as smaller streams, which is believed to result from the
larger drainage areas and greater complexity of upstream contributing areas. This divergence was
observed most clearly when the Base Flow Index (BFI) was mapped. Thus, because concentrations
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
are likely to be lower in these larger streams as a result from dilution from a larger drainage area,
EPA chose to terminate the analysis once a stream of order 6 was reached.
EPA based navigation of the hydrologic network on NHDPlus catchments (McKay et al., 2017)
and a crosswalk of data between these catchments and HUC12s provided within EPA's EnviroAtlas
data layers (Pickard et al., 2015). Navigation of the network using NHDPlus was completed by
identifying "from" and "to" designations for each catchment. All catchments except for the very
first or last should have both designations. This network was used to identify catchments located
within and at the outlet of each HUC12 within the EnviroAtlas framework. EnviroAtlas navigation
of HUC12s identifies three attributes: HUC12, ToHUC12 and OutletCOMID (i.e., NHDPlus
catchment at outlet of a HUC12). EPA used the completed navigation for HUC12s to identify the
HUC10 outfalls (i.e., when the next downstream HUC12 fell within a different HUC10, that
HUC12 was designated as an outlet). Exposures were calculated either at the outfall of each HUC10
or at the outfall of an individual HUC12 if it discharged directly to a higher order stream.
Several issues were identified during the navigation process. Although NHDPlus incorporates
information from the USGS Watershed Boundary Dataset, there are discrepancies between the
boundaries in the two datasets that can derail automated navigation. These discrepancies were
identified through QC checks run on automated navigation by review of generated tabular data,
comparison of the cumulative drainage areas for outlet catchments and the associated HUC12, and
visual inspection of the HUC12 routing network.
Closed Basins: These basins have an internal sink to which they drain (i.e., losing streams). They
do not flow to a larger stream network or out to the coast. These basins may consist of any
number of HUC12s (i.e., from one to several dozen). Some closed basins could be identified
from the original HUC12 navigation which flagged the downstream HUC12 as "Closed Basin."
Others had to be mapped and visually identified. All closed basins were removed from the
modeling analysis because the disconnected hydrology introduced a great deal of uncertainty
into the evaluation.
Scale Issues: There were a small number of HUC12s that were oddly shaped or that were smaller
than the identified outlet catchment. Depending on the location of these HUC12s and the
magnitude of the differences, the outlet catchment was either adjusted to fit the HUC12 or the
HUC12 was removed from the model (i.e., coastal outfalls).
Catchment Issues: The automated process used to define outlet catchments defined for each
HUC12 did not account for instances where the NHDPlus flowline through the catchment was
labeled as a connector, a waterbody, a canal, or some artificial pathway (i.e., path through a
waterbody to ensure continuous flow lines). Upon finding a null, outlier (e.g., extremely small
cumulative drainage area for an outlet catchment), or other confounding value (e.g., negative
cumulative streamflow), EPA visually inspected the specific catchment or HUC12 to determine
a remedy. In almost all cases the remedy was to skip over the individual, anomalous catchment
and define the next downstream catchment as the outlet. Other remedies were specific to the
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
case and may have used the catchment upstream as the outlet or removal of the HUC12 from
the analysis for one of the reasons above.
To support the spatial resolution needed for the evaluation, EPA created a navigation text file for
HUClOs that mimics the file received for the EnviroAtlas navigation of HUC12s. The HUC10 file
contains three fields: HUC10, ToHUCIO, and OutletCOMID. The OutletCOMID is the identifier
of the NHDPlus catchment at the outlet of the HUC10. This catchment corresponds to an outlet
listed in the HUC12 file; therefore, an outlet HUC12 was identified by joining the two files.
HUClOs are the primary spatial resolution at which model results are aggregated for the evaluation,
although individual model runs are conducted at the HUC12 scale. The routing through the
hydrologic network was completed to determine cumulative impacts as a post-processing step. In
total, modeling was conducted for a total of 7,999 modeled HUClOs (comprised of 32,998 HUC12s).
As noted previously, the HUC12 network constructed for this modeling effort had some HUC12s
that were not modeled, resulting in the fragmentation of some HUClOs. Because of this order of
processing, all routing through the network can be calculated, including routing within only the
fragmented pieces and the entire connected network. For instance, the drainage area captured by
the outlet catchment did not always capture the entire HUC12 because of the confluence of
multiple streams at the outlet. Instead, that flow is reflected in the downgradient HUC12. In
addition, overlay of the Economic Feasibility Zone (EFZ) layer created disconnects and gaps within
the HUC12 network. The following list summarizes the discrepancies and disconnects identified
and describes the approaches used to address the issues:
¦ For 4,400 HUClOs (55%), all HUC12 were included and a single exposure point was modeled at
the outfall of the HUC10. No additional steps were necessary to address these areas.
¦ For 3,599 HUClOs (45%), one or more HUC12s within the HUC10 did not contribute
constituent mass to the HUC10 outfall. These HUClOs had one or more exposure point modeled
at either the outfall of the HUC10 or at the outfall of individual HUC12 within the HUC10:
o 1,635 HUClOs (20%) are intersected by stream of order 6 or higher, resulting in a series of
tributaries within the HUC 10 that feed into the large-order stream. EPA modeled exposures
at the outfall of each individual HUC12 that discharged into a large-order stream. HUC12s
that fell along the flow path of the large-order stream were not modeled.
o 1,964 HUClOs (25%) have one or more HUC12s that do not contribute to the HUC10 outfall
due to areas without agricultural land use, areas outside the EFZ, or other similar causes. EPA
modeled flow through these areas the same as every other HUC12. However, it was assumed
that these areas contributed zero constituent mass to the downstream mainstem flow.
Watershed Attributes: To complete the evaluation, a number of attributes were needed for each
of the assessed spatial units: HUC12s and headwaters. EPA compiled the attribute information
using various base data and Value Added Attributes available within and supplemental to
NHDPlus. (rather than summing the incremental flows of catchments within the HUC 12)
Beneficial Use Evaluation of FGD Gypsum in Agriculture
Appendix E: Probabilistic Analysis
-------
Table E-3. Watershed attributes compiled from NHDPIus
Attribute
Source
Description
Cumulative
Baseflow
Index (BFI)
EPA's StreamCat
Database: BFIs for
the entire upstream
watershed (BFIWs)
Source: EPA's Stream-Catchment database (StreamCat; Hill et al., 2016)
provides geospatial attributes indexed to the NHDPIus version 2 dataset. The
BFI is the ratio of base flow to total flow, expressed as a percentage. The BFIWs
attribute summarizes the ratio for the entire upstream watershed. Therefore,
for each headwater and HUC12 the BFIWs corresponding to the outlet
catchment was chosen as the parameter value.
Stream
Length
Calculated based
on NHDPIus
flow/line length
Source: NHDPIus dataset field LengthKm in Flowline file (McKay et al., 2017)
Headwaters: Sum the segment lengths for the individual stream segments
(i.e., stream orders 1 and 2)
Headwater HUC12s: Navigate upstream from the outlet and determine the
longest path. Use the maximum length found as the stream length.
All other HUC12s (with an upstream HUC12): Determine longest path
navigated through the HUC12 from inflow to outlet by navigating the flow path
from the outlet of target HUC12 upstream to the outlet of upstream HUC12.
Then sum the flowline lengths to determine the stream length.
Cumulative
and
Incremental
Streamflow
Streamflow
Velocity
NHDPIus VAA:
Q0001C, and
Qlncr0001C
V0001C
Source: NHDPIus dataset fields Q0001C (cumulative) and Qlncr0001C
(incremental) in file EROM MA0001 (McKay et al., 2017)
All Flow estimates are in cubic feet per second (cfs) and represent the flow at
the bottom (downstream end) of the NHDFIowline feature.
All Velocity computations are in feet per second (fps) using the Jobson Method
(USGS, 1996) and represent the velocity at the bottom of the NHDFIowline
feature.
For incremental flows, the incremental flow from each catchment within the
assessment unit are summed to provide a total incremental flow for the unit
(e.g., if there are 5 catchments within a HUC12 then the 5 incremental flow
values are summed). For cumulative flow, the value corresponding to the outlet
catchment of each assessment unit was selected.
Percent
Cropland
Calculated based
on field area
Source: Compiled for this evaluation (See Appendix C: Use Characterization)
Sum of all field area contained within the corresponding headwater or HUC12
area.
EPA compiled these data based on the identified outlet catchments using an automated process.
As the data were compiled into a tabular format for each assessment unit, EPA identified places
where the available data were missing or anomalous. A summary of the issues by assessment unit
type are described below.
¦ There were eight modeled headwaters (0.02%) missing BFIWs values. All eight were located
in tidal areas and were removed from the analysis.
¦ There were 73 modeled HUC12s (0.2%) missing BFIWs values. EPA removed four of these
HUC12s (0.01%) from the modeling analysis because they were identified as either tidal or
closed basins. For 26 of the modeled HUC12s (0.08%), EPA used the BFIWs value reported for
the next downstream catchment. Finally, for 43 of the modeled HUC12s (0.1%), EPA used an
average BFIWs from all the catchments within the HUC12 because both the outlet and the
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ -|q
Appendix E: Probabilistic Analysis
-------
next downstream catchment were missing values. The variability of BFIWs within a single
HUC12 was generally small and so an average was determined to introduce minimal additional
uncertainty.
¦ There were 43 modeled headwaters (<0.01%) where the incremental flow values reported were
less than or equal to zero. EPA removed 15 of these headwaters from the modeling analysis
after visual inspection because they were located in the middle of a waterbody, such as a
drainage canal or a "connector" (an artificial designation used to ensure that all streams
connect), which do not reflect typical flowing streams. For the remaining 28 headwaters, EPA
used the value reported for the next downstream catchment.
¦ There were 1,318 modeled HUC12s (4%) with incremental flow values less than zero. For 374
of these HUC12s (1%), EPA calculated the incremental flow based on the difference between
the cumulative flows reported for that HUC 12 and the one immediately upgradient. For the
remaining 944 HUC 12s (3%), all efforts to calculate an alternate incremental flow with
NHDPlus data resulted in negative values. Many of these HUC12s had lakes or other larger
water bodies located at the outfall, which likely caused issues in the reported flows. Therefore,
EPA assigned a fixed value of 10 cfs based on a typical values reported for HUC12s with data
and best professional judgment.
Lakes: In addition to rivers and streams, an effort was made to evaluate lakes, reservoirs and other
lentic water bodies, referred to collectively as "lakes" in this discussion for simplicity. To locate
these lakes, EPA selected any water body classified as either a lake or reservoir in NHDPlus. EPA
identified two broad classes of lakes:
¦ Flowthrough lakes are those that fall along the navigated stream network and contribute flow
to downgradient streams. These lakes are frequently located entirely within a single HUC12.
Given the larger cumulative drainage area upgradient of these lakes, the majority of the water
that flows through these water bodies originates from upgradient streams. As a result, it is
expected that the long-term concentrations from mixing within these lakes will be similar to
adjacent streams. Thus, for the purposes of this evaluation, EPA treated these water bodies the
same as streams.
¦ Terminal lakes are those that do not fall within the stream network. Terminal lakes were
designated as such if they touched an NHD flowline with a terminal flag (i.e., NHD attribute)
set to true. These water bodies were often located entirely within a single catchment. Because
these water bodies receive flow primarily from an isolated drainage area, it is more likely that
long-term concentrations could differ from nearby streams. However, there was not enough
data (e.g., depth, volume, percent of catchment area drained) to model these water bodies
without a number of assumptions that would introduce a great deal of uncertainty into model
results. Thus, EPA did not model these water bodies in the evaluation.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £
Appendix E: Probabilistic Analysis
-------
E.I .5 Surface Water Characteristics
Data on regional surface water characteristics were collected from the legacy STORET database.
STORET is the largest single source of water quality data in the United States, containing over 275
million analyses performed on more than 45 million samples collected from 800,000 stations across
the country between 1960 and 1998. However, the STORET website states:
"The EPA does not change or filter incoming data. This means that when pulling
data out of the Warehouse, users must be aware that they are responsible for
screening the data for their use. "
EPA notes that there is a high degree of variability in these data due to differences in quality
assurance/quality control testing, bias towards samples collected at site locations known to have
contamination problems, and bias towards samples collected during critical periods (e.g., summer
low flows). To account for these factors, EPA used the STORET data as discussed below.
Temperature: Median surface water temperatures were collected for each hydrologic region and
assigned to each water body within that region. Median values were selected to capture reflect
annualized values. Table E-4 provides the temperatures used for each hydrologic region.
Table E-4. Regional Surface Water Temperature
Hydrologic
Region
Median Surface Water
Temperature (°C)
1
14
2
16
3
21
4
14
5
17
6
18
7
15
8
20
9
10
10
13
11
17
12
21
13
16
14
9
15
17
16
9
17
11*
* Legacy STORET data not available for region 17 at this time.
Assigned median temperature of 11 based on professional
judgment to represent cooler surface water temperature in
the mountainous Pacific northwest region.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 12
Appendix E: Probabilistic Analysis
-------
Total Suspended Solids (TSS): Minimum, maximum and weighted geometric means of these
collected values were used to define log triangular distributions regionally for streams. The
triangular distribution was selected because it is typically used to describe a population for which
there is only limited sample data, but there is sufficient knowledge to determine that the
distribution is modal, rather than uniform, as was the case here. Geometric means weighted by the
annual number of measurements were used because the actual distribution around the median is
unknown. Once developed, these distributions were sampled during the preparation of the input
data files. Table E-5 provides the TSS values used to define the distribution for each region, along
with the number of the modeled facilities assigned to that region.
Table E-5. Surface Water Tota
Suspended Solids (TSS) Distributions
Hydrologic
Number of
Annual Median
Values
Annual Median TSS
(log-triangular distribution)
Region
Minimum
Weighted
Geometric Mean
Maximum
1
33
3.2
8
40
2
38
10
32
316
3
36
6.3
25
79
4
37
6.3
25
794
5
38
4
25
100
6
28
5
16
316
7
37
32
63
1,585
8
38
50
158
316
9
35
13
32
3,162
10
38
10
126
398
11
38
25
200
794
12
35
40
79
1,995
13
37
32
200
79,433
14
38
16
158
5,012
15
37
20
200
19,953
16
33
4
16
2,512
17
37
2
6
316
Suspended Sediment Partitioning: The model partitions constituent mass between surface water,
suspended solids and benthic sediment with linear partition coefficients. This approach assumes
that equilibrium is maintained among these dissolved constituents within the water column and
constituents in suspended solids and bed sediment. Table E-6 provides distributions for the
partitioning coefficients used by the surface water model. These distributions were derived from
published empirical data presented in U.S. EPA (2005a), Allison et al. (2003) for manganese, and
ORNL (1984) for iron. These data were sampled during the preparation of the input files in the
Monte Carlo process.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 13
Appendix E: Probabilistic Analysis
-------
Table E-6. Log Distribution of Sediment/Water and Suspended Solids/Water Partition Coefficients
Constituent
Sediment/Water
Suspended Solids/Water
Min
Mean
Max
Standard
Deviation
Min
Mean
Max
Standard
Deviation
Antimony
0.6
3.6
4.8
1.8
--
-
-
-
Arsenic
1.6
2.4
4.3
0.7
2.0
3.9
6.0
0.5
Cadmium
0.5
3.3
7.3
1.8
2.8
4.9
6.3
0.6
Chromium III
1.9
4.9
5.9
1.5
3.9
5.1
6.0
0.4
Chromium VI
0
1.7
4.4
1.4
3.6
4.2
5.1
0.5
Iron
--
-
-
-
N/A
1.4
N/A
N/A
Lead
2.0
4.6
7.0
1.9
3.4
5.7
6.5
0.4
Manganese
2.4
3.2
4.7
0.7
4.5
4.7
5.3
0.2
Mercury (divalent)
3.8
4.9
6.0
0.6
4.2
5.3
6.9
0.4
Mercury (methyl)
2.8
3.9
5.0
0.5
4.2
4.9
6.2
0.7
Nickel
0.3
3.9
4.0
1.8
-
-
-
-
Selenium IV
1.0
3.6
4.0
1.2
3.8
4.4
4.8
0.4
Selenium VI
-1.4
0.6
3.0
1.2
3.1
3.8
4.6
1.0
Thallium
-0.5
1.3
3.5
1.1
3.0
4.1
4.5
1.0
Zinc
1.5
4.1
6.2
1.6
-
-
-
-
— Constituent not evaluated for this pathway.
N/A = data not available.
E.2 Water Mass Balance Model
The hydrologic module of the land application unit (LAU) model was used to estimate long-term
water balance in the field. First-order partitioning was assumed to distribute soluble constituent
mass between the overland and subsurface transport pathways. Figure E-4 depicts the conceptual
model for water flow used in this evaluation.
Z-axis ^
I
| Overland Runoff
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix E: Probabilistic Analysis
-------
The LAU model generates estimates of long-term average runoff and infiltration. Infiltration
contributes to regional aquifer flow (qAquifer) within the saturated thickness of the aquifer (B),
modeled by EPACMTP; baseflow from aquifer to the surface water body (q ef w) is derived from
NHDPlus. Runoff contributes to surface water body flow. The following sections describe the
hydrology model used to simulate the movement of water in and around farm fields and introduce
how data from NHDPlus are used in various water balance calculations.
The LAU hydrologic model (U.S. EPA, 2003c) was used to simulate watershed runoff and ground
water recharge ("infiltration"). The hydrology module is based on a daily soil moisture balance
performed within the root zone of the soil column. At the end of a given day, the soil moisture in
the root zone is the net moisture balance from the previous day with addition of water from
precipitation and residual moisture in FGD gypsum and subtraction of water losses through runoff,
infiltration and evapotranspiration.
Data on local climate and hydrogeologic environment associated with each HUC10 were used to
determine long term rates of infiltration and runoff for agricultural fields within the HUC10.
Precipitation is undifferentiated between rainfall and frozen precipitation; frozen precipitation is
treated as rainfall on an annualized basis. As described above, available daily precipitation data
from climate stations associated with a HUC10 were coordinated with average soil characteristics
for the farmland within the same region and presented to the LAU module.
Potential evapotranspiration (PET) is the demand for soil moisture from evaporation and plant
transpiration. When soil moisture is abundant, actual evapotranspiration (ET) equals PET. When
soil moisture is limiting, ET will be less than PET. The extent to which it is less under limiting
conditions has been expressed as a function of PET, available soil water, and available soil water
capacity. Water that is not lost to evapotranspiration is available to runoff or infiltrate.
Runoff is based on the Soil Conservation Service curve number procedure (USDA, 1986) and is a
function of current and antecedent precipitation, as well as land use. Land use catalogued by cover
type (e.g., woods, meadow, impervious surfaces), treatment or practice (e.g., contoured, terraced),
hydrologic condition, and hydrologic soil group.
Soil moisture in excess of the soil's field capacity, if not lost through evapotranspiration, is available
for gravity drainage from the root zone as infiltration to subroot zones (Dunne and Leopold, 1978).
The rate of infiltration is limited by the saturated hydraulic conductivity of the unsaturated soil
(Ksat). If infiltration exceeds the Ksat, a feedback loop is triggered that increases the previously
calculated runoff volume by the amount of excess soil moisture (i.e., above field capacity and Ksat).
This adjustment is made to preserve water balance and assumes that the runoff curve number
method, which is not highly sensitive to soil moisture, has admitted more water into the soil
column than can be accommodated. After the runoff is increased to account for this excess, the
ET, infiltration, and soil moisture are updated to reflect this modification and preserve the water
balance. The resulting long-term average overland runoff and ground water infiltration rates
determine the rate that mass is depleted from soils receiving FGD gypsum.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 15
Appendix E: Probabilistic Analysis
-------
E.3 Source Term Model
The distribution of constituent mass present in the applied FGD gypsum and released to overland
runoff and infiltration is calculated prior to ground water or surface water model runs. Dissolved
concentrations are dependent on the mass balance of water calculated for each release pathway, as
well as both the constituent concentrations present in and released from FGD gypsum; leaching
behavior (i.e., availability or solubility-limited); soluble fraction of constituent mass; gypsum
application rate, frequency and duration that are sampled from available distributions.
Measured leachate concentrations were adjusted based on leaching behavior prior to use in the
model. For constituents with solubility-controlled leaching behavior, measured concentrations
were used without further adjustment. For constituents with availability-limited behavior, the
leachate concentration was increased to ensure depletion of the soluble content. As discussed in
Section 5 (Screening Analysis), this was done to account for potential for the probabilistic analysis
to combine high-end values for bulk concentration, leachate concentration and soluble fraction
that could result in scenario where available content does not deplete within the year. While this
will result in higher leachate concentrations than will actually occur in the field, it will not result
in a dramatic overestimation of releases. Any concentrations higher than that needed to deplete
available content will result in faster depletion and the exact same annualized concentration. It is
possible that this adjustment could push some constituents above solubility limits; however, past
studies have found that similar adjustments provided a reasonable estimate of field leaching (U.S.
EPA, 2014a).
Applied leachate and runoff concentrations for annually applied gypsum uses are equal to the
calculated leachate concentration if available soluble mass is not depleted from soils between
applications. In these instances, leaching is assumed to persist for a time required to deplete all
applied soluble constituent mass. However, if the applied soluble mass depletes prior to the next
application, an effective dissolved concentration is determined for both pathways such that soluble
mass of one application depletes in exactly one year. The effective dissolved concentration is
applied for as many years as FGD gypsum is applied to the field. This same approach is used to
determine applied constituent concentrations in runoff water for ten-year application frequency
uses: an effective concentration is calculated if soluble mass depletes prior to the next gypsum
application; otherwise, the calculated concentration was used. The same rules for specifying how
long leaching occurs. The result in all these scenarios is a constant dissolved constituent
concentration release for as long as it takes to deplete the soluble mass.
Ten-year application frequencies were handled in a slightly different manner for the ground water
pathway. The calculated leachate concentration was always applied regardless of how much time
was required to deplete soluble constituent mass in gypsum application. If the time required to
deplete the available soluble constituent mass in a single gypsum application extends beyond the
next scheduled application, mass is released at a constant rate for as long as it takes to deplete all
applied mass for all gypsum applications. If the soluble constituent mass from a single gypsum
application depletes before the next gypsum application, the timing of soluble constituent mass
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 15
Appendix E: Probabilistic Analysis
-------
releases will respect the depletion time resulting in a repeating square wave profile where each
wave persists for a time equal to the depletion time for a single application, and the time between
square waves equals ten years minus the depletion time.
E.4 Ground Water Model
Ground water modeling was conducted with EPACMTP (U.S. EPA, 2003a). This model consists of
two steady state flow modules that address subsurface flow through unsaturated and saturated
zones. Data requirements for EPACMTP ground water flow calculations are satisfied with HUC10-
based distributions of soil megatextures corresponding to SSURGO classifications within the HUC,
hydrogeologic environment assignments to a HUC 10, and infiltration rates from the LAU module
mentioned above. Specific soil and aquifer parameters are drawn from megatexture and
hydrogeologic environment assignments from a preliminary simulation of EPACMTP using
established distributions as described in the Technical and Parameter/Data Background documents
(U.S. EPA, 2003a,b). A database of aquifer parameters is developed from the preliminary simulation
and reused for all subsequent simulations of FGD gypsum Uses and constituents.
EPACMTP consists of two coupled modules that address subsurface transport through unsaturated
and saturated zones. These modules treat soils as uniform, porous media and do not account for
preferential pathways or facilitated transport. EPA assumed that farm drinking water wells are
located upstream from surface water bodies, and so did not consider interception of ground water
by surface water bodies prior to reaching the well. This is considered a reasonable assumption
given the scattered spatial distribution of farm fields across the landscape.
In the unsaturated zone, the flow of water is driven primarily by gravity. Therefore, flow is
modeled entirely in the vertical direction (i.e., no lateral flow). This assumption can be made
because the scale of lateral migration due to dispersion will be orders of magnitude less than the
scale of vertical migration through areas receiving application of FGD gypsum (U.S. EPA, 2003a).
The solution to the governing equation for unsaturated zone flow yields estimates of vertical Darcy
velocity and average water content used to simulate contaminant transport. Darcy velocity
influences constituent advection and water content is used to determine equilibrium partitioning
of constituent mass between dissolved and sorbed phases.
In the saturated zone, flow is controlled primarily by the hydraulic conductivity of the aquifer and
the regional hydraulic gradient. Ground water flow velocities are the principal output of the
solution to the governing equation for steady state ground water flow. Contaminant transport
within the unsaturated and saturated zones requires flow velocities to advect and disperse dissolved
constituent mass in the porous media. EPACMTP assumes that movement of constituent mass is
driven primarily by the ground water advection. However, flow may be altered both by mounding
underneath the field from high volumes of leachate, which encourages spreading of the constituent
plume in all directions, and by uncontaminated recharge from precipitation that falls around the
field, which increases mixing in the vertical direction. In addition to advection, EPACMTP also
accounts for the mixing of ground water due to dispersion, which occurs to some degree in all
directions.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ 17
Appendix E: Probabilistic Analysis
-------
During transport through both the unsaturated and saturated zones, constituents may sorb to the
surrounding soils. This process is represented by soil-water partitioning coefficients (Kd values),
which is the ratio of constituent mass sorbed to the soil and dissolved in solution at equilibrium.
For most inorganics, Kd values are strongly dependent on the concentration present in leachate
and generally decrease with higher concentrations. The Kd values used in this analysis were drawn
from constituent-specific distributions of Kd values versus leachate concentrations, also known as
sorption isotherms. EPA selected nonlinear isotherms as the most representative of changes in
sorption as leachate concentrations decrease during transport through soil and ground water. This
approach is believed to be appropriate and reasonable because the vast majority of leachate
concentrations are low enough to fall in the linear region of the nonlinear isotherms. Nonlinear
isotherms were generated by Metal Speciation Equilibrium Model for Surface And Ground Water
(MINTEQA2) (U.S. EPA, 2001) for use with EPACMTP. The development of these isotherms is
described in detail in Appendix G of the EPACMTP Technical Background Document {U.S. EPA,
2003a).
Leachate concentrations applied to the ground water pathway described above were used to select
Kd values for the unsaturated zone module, while the soil pore water concentrations at the
boundary of the unsaturated and saturated zones were used to select Kd values for the saturated
zone module. Constituents with low Kd values will have low retardation factors and may move at
nearly the same velocity as the ground water. Constituents with high Kd values will have high
retardation factors and may move much more slowly than ground water. The subsurface migration
of some constituents may be very slow, and it may take a substantial amount of time for the
constituent plume to reach the downgradient receptors. As a result, the maximum concentration
may not occur until thousands of years after FGD gypsum has ceased. To prevent prohibitive model
run times, while not missing significant risks to potential receptors, EPA ran the model until either
the observed ground water concentration of a constituent at the receptor point peaked and then
fell below a model-specified minimum concentration (10 16 mg/L), or the model had been run for
a total duration of 10,000 years.
E.5 Surface Water Model
Ground water flow velocity at the ground-surface water interface is used to estimate total aquifer
flow in the vicinity of the surface water body. To predict the mass flux of a constituent from an
aquifer into a stream, the distribution of constituent concentrations and volumetric ground water
fluxes along the upgradient edge of the stream must be known. The volumetric ground water flux
depends on the difference between the stream stage and the hydraulic head in the aquifer. If the
hydraulic head in the aquifer is higher than the stream stage, the ground water from the aquifer
will enter the stream as baseflow. When this occurs, the stream is said to be a gaining stream.
Baseflow into each stream was estimated using the BFI, which captures mean baseflow activity for
a stream, thereby supporting the assumption that all streams included in this analysis are gaining
streams. Figure E-5 depicts a generalized scenario for ground water interacting with a surface water
body.
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ -|g
Appendix E: Probabilistic Analysis
-------
As shown in this figure, the water body does not always completely intercept the ground water
plume. This can occur if there is a bend in a stream or if only a portion of smaller headwater streams
intersect the plume. The centroid of the water body is located at the point (Xsw, Ysw) where Xsw
is the distance to the surface water body, and Ysw is assumed to be the same distance from the
plume centerline assigned to the receptor well. The total mass flux is determined using the
baseflow rate and the output constituent concentration from EPACMTP according to the equation
below (variable names are only defined the first time appear).
The ground water volumetric flux per unit area of stream bed is governed by the difference in
ground water hydraulic head, hydraulic conductivity of streambed, and streambed thickness.
Baseflow was estimated based on the BFI and flow of each water body segment. Stream flow
contributed by runoff to a stream segment was calculated from the simulated runoff rate used to
estimate runoff concentrations by dividing the simulated rate by the inverted BFI for that HUC.
This flow rate was calculated as a first step to relate mass loading from runoff and ground water
discharge, which is estimated based on flow rates from NHDplus.
(E-l) QLfu =
Where:
qRO-Actch'BFI Qro" ' BFI
100- BFI 100- BFI
Ac tch ~~
Area of the HUC12 contributing flow to main stem [m2]
BFI
NHDPlus mean annual base-flow index [%]
nLAU
^ F
Modeled ground water discharge (baseflow) to surface water [m3/yr]
c
1
Modeled overland runoff to surface water [m3/yr]
Qro
Modeled specific runoff rate (depth) from total drainage area [m/yr]
Beneficial Use Evaluation of FGD Gypsum in Agriculture £ -|g
Appendix E: Probabilistic Analysis
-------
Mass flux from ground water can be estimated by integrating the product of baseflow flux per unit
area and constituent concentration over the total baseflow area. The total baseflow area is
determined by the baseflow depth, Dbf, and the length of the water body intercepting the ground
water plume.
(E-2) mGW
Where:
= (q4T')(l0008,J P™ rUVsw.y.z,t)dydz
V B ) VI,000 mgj Jy=Ymin Jz= 0
mHUC
mGW
Incremental mass flux from ground water to HUC12 water body (g/yr)
B
Saturated thickness of the aquifer (m)
D F
Depth of baseflow (m)
^Aquifer ~~
Volumetric flow rate per unit width of aquifer (m2/yr)
Ym x
Leftmost intersection of plume and stream bed with respect to ground water flow
direction (m)
^Min
Rightmost intersection of plume and stream bed with respect to ground water
flow direction (m)
c
Chemical concentration at surface water boundary (mg/L)
X
Horizontal distance of stream from the downgradient edge of the field (m)
Xsw
Distance of stream from the downgradient edge of the field (m)
y
Horizontal distance from the plume center line along a vertical plane at the
upgradient side of the surface water body (m)
Ysw
Distance of stream centroid from the plume centerline (m)
z
Vertical distance from the top of a vertical plane at the upgradient side of the
surface water body, positive downward (m)
The ground water mass flux to surface water was treated as a direct load to the water body. The
corresponding contributions from runoff were calculated separately as a mass flux to obtain the
same units as contributions from ground water. The mass flux of a dissolved constituents in
overland runoff to a water body was estimated more directly as the product of the calculated
concentration in runoff from the field, the annualized runoff rate and the area of the field receiving
FGD gypsum application (relative to the total area of the drainage area).
(E-3) m^oU — CR0 " qRo ' AFGD
Where:
Afgd ~ Area of the HUC12 receiving FGD application [m2]
CR0 - Concentration in runoff water [mg/L = g/m3]
itirq0 - Mass flux from runoff water [g/yr]
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 20
Appendix E: Probabilistic Analysis
-------
The mass loading calculated for ground water and surface water (m^oU) were summed
together to obtain a total loading to the water body. To convert mass load back to a concentration,
the annual flow through the headwater stream was derived from the calculated runoff volume
from the entire drainage area scaled by the BFI for that area.
(F 41 0LAU = nLAU ( 100 ^ = nLAU ( 1QQ\
( J Total Qro \100 — BFI/ ^ f V BFI y
Where:
QtA^ - Annual average incremental flow through a stream segment [m3/yr]
This total mass load was used together with the flow in the water body segment (Qt^ ) are used
to calculate the concentration in the water column (C™tGt) and sediment (C^uc) based on the
steady-state model documented in EPAs Human Health Risk Assessment Protocol (HHRAP) for
Hazardous Waste Combustion Facilities (U.S. EPA, 2005b). These equations were used to calculate
the distribution of constituent mass between dissolved and sorbed phases within the water column.
The following are key assumptions of the model and specifics on how it was applied in this
evaluation:
¦ The model assumes steady-state flow and transport conditions. Long-term average annual
stream flow and climatic data and assumptions were used.
¦ The model accounts for constituent loadings into the water body through ground water
discharge and waste outfalls, and direct air deposition. The sources relevant to this beneficial
use evaluation are discharges from ground water and overland runoff.
¦ The model estimates the rate of incorporation, or burial, of constituents into bed sediments as
a function of the rate at which sediments deposit from the water column onto the surficial
sediment layer. The burial rate was set to zero (a protective assumption) because of the lack of
national data available to estimate this process.
¦ The model can incorporate separate decay rate constants for the water column and the benthic
sediments to allow for consideration of decay mechanisms that remove constituents from the
water body. However, because all inorganics considered are persistent in the environment,
degradation was not relevant to this risk assessment.
The following equations used the mass load and stream flow to calculate the concentrations in
each headwater stream. The resulting concentrations were used to calculate exposures for
ecological receptors living in and around the water body.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ ^
Appendix E: Probabilistic Analysis
-------
(E-7) CHUC
3rhHUC
wt t
nLAU ¦ f ¦( dz ^
T t rw ter ^dz _ dfa J
rp ri rHUC -rHUC-f ( dz ^
^wct t — Lwt t % ter ^ ^ J
fP pHUC _ pHUC . c . c . ( ^
v*^dw *-wt t 1w ter Mi ve ^ j
(u t n'v pHUC pHUC . c . ^z
Lb — Lwt t rjjenthic .
Qb
Where:
db - Depth of upper benthic layer [0.03 m]
dz - Depth of the water body [m]
hi ir
Cb - Total concentration in HUC12 stream bed sediment [g/m3 or mg/L]
hi ir
Cdw - Dissolved water body concentration in HUC12 stream [g/m3 or mg/L]
C™ct - Total water body concentration in HUC12 stream [g/m3 or mg/L]
fbenthic ~ Fraction of constituent mass sorbed to benthic sediment [unitless]
fw ter ~ Fraction of constituent mass in the total water column [unitless]
The following equations were used to calculate the fraction of constituent mass that partitions
between the water (i.e., dissolved), suspended solids and benthic sediment.
(E-ll) fd =
(E-12) fw ter =
1 + (Kd W • TSS) (woo m|) ( I oorfnig)
(^m
(SidTt)(^)+^[bsp+(Kdb 'bsc)1
^[bsp + (Kdb bsc)]
(E"13) fbenthic —
(^3^-) (^r) + [bsp + (Kdb bsc)l
Where:
bsc - Bed sediment particle concentration [1 g/cm3 or 1 kg/L]
bsp - Bed sediment porosity [0.6 cm3/cm3]
fd - Fraction of constituent mass in water column that is dissolved [unitless]
kb - benthic burial rate constant [1/yr]
Kdb - Sediment-water partition coefficient [mL/g]
Kd w - Suspended sediment-water partition coefficient [mL/g]
TSS - Total suspended solids [mg/L; Table E-5]
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 22
Appendix E: Probabilistic Analysis
-------
Of the constituents found in FGD gypsum, only mercury has the potential to volatilize to any
appreciable degree. Additional data and equations were necessary to account for this loss pathways.
When modeling mercury, the following set of equations were substituted for Equations E-7 to
account for mercury volatilization.
(E-7 a) C^ut =
rhHUC
-fwter-^+ (kwt " Aw ¦ dz)
(E-7b) kwt — fw ter ¦ f^i ye ' kv + fbenthic ' kb
Kv
(E-7c) kv =
dw
(E-7d) Kv =
1
kZ +
(Kg ' R ¦ Tw)
-l
g(Tw-293)
(E-7e) Kl =
Where:
0
Aw ~~
Dw -
H
Kg -
Kl -
Kv -
kv -
kwt ~~
R
( lm2 ) D
\1000 cm2/ v
u /3.1536 x 107 sec
yr
Temperature correction factor [unitless; 1.026]
Surface area of water body [m2; NHDPlus]
Diffusivity of mercury in water [1.77 x 10"5 cm2/sec]
Henry's Law constant for mercury [7.1 x 10"10 atnr m3/mol]
Gas phase transfer coefficient for mercury [36,525 m/yr]
Liquid phase transfer coefficient [m/yr]
Overall constituent transfer coefficient from liquid to gas phase [m/yr]
Water column volatilization rate constant [1/yr]
Total water body dissipation rate constant [1/yr]
Universal gas constant [8.205 x 10 5 atm'm3/mol'K]
Water body temperature [K; Table E-4]
Water body current velocity [m/sec; Table E-3]
Calculations for the concentrations in mainstem streams used the same set of equations described
above for headwater streams. There are two major differences between mainstem and headwater
streams. The first is that the mass flux from overland runoff and baseflow are calculated for each
entire HUC12. The second is that the total mass flux through each HUC12 outfall also includes
contributions from any upstream HUC12. The mass contribution from each HUC12 to next was
calculated by multiplying the total water concentration at the outfall (Cj^ t) by the incremental
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 23
Appendix E: Probabilistic Analysis
-------
annual average flow from NHDPlus (Q^crement ) t0 obtain a mass flux (dissolved and adsorbed)
contributed by that HUC12. At each HUC12 outfall, the mass loading from the current and all
upstream HUC12 were summed together. When summing constituent mass from upgradient
HUC12, the model runs used to characterize each HUC12 were allowed to vary, which resulted in
combinations that captured a range of application areas and rates across the landscape. The total
mass loading (riioutfaii) at each HUC10 outfall was used together with the total NHDPlus flow in
that water body segment (Qt™ ) to calculate the concentration in the water column (C™tGt) and
sediment (C^uc). This process was done at each HUC10 outfall until the stream reached either a
coastline or another stream of order 6 or higher. These resulting concentrations in each HUC10
outfall were used to calculate exposures from fish ingestion.
The use of NHDPlus flow rates to accumulate mass between HUC12 conserves concentration in
the water column, but not mass. This approach was selected for several of reasons. First, the spatial
resolution of weather data used in the LAU model is somewhat limited. Use of NHDPlus flow rates
better capture variability in the relative contributions from adjacent HUCs. Next, there are a
number of upgradient HUCs that fell outside the economic feasibility zone and so were not
modeled. Use of NHDPlus avoided the need to calculate runoff over a much wider area. Finally, in
areas where calculated runoff was zero, this would have resulted in zero flow from that segment
of the water body. Thus, use of NHDPlus flows captured any baseflow from these areas.
E.6 Soil Model
Soil concentrations are dependent on the frequency and duration of FGD gypsum applications, the
constituent concentrations present in and released from gypsum, and the fraction of constituent
mass that is soluble. This model assumes that FGD gypsum is initially applied on the soil surface
and eventually tilled into the earth. Therefore, long-term soil concentrations are calculated based
on mixing within the top 20 cm of the soil column.
Long-term soil concentrations in farm fields receiving FGD gypsum applications are determined
over up to a 100-year period assuming first-order losses of the soluble fraction of constituent mass
to the subsurface (i.e., leaching) and runoff at a rate equal to the assigned leachate and runoff
concentrations determined above. The general calculation for each year is as follows assuming the
soil is initially free of constituent mass is:
1. For each year of application, calculate constituent mass added to field. Track soluble and
insoluble fractions of applied constituent mass separately.
2. Calculate soluble losses to leaching and runoff for a year.
3. Subtract soluble losses from soluble mass fraction on soil.
4. Sum the insoluble and soluble fractions of the current year to the previous year's total soil
concentration
If the total soil concentration at the end of the current year is greater than the previous maximum,
update the maximum soil concentration to be equal to the current year.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 24
Appendix E: Probabilistic Analysis
-------
E.7 References
Allison, J. 2003. "3MRA Kds Checked and Revised." E-mail correspondence with J. Allison of
Allison Geoscience Consultants. June 13.
ORNL (Oak Ridge National Laboratory). 1984. "A Review and Analysis of Parameters for Assessing
Transport of Environmentally Released Radionuclides through Agriculture." ORNL-5786.
Prepared by C.F. Baes III, R.D. Sharp, A.L. Sjoreen and R.W. Shor Office for the EPA Office
of Air and Radiation under Interagency Agreement AD-89-F-2-A106. Oak Ridge, TN.
September.
Bailey, R.G., P.E. Avers, T. King, W.H. McNab eds. 1994. Ecoregions and subregions of the United
States (map). Washington, DC: U.S. Geological Survey. Scale 1:7,500,000. Colored.
Accompanied by a supplementary table of map unit descriptions compiled and edited by
McNab, W.H. and Bailey, R.G. Prepared for the USDA Forest Service.
Bear, E. 1972. Dynamics of Fluids in Porous Media. American Elsevier, New York.
Carsel, R.F. and R.S. Parrish. 1988. "Developing Joint Probability Distributions of Soil Water
Retention Characteristics." Water Resources Research. 24(5):755-769.
Carsel, R.F., R.S. Parrish, R.L. Jones, J.L. Hansen and R.L. Lamb. 1988. "Characterizing the
Uncertainty of Pesticide Leaching in Agricultural Soils." Journal of Contaminant Hydrology.
2:111-124.
Dunne, Thomas, and Luna B. Leopold. 1978. Water in Environmental Planning. New York: W.H.
Freeman and Company.
Hill, R.A., M.H. Weber, S.G. Leibowitz, A.R. Olsen and D.J. Thornbrugh. 2016. "The Stream-
Catchment (StreamCat) Dataset: A Database of Watershed Metrics for the Conterminous
United States." Journal of the American Water Resources Association. 52:120-128.
Hunt, C.D. 1979. "National Atlas of the United States of America—Surficial Geology." NAC-P-
0204-75M-0 [map file: geol75m], U.S. Geological Survey.
McKay, L., T. Bondelid, T. Dewald, C. Johnston, R. Moore and A. Rea. 2017. NHDPlus Version 2:
User Guide.
Newell, C.J., L.P. Hopkins, and P.B. Bedient. 1989. Hydrogeologic Database for Ground Water
Modeling. API Publication No. 4476. American Petroleum Institute, Washington, DC.
Newell, C.J., L.P. Hopkins and P.B. Bedient. 1990. "A Hydrogeologic Database for Ground-Water
Modeling." Groundwater. 28(5):703-714.
Pickard, B.R., J. Daniel, M. Mehaffey, L.E. Jackson and A. Neale. 2015. "EnviroAtlas: A New
Geospatial Tool to Foster Ecosystem Services Science and Resource Management." Ecosystem
Services. 14:45-55.
USDA (Department of Agriculture). 1986. "Urban Hydrology for Small Watersheds." Technical
Release 55. Prepared by the Natural resources Conservation Service. Washington, DC. June.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 25
Appendix E: Probabilistic Analysis
-------
USDA (Department of Agriculture). 2016. Farm Service Agency. Available online at
https://www.fsa.usda.gov/programs-and-services/aerial-photography/imagery-
pr oducts/common-land-unit-clu/index
USDA (Department of Agriculture). 2017. Soil Survey Staff, Natural Resources Conservation
Service, United States Department of Agriculture. Web Soil Survey.
U.S. EPA (Environmental Protection Agency). 1997. "EPA's Composite Model for Leachate
Migration with Transformation Products. EPACMTP: User's Guide." Prepared by
HydroGeoLogic, Inc. for the EPA Office of Solid Waste under Contract Number 68-W4-0017.
Washington, DC.
U.S. EPA. 1998. "Methodology for Assessing Health Risks Associated with Multiple Pathways of
Exposure to Combustor Emissions." EPA 600/R-98/137. Prepared by the EPA Office of
Research and Development. Cincinnati, OH. December.
U.S. EPA. 2001. Revisions in Input Data and Modeling Procedures for Using MINTEQA2 in
Estimating Metal Partition Coefficients. Prepared by Allison Geoscience Consultants, Inc. for
the Office of Solid Waste, Washington, DC.
U.S. EPA. 2003a. "EPA's Composite Model for Leachate Migration with Transformation Products
(EPACMTP). Technical Background Document." EPA53-R-03-002. Prepared by
HydroGeoLogic, Inc. and RMC, Inc. for the EPA Office of Solid Waste under Contract
Number 68-W-01-004. Washington, DC. April.
U.S. EPA. 2003b. "EPA's Composite Model for Leachate Migration with Transformation Products
(EPACMTP). Parameters/Data Background Document." EPA53-R-03-003. Prepared by
HydroGeoLogic, Inc. and RMC, Inc. for the EPA Office of Solid Waste under Contract
Number 68-W-01-004. Washington, DC. April.
U.S. EPA. 2003c. "Multimedia, Multipathway, and Multireceptor Risk Assessment (3MRA)
Modeling System. Volume I: Modeling System and Science. SAB Review Draft." EPA530-D-
03-00la. Prepared by the EPA Office of Research and Development and Office of Solid Waste.
July.
U.S. EPA. 2005a. "Partition Coefficients for Metals in Surface Water, Soil, and Waste." EPA/600/R-
05/074. Prepared by J.D. Allison and T.L. Allison of HydroGeoLogic, Inc. and Allison
Geoscience Consultants, Inc. for the EPA Office of Research and Development under Contract
Number 68-C6-0020. Washington, DC. July.
U.S. EPA. 2005b. "Human Health Risk Assessment Protocol for Hazardous Waste Combustion
Facilities." EPA/530/R-05/006. Prepared by the EPA Office of Solid Waste and Emergency
Response. Washington, DC. September.
USGS (United States Geological Survey). 1996. "Prediction of Travel Time and Longitudinal
Dispersion in Rivers and Streams." Water Resources Investigations Report 96-4013. Prepared
by H.E. Jobson of the United States Geological Survey. Reston, VA.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 25
Appendix E: Probabilistic Analysis
-------
USGS. 2013. "National Hydrography Geodatabase: The National Map." Prepared by the United
States Geological Survey. Available online at: https://viewer.nationalmap.gov/basic/. Last
Accessed on 5/3/2018.
Beneficial Use Evaluation of FGD Gypsum in Agriculture ^ 27
Appendix E: Probabilistic Analysis
------- |