wEPA
United States
Environmental Protection
Agency
Environmental Impact and Benefits
Assessment for Proposed Effluent
Guidelines and Standards for the
Construction and Development Category
November 2008
-------
U.S. Environmental Protection Agency
Off ice of Water (4303T)
1200 Pennsylvania Avenue, NW
Washington, DC 20460
EPA-821-R-08-009
-------
ACKNOWLEDGEMENTS AND DISCLAIMER
This document was prepared by Environmental Protection Agency Office of Water staff. The
following contractors provided assistance in performing the analyses supporting the conclusions
detailed in this document.
Abt Associates, Inc.
Stratus Consulting, Inc.
Office of Water staff have reviewed and approved this document for publication. Neither the
United States Government nor any of its employees, contractors, subcontractors, or their
employees makes any warranty, expressed or implied, or assumes any legal liability or
responsibility for any third party's use of or the results of such use of any information, apparatus,
product, or process discussed in this document, or represents that its use by such a party would
not infringe on privately owned rights.
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table of Contents
1 Introduction 1-1
2 Sediment 2-1
2.1 Sediment Behavior in Surface Waters 2-2
2.1.1 Important Sediment Characteristics 2-2
2.1.2 Sediment Transport 2-3
2.1.3 Turbidity 2-6
2.2 Documented Relationship between Construction and Sediment 2-7
2.2.1 Impacts of Construction Site Discharge on Sediment Levels in Surface Waters 2-7
2.2.2 Impacts of Construction Site Sediment Discharge on the Aquatic Ecosystem 2-13
2.3 Receiving Water Impacts of Sediment 2-16
2.3.1 Sediment and Turbidity Impaired Waters in the United States 2-16
2.3.2 Aquatic Life Impacts of Sediment 2-22
3 Other Pollutants and Impacts from Construction Discharges 3-1
3.1 Documented Pollutants Other than Sediment from Construction Sites 3-2
3.1.1 Nutrients 3-2
3.1.2 Fob/cyclic Aromatic Hydrocarbons 3-3
3.1.3 Metals 3-3
3.2 Receiving Water and Aquatic Life Impacts of Pollutants Other than Sediment 3-3
3.2.1 Nutrients 3-3
3.2.2 Fob/cyclic Aromatic Hydrocarbons 3-4
3.2.3 Metals 3-5
4 Overview of Benefits from Regulation 4-1
4.1 Conceptual Framework for Valuation of Environmental Services 4-1
4.2 Applicable Benefit Categories 4-3
4.2.1 Market Benefits 4-3
4.2.2 Nonmarket Benefits 4-6
4.3 Summary of Effects 4-9
5 Pollutant Load Modeling 5-1
6 Water Quality Modeling 6-1
6.1 SPARROW Model Documentation 6-1
6.1.1 General Overview 6-1
6.1.2 Modeling Concept 6-1
6.1.3 Model Infrastructure 6-2
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.1.4 Model Specification 6-4
6.1.5 Model Estimation 6-6
6.2 SPARROW Sediment Model 6-6
6.2.1 Sediment Model Data Sources 6-6
6.2.2 Preliminary Model Estimation Results 6-8
6.2.3 Model Simulation 6-9
6.3 Water Quality Modeling Results 6-10
6.3.1 Estimated Changes in TSS Concentration in Watersheds with Most Developed Acres
(1992-2001) 6-13
6.3.2 Estimated Changes in TSS Concentrations for all RF1 Watersheds Receiving Sediment
Loading from Construction Sites 6-17
6.3.3 Impacts on Reservoir Sedimentation 6-24
6.3.4 Hypothetical "No Discharge" Scenario 6-24
6.4 Extension of SPARROW to Estuaries and Coastal Waters 6-25
6.5 Use of APEX-SWAT to Evaluate Construction Sediment Regulatory Options: Case Study, Dallas,
TX 6-33
6.5.1 Overview 6-33
6.5.2 Objectives 6-34
6.5.3 APEX-SWAT 6-34
6.5.4 Dallas-Fort Worth Study Setting 6-37
6.5.5 Sources of Data 6-38
6.5.6 Experimental Design 6-39
7 Benefits to Navigation 7-1
7.1 Data Sources 7-2
7.2 Identifying Waterways That Are Dredged and the Frequency of Dredging 7-3
7.2.1 Determining Dredging Job Locations 7-3
7.2.2 Identifying the Frequency of Dredging 7-4
7.3 Estimating the Navigational Maintenance Cost per Cubic Yard of Sediment Removed 7-5
7.4 Estimating the Total Cost of Navigable Waterway Maintenance Under the Baseline and Post-
Compliance Scenarios 7-6
7.4.1 Estimating the Reduction in Sediment Dredging in Navigable Waterways Due to the
Reduction in Discharge from Construction Sites 7-6
7.4.2 Estimating the Total Cost of Navigable Waterway Maintenance Under the Baseline and
Post-Compliance Scenarios 7-6
7.4.3 Sensitivity Analysis 7-8
7.4.4 Annualizing Future Dredging Costs 7-9
7.5 Estimating the Cost Savings from Decreased Dredging ofNavigable Waterways 7-11
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
7.6 Sources of Uncertainty and Limitations 7-14
8 Benefits to Water Storage 8-1
8.1 Review of Literature on Reservoir Sedimentation 8-1
8.2 Data Sources 8-3
8.3 Estimating the Unit Cost of Sediment Removal from Reservoirs 8-4
8.4 Estimating the Total Cost of Reservoir Dredging Under Different Policy Options 8-4
8.4.1 Estimating Sediment Accumulation in Reservoirs and the Amount of Sediment Expected
To Be Dredged 8-4
8.4.2 Estimating the Total Cost of Reservoir Dredging Under the Baseline Scenario and Policy
Options 8-5
8.5 Estimating the Cost Savings from Reduced Reservoir Dredging 8-6
8.6 Sources of Uncertainty and Limitations 8-8
9 Benefits to Drinking Water Treatment 9-1
9.1 Sediment Effects on Drinking Water Treatment Costs 9-1
9.2 Avoided Costs of Drinking Water Treatment from Reducing Sediment Discharge from
Construction Sites 9-1
9.3 Data Sources 9-4
9.4 Modeling Sediment Concentrations and Reductions 9-4
9.5 Identifying Reaches with Drinking Water Intakes and Their Intake Volumes 9-5
9.6 Estimating the Cost of Chemical Turbidity Treatment 9-6
9.7 Estimating the Cost of Sludge Disposal 9-7
9.7.1 Estimating the Amount of Sludge Generated 9-8
9.7.2 Calculating the Probability That the Sludge Will Require Off-Site Disposal 9-9
9.7.3 Estimating the Distance to the Disposal Location 9-9
9.7.4 Calculating the Cost of Sludge Disposal 9-9
9.8 Sensitivity Analysis 9-10
9.9 Estimating the Total Costs of Drinking Water Treatment Under the Baseline and Post-Compliance
Scenarios 9-10
9.10 Estimating Cost Savings from Lower Sediment and Turbidity in Drinking Water Influent 9-11
9.11 Sources of Uncertainty/Limitations 9-14
10 Benefits from Water Quality Improvements 10-1
10.1 Water Quality Index 10-1
10.1.1 WQI Methodology for Rivers and Streams 10-1
10.1.2 WQI Methodology for Estuaries 10-3
10.1.3 Relation between WQI and Suitability for Human Uses 10-5
November 2008 iii
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
10.1.4 Sources of Data on Ambient Water Quality 10-5
10.1.5 Estimated Changes in Water Quality (AWQI) from the Regulation 10-7
10.2 Total Benefits of Water Quality Improvements 10-12
10.3 Estimating Total WTP for Water Quality Improvements 10-18
10.4 Uncertainty and Limitations 10-19
11 Total Estimated Benefits 11-1
11.1 Summary of the Estimated Benefits 11-1
11.2 Sources of Uncertainty and Limitations 11-5
11.2.1 Water Quality Model Limitations 11-5
11.2.2 Focus on One Pollutant of Concern (Sediment) 11-6
11.2.3 Omission of Several Benefit Categories from the Analysis of Monetized Benefits 11-7
12 References 12-1
Appendix A - Threatened and Endangered Aquatic Species A-l
Appendix B - SPARROW Model Documentation B-l
B.I General Overview B-l
B.2 Modeling Concept B-2
B.2.1 Objectives of SPARROW Modeling B-2
B.2.2 SPARROW Mass Balance Approach B-4
B.2.3 Time and Space Scales of the Model B-4
B.2.4 Accuracy and Complexity of SPARROW Models B-6
B.2.5 Comparison of SPARROW with Other Watershed Models B-7
B.2.6 Model Infrastructure B-9
B.2.7 Use of Monitoring Data B-10
B.2.8 Model Specification for Monitoring Station Flux Estimation B-ll
B.2.9 Tools for Flux Estimation B-ll
B.2.10 Guidance for Specifying Monitoring Station Flux Models B-12
B.2.11 Stream Network Topology B-13
B.2.12 Watershed Sources and Explanatory Variables B-15
B.3 Model Specification B-16
B.3.1 Model Equation and Specification of Terms B-17
B.3.2 Contaminant Sources B-19
B.3.3 Landscape Variables B-20
B.3.4 Stream Transport B-20
B.3.5 Reservoir and Lake Transport B-21
B.4 Model Estimation B-22
B.4.1 Evaluation of Model Parameters B-22
November 2008 iv
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
B.5 Model Prediction B-23
Appendix C - A Preliminary SPARROW Model of Suspended Sediment for the Conterminous
United States (Schwarz 2008a) C-l
Appendix D - Water Quality Index Tables D-l
Appendix E - Meta-Analysis Results E-l
E.I Literature Review of Water Resource Valuation Studies E-l
E.I.I Identifying Water Resource Valuation Studies E-2
E.I.2 Description of Studies Selected for Total WTP Meta-Analysis E-4
E.2 Total WTP Meta-Analysis Regression Model and Results E-l
E.2.1 Metadata Total WTP Regression Model E-8
E.2.2 Total WTP Regression Model and Results E-12
E.2.3 Interpretation of Total WTP Regression Analysis Results E-14
E.2.4 Model Selection E-18
E.3 Model Limitations E-19
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
List of Tables
Table 2-1: Sediment Grade Scale 2-2
Table 2-2: Studies Quantifying Sediment Increases in Surface Waters 2-7
Table 2-3: Summary of Sediment Concentration Findings from Selected Studies 2-9
Table 2-4: Summary of Sediment Yield Findings from Selected Studies 2-9
Table 2-5: Summary of Turbidity Findings from Selected Studies 2-10
Table 2-6: Studies Quantifying Sediment Increases and Their Impacts on Aquatic Life 2-13
Table 2-7: Sediment Criteria for Surface Water Quality, by State 2-16
Table 2-8: "Sediment/Siltation" Impairment in 305(b)-Assessed Waters : 2-19
Table 2-9: "Turbidity" and "Suspended Solids" Impairment in 305(b)-assessed Waters1 2-19
Table 2-10: SPARROW Distribution of TSS Concentrations1'2 2-21
Table 3-1: Nutrients in Studies of Construction Site Discharges 3-2
Table 3-2: Nutrient Impairment in 305(b)-Assessed Waters1 3-4
Table 3-3: Metal Impairment in 305(b)-Assessed Waters1 3-5
Table 4-1: Summary of Benefits from Reducing Sediment Discharges from Construction Sites 4-10
Table 5-1: Baseline Construction Sediment Loading Summary and Post-Compliance Reductions 5-2
Table 5-2: Baseline Construction Sediment Loading Summary and Option 1 Reductions by EPA Region
5-2
Table 5-3: Baseline Construction Sediment Loading Summary and Option 2 Reductions by EPA Region
5-3
Table 5-4: Baseline Construction Sediment Loading Summary and Option 3 Reductions by EPA Region
5-3
Table 6-1: Estimation Results for the SPARROW Suspended Sediment Model 6-9
Table 6-2: Top Percent of Reaches by Increase in Developed Land Area (1992-2001) 6-14
Table 6-3: Distribution of Top RF1 Watersheds by Increase in Developed Land Area (1992-2001) 6-14
Table 6-4: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 1% of RF1
Watersheds by Increase in Developed Land Area (1992-2001) 6-15
Table 6-5: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 10% of RF1
Watersheds by Increase in Developed Land Area (1992-2001) 6-15
Table 6-6: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 25% of RF1
Watersheds by Increase in Developed Land Area (1992-2001) 6-16
Table 6-7: TSS Concentration Distribution Based on SPARROW Output for 43,821 RF1 Reaches
Receiving Construction Sediment Discharges 6-17
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-8: Summary of Baseline TSS Concentration in RF1 Reaches Receiving Construction Sediment
Discharges, by EPA Region 6-18
Table 6-9: Summary of Option 1 TSS Concentration in RF1 Reaches Receiving Construction Sediment
Discharge, by EPA Region 6-18
Table 6-10: Summary of Option 2 TSS Concentration in RF1 Reaches Receiving Construction Sediment
Discharge, by EPA Region 6-19
Table 6-11: Summary of Option 3 TSS Concentration in RF1 Reaches Receiving Construction Sediment
Discharge, by EPA Region 6-19
Table 6-12: Summary of Estimated Changes in TSS Concentration by Policy Option 6-21
Table 6-13: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 1.6-21
Table 6-14: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 2.6-22
Table 6-15: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 3.6-23
Table 6-16: Sediment Accumulation in Reservoirs by Policy Option and EPA Region1 6-24
Table 6-17: Total Yearly Sediment from All Sources for Baseline and Zero Discharge Scenarios Based on
SPARROW Output for 62,370 RF1 Reaches 6-25
Table 6-18: TSS Concentrations (mg/L) Estimated by DCP, Southeastern Estuaries 6-28
Table 6-19: TSS Concentrations (mg/L) Estimated by DCP, Gulf of Mexico Estuaries 6-29
Table 6-20: TSS Concentrations (mg/L) Estimated by DCP, Northeastern Estuaries 6-30
Table 6-21: TSS Concentrations (mg/L) Estimated by DCP, West Coast Estuaries 6-31
Table 6-22: Comparison between SPARROW-DCP Estimated and Observed Suspended-Sediment
Concentrations (mg/L) 6-32
Table 7-1: Dredging in U.S. Navigable Waterways, 1995-2006 7-3
Table 7-2: Dredging of Navigable Waterways Performed or Contracted by USAGE, 1995-2006 7-3
Table 7-3: Dredging Jobs and Recurrence Intervals, 1995- 2006 7-5
Table 7-4: Costs of Recurring Dredging, 1995-2006 7-6
Table 7-5: Annualized Dredging Costs Under the Baseline Scenario (millions of 2008$) 7-10
Table 7-6: Annualized Reductions in Dredging and Costs Under Option 1 7-12
Table 7-7: Annualized Reductions in Dredging and Costs Under Option 2 7-12
Table 7-8: Annualized Reductions in Dredging and Costs Under Option 3 7-13
Table 8-1: Average Annual Sedimentation Deposition Recorded by RESIS 8-3
Table 8-2: Reservoir Dredging Under the Baseline Scenario 8-6
Table 8-3: Reduction in Reservoir Dredging and Costs Under Option 1 8-7
Table 8-4: Reduction in Reservoir Dredging and Costs Under Option 2 8-8
Table 8-5: Reduction in Reservoir Dredging and Costs Under Option 3 8-8
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 9-1: Public Water Intakes and Estimated Average Daily Flow by EPA Region 9-6
Table 9-2: Drinking Water Turbidity Treatment Costs Under the Baseline Scenario 9-11
Table 9-3: Reduction in Drinking Water Treatment Costs Under Option 1 9-13
Table 9-4: Reduction in Drinking Water Treatment Costs Under Option 2 9-13
Table 9-5: Reduction in Drinking Water Treatment Costs Under Option 3 9-14
Table 10-1: Original and Revised Weights for Freshwater WQI Parameters 10-2
Table 10-2: Freshwater Water Quality Subindex 10-3
Table 10-3: Weights for Estuarine WQI Parameters 10-4
Table 10-4: Estuarine Water Quality Subindex 10-4
Table 10-5: Water Quality Classifications 10-5
Table 10-6: Percentage of River Miles in Conterminous 48 States by WQI Classification for EPA
Regions: Baseline Scenario 10-7
Table 10-7: Estimated Water Quality Improvements Under Option 1 10-9
Table 10-8: Estimated Water Quality Improvements Under Option 2 10-10
Table 10-9: Estimated Water Quality Improvements Under Option 3 10-11
Table 10-10: Independent Variable Assignments 10-14
Table 10-11: Average Household Willingness to Pay for Water Quality Improvement by Region (2008$)
10-17
Table 10-12 :Regional Willingness to Pay for Water Quality Improvement (Millions 2008$) 10-18
Table 11-1: Annual Total National Benefits by Benefits Category (millions of 2008$) 11-2
Table 11-2: Annual Total National Benefits Under Option 1 (thousands of 2008$) 11-4
Table 11-3: Annual Total National Benefits Under Option 2 (thousands of 2008$) 11-4
Table 11-4: Annual Total National Benefits Under Option 3 (thousands of 2008$) 11-5
Table A-l: List of Federal Threatened and Endangered Aquatic Species Potentially Impacted by Sediment
A-l
Table D-l: Estimated Water Quality Improvements Under Option 1 D-2
Table D-2: Estimated Water Quality Improvements Under Option 2 D-5
Table D-3: Estimated Water Quality Improvements Under Option 3 D-8
Table E-l: Selected Summary Information for Studies E-5
Table E-2: Variables and Descriptive Statistics for the Total WTP Regression Model E-ll
Table E-3: Estimated Multilevel Model Results for the Trans-log and Semi-log Total WTP Regression
Models: WTP for Aquatic Habitat Improvements E-14
Table E-4: Comparison of WTP for Different Changes in WQI Based on Semi-log and Trans-log Models
E-19
November 2008
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Introduction
The nature of construction activities is to change, often significantly, many elements of the natural
environment. Construction activities typically include clearing the land of vegetation, excavating earth,
and compacting soil, all of which lead to increased stormwater runoff and higher erosion rates.
Construction activities currently affect approximately 590,000 acres in the conterminous United States
each year (USEPA, 2008c).
Construction site discharges have been documented to increase the loadings of pollutants to surface
waters. The most prominent and widespread pollutant is sediment. The level of sediment is often
identified through the measurement of other pollutants in the water body, most notably turbidity,
suspended solids, total suspended solids (TSS), suspended sediment concentration (SSC), or
settleable solids. Other documented pollutants include metals, nutrients, and polycyclic aromatic
hydrocarbons (PAHs). These pollutants can derive from construction equipment and materials or from
contamination of a site prior to the start of construction activity. Other possible pollutants include
pesticides, other toxic organics, and materials that exert biological oxygen demand (BOD) in surface
waters. Construction activities mobilize these pollutants when they disturb soil and increase stormwater
runoff within and from a site, making the pollutants available for discharge to surface waters. Pollutants
can also enter surface waters during excavation dewatering, construction equipment washout, equipment
operation in or near surface waters, and dredging and filling in or near surface waters.
Increases in sediment and other pollutant discharges from construction sites and consequential increases
in water body pollutant levels can have adverse impacts on aquatic ecosystems and on human uses of
aquatic resources. Ecological impacts from construction pollutant discharges range from temporary to
permanent and include both physical impacts on water bodies and biological impacts on aquatic
ecosystems. Under a worst-case scenario, a water body may become uninhabitable to some or all varieties
of plant and animal life due to high levels of sediment or other pollutants in the water body. Increased
sediment and turbidity levels from construction discharges can also impact human uses of water
resources, including drinking water supply, recreation, navigation, and fishing, as well as impair proper
functioning of stormwater management systems.
EPA reviewed literature pertaining to various aspects of construction site discharges, including their
magnitude, impacts on receiving waters, effects on aquatic life, and impacts on human uses of water
resources. Information on sediment discharges and their ecological impacts is summarized in Chapter 2.
Chapter 3 presents information available on discharges of pollutants other than sediment and on their
effects on receiving water bodies.
To quantify national water quality impacts from construction site discharges, EPA used the Spatially
Referenced Regressions on Watershed Attributes (SPARROW) modeling methodology. EPA used results
from the model to better understand both current conditions and the nature of water quality changes that
could result from regulatory actions EPA is proposing to reduce construction site sediment discharges.
Chapter 5 summarizes the regulatory options EPA considered and their associated sediment loading
reductions. Chapter 6, Appendix B, and Appendix C discuss the SPARROW methodology and the results
of EPA's analysis. Because of data limitations, EPA was not able to conduct a similar analysis for
construction site pollutants other than sediment.
November 2008 1-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
The Agency expects that reduced discharges of sediment and other pollutants from construction sites will
enhance environmental services provided by the affected water bodies and, as a result, human welfare.
Human activities and uses expected to benefit from reduced sediment discharge from construction sites
include recreation, commercial fishing, public and private property ownership, navigation, and water
supply and use. In analyzing benefits of the regulation, EPA quantified economic benefits of reduced
dredging of navigable waterways, reduced dredging of water storage facilities (reservoirs), and reduced
drinking water treatment costs. Chapters 7-9 provide details of the methods used in analyzing these
benefit categories and the results of EPA's analysis. Other benefit categories, including reduced flood
risk, increases in property values, and decreased wear on industrial water-cooling equipment and
commercial fishing, are discussed qualitatively in Chapter 4. EPA estimates that Options 1, 2, and 3 will
reduce expenditures on navigable waterway dredging by approximately $1, $13, and $27 million per year
respectively. Expenditures on reservoir dredging are expected to decrease by $600,000 per year under
Option 1, by $17.6 million per year under Option 2, and by $30.6 million per year under Option 3.
Reductions in drinking water treatment costs are estimated to amount to $200,000, $7.4 million, and
$13.1 million annually under Options 1, 2, and 3, respectively. Overall, EPA expects this regulation to
save governments and private entities between $1.8 and $70.9 million each year, depending on the policy
option.
EPA also expects that reductions in sediment discharges to surface waters resulting from the regulation
will enhance or protect aquatic ecosystems currently under stress. The drop in sediment loading is
expected to improve the protection of resident species; enhance the general health offish, invertebrate,
and other aquatic organism populations; increase their propagation to waters currently impaired; and
expand fisheries for both commercial and recreational purposes. Improvements in water quality such as
decreased turbidity will also favor increased recreational activities such as swimming, boating, and
outings. Finally, the Agency expects that the regulation will augment nonuse values (e.g., option,
existence, and bequest values) of the affected water resources. EPA used a meta-analysis of surface water
valuation studies to estimate the total value of nonmarket benefits (including use and nonuse values)
stemming from the regulation. The Agency estimates that mean total willingness to pay (WTP) for water
quality improvements resulting from the regulation ranges from $2 million to $500 million, depending on
the regulatory option under consideration. Chapter 10, Appendix D, and Appendix E of this report provide
details on EPA's analysis of nonmarket benefits. EPA expects that nonmarket benefits resulting from
Option 1 will be approximately $16.6 million per year. Options 2 and 3 have more substantial nonmarket
benefits, estimated at $295 million and $400 million per year, respectively. Combining these nonmarket
benefits estimates with the avoided cost estimate produces total benefits estimates of $18.4 million per
year for Option 1, of $333.3 million per year for Option 2, and of $470.6 million per year for Option 3.
EPA was not able to monetize benefits from reducing discharges of other pollutants that can be present in
discharges from construction sites (e.g., nitrogen, phosphorus, PAHs, and metals) due to a lack of
quantitative data on the frequency and magnitude of these discharges. Chapter 3 of this report
summarizes the ecological effects associated with discharges of pollutants other than sediment from
construction activity. Reducing discharges of these pollutants is expected to generate a wide range of
benefits. For example, reducing discharges of nutrients would reduce the rate of eutrophication in
receiving waters, leading to fewer beach closings, greater productivity of fisheries in these waters, and
greater enjoyment of these water resources. Reducing discharges of toxic pollutants such as PAHs and
metals from construction sites would reduce the level of these pollutants in the affected water bodies,
leading to healthier aquatic communities and lower pollutant concentrations in water body sediment.
Because sediment dredged from water body beds or sludge waste from drinking water treatment is more
November 2008 1-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
costly to dispose of when it contains elevated levels of toxic pollutants, lower levels of toxic pollutants
are expected to lower the costs of sludge and dredged materials disposal. As these pollutants also factor
into the determination of the water quality of a surface water body, reducing ambient concentrations of
these pollutants is likely to increase the value of affected water resources.
November 2008 1-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
2 Sediment
Sediment is the most commonly documented pollutant in construction site discharges. Dramatic increases
in sediment loadings downstream of construction sites have been documented by dozens of studies over
the past four decades. An early and frequently cited study by Wolman and Schick (1967) performed in the
Baltimore, Maryland area found that suspended sediment concentrations in streams receiving construction
site discharges were anywhere between 2 and 200 times greater than those in streams in rural or wooded
areas.
Sediment from construction sites can consist of both inorganic and organic materials. Sediment influences
several measures of water quality and pollutant levels:
> Total suspended solids (TSS) - a measure of the amount of suspended inorganic and organic
material in the water column, usually expressed in milligrams per liter.
> Turbidity - a measure of the ability of light to penetrate water. Turbidity is typically measured
with a nephelometer (a type of turbidimeter) and expressed in nephelometric turbidity units
(NTUs). Turbidity levels are related to the quantity, size, shape, and color of particles suspended
in the water column (many small particles reduce light penetration more than a few large particles
of the same mass). Higher NTU levels mean that less light is able to penetrate water.
> Total dissolved solids (TDS) - a measure of the dissolved particles in water, usually expressed in
milligrams per liter. It is often used as an indicator of drinking water quality.
Sediment can also settle out of the water column, creating a layer of sediment on a water body's bed. This
process is commonly called siltation or sedimentation, and the sediment can be referred to as settleable
solids.
Episodic precipitation events are the primary cause of construction site sediment discharges to surface
waters. Most sediment discharge, therefore, takes place during or shortly after precipitation events. Once
a precipitation event ceases, discharges from construction sites generally cease within a relatively short
time period. Individual construction sites are transient as well, with activity typically lasting several
months to several years. Sediment discharges due to active construction at an individual site are therefore
typically transient, too. However, impacts from a construction site's sediment discharges may persist
beyond the lifespan of a precipitation event or of a site's active construction. Sediment discharges
typically contain large fractions of persistent matter that can remain in surface waters for long periods of
time. During this time, sediments can shift multiple times between suspension in the water column and
deposition on a water body bed, depending on currents and other disturbance factors. As time passes, the
sediments can migrate downstream and impact new waterbodies far from their original point of entry into
the surface water network.
Because of sediment's ability to migrate downstream, a waterbody downstream of a watershed containing
a consistent level of construction activity from year to year can experience a relatively consistent level of
impact from construction sediment discharges, even though the timing of individual precipitation events
and locations of individual construction sites within the watershed may change from year to year.
Some water bodies contain aquatic ecosystems that are adapted to sediment influx, whereas others contain
ecosystems adapted to low levels of sediment input. The appropriate level of sediment varies from
November 2008 2-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
waterbody to waterbody. However, even ecosystems adapted to sediment input may become impaired if
sediment input is excessive or if the ecosystem's resiliency is taxed by other stressors.
High levels of sediment can have profound effects on aquatic ecosystems, impairing photosynthetic
activity, reducing available food, and burying habitat, thereby causing fish and other organisms to
relocate or die. The sections below provide additional detail on construction site sediment discharges and
the consequences of increased sediment loads to receiving waters and their ecosystems. For a detailed
discussion of the process of sediment discharge from construction sites, see EPA's Development
Document for Proposed Effluent Guidelines and Standards for the Construction and Development
Category (USEPA 2008b). Elevated sediment loadings can create many adverse economic impacts,
including decreased reservoir capacity, increased navigable channel dredging, increased drinking water
treatment costs, and decreased water body recreational value. These and other economic impacts are
discussed in Chapter 4 and Chapters 7-10.
2.1 Sediment Behavior in Surface Waters
Sediments, once eroded from land surfaces and delivered to surface water bodies, are transported and
ultimately deposited within the riverine flow system (including banks, floodplains, or terraces); within
lakes, reservoirs, or other impoundments; or in estuaries, bays, or elsewhere on the continental margins.
Factors influencing sediment impact duration are the intensity, quantity and composition of the sediment
discharges and the nature of the receiving water. Some surface waters are able to purge or absorb
sediment inputs within relatively short time frames. Other surface waters allow sediment accumulations to
persist for long periods of time.
2.1.1 Important Sediment Characteristics
Important characteristics of sediments that influence transport and fate include the size, shape, density,
fall velocity, and concentration of sediment particles (Shen and Julien 1993). Commonly used definitions
of particle size include nominal diameter, or diameter of a spherical particle of equivalent volume, and
sieve diameter, which is the smallest square sieve opening through which the particle will pass (units of
mm or jam). Table 2-1 presents the broad categories of the sediment size scale.
Table 2-1: Sediment Grade Scale
Category
Size range
(mm)
Boulders,
Cobbles
64 - 4,000
Gravel
2-64
Sand
0.062-2
Silt
0.004-0.062
Clay
0.00024 - 0.004
Source: Julien (1995)
Particle size is important in the context of sediment transport because it helps determine how quickly a
particle will settle out of suspension in the water column onto the bed of a water body. The fall or settling
velocity of sediment particles can be estimated using a form of Stokes' law (Julien 1995; Chapra 1997):
•ds2 (Eq.2-1)
Where:
November 2008
2-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
co = settling velocity (cm s"1)
a = dimensionless form factor accounting for geometry (ot=l for sphere)
g = gravitational acceleration (g=9 81 cm s"2)
ps = density of sediment particle (g cm"3)
pw = density of water (g cm"3)
Um = dynamic viscosity (g cm"1 s"1)
ds = particle diameter (cm)
This formula indicates settling velocity increases as particle size increases, assuming all other factors are
equal. Stokes-type settling is most directly applicable to the behavior of sediment in detention basins and
some reservoirs and lakes, in which flow velocities are low and turbulent energy in the water column is
minimal. In streams and rivers, settling velocity operates in conjunction with turbulent energy affecting
particle suspension to determine sediment settling dynamics. Since turbulent energy is required to offset
the force of gravity (settling) in order to keep a particle in suspension, a greater amount of turbulent
energy is required to suspend larger, heavier particles than for smaller particles.
The particle size distribution describes the percentage by weight of sediment materials in each size
category for a given sediment sample. The sediment size d50, or median grain size, is the particle size for
which 50 percent of particles by weight are larger and 50 percent are smaller, and is often used to
summarize a sediment sample. Similarly, d90 indicates a relatively coarse fraction, for which 90 percent of
particles are smaller.
2.1.2 Sediment Transport
Sediment in Streams and Rivers
The transport and fate of sediments within river and stream channels are controlled by two general sets of
influences. These are the factors determining the stream transport capacity and the rate of supply of the
sediments themselves, along with their physical and chemical characteristics.
Sediments conveyed by stream channels, known as total sediment load, are typically partitioned into bed
load and suspended load, although the boundary separating these regimes is not rigid, but rather
determined by flow conditions. Sediment particle size is the primary distinguishing factor, with larger
particles moving as bed load and finer particles as suspended load.
Bed load consists of the movement of particles along the streambed, by rolling, sliding, and hopping.
Hopping, or saltation, involves the brief suspension of sediment particles by turbulent flow. Particles
moving as bed load typically consist of sand and coarser size fractions, and the bed layer thickness is
approximately twice the bed load particle diameter. Bed load is, in most settings, intermittent, and bed
load particles may not move for long periods during low flows or between flood events.
Suspended sediment load is "the rate of movement of sediment particles that are supported by the
turbulent motion of the stream flow" (Shen and Julien 1993) and consists primarily of the clay and silt
fractions of the particle size distribution. It is commonly measured as Suspended Sediment Concentration
(SSC) or, alternatively, as Total Suspended Sediment (TSS). Although SSC and TSS are interpreted in a
November 2008 2-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
similar manner, as measures of the concentration of suspended solid-phase material in surface water
bodies, Gray et al. (2000) has determined that differences in laboratory protocols result in systematic
differences between SSC and TSS when measuring a given ambient concentration of suspended sediment.
SSC measurements are systematically higher than TSS, particularly when the sand component of total
sediments exceeds roughly 25 percent of total (dry) sediment mass. The total sediment load consists of
the sum of bed load and suspended load, respectively. The total sediment flux (total mass of sediment
passing through a given stream cross-section per unit of time) is typically difficult to measure directly. To
obtain estimates of total sediment flux, bed load and suspended load are sampled, and the total flux is
estimated by multiplying each sampled concentration by the respective volume of flow.
Several hydraulic and geomorphologic factors determine the capacity of a given stream reach to mobilize
and to transport sediment, both as bed load and suspended load. A partial list of these factors includes
channel width, flow depth and cross-sectional geometry, bed slope and roughness, and discharge velocity
and volume. In general, the more turbulent energy available for suspension and mobilization of sediment,
the greater the sediment transport capacity per unit of stream width, and the larger the sediment particles
that can be moved. Several empirical formulas have been developed to estimate sediment transport
capacity on the basis of flow- and channel- related variables. One example, developed by Simons et al.
(1981) takes the form of a power law:
qs=Cl-h^-V^ (Eq.2-2)
Where:
qs = sediment transport per unit stream width (ft2/s)
h = depth of flow in ft
V = flow velocity in ft/s
ci, c2, c3 = empirical coefficients calibrated to reflect channel slope (gradient) and median
particle size diameter.
The unit transport capacity is strongly related to flow velocity (V), with values of c3 ranging from 3.3 to
3.95. Many, if not most, of the hydraulic factors controlling sediment transport capacity, including
channel geometry, depth and velocity of flow, vary extensively in time and space within a given river
system.
Most North American rivers and streams possess a strong seasonal discharge cycle, with spring discharge
volumes typically many times larger than late summer - autumn flows. Intense or prolonged rainfall
events can also generate flood pulses of hourly to daily duration, which often have significant turbulent
energy. In addition, as Eq. 1-2 indicates, sediment transport capacity is a non-linear function of flow-
related variables. The movement of sediments, both as bed load and as suspended load, is thus highly
non-uniform in time for most river systems. The majority of annual sediment flux, and particularly the
movement of coarse or highly cohesive sediment particles may occur over a relatively short period of
time within a single flood event. Between such events, sediments are typically stored within the channel.
In developing annual or long-term sediment budgets for specific river reaches, it is therefore necessary to
utilize sediment concentration samples and associated flow volumes representative of the entire range of
annual flow conditions. Long-term annual budgets are derived by estimating the sediment mass fluxes for
November 2008 2-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
each rate of flow, and weighting these fluxes by the fraction of the year for which the flow conditions
typically occur.
Erosion and deposition of sediments within a river course also exhibit spatial patterns strongly related to
stream morphology. River reaches having smaller cross-sectional flow area, steeper slopes and higher
flow velocities, typical of upper catchments, discourage the deposition of sediments. By contrast, wider
channels with lower bed slopes and associated velocities, typical of downstream reaches, act as regions of
relative sediment deposition. Sediments are deposited preferentially in regions of low-energy flow,
including pools and the inside of bends, where bars form (Chapra 1977). Such deposition zones are not
permanent features of the watercourse, since these deposits can be scoured and moved downstream during
major flow events.
Hydraulic and geomorphologic variables provide one set of controls on the sediment transport capacity of
a given river reach. Actual sediment transport can also be regulated by the rate of supply of sediments
(Figure 2-1, Julien 1995). Within a particular reach, observed sediment fluxes can originate either from
land surface (erosion) processes adjacent to the reach, from streambank erosion, from upstream reaches,
or from the re-mobilization of sediments previously deposited within the reach.
Figure 2-1: Sediment Transport Capacity and Supply Curves
I
£
1
w
,t
Of
e»
•e
Sedlimnt Transport Capacity
Capacity-Limited
Washload Bed Material Load
Grain Size (d,)
Source: Julien (1995)
One consequence of the competing controls on sediment transport is that increases/reductions in land
surface erosion and sediment delivery rates may not result in directly proportional increases/decreases in
observed sediment loads in downstream reaches. For example, a reduction in sediments delivered to a
particular reach resulting from improved land management may result in decreased total sediment loads
within that reach, resulting in turn in less stream energy expended on turbulent suspension of sediment.
This energy is then available for the re-suspension of sediments previously deposited within the reach,
which will be reflected in sediment concentrations at points downstream. In situations where sediment
supplies are strongly limited, such as in reaches below dams or other sediment-trapping impoundments,
flows may possess considerable excess transport capacity. Such sediment-starved flows may entrain
gravel or other large particles, and result in bank erosion and bed incision (Kondolf, 1995).
November 2008
2-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
The episodic movement of sediments within river systems at seasonal, annual, decadal and even longer
timesteps must be taken into account in efforts to link land use and management policies with in-stream
water quality indices. The geomorphologic perspective on erosion and sedimentation emphasizes an
understanding of the long-term frequency-magnitude relationships of the flow events that are largely
responsible for re-distributing sediments within river systems, thereby resulting in observed hydraulic and
topographic factors. Effective discharge (Wolman and Miller, 1960) is defined as the rate of discharge
associated with the greatest contribution to sediment transport over time (alternatively, the flow
instrumental in determining channel morphology). This is often found to be the maximum product of
discharge magnitude and event frequency. For temperate regions within the U.S., effective discharge is
often found to correspond to flow events having a recurrence interval of roughly 1.5 years (Dunne and
Leopold, 1978).
Sediment in Lakes and Reservoirs
Lakes and reservoirs have lower flow velocities and greater depth of flow than streams and rivers and
often act as deposition locations (sinks) for sediments. The lake and reservoir flow environment more
closely approximates the still-water conditions under which Stokes' law (Section 2.1.1) applies to
sediment particles to predict settling velocity. The assumption of Stokes-type settling is often used to
predict the rate of sedimentation in lakes and reservoirs by comparing the time required for a particle to
settle with the particle's transit time through the water body. Several factors influence settling and transit
times, including the origin of the impoundment (natural versus artificial), size, shape, depth, and
regulation of outflow (Chapra, 1977). These characteristics vary widely among lakes and reservoirs.
Sediment deposition in lakes and reservoirs often begins with the formation of deltas at the point of water
inflow, where incoming stream flow decelerates and the heaviest particles settle (Julien 1995). Current
and wind-induced turbulence can keep finer sediments in suspension, creating a deposition pattern called
focusing in which coarser particles are deposited in shallower waters and finer particles in deeper waters,
often in the centers of lakes (Chapra 1997).
Sediment in Estuaries
Estuaries, like lakes and other impoundments, exhibit wide variation in benthic geometry, residence times
or flushing rates, vertical mixing and stratification, wave exposure, and other factors that govern the
transport and fate of sediments. Within estuary systems, flow regimes transition first from unidirectional
turbulent channel flow controlled primarily by topographic gradient and discharge rate to a tidal river
reach in which downstream flow is influenced by the tidal cycle. The flow regime transitions again in the
estuary proper, in which discharge and tidal forces are offsetting and then again in the bay or open ocean
where tidal, wave, or combinations of these forces dominate. As sediment-laden freshwater decelerates
and encounters tidal cycles upon entering the estuary, a null zone is formed in which channel discharge
and tidal action largely cancel each other, favoring the deposition of suspended sediment (Chapra 1997).
In estuaries, lakes and other impoundments, deposited sediments can be disturbed and re-distributed due
either to extreme natural events, including floods and high winds, or by human impacts and interventions.
2.1.3 Turbidity
Turbidity is a non-conventional pollutant and an important dimension of water quality in streams, lakes
and estuaries that is closely related to, but distinct from Total Suspended Solids (TSS). Turbidity refers
generally to the extent to which light can penetrate the water column, with greater depths of light
transmission associated with lower turbidity and higher water quality. Turbidity is commonly measured
using a Secchi disk, lowered into the water column, and the depth at which the black-and-white pattern on
November 2008 2-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
the disk remains visible is the Secchi depth. It is typically assumed that light can penetrate to roughly 2-3
times the Secchi depth. Turbidity is also measured in nephelometric turbidity units, using a turbidimeter
containing a photocell.
Turbidity is often used as an indicator of photosynthetic activity, although all types of particulate matter
within the water column will affect turbidity measurements. Particulate matter consists of mineral
particles ("sediment"), algae and related photosynthetic organisms, and organic detritus. Within most
lakes, photosynthetic organisms are the major source of turbidity, although sediments can be an important
factor in determining water clarity. Sediments can be introduced via erosion and riverine transport
processes, particularly during floods; by human activities (e.g., dredging), by the actions of bottom-
feeding fish such as carp; or by re-suspension due to wind and wave action. In most water bodies,
turbidity also has a seasonal component, as phytoplankton concentrations vary due to water temperature,
dissolved oxygen and the availability of nutrients.
2.2 Documented Relationship between Construction and Sediment
There are a significant number of studies addressing sediment discharges from construction sites, as well
as multiple state water quality reports that document construction sites associated with surface water
impairment with sediment. EPA conducted a literature search that yielded 58 studies on this topic: 17 case
studies that measure sediment increases due to construction, 5 case studies that measure this and also
study the impacts on the aquatic ecosystem, and 36 studies that qualitatively discuss cases of construction
sites increasing sediment levels in surface waters and impacting aquatic life. These studies are
summarized in the sections below.
2.2.1 Impacts of Construction Site Discharge on Sediment Levels in Surface Waters
Table 2-2 lists 17 studies which focus on quantifying impacts of construction site discharges on sediment
levels in surface waters. EPA found an additional 38 documents that discuss this topic qualitatively.
These documents are discussed at the end of this section.
The earliest of the quantitative studies EPA located was conducted from 1961 to 1964 by the U.S.
Geological Survey (Vice et al. 1969). Study of this topic has continued to the present with two studies
published in 2006 (Cleveland and Fashokun and Hedrick et al.). Overall, this body of studies finds that
construction activity increases sediment discharge, subsequently increasing the amount of sediment in
affected water bodies. An exception is Hedrick et al. (2006) which finds no statistical difference in
sediment levels in a stream affected by construction compared to those of an unaffected stream. The
authors attribute this observation partially to the successful implementation of sediment abatement
practices at the construction site and partially to the fact that construction took place during the winter
when the ground was mostly frozen (Hedrick et al. 2006) and therefore more resistant to erosion.
Table 2-2: Studies Quantifying Sediment Increases in Surface Waters
Author(s) Year Study Name
Barrett et al. 1995 Effects of Highway Construction and Operation on Water Quality and
Quantity in an Ephemeral Stream in the Austin, Texas Area
Carline et al. 2003 Effects of Bridge and Road Construction on Spring Creek: Route 26
Transportation Improvements Project
Cleveland and Fashokun 2006 Construction-Associated Solids Loads with a Temporary Sediment
Control BMP
Daniel et al. 1979 Sediment and Nutrient Yield from Residential Construction Sites
November 2008 2-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-2: Studies Quantifying Sediment Increases in Surface Waters
Author(s)
Year
Study Name
Duck 1984 The Effect of Road Construction on Sediment Deposition In Loch
Earn, Scotland
Hedrick et al. 2006 Impact of Tunnel Reconstruction on Stream Water Quality in Great
Smoky Mountains National Park
Helsel 1985 Contributions of Suspended Sediment from Highway Construction
and Other Land Uses to the Olentangy River, Columbus, Ohio
Lubliner and Golding
Nelson and Booth
2005 Stormwater Quality Survey of Western Washington Construction
Sites, 2003-2005
2002 Sediment Sources in an Urbanizing, Mixed Land-Use Watershed
Novotny and Chesters
Owens et al.
1989 Delivery of Sediment and Pollutants from Nonpoint Sources: A Water
Quality Perspective
2000 Soil Erosion from Two Small Construction Sites, Dane County,
Wisconsin
Reed 1980 Suspended-Sediment Discharge in Five Streams Near Harrisburg,
Pennsylvania, Before, During, and After Highway Construction
Stephens et al. 1996 Potential Effects of Coal Mining and Road Construction on the Water
Quality of Scofield Reservoir and Its Drainage Area, Central Utah,
October 1982 to October 1984
Vice et al. 1969 Sediment Movement in an Area of Suburban Highway Construction,
Scott Run Basin, Fairfax County, Virginia, 1961-64
Ward and Appel
1988 Suspended-Sediment Yields in the Coal River and Trace Fork Basins,
West Virginia, 1975-84
Wong 2005 Water Quality in the Haiawa, Haiku, and Kaneohe Drainage Basins
Before, During, and After H-3 Highway Construction, Oahu, Hawaii,
1983-99
Yorke and Herb 1978 Effects of Urbanization on Streamflow and Sediment Transport in
Maryland
Table 2-3, Table 2-4, and Table 2-5 summarize findings from the studies. Depending on the objective of
the study, sediment in receiving waters is measured either as a concentration (Table 2-3) or as a total yield
from the surface water's watershed basin (Table 2-4). Sediment concentrations are measured in
milligrams per liter (mg/L) and units of mass are used to measure sediment yield. Sediment yield is
expressed in terms of total yield or in terms of an increase due to construction site discharge and is
sometimes specified per unit area or period of time. Several studies also measure turbidity in the receiving
water (Table 2-5), as it is closely related to suspended sediment levels and is also an indicator of water
quality. Turbidity, a measure of the extent to which light is scattered, is commonly measured in NTUs,
but also may be measured in other units such as formazin turbidity units (FTUs). Both units measure the
light-scattering properties of the water, but each is specific to the measurement device and its calibration.
In many of the studies, measurements are taken from the surface water area adjacent to the construction
site, though in some cases measurements are taken further downstream.
November 2008
2-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-3: Summary of Sediment Concentration Findings from Selected Studies
Study
Barrett etal. (1995)
Barton (1977)
Chisholm and Downs (1978)
Cleveland and Fashokun (2006)
Cleveland and Fashokun (2006)
Cline etal. (1982)
Cline etal. (1982)
Cline etal. (1982)
Cline etal. (1982)
Duck (1984)
Helsel (1985)
Line and White (2007)
Lubliner and Golding (2005)
Lubliner and Golding (2005)
Lubliner and Golding (2005)
Owens et al. (2000)
Owens et al. (2000)
Wong (2005)
Young and Mackie (1991)
Location
Austin, TX
Ontario, Canada
Boone County, WV
Harris County, TX
Harris County, TX
Rocky Mountains, CO
Rocky Mountains, CO
Rocky Mountains, CO
Rocky Mountains, CO
Loch Earn, Scotland
Olentangy River, OH
Piedmont Region, NC
Northwest WA
Northwest WA
Northwest WA
Dane County, WI
Dane County, WI
Honolulu, HI
Northwest Territories, Canada
Unaffected Site
34.8
2.0
Affected Site
179.0
50.0
— 3,051.2
122.5
1,227.1
16.4
2.2
3.8
2.6
1.4
125.0
349
57.3
1,410.3
10.8
18.9
91.0
121.6
134.3
261.0
6,170
— 33.7
—
—
131.2
628.7
— 2,400.0
—
—
15.0
15,000.0
7.0
93.0
Sediment Concentration (mg/L)
Table 2-4: Summary of Sediment Yield Findings from Selected Studies
Study
Carline et al. (2003)
Daniel etal. (1979)
Downs and Appel ( 1986)
Nelson and Booth (2002)
Reed (1980)
Selbig et al. (2004)
Selbig et al. (2004)
Stephens etal. (1996)
Vice etal. (1969)
Ward and Appel (1988)
Wolman and Schick (1967)
Yorke and Herb (1978)
Location
Centre County, PA
Southeast WI
Coal River, WV
Northwest WA
Harrisburg, PA
Dane County, WI
Dane County, WI
Scofield, UT
Fairfax, VA
Coal River, WV
Maryland
Rock Creek, MD
Sediment Yield
Unaffected Site
Affected Site
— 182.0
— 19.2
107.6
220.0
— : 43
71.0
42.2
3.8
0.2
2,890.0
746.0
294.3
60.3
4.2
0.7
33,320.0
1,480.0
— 97,333.3
— i 32.7
Units
metric tons/year
metric tons/hectare
tons/sq. mi.
tons
tons
tons
tons
metric tons/sq. km.
tons
tons/sq. mi.
tons/sq. mi./year
tons/acre
November 2008
2-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-5: Summary of Turbidity Findings from Selected Studies
Study Author
Barrett, etal. (1995)
Chisholm & Downs (1978)
Cleveland & Fashokun (2006)
Cleveland & Fashokun (2006)
Lubliner & Golding (2005)
Lubliner & Golding (2005)
Lubliner & Golding (2005)
Lubliner & Golding (2005)
Lubliner & Golding (2005)
Lubliner & Golding (2005)
Wong (2004)
Location
Austin, TX
Boone County, WV
Harris County, TX
Harris County, TX
Northwest WA
Northwest WA
Northwest WA
Northwest WA
Northwest WA
Northwest WA
Honolulu, HI
Unaffected site
28
Affected site
112
2.31 | 7.11
95.62
71. 92
141. 12 | 159.22
2.2
2.2
5.2 6.4
5.5
5.2
6.4 25.7
9.8
45.0
17.6 12.0
4.03
4.9
Turbidity (NTU)
1 Averages of multiple measurements at comparable measurement sites on two tributaries to the Little Coal River.
2 Measured in FTU.
3 "Unaffected site" measurement is measurement before start of construction activity.
Because construction site sediment loadings are highly dependent on site conditions, rainfall, nature of
construction activity, presence and effectiveness of sediment abatement practices, and many other factors,
the observations summarized here are only intended to give a sense of the range of documented sediment
loadings. The enormous variation in measurements from different sites and different studies underscores
the difficulty of drawing general conclusions about sediment loadings from all construction sites based on
information available in the literature. EPA estimates more than 590,000 acres of land, encompassing a
wide variety of sediment discharge conditions, undergo construction activity each year in the United
States (USEPA 2008c). The studies summarized here provide valuable information on actual construction
sites, but do not describe conditions resulting from all current construction activity taking place in the
United States. Therefore, EPA did not rely on this collection of studies to quantitatively estimate current
national water quality impacts from construction site sediment discharges and how they would decrease
under the proposed regulatory options. Instead, EPA used the Spatially Referenced Regressions on
Watershed Attributes (SPARROW) water quality modeling methodology to estimate surface water
sediment levels associated with construction activity. This methodology and the results of the analysis are
discussed in more detail in Chapter 6 of this document.
However, it is possible to conclude from the available studies that construction activities do lead to higher
levels of sediment discharge to surface waters and higher levels of sediment in surface waters. All studies
comparing areas under construction to areas unaffected by construction conclude that construction areas
produce higher sediment loadings than their undisturbed counterparts. Although certain observations in
some studies, notably Cleveland and Fashokun (2006) and Cline et al. (1982), show some undisturbed
areas producing higher sediment levels than construction-affected sites, these particular observations are
associated with special circumstances noted by the authors. In Cleveland and Fashokun (2006), TSS
levels in a drainage ditch upstream of a construction site were higher than levels downstream of the site
only during nonstorm periods, whereas most sediment leaves construction sites during or shortly after
precipitation events. The authors note that, based on their observations, construction activity caused a six-
fold increase in total sediment leaving a construction site. In Cline et al. (1982), a reservoir between the
observation points upstream and downstream of a construction site released clear water into the affected
stream, diluting TSS concentrations at the downstream observation point.
In total, 15 of the 17 studies measuring sediment levels in surface waters near or downstream of
construction sites find that construction activity contributed significantly to downstream sediment levels.
November 2008
2-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Of the two studies that do not document a consistent impact, Lubliner and Golding (2005) find that of 183
construction sites visited, only 44 were discharging at the time of their visit. The authors attribute the
absence of discharge at 139 sites to lower intensity rainfall, permeable soils, the use of sediment
abatement practices, and the difficulty of timing their visits to coincide with discharge events at the sites
they had chosen. Of the 44 discharging sites, 6 were discharging directly to surface waters, 2 of which
caused noticeable increases in TSS and turbidity levels downstream of the construction site. The
remaining 38 sites allowed stormwater to seep into the ground or discharge to a city stormwater system
(Lubliner and Golding 2005). The second study, by Hedrick et al. (2006), finds no statistical difference
between sediment loadings at affected and unaffected sites, which they attribute to the effective
implementation of sediment abatement practices and the resistance of frozen ground to erosion.
An early study by Vice et al. (1969) finds that highway construction areas constituted between 1 and
10 percent of a total watershed's basin area at any given time but contributed 85 percent of that
watershed's total sediment yield. Measurements were taken nearly 2 miles downstream from the
construction area, indicating that the construction impacts were not localized to the active construction
site, and discharged sediment was carried downstream. Reed (1980) monitored a highway construction
site in central Pennsylvania before, during, and after construction activity. He finds that sediment
discharge increased two- to four-fold during the active construction period over pre-construction levels,
even in the presence of sediment abatement practices, but quickly returned to preconstruction levels once
construction activity was complete. Daniel et al. (1979) investigated water quality impacts of residential
construction sites in southeastern Wisconsin and find that construction activities yielded significantly
higher amounts of both sediments and nutrients (see the discussion of nutrients in Chapter) than
comparable agricultural land use. This is notable because agricultural land use has been identified as an
important source of water quality degradation through discharges of both sediments and nutrients, as well
as other pollutants (Clark et al. 1985), though it encompasses a larger percentage of total land area in the
United States on an annual basis than construction activity. Nelson and Booth (2002) calculate in their
study of Northwest Washington that construction sites had the highest sediment yield per unit area of all
land uses they examined.
More recently, Barrett et al. (1995) measured water quality indicators upstream and downstream of a
highway construction site in the Austin, Texas area. They find that sediment yields from the site increased
due to construction activity and that turbidity levels in the impacted stream increased more than TSS
levels. The same study also finds increases in metal contamination downstream of construction sites that
were closely correlated to sediment discharge levels (see the discussion of metals in Chapter 3). Similar
to Reed (1980), Barrett et al. (1995) finds these increases to be temporary, dissipating quickly after
construction ended.
Another recent study by Owens et al. (2000) examining water quality impacts from one residential and
one commercial construction site in Wisconsin reports substantial increases in TSS concentrations during
the construction phase at both sites. The commercial site created higher sediment concentrations than the
residential site, though the authors believe this observation is partially explained by the fact that the
residential construction took place during the winter, when rainfall intensity tends to be much lower,
whereas the commercial construction took place during the summer, when rainfall intensity is higher. A
2006 study by Cleveland and Fashokun on construction of a highway in Texas finds that construction
activity resulted in a six-fold increase in total sediment leaving the construction site, despite the presence
of a rock filter dam constructed to reduce sediment discharge from the site. Most recently, Line and White
(2007) compared water quality in a developing area with an analogous water in an undeveloped area.
Their findings suggest that construction activities significantly alter sediment loadings above increases
November 2008 2-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
that may be attributed to overall changes in land use, particularly during the initial land-clearing phase.
Overall TSS export from the developing area was 95 percent greater than from the undeveloped area.
Five of these case studies measure turbidity as well as TSS levels in surface waters affected by
construction site sediment discharges. Barrett et al. (1995) recorded a substantial increase in turbidity
from 28 to 112 NTU at the affected measurement site, noting that this is greater than the increase in TSS
due to the ineffectiveness of construction site silt fences for reducing turbidity. Chisholm and Downs
(1978) note higher turbidity in a reach that was receiving construction sediment discharges compared to a
nearby reach that was not. Similar to their TSS findings, Cleveland and Fashokun (2006) find that
turbidity was higher in the upstream site during non-storm conditions but during storms was higher at the
downstream site. Downstream turbidity values during construction activity were also higher compared to
measurements taken before construction started. As mentioned earlier, Lubliner and Golding (2005) note
an increase in sediment levels at two of six construction sites discharging to surface waters which led to
four- to five-fold increases in turbidity levels. Wong (2004) measured turbidity before and during
highway construction activity in Hawaii and finds that average turbidity was higher during the
construction period over a range of different sites and time periods.
In addition to measuring increases in surface water TSS concentrations and turbidity levels due to
construction site discharges, seven of the above studies discuss increases in sediment deposition, or
sedimentation, also resulting from increases in sediment discharges to surface waters. Duck (1985)
recorded 1,824 metric tons of sediment deposited on a lake bed during a two month construction period.
The author estimates that the lake, under natural conditions, would have otherwise taken 20 to 25 years to
accumulate this quantity of sediment. Similarly, Stephens et al. (1996) note increased sediment deposition
in the Scofield Reservoir in Utah due to construction activities discharging sediment to its tributaries.
As well as increasing the quantity of sediment on the surface water bed, sedimentation can also change
the composition of a surface water bed's substrate. Vice et al. (1969) estimate that 50 percent of the
sediment discharged from observed construction activity was deposited in receiving surface waters.
Barton (1977) finds that deposition increased 10-fold due to construction site discharges to a small stream
in Ontario, resulting in decreased proportions of organic matter in the sediment. Chisholm and Downs
(1978) recorded 8 to 10 inches of silt and clay deposited on the bed of a small stream in West Virginia.
Cline et al. (1982) recorded an increased percentage of fine sediments in the substrate of the observed
surface water.
None of these studies evaluate how long sediment deposits remain on surface water beds. Barrett et al.
(1995) note that increased sediment deposition in the creek under study was ameliorated within one year
of cessation of construction activity. Sediment deposited in surface waters with higher energy currents
such as those found in many rivers and streams is more likely to be re-suspended in the water column and
carried downstream than sediment deposited in water bodies with lower energy currents, such as many
lakes and reservoirs. These water bodies can lose capacity over time due to sediment accumulation.
In addition to the studies presented above, there are a significant number of other cases in which
construction activity has been implicated in elevated downstream sediment loadings and ecological
degradation. A number of these cases have been reported by state environmental protection departments
as part of their assessment of their state's surface water quality. A review of the state of Ohio's water
quality reports from the early 1990s through 2006 yielded 31 reports specifically identifying water quality
degradation due to construction activity (Ohio EPA 1994, 1995, 1996a, 1996b, 1996c, 1997a, 1997b,
1997c, 1997e, 1997f, 1998a, 1998b, 1998c, 1999a, 1999b, 1999c, 1999d, 2000, 2001b, 2003a, 2003b,
2004, 2005, 2006a, 2006b, 2006c, 2006d, 2007b, 2007c). Ohio EPA (1997f), a 1995 water quality study
November 2008 2-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
of the Twin Creek watershed, focuses on construction as the source of pollution in the Twin Creek basin.
A golf course construction site had several negative impacts on water quality and aquatic life in several of
the water bodies evaluated. Turbidity in the water column was observed up to 10 miles downstream from
the construction site and was attributed to uncontrolled construction site stormwater discharges to the
surface water. The authors caution that construction site discharges pose a threat to this otherwise high-
quality watershed, because though each construction site is short-lived, ongoing development can make
construction's influence on water quality pervasive and substantial overtime.
The states of Michigan and Missouri have also produced reports that identify construction as a source of
water quality degradation (Missouri DNR 2004, 2005; Michigan DEQ 2006).
In addition to these state reports, the World Wildlife Fund (WWF) identifies construction sediment
discharges as a threat to the health of the Tennessee, Cumberland, and Mobile river basins (Buckner et al.
2002).
Helsel (1985) and Ward and Appel (1988) both note that although construction site sediment discharges
increase sediment levels in nearby water bodies, the small area of construction sites relative to the larger
area of the entire watershed render these increases less perceptible at a larger scale.
The studies summarized in this section provide strong evidence linking the land disturbance associated
with construction activities to elevated sediment and turbidity levels in downstream surface waters. These
increases have been documented over the past half-century in many regions of the country, over a wide
range of site conditions, in all seasons, and with varying levels of sediment abatement practices,
underscoring the widespread nature of the problem of sediment pollution from construction sites.
2.2.2 Impacts of Construction Site Sediment Discharge on the Aquatic Ecosystem
Sediment has multiple well-documented impacts on aquatic ecosystems. These impacts are described in
more detail in Section 2.2. EPA's literature search yielded six studies that measure the impact of
construction sediment discharges on both water quality and the aquatic ecosystem. Table 2-6 lists these
studies.
Table 2-6: Studies Quantifying Sediment Increases and Their Impacts on Aquatic Life
Author(s) Year Study Name
Barton 1977 Short-term Effects of Highway Construction on the Limnology of a
Small Stream in Southern Ontario
Chisholm and Downs 1978 Stress and Recovery of Aquatic Organisms as Related to Highway
Construction Along Turtle Creek, Boon County, West Virginia
Clineetal. 1982 The Influence of Highway Construction on the Macroinvertebrates
and Epilithic Algae of a High Mountain Stream
Selbig et al. (USGS) 2004 Hydrologic, Ecologic, and Geomorphic Responses of Brewery Creek
to Construction of a Residential Subdivision, Dane County,
Wisconsin, 1999-2002
Wolman and Schick 1967 Effects of Construction on Fluvial Sediment, Urban and Suburban
Areas of Maryland
Hunt and Grow 2001 Impacts of Sedimentation to a Stream by a Non-Compliant
Construction Site
Like the studies described in Section 2.2.1, which focus primarily on the magnitude of sediment yield or
surface water sediment concentration change created by construction sites, these six studies provide
November 2008 2-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
quantitative measurements of sediment loadings to impacted surface waters as part of their assessment of
the biological communities in the affected water bodies. The magnitudes of the sediment loadings
documented by these studies are listed in Table 2-3, Table 2-4, and Table 2-5. All of the studies find that
sediment yields or concentrations in surface waters affected by construction sites were higher than those
in unaffected waters.
Wolman and Schick (1967, p. 451) find that increased sediment deposition led to changes in the structure
of the aquatic community immediately downstream of construction activity. They list "...blanketing of
bottom-dwelling flora and fauna, alteration of the flora and fauna due to changes in light transmission and
abrasive effects of sediment, and alterations of species offish due to changes produced in the flora and
fauna upon which the fish depend..." as the impacts observed in a stream reach downstream of
construction activity.
Barton (1977) and Chisholm and Downs (1978) conducted studies on the effects of highway construction
on aquatic ecosystems in southern Ontario and West Virginia, respectively. Barton (1977) notes that
standing fish crops were significantly reduced immediately downstream of construction activity that
caused TSS concentrations to increase from 2 mg/L to 50 mg/L. The study also notes that the composition
of the benthic macroinvertebrate community shifted, though no species were eliminated. Chisholm and
Downs (1978) find more severe impacts to the benthic macroinvertebrate community, noting a reduction
in total abundance of organisms and the elimination of certain benthic species due to TSS concentrations
greater than 3,000 mg/L caused by construction site discharges.
Cline et al. (1982) evaluated the effects of highway construction on the macroinvertebrate community in a
Rocky Mountain stream and found it to be altered by construction, though not as severely as anticipated.
Sediment concentrations downstream of areas affected by construction activity were 9 to 45 times greater
than concentrations downstream of unaffected areas. Impacts observed were similar to those found by
Barton (1977) and Chisholm and Downs (1978) and included a reduction in density, abundance, and
diversity, though no elimination of species was found. The authors also note reduced organic content in
the epilithon (rock-coating material) and fewer algal species.
In three of these studies, the biological effects of increased sediment levels adjacent to or a short distance
downstream of the construction site are found to be temporary, typically lasting less than one year after
the cessation of construction activity (Barton 1977; Chisholm and Downs 1978; Cline et al. 1982).
Selbig et al. (2004) find no statistical difference between the sediment loadings from a site affected by
construction and one unaffected by construction, though the construction-affected site did have
marginally higher mean sediment levels measured from 30 to 100 feet downstream of construction
activity. They also find no substantive changes to the biological communities in the affected stream
reaches. The authors attribute this lack of observable impact to the effective implementation of several
different sediment abatement practices at the construction site.
In a 1999 study of the water impact of noncompliant construction activities on Ohio's Turkey Creek,
Hunt and Grow (2001) attributed the significant decrease in fish species diversity, quantity, and weight to
the significant increase of silt and clay in the creek's substrate due to uncontrolled discharge from the
construction site. The number offish species decreased from 26 at the upstream site to 19 at the
downstream site. The number offish in the equally sized study areas decreased more drastically, from 525
to 230. Pollutant-tolerant species increased by 237 percent, while pollutant-intolerant species decreased
by 33 percent. The weights offish also dropped by 62 percent. The overall rating of the aquatic habitat
declined significantly at the downstream site due to these negative impacts attributed to uncontrolled
construction site discharge.
November 2008 2-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
In addition to these studies that examine the effects of quantified increases in sediment levels on aquatic
ecosystems, there are multiple studies that recognize upstream construction activity as the source of
downstream aquatic life degradation in specific locations. Water quality reports from the Ohio EPA
evaluate biological indicators of water quality and, where possible, identify the water body's sources of
impairment. Multiple reports identify fish and macroinvertebrate communities impaired by elevated
suspended or deposited sediment levels deriving from construction activity upstream. Several types of
construction are identified as sediment sources in the studies, including road (Ohio EPA 1997e, 1998c,
1999b, 200la), bridge (Ohio EPA 2001a, 200Ib), highway (Ohio EPA 1994, 1998b, 1998d, 1999d, 2005,
2005, 2007a, 2007b), residential (Ohio EPA 1997d, 2006d), commercial (Ohio EPA 2005), golf course
(Ohio EPA 1997f, 1998e, 1999d, 2004, 2006b, 2007c), and airport (Ohio EPA 1996a, 1998c, 2001b). In
particular, Ohio EPA (1997f), a 1995 water quality study of the Twin Creek watershed, finds that
construction of a golf course severely impacted nearby stream habitat, where the investigators found none
of the live fingernail clams that were present throughout the rest of the watershed but found instead
pollutant-tolerant midges which were not common in the overall watershed. They attribute this to the four
to six inches of silt that had been introduced to the stream bed from construction discharges.
The state of California has also expressed concern about discharges from construction sites and
specifically discusses the impacts of these discharges on threatened and endangered California coho
salmon and steelhead trout and their habitat (McEwan and Jackson 1996; California Department of Fish
and Game 2004). The state noted, for example, that discharges from the construction of Interstate 5 in
California depleted spawning gravels for coho salmon (California Department of Fish and Game 2004).
The World Wildlife Fund documents the most severe aquatic ecosystem impact, describing an incident in
Tennessee where road construction crews failed to implement appropriate sediment controls and a rain
event deposited more than four feet of sediment on the bed of a nearby creek, eliminating aquatic
organisms and contaminating the water supply of a downstream town (Buckner et al. 2002).
The studies summarized in Sections 2.2.1 and 2.2.2 above document impacts associated with known
construction sites. Given the large number of construction sites active every year, the large number of
sediment-impaired waters in the United States (see Section 2.3.1), and limitations in resources available
for impact study and documentation, the documentation of impacts summarized here cannot describe all
impacts associated with sediment discharge from construction sites or all known incidences of impact.
This fact is made clearer by reports of sediment discharge from construction sites from members of the
public that have not been published as studies. For example, several reports were submitted as sworn
testimony in (NRDC et al. v. U.S. EPA, 2005a, 2005b, 2005c, 2005d, 2005e). These reports describe
various testifier-observed construction site impacts including sediment-laden and turbid discharges,
elevated surface water turbidity and sediment levels, creek and lake sedimentation, loss of recreational
opportunities, and compromised surface water aesthetics.
Section 2.3.2 summarizes studies of aquatic organism and ecosystem impacts associated with excessive
sediment levels in surface waters. These studies are not specific to sediment discharges from construction
sites; however, given the wide range of characteristics associated with construction site sediments, they
are likely to exhibit a range of behaviors and impacts similar to those of sediment from other pollutant
sources.
November 2008 2-15
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
2.3 Receiving Water Impacts of Sediment
2.3.1 Sediment and Turbidity Impaired Waters in the United States
When additional sediment enters a surface water, it can raise the TSS, TDS, and turbidity levels in the
water column, increase deposition of sediment on the water body's bed, and impact the surface water's
overall quality. Sediments, whether suspended, dissolved, or settled, play an important role in an
ecosystem, providing habitat for benthic species, spawning grounds for fish, nutrient transport, and other
roles. Appropriate levels of sediment are water body-specific, as sediment levels vary naturally among
different types of water bodies and according to the geology, topography, stream gradient, water body
morphology, vegetative land cover, climate, soil erodibility and other landscape characteristics of the
contributing watershed (USEPA 2006b). Changing the quantity and behavior of sediments in an aquatic
ecosystem, however, can have adverse consequences for its members (USEPA 2006b). Elevated sediment
levels also negatively impact human use of aquatic resources for recreation, drinking water supply,
irrigation, and navigation and can impair stormwater management systems and surface water aesthetics.
State Sediment Criteria
Many states have developed criteria to designate appropriate sediment levels for either all waters or for
specific surface waters within the state. Many of these criteria are narrative statements describing the
general nature of healthy sediment levels without attempting to provide numeric guidelines. Some states
have also set numeric criteria for TSS and/or turbidity levels. In addition, specific Total Maximum Daily
Loads (TMDLs) developed for water bodies within a state that are impaired for sediment are likely to
contain numeric sediment standards for that water body.
Table 2-7 summarizes state sediment criteria for surface water quality. States using narrative criteria are
noted with a "Yes" in the "Narrative TSS" or the "Narrative Turbidity" column.
Table 2-7: Sediment Criteria for Surface Water Quality, by State
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Numeric TSS
(mg/L) Narrative TSS
-
-
-
I
|
| Yes
|
-
-
Yes
10-553 I Yes
| Yes
1000b |
| Yes
| Yes
| Yes
Yes
Yes
Numeric
Turbidity (NTU)
50
5-25g
10-50g
10-758
Narrative
Turbidity
-
Yes
Yes
Yes
|
I
|
10
-
29
-
a
Yes
Yes
| Yes
1
| Yes
25
Yes
| Yes
-
25-50
Yes
Yes
November 2008
2-16
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-7: Sediment Criteria for Surface Water Quality, by State
State
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Puerto Rico
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Numeric TSS
(mg/L) Narrative TSS
-
-
| Yes
|
-
Yes
Yes
I Yes
|
25C | Yes
| Yes
25
Yes
I Yes
500d | Yes
30 |
1-7.256'1 |
1
Yes
Yes
|
1
Yes
53-263 (at any
time); 30-150
Yes
(monthly
average)8
| Yes
| Yes
35-90g Yes
-
Yes
-
-
| Yes
Yes
Numeric
Turbidity (NTU)
Narrative
Turbidity
-
150 at any time;
50 monthly
average
Yes
| Yes
|
-
50
Yes
Yes
I Yes
|
| Yes
10
Yes
30-50 at any 1
time; 10-15 ;
monthly average8
Yes
I
10-50g |
1
|
1
10% increase
above
background
40-100 NTUh
Yes
Yes
|
5-10 NTUg |
10% above
background1
Yes
Yes
| Yes
| Yes
10-15 NTUg
Yes
10-25 NTUg
Yes
5-10 NTU or 10-
20% increase8
10 NTU or 10%
increase
-
| Yes
10-15 NTUg
Yes
November 2008
2-17
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-7: Sediment Criteria for Surface Water Quality, by State
State
Numeric TSS
(mg/L) Narrative TSS
Numeric
Turbidity (NTU)
Narrative
Turbidity
a. Varies based on several parameters
b. Individual water body criteria from Illinois EPA (2006)
c. Individual water body criteria from Nevada DEP (2003)
d. Standard is for total dissolved solids (TDS)
e. Individual water body criteria from Ohio EPA (2000)
f Individual water body criteria from Ohio EPA (2005)
g. Varies based on water body classification
h. Specific to the Neshaminy Creek Basin (USEPA 2006b)
i. Specific to trout stock waters
Sources: USEPA (2006b); Illinois EPA (2006); Nevada DEP (2006); Ohio EPA (2000);
Ohio EPA (2005)
Sediment and Clean Water Act (CWA) Sections 303(d) and 305(b) Impaired Waters
Under Section 305(b), states are required to submit to EPA every two years a report describing the water
quality of all waters in the state. States have determined the appropriate uses of water bodies within their
jurisdiction. Typical uses include recreation, drinking water supply, navigation, and wildlife habitat,
among others. If a water body fails to meet any one of its designated uses, Section 305(b) of the CWA
requires states to list that water body as "impaired." If the water body currently meets its designated uses
but may not in the future, the state may list that water body as "threatened." Along with this list, states are
required to submit under Section 303(d) of the CWA a list of all impaired and threatened water bodies
still requiring Total Maximum Daily Load (TMDL) criteria (USEPA 2005a).
EPA's 2002 National Water Quality Inventory: Report to Congress (USEPA 2007d), which summarizes
the states' 305(b) reports, lists "sediment/siltation" as the most prevalent cause of pollution in rivers and
streams in terms of number of miles affected and the fourth most prevalent for lakes, ponds, and
reservoirs. Table 2-8 and Table 2-9 present impairment of water bodies from the 2002 305(b) reporting
cycle (the most recent year for which data are publicly available). It is important to note that the water
bodies assessed do not represent all water bodies in the United States. The aforementioned report
indicates that only 19 percent of rivers and streams; 37 percent of lakes, ponds, and reservoirs; and
35 percent of estuaries were evaluated for the 2002 reporting cycle (USEPA 2007d).
Sediment-related impairment in the 305(b) data falls into several categories of pollutant causes, including
"sediment/siltation," "turbidity," and "suspended solids." According to documentation for WATERS,
"sediment/siltation" impairment is related to settled sediments, while "turbidity" impairment is caused by
suspended particles, and this parent cause includes "suspended solids."
Of the assessed rivers and streams, more than 103,000 miles were reported as impaired due to
"sediment/siltation," along with nearly 1,264,000 acres of lakes, ponds, and reservoirs. "Turbidity" and
"suspended solids" impair close to 29,500 miles of rivers and streams, and nearly 839,000 acres of lakes,
ponds, and reservoirs. Nearly 119 square miles of bays and estuaries are reported as impaired by
"sediment/siltation" and 1,689 square miles by "turbidity" or "suspended solids."
November 2008
2-18
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 2-8: "Sediment/Siltation" Impairment in 305(b)-Assessed Waters 1
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
River, Stream
(miles)
Assessed
21,699.5
Impaired
5,276.8
12,814.8 4,874.3
127,468.7
66,507.3
128,144.0
49,969.2
7,541.4
6,766.5
22,942.4
3,514.5
56,467.7 9,512.9
110,169.4
6,892.0
41,477.6 19,362.1
80,414.5
695,132.8
17,271.3
103,954.1
Lake, 2 Pond, Reservoir
(acres)
Assessed
1,491,595.5
409,319.9
271,042.5
1,173,482.6
3,699,833.3
2,726,913.2
693,980.7
1,912,016.9
1,272,016.2
1,181,681.6
14,831,882.3
Impaired
175,527.5
50,960.0
44,378.5
19,554.0
155,925.9
179,666.7
9,958.0
252,899.0
279,590.1
95,725.8
1,264,185.4
Bay, Estuary
(sq. miles)
Assessed
1,018.7
Impaired
0.1
1,675.9 0.5
4,996.1
35.2
329.8 0.0
86.0 0.0
7,802.7
6.0
0.0 0.0
0.0
0.0
958.8 73.0
13,577.7 3.8
30,445.7 118.7
1 Values are preliminary pending correction of EPA's 305(b) dataset, publicly available through the WATERS database, which links
305(b)-assessed waters to their causes of impairment. The 2002 305(b) report (USEPA 2007d) lists 100,446 miles of rivers and streams
and 1,317,938 acres of lakes and reservoirs impaired by "sediment/siltation."
2 Does not include Great Lakes
Source: WATERS (USEPA 2008g)
Table 2-9: "Turbidity" and "Suspended Solids" Impairment in 305(b)-assessed Waters1
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
River, Stream
(miles)
Assessed
21,699.5
Impaired
1,719.0
12,814.8 23.7
127,468.7
525.9
66,507.3 458.6
128,144.0
49,969.2
6,292.7
9,936.3
56,467.7 135.0
110,169.4
41,477.6
3,340.6
5,913.1
80,414.5 1,111.1
695,132.8
29,455.9
Lake 2, Pond, Reservoir
(acres)
Assessed
1,491,595.5
409,319.9
271,042.5
1,173,482.6
3,699,833.3
2,726,913.2
693,980.7
1,912,016.9
1,272,016.2
1,181,681.6
14,831,882.3
Impaired
21,275.1
0.0
0.0
1,810.0
136,122.5
469,765.5
95,423.0
23,483.5
86,962.7
3,803.5
838,645.8
Bay, Estuary
(sq. miles)
Assessed
Impaired
1,018.7 j 9.3
1,675.9 0.0
4,996.1
1,528.1
329.8 7.1
86.0
7,802.7
0.0
52.0
0.0 0.0
0.0
0.0
958.8 ! 37.9
13,577.7 54.9
30,445.7
1,689.3
1 Values are preliminary pending correction of EPA
305(b) assessed waters to their causes of impairment
2007d).
2 Does not include Great Lakes
Source: WATERS (USEPA 2008g)
s 305(b) dataset publicly available through the WATERS database, which links
Turbidity was not among the pollutants listed in the 2002 305(b) report (USEPA
Figure 2-2 presents the United States divided into EPA regions, which are referred to in the tables above
and throughout the remainder of this document.
November 2008
2-19
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 2-2: EPA Regions
Rl
I i ,-rt '
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
summarizes the distribution of current TSS concentrations in the Reach File Version 1.0 (RF1): network
as estimated by EPA's analysis. The RF1 network consists of approximately 650,000 miles of rivers and
streams and associated lakes and reservoirs in the conterminous 48 states. The median TSS concentration
as estimated by this analysis is 261.91 mg/L when weighted by reach length. The range of TSS
concentrations falls between 24.79 mg/L and 3,203.47 mg/L at the 5th and 95th percentiles respectively.
The average TSS concentration is 704.88 mg/L, indicating that concentrations in the 90th to 95th percentile
range are high enough to shift the average concentration significantly higher than the median
concentration.
Table 2-10: SPARROW Distribution of TSS Concentrations
1,2
JVTI
Reach
Count
62,370
JVTI
River
Miles
650,043
Average TSS
(mg/L)1
704.88
5th
Percentile
24.79
25th
Percentile
92.71
50th
Percentile
261.91
75th
Percentile
768.51
95th
Percentile
3,203.47
1 Concentrations weighted by reach length. Statistics exclude potential outliers (defined as values above the 95th percentile (TSS = 6,049.7
mg/L)).
2 Results are reflective of both polluted and unpolluted waters.
Source: USEPA analysis
Figure 2-3 shows total RF1 river miles and their current TSS concentrations as predicted by SPARROW
(not weighted by reach length). As shown below, the majority of waters have TSS concentrations below
1,000 mg/L, though substantially higher concentrations up to and exceeding 7,800 mg/L can be found in
some U.S. waters.
The Reach File Version 1.0 (RF1) database contains location, length, and other information for 62,370 reaches in the
conterminous United States (USEPA 2007d). Each individual reach is given a unique identifier known as an RF1 ID, which
is used throughout this Environmental Assessment to identify individual waters benefitting from water quality
improvements expected form this regulation. See the RF1 documentation for more information at
http://www.epa.gov/waters/doc/rfl meta.html.
November 2008
2-21
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 2-3: TSS Concentrations by Total RF1 River Miles as Predicted by SPARROW for Baseline Conditions
160,000
150,000 :
140,000 :
130,000 :
120,000 :
110,000
ro
'5
Total: 62,370 river reaches (650,043 river miles).
40,000
30,000
20,000
10,000
TSS Concentration
Source: USEPA analysis
A certain level of sediment is present in surface waters under natural, undisturbed conditions, and some
level of sediment is necessary for the natural biological function of most surface waters. Current levels of
sediment discharge to the United States surface water network are much higher, however, than would be
observed under natural, undisturbed conditions due to sediment discharges from construction, agriculture,
eroding streambeds due to land use change, and other human activities.
Sediment from construction sites (and other sources) can have a wide range of effects on aquatic
ecosystems located downstream of those construction sites. Section 2.2.2 discusses the types of impacts
sediments can have on aquatic life. Chapter 4 and Chapters 7-10 describe the impacts sediments can have
on human use of aquatic resources, including aquatic life.
2.3.2 Aquatic Life Impacts of Sediment
Increased levels of sediment in surface waters have been shown to have negative impacts on aquatic life.
These effects are documented for primary producers, such as aquatic plants, as well as for mollusks,
benthic macroinvertebrates, fish, and other aquatic organisms. The various manifestations of elevated
sediment levels (elevated TSS, TDS, turbidity, and sedimentation levels) affect different aquatic
organisms in different ways. Some organisms are directly impacted. Other organisms are harmed
indirectly as direct impacts on some organisms cascade upward or downward through the ecosystem. For
example, aquatic vegetation reduces photosynthesis and growth under high turbidity levels and resulting
November 2008
2-22
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
low light conditions. Reduced vegetation growth can reduce food availability for a variety of aquatic
organisms, depressing their survival rates (Henley et al. 2000).
Sediment impacts have been researched since at least the 1960s through both laboratory dose-response
and in-stream field studies, though many aspects remain to be explored. Much research has been done on
the direct effects of sediment on fish, particularly salmon because of their economic importance. Many
more recent studies address the interconnected nature of aquatic ecosystems and examine a broader range
of organism types. Documented effects and relevant studies are organized in the discussion below by the
type of aquatic organism studied (primary producer, invertebrate, fish) and by the type of organism
reaction. A section specific to threatened and endangered species follows this discussion.
Primary Producers
Primary producers such as algae, phytoplankton, and submerged aquatic vegetation play a critical role in
the health of aquatic ecosystems by transforming sunlight into energy and providing food for other
aquatic organisms. Their existence and productivity is therefore crucial to the survival of the entire
ecosystem. A compounding factor of reduced quantities of aquatic macrophytes and algae due to
sediment impacts in an aquatic ecosystem is that there are fewer stems, leaves, and other materials to slow
water flow and dislodge suspended sediment from the water column. This effect makes recovery more
difficult and allows elevated TSS concentrations to persist further downstream (Wood and Armitage
1997).
A 1997 study by Wood and Armitage identifies four primary ways in which elevated sediment levels can
affect primary producers in aquatic ecosystems:
> Reduce light penetration,, thus reducing photosynthesis levels
> Damage aquatic vegetation leaves and stems through abrasion
> Prevent algal cells from attaching to water body substrate
> Reduce organic content of periphytic cells.
Light reduction: A reduction in primary production as a result of increased turbidity due to elevated TSS
concentrations is the most common impact that has been documented by LaPerriere et al. (1983), Lloyd
(1987), and Rivier and Seguier (1985) (all as cited in Waters 1995), Edwards (1969) and Meyer and
Heritage (1941) (both as cited in Kerr 1995). LaPerriere et al. (1983) find that a moderately turbid stream
with turbidity measurements of approximately 170 NTU had oxygen production reduced by 0.08-
0.64 g/m2 per day. In highly turbid streams exceeding 1,000 NTU, primary production was undetectable.
Lloyd (1987) develops a model relating turbidity and primary production. The model indicates that in a
clear, shallow stream, a turbidity increase of 5 NTU decreases primary production between 3 percent and
13 percent, and an increase of 25 NTU reduces primary production by up to 50 percent (Waters 1995).
The turbidity levels recorded by studies of construction sites often exceed 100 NTU and are high enough
to cause a reduction in primary production. A study by Guenther and Bozelli (2004) attempts to determine
the ultimate cause of reduced photosynthetic activity in turbid waters, which they postulate could be a
result of either lower light levels or of increased sinking of phytoplankton due to their bonding to
sediment particles. The study's results suggest that reduced light penetration is the primary cause of lower
primary production.
Physical damage: Algae and aquatic macrophytes can also be physically damaged by elevated TSS
concentrations. This effect is noted by Lewis (1973), cited in Kerr (1995), from an experiment in which
concentrations in excess of 100 milligrams per liter of suspended coal particles are shown to cause severe
November 2008 2-23
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
abrasive damage to the leaves of the aquatic moss Eurhynchium riparioides over a period of 3 weeks. The
damage substantially lowered the moss's ability to produce chlorophyll.
Reduced organic periphy ton: Periphyton consist of the collection of aquatic organisms, bacteria, and
dead organic matter that are attached to submerged surfaces in aquatic ecosystems. A reduction or
complete elimination of periphytic cells due to elevated sediment levels is described by Van
Nieuwenhuyse and LaPerriere (1986), Pain (1987) (both as cited in Waters (1995)), and Samsel (1973, as
cited in Kerr 1995). This impact was manifested by a general reduction in the number and diversity of
aquatic primary producers. Graham (1990), as cited in Waters (1995), finds that a decrease in the
percentage of organic matter in periphytic material may result from additional sediment particles binding
to the periphyton. Herbivores consuming the periphyton also consume the elevated quantities of sediment,
reducing the nutritional benefit from their feeding (Ryan 1991).
Invertebrates
Aquatic invertebrates, including zooplankton, aquatic insects, and bivalves (mussels and clams), are an
important link in the aquatic food chain between primary producers and larger animals such as fish
(Waters 1995). The effects of elevated sediment levels on invertebrates, particularly on benthic
macroinvertebrates, have been the focus of a large number of studies over the past several decades. The
impacts of sediment on invertebrates can be classified into the following categories identified by Kerr
(1995) and Wood and Armitage (1997):
> Loss of habitat
> Altered movement and increased drifting
> Altered species composition
> Physiological changes
> Reduced feeding activity
> Reduced growth rates
> Toxicity and direct mortality
Loss of habitat: Increased sediment levels in water bodies can degrade benthic macroinvertebrate habitat.
Sediment settles from the water column onto the substrate to which sedentary organisms attach and also
fills small nooks and crevices in water body beds, known as interstitial spaces, where a variety of species
live, reproduce, and seek protection from predators. Siltation of previously coarse substrates is implicated
in the reduction of mussel populations by Ellis (1936) and Marking and Bills (1980), both as cited in
Waters (1995). Waters (1995) also cites a severe sedimentation case documented by Cordone and
Pennoyer (1960) in which "bottom fauna densities and biomass... were reduced to less than 10% of
former levels" after a "gravel-washing operation covered stream bottoms up to one foot in depth or more
and reduced the substrate to a 'hard, bedrock-like' condition." Ryan (1991), as cited in Henley et al.
(2000), concludes that as little as a 12 percent to 17 percent increase in interstitial fine sediment could
result in a 16 percent to 40 percent decrease in macroinvertebrate abundance.
Altered movement and increased drifting: Movement downstream (drifting) of benthic
macroinvertebrates has been found to increase with elevated sediment concentrations (Kerr 1995).
Downstream drift reduces organism abundance in the impacted surface water. Increases in invertebrate
drift have been shown to be specifically related to the turbidity caused by suspended sediments (Waters
November 2008 2-24
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
1995). White and Gammon (1977), as cited in Waters (1995), postulate that prolonged high
concentrations of suspended sediment could reduce benthic macroinvertebrate population levels.
Altered species composition: Loss of habitat and increased drifting can culminate in a change in the
species composition of an aquatic community as sediment-sensitive species are replaced by sediment-
tolerant species. Gammon (1970), as cited in Kerr (1995), finds that TSS concentrations between 40 mg/L
and 120 mg/L reduced macroinvertebrate density by 25 percent, and greater TSS concentrations caused
more substantial reductions of 60 percent. Organisms intolerant of turbid waters include mayflies, clams,
and bryozoans (Cooper (1987), as cited in Kerr (1995)). Sediment discharges from construction activity
are implicated in reduced invertebrate abundance and modified community composition in several
studies, including Chisholm and Downs (1978), Cline et al. (1982), Ohio EPA (2003a), and Levine et al.
(2003).
Physiological changes: The physical effects of elevated sediment levels on invertebrate species can be
severe. Sediment particles can clog an organism's filter feeding apparatus or digestive system (Kerr
1995), resulting in injury to or death of the organism.
Reduced feeding activity: Another effect of elevated TSS and turbidity levels on aquatic invertebrates is a
reduction in their feeding activity. Kerr (1995) notes that this impact is predominant in filter feeding
species such as mussels and clams and cites Ellis (1936), which finds that under turbid conditions,
mussels and clams reject silt-laden food and, under highly turbid conditions, close their shells completely.
Waters (1995) also discusses sediment's impact on feeding, citing Ellis (1936) and Aldridge (1987), who
find that sediment additions impaired feeding in three species of bivalves. Reduced feeding activity can
result in impaired functioning or death of organisms.
Reduced growth rates: Kerr (1995) cites three studies (Appleby and Scarratt (1989), Kirk (1992), and
Paffenhofer (1972)) that find reduced growth rates in invertebrates subjected to elevated sediment levels
due, potentially, to reduced feeding activity or reduced food value.
Toxicity and direct mortality: Sediment can directly or indirectly increase invertebrate mortality levels.
Forbes et al. (1981), as cited in Kerr (1995), finds that suspended sediment are acutely toxic to amphipods
(shrimp-like crustaceans). Henley et al. (2000) concludes from a literature review that increased sediment
concentrations have a negative effect on the survival rates of freshwater mussels with the magnitude of
the impact being species-specific. The review also suggests that many of the more than 30 extinct species
of North American mussels died out due to a loss of habitat, most likely from sedimentation of the
substrate (Henley et al. 2000).
The overall effect of the various impacts from elevated sediment levels on aquatic invertebrates is a
general reduction in their abundance, as documented in the literature reviews conducted by Kerr (1995),
Waters (1995), and Henley et al. (2000). Studies of specific construction sites also find that
macroinvertebrate communities were impaired downstream of construction sites, citing elevated levels of
TSS and turbidity as the causes (Chisholm and Downs 1978; Cline et al. 1982; Fossati et al. 2001; Levine
et al. 2003, 2005; Selbig et al. 2004; Ohio EPA 1996a, 1997f, 1998d, 1999d, 2006b).
Fish
A substantial amount of research has been done to identify and quantify the effects of elevated sediment
levels on a variety offish species. The observed impacts generally fall into the following categories:
> Avoidance
> Feeding and hunting
November 2008 2-25
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
> Breeding and egg survival
> Juvenile survival
> Physical damage
> Mortality and population level changes
Avoidance: Because fish are much more mobile than aquatic plants, plankton, or invertebrates, they are
capable of actively avoiding waters with high levels of suspended sediment and/or turbidity if migration
routes to alternative habitats are available. Barton (1977) notes a relocation offish populations away from
areas with elevated TSS levels in a small stream in southern Canada. Servizi and Martens (1987, as cited
in Reid et al. 2003) determine that turbidity levels in excess of 37 NTU elicit avoidance behavior among
fish populations. These types of avoidance behaviors may reduce or eliminate fish populations in stream
reaches with sustained elevated sediment levels.
Feeding and hunting: Elevated TSS and turbidity levels have been shown to reduce feeding rates in fish
through studies on bluegills, arctic grayling, Coho salmon, and steelhead (Kerr 1995). This effect is
generally believed to be a result of decreased visibility caused by turbidity, as many fish rely on their
vision to locate food or prey. De Robertis et al. (2003) find support for this theory in a study where
piscivorous fish, which visually locate their prey at longer distances, were more severely affected by
increases in turbidity levels than were planktivorous fish, which locate their prey at much shorter
distances. Research into the impacts of elevated sediment levels by USEPA (2004a) concludes that
turbidities greater than 25 NTU or TSS levels of 2,000-3,000 mg/L or greater are sufficient to reduce the
predation abilities offish. This reduction in predation can benefit prey species that use turbid water to
hide from predators (Rowe et al. 2003). However, a paper by the Canadian Department of Fisheries and
Oceans (2000) suggests that as turbidity levels continue to increase, the advantage of reduced predation is
outweighed by the prey fishes' own increased difficulty in locating food. A compounding effect of
elevated sediment levels is decreased food availability due to invertebrate population decline and
decreased primary productivity, as described in the sections above.
Breeding and egg survival: In addition to reducing habitat for invertebrates, elevated sediment levels also
reduce spawning habitat for multiple fish species. Saunders and Smith (1965, as cited in Kerr 1995) find
that increased TSS levels reduced fish spawning activity. Other studies document lower survival rates for
eggs spawned in surface waters impacted by elevated TSS and turbidity levels (Chapman 1988 as cited in
Canadian Department of Fisheries and Oceans 2000). This effect is most likely due to the fact that
sediments, particularly the finer fractions, can reduce the level of oxygen available to eggs deposited in
the substrate (Kerr 1995). Bjornn et al. (1977), as cited in Kerr (1995), find that successful emergence of
fish fry from eggs begins to decline when fine sediment levels in the substrate are in excess of 20-30
percent. Reynolds (1989, as cited in Henley et al. 2000) finds that increases in turbidity levels are
associated with increases in sac fry mortality in arctic grayling. These reductions in spawning and
survival of egg and fry have the potential to reduce fish population levels in surface waters impacted by
sediments.
Juvenile survival: Research into the effects of sediment on fish populations has noted that high sediment
concentrations can impact juvenile fish more severely than adults. Juvenile Coho and Chinook salmon
exhibited abnormal surfacing behavior under higher sediment concentrations, rendering them more
vulnerable to avian predators (Canadian Department of Fisheries and Oceans 2000). The Canadian
Department of Fisheries and Oceans (2000) study also notes that excessive sediment levels can act as a
stressor to juvenile fish and consequently increase their predation risk. As with a reduction in fish egg and
November 2008 2-26
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
fry survival, reduced survival of juvenile fish to adulthood has the potential to reduce fish population
levels.
Physical damage: Sediment particles, particularly more angular particles, are capable of causing physical
damage to fish by clogging gills and impairing breathing or through skin abrasion (Bell 1973, as cited in
Waters 1995). Kerr (1995) notes that fish can withstand short episodes of high sediment concentrations by
producing mucus to protect their skin and gills, but this reaction stresses fish. Reid et al. (2003) finds that
elevated sediment levels reduces the ability of rainbow trout to withstand hypoxic conditions and thereby
reduced their average life span.
Mortality and population level changes: The potential cumulative effect of these different impacts from
elevated sediment levels is a decrease in fish population levels in impacted areas. Reductions can take
place through direct mortality in the short term and through reduced reproductive success in the long
term. Newcombe (1994) and Newcombe and Jensen (1996), (both cited in Henley et al. 2000) find that
elevated sediment concentrations are associated with increased mortality in at least 14 species offish.
USEPA (2004a) states that suspended sediment concentrations between 500 and 6,000 mg/L can result in
significant (greater than 50 percent) mortality. Berkman and Rabeni (1987), as cited in Wood and
Armitage (1997), noticed a decline in the overall abundance offish stocks as sediment levels increased in
a river in northeast Missouri. Barton (1977), Graves and Burns (1970), and Ohio EPA (1999c) find that
standing fish crops were reduced immediately downstream from construction activity.
The severity of impacts on fish populations is related to the intensity and duration of elevated sediment
levels (Anderson et al. 1996, as cited in USEPA 2004a). Short-term increases in TSS and turbidity levels
can occur during spring thaws, storms, and other natural events. However, a construction project may
produce sediment levels significantly higher those associated with natural events, for longer periods of
time, and/or at times when an aquatic ecosystem is unaccustomed to receiving such sediment inputs,
exposing fish populations to the impacts described above.
Threatened and Endangered Species
Threatened and endangered (T&E) and other special status species can be adversely affected in several
ways by construction activities. This section discusses the potential for impacts from construction site
sediment discharges on aquatic T&E species. Construction activity can jeopardize aquatic T&E species
through increased sediment levels, a significant contributor to aquatic plant, invertebrates, and fish injury
and mortality as discussed in the section above. For example, elevated sediment concentrations adversely
affect aquatic plants and microscopic primary producers through reduced light penetration, which
impedes photosynthesis (Kerr 1995). Elevated sediment levels impact aquatic invertebrates and fish
through habitat modification and destruction as well as through direct adverse physical impacts on
species. Some fish species cannot survive in water with high levels of suspended sediments, and
sediments that settle on river or stream bottoms can smother fish egg deposits (NOAA 2004). The
potential further decline of listed species or critical habitat for these species from construction activity is
particularly important because these species are already rare and at risk of irreversible decline.
The Federal trustees for T&E species are the Department of the Interior's U.S. Fish and Wildlife Service
(USFWS) and the Department of Commerce's National Marine Fisheries Service (NMFS). The
departments are also referred to collectively herein as the Services. The USFWS is responsible for
terrestrial and freshwater species (including plants) and migratory birds, whereas the NMFS deals with
marine species and anadromous fish (USFWS 2008a). Various state agencies and departments have state-
level jurisdiction over T&E species.
November 2008 2-27
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
A species is federally listed as "endangered" when it is likely to become extinct within the foreseeable
future throughout all or part of its range if no immediate action is taken to protect it. A species is listed as
"threatened" if it is likely to become endangered within the foreseeable future throughout all or most of
its range if no action is taken to protect it. Species are selected for listing based on petitions, surveys by
the Services or other agencies, and other substantiated reports or field studies. The 1973 Endangered
Species Act outlines detailed procedures used by the Services to list a species, including listing criteria,
public comment periods, hearings, notifications, time limits for final action, and other related issues
(USFWS 1996).
A species is considered to be federally threatened or endangered if one or more of the following listing
criteria apply (USFWS 2007):
> The species' habitat or range is currently undergoing or is jeopardized by destruction,
modification, or curtailment.
> The species is overused for commercial, recreational, scientific, or educational purposes.
> The species' existence is vulnerable because of predation or disease.
> Current regulatory mechanisms do not provide adequate protection.
> The continued existence of a species is affected by other natural or manmade factors.
States and the federal government have also included species of "special concern" on their lists. These
species have been selected because they are (1) rare or endemic, (2) in the process of being listed, (3)
being considered for listing in the future, (4) found in isolated and fragmented habitats, or (5) considered
a unique or irreplaceable state resource.
The federal government and individual states develop and maintain lists of species that are considered
endangered, threatened, or of special concern. The federal and state lists are not identical: a state does not
list a species that is on the federal list if it is extirpated in the state. States may also list a species that is
not on the federal list if the species is considered threatened or endangered at the state, but not federal,
level.
This discussion focuses on federally listed T&E species based on information in the USFWS Threatened
and Endangered Species System (TESS) database (USFWS 2008b), available at
http://www.fws.gov/endangered/wildlife.html. Information on both federal and state listed species is
available online in the NatureServe database (NatureServe 2008) at http://www.natureserve.org/explorer/.
Additional information on state-listed species is available from state T&E coordinators.
Numerous physical and biological stressors have resulted in the listing of multiple aquatic species as
threatened or endangered. The major factors include habitat destruction or modification, displacement of
populations by exotic species, dam building and impoundments, increased sediment and turbidity levels
in the water column, sedimentation, various point and nonpoint sources of pollution, poaching, and
accidental catching.
As discussed in the sections above and in Chapter XX3XX, aquatic organisms are susceptible to adverse
effects from pollutant discharges from construction activity. Increased sedimentation, TSS, and turbidity
levels in water bodies downstream of construction activity can depress organism growth, survival, and
reproduction, thereby leading to further decline in an impacted T&E species population.
Because construction activities subject to NPDES permitting occur in every state, there are many
potentially affected T&E aquatic species. The current USFWS list of T&E species includes 255 aquatic
November 2008 2-28
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
species in five species groups: fish, amphibians, clams, crustaceans, and corals (USFWS 2008b). For a
complete list of threatened and endangered aquatic species, see Appendix A. Although the USFWS list of
T&E species does not provide information on causes of endangerment for each species, a review of the
ecological literature suggests that sedimentation may have adverse impacts on species that are already in
peril either directly or indirectly through impacts on supporting food chains (Cloud 2004; NatureServe
2008; USFWS 2008b).
Several T&E species are thought to be particularly susceptible to excessive sedimentation. One such
example is Amblema neislerii, an endangered freshwater mussel commonly known as the fat threeridge.
As the NatureServe Explorer database notes, "the species and its habitats continue to be impacted by
excessive sediment bed loads of smaller sediment particles, changes in turbidity, [and] increased
suspended solids..." (NatureServe 2008). Other similarly sediment-susceptible species of mollusks
include the oyster mussel, purple bankclimber, Louisiana Pearlshell, Alabama Heelsplitter, and Ouachita
rock pocketbook (USFWS 2008b).
Many fish species are also vulnerable to sedimentation. Sedimentation is a risk factor for a number of
shiner species, including the Arkansas river shiner, Pecos bluntnose shiner, and blue shiner (NatureServe
2008). The NatureServe Explorer notes that the biggest protection need for the threatened blue shiner
(Cyprinella caeruled), native to the southeastern United States, is "preventing] siltation of habitat,
especially during the spawning period" (NatureServe 2008). Sedimentation is also listed as a threat factor
to a number of minnow and darter species (e.g., loach minnow, spikedace, and leopard darter) (Cloud
2004; NatureServe 2008; USFWS 2008b). Sedimentation has also contributed to the threatened status of
some populations of rainbow trout and several salmon species in the Pacific Northwest (NatureServe
2008). California coho salmon and steelhead trout are listed and threatened and endangered by the state of
California. The state has expressed concern about the impact of sediment discharges from construction
sites on habitat throughout these species' range (McEwan and Jackson 1996; California Department of
Fish and Game 2004). The state noted, for example, that discharges from the construction of Interstate 5
in California depleted spawning gravels for coho salmon (California Department of Fish and Game 2004).
November 2008 2-29
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
3 Other Pollutants and Impacts from Construction Discharges
In addition to mobilizing and discharging sediment, settleable solids, TDS, TSS, and turbidity to surface
waters, construction activity has the potential to contribute many other kinds of pollutants to adjacent
surface waters. These pollutants, though not as well studied as sediment in the construction context, have
the potential to impact water quality and therefore aquatic life and human use of water resources.
Construction activity involves the use of a wide variety of materials with the potential to produce metal,
wood, plastic, and other debris that can be washed into nearby waters if not properly managed. In
addition, chemicals associated with construction materials and equipment such as wood treated with
creosote and chromated copper arsenate, paint, adhesives, solvents and vehicle oils, fuel, and grease can
leach or wash from equipment and materials during rain events and subsequently discharge to surface
waters. Asphalt and concrete pavement and building materials contain chemicals that can alter water pH
or have toxic effects in aquatic environments. Concrete wastes have an elevated pH due to the lime they
contain. In addition, during the manufacturing process, concrete is often mixed with industrial wastes that
can elevate metal levels in concrete waste. Asphalt materials contain a variety of organic compounds,
including polycyclic aromatic hydrocarbons (PAHs) and additives whose environmental effects have not
been thoroughly characterized. Coal tar-based sealcoat for driveway and parking lot surfaces, in
particular, contains very high levels of PAHs (Van Metre et al. 2006) and has been documented as a cause
of a fish kill (Montgomery County 2000). Commonly used landscaping materials such as fertilizer,
mulch, lime, and pesticides can also wash from construction sites into surface waters, contributing
nutrients, oxygen-demanding material, and toxic pollutants, including metals present in many fertilizers
and pesticide residues.
In addition to pollutant discharges from construction equipment and materials, land disturbance
associated with construction can mobilize pollutants already present on a site when construction begins,
These pollutants can be natural in origin (e.g., nutrients, oxygen-demanding materials, or metals naturally
present in a site's soil), or they can derive from earlier contamination of a site by human activity. Former
industrial activities can deposit a wide range of pollutants in a site's soil, including toxic metals and
organic compounds. Land previously used for agriculture can contain high levels of pesticide residues,
nutrients from fertilizers, or salts from irrigation.
Pollutants other than sediment have not been as widely researched or documented in the construction
environment. This is despite the fact that many of the pollutants described above adsorb to or otherwise
travel with sediment discharges. EPA was able to locate 11 studies and reports that describe pollutants
other than sediment found in construction site discharges or at elevated levels immediately downstream of
construction sites. The more frequently described pollutants are:
> Nutrients (nitrogen and phosphorus compounds)
> Polycyclic aromatic hydrocarbons (PAHs)
> Metals
These pollutants are discussed in additional detail in the sections below. Other pollutants such as
ammonia (Ohio EPA 1997f), fecal coliform (Ohio EPA 1997f, 1998c, 1998d), and chemicals from spills
were also described in some studies. In general, however, they do not seem to be well characterized.
November 2008
3-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
An additional impact from construction activity is that vegetation removal, soil compaction, and modified
surface gradients can increase the volume and intensity of water runoff discharging from a construction
site. The increased flow can cause changes in the hydrograph of a receiving stream, causing it to exhibit
more frequent and larger peak flows, followed by periods of reduced base flow and possibly desiccation.
The stream may incise or widen to accommodate the periodic elevated flow volumes (Wheeler et al.
2003). The changes in streambed morphology can increase sediment erosion, destabilize stream banks,
elevate sediment levels in downstream waters, and cause long-term changes in aquatic and riparian
habitat. These changes can result in the destruction of previously existing aquatic communities.
3.1 Documented Pollutants Other than Sediment from Construction Sites
3.1.1 Nutrients
Nutrients, primarily phosphorus and nitrogen compounds, have the potential to discharge from
construction sites in stormwater discharges, elevating nutrient levels in receiving waters. Nitrogen is
typically found in soluble forms. Phosphorus may be released in either soluble form or attached to
sediment (Daniel et al. 1979). Three of the studies discussed in Chapter 2 measured nutrient as well as
sediment levels in impacted surface waters. Two additional papers discuss nutrients in construction site
discharges qualitatively. Table 3-1 lists these studies and the specific nutrients they describe.
Table 3-1: Nutrients in Studies of Construction Site Discharges
Author(s)
Daniel et al.
Novotny and
Chesters
Harbor et al.
Ohio EPA
Fossati et al.
Year
1979
1989
1995
1997
2001
Study Name
Sediment and Nutrient Yield from Residential Construction
Sites
Delivery of Sediment and Pollutants from Nonpoint Sources: A
Water Quality Perspective
Reducing Nonpoint Source Pollution from Construction Sites
Using Rapid Seeding and Mulching
Biological and Water Quality Study of Twin Creek and
Selected Tributaries 1995
Impact of Sediment Releases on Water Chemistry and Macro-
invertebrate Communities in Clear Water Andean Streams
(Bolivia)
Specific Nutrients
Phosphorus,
Nitrogen
Phosphorus
Phosphorus
Nitrate, Nitrite,
Ammonia
Phosphates
Daniel et al. (1979) reported that elevated nitrogen and phosphorus loadings in waters receiving
construction site discharges were 90 and 99 percent (respectively) explained by elevation in sediment
loadings from residential construction sites sampled in Wisconsin. Novotny and Chesters (1989) also
studied construction sites in Wisconsin and concluded that nutrient discharges are closely related to
sediment discharge levels because water discharging from construction sites carries both soil and the
nutrients it contains. Harbor et al. (1995) corroborated this conclusion, and also noted that certain erosion
control measures reduced sediment discharge by up to 86 percent and phosphorus discharge by up to 80
percent.
Though Ohio EPA (1997f) and Fossati et al. (2001) do not include detailed discussions of nutrient levels
in sediment at construction sites, they do note that increased levels of nutrients have accompanied
increases in sediment levels immediately downstream of construction activity. Ohio EPA (1997f) directly
attributes these increases to the lack of proper construction site discharge management at a golf course.
Elevated levels of TSS, nitrogen compounds (including ammonia, which has both nutrient and toxic
November 2008
3-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
properties), and fecal coliform were recorded at a chemical sampling site about half of a mile downstream
of the construction discharge site.
3.1.2 Polycyclic Aromatic Hydrocarbons
Polycyclic aromatic hydrocarbons (PAHs) are not frequently documented as construction site discharge
pollutants. However, Ohio EPA water quality reports recognize construction sites, specifically road
construction sites, as possible contributors to elevated PAH levels in at least three rivers: Little Cuyahoga
River, Little Miami River, and Scioto River (Ohio EPA 1998d, 1998e, 1999b). PAHs are found in the
asphalt and tar used in road construction and repair, and it is possible for PAHs to leach into sediment and
construction site discharges during rain events (Ohio EPA 1998d).
3.1.3 Metals
Like PAHs, metals have not been frequently documented as pollutants in construction site discharges.
However, an Ohio EPA report on Raccoon Creek attributes increased metal concentrations to pressure-
treated wood products used at a nearby residential construction site (Ohio EPA 1997d). Barrett et al.
(1995) recorded concentrations of iron and zinc increasing much more than those of other metals as TSS
concentrations increased in surface waters due to construction site discharges. The study found that TSS
concentrations explained 40 to 94 percent of the variance in iron concentration (depending on the
measurement site), but identified no direct correlation to other parameters and did not hypothesize as to
the ultimate source of the iron or zinc increases.
3.2 Receiving Water and Aquatic Life Impacts of Pollutants Other than
Sediment
3.2.1 Nutrients
Nutrients are an impairment source for a large number of surface waters in the United States, ranking just
behind sediment in total number of listings. Nutrient impairment (including associated algal growth and
algal blooms) impairs the second-largest number of stream and river miles and the fifth-largest number of
lake, pond, and reservoir acres. Table 3-2 summarizes the total number of stream and river miles and lake,
pond, and reservoir acres reporting nutrient or algal-related impairment in the 2002 305(b) assessment
data. This information was obtained from the EPA WATERS database (USEPA 2008g). As with
sediment listings, it should be noted that the 305(b) data only assess 19 percent of rivers and streams; 37
percent of lakes, ponds, and reservoirs; and 35 percent of bays and estuaries, so the actual number of
waters impaired by nutrient pollution is likely to be much higher than the 2002 305(b) data indicate.
As shown in Table 3-2, nutrients impair more than 21,000 miles of rivers and streams; more than
1.7 million acres of lakes, ponds, and reservoirs; and more than 1,800 square miles of bays and estuaries.
Region 5 shows the greatest length of assessed rivers and streams impaired, as well as the largest number
of impaired lake, pond, and reservoir acres. Region 3, which includes the Chesapeake Bay, accounts for
almost all nutrient-impaired bay and estuary area in the United States, with more than 1,800 square miles.
According to WATERS, nutrients are also the cause of impairment for 13 percent of 303(d)-listed surface
waters (USEPA 2008g), the more severely impaired waters requiring TMDL development.
November 2008 3-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 3-2: Nutrient Impairment in 305(b)-Assessed Waters1
EPA
Region
1
2
o
5
4
5
6
7
8
9
10
Total
River, Stream
(miles)
Assessed Impaired
21,699.5
12,814.8
3,080.1
0.0
127,468.7 3,447.2
66,507.3
128,144.0
1,525.6
9,315.6
49,969.2 873.8
56,467.7
110,169.4
784.6
858.7
41,477.6 269.7
80,414.5
695,132.8
1,525.0
21,680.1
Lake, 2 Pond, Reservoir
(acres)
Assessed Impaired
1,491,595.5
409,319.9
210,600.4
140.8
271,042.5 41,053.4
1,173,482.6
23,414.0
3,699,833.3 I 964,737.0
2,726,913.2 90,150.4
693,980.7
1,912,016.9
89,630.0
265,414.6
1,272,016.2 35,778.6
1,181,681.6
14,831,882.3
71.7
1,720,990.9
Bay, Estuary
(sq. miles)
Assessed Impaired
1,018.7
1,675.9
0.0
0.0
4,996.1 1,831.0
329.8
86.0
0.0
0.0
7,802.7 6.0
0.0
0.0
0.0
0.0
958.8 7.8
13,577.7
30,445.7
0.0
1,844.7
1 Values are preliminary pending correction of EPA's publically available 305(b) dataset
2 Does not include Great Lakes
Source: WATERS (USEPA 2
While nutrients are necessary for the primary production (photosynthetic activity) that forms a
fundamental part of the aquatic food chain, nutrients are also of concern as a pollutant in surface waters
because they can cause eutrophication. Eutrophication is a process by which excessive nutrient levels
increase the growth rates of primary producers, particularly algae, which can drastically alter a water
body's ecological balance and also impair human use of the water body.
Excessive algal growth rates can lead to algal blooms, certain types of which can be both directly and
indirectly harmful to both aquatic organisms and human beings. Harmful algal blooms can be directly
toxic to aquatic organisms, resulting in impairment of populations and even local species extinctions
(Burkholder 1998). Algal blooms known as brown or red tides are intense growths of algal species which
release neurotoxins into the water column. The neurotoxins can harm some forms of aquatic life. These
types of blooms usually necessitate beach closings and temporary bans on seafood harvesting from the
water body because consuming shellfish from affected waters can cause shellfish poisoning in humans
(Burkholder 1998).
Algal blooms caused by excessive nutrient input into a water body can significantly alter aquatic
community structure. Increased algal growth can reduce available light for submerged aquatic vegetation,
reducing photosynthesis levels and decreasing food and habitat for benthic organisms. Decomposition of
excessive algal biomass reduces dissolved oxygen levels in surface waters, lessening the amount available
to other aquatic organisms and either impairing their functioning or suffocating them.
Other types of algal blooms may not be as directly harmful to humans but can still cause economic
damage because the increased levels of organic material associated with the blooms may clog drinking
water intakes or cause an unpleasant taste in drinking water (Boyd 1990).
3.2.2 Polycyclic Aromatic Hydrocarbons
Because PAHs are not specified as a type of 305 (b) impairment in the EPA WATERS database, surface
waters in the United States are not listed as impaired by this specific pollutant. However, PAHs are a
known carcinogen in many animals and are suspected to be carcinogenic in humans as well, though no
November 2008
3-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
specific information on concentrations of concern is currently available (USEPA 2008e). PAHs tend to
travel attached to sediment, settling with it on water body beds. They may then be ingested by benthic
dwelling or feeding organisms, facilitating PAH transfer up the food chain (USEPA 2004b). PAHs also
have bioaccumulation potential and have been found to cause birth defects and liver and blood problems
in some animals (USEPA 2008e). PAHs in benthic sediments have been specifically linked to liver
abnormalities in bottom-dwelling fish species (Malins et al. 1988; Vethaak and Rheinhallt 1992).
3.2.3 Metals
Metals as a group (excluding mercury, which is considered separately) are considered an impairment
category in the EPA WATERS database. As with sediment and nutrient listings, it should be noted that
the 305(b) data only assess 19 percent of rivers and streams; 37 percent of lakes, ponds, and reservoirs;
and 35 percent of bays and estuaries, so the actual number of waters impaired by nutrient pollution is
likely to be much higher than the 2002 305(b) data indicate.
Table 3-3 details 305(b) metal impairment listings by EPA region. More than 33,000 miles of assessed
rivers and streams were impaired by metals, along with nearly 1.3 million acres of lakes, ponds, and
reservoirs and 4,700 square miles of bays and estuaries. Region 1 shows the greatest length of assessed
rivers and streams with metal impairment at 11,606, while Region 6 accounts for the largest share of lake,
pond, and reservoir metal impairment. As with nutrient impairment, Region 3 shows the largest area of
bays and estuaries impaired by metals, though Region 6 also reported nearly 1,700 square miles of metal-
impaired bays and estuaries in 2002. WATERS data on current 303(d) lists show that metals account for
12.5 percent of all 303(d) waters awaiting TMDL development.
Table 3-3: Metal Impairment in 305(b)-Assessed Waters1
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
River, Stream
(miles)
Assessed j Impaired
21,699.5
11,606.2
12,814.8 0.0
127,468.7
66,507.3
5,284.2
517.2
128,144.0 6,824.4
49,969.2
56,467.7
3,376.0
2,151.1
110,169.4 1,178.3
41,477.6
293.0
80,414.5 2,324.6
695,132.8 1 33,555.0
Lake 2, Pond, Reservoir
(acres)
Assessed
1,491,595.5
Impaired
398,797.8
409,319.9 0.0
271,042.5
1,173,482.6
30,168.4
32,155.0
3,699,833.3 55,137.0
2,726,913.2
693,980.7
634,326.8
135,478.0
1,912,016.9 1,681.6
1,272,016.2
169.0
1,181,681.6 0.0
14,831,882.3 1 1,287,913.6
Bay, Estuary
(sq. miles)
Assessed
1,018.7
Impaired
21.3
1,675.9 0.0
4,996.1
2,996.5
329.8 ! 0.0
86.0 0.0
7,802.7
0.0
1,689.4
0.0
0.0 0.0
958.8
7.8
13,577.7 12.0
30,445.7 | 4,727.0
1 Values are preliminary pending correction of EPA's publically available 305(b) dataset
2 Does not include Great Lakes
Source: WATERS (USEPA 2008g)
Metals tend to adsorb to sediment in the aquatic environment. Metals can therefore accumulate on water
body beds as they settle out of the water column with sediments. Metals deposited in the substrate may be
consumed by benthic organisms and then progress through the food chain, potentially harming larger
organisms. Trace amounts of metals are necessary for normal biological functions in many aquatic
species. However, concentrations above the trace level can be acutely or chronically toxic, adversely
affecting behavior and reproductive success in both aquatic species and waterfowl. Aquatic ecosystem
November 2008
3-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
exposure to metals has also been shown to reduce photosynthetic efficiency and algal colonization,
lowering primary production rates and diminishing food available to other organisms.
When present at sufficiently high levels, metals can cause dredged sediment to be classified as toxic,
making its disposal more costly. During dredging or flooding, sediment containing metals may be
reintroduced into the water column because of the disturbance. Metals attached to suspended sediment
can be transported downstream, potentially contaminating areas far from their point of origin. Disturbance
of contaminated sediments can also release metal contaminants into water drawn from surface waters for
drinking water supply.
November 2008 3-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Overview of Benefits from Regulation
This chapter provides an overview of the potential benefits to society related to the reduced sediment
discharges from construction sites that will result from the proposed regulation. Sediments and other
pollutants from construction sites may have a wide range of effects on water resources located in the
vicinity of construction sites. These environmental changes affect economic productivity (e.g., drinking
water supply and storage and navigation) as well as environmental services valued by humans (e.g.,
recreation, public and private property ownership, existence services such as aquatic life, wildlife, and
habitat designated uses). Related market benefits (e.g., avoided costs of producing various market goods
and services) and nonmarket benefits are additive (Freeman 2003). In all cases, benefits are
conceptualized and estimated based on established welfare theoretic models (Freeman 2003; Just et al.
2004).
This chapter provides a conceptual framework for understanding the benefits likely to be achieved by the
regulation, a qualitative discussion of those benefits, and a review of the benefit estimation methods used
in EPA's analysis of this regulation. The following chapters quantify and estimate the economic value of
these benefit categories.
4.1 Conceptual Framework for Valuation of Environmental Services
The economic benefits of the proposed regulation reflect the money-metric values associated with the
resulting improvements in water and other resources. The conceptual approach for estimating the
monetary value of benefits of a policy event involves an evaluation of changes in social welfare realized
by consumers and producers, and quantification of these changes in money-metric terms. Such measures
are based on standardized and widely accepted concepts within applied welfare economics (Just et al.
2004). They reflect the degree of well-being derived by economic agents (e.g., people and firms) given
different levels of goods and services, including those associated with environmental quality. For market
goods, analysts typically use money-denominated measures of consumer and producer surplus, which
provide an approximation of exact welfare effects (Freeman 2003). For nonmarket goods, such as aquatic
habitat, values must be assessed using nonmarket valuation methods (Freeman 2003). In such cases,
valuation estimates are typically restricted to effects on individual households (or consumers), and either
represent consumer surplus or analogous exact Hicksian welfare measures (e.g., compensating variation).
The choice of welfare measures (i.e., value) is often determined by the valuation context.
The economic value of the changes in environmental services provided by surface waters from the
regulation is a product of three sets of functional relationships (Freeman 2003):
> The effect of the regulation on environmental quality (Q)
> Human uses of the affected resources and their dependence on environmental quality (X)
> The economic value of the uses of the environment (V).
The first functional relationship relates the proposed regulation to changes in environmental quality. In
this case, water quality is a function of sediment loadings (S), in-stream pollutant concentrations (vector
P), and other relevant attributes (vector A) such as stream flow and velocity. Because the proposed
regulation affects private activities that influence sediment discharges, sediment loadings are a function of
private responses to the proposed regulation (R). This relationship may be represented as:
November 2008
4-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Q = f(S(R),P,A) (Eq.4-1)
For this analysis, water quality changes are estimated in terms of sediment concentrations in receiving
waters using SPARROW (see Chapter 6 and Appendix A for details).
The second set of functional relationships describes the human values of the affected water resources and
their dependence on water quality. In the following analyses, sediment discharges is treated as an "input"
to the water quality, and, as a result, as an "input" to ecological services provided by the affected water
bodies. Like any other input, the value of changes in sediment discharges is determined by its impact on
water quality. Water quality, in turn, influences human uses of the affected resources, leading to changes
in use values. It may also lead to changes in ecosystem services that generate nonuse values.
For example, water quality is often characterized based on its suitability for recreational activities (e.g.,
whether the water is beatable, fishable, or swimmable) or its ability to support specified uses (e.g.,
primary/secondary contact recreation, public drinking water supply, or commercial fishing). To link water
quality changes from reduced sediment discharges to effects on human uses and support for aquatic and
terrestrial species habitat, EPA used a Water Quality Index (WQI) for this analysis (McClelland 1974;
USEPA 2004b). Chapter 10, Section 10.1.1, provides detail on the WQI and its application to the benefits
analysis for the regulation. The WQI presents water quality in a way linked to suitability for varying
human uses, but does not in itself identify associated changes in human behavior. Behavioral changes and
associated welfare effects are implied in the proposed benefit transfer approach for measuring economic
values. Additional details of this approach are provided below and in the following chapters.
In addition to water quality (Q), the level of human uses or other ecological services (vector X) of the
environmental resource depends on a variety of other factors. If vector/represents these other inputs into
the production of environmental services or uses based on the affected water resource, the second
functional relationship can be expressed as:
X = f(Q,I) (Eq.4-2)
The vector of environmental services or human uses (X) affected by changes in water quality and
sediment discharges could include (van Houtven et al. 2005):
> Services received through purchases of market goods (Z) produced with surface water resources (e.g.,
commercial fish)
> Goods, services, and activities produced within the household (H), such as recreation days
> Public sector services (G) such as drinking water quality
> Existence values (N) associated with water quality improvements.
The third set of relationships defines the economic value (V) of the ecological services or human uses of
the environment. This relationship "embodies the value judgment society has adopted for economic
purposes" (Freeman 2003). For simplicity, and following standard approaches in benefit cost analysis
(Boardman et al. 2001), this analysis assumes that the value function is a simple aggregation of
individuals' values.2
This function could also incorporate weights or environmental justice concepts.
November 2008 4-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
The relationship between individual or household utility (U) and water quality (Q) can be more formally
described though the following utility function (van Houtven et al. 2005):
U = U(Z(Q),H(Q,I),G(QIN(Q\Y;B) (Eq. 4-3)
Individuals are assumed to derive utility from Z, H, G, and N, as well as from other goods and services
(Y), which are not related to water resources. A subset of household goods and services (H), such as
recreation, are "produced" with water quality (Q) and other inputs (I) such as travel. To account for
variation in this utility structure across individuals, B denotes a vector of individual or household
characteristics, such as demographic characteristics, that influence the form or parameters of the utility
function.
Maximization of Eq. 4-3 with respect to an income (M) constraint produces the following indirect utility
function:
V = V(Q,P,M;B) (Eq.4-4)
Where P represents the vector of prices associated with goods and services (Z, H, G, and Y).
The total dollar value associated with a change in water quality from Q0 to Q\ (holding other variables
constant) can therefore be expressed using the compensating surplus (CS) measure from the following
equation:
V(Q0 ,P0J0-B) = V(Q, ,P0J0-CS'B) (Eq. 4-5)
Other things being equal, an individual is better off with a higher level of water quality (Qi) than a lower
level (Qo). CS in this equation represents the individual's maximum willingness, to pay for the specified
water quality improvement, leaving the individual just as well off after payment and higher water quality.
From the above, CS (or willingness to pay, WTP) may be derived as a function of pre- and post-change
environmental quality, prices, income, and individual attributes:
CS = CS(Q0,Ql,P0,Y0;B) (Eq. 4-6)
This function also provides the basic conceptual foundation for constructing a benefit transfer function,
which can be used to predict WTP values for defined changes in water quality (van Houtven et al. 2005).
4.2 Applicable Benefit Categories
There are likely to be several categories of benefits of the regulation; some benefits are linked to direct
use of market goods and services, and others pertain to nonmarket goods and services. The following
sections outline the most important categories of the expected benefits, and approaches used to estimate
each category. In cases where estimates were not possible due to data limitations or other constraints, this
is also noted below.
4.2.1 Market Benefits
Some of the social benefits expected from the regulation will manifest themselves in economic markets
through changes in price, cost, or quantity of market-valued activities. For benefit endpoints traded in
markets, such as increased yields from commercial fisheries, benefits can be measured using standard
November 2008 4-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
market approaches based on data including market prices and quantities (Boardman et al. 2001).
Competitive prices can also be used to measure benefits related to avoided costs associated with market
goods and services. In analyzing benefits of the regulation, EPA used the avoided cost method to
monetize three benefit categories: (1) maintenance of navigational waterways, (2) reservoir dredging, and
(3) drinking water treatment. Other applicable market benefits (e.g., commercial fishing) are discussed
qualitatively. The latter approach represents an appropriate measure of social benefits for cases in which a
policy change reduces costs to producers of a market-supplied good, but in which price effects are
minimal (Boardman et al. 2001, p. 70-74).
Maintenance of Navigational Waterways
The primary effect of sediment discharges on navigation and transportation is accumulation of sediment
in navigational channels and harbors (USEPA 2004b). Keeping these areas passable requires extensive
dredging, which can be costly. Additionally, since dredged material may contain significant quantities of
toxics and heavy metals, dredging can create new environmental problems (Clark et al. 1985).
Implementation of the regulation is expected to result in less frequent dredging of navigable waterways
due to reduced sediment discharges from construction sites and, as a result, cost savings to the
government and private entities responsible for maintenance of navigational shipping channels, harbors,
and other waterways. Chapter 7 of this report details methods and presents results of EPA's analysis of
benefits to navigation expected from the regulation.
Water Supply and Use
> Reservoir Dredging. Water storage facilities (e.g., reservoirs) serve many functions, including
providing drinking water, flood control, hydropower supply, and recreational opportunities. Sediment
in streams can be carried into reservoirs, where it can settle and build up layers of silt over time. An
increase in sedimentation rates will reduce reservoir capacity. To replace this capacity, sediment must
be dredged from reservoirs, or new reservoirs must be constructed (Clark et al. 1985). Sediment can
also affect evaporation rates at reservoirs. Turbid waters tend to be sharply stratified, with a warm
layer at the surface and a cooler layer below. Because warm water evaporates faster, turbidity can
cause higher rates of water loss from reservoirs. However, turbidity may also increase the reflectivity
of water, which makes water absorb solar heat more slowly than it otherwise would, and reduces
evaporation rates (Clark et al. 1985). The overall effect from temperature stratification could be
positive or negative. The regulation is expected to reduce the amount of sediment entering reservoirs
and, as a result, the need for sediment mitigation measures in these reservoirs and the cost of reservoir
maintenance. Chapter 8 of this report details methods and presents results of EPA's analysis of
benefits from reduced sedimentation of reservoirs due to the regulation.
> Drinking Water Treatment and Household Water Use. Discharges from construction sites can affect
the quality and cost of providing drinking water. Sediment and other discharged pollutants negatively
affect water quality, and require increased spending on treatment measures such as settlement ponds,
filtration, and chemical treatment. There is an additional cost of removing sludge that is created
during the treatment process (USEPA 2007a). EPA used the avoided cost of treating drinking water
for sediment and disposing of sludge resulting from reduced turbidity in source water as a measure of
benefits resulting from the proposed regulation. Chapter 9 of this report details methods and presents
results of EPA's analysis of benefits to drinking water treatment plants. The presence of pollutants or
dissolved minerals in drinking water may also affect the flavor and odor of water. Although EPA did
not estimate a value of improved taste and odor of drinking water, this benefit is implicitly accounted
November 2008 4-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
for in the WTP for changes in environmental services provided by surface waters affected by
construction site discharges stemming from the regulation (see Chapter 10 of this report for details).
> Industrial Water Use. Sediment may have negative effects on industrial water users. Suspended
sediment increases the rate at which hydraulic equipment, pumps, and other equipment wear out,
causing accelerated depreciation of capital equipment. Sediment can also clog cooling water systems
at power plants and other large industrial facilities. However, at least one positive effect is possible as
well. Since turbidity may reduce the rate at which water bodies absorb solar heat, more turbid water
bodies may supply cooler water, which in turn could improve the efficiency of cooling water systems
(Clark et al. 1985). The proposed regulation is expected to benefit industrial water users by reducing
sediment concentrations in source waters and thus increasing the useful life of industrial equipment.
The Agency, however, was not able to quantify this benefit category due to data limitations.
> Agricultural Water Use. Irrigation water that contains sediment or other pollutants from construction
sites can harm crops and reduce agricultural productivity. Suspended sediment can form a crust over a
field, reducing water absorption, inhibiting soil aeration, and preventing emergence of seedlings.
Sediment can also coat the leaves of plants, resulting in inhibited plant growth and reduced crop value
and marketability. Furthermore, irrigation water that contains dissolved salts or pollutants can harm
crops and damage soil quality (Clark et al. 1985). The proposed regulation is expected to benefit
agricultural producers by reducing sediment discharges and, as a result, sediment deposition on farm
land; this would lead to improvements in land productivity and enhanced marketability of agricultural
products. The Agency, however, was not able to quantify this benefit category due to data limitations.
Commercial Fishing
Sediment and other discharges from construction sites can greatly reduce fish populations by inhibiting
reproduction and survival of an aquatic species. These population reductions would reduce the size of
commercial harvest. These changes negatively affect subsistence anglers, commercial anglers and fish
sellers, and consumers offish and fish products. Chapter 2 of this report provides some detail on aquatic
life impacts from increased sediment concentrations in fresh and marine waters. Reducing sediment
discharges from construction sites is expected to enhance aquatic life habitat and thus contribute to
reproduction and survival of commercially harvested species, which in turn will lead to an increase in
producer and consumer surplus. EPA did not quantify market benefits from improved commercial harvest
resulting from this regulation due to data limitations. The nonmarket value of improved habitat for
commercial fish species is implicitly accounted for in the WTP for changes in environmental services
provided by surface waters affected by construction site discharges stemming from the regulation (see
Chapter 10 of this report for details).
Public and Private Property Ownership
> Property Values. Aesthetic degradation of land and water resources resulting from sediment
discharges (e.g., increased water turbidity) can reduce the market value of property and thus affect the
financial status of property owners. For example, a hedonic price study by Bejranonda et al. (1999)
found that "the rate of sediment inflow entering the lakes has a negative influence on lakeside
property rent" (p. 216). Sediment discharges also has a significant impact on stream morphology. For
example, higher coarse sediment load leads to an increase in width of the river bed and, as a result,
bank erosion (Wheeler et al. 2003). A 1993 study of Lake Erie's housing market found that "erosion-
prone lakeshore property will be discounted" (Kriesel 1993). Stabilization of stream banks leads to an
increase in the value of surrounding property (Streiner 1996). Therefore, the proposed regulation is
expected to enhance nearby property value in two ways: (1) by improving aesthetic quality of land
November 2008 4-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
and water resources (e.g., reduced turbidity) and (2) by improving stream morphology and stabilizing
river banks.
Due to data limitations, EPA did not estimate changes in market values of properties located near
water bodies expected to benefit from reduced sediment discharges from construction sites as a result
of this regulation. The nonmarket component (i.e., increased satisfaction with the property) is
implicitly accounted for in WTP for improvements in environmental services provided by surface
waters affected by construction site discharges (see Chapter 10 of this report for details).
> Flood Damages. Sedimentation can also increase the severity of property damages from flooding.
Sedimentation of river beds can reduce river capacity and result in higher flood levels and more
frequent flooding. Additionally, sediments carried by flood waters can damage property and can
be expensive to remove, particularly in developed areas (Clark et al. 1985). Clark et al. (1985)
estimated flooding damages attributable to sediment discharges to be $1.5 billion (2008$),
annually. This regulation is expected to reduce flooding damages by reducing sediment
discharges from construction sites. The Agency, however, was unable to estimate benefits from
reduced flood damages due to data limitations.
> Ditch Maintenance. Sediment in discharges from construction sites can clog ditches and culverts.
Such sedimentation can increase flooding, which may result in damages to bridges, roads,
farmland, and other structures in the nearby areas. Preventing these damages may require
increased maintenance efforts such as dredging (Clark 1985). The proposed regulation is
expected to reduce the costs of ditch maintenance by reducing the amount of sediment deposited
in ditches. This benefit category is not quantified due to data limitations.
4.2.2 Nonmarket Benefits
EPA expects that this regulation will reduce the sediment discharges to surface waters and will enhance or
protect aquatic ecosystems currently under stress. The decrease in sediment loading is expected to
improve protection of resident species, enhance the general health offish and invertebrate populations,
increase their propagation to waters currently impaired, and expand fisheries for both commercial and
recreational purposes. Improvements in water quality such as decrease in turbidity will also favor
increased recreational activities such as swimming, boating, and outings. Finally, the Agency expects that
the regulation will augment nonuse values (e.g., option, existence, and bequest values) of the affected
water resources.
The expected nonmarket benefits of water quality improvements fall into two broad categories:
nonmarket use benefits and nonmarket nonuse benefits. The following sections describe these benefit
categories and the nonmarket approaches to valuing these benefits in more detail.
Nonmarket Use Benefits
Direct use benefits include the value of improved environmental goods and services used and valued by
people (even if these services and goods are not traded in markets). Recreation is a typical example of
nonmarket direct use of water resources. Improved water quality from reducing sediment discharges from
construction sites may translate into two components of recreational benefits: (1) an increase in the value
of a recreational trip resulting from a more enjoyable experience, and (2) an increase in recreational
participation.
Adverse impacts of sediment discharges and, as a result, increased turbidity of surface waters on
recreational activities are summarized below.
November 2008 4-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
> Outings. Activities that take place near water such as hiking, jogging, picnicking, and wildlife
may be adversely affected by sediment and other pollutant discharges into the water. While these
activities do not involve contact with the water, murky and visually unpleasant water and odors
may greatly detract from the enjoyment gained through these activities. Decreases in fish
populations may cause a reduction in wildlife near the resource, affecting wildlife viewing.
Pollutants may negatively affect local flora and fauna, reducing the aesthetic appeal of the area
near the resource and negatively impacting wildlife viewing.
> Recreational Fishing. As noted in Chapter 2, Section 2.3.2, Aquatic Life Impacts of Sediment,
sediment and other discharges from construction sites can reduce fish populations by inhibiting
reproduction and survival of an aquatic species. This may lead to fewer and smaller fish, and a
reduction of the game fish population. In addition, sediments and other pollutions reduce the
aesthetics of the water body, which may reduce the anglers' utility of their fishing experience.
Additionally, sediment may affect recreational anglers by reducing the distance over which fish
can see lures, resulting in lower catch rates (Clark et al. 1985).
> Boating. Polluted water greatly reduces the aesthetic appeal of recreational boating activities.
Turbidity caused by sediment discharges may affect the safety of boating. Turbidity may obscure
underwater obstacles, making collisions more likely. Increased sediment concentrations may
create sandbars, increasing the chances of running aground. Clark et al. (1985) estimated that
turbidity (from all sources) may be responsible for as many as 200 boating fatalities and many
more injuries each year. At $7 million a life, the value of preventing these fatalities is $1.4 billion
per year. Even if reducing discharges from construction sites prevents a few fatalities and injuries
each year, the expected monetary value of preventing boating accidents can be significant.
> Swimming. Turbidity and other problems associated with sediment discharges may greatly reduce
a swimmer's aesthetic enjoyment of a resource. Additionally, turbidity may create safety hazards
for the swimmer by reducing the ability to see underwater hazards or increasing diving accidents
by impairing the ability to gauge water depth.
> Hunting. Similar to the effect on outings, discharged pollutants may greatly detract from the
hunters' aesthetic enjoyment of a water resource. Damage to flora and fauna may also cause a
reduction in the game population, reducing the number and quality of the game available.
The proposed regulation is expected to enhance recreational uses of water resources affected by sediment
discharges from construction sites, thereby resulting in welfare gain to recreational users of these
resources.
Nonmarket Nonuse Benefits
Even if no human activities (or uses) are affected by environmental changes caused by sediment and other
pollutants from construction sites, such environmental changes may still affect social welfare. For a
variety of reasons, including bequest, altruism, and existence motivations, individuals may value the
knowledge that water quality is being maintained, that ecosystems are being protected, and that
populations of individual species are healthy completely independent of their use value. It is often
difficult to quantify the relationship between changes in pollutant discharges and the improvements in
societal well-being that are not associated with current use of the affected ecosystem or habitat. That these
values exist, however, is indisputable, as evidenced, for example, by society's willingness to contribute to
organizations whose mission is to purchase and preserve lands or habitats to avert development (although
some portion of these donations may be motivated by use values). Notwithstanding challenges involved
November 2008 4-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
in estimation of nonuse values, there is a substantial literature devoted to such issues (Bateman et al.
2002). This literature provides insight into analysts' ability to estimate nonuse values within various types
of policy contexts, and for various types of resources.
Nonmarket Valuation Methods
It is frequently difficult to quantify and attach economic values to ecological benefits. The difficulty
results from imperfect understanding of the relationship between changes in sediment discharges from
construction sites and the specific ecological changes, from a lack of water quality monitoring data for
many locations, and from time lags between water quality changes and changes in species population and
composition. In addition, it is difficult to attach monetary values to these ecological changes because they
often do not occur in markets in which prices or costs are readily observed.
A variety of nonmarket valuation methods exist for estimating nonmarket use value, including both
revealed and stated preference methods (Freeman 2003). Where appropriate data are available or may be
collected, revealed preference methods can represent a preferred set of methods for estimating use values.
These methods use observed behavior to infer users' value for environmental goods and services.
Examples of revealed preference methods include travel cost, hedonic pricing, and random utility (or site
choice) models. Compared to nonuse values, use values are often considered relatively easy to estimate,
due to their relationship to observable behavior, the variety of revealed preference methods available, and
public familiarity with the recreational services provided by surface water bodies.
In contrast to direct use values, nonuse values are often considered more difficult to estimate. Stated
preference methods, or benefit transfer based on stated preference studies, are the generally accepted
techniques for estimating these values (USEPA 2000b; U.S. OMB 2003). Stated preference methods rely
on carefully designed surveys, which either (1) ask people about their WTP for particular ecological
improvements, such as increased protection of aquatic species or habitats with particular attributes, or (2)
ask people to choose between competing hypothetical "packages" of ecological improvements and
household cost (Bateman et al. 2003). In either case, values are estimated by statistical analysis of survey
responses.
Nonuse values may be more difficult to assess than use values for several reasons. First, nonuse values
are not associated with easily observable behavior. Second, nonuse values may be held by both users and
nonusers of a resource. Because nonusers may be less familiar with a resource, their values may be
different from the nonuse values for users of the same resource. Third, the development of a defensible
stated preference survey is often a time- and resource-intensive process. Fourth, even carefully designed
surveys may be subject to certain biases associated with the hypothetical nature of survey responses
(Mitchell and Carson 1989; Bateman et al. 2003). Finally, efforts to disaggregate total WTP into its use
and nonuse components have proved troublesome (Carson et al. 2000).
To evaluate nonmarket benefits of this regulation, EPA developed a benefit transfer approach based on a
meta-analysis of surface water valuation studies. Benefit transfer may be described as the "practice of
taking and adapting value estimates from past research ... and using them ... to assess the value of a
similar, but separate, change in a different resource" (Smith et al. 2002, p. 134). It involves adapting
research conducted for another purpose to estimate values within a particular policy context (Bergstrom
and De Civita 1999). Although the use of primary research to estimate values is generally preferred, the
realities of the policy process often dictate that benefit transfer is the only option for assessing certain
types of nonmarket values (Rosenberger and Johnston 2007). In the benefit transfer used for analyzing
benefits of the regulation, the meta-analysis is of WTP that incorporates both use and nonuse values.
November 2008 4-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Although the potential limitations and challenges of benefit transfer are well established (Desvousges et
al. 1998), the Agency also emphasizes that benefit transfers are a nearly universal component of benefit
cost analyses conducted by and for government agencies. As noted by Smith et al. (2002, p. 134), "nearly
all benefit cost analyses rely on benefit transfers, whether they acknowledge it or not." Benefits transfer
methods may be placed into three general categories: (1) transfer of an unadjusted fixed-value estimate
generated from a single study; (2) the use of expert judgment to aggregate or otherwise alter benefits to be
transferred from a site or set of sites, and (3) estimation of a value estimator model or benefits transfer
function, often based on data gathered from multiple sites (Bergstrom and De Civita 1999).
Given the generally unreliable performance of unadjusted single-site transfers, EPA used meta-analysis of
45 surface water valuation studies that use stated preference techniques to estimate a benefit function that
allows forecasting of total WTP estimates (including use and nonuse values) in a variety of policy
contexts, thereby providing more valid benefit estimates for transfer applications (Johnston et al. 2005;
USEPA 2000b). Because stated preference surveys describe a range of environmental services
corresponding to different levels of water quality (e.g., suitable for boating or swimmable), the total WTP
derived from meta-analysis implicitly accounts for a wide range of nonmarket use and nonuse values,
including recreation, aesthetic value of near-water properties, enhanced quality of source water for
drinking and household use, as well as existence and bequest values.
The technical details involved in the estimation of original meta-analyses are presented in Appendix B as
well as in sources such as Johnston et al. (2005; 2006), Bateman and Jones (2003), Shrestha et al. (2007),
Rosenberger and Phipps (2007), and USEPA (2004c). Chapter 10 of this report describes application of
the meta-analysis results for estimating benefits of water quality improvements resulting from the
regulation.
4.3 Summary of Effects
Ultimately, changes in sedimentation related to the proposed regulation affect humans through influences
on the production of goods and services that enter into individuals' utility functions. Table 4-1
summarizes some of the important ways in which sediment discharges from construction sites affects the
quantity and quality of services provided by the natural environment and how these services in turn
provide economic value to society through various use and nonuse related societal welfare mechanisms.3
EPA was not able to bring the same depth of analysis to all of the environmental services affected by the
regulation categories, however, because of imperfect understanding of the link between discharge
reductions and benefit categories, and how society values some of the benefit events. EPA was able to
quantify and monetize some benefits (including maintenance to navigable waterways, reservoir dredging,
drinking water treatment, and WTP for improvements in large river and streams), quantify but not
monetize other benefits (changes in TSS concentrations in coastal waters), and assess still other benefits
only qualitatively (e.g., commercial fishing, flood damages, and property values).
3 Underlying this table is a conceptual framework based on the valuation of ecosystem services (Daily 1997; Simpson and
Christensen 1997; van Houtven et al. 2005).
November 2008 4-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 4-1: Summary of Benefits from Reducing Sediment Discharges from Construction Sites
Activity
Recreation
> Outings
> Boating
> Swimming
> Fishing
Commercial Fishing and
Shellfishing
Property Ownership
Water Conveyance and
Supply
> Water conveyance
> Water storage
> Water treatment
Water Transportation
Water Use
> Industrial
> Municipal
> Agricultural
Knowledge (No Direct
Uses)
Environmental Service Flows
Potentially Affected by
Discharges from Construction
Sites
Aesthetics, water clarity, water
safety, degree of sedimentation,
weed growth, fish and shellfish
populations
Fish and shellfish populations
Aesthetics, safety of property from
flooding, property value
Turbidity, degree of sedimentation
Degree of sedimentation of
navigational channels, harbors, and
other waterways
Turbidity
Environmental health
Entities Affected by Environmental Changes
Firms
Governments
Individuals
Use & Nonuse Values
•/ Use & Nonuse Values
•/ •/ Use & Nonuse Values
^
Use Values
>/ >/ Use Values
>/ >/ Use Values
Nonuse
Methods Used in EPA's Analysis of Benefits from
the Regulation
Benefits Transfer
Qualitative Discussion
Benefits Transfer
Avoided cost to governments, including:
> Reservoir dredging
> Drinking water treatment
Avoided cost to firms and governments
Qualitative Discussion
Benefits Transfer
November 2008
4-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Pollutant Load Modeling
To estimate the environmental changes that would occur as a result of the regulation, EPA modeled
construction site pollutant discharges under baseline conditions and under each of several regulatory
scenarios. EPA used the Revised Universal Soil Loss Equation (RUSLE, pronounced "Russell") (Renard
1997) to estimate sediment generated from a series of model construction sites. By evaluating the particle
size distribution of soils present in areas surrounding 11 indicator cities, EPA estimated the fraction of
sediment discharges that would be removed under baseline conditions and under a series of regulatory
options. EPA scaled up the model site results to the state level based on the amount of construction
activity occurring in each state. EPA then allocated state-level loads to RF1 watersheds within the state
based on the percentage of development occurring in each RF1 watershed.
The procedures that EPA used in the RUSLE analysis and in the pollutant load reductions estimates are
described in Chapter 10 of EPA's Development Document for Proposed Effluent Guidelines and
Standards for the Construction and Development Category (USEPA 2008b). The results of this analysis
are also detailed in that document.
EPA is proposing to establish effluent limitations guidelines (ELGs) and new source performance
standards (NSPS) for stormwater discharges from the construction and development industry. These
guidelines and standards would require discharges from certain construction sites to meet a numeric
turbidity limit. The guidelines and standards would also require all construction sites currently required to
obtain a National Pollutant Discharge Elimination System (NPDES) to implement a variety of best
management practices (BMPs) designed to limit erosion and control sediment discharges from
construction sites. EPA evaluated three options in developing the proposed rule. These options are
described below:
> Option 1 would establish minimum sizing criteria for sediment basins used at construction sites
with 10 or more disturbed acres draining to one location. Under this option, permittees would be
required to install sediment basins that either provide 3,600 cubic feet per acre of runoff storage,
or are designed to store runoff from the local two-year, 24-hour storm event. This option also
includes requirements for implementing a variety or erosion and sediment controls on all
construction sites that are required to obtain a permit.
> Option 2 contains the same requirements as Option 1. In addition, construction sites of 30 or more
acres would be required to meet a numeric turbidity limit in stormwater discharges from the site.
The numeric turbidity standard would be applicable to stormwater discharges for all storm events
up to the local two-year, 24-hour event. The turbidity standard would only apply to construction
sites located in areas where the rainfall runoff erosivity factor (R-factor) as defined in RUSLE is
greater than or equal to 50 and if the soils on the site contain 10 percent or more by mass of soil
particles smaller than 2 microns in diameter.
> Option 3 contains the same requirements as Option 1. Option 3 also requires all sites with 10 or
more acres of disturbed land to meet a numeric turbidity standard. The turbidity standard would
apply to all sites, regardless of soil types or R-factor. The turbidity standard would apply to all
stormwater discharges for all storm events up to the local two-year, 24-hour event.
Table 5-1 presents a national summary of sediment loadings calculated by EPA for the baseline and three
policy options, as well as reductions under each policy option. Option 1 is expected to result in a
November 2008 5-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
reduction of 1.2 percent of construction sediment discharges nationally, while Option 2 and Option 3 are
expected to result in reductions of 46.0 percent and 87.8 percent, respectively. Total reductions under
Option 2 are 13.21 million tons of sediment, and 25.21 million tons under Option 3, compared to just
340,000 tons under Option 1.
Table 5-1: Baseline Construction Sediment Loading
Summary and Post-Compliance Reductions
Scenario
Baseline
Option 1
Option 2
Option 3
Sediment
Loading
(million tons)
28.70
28.36
15.49
3.49
Sediment
Reduction (million
tons)
-
0.34
13.21
25.21
% Reduction
from
Baseline
-
1.2%
46.0%
87.8%
Source: USEPA analysis
Table 5-2, Table 5-3, and Table 5-4 present estimates of the amount of sediment entering reaches from
construction related activity under each of these three options, by EPA region. Each table shows the
baseline amount of sediment entering reaches as well as post-compliance conditions and reductions for a
regulatory option. Under Option 1, Region 4 shows the largest absolute and percentage reduction in
sediment loadings, at 2.63 million tons and 2.3 percent, respectively. Reductions in other regions are
1 percent or less of baseline sediment loadings with the exception of Region 8, which shows a 1.5 percent
reduction. Option 2 shows more significant reductions in all regions, with Region 4 again showing the
largest reduction in terms of absolute mass of sediment prevented from entering surface waters (4.85
million tons), though Region 7 shows a larger percentage reduction, at nearly 60 percent of construction
sediment discharges prevented from entering waters. Option 3 reduces construction site sediment
discharges by 87 to 88 percent in all regions, with Region 4 showing the largest absolute reduction of
11.46 million tons of construction sediment discharges prevented from entering surface waters.
Table 5-2: Baseline Construction Sediment Loading Summary and Option 1
Reductions by EPA Region
Option 1
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
National
Total
Baseline Sediment
Loading
(million tons)
0.11
0.52
2.63
13.01
2.78
3.71
4.64
0.35
0.11
0.83
28.70
Sediment
Loading (million
tons)
0.11
0.51
2.63
12.72
2.76
3.71
4.63
0.35
0.11
0.83
28.36
Sediment
Reduction
(million tons)
0.00
0.01
0.00
0.30
0.02
0.00
0.00
0.01
0.00
0.00
0.34
% Reduction
from Baseline
0.5%
1.0%
0.0%
2.3%
0.9%
0.0%
0.1%
1.5%
0.0%
0.0%
1.2%
Source: USEPA analysis
November 2008
5-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 5-3: Baseline Construction Sediment Loading Summary and Option 2
Reductions by EPA Region
Option 2
EPA
Region
1
2
o
5
4
5
6
7
8
9
10
National
Total
Baseline Sediment
Loading (million
tons)
0.11
0.52
2.63
13.01
2.78
3.71
4.64
0.35
0.11
0.83
28.70
Sediment
Loading (million
tons)
0.08
0.27
1.06
8.16
1.39
1.66
1.87
0.27
0.10
0.62
15.49
Sediment
Reduction
(million tons)
0.03
0.25
1.57
4.85
1.39
2.06
2.77
0.08
0.01
0.21
13.21
% Reduction
from Baseline
29.2%
47.6%
59.6%
37.3%
50.0%
55.4%
59.8%
22.3%
7.3%
25.2%
46.0%
Source: USEPA analysis
Table 5-4: Baseline Construction Sediment Loading Summary and Option 3
Reductions by EPA Region
Option 3
EPA
Regions
1
2
3
4
5
6
7
8
9
10
National
Total
Baseline Sediment
Loading (million
tons)
0.11
0.52
2.63
13.01
2.78
3.71
4.64
0.35
0.11
0.83
28.70
Sediment
Loading (million
tons)
0.01
0.06
0.32
1.56
0.35
0.47
0.57
0.04
0.01
0.10
3.49
Sediment
Reduction
(million tons)
0.09
0.46
2.31
11.46
2.43
3.24
4.07
0.31
0.09
0.73
25.21
% Reduction
from Baseline
87.8%
87.9%
88.0%
88.0%
87.5%
87.3%
87.7%
88.0%
88.0%
88.3%
87.8%
Source: USEPA analysis
November 2008
5-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6 Water Quality Modeling
6.1 SPARROW Model Documentation4
6.1.1 General Overview
SPARROW (SPAtially Referenced Regressions On Watershed attributes) is a watershed modeling
technique for estimating contaminant source contributions and transport in surface waters. SPARROW
employs a statistically estimated nonlinear regression model with contaminant supply and process
components, including surface-water flow paths, non-conservative transport processes, and mass-balance
constraints. Regression equation parameters are estimated by correlating stream water-quality records
with GIS (Geographic Information System) data on pollutant sources (point and nonpoint) and climatic
and hydrogeologic properties (e.g., precipitation, topography, vegetation, soils, and water routing) that
affect contaminant transport. The SPARROW parameter-estimating procedure also provides measures of
uncertainty in water-quality predictions. SPARROW model infrastructure consists of a detailed watershed
- stream reach network to which all monitoring stations and watershed properties are spatially referenced.
Digital elevation models (DEMs) are used to delineate watershed boundaries and to identify overland
flowpaths. The spatially distributed model structure allows separate statistical estimation of land- and
water-related rates of pollutant delivery from sources to streams, and transport of pollutants to
downstream locations within the stream network (reaches, reservoirs and estuaries). Mechanistic
separation of terrestrial and aquatic features of large watersheds and improved parameter estimation
techniques represent significant advances in water-quality modeling of important contaminant sources
and watershed properties that control transport over large spatial scales. SPARROW has been applied in
the analysis of suspended sediment, surface-water nitrogen and phosphorus nutrients, pesticides, organic
carbon, and fecal bacteria, and is potentially applicable to other measures of water quality. The
SPARROW software is written in Statistical Analysis System Interactive Matrix Language (SAS IML).
The remainder of this chapter provides a brief description of SPARROW modeling concepts, model
specification, data sources, and estimation results. Appendix B of this report provides more detail on the
nature of the general SPARROW methodology. Appendix C provides more detail on the suspended
sediment SPARROW model developed by USGS (Schwarz 2008a).
6.1.2 Modeling Concept
The broad objective of SPARROW modeling is to establish a mathematical relationship between water-
quality measurements made at a network of monitoring stations and attributes of the contributing
watersheds. The model may be used to satisfy a variety of water-quality information objectives. One
objective is to describe past or present water-quality conditions for a state or region on the basis of
monitoring data. A second objective is to identify and quantify the sources of pollution that give rise to
in-stream water-quality conditions. A third objective is water quality simulation: the ability to portray
counterfactual conditions for specified inputs. Water-quality simulations can depict the in-stream effects
of changes in contaminant sources associated with alternative potential pollution-control strategies,
providing an important step in the analysis of benefits associated with specific Best Management
Discussion adapted from Schwarz et al., USGS (2006).
November 2008 6-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Practices (BMP) or regulatory approaches. For the analysis described in this document, SPARROW is
used to evaluate the effectiveness of a range of regulatory options for controlling and mitigating
sediments associated with construction activities, with explicit focus on the impacts of such options on in-
stream water quality as measured by Total Suspended Solids (TSS). Determining regulatory option
impacts on national water quality through physical experimentation rather than through modeling would
be extremely difficult. A fourth objective of SPARROW modeling is to test one or more hypotheses about
the nature and importance of factors and processes that may have influenced water quality at the locations
where samples were collected.
SPARROW models are expressed in the form of a mass balance. Because the dependent variable in
SPARROW models (the mass of contaminant that passes a specific stream location per unit time, also
referred to herein as loading) is linearly related to all sources of contaminant mass in the model, all
accounting rules relating to the conservation of mass apply. Mass accounting in SPARROW models is
also supported by the explicit spatial structure defined by the stream network. The imposition of mass
balance greatly improves the interpretability of model coefficients. For example, assuming mass balance,
the coefficient associated with the reach time-of-travel variable is interpreted as a first-order decay rate.
Mass balance provides a basis for contaminant loading accounting, whereby loading can be allocated to
its various sources, both spatially and topically. Spatial variability is modeled as a function of natural and
human-related properties of watersheds that influence the supply and transport of contaminants. Estimates
of long-term mean loading are developed from water-quality and streamflow monitoring data that are
regularly collected at fixed locations on streams and rivers.
6.1.3 Model Infrastructure
A flow diagram is provided in Figure 6-1 to illustrate the functional linkages between the major spatial
components of SPARROW models. Loading estimates at monitoring stations are derived from station-
specific models that relate contaminant concentrations from individual water-quality samples to
continuous records of streamflow and time. The stream reach, inclusive of its incremental contributing
drainage basin, is the most elemental spatial unit of the SPARROW model infrastructure. Explanatory
data (e.g., climate, topography, land use) are frequently compiled according to geographic units that are
not coincident with the drainage basin boundaries of river reaches. These data may be collected at
different spatial scales and according to spatial units that reflect political boundaries (e.g., counties) or
other non-hydrologic features of the landscape.
November 2008 6-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-1: Schematic of the Major SPARROW Model Components
Industrial / Municipal
Point Sources
Aquatic Transport
Streams
Reservoirs
In-Stream Flux Prediction
Calibration minimizes differences
between predicted and calculated
mean-annual loads at the
monitoring stations
SPARROW
Model Components
Water-Quality and
Flow Data
Periodic measurements
at monitoring stations
Monitoring
Station Flux
Estimation
Rating Curve Model
of Pollutant Flux
Station calibration to
monitoring dafa
Mean-Annual Pollutant
Flux Estimation
Evaluation of Model
Parameters and Predictions
Source: Modified from Alexander, Elliott, et al. (2002)
The estimation of a SPARROW model requires estimates of long-term mean loading from a spatially
distributed set of monitoring stations, having sufficiently long periods of record, inclusive of a wide range
of spatial scales and expressing considerable variation in predictor variable conditions. Long-term mean
loading is obtained by relating infrequently collected water-quality samples to continuous measurements
of streamflow and functions of time. In order to have consistency across stations having different periods
of record, it is necessary to detrend the loading estimates to a common base year prior to computing the
long-term mean.
A vector- or raster-based digital representation of the stream and river network topology is the most
fundamental component of the spatial infrastructure that supports the SPARROW model (Figure 6-2).
The node topology is defined according to an upstream (from-) and downstream (to-) node attribute table.
November 2008
6-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-2: Schematic Illustrating a Vector Stream Reach Network with Node Topology and
Water/Contaminant Reach-Node Routing Table
The reach-type indicator has possible values of "0" (Stream Reach), "1"(lmpoundment Reach), and "2"
Outlet Reach for Impoundment.
7.-? 72
.13
12
77
1
Reservoir
"shorelines
Divergent
Reach
Reach Node
'Stream Reach
Reach-Node Attribute Table
Reach Up-
Number stream
Node
Down- Diversion Reach-Type
stream Fraction Indicator
Node
13
10
12
11
9
8
7
5
6
4
2
3
1
13
10
12
11
9
8
7
6
6
5
4
3
2
10
9
11
9
8
4
6
5 {
5 (
4
2
2
1
.0
.0
.0
.0
.0
.0
.0
17
13
.0
.0
.0
.0
0
1
0
1
2
0
0
0
0
0
0
0
0
SPARROW separately estimates the sediment and contaminant attenuation in reservoirs and lakes. The
assessment of pollutant loadings to coastal estuaries requires an expanded reach network that includes
shoreline features and the identification of reaches that terminate at estuaries. Shoreline reaches are used
to define coastal drainage areas - areas that discharge runoff directly to the estuary without transport
through a stream reach. The SPARROW node and routing architecture also can fully support the
modeling of contaminant transport along "off-reach" (landscape) flow paths according to flow directions
defined by landscape topography as reflected, for example, in DEMs.
The explanatory variables evaluated in SPARROW reflect current knowledge of natural and human-
related sources and the physical, chemical, or biological properties of the terrestrial and aquatic
ecosystems that affect supply and transport of contaminants in watersheds. Point- and nonpoint-source
variables may include direct measures of the supply of contaminant mass to the landscape and streams
and reservoirs (e.g., municipal and industrial wastewater discharge, fertilizer application), or they may
serve as surrogate indicators of the contaminant mass supplied by point and diffuse sources in watersheds,
such as land-use/land-cover data or census data on human and livestock populations. Climatic and
landscape properties that affect contaminant transport may include measures of water-balance terms (e.g.,
solar radiation, precipitation, evaporation, evapotranspiration), soil characteristics (e.g., permeability,
moisture content), water-flow path properties (e.g., slope, hydraulic roughness, topographic index), or
management practices and activities (e.g., tile drains, conservation tillage, BMPs).
6.1.4 Model Specification
The specification of a SPARROW model consists of identifying the explanatory variables and functional
forms. Conceptually, the contaminant load or loading leaving a reach is the sum of two components. The
first component is the load generated within upstream reaches that is delivered to the reach via the stream
network. Losses of loading from the stream network may occur at points where flow is diverted, and
November 2008
6-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
loading will generally be attenuated by stream or reservoir processes. The second component consists of
source loading that is generated within the reach's incremental watershed and delivered to the stream
network within the reach. A number of source-dependent processes affect the amount of source loading
reaching the stream network and transported to the reach's downstream outlet node. For loading
originating on the landscape, the processes affecting delivery to the stream network are called land-to-
water delivery processes, and may include both surface and sub-surface elements. Land-to-water delivery
processes determine the amount of contaminant generated within an incremental drainage area delivered
to the area's corresponding reach.
Large digital spatial data sets are now readily available, and advances in digital topographic and stream
network data [e.g., National Hydrography Dataset (NHD), National Land Cover Database (NLCD)] have
made spatially distributed modeling more feasible. Inclusion in the SPARROW model of detailed
information on the geographic locations where contaminants are released in watersheds has particular
relevance to the policy and management applications of the model. The source terms used in the model
can be generally classified as intensive and extensive measures of contaminant mass. Intensive measures
are direct measures of pollutant mass, such as fertilizer application, livestock waste, atmospheric
deposition, or sewage-effluent loadings. The associated source parameter is a dimensionless coefficient
that, together with standardized expressions of the land-to-water delivery factor, describes the proportion
or fraction of the source input that is delivered to streams. Extensive measures of contaminant mass are
surrogate indicators and include measures of watershed properties such as specific land-use area and
sewered population, considered to be proportional to the actual mass loadings generated by a particular
type of a contaminant source. The associated model coefficients are expressed as the contaminant mass
generated per unit of the source type (e.g., kilograms kilometer"2 year"1; kilograms person"1 year"1).
Combined with the land-to-water delivery factor, the standardized source coefficient indicates the mean
quantity of contaminant mass per unit of the surrogate source measure that is delivered to streams. For
land-use terms, the standardized coefficient gives what is frequently cited as an export coefficient.
Landscape variables in SPARROW describe properties of the landscape that relate to climatic, natural- or
human-related terrestrial processes affecting contaminant transport. The model structure allows tests of
hypotheses about the influence of specific features of the landscape on contaminant transport. Landscape
variables may include water-balance terms (e.g., precipitation, evapotranspiration) related to climate and
vegetation, soil properties (e.g., organic content, permeability, moisture content), topographic water flow-
paths variables (e.g., overland flow, topographic index, and slope), or management practices and
activities, including tile drainage, conservation tillage practices, and BMPs related to stream riparian
properties. Particular types of land-use classes, such as wetlands or impervious cover, may also be
potentially used to describe transport properties of the landscape.
Stream attenuation processes that act on contaminant loading as it travels along stream reaches are
frequently modeled as first-order reaction rate processes. A first-order decay process implies that the rate
of removal of the contaminant from the water column per unit of time is proportional to the concentration
or mass that is present in a given volume of water (a zero-order process corresponds to a constant rate of
removal per unit of time). Accordingly, in basic forms of the SPARROW model, the fraction of the
contaminant mass originating from the upstream node and transported along a reach to its downstream
node is estimated as a function of the mean water time of travel. Attenuation processes that act on
contaminant mass as it travels through a lake or reservoir are often modeled as a net removal process,
with the loss coefficient expressed as either a first-order reaction rate or a mass-transfer coefficient (also
referred to as an apparent settling velocity). These mass-balance models typically assume steady-state and
uniformly mixed conditions in the water body.
November 2008 6-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.1.5 Model Estimation
The SPARROW model equation is a nonlinear function of its parameters, and the model must be
estimated using nonlinear techniques. SPARROW utilizes a nonlinear weighted least squares (NWLS)
estimation method, implemented in SAS. Model parameters are evaluated for statistical significance and
the quantification of uncertainty. The key parameter statistics include the estimated mean coefficient
values, estimated variance of these coefficient estimators, and measures of statistical significance based
on the t statistics. Parameters are also evaluated for physical interpretability, to test hypotheses about the
importance of different contaminant sources and the hydrologic and biogeochemical processes that are
represented in the model. The interpretability of the parameters and their relation to specific processes are
enhanced in SPARROW by the use of a mass balance, mechanistic structure that explicitly separates the
terrestrial and aquatic properties of watersheds and accounts for nonlinear interactions among watershed
properties. The sign of SPARROW model coefficients can be evaluated to determine the direction of
influence of any explanatory variable on in-stream estimates of mean-annual loading. Coefficients
associated with source inputs expressed in areal units describe the mass per unit area delivered to streams
from these land areas. Other source coefficients that are expressed in dimensionless units provide a
measure of the fraction of the contaminant that is delivered from each source to streams, rivers, and
reservoirs.
6.2 SPARROW Sediment Model5
The following describes preliminary results from a SPARROW suspended sediment model developed for
the coterminous United States and described in Schwarz (2008a), which is reprinted in full in Appendix C
of this document. The analysis is based on loading estimates compiled from more than 1,800 long-term
monitoring stations operated by the U.S. Geological Survey (USGS) during the period 1975-2007. The
SPARROW model is structured on the Reach File 1 (RF1) stream network, consisting of approximately
62,000 reach segments. The reach network has been modified to include more than 4,000 reservoirs, an
important landscape feature affecting the delivery of suspended sediment. The model identifies six
sources of sediment, including the stream channel and five classes of land use: urban, forested, Federal
nonforested, agricultural and other, and noninundated land. The delivery of sediment from landform
sources to RF1 streams is mediated by soil permeability, erodibility, slope, and rainfall; streamflow is
found to affect the amount of sediment mobilized from the stream channel. The results show agricultural
land and the stream channel to be major sources of sediment loading. Per unit area, Federal nonforested
and urban lands are the largest landform sediment sources. Reservoirs are identified as major sites for
sediment attenuation.
6.2.1 Sediment Model Data Sources
The spatial framework of the SPARROW sediment model is the vector-based 1:500,000 scale River RF1
hydrography, originally developed by the EPA (1996) and enhanced to include areal hydraulic load
information for selected reservoirs, shoreline reaches, and reach catchment areas derived from the USGS
Hydro IK DEM (USGS 2006a). The enhanced network, consisting of 62,776 reach segments, including
shoreline reaches, 61,214 delineated reach catchments, and 2,171 individual reservoirs, has been used to
support numerous national SPARROW modeling efforts for the coterminous United States. The RF1
reach network for the current SPARROW sediment model was further enhanced by inclusion of areal
Adapted from Schwarz (2008a).
November 2008 6-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
hydraulic load information for approximately 2,000 additional large reservoirs from the National
Inventory of Dams (NID, USAGE 1996). The dependent variable in the SPARROW sediment model is
given by long-term mean sediment loading, detrended to the base year 1992. In-stream sediment
concentrations and stream discharge measurements over the water year (WY) period 1975-2007, have
been obtained from the National Stream Quality Accounting Network (USGS 2006b), the USGS
NAWQA Program (USGS, 2001), and the USGS National Water Information System (NWIS; USGS
2008), a database encompassing USGS water quality monitoring stations and water quality monitoring
activities done in cooperation with State government. The sediment model is estimated using 1,828
monitoring stations on the RF1 stream network (Figure 6-3).
While the RF1 network provides reasonably comprehensive national coverage of major rivers, streams
and other surface water bodies, coverage is limited in certain important respects. First, RF1 network
coverage is limited to the coterminous United States, thus excluding Alaska and Hawaii. In addition,
while RF1 l:500,000-scale network reaches have associated data or estimates of stream discharge and
velocity that are required to specify the SPARROW model, the network excludes the majority of the
nation's total stream mileage, and smaller streams in particular. The linear coverage of the RF1 network is
approximately 700,000 miles (USEPA 2007e). By contrast, coverage of the USGS National Hydrographic
Dataset (NHD) at 1:24,000 - 1:100,000 scale is currently over 7 million miles (Smiley 2007). Non-
homogeneity of coverage may lead to an under-estimation of the impacts of construction-related sediment
on smaller stream reaches. As construction activities may be concentrated along lower-order streams not
included in the RF1 network, the relative share of total sediments contributed by construction activities in
these reaches may be high during active construction. By contrast, the specific impacts of construction
activities may diminish in magnitude relative to contributions from spatially extensive and diffuse land
uses, including agriculture, at the level of RF1 reaches.
Data on land cover and land use have been developed from the 2001 USGS National Land Cover Data Set
Retrofit Change Product (Multi-Resolution Land Characteristics Consortium (MRLC), 2001), derived
from Landsat TM/ETM remotely sensed imagery at 30-meter resolution and classified according to the
eight Anderson Level I categories. For use within SPARROW, data were transformed to 1 km2 cells
within a Lambert map projection as consistent with Hydro IK. The 1 km2 cells were then resolved to
catchments associated with specific RF1 reaches. For model estimation, land use was given by the 1992
values of the 2001 NLCD Retrofit Change Product. The 2001 values of the Retrofit Change Product were
used to simulate water quality conditions for 2001. Federal range and barren land was included in the
model separately from private land. Federal land extent, taken from the Federal Land coverage of the
National Atlas (USGS 2003) and transformed to 1 km2 cells, was apportioned using the 1 km2
transformation of the 1992 NLCD Change Product for Anderson Level 1 range and barren land classes.
For model simulation of 2001 conditions, federal range and barren land were similarly estimated using the
land use estimates from the 2001 NLCD Retrofit Change Product.
November 2008 6-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-3: Location of 1,828 Water Quality Monitoring Stations Used in the SPARROW Sediment Model,
in Relation to the Reach File 1 (RF1) Reach Network
Variables determining the delivery of contaminants from the land to RF1 streams include soil erodibility
(RUSLE k-factor), soil permeability, mean slope, and precipitation (RUSLE r-factor). Slope, soil
erodibility, and permeability were obtained from the State Soil Geographic (STATSGO) database
(USDA, 1994), converted to a 1-km2 grid and averaged over RF1 catchments. The RUSLE rainfall factor
was derived by interpolating a digitized national map of rainfall factor isoline contours, creating a
continuous 1 km2 grid surface, which was subsequently averaged over individual RF1 catchments.
6.2.2 Preliminary Model Estimation Results
The nonlinear least squares estimation results of a preliminary version of the SPARROW suspended
sediment model are given in Table 6-1. The preliminary model includes six source terms: five measured
by area of specific land use (expressed in km2), and the sixth by the length of stream channel. The five
land use sources are urban land, forested land, federal nonforested land, agricultural land, and other land.
The federal land class excludes federal forested land, which is incorporated in the forested land class.
Agricultural land includes cropland, pasture land and orchards. Other land consists of nonfederal range
and barren land. Among all the land classes, only wetlands and land covered by water, ice or snow are
excluded as a potential source. The source described as "streambed" relates to stream channels as a direct
source of sediment, and is measured in terms of stream length (expressed in meters). The model specifies
two in-stream sediment attenuation processes: attenuation in streams, distinguished by three streamflow
classes (below 500 cfs, 500-1,000 cfs, and greater than 1,000 cfs); and reservoir attenuation, specified as a
function of areal hydraulic load.
November 2008
6-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-1 : Estimation Results for the SPARROW Suspended Sediment Model
Parameter
Source Coefficients
Urban land
Forested land
Federal nonforested land
Agricultural land
Other land
Streambed (reach length)
Land to Water Delivery
Slope
Soil permeability
R-factor
K-factor
Flow [< 500 cfs] (Reach)
Flow [> 500 cfs] (Reach)
Stream Attenuation Factors
Travel time (Q < 500 cfs)
Travel time (500 < Q < 1,000 cfs)
Travel time (Q > 1,000 cfs)
Reservoir settling velocity
Number of Observations
RMSE
R-square
Units
kg/km2/yr
kg/km2/yr
kg/km2/yr
kg/km2/yr
kg/km2/yr
kg/m/yr
Factors
day'1
day'1
day'1
m/yr
1,828
1.414
0.711
Estimate
47,130
634
64,344
18,047
11,343
28.80
0.804
-0.778
0.821
1.292
0.154
0.721
-0.007
-0.233
0.009
36.49
Standard Error
9,925
898
12,411
3,623
3,186
6.40
0.087
0.094
0.081
0.279
0.100
0.354
0.016
0.057
0.047
5.552
p-value
0.000
0.480
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.125
0.042
0.673
0.000
0.854
0.000
Source: Schwarz 2008a
With the exception of forested land, all of the source variables are highly statistically significant. The
largest intrinsic sediment yield is associated with federal range and barren land; urban land has the second
highest intrinsic yield. Stream channels are also a statistically significant source of sediment. Land-to-
water delivery for land sources is strongly mediated by the four delivery variables: soil permeability, soil
erodibility, slope, and rainfall. With the exception of soil permeability, an increase in these factors results
in an increase in the share of sediment delivered to streams. Permeable soils reduce the delivery of
sediment, presumably because more runoff infiltrates, leaving less overland flow to carry sediment.
Greater streamflow causes an increase in the amount of sediment generated from stream channels, with
the largest effect associated with streams with flow greater than 500 cfs. Reservoir retention is statistically
significant and indicates sediment settles at a mean velocity of 36 meters per year. The preliminary model
indicates medium sized streams (flow 500-1,000 cfs) have a statistically significant negative rate of in-
stream attenuation, indicating that medium streams are a source of sediment, in addition to the stream
channel source identified above. The negative rate of in-stream attenuation implies that sediment loading
is proportionally enhanced as it passes through medium sized streams. This enhancement process is
inconsistent with mass balance and is yet to be explained. In-stream attenuation in small and large streams
is not significantly statistically different from zero. Thus, the preliminary model does not find evidence
for sediment loss in streams.
6.2.3 Model Simulation
The estimated SPARROW suspended sediment model for base water year 1992 is used to simulate water-
quality conditions for 2001, with and without proposed changes in the regulation of construction activity.
The simulation of suspended sediment loading for water year 2001 without changes in regulation is
obtained with all land use-related source variables set to 2001 values based on the NLCD Retrofit Change
Product. Flow weighted concentration is estimated by dividing simulated loading estimates by mean
streamflow over the period WY 1975-2006. Although the SPARROW model does not explicitly include a
November 2008
6-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
term for construction loading, such loading is implicitly accounted for in the urban land component of the
model. Therefore, the 2001 pre-compliance loading from construction (the "base-case" scenario loading)
is incorporated in the 2001 loading attributed to urban land that is obtained by evaluating the urban land
variable in the SPARROW model using the 2001 NLCD Change Product value.
The absence in the SPARROW model of an explicit term for construction loading requires the
development of an indirect method for assessing changes in sediment loading arising from different
construction industry regulation scenarios. Suspended sediment loading under alternative regulation
scenarios has been estimated by EPA using a variation of the USLE. The USLE method determines the
amount of soil that is mobilized and delivered, under a regulatory scenario, to the edge of a construction
site. To evaluate the impact that changes in these loadings have on RF1 stream sediment loading and
concentration, it is first necessary to assess the rate at which "edge of site" loads are subsequently
delivered to RF1 streams. We use the estimated rate of delivery from agricultural land (explicitly included
in the model) that can be factored into a mobilization ("edge of site") delivery component and a stream-
delivery component. The method uses edge-of-site measurements of agricultural sediment erosion from
the 1992 National Resources Inventory (NRI) to isolate the stream-delivery component from the overall
rate of delivery from agricultural land and applies this component to the change in construction loading to
determine the change in loading to RF1 streams. Thus, the approach assumes that the delivery of sediment
from the edge of a site to an RF1 stream is the same for both construction and agriculture activities;
although the mobilization and delivery of sediment to the edge of the site between these activities is
allowed to differ.
6.3 Water Quality Modeling Results
This section presents the results of water quality modeling using the SPARROW model for sediment.
SPARROW simulations generate estimates of water sediment content as Suspended Sediment
Concentration (SSC), as consistent with water quality gauging records used to estimate SPARROW
sediment models. By contrast, ambient sediment concentrations measured as Total Suspended Solids
(TSS) are required in the calculation of the water quality index (details of the water quality index can be
found in Chapter 10). Although SSC and TSS are interpreted in a similar manner as measures of the
concentration of suspended solid-phase material in surface water bodies, Gray, et al. (2000) have
determined that differences in laboratory protocols result in systematic differences between SSC and TSS
when measuring a given ambient concentration of suspended sediment. SSC measurements are
systematically higher than TSS, particularly when the sand component of total sediments exceeds roughly
25 percent of total (dry) sediment mass. Therefore, for the analysis described in this document,
SPARROW estimates of SSC have been divided by a factor of 1.3 to obtain the corresponding TSS
values.
Construction activity discharges impact the majority of RF1 reaches in the coterminous United States.
During the nine-year period from 1992 through 2001, approximately 72 percent of RF1 reach watersheds
contained some level of construction. Construction discharges differ from other types of industrial
discharges in this respect. Other types of industrial dischargers tend to be fewer in number and less widely
distributed across the United States.
The intensity of construction activity, however, varies widely among individual RF1 watersheds. Figure
6-4 illustrates the highly uneven distribution of construction acres among RF1 watersheds during the
1992-2001 time period. Figure 6-5 shows the cumulative distribution of acres of construction by RF1
watersheds in decreasing order of construction acres by watershed. As the two figures show, relatively
November 2008 6-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
few watersheds account for a large fraction of acres of construction nationally. In Figure 6-5, for
example, the 4 percent of RF1 watersheds containing the largest number of construction acres per
watershed encompassed, as a group, almost 3 million construction acres, or 56 percent of all construction
acreage for the 1992-2001 time period.
Total Acres of Construction per Percentile Group
(1992-2001)
Figure 6-4: Plot of Construction Acres by RF1 Watershed Percentile Group (1992-2001)
1 snn nnn
1 snn nnn
1 4nn nnn
1 9nn nnn
1 nnn nnn
snn nnn
snn nnn J
400,000 -j
200,000 -j
O-l
1
M
We
1
L
LI,,,,,,,
IIIIIIIIIIIIIIIIIIIIMluiii.. 1 1 1 1 1 1 1 i-
% 6% 1 1 % 1 6% 21 % 26% 31 % 36% 41 % 46% 51 % 56% 61 % 66% 71 % 76% 81 % 86% 91 % 96%
Dst construction acres per Least construction acres per
itershed watershed
RF1 Watershed Percentile Rank Based on Total Construction Acres
November 2008
6-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-5: Cumulative Plot of Construction Acres by RF1 Watershed (1992-2001)
1% 6% 11% 16% 21% 26% 31% 36% 41% 46% 51% 56% 61% 66% 71% 76% 81% 86% 91% 96%
Most construction acres Least construction acres
per watershed per watershed
RF1 Watershed Percentile Rank Based on Total Construction Acres
Because construction sites are distributed so unevenly among watersheds, this chapter summarizes water
quality information in two ways. The first set of tables included in this chapter presents information for
the relatively small number of RF1 watersheds that contain most construction acreage for the 1992-2001
time period. The second set of tables presents information for all RF1 reaches directly receiving
discharges from construction sites, regardless of the level of construction activity in their watersheds.
An important factor to consider when examining the data summarized in the tables below is that episodic
precipitation events are the primary cause of construction site TSS discharges to surface waters. Most
TSS discharge, therefore, takes place during or shortly after precipitation events. Once a precipitation
event ceases, discharges from construction sites generally cease within a short time period. The estimates
presented in the tables below, however, are not intended to reflect surface water TSS concentrations
associated with individual, episodic storm events. They instead represent average concentrations
estimated to exist in RF1 reaches over multi-year time periods. EPA uses these long-term estimates
because the data and modeling resources currently available to EPA do not permit a finer level of time
resolution.
EPA expects that, in general, surface water TSS concentrations in the short time periods following storm
events would be higher than the concentrations presented in the tables below. Sediment concentration
reductions associated with the three proposed regulatory options would be higher, as well. During periods
when no precipitation event is available to cause a sediment discharge from a construction site, surface
water sediment concentrations would generally be lower.
November 2008
6-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
A second important factor to consider when examining the data in the tables below is that construction
site sediment discharges also have strong geospatial variability. Geospatial differences occur as a result of
variation in soil type, topography, distance from construction site to surface water, receiving surface
water characteristics, and weather patterns, among other factors. These differences are reflected to a
certain extent by the differences in surface water TSS concentrations among different groups of reaches,
as reflected by reaches having progressively greater construction acreage within their watershed
(presented in Table 6-2 through Table 6-6) and among the various EPA Regions (presented in Table 6-4,
Table 6-5, and Table 6-6). In general, however, the data presented below are averaged over large
geographic regions because of limitations in the data and modeling resources currently available to EPA.
Despite the limited detail they provide on temporal and geospatial variations in TSS concentration, the
summaries are nevertheless useful for interpreting broad-scale changes in water quality as a result of the
proposed regulatory options.
When summarizing TSS concentrations, EPA calculated statistics based on reach-length weighted TSS
concentrations to account for the wide variation among reaches, as noted in the tables. EPA believes that
this adjustment provides a more representative characterization of the distribution of TSS concentration
by giving more weight to the longer reaches. Additionally, EPA has generally excluded from the
distribution TSS concentrations estimated on reaches that have baseline TSS concentrations above the
95th percentile (6,049.7 mg/L), since these reaches could potentially be considered outliers.
6.3.1 Estimated Changes in TSS Concentration in Watersheds with Most Developed
Acres (1992-2001)
Table 6-2 summarizes the characteristics of different groups of modeled watersheds in terms of the relative
amount of development they contain. The table provides information for three particular subgroups that
represent the "top" watersheds in terms of the amount of developed land during the period of 1992-2001:
top 1 percent, top 10 percent, and top 25 percent. As discussed above, development is not evenly
distributed during the nine-year period under examination and each of the watershed subsets therefore
account for a relatively disproportionate share of construction acres. For example, the top 10 percent of
watersheds account for nearly three-quarters of the increase in developed land in the coterminous United
States, while the top 25 percent account for nearly all developed land (91.8 percent). These watersheds
are also distributed unevenly across EPA regions both in terms of total river miles and the number of
water body reaches, as shown in Table 6-3. For example, approximately 22 percent of river miles within
the top 1 percent are found in Region 4 even though the region accounts for approximately 15 percent of
overall RF1 river miles.
These differences are reflected in TSS concentration estimates shown in Table 6-4, Table 6-5, and Table
6-6 for each of the three top groups, respectively, and across the three regulatory options (see Chapter 5
for option descriptions). As shown in the tables, the top 1 percent of watersheds in terms of construction
acres tend to have the highest sediment concentrations, both under current conditions and under the three
options. While the three top groups result in a similar fraction of miles showing improvements in TSS
concentrations across the three groups, reaches within the top 1 percent of watersheds exhibit a higher
magnitude of TSS concentration reductions than reaches in watersheds that contain comparatively fewer
construction acres.
November 2008 6-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-2: Top Percent of Reaches by Increase in Developed Land Area (1992-2001)
Percent of RF1 Watersheds
Top l%ofRFl Watersheds
Top 10%ofRFl Watersheds
Top 25% of RF1 Watersheds
Coterminous U.S.
(100% of Reaches)
Number of RF1
Watersheds1
612
6,112
15,310
61,121
% of Total
Increase in
Coterminous U.S.
Developed Land
31.5%
74.9%
91.8%
100.0%
Total Increase in
Developed Land
Area (1992-2001)
(million acres)
1.7
4.0
4.9
5.3
Average Increase
in Developed
Land Area per
RF1 Watersheds
(1992-2001)
(acres)
2,727.5
648.5
317.6
86.6
Total Land Area
(million acres)
94.5
484.8
870.4
1,885.3
%of
Coterminous U.S.
Land Area
5.0%
25.7%
46.2%
100.0%
1 RF1 watersheds included in count are those for which both NLCD and SPARROW data are available.
Top 1% of RF1 watersheds had 1,304 acres or greater of developed land area (1992-2001); Top 10% of RF1 watersheds had 178 acres or greater of developed land area (1992-2001); Top 25% of
watersheds had 52 acres or greater of developed land area (1992-2001).
Table 6-3: Distribution of Top RF1 Watersheds by Increase in Developed Land Area (1992-2001)
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
TOTAL
Top 1% of Watersheds
River Miles
146
122
917
3,058
2,419
2,999
1,400
853
1,643
342
13,897
% of Miles
1%
1%
7%
22%
17%
22%
10%
6%
12%
2%
100%
Top 10% of Watersheds
River Miles
1,872
2,834
9,757
27,552
16,761
24,913
12,412
9,535
6,072
3,860
115,568
% of Miles
2%
2%
8%
24%
15%
22%
11%
8%
5%
3%
100%
Top 25% of Watersheds
River Miles
5,178
6,714
19,453
57,298
34,112
46,585
26,276
26,285
12,191
11,061
245,153
% of Miles
2%
3%
8%
23%
14%
19%
11%
11%
5%
5%
100%
Coterminous U.S.
(100% of Reaches)
River Miles
18,303
16,099
33,601
94,458
71,510
98,642
60,887
130,190
56,463
69,398
649,552
% of Miles
3%
2%
5%
15%
11%
15%
9%
20%
9%
11%
100%
November 2008
6-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-4: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 1% of RF1 Watersheds by Increase in Developed
Land Area (1992-2001)
Compliance
Scenario
Baseline
Option 1
Option 2
Option 3
Median
TSS
(mg/L)1
350.59
350.59
337.59
296.41
River-
Mile
Weighted
Average
TSS
(mg/L)1
703.72
703.07
680.48
660.78
Total Improved
RF1
Miles
-
4,314
12,619
13,031
%of
Total
Miles*
-
31.01%
90.71%
93.67%
Reduction in TSS Concentrations (mg/L)
0< TSS <1
RF1
Miles
-
3,532
4,176
2,007
%of
Total
Improved
Miles
-
25.39%
30.02%
14.43%
1< TSS <5
RF1
Miles
-
425
3,480
3,941
%of
Total
Improved
Miles
-
3.06%
25.02%
28.33%
5< TSS <10
RF1
Miles
-
Ill
1,025
1,720
%of
Total
Improved
Miles
-
0.80%
7.37%
12.36%
10< TSS <50
RF1
Miles
-
220
2,536
2,665
%of
Total
Improved
Miles
-
1.58%
18.23%
19.16%
50< TSS
RF1
Miles
-
25
1,403
2,698
%of
Total
Improved
Miles
-
0.18%
10.09%
19.39%
*Total miles (13,911) are based on top 1% of reaches by area of urban development from 1992 to 2001.
1 River-mile weighted.
Table 6-5: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 10% of RF1 Watersheds by Increase in Developed
Land Area (1992-2001)
Compliance
Scenario
Baseline
Option 1
Option 2
Option 3
Median
TSS
(mg/L)1
248.34
248.05
239.16
231.65
River-
Mile
Weighted
Average
TSS
(mg/L)1
642.70
642.43
631.25
622.91
Total Improved
RF1
Miles
-
32,455
106,876
109,397
%of
Total
Miles*
-
28.08%
92.46%
94.64%
Reduction in TSS Concentrations (mg/L)
0< TSS <1
RF1
Miles
-
29,358
42,851
27,264
%of
Total
Improved
Miles
-
25.40%
37.07%
23.59%
1< TSS <5
RF1
Miles
-
2,242
30,838
23,495
%of
Total
Improved
Miles
-
1.94%
26.68%
20.33%
5< TSS <10
RF1
Miles
-
395
11,396
33,818
%of
Total
Improved
Miles
-
0.34%
9.86%
29.26%
10< TSS <50
RF1
Miles
-
365
17,267
15,977
%of
Total
Improved
Miles
-
0.32%
14.94%
13.82%
50< TSS
RF1
Miles
-
96
4,523
8,843
%of
Total
Improved
Miles
-
0.08%
3.91%
7.65%
*Total miles (115,588) are based on top 10% of reaches by area of urban development from 1992 to 2001.
1 River-mile weighted.
November 2008
6-15
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-6: Estimated Changes in RF1 Reach TSS Concentration by Policy Option: Top 25% of RF1 Watersheds by Increase in Developed
Land Area (1992-2001)
Compliance
Scenario
Baseline
Option 1
Option 2
Option 3
Median
TSS
(mg/L)1
248.39
248.34
241.80
236.29
River-
Mile
Weighted
Average
TSS
(mg/L)1
632.84
632.68
625.02
619.46
Total Improved
RF1
Miles
-
69,845
225,465
231,932
%of
Total
Miles*
-
28.49%
91.97%
94.61%
Reduction in TSS Concentrations (mg/L)
0< TSS <1
RF1
Miles
-
65,058
104,586
74,942
%of
Total
Improved
Miles
-
26.54%
42.66%
30.57%
1< TSS <5
RF1
Miles
-
3,714
65,838
75,640
%of
Total
Improved
Miles
-
1.51%
26.86%
30.85%
5< TSS <10
RF1
Miles
-
536
21,668
29,476
%of
Total
Improved
Miles
-
0.22%
8.84%
12.02%
10< TSS <50
RF1
Miles
-
441
27,205
39,725
%of
Total
Improved
Miles
-
0.18%
11.10%
16.20%
50< TSS
RF1
Miles
-
96
6,169
12,149
%of
Total
Improved
Miles
-
0.04%
2.52%
4.96%
*Total miles (245,153) are based on top 25% of reaches by area of urban development from 1992 to 2001.
1 River-mile weighted.
November 2008
6-16
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.3.2 Estimated Changes in TSS Concentrations for all RF1 Watersheds Receiving
Sediment Loading from Construction Sites
Table 6-7 summarizes national average reach-length weighted TSS concentrations in RF1 reaches as
estimated by SPARROW under both the baseline scenario representing current conditions and the three
regulatory options (see Chapter 5 for option descriptions). Information is presented for all reaches in the
RF1 network that receive sediment loading from construction sites.
Table 6-7 also presents information on the distribution of TSS concentrations under the baseline scenario
representing current conditions and the three regulatory options. TSS concentrations in the RF1 network
vary over a wide range. Values for the 5th and 95th percentile bounds under baseline conditions are 25.11
and 3,035.28 mg/L respectively. Median (50th percentile) TSS concentrations are approximately
261.49 mg/L under the baseline scenario. Under Option 1, the median concentration decreases 0.17 mg/L
to 261.32 mg/L. Median concentrations under Option 2 decrease 4.88 mg/L to 256.61 mg/L; under Option
3, they decrease 7.72 mg/L to 253.77 mg/L. These numbers represent the national average concentration
reduction and do not provide information on the geospatial variability in TSS concentration reduction,
which is high. TSS concentrations in Table 6-7 representing the 5th, 25th, 50th, 75th, and 95th percentiles
provide information on the distribution of TSS concentration at the national level.
Table 6-7: TSS Concentration Distribution Based on SPARROW Output for 43,821 RF1 Reaches
Receiving Construction Sediment Discharges
Distribution of TSS Concentrations
4,2
Policy
Option
Baseline
Option 1
Option 2
Option 3
Reach
Count
43,821
43,821
43,821
43,821
River
Miles
517,982
517,982
517,982
517,982
Average
TSS
(mg/L)1'2
678.20
678.11
673.69
670.41
5th
Percentile
25.11
25.04
23.47
20.92
25th
Percentile
96.51
96.43
94.18
91.68
50th
Percentile
(Median)
261.49
261.32
256.61
253.77
75th
Percentile
723.68
723.60
715.05
711.39
95th
Percentile
3,035.28
3,035.28
3,029.70
3,019.24
1 Based on reach-length weighted distribution
2 Excludes potential outliers (i.e., reaches with baseline TSS concentration above the 95th percentile)
Source: USEPA analysis
Table 6-8 through Table 6-11 detail TSS concentrations by EPA region under baseline conditions and for
each of the three regulatory options considered. The highest average TSS concentrations are found in
Regions 6, 7, 8, and 9 which have average reach-length weighted TSS concentrations greater than
900 mg/L under all scenarios. Median TSS concentrations in these regions—ranging from 333 to
919 mg/L under baseline conditions—are also higher than median concentrations in Regions 1, 2, 3, 4, 5,
and 10 where the highest median TSS concentration is 185 mg/L. Under Option 1, TSS concentrations
decline a small amount (by 0.3 percent) from baseline concentrations in Region 4 and are nearly identical
in all other regions. Under Options 2 and 3, reductions in TSS concentrations are greater and are present
in all regions.
November 2008
6-17
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-8: Summary of Baseline TSS Concentration in RF1 Reaches Receiving Construction
Sediment Discharges, by EPA Region
EPA
Region
1
2
3
4
5
6
1
8
9
10
National
Reach
Count
1,863
1,467
2,673
8,870
5,170
5,774
4,019
6,931
2,471
4,583
43,821
River
Miles
15,599
15,439
32,676
91,882
68,476
83,009
56,489
81,993
34,393
38,026
517,982
Average
TSS
(mg/L)1'2
76.61
109.44
188.39
229.24
297.41
1,161.98
1,208.77
1,326.23
947.23
236.80
678.20
Distribution of TSS Concentrations1
5th
Percentile
6.19
10.26
30.16
27.32
16.54
64.68
161.91
63.42
28.07
19.41
25.11
25th
Percentile
25.46
33.80
73.23
86.95
58.88
233.85
443.79
271.95
140.02
52.92
96.51
50th
Percentile
(Median)
40.19
75.49
135.12
174.16
185.23
522.55
918.55
782.73
332.81
108.85
261.49
75th
Percentile
58.56
134.93
225.28
307.20
425.17
1,602.60
1,662.43
1,931.92
1,237.99
245.39
723.68
95th
Percentile
143.52
305.92
484.68
592.39
890.25
4,376.59
3,372.74
4,438.09
4,113.00
880.86
3,035.28
1 Based on reach-length weighted distribution
2 Excludes potential outliers (i.e., reaches with baseline TSS
Source: USEPA analysis
concentration above the 95th percentile)
Table 6-9: Summary of Option 1 TSS Concentration in RF1 Reaches Receiving Construction
Sediment Discharge, by EPA Region
Distribution of TSS Concentrations1
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Reach
Count
1,863
1,467
2,673
8,870
5,170
5,774
4,019
6,931
2,471
4,583
43,821
River
Miles
15,599
15,439
32,676
91,882
68,476
83,009
56,489
81,993
34,393
38,026
517,982
Average
TSS
(mg/L)1'2
76.61
109.43
188.39
228.79
297.40
1,161.98
1,208.76
1,326.22
947.23
236.80
678.11
5th
Percentile
6.19
10.26
30.16
27.30
16.54
64.68
161.91
63.42
28.07
19.41
25.04
25th
Percentile
25.46
33.80
73.23
86.83
58.88
233.85
443.79
271.93
140.02
52.92
96.43
50th
Percentile
(Median)
40.19
75.49
135.09
173.56
185.23
522.55
918.55
782.73
332.81
108.85
261.32
75th
Percentile
58.56
134.93
225.28
306.10
425.17
1,602.60
1,662.43
1,931.79
1,237.99
245.39
723.60
95th
Percentile
143.52
305.91
484.68
590.91
890.24
4,376.59
3,372.72
4,437.80
4,113.00
880.86
3,035.28
1 Based on reach-length weighted distribution
Excludes potential outliers (i.e., reaches with baseline TSS concentration above the 95th percentile)
Source: USEPA analysis
November 2008
6-18
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-10: Summary of Option 2 TSS Concentration in RF1 Reaches Receiving Construction
Sediment Discharge, by EPA Region
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Reach
Count
1,863
1,467
2,673
8,870
5,170
5,774
4,019
6,931
2,471
4,583
43,821
River
Miles
15,599
15,439
32,676
91,882
68,476
83,009
56,489
81,993
34,393
38,026
517,982
Average
TSS
(mg/L)1'2'
76.33
108.97
184.63
222.60
296.52
1,152.75
1,199.25
1,325.85
946.91
231.62
673.69
Distribution of TSS Concentrations1
5th
Percentile
6.13
10.25
28.68
26.77
16.20
54.06
158.86
63.42
28.07
12.48
23.47
25th
Percentile
25.17
33.75
71.75
84.59
58.23
224.94
434.18
271.89
140.01
47.92
94.18
50th
Percentile
(Median)
39.75
74.89
132.22
167.76
185.18
512.77
913.59
782.73
332.81
105.14
256.61
75th
Percentile
58.25
133.98
222.62
300.46
422.97
1,593.86
1,650.71
1,931.36
1,237.99
242.08
715.05
95th
Percentile
143.36
305.14
475.38
574.48
889.99
4,376.36
3,284.40
4,436.97
4,112.90
867.20
3,029.70
1 Based on reach-length weighted distribution.
2 Average excludes potential outliers (values above the 95th percentile).
Source: USEPA analysis
Table 6-11: Summary of Option 3 TSS Concentration in RF1 Reaches Receiving Construction
Sediment Discharge, by EPA Region
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Reach
Count
1,863
1,467
2,673
8,870
5,170
5,774
4,019
6,931
2,471
4,583
43,821
River
Miles
15,599
15,439
32,676
91,882
68,476
83,009
56,489
81,993
34,393
38,026
517,982
Average
TSS
(mg/L)1'2'
76.33
108.97
184.63
222.60
296.52
1,152.75
1,199.25
1,325.85
946.91
231.62
670.41
Distribution of TSS Concentrations1
5th
Percentile
5.87
9.98
27.56
19.93
15.74
49.93
158.60
62.97
28.07
0.00
20.92
25th
Percentile
25.02
33.57
71.28
80.71
57.32
221.93
427.47
268.12
139.55
42.00
91.68
50th
Percentile
(Median)
39.62
74.31
130.96
163.53
184.78
511.88
909.96
779.09
332.35
100.99
253.77
75th
Percentile
58.14
133.89
221.41
293.93
421.09
1,590.63
1,645.77
1,929.57
1,225.99
237.22
711.39
95th
Percentile
142.02
304.46
466.99
560.85
888.25
4,374.28
3,284.26
4,431.54
4,112.12
860.97
3,019.24
1 Based on reach-length weighted distribution.
2 Excludes potential outliers (i.e., reaches with baseline TSS concentration above the 95th percentile).
Source: USEPA analysis
The following tables describe the number of RF1 reach miles improved and the distribution of TSS
concentration reductions for each of three options analyzed for this regulation. As shown in Table 6-12,
which summarizes the effects of sediment reduction at the national level, Option 3 improves 94.3 percent
of modeled RF1 river miles receiving construction sediment loading. Option 2 is predicted to lead to
improvements in 90.9 percent of RF1 river miles, and Option 1 is predicted to improve 29.9 percent of
RF1 river miles. Under Option 3, approximately half of the improved river miles show reductions in TSS
concentrations of less than one mg/L, and another 26.9 percent show reductions in TSS concentrations
November 2008
6-19
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
between 1 and 5 mg/L. Nearly 3 percent of improved RF1 river miles are expected to see larger
reductions in TSS concentrations of 50 mg/L or more. Under Option 2, 62.6 percent of improved RF1
river miles experience TSS reductions of less than 1 mg/L, and 7.1 percent are improved by a reduction in
TSS concentrations greater than 10 mg/L. Of the nearly 30 percent of RF1 river miles improved under
Option 1, 96.4 percent show improvements of less than 1 mg/L.
Table 6-13 summarizes Option 1's sediment reductions at the regional level. Nationwide, 96.4 percent of
improved RF1 river miles under Option 1 experience TSS concentration reductions less than 1 mg/L. Out
often regions, only EPA Region 4 is estimated to experience larger improvements in surface water
quality, with 8.6 percent of improved RF1 river miles projected to have TSS concentrations reduced
between 1 and 5 mg/L and 2.5 percent of river miles expected to have TSS concentrations reduced by
5 mg/L or greater.
As shown in Table 6-14, Option 2 is estimated to reduce TSS concentrations by greater than 50 mg/L in
7,014 river miles nationwide that receive construction sediment loading, and 33,577 river miles by 10 to
50 mg/L. Smaller improvements of 1 to 5 mg/L are predicted in 106,250 RF1 river miles, and 294,802
river miles are expected to experience modest improvements in TSS concentrations (less than 1 mg/L).
EPA Region 4 is estimated to benefit the most in terms of total improved river miles (91,296 miles).
Regions 6 and 7 are expected to see a greater percentage of reaches with reductions in TSS greater than
10 mg/L. Regions 1, 2, 5, 8, and 9 are predicted to have more than 75 percent of TSS concentration
reductions be less than 1 mg/L.
Option 3 (Table 6-15) is estimated to reduce TSS concentrations by more than 50 mg/L in 14,106 RF1
miles receiving construction sediment loading and by between 10 and 50 mg/L in 50,845 miles, meaning
that more than 13 percent of improvements under this option are TSS reductions of 10 mg/L or greater.
Region 4 has the largest gain in terms of number of river miles (91,297 miles) improved, though Region 7
is expected to have a greater proportion of improved reaches with larger reductions in sediment
concentration, with nearly 28 percent of improved RF1 miles having 10 mg/L or greater reductions in
TSS concentrations. Regions 1, 2, 5, 8, 9, and 10 have smaller reductions in most of their improved
reaches though these regions do have more reaches with TSS reductions 1 mg/L or greater than they do
under Options 1 and 2.
November 2008 6-20
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-12: Summary of Estimated Changes in TSS Concentration by Policy Option
Compliance
Scenario
Option 1
Option 2
Option 3
Reduction in TSS Concentrations (mg/L)
0< A TSS <1
RF1
Miles
149,029
294,803
249,489
%of
Improved
Miles
96.36%
62.63%
51.10%
1< A TSS <5
RF1
Miles
4,388
106,250
131,368
%of
Improved
Miles
2.84%
22.57%
26.91%
5< A TSS <10
RF1
Miles
639
29,085
42,416
%of
Improved
Miles
0.41%
6.18%
8.69%
10< A TSS <50
RF1
Miles
506
33,578
50,844
%of
Improved
Miles
0.33%
7.13%
10.41%
50< A TSS
RF1
Miles
96
7,014
14,105
%of
Improved
Miles
0.06%
1.49%
2.89%
Total Improvements
RF1
Miles
154,658
470,730
488,224
% of Total
RF1 Miles1
29.86%
90.88%
94.26%
1 Calculated as a percentage of the RF1 river miles that receive construction sediment loading (517,982 miles in total), as modeled by SPARROW.
Source: USEPA analysis
Table 6-13: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 1
Reduction in TSS Concentration (mg/L)
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Total
Total
Improved
River Miles
3,165
3,048
917
48,589
32,874
3,836
12,581
46,256
505
2,896
154,667
0 < A TSS < 1
River
Miles
3,165
3,048
917
43,187
32,846
3,836
12,458
46,171
505
2,896
149,029
%of
Total
Improved
Miles
100.00%
100.00%
100.00%
88.88%
99.91%
100.00%
99.02%
99.82%
100.00%
100.00%
96.35%
1 < A TSS < 5
River
Miles
0
0
0
4,170
28
0
123
67
0
0
4,388
%of
Total
Improved
Miles
0.00%
0.00%
0.00%
8.58%
0.09%
0.00%
0.98%
0.14%
0.00%
0.00%
2.84%
5 < A TSS < 10
River
Miles
0
0
0
630
0
0
0
9
0
0
639
%of
Total
Improved
Miles
0.00%
0.00%
0.00%
1.30%
0.00%
0.00%
0.00%
0.02%
0.00%
0.00%
0.41%
10 < A TSS < 50
River
Miles
0
0
0
506
0
0
0
9
0
0
515
%of
Total
Improved
Miles
0.00%
0.00%
0.00%
1.04%
0.00%
0.00%
0.00%
0.02%
0.00%
0.00%
0.33%
50 < A TSS
River
Miles
0
0
0
96
0
0
0
0
0
0
96
% of Total
Improved
Miles
0.00%
0.00%
0.00%
0.20%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.06%
Source: USEPA analysis
November 2008
6-21
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-14: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 2
Reduction in TSS Concentration (mg/L)
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Total
Total
Improved
River Miles
15,391
15,367
32,308
91,296
68,136
75,721
49,840
64,596
27,026
31,048
470,729
0 < A TSS < 1
River
Miles
14,659
13,760
16,846
38,783
56,203
27,413
19,574
58,576
26,231
22,757
294,802
%of
Total
Improved
Miles
95.24%
89.54%
52.14%
42.48%
82.49%
36.20%
39.27%
90.68%
97.06%
73.30%
62.63%
1 < A TSS < 5
River
Miles
604
1,412
11,777
32,405
9,786
25,861
14,590
4,903
696
4,216
106,250
%of
Total
Improved
Miles
3.92%
9.19%
36.45%
35.49%
14.36%
34.15%
29.27%
7.59%
2.58%
13.58%
22.57%
5 < A TSS < 10
River
Miles
105
191
1,859
8,785
1,269
9,273
5,451
718
14
1,421
29,086
%of
Total
Improved
Miles
0.68%
1.24%
5.75%
9.62%
1.86%
12.25%
10.94%
1.11%
0.05%
4.58%
6.18%
10 < A TSS < 50
River
Miles
23
4
1,665
9,278
811
11,010
8,519
399
49
1,819
33,577
%of
Total
Improved
Miles
0.15%
0.03%
5.15%
10.16%
1.19%
14.54%
17.09%
0.62%
0.18%
5.86%
7.13%
50 < A TSS
River
Miles
0
0
161
2,045
67
2,164
1,706
0
36
835
7,014
% of Total
Improved
Miles
0.00%
0.00%
0.50%
2.24%
0.10%
2.86%
3.42%
0.00%
0.13%
2.69%
1.49%
Source: USEPA analysis
November 2008
6-22
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-15: Water Quality Impacts: Improvements in TSS Concentrations from Baseline to Option 3
Reduction in TSS Concentration (mg/L)
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Total
Total
Improved
River Miles
15,392
15,367
32,309
91,297
68,135
75,720
49,839
71,666
30,746
37,755
488,226
0 < A TSS < 1
River
Miles
13,572
12,165
13,046
22,405
49,131
19,929
15,142
53,098
24,774
26,226
249,488
%of
Total
Improved
Miles
88.18%
79.16%
40.38%
24.54%
72.11%
26.32%
30.38%
74.09%
80.58%
69.46%
51.10%
1 < A TSS < 5
River
Miles
1,465
2,783
13,882
36,548
14,972
23,368
15,542
14,387
4,133
4,290
131,370
%of
Total
Improved
Miles
9.52%
18.11%
42.97%
40.03%
21.97%
30.86%
31.18%
20.08%
13.44%
11.36%
26.91%
5 < A TSS < 10
River
Miles
193
334
2,573
12,637
2,286
13,413
5,382
2,651
908
2,040
42,417
%of
Total
Improved
Miles
1.25%
2.17%
7.96%
13.84%
3.36%
17.71%
10.80%
3.70%
2.95%
5.40%
8.69%
10 < A TSS < 50
River
Miles
138
85
2,362
15,192
1,515
15,050
10,719
1,376
831
3,577
50,845
%of
Total
Improved
Miles
0.90%
0.55%
7.31%
16.64%
2.22%
19.88%
21.51%
1.92%
2.70%
9.47%
10.41%
50 < A TSS
River
Miles
24
0
446
4,515
231
3,960
3,054
154
100
1,622
14,106
% of Total
Improved
Miles
0.16%
0.00%
1.38%
4.95%
0.34%
5.23%
6.13%
0.21%
0.33%
4.30%
2.89%
Source: USEPA analysis
November 2008
6-23
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.3.3 Impacts on Reservoir Sedimentation
Construction discharges affect not only sediment concentrations in receiving water bodies, but they also
affect the amount of sedimentation that takes place in reservoirs. As described in Section 6.2, reservoirs
provide major sites for sediment attenuation within the hydrographic network. This attenuation is
estimated in SPARROW for more than 4,000 reservoirs distributed throughout the RF1 network. The
model estimates that more than 449 million cubic yards accumulate under baseline conditions. The
accumulating sediment may hinder reservoir functions and need to be periodically dredged, as discussed
in greater detail in Chapter 8. The policy options reduce this amount to various degrees. As shown in
Table 6-16, Option 1 reduces the amount of sediment accumulating in reservoirs slightly as compared to
current conditions (by 123,600 cubic yards or approximately 0.03 percent). Most of this reduction is
concentrated in Region 4. Options 2 and 3 show much greater reductions in reservoir sedimentation, with
estimated reductions of 5.9 and 9.8 million cubic yards, respectively (or 1.3 and 2.2 percent). They also
show more geographically distributed effects, with significant reductions observed in Regions 6, 4, 7, and
10.
Table 6-16: Sediment Accumulation in Reservoirs by Policy Option and EPA Region
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Sediment Accumulation
(thousand cubic yards)
Baseline
1,318
2,769
5,457
46,353
36,094
130,271
42,126
127,509
42,583
14,918
449,399
Option 1
1,318
2,769
5,457
46,233
36,093
130,271
42,126
127,509
42,583
14,918
449,275
Option 2
1,308
2,758
5,378
44,426
35,940
127,468
41,412
127,500
42,579
14,732
443,502
Option 3
1,298
2,749
5,342
43,012
35,854
126,000
41,129
127,444
42,537
14,195
439,559
Reduction in Sediment Accumulation
(thousand cubic yards)
Option 1
0.1
0.1
0.2
120.6
1.4
0.2
0.3
0.7
0.1
0.0
123.7
Option 2
9.6
11.6
78.9
1,927.7
154.1
2,802.6
714.0
9.0
3.6
185.8
5,896.9
Option 3
20.3
20.4
115.4
3,341.0
240.4
4,270.5
997.8
65.7
45.2
722.9
9,839.6
1 Excludes reservoirs located on "outlier" reaches (i.e., reaches with baseline TSS concentration above the 95th percentile).
6.3.4 Hypothetical "No Discharge" Scenario
Table 6-17 presents SPARROW output for the baseline scenario representing current conditions and a
hypothetical scenario in which all sediment discharges from construction activity are prevented from
entering surface waters. SPARROW estimates that approximately 767 million cubic yards of sediment are
transported in RF1 reaches each year, and 450 million cubic yards settle in reservoirs. SPARROW
estimates that removing all construction site sediment contributions would result in a 3.48 percent
reduction in national sediment loadings in RF1 reaches and a 2.49 percent reduction in national reservoir
sediment attenuation. It is important to note that a certain level of sediment is present in surface waters
under natural, undisturbed conditions and that some sediment is necessary for the natural biological
function of surface waters. Current levels of sediment discharge to the U.S. surface water network are
much higher, however, than would be observed under natural, undisturbed conditions due to sediment
discharges from construction, agriculture, and eroding streambeds associated with land use change and
November 2008
6-24
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
other human activity. Therefore, while construction sediment discharges represent approximately
3.48 percent of total sediment in surface waters, they represent a greater percentage of total sediment
pollution in the nation's waters. Given the relatively small percentage of U.S. land area dedicated to
construction activity on an annual basis (-0.03% of coterminous United States), construction sites have a
disproportionately high rate of sediment contribution to surface waters relative to other land uses.
Table 6-17: Total Yearly Sediment from All Sources for Baseline and Zero Discharge Scenarios
Based on SPARROW Output for 62,370 RF1 Reaches
Scenario
Baseline
Zero Discharges
River
Miles
650,043
650,043
In-Stream
Sediment
Flux
(millions of
cubic yards %
per year)1"2 Reduction
317.51
302.01 4.88%
Reservoir
Sediment
Attenuation
(millions of
cubic yards
per year)u
449.78
438.56
Total
Sediment
Flux
(millions of
% cubic yards
Reduction per year)1
767.29
2.49% 740.57
Reduction
-
3.48%
1 Based on all reaches estimated by SPARROW (62,370 RF1 Reaches). One cubic yard = 1,127kg.
2 In-stream sediment flux estimated by subtracting total sediment flux minus total reservoir sediment attenuation.
3 Total includes reaches that do not receive construction sediment loading. Values are therefore slightly greater than reported in other tables that
include only reaches that receive construction sediment loading (449.40 million cubic yards for the baseline scenario).
Source: USEPA analysis
6.4 Extension of SPARROW to Estuaries and Coastal Waters
EPA used SPARROW output to evaluate sediment deliveries to estuaries, as well. Estimates of ambient
concentrations of sediment [Total Suspended Solids (TSS)] within estuaries were calculated using the
Dissolved Concentration Potential (DCP) approach (Versar, Inc./USEPA, 1997). The DCP approach was
developed by the National Oceanic and Atmospheric Administration (NOAA) to predict the ambient
concentration of conserved contaminants that are introduced into an estuary and are then subject to
mixing and dilution. An advantage to this approach is that published values of DCP have been provided
by NOAA for most major estuaries in the coterminous United States. Using this approach, EPA was able
to estimate TSS concentrations associated with annual quantities of sediment discharged into each estuary
(as predicted by SPARROW) without detailed numerical simulation modeling of estuaries. The DCP
approach presumes that water quality constituents are dissolved, conservative pollutants dispersed under
well-mixed, steady-state conditions. These criteria are not strictly met for TSS, although the finest particle
fractions approximately met these criteria due to their extremely low settling velocities and relatively
stable composition. Because current and proposed sediment abatement practices remove coarser particle
fractions first, construction site sediment discharges are expected to contain primarily finer particles.
Therefore, the DCP methodology criteria are assumed to approximately applicable for estimating changes
in estuary water quality associated with changes in construction site sediment discharges, and the
approach may be capable of generating credible results for most major estuaries.
NOAA has calculated DCP values for most major estuaries in the Southeast (NOAA and USEPA, 1989a);
Northeast (NOAA and USEPA, 1989b); Gulf of Mexico (NOAA and USEPA, 1989c); and West Coast
(NOAA and USEPA, 1991) of the coterminous United States. Calculated DCP values for 22 West Coast,
24 Northeast, 23 Gulf, and 17 Southeastern U.S. estuaries are presented in tabular form in USEPA (1997)
and in Table 6-18 to Table 6-21. Data on these and additional estuaries are found via the NOAA Coastal
Geospatial Data Project (http://coastalgeospatial.noaa.gov).
November 2008
6-25
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Total sediment loadings estimated in SPARROW for reaches draining to estuaries and shoreline reaches
draining to estuaries were aggregated on the basis of 5-digitNOAA Estuarine Drainage Area (EDA)
codes. Estimated loadings are calculated in SPARROW as Suspended Sediment Concentrations (SSC).
Suspended Sediment Concentrations needed to be adjusted before the application of DCP methods, for
two reasons. First, reported SSC values include all particle size fractions and coarse sediment fractions
are not assumed to behave as conserved within the water column due to higher settling velocities.
Approximately 127,000 sediment records for which particle size distributions were recorded, collected at
1805 water quality gauging stations obtained from USGS (Schwarz 2008b) were analyzed. It was found
that on average over samples, 78.4 percent of suspended sediment mass was classified as "fine" (diameter
< 0.0625 mm). This percentage varied to some extent by 2-digit USGS Water Resources Regions, from
71 percent (Region 14) to 89.5 percent (Region 9); and was subject to even wider at-site inter-seasonal
and inter-annual variations. It is assumed that the implied sediment size distribution of SPARROW output
is equivalent to observed distributions. However, only 15 of the 1,805 water quality gauging stations for
which sediment size fraction information was available were terminal or coastal reaches (draining directly
into bays or estuaries), so that no regional or estuary-specific sediment size distributions could be
identified. A second consideration is that USGS (2000) has determined that Suspended Sediment
Concentration (SSC) and Total Suspended Solids (TSS), while carrying the same physical interpretation,
are subject to systematic differences in measurement due to differing laboratory protocols. SSC values in
the SPARROW output are divided by 1.3 to adjust for this difference, as described in Section 6.3: Water
Quality Modeling Re suits. The use of this adjustment factor yields TSS concentrations that are roughly 77
percent of corresponding SSC values. Since this adjustment is reasonably close to the nation-wide
(average of samples) percentage of fine particles in SSC (78 percent), SPARROW SSC values were
converted to TSS equivalent, and the TSS values assumed to conform approximately to the assumptions
underlying DCP calculations. In order to determine how well these assumptions are reflected in observed
estuarine water quality conditions, data on suspended sediment size distribution for 623 samples taken
within San Francisco Bay were obtained from the USGS (http://sfbay.wr.usgs.gov/access/wqdata/). The
percentage of fine particles (< 0.0625 mm) was 71 percent for these samples. It is evident that suspended
sediment accounting within estuaries is complicated, and ambient TSS levels, in addition to the
percentage of fine particles, are observed to reflect many influences beyond the annual riverine input of
sediment. As an additional check on SPARROW outputs to estuaries, the aggregated annual loadings per
unit of contributing watershed area were calculated for each channel contributing sediment to estuaries,
and found to be reasonable in most cases. Annual sediment loads as estimated by SPARROW were
aggregated by estuary, unit-converted and divided by 10,000 tons per year (the DCP unit loading), and
multiplied by DCP coefficients to obtain ambient estuarine TSS concentrations. Baseline estimates of
estuarine TSS varied from below 1 mg/L (Northeastern estuaries) to a maximum of 830 mg/L (San
Antonio Bay, TX), and averaged 71 mg/L over all estuaries and all scenarios. Estimates of TSS as
calculated using DCP appear in Table 6-18 for Southeastern Estuaries, Table 6-19 for Gulf of Mexico
Estuaries, Table 6-20 for Northeastern Estuaries, and Table 6-21 for West Coast Estuaries. A number of
estimated TSS values fall outside of anticipated ranges, and the sources of variation across estuaries
require careful investigation. Negative concentrations under Option 3 for Indian River, FL, Tampa Bay,
FL, Puget Sound, WA and Yaquina Bay, OR represent artifacts of SPARROW estimation and are not to
be interpreted physically.
Average reductions in estuarine TSS concentrations relative to baseline values were less than 1 percent
for Option 1; approximately 3.3 percent for Option 2; and 8.7 percent for Option 3. Considerable regional
variation was observed. In Northeastern estuaries, very little change in TSS concentrations were observed
across management options. This primarily reflects the relatively low baseline TSS concentrations,
November 2008 6-26
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
averaging only 13 mg/L across Northeastern estuaries. By contrast, baseline TSS concentrations across all
estuaries averaged 71 mg/L, and TSS concentrations in Gulf of Mexico estuaries averaged 149 mg/L. The
largest regional changes observed, a 13 percent average decrease relative to baseline TSS concentrations
for Option 3, were found in the West Coast estuaries.
A preliminary examination was made of SPARROW-simulated TSS concentrations relative to measured
values from monitoring data. Measured observations were obtained from the EPA National Coastal
Assessment (http://www.epa.gov/emap/nca/html/data/index.html). and augmented for San Francisco Bay
with data from USGS (http://sfbav.wr.usgs.gov/access/wqdata/). Estimated TSS concentrations obtained
from SPARROW predictions and application of the DCP approach in general exceed measured values,
although agreement is remarkably good in many instances. In particular, SPARROW/DCP estimates of
TSS for Delaware Bay (40.33 mg/L) are very close to observed values (38.45), and virtually identical for
Long Island Sound (9.75 predicted vs. 9.8 observed). Results for San Francisco Bay illustrate the
difficulty in comparing model predictions with observations. SPARROW/DCP TSS estimates of 90 mg/L
are slightly below the NOAA EMAP average of 71 observations (132 mg/L); but above the mean of the
larger USGS sample (34 mg/L). These and related results suggest that SPARROW loading predictions,
transformed by DCP assumptions to estuarine ambient concentrations (TSS), provide a reasonable basis
for evaluating changes in TSS due to specific regulatory options, although predicted TSS values may
differ from physical observation in absolute value in certain settings. Results are summarized in Table
6-22 for seven major estuaries.
November 2008 6-27
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-18: TSS Concentrations (mg/L) Estimated by DCP, Southeastern Estuaries
Estuary
Albemarle/Pamlico Sound, NC, VA
Altamaha River, GA
Biscayne Bay, FL
Bogue Sound, NC
Broad River, SC
Cape Fear River, NC
Charleston Harbor, SC
Indian River, FL
N and S Santee Rivers, SC
New River, NC
Ossabaw Sound, GA
Savannah River, SC, GA
St. Catherines/Sapelo Sound, GA
St. Helena Sound, SC
St. Johns River, FL
St. Andrews/St. Simons Sound, GA
WinyahBay, SC, NC
DCP
0.14
0.37
0.4
1.47
4.92
0.61
0.44
1.02
2.91
7.68
1.99
0.43
7.56
0.95
0.83
2.43
0.39
Base
TSS (mg/L)
34.3
82.8
0.2
3.0
16.3
69.3
2.6
5.3
456.5
19.0
81.0
61.1
17.9
10.8
13.5
47.4
108.6
Option 1
TSS (mg/L)
34.2
82.3
0.2
3.0
16.3
69.2
2.6
5.3
456.2
19.0
80.9
61.0
17.8
10.8
13.5
47.3
108.5
% change
rel. to base
-0.1%
-0.6%
0.0%
-0.1%
0.0%
-0.1%
0.0%
0.0%
-0.1%
-0.1%
-0.2%
-0.2%
-0.7%
0.0%
0.0%
-0.3%
-0.1%
Option 2
TSS (mg/L)
33.9
78.6
0.2
3.0
15.9
68.7
2.6
5.2
450.5
18.9
79.9
58.9
16.8
10.8
13.5
46.0
107.4
% change
rel. to base
-1.0%
-5.1%
0.0%
-0.8%
-2.7%
-0.9%
-2.3%
-1.1%
-1.3%
-0.5%
-1.4%
-3.7%
-6.2%
-0.8%
-0.3%
-2.9%
-1.1%
Option 3
TSS (mg/L)
33.7
74.7
0.2
2.9
14.5
68.3
2.4
N/A
441.1
18.9
78.9
54.4
15.8
10.5
7.4
44.7
106.4
% change
rel. to base
-1.7%
-9.7%
-4.7%
-1.4%
-10.9%
-1.5%
-9.2%
N/A
-3.4%
-0.9%
-2.7%
-11.0%
-11.9%
-3.3%
-45.4%
-5.7%
-2.0%
Source: USEPA analysis
November 2008
6-28
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-19: TSS Concentrations (mg/L) Estimated by DCP, Gulf of Mexico Estuaries
Estuary
Apalachee Bay, FL, GA
Apalachicola Bay, FL
Arkansas Bay, TX
Base
DCP TSS (mg/L)
0.36
0.17
6.02
Atchafalaya and Vermillion Bays, LA 0.04
Brazos River, TX
Calcasieu Lake, LA
Charlotte Harbor, FL
Choctawhatchee Bay, FL, AL
Corpus Cristi Inner Harbor, TX
Galveston Bay, TX
Laguna Madre, TX
Matagorda Bay, TX
Mississippi River, LA, MS
Mississippi Sound, MS, LA, AL
Mobile Bay, AL
Pensacola Bay, FL
Perdido Bay, FL, AL
Sabine Lake, LA, TX
San Antonio Bay, TX
St. Andrew Bay, FL
Suwanee River, FL
Tampa Bay, FL
Ten Thousand Islands, FL
1.11
1.18
0.76
0.7
4.67
0.4
0.34
0.81
0.01
0.17
0.08
0.46
2.89
0.38
1.3
0.76
0.38
1.03
1.94
8.2
27.7
138.6
251.0
595.6
48.3
15.5
51.6
554.9
93.3
24.0
186.5
194.8
128.0
86.9
43.9
35.6
76.3
829.4
2.3
24.6
10.5
1.7
Option 1 % change
TSS (mg/L) rel. to base
8.2
27.1
138.6
251.0
595.6
48.3
15.5
51.5
554.9
93.3
24.0
186.5
194.8
128.0
86.7
43.9
35.5
76.3
829.4
2.3
24.6
10.5
1.7
0.0%
-2.0%
0.0%
0.0%
0.0%
0.0%
0.0%
-0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
-0.2%
-0.1%
-0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Option 2
TSS (mg/L)
8.2
22.7
138.3
248.0
593.6
47.5
15.5
51.2
549.1
85.1
24.0
185.4
193.3
119.6
85.2
43.6
35.2
67.0
817.0
2.3
24.5
10.3
1.7
% change
rel. to base
-0.5%
-18.2%
-0.3%
-1.2%
-0.3%
-1.7%
-0.1%
-0.6%
-1.0%
-8.7%
-0.3%
-0.6%
-0.8%
-6.5%
-2.0%
-0.7%
-1.1%
-12.3%
-1.5%
-0.2%
-0.4%
-1.8%
0.0%
Option 3
TSS (mg/L)
4.7
18.0
138.1
246.7
592.5
47.0
13.5
50.5
546.0
80.8
23.9
184.7
192.7
115.5
83.9
43.0
34.4
61.9
810.4
1.7
23.7
N/A
1.6
% change
rel. to base
-43.0%
-35.1%
-0.4%
-1.7%
-0.5%
-2.9%
-13.0%
-2.1%
-1.6%
-13.4%
-0.4%
-1.0%
-1.1%
-9.8%
-3.4%
-2.1%
-3.4%
-18.9%
-2.3%
-23.9%
-3.4%
N/A
-2.4%
Source: USEPA analysis
Table 6-20: TSS Concentrations
(mg/L) Estimated
by DCP,
Northeastern Estuaries
November 2008 6-29
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Estuary
Barnegat Bay, NJ
Blue Hill Bay, ME
Buzzards Bay, MA
Cape Cod Bay, MA
Casco Bay, ME
Chesapeake Bay, VA, MD, DE, PA
Chincoteague Bay, MD, VA
Delaware Bay, DE, NJ, PA, MD
Englishman Bay, ME
Gardiners Bay, NY
Great Bay, NE, NH
Great South Bay, NY
Hudson River/Raritan Bay, NY, NI
Long Island Sound, NY, CT, MA
Massachusetts Bay, MA
Merrimack River, NH., MA
Muscongus Bay, ME
Narragansett Bay, MA, RI
Narragauges Bay, ME
Passaquoddy Bay, ME
Penobscot Bay, ME
Saco Bay, ME, NH
Sheepscop Bay, ME, NH
Base
DCP TSS(mg/L)
1.36
1.03
1.04
0.69
0.61
, DC 0.072
3.08
0.14
0.92
1.77
1.54
5.07
, MA, CT 0.2
0.054
0.27
1.01
2.25
0.52
1.54
0.27
0.13
0.45
0.088
3.7
1.7
3.8
0.5
3.3
92.9
2.4
40.3
4.7
2.0
7.6
12.0
52.4
9.7
2.4
32.0
6.8
8.4
4.3
1.5
7.1
4.2
0.2
Option 1
TSS (mg/L)
3.7
1.7
3.8
0.5
3.3
92.9
2.4
40.3
4.7
2.0
7.6
12.0
52.4
9.7
2.4
32.0
6.8
8.4
4.3
1.5
7.1
4.2
0.2
% change
rel. to base
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Option 2
TSS (mg/L)
3.7
1.7
3.8
0.5
3.3
91.4
2.4
39.9
4.7
2.0
7.6
12.0
52.1
9.7
2.4
31.9
6.6
8.4
4.3
1.5
7.1
4.2
0.2
% change
rel. to base
-0.2%
-2.0%
-0.1%
-0.1%
-0.6%
-1.6%
-0.9%
-1.2%
-0.1%
-0.2%
-0.4%
-0.2%
-0.7%
-0.2%
-0.2%
-0.1%
-3.8%
-0.1%
-0.3%
-0.1%
-0.7%
-0.3%
-1.7%
Option 3
TSS (mg/L)
3.7
1.6
3.7
0.5
3.3
90.6
2.4
39.7
4.7
2.0
7.5
12.0
51.8
9.7
2.3
31.7
6.3
8.3
4.3
1.5
7.0
4.2
0.2
% change
rel. to base
-0.7%
-3.8%
-1.2%
-1.4%
-1.1%
-2.5%
-1.5%
-1.7%
-0.2%
-0.3%
-1.4%
-0.4%
-1.3%
-0.5%
-2.2%
-0.8%
-7.4%
-0.9%
-0.5%
-0.1%
-1.3%
-0.8%
-3.3%
Source: USEPA analysis
Table 6-21: TSS Concentrations
(mg/L) Estimated
by DCP, West Coast Estuaries
November 2008
6-30
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Estuary
Alsea River, OR
Columbia River, WA, OR
Coos Bay, OR
Eel River, CA
Grays Harbor, WA
Humboldt Bay, CA
Klamath River, OR, CA
Monterey Bay, CA
Nehalam River, OR
Netarts Bay, OR
Puget Sound, WA
Rogue River, OR
San Diego Bay, CA
San Francisco Bay, CA
San Pedro Bay, CA
Santa Monica Bay, CA
Siletz Bay, OR
Siuslaw River, OR
Tillamook Bay, OR
Umpqua River, OR
WillapaBay, WA
Yaquina Bay, OR
DCP
2.086
0.032
2.27
0.39
0.278
8
0.31
1.05
1.556
12.564
0.039
0.556
12.31
0.104
4.41
1.371
2.424
1.853
1.049
0.793
0.466
3.356
Base
TSS (mg/L)
37.4
76.0
51.3
65.3
21.2
38.7
79.7
148.4
38.5
15.7
9.7
125.2
88.9
90.4
104.4
40.9
42.8
46.5
22.1
109.5
12.1
25.7
Option 1
TSS (mg/L)
37.4
76.0
51.3
65.3
21.2
38.7
79.7
148.4
38.5
15.7
9.7
125.2
88.9
90.4
104.4
40.9
42.8
46.5
22.1
109.5
12.1
25.7
% change
rel. to base
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Option 2
TSS (mg/L)
36.5
75.0
45.5
65.3
18.7
38.6
79.6
148.4
37.2
14.6
6.9
124.9
88.9
90.4
104.4
40.9
39.5
45.6
20.7
109.0
9.4
9.3
% change
rel. to base
-2.2%
-1.3%
-11.4%
0.0%
-11.6%
-0.2%
-0.2%
0.0%
-3.4%
-7.4%
-28.4%
-0.2%
0.0%
0.0%
0.0%
0.0%
-7.9%
-1.9%
-6.3%
-0.5%
-22.3%
-63.8%
Option 3
TSS (mg/L)
35.0
72.7
34.7
65.1
11.6
37.9
78.5
148.3
34.8
12.4
N/A
124.3
88.9
90.2
104.3
40.9
33.2
44.0
18.1
108.0
1.5
N/A
% change
rel. to base
-6.4%
-4.3%
-32.4%
-0.3%
-45.2%
-1.9%
-1.6%
0.0%
-9.5%
-21.0%
N/A
-0.7%
-0.1%
-0.3%
-0.1%
0.0%
-22.5%
-5.4%
-17.9%
-1.4%
-87.2%
N/A
Source: USEPA analysis
November 2008
6-31
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 6-22: Comparison between SPARROW-DCP Estimated and Observed Suspended-Sediment Concentrations (mg/L)
Estuary
Chesapeake Bay
Columbia River
Delaware Bay
Hudson River
Long Island Sound
Puget Sound
San Francisco Bay (1)
San Francisco Bay (2)
DCP
0.072
0.032
0.14
0.2
0.054
0.039
0.104
0.104
Area
(km2)
168,794
617,376
33,777
41,793
42,222
29,500
116,772
116,772
SPARROW
baseline
92.92
76.02
40.33
52.44
9.75
9.68
90.39
90.39
Data
mean
13.82
8.33
38.45
16.07
9.80
4.31
132.13
33.88
Data std.
dev.
7.88
13.36
32.01
22.97
11.18
1.36
470.58
46.16
Coefficient
variation
0.57
1.60
0.83
1.43
1.14
0.32
3.56
1.36
Data
count
94
132
90
40
176
60
71
5176
Data
source
b
c
a
a
a
d
e
f
Sources:
a. EMAP VA Province.
b. EMAP Mid-Atlantic, VA Province.
c. EMAP West Coast/Oregon DEQ.
d. EMAP West Coast/Washington Dep. Env.
e. EMAP West Coast/NOAA.
f. USGS Water Quality of San Francisco Bay.
November 2008 6-32
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.5 Use of APEX-SWAT to Evaluate Construction Sediment Regulatory
Options: Case Study, Dallas, TX
6.5.1 Overview
The micro-watershed model APEX (Agricultural Policy - Environmental extender; Williams et al.
2008), in conjunction with the watershed-scale model SWAT (Soil and Water Assessment Tool; Arnold
et al. 1998), is being used to evaluate a range of regulatory options for mitigating sediment associated
with construction activities. The APEX-SWAT study is configured to complement the nationwide
SPARROW model evaluation described in Sections 6.1-6.3 of this document. The studies are
complementary in several respects: APEX-SWAT employs a physically based simulation modeling
approach to the evaluation of regulatory options, whereas SPARROW employs what is primarily a
statistically estimated nonlinear regression approach. The SPARROW study is national in scope, at the
spatial resolution of the Reach File 1 (RF1) stream network, while the APEX-SWAT approach involves a
case study of a single urban area - Dallas, Texas on the Trinity River - with spatial resolution at
plot/multi-field to sub-catchment scale. Finally, SPARROW is specified to predict the long-term annual
or seasonal mean contaminant loadings (long-term annual loading associated with specific land use and
management practices); and APEX-SWAT examines transient conditions, with respect to both land
disturbance and to weather and flow conditions.
The APEX-SWAT study is motivated by several insights emerging from the national SPARROW
sediment modeling study. The fundamental spatial unit within the national SPARROW model is the RF1
stream reach and associated catchment. Even in rapidly urbanizing regions, sediment loadings associated
with construction activities, while perhaps the dominant sediment sources within the specific tributary
reaches containing active construction sites, may contribute only modestly to observed sediment loadings
as measured or simulated at the RF1 level, depending on the magnitude of water and sediment
contributions from other tributaries and sediment sources within the RF1 catchment. Spatial averaging of
loadings at the RF1 level may obscure the degree to which significant changes (and hence potential
benefits) occur within specific construction-affected reaches in response to respective sediment abatement
practices.
As a related issue, while SPARROW is in principle able to account for the influences of the position of
the sediment source (construction site) within the landscape via statistical evaluation of land-to-water
delivery processes over wide areas, in practice the sensitivity of such insights is limited by the spatial
resolution of model landscape units. In the national SPARROW model, this resolution is 1 km2. By
contrast, APEX-SWAT simulation can be conducted over a range of spatial resolutions down to 1 hectare
or finer, if desired, and details of the overland flowpath, such as length, slope, hydraulic roughness, and
vegetative cover, can be simulated explicitly. Finally, APEX-SWAT simulation can be used to examine
the influences of rainfall intensity and duration, overlaid on variations in antecedent soil moisture and
seasonally determined vegetative cover conditions, on soil erosion and sediment mobilization associated
with a specific sediment abatement practice. Via continuous simulation, decades of daily (and sub-daily)
weather variability can be evaluated.
The sections below provide additional details on the APEX-SWAT analysis. Results from the analysis are
not available at this time. EPA plans to present the results from this analysis in a future version of this
document.
November 2008 6-33
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.5.2 Objectives
The primary objective of the APEX-SWAT model case study is to use continuous, dynamic simulation to
evaluate the respective water quality impacts of a range of sediment abatement options designed to reduce
or to eliminate the delivery to surface waters of sediments and associated contaminants originating on
construction sites. Related objectives are (1) to evaluate the influence of the construction site setting
within the landscape on loadings to the surface flow system; (2) to evaluate the influence of
meteorological characteristics of the study setting on sediment loadings; and (3) to assess the spatial
variation in sediment loadings at tributary scale within the larger watershed. Attributes of interest with
respect to landscape setting include proximity of the construction site to the nearest channel;
characteristics of the construction site (e.g., size, shape, slope) and of the intervening flowpath, including
length, slope, hydraulic roughness, and vegetative cover; and the impacts of buffers, filters or related
structures on sediment loadings. Attributes of interest with respect to meteorological conditions include
rainfall intensity, duration, inter-storm duration and antecedent soil moisture conditions. The intent is to
examine the combined influences of sediment abatement practices themselves, landscape attributes, and
meteorological conditions on sediment loadings, in order to acquire a more comprehensive understanding
of the likely water quality impacts of proposed sediment abatement options under heterogeneous
conditions.
6.5.3 APEX-SWAT
The farm- and field-scale model APEX (Williams et al. 2008) has been combined with the watershed
simulation model SWAT (Arnold et al. 1998) to provide a modeling framework within which detailed
evaluations of soil conservation, tillage, and nutrient management practices and their impacts on soil,
nutrient, and contaminant runoff at field scale can be integrated to provide estimates of water quality
impacts at watershed scale. APEX-SWAT is a central component of the U.S. Department of Agriculture
(USDA) Conservation Effects Assessment Program. This program was initiated to support the objectives
of the 2002 Farm Bill, and emphasizes the use of simulation modeling to quantify the environmental
benefits of conservation practices at the national scale (Santhi et al. 2005). The APEX-SWAT integration
was initially conducted in the context of the EPA-funded "Livestock and the Environment: A National
Pilot Project (NPP)" initiated in 1992 to evaluate livestock contributions to environmental problems at the
watershed scale (Gassman et al. 2005). The respective models were integrated within the "Comprehensive
Economic Environmental Optimization Tool - Livestock and Poultry" (CEEOT-LP) modeling system
developed within the context of the NPP (Gassman et al. 2002; Osei et al. 2000a; Osei et al. 2000b). The
basic approach used by APEX-SWAT is to simulate the impacts of specific soil and nutrient management
practices at field scale using APEX, and to input these edge-of-field sediment and nutrient loadings
(dissolved and advected) into SWAT. SWAT subsequently routes these loadings through the stream
network, within which their impact on ambient water quality can be evaluated (Gassman et al. 2005).
APEX
APEX is an evolution of the EPIC (Erosion-Productivity Impact Calculator; subsequently Erosion-Policy
Impact Climate) model (Williams et al. 1984). EPIC was developed from 1981 onward and used
extensively in the context of the Second Resources Conservation Act Appraisal (1985). EPIC contains
nine distinct simulation components: weather, hydrology, erosion, nutrients, soil temperature, plant
growth, plant environmental control, tillage and economic budgets. EPIC was designed as a field-scale
model, where a "field" is conceptualized as a parcel that is relatively homogeneous with respect to
weather, soil characteristics, topography, cropping patterns, and land management practices. It has been
November 2008 6-34
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
used extensively in a wide range of studies examining soil erosion and conservation, crop yield response
to inputs and management, nutrient and carbon cycling, water quality, and economic analysis of
conservation policies and practices (Gassman et al. 2005). EPIC is a continuous simulation model
operating at daily time steps. Soil erosion is calculated using one of nine established procedures,
including the Universal Soil Loss Equation (USLE; Wischmeier and Smith 1978), Modified Universal
Soil Loss Equation (MUSLE; Williams 1995) and Revised Universal Soil Loss Equation (RUSLE;
Renard 1997) for water (rainfall, hydraulic) erosion; and the Wind Erosion Stochastic Simulator (WESS;
Potter et al. 1998) for wind erosion.
APEX was developed primarily to enable the simulation of "farms" consisting of several "fields" as
defined above, permitting the physical and economic evaluation of farming systems that are
heterogeneous with respect to both physical and spatial attributes and management practices. APEX
utilizes EPIC procedures at field level, and combines these to enable, for example, the evaluation of
sediment and nutrient runoff as generated within one plot and attenuated by a second, down-slope plot
containing filter or buffer strips. Like EPIC, APEX is a continuous simulation model operating at daily
time steps. It possesses, in addition, the capacity to route runoff and associated contaminants between
sub-units.
SWAT
SWAT is a physical-conceptual model at watershed scale. It combines simulation components that are
rigorously physical with others that are conceptual or empirical. SWAT is a continuous simulation model
operating at daily time steps, and has been used successfully over a range of timeframes. Typical
applications involve annual or multi-year simulations, but SWAT has also been used to simulate the
impacts of global climate change over longer periods. It is a comprehensive watershed simulation
modeling system, integrating climate, hydrology, soil dynamics, biophysics, and water quality simulation
capabilities and encompasses both land-phase watershed processes and receiving waters. SWAT,
developed by the USDA Agricultural Research Service (ARS) (Neitsch et al. 2002, 2005), represents an
evolution of the Simulator for Water Resources in Rural Basins (SWRRB) model (Arnold et al. 1995) and
integrates components and algorithms from numerous pre-existing ARS models. These include
Chemicals, Runoff and Erosion from Agricultural Management Systems (CREAMS; Knisel 1980),
Groundwater Loading Effects on Agricultural Management Systems (GLEAMS; Leonard et al. 1987),
and EPIC (Williams et al. 1984). Current versions include SWAT 2000 (2001), integrated within EPA's
BASINS 3.0 framework, and SWAT 2005, the most recent version. BASINS is a multipurpose
environmental analysis system that integrates a geographic information system (GIS), national watershed
data, and a range of environmental assessment and modeling tools (USEPA 2008a).
The typical simulation domain of SWAT is a large watershed; from thousands of acres to thousands of
square miles. Watersheds are partitioned into sub-basins, each of which is a mass balance (e.g., water,
sediment, nutrient) accounting framework. Runoff, sediment, nutrients, and other associated constituents
are generated within the sub-basin according to physical process models, and are routed to the sub-basin
tributary via algorithms that reflect, among other factors, travel time and distance. From that point, they
enter the model channel routing model subsystem, which simulates process dynamics for individual water
quality constituents. There are no rigid or uniform criteria for subdividing watersheds into sub-basin
accounting units, although this is typically done on the basis of similarity in land use, climate, and related
factors. Similarly, there is no general restriction on sub-basin size distribution. A given SWAT model
implementation may consist of large, coarsely parameterized units mixed with smaller, more detailed
units.
November 2008 6-35
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Sub-basin surfaces are in turn subdivided into hydrologic response units (HRU), defined as unique
soil/land use/management combinations within the sub-basin. The HRU is the spatial unit for which the
land phase of the hydrologic cycle is simulated. Simulated outputs from each HRU (e.g., runoff,
suspended sediment, nutrients) are summed over the sub-basin and introduced to the sub-basin channel,
without regard to the spatial location of the HRU within that sub-basin. For situations in which it is
important to take into account the overland flowpath of water and associated loadings, sub-basins can be
further resolved into hill slope elements, which are connected via well-defined flowpaths as specified in a
routing table. Hill slope elements are sub-basins (rather than HRUs), and as such are autonomous water
and constituent accounting units. Hill slope discretization is used in evaluating the effectiveness of filter
strips. It is also possible to introduce a grid system in areas of special interest.
SWAT consists of numerous simulation modules, described in detail in the SWAT Theoretical
Documentation (Neitsch et al. 2002, 2005). Within the general sub-basin (HRU)-channel network
configuration, SWAT simulation modules include climate (including a weather generator); surface runoff;
evapotranspiration; soil water balance accounting; groundwater; nutrient and pesticide accounting (N, P
dynamics, bacteria); erosion and sediment transport; crop growth; agricultural land and water
management; urban runoff and contaminant wash-off; channel routing of water, sediments, nutrients,
bacteria, and heavy metals; and impoundments, including regulated and unregulated reservoirs, ponds,
and wetlands.
Surface runoff is simulated via two options. The first is the SCS Curve Number procedure (SCS 1972),
which predicts runoff as a function of daily rainfall, hydrologic soil group, land management practice, and
antecedent moisture conditions. The SCS curve number is a widely used empirical procedure, and is
useful for simulating changes in runoff associated with changes in any of the conditioning factors in a
computationally efficient manner. The second approach is based on the Green and Ampt (1911)
infiltration model, which involves an idealization of the complex physical process of soil water
infiltration. Precipitation is presumed to generate runoff at a rate equal to the difference between rainfall
intensity and calculated infiltration capacity. Peak runoff rate, important in the simulation of soil erosion,
is estimated via the rational method, where peak runoff is a function of rainfall intensity, watershed
contributing area, and time of concentration. For larger sub-basins, SWAT utilizes a surface runoff lag
procedure to simulate the overland routing of runoff to the sub-basin channel.
Erosion of soil is simulated in SWAT using the MUSLE (Williams 1995), an evolution of the widely used
USLE (Wischmeier and Smith 1978). It is expressed as:
SED = 114Qsurf • qpeak • AHRU )°56 • KUSLE • CUSLE • PUSLE • LSUSLE • CFRG, (Eq. 6-1)
where SED is the daily sediment yield (MT), Qsulfis surface runoff volume (mm/Ha), qpeak is peak runoff
rate (m3/s), Amu is the HRU area (Ha), K is the USLE soil credibility factor (empirical), C is the USLE
cover and management factor, P is the USLE support practice factor, LS is the USLE topographic factor,
and CFRG is the soil coarse fragment factor. Peak runoff and runoff volume are obtained as described
above, and other components are calculated within SWAT on the basis of fixed (soil) and temporally
varying (land cover, management) variables. In the land phase of SWAT, sediment moves with surface
runoff, subject to the same lags for larger spatial units.
Within the channel network component of SWAT, water is routed to the basin outlet on the basis of either
the variable storage routing or the Muskingum routing methods, respectively. In each case, rates and
velocities are determined by Manning's formula. Channel flows are subject to transmission and
evaporation losses, and to bank storage. SWAT versions through 2000 simulated sediment routing
November 2008 6-36
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
through channels on the basis of stream power theory. In SWAT 2005, a simplified form of channel
routing based on peak channel velocity is used. SWAT simulates four basic types of impoundments.
Ponds, wetlands, and depressions (potholes) are by definition located within sub-basins and do not
impound water flowing within the main channel system. Reservoirs are located on the main channel
network, and each is subject to a separate water and sediment balance. For reservoirs, release can either
be stipulated as a schedule, or releases can reflect daily, monthly, or average measured values. Outflow
from ponds is calculated as a function of "target" storage, which varies seasonally. Wetlands and
depressions release water whenever simulated volume exceeds capacity, or through release operations and
engineered drainage (e.g., tiles). All impoundment structures can act as sediment traps. Sediments
entering impoundments with inflow are assumed to be completely mixed, and subject to settling. Settling
occurs when sediment concentrations exceed equilibrium values, expressed as megagram per cubic meter
(Mg/m3) of detained storage. Settling is simulated by the following equation:
C \
^Sed,eq)
' sed,eq
(Eq. 6-2)
where C^/is the final sediment concentration (Mg/m3), Csedii is the initial concentration (Mg/m3), Csedigq
is the equilibrium concentration (Mg/m3), k is a decay constant (day"1), t is the model time step (1 day),
and dso is the median particle size of inflow sediment (|Jm).
SWAT is intended to provide a comprehensive toolbox for predicting the impacts of land management
decisions on water, sediment, nutrient, pesticide, and bacteria yields from predominantly rural and
agricultural watersheds, and for Total Maximum Daily Load (TMDL) analysis as a component of
BASINS. SWAT was designed to simulate large basins, but has been used effectively at a range of scales,
from field to major watershed. SWAT models have been implemented for the Upper Mississippi Basin
(Gassman et al. 2003) and over the entire coterminous United States, at basin scale. SWAT's quasi-
physical specification is designed to enable simulation of ungauged basins, although in practice, the
majority of published SWAT studies have involved calibration and validation on the basis of physically
measured data. SWAT's design and functionality enable the linkage of upland land management activities
influencing erosion with receiving water quality (sediment delivery) inclusive of channel routing and
storage. With respect to BMPs, SWAT by design possesses the capability to simulate (1) erosion control
via soil cover and management factors, (2) runoff control via flowpath modifications, and (3) sediment
control via settling basins (ponds, reservoirs) and vegetated buffers (filter strips).
6.5.4 Dallas-Fort Worth Study Setting
Dallas-Fort Worth was selected as the location for an APEX-SWAT case study, primarily on the basis of
an established SWAT model of the Trinity River watershed containing Dallas. Dallas, located at
approximately 32°, 47' N Lat; 96°, 48' W Lon.; and 131m elevation in Northeastern Texas, is a major
metropolitan area occupying roughly 385 mi2 and containing a population of approximately 1.2 million
(2006). Dallas is located within the Trinity River basin (USGS 08057000) in a generally low-relief
setting. The Trinity is contained within 15m levees through much of downtown Dallas, although efforts
are currently underway to redevelop the urban river corridor as parkway and recreational facilities. Dallas
has a humid sub-tropical climate, receiving around 37 inches (940 mm) of precipitation annually, with
springtime (March-June) the rainiest period. Severe thunderstorms are common, particularly in spring
and early autumn. Average temperatures range between 55 °F in January and 96°F in July. The Trinity
River drains 6,106 mi2 above Dallas. Annual average discharge ranges between 115 cubic feet per second
(cfs) (1956) and 7,154 cfs (1982), averaging 1,844 cfs for the 1931-2007 period (USGS 2008). The East
November 2008 6-37
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Fork of the Trinity, containing one of the sample sub-watersheds, lies to the east of Metro Dallas and
joins the Trinity River south and east of Dallas County.
The selection of sub-catchments within the Trinity and East Fork Trinity watersheds for high-resolution
modeling in APEX-SWAT was guided primarily by three criteria. First, sub-watersheds must contain
locations of recent significant urban development. Second, developed regions should not receive
extensive inflow from upstream of imputed urbanizing areas, as this would tend to obscure local
contributions to overall sediment (TSS) loadings. Finally, candidate sub-catchments should drain freely to
the main stem of the Trinity (or primary tributaries); and not to lakes, reservoirs, or other impoundments
that would complicate the simulation of sediment loadings downstream. Location and extent of urban
development are of primary significance in this process. The Multi-Resolution Land Characteristics
Consortium (MRLC) has developed comprehensive National Land Cover Databases (NLCD) for 1992
and 2001, respectively, derived from the Landsat Thematic Mapper-Extended Thematic Mapper imagery
at 30-meter resolution. The 1992 NLCD coverages were re-projected and re-classified to achieve
consistency with the 2001 coverages at 30-meter pixel level. Locations of land cover conversion from
nonurban to urban uses, and from pervious to impervious surface characteristics, were identified via
pixel-by-pixel comparison between the 1992 and 2001 datasets, respectively. Converted pixels serve to
identify areas where construction activities occurred within the 1992-2001 period. This dataset, identified
as the "NLCD 1992/2001 Retrofit Land Cover Change Product," is available at the MRLC Web site
(http://www.mrlc.gov). As nonurban to urban land conversion consists of (1) agriculture-to-urban,
(2) forest-to-urban, (3) grassland/shrub-to-urban, and (4) open water to urban categories, respectively, it
should be noted that this procedure omits at least one category of construction activity, resulting when
previously developed land is re-converted or re-developed. Thus, the procedure described identifies only
nonurban to urban land conversion.
6.5.5 Sources of Data
General datasets required for successful SWAT implementation include climate (e.g., precipitation,
temperature, radiation, humidity, wind speed); topography; soils; land use/land cover; stream reach
network; and, ideally, data on stream discharge and water quality to enable calibration and validation.
Most of these data are available from various sources including USGS hydrologic unit boundaries,
topographic data at 30-meter resolution from the USGS National Elevation Dataset, soils data from the
Soil Survey Geographic (SSURGO) database, and 1:24,000 land cover data from USDA's Natural
Resources Conservation Service. Alternatively, land cover data from NLCD 2001 are available (MRLC
2001). Gridded monthly climate data (e.g., precipitation, maximum and minimum temperature, dewpoint)
are available from the Parameter-Elevation Regression on Independent Slopes Model (PRISM; OSU
2007) for the entire United States. Monthly data are in general too coarse for most SWAT applications,
and daily or finer resolution data for specific stations are available from the NOAA National Climatic
Data Center. Weather data can also be simulated within SWAT, although an appropriate basis for
calibration of the weather generator is required. Measured discharge data are available from USGS, and
water quality data from USGS and EPA's STORET environmental data system, a repository for water
quality, biological, and physical data (USEPA 2008f). Data for specific locations of construction activities
will be gathered on a case-by-case basis, and probably represent the binding constraint to data acquisition
for detailed SWAT simulations of construction activity.
November 2008 6-38
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
6.5.6 Experimental Design
The basic methodological approach to evaluating regulatory options for construction site sediment
abatement consists of (1) simulating a water quality baseline, and subsequently (2) generating alternative
scenarios corresponding with specific sediment abatement options. The baseline assumes that
construction activities occur at locations within the target sub-basins identified from the NLCD
1992/2001 Retrofit Land Cover Change Product ("retrofit product"). The retrofit product, overlaid on the
local flow network and sub-watershed boundaries, appears in Figure 6-6 (Trinity River within Dallas
County) and Figure 6-7 (East Fork Trinity River). Because areas of nonurban to urban land conversion
span the 1992-2001 period, an assumption is made that the progress of development was uniform over
this period. Assuming that approximately 10 percent of the developed land identified in the retrofit
product is subject to development in a given year during this period (based on a further assumption that
soils on developed parcels are exposed to erosive forces for one year or parts of one year only), an
additional set of assumptions is required to estimate the distribution of areal extent of such developed
parcels. This is accomplished on the basis of a construction site parcel size distribution discussed in
EPA's Development Document for Proposed Effluent Guidelines and Standards for the Construction and
Development Category (2008). Locations of target parcels within the sample sub-watersheds are selected
to represent a range of settings, with emphasis on distance to watercourse and, to the extent applicable,
slope, topography and soils. For the baseline simulation, no sediment remediation is assumed, so that the
extent of sediment mobilization and delivery are determined by soil and other site characteristics,
topography, rainfall erosivity, and related physical factors.
Once the sample parcels have been specified, the baseline simulation consists of a 30-year continuous
simulation using meteorological records from the study area, augmented by the SWAT weather generator
as required. Loadings of sediment, measured as Total Suspended Solids (TSS), are recorded at the outlets
of each of the sample sub-catchments; and at representative points along the main stem of the Trinity
River above and below the points of confluence. Under the baseline simulation, the locations and
characteristics of sample parcels remain unchanged, since the objective is to evaluate the impacts of
weather, season, and antecedent soil moisture conditions, respectively. Each successive scenario then
involves the substitution of a reduced sediment loading, corresponding to a specific regulatory option (see
description of options in Chapter 5.
A useful output is the spatial distribution of water quality (loadings) impacts, which will reflect both the
spatial distribution of construction activities, the impacts of specific regulatory options, and other factors
contributing to the background sediment loadings within the larger watershed.
November 2008 6-39
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-6: Land Conversion Watershed Boundaries and River Network, Dallas County
LULC Change
Streams ^^ Agriculture to Urban
j j County Boundary ^^) Forest to Urban
Watertaodies and Wetlands H| Grassland/Shrubto Urban
Subvyatersried Used for Modeling • • Open Water to Urban
| | Analysis Watershed Boundary
25 5 75
November 2008
6-40
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 6-7: Land Conversion Watershed Boundaries and River Network, East Fork, Trinity River
Kaufman^
./
LULC Change
^B Agriculture to Urban
[""j County Boundary ^^| Forest to Urban
Waterbodies and Wetlands ^H Grassland/Shrubto Urban
Subwatersried Used for Modeling ^^| OpenWaterto Urban
Analysis Watershed Boundary
\0 1.25 2.5 3.75 5
November 2008
6-41
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Benefits to Navigation
Navigable waterways, including rivers, lakes, bays, shipping channels and harbors, are an integral part of
the United States' industrial transportation network. Navigable channels are prone to reduced
functionality due to sediment build-up, which can reduce the navigable depth and width of the waterway
(Clark et al. 1985). Removing sediment to keep navigable waterways passable requires dredging. The
U.S. Army Corps of Engineers (USAGE) spends an average of more than $572 million (2008$) every
year to dredge waterways and keep them passable (USAGE 2007).
Implementation of the C&D regulation is expected to result in less frequent dredging of navigable
waterways due to reduced sediment runoff from construction sites and. This will result in cost savings to
the government and private entities responsible for maintenance of navigational shipping channels,
harbors, rivers, and other waterways. For example, prior studies of benefits of the Conservation Reserve
program found that a reduction of 205 million tons of sediment erosion from farmland would result in
approximately $526 million (2008$) in benefits to navigation from decreased dredging (Ribaudo 1989).
In addition to the avoided costs when the need for dredging is reduced, there are also environmental
benefits. Dredging is itself an environmentally disruptive activity because dredged material may contain
significant quantities of toxic substances and heavy metals. Dredging can disturb contaminants that have
settled to the bottom of the waterway, reintroducing the possibility of their uptake by fish and other
aquatic organisms. Dredged sediment may be disposed of in another section of the water body or
watershed, relocating the problem rather than removing it. Additionally, if the sediment removed from a
site is contaminated, it could add to pollution at the disposal site (Clark et al. 1985). Finally, unless there
is overdredging to compensate or sedimentation is monitored so that dredging activity may be timed
optimally, water bodies will be on average less navigable even with a dredging program.
This chapter presents EPA's analysis of the navigable waterway maintenance costs that would be avoided
by implementation of the C&D regulation. This approach represents an appropriate measure of social
benefits for cases in which a policy change reduces costs to producers (in this case "producers" of
navigable waterways), but in which price effects are minimal (cf Boardman et al. 2001, p. 70-74).
Additional environmental benefits of reduced dredging are not estimated due to the highly place-specific
nature of these benefits and difficulty in estimating these costs over the national scale. Hence, in this
regard the Agency's benefit estimate in this chapter is understated. This analysis includes the following
steps:
> Identifying navigable waterways that are regularly dredged and estimating the frequency of
regular dredging in each waterway
> Estimating the navigable waterway maintenance cost per cubic yard of sediment dredged
> Estimating the total cost of navigable waterway maintenance under the baseline and post-
compliance scenarios
> Estimating cost saving from decreased dredging of navigable waterways due to the reduction in
sediment runoff from construction sites.
November 2008
7-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
7.1 Data Sources
EPA relied on the USAGE Dredging Information System in analyzing benefits to navigation from the
proposed C&D regulation. The dredging database catalogs all USAGE dredging contracts from 1990 to
2006 and all USACE-conducted dredging jobs from 1995 to 2006, and provides information on the
location, dates, cost, and amount of sediment dredged for each dredging job or contract. For the purpose
of this analysis, a "dredging occurrence" refers to a single record in these data, containing a location and
time of dredging, while a "dredging job" refers to a single location of dredging that may have more than
one dredging occurrence, and thus be repeated in the data.
Where the USAGE Dredging Information System did not contain cost or quantity of sediment dredged for
a listed dredging occurrence, EPA estimated the missing information from other jobs in the same EPA
region. For each region, the Agency calculated 10th, 50th, and 90th percentile costs and quantity boundaries
and used these to fill in all incomplete records in that region. EPA included 50th percentile (median)
estimates in the summary data presented in Table 7-1 through Table 7-4 and in the midpoint estimate of
total dredging costs. The 10th and 90th percentile cost and amount dredged estimates were used in the low
and high total dredging cost estimates, respectively. For more details, see Section 7.4.3: Sensitivity
Analysis.
Table 7-1 summarizes dredging information for dredging jobs for the period 1995-2006, including
dredging performed by USAGE and dredging contracted to other firms. The cost data were adjusted to
2008$ using the construction costs of the producer price index (PPI) (Construction Cost Index 2008).
During this period, USAGE funded more than 3,700 dredging occurrences, worth more than $8.4 billion
(2008$). This dredging removed more than 2.4 billion cubic yards of sediment from more than 1,100
navigable reaches. The region with the most dredging was Region 4, reporting more than 1,000 dredging
occurrences between 1995 and 2006, nearly 100 per year. Region 6 reported the largest volume of
sediment removed, with more than 1.3 billion cubic yards removed from navigable waterways over this
12-year period. Though benefits are also dependent on the amount of sediment reduction in a region,
regions with more frequent dredging are more likely to experience greater benefits from reduced
sedimentation in navigable waterways.
November 2008 7-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-1 :
EPA
Region
1
2
o
3
4
5
6
7
8
9
10
Total
Source: Dredg
Dredging in
Number of
Dredging
Occurrences
75
307
283
1,037
703
801
21
1
134
352
3,714
U.S. Navigable Waterways, 1995-2006
Total Sediment
Removed
(millions of cubic
yards)
3
106
113
510
104
1,370
11
1
66
152
2,435
Total Cost
(millions of 2008$)
$40
$754
$800
$2,954
$416
$2,466
$25
$2
$442
$551
$8,451
ing Information System (USAGE 2007)
Table 7-2 summarizes dredging activity from 1995 through 2006 by type of entity performing dredging
work (private or government). As noted above, USAGE spent more than $8.4 billion (2008$) over the
past 12 years to fund more than 3,700 dredging occurrences. The majority of dredging work (57 percent)
was contracted out, with USAGE performing about 43 percent of the work over this period. The
individual costs of these dredging occurrences range considerably, from less than $10,000 to more than
$100 million, but averaging about $2.6 million (2008$).
Table 7-2: Dredging of Navigable Waterways Performed or Contracted by USAGE,
1995-2006
USAGE
Contract
Total
Source: Dredg
Reported
Occurrences
of Dredging
1,613
2,101
3,714
Percent of
Total
Reported
Dredging
Occurrences
43.4%
56.6%
100.0%
Reported
Sediment
Dredged
(Millions of
cubic yards)
485
1,950
2,435
Percent of
Total Reported
Sediment
Dredged
19.9%
80.1%
100.0%
Total
Reported
Cost
(Millions
of 2008$)
$1,828
$6,622
$8,451
Percent
of Total
Reported
Cost
21.6%
78.4%
100.0%
ing Information System (USAGE 2007)
7.2 Identifying Waterways That Are Dredged and the Frequency of Dredging
7.2.1 Determining Dredging Job Locations
Location information is given by latitude-longitude coordinates for some jobs, but for others, location
information was limited to the name of the job (usually the waterway dredged) and the USAGE district
that performed the job. To identify the nearest reach segment, EPA used an unprojected version of the
ERF 1.2 (Enhanced Reach File) from USGS. For this analysis, EPA researched latitude-longitude
November 2008
7-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
coordinates for jobs where they were not provided and linked them to Reach File Version 1.0 (RF1)
reaches. Each latitude/longitude of interest was matched to the nearest point in the ERF 1.2 universe of
points using a spherical model of the earth and a standard haversine distance formula. To associate the
point with an RF1 reach number, EPA used the "RR" (River Reach) attribute from the ERF 1.2 attribute
table. No reach types were excluded from consideration in the nearest reach calculation. Because these
coordinates may be inexact, dredging activity was aggregated to the state level for the calculations in this
analysis and is presented at a regional level.
7.2.2 Identifying the Frequency of Dredging
Continuous sediment deposition in a navigable waterway is likely to require repeated dredging to
maintain navigability. Between 1995 and 2006, 1,166 sites required dredging on at least one occasion,
and 475 required dredging more than once. Table 7-3 shows these statistics for each EPA region. The
number of dredging jobs varies by region, and is likely to be influenced by the size of the region, its
number of navigable waterways, and their economic importance. Because the data available from USAGE
reflect a relatively short period of time (12 years) and dredging data show a large degree of variability in
the frequency of dredging occurrences, EPA calculated an average frequency of recurrence for each
dredging job by dividing 12 years by the number of occurrences of each job over these 12 years.6 The
EPA-estimated regional average recurrence intervals range from 8 to 12 years, including jobs that
occurred only once in 12 years. For all dredging jobs in the United States, the average frequency of
recurrence is approximately 9 years. Excluding jobs that occurred only once over this period, the regional
average recurrence interval ranges from 3.5 to 6 years, averaging slightly less than 4 years at the national
level.
Prior studies (Clark et al. 1985; Ribaudo 1989) of the cost savings from reduction in sedimentation of
navigable waterways, shipping channels, and harbors assumed a linear relationship between sediment
runoff and dredging activity, implying that every ton of sediment entering navigable waterways would
eventually be dredged up. Because dredging is an expensive and environmentally disruptive activity, it is
unlikely that every navigable waterway that receives sediment will eventually be dredged.
For this analysis, EPA assumed that navigable waterways that are regularly dredged would benefit from
reducing the amount of sediment that needs to be dredged up. Although other waterways may benefit
from reduced sedimentation, EPA estimated based its estimation of reduced dredging activity based on
historic dredging data from USAGE. For dredging jobs that occurred only once between 1995 and 2006,
EPA assumed that this job or one similar to it will occur over the same time frame in the future. However,
these single occurrence jobs are excluded from the low range estimate of the analysis (see Section 7.4.3:
Sensitivity Analysis).
As described later in this section, EPA used the number of days between dredging occurrences and imposed a minimum
number of intervening days for a dredging occurrence to be considered unique. Occurrences separated by fewer than this
minimum number of days were considered as a single occurrence.
November 2008 7-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-3:
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Dredging Jobs and
Number of Jobs
41
92
118
268
277
211
16
1
48
94
1,166
Recurrence
Number of
Recurring
Jobs
10
43
42
100
126
96
1
0
20
37
475
Intervals, 1995-
Average
Interval
(years)
10.0
8.2
9.3
8.8
8.4
8.1
11.6
12.0
8.6
8.8
8.7
2006
Average Interval
for Recurring
Jobs (years)
3.8
3.9
4.5
3.5
4.0
3.5
6.0
N/A
3.8
4.0
3.8
Source: USAGE Dredging Information System (2007)
EPA used the USAGE dredging data to identify navigable waterways that are dredged (dredging jobs), as
well as the frequency at which dredging activity occurs. For each dredging job included in the analysis,
the Agency identified:
> The number of occurrences of this dredging job
> The time elapsed between occurrences of dredging.
EPA excluded from this analysis any occurrences where the time elapsed since the last dredging
occurrence was less than 30 days due to uncertainty as to whether these occurrences were continuations of
an earlier occurrence of dredging. From this information, EPA established a dredging schedule for each
recurring dredging job for the next 12 years. For the 691 jobs that only occurred once from 1995 to 2006,
the number of future dredging jobs is set to one, except in the low estimate, where it is set to zero (see
Section 7.4.3: Sensitivity Analysis). For other dredging jobs, EPA divided 12 years by the number of
occurrences of that job between 1995 and 2006 to obtain an average interval of dredging for that job in
years.
7.3 Estimating the Navigational Maintenance Cost per Cubic Yard of Sediment
Removed
Table 7-4 shows the average cost of dredging for each EPA region. Average unit dredging costs vary
considerably over these regions, from around $1.80 per cubic yard in Region 6 to more than $ 11 per cubic
yard in Region 1. The average unit cost of dredging for the entire United States is $3.40 per cubic yard.
November 2008 7-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-4: Costs of Recurring Dredging, 1995-2006
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
Total
Total Sediment
Total Cost Removed Average Cost
(millions of 2008$) (millions of cubic yards) per cubic yard
$44
$930.7
$932.4
$3,449.4
$541.4
$2,938.0
$26.0
$2.3
$500.6
$599.6
$9,963.8
3.9
131.6
137.5
639.9
120.8
1,644.2
10.9
0.5
76.3
162.1
2,927.7
$11.11
$7.07
$6.78
$5.39
$4.48
$1.79
$2.39
$4.49
$6.56
$3.70
$3.40
Source: USAGE Dredging Information System (2007)
Though USAGE does not disaggregate its reported dredging costs, the cost of sediment dredging typically
includes the following components (Sohngen and Rausch 1998):
> The cost of dredging sediment from the bed and loading onto the boat
> The cost of transporting the material to a disposal facility
> The cost of confining or disposing of the dredged sediment.
7.4 Estimating the Total Cost of Navigable Waterway Maintenance Under the
Baseline and Post-Compliance Scenarios
7.4.1 Estimating the Reduction in Sediment Dredging in Navigable Waterways Due to
the Reduction in Discharge from Construction Sites
EPA estimated changes in sediment deposition in navigable waterways under different regulatory
scenarios using output from SPARROW. Chapter 6 provides a description of SPARROW as well as input
data used in this analysis. Using SPARROW output, EPA first determined sediment settlement in each
state under the baseline scenario (no regulation in place), and then for the different policy options. The
amount of sediment dredged in a state under a particular post-compliance scenario was divided by the
amount dredged in the baseline scenario to obtain the percentage of sediment requiring dredging in that
state under the post-compliance scenario. This calculation was performed at the state level rather than the
reach level to adjust for any error in assigning river reach identifiers to dredging jobs.
EPA multiplied this statewide percentage by the average quantity of sediment dredged by each dredging
job in a state to obtain an estimate of the amount of sediment requiring dredging by each occurrence of
that job in the post-compliance scenario.
7.4.2 Estimating the Total Cost of Navigable Waterway Maintenance Under the Baseline
and Post-Compliance Scenarios
EPA estimated future dredging costs over a period of 12 years from 2009 to 2021 under both the baseline
and post-compliance scenarios. For this cost calculation, average cost per cubic yard of sediment dredged,
November 2008 7-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
total amount of sediment dredged, and the average interval between dredging jobs were calculated for
each relevant job in the USAGE Dredging Information System. For the post-compliance scenarios, total
amount of sediment dredged was determined as described in Section 7.4.1: Estimating the Reduction in
Sediment Dredging in Navigable Waterways Due to the Reduction in Discharge from Construction Sites.
Because costs will occur whenever the water body is dredged rather than on an annual basis, each reach
will have a unique stream of costs derived from its individual dredging frequency, cost, dredging volume,
and reduction in sediment. Thus, EPA calculated the present value of dredging costs over the 12-year
period using a dredging schedule for each dredging job and then discounted the estimated values using
both 3 and 7 percent annual interest rates. EPA estimated the present value of each dredging job as
follows:
> Estimating the average recurrence interval for the job over the 12-year period for each job by
dividing 12 years by its number of occurrences in the USAGE dredging data. This produces / in
the equation below.
> Entering this information along with the cost per cubic yard dredged, number of cubic yards
dredged, and percentage of sediment requiring dredging in the post-compliance scenario into the
formula below:
12/1
^=z
'Oh*R *C~
^ , pc . (Eq. 7-1)
l + (d*I? \
Where:
PV = Present Value
Z is the sum of the expression in brackets for the number of dredging occurrences of this
dredging job
/ = recurrence interval in years (including fractions of a year) for this dredging job
12/7 = number of periods between recurrence intervals for which costs will be discounted
Qb = cubic yards dredged under the baseline scenario
Rpc = percentage of cubic yards remaining in the post-compliance scenario
C = cost per cubic yards dredged
d = annual discount rate 7
d*I = the discount rate for the period, being the annual rate d prorated for the length of the
period / in years
n = the number of the specific dredging occurrence.
7 EPA estimated costs using both 3 and 7 percent discount rates, in accordance with OMB (2003, 33-34).
November 2008 7-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
For example, if the Mississippi River is scheduled to be dredged every 2 years between 2009 and 2021,
there will be no dredging costs for year 1, but the dredging costs in year 2 will be equal to the cost per
cubic yard multiplied by the amount of sediment dredged, then discounted by 2 years. There will again be
no dredging costs for year 3. Dredging costs will then be incurred in year 4 as in year 2, but discounted by
4 years, and so on. By the end of 2021, the Mississippi River will have been dredged six times. The sum
of the discounted costs of these six dredging occurrences is the total present value of the dredging cost of
the Mississippi River for the next 12 years. Repeating this process for every RF1 reach in this analysis
yields an estimate of national dredging costs over the next 12 years.
Alternately, the frequency can be interpreted as occurring at the beginning of each interval rather than at
the end, so that in the example above, the Mississippi River would be dredged in year 1, then in year 3,
etc., and discounted accordingly. This approach was taken for EPA's high estimate (see below).
7.4.3 Sensitivity Analysis
To account for uncertainty in projecting future dredging costs, EPA adjusted the assumptions described in
Section 7.2: Identifying Waterways That Are Dredged and the Frequency of Dredging to provide a range
of benefits estimates. EPA estimated low, midpoint, and high levels of dredging activity for this
sensitivity analysis as follows:
> Low estimate:
D Assuming a minimum recurrence interval (below which occurrences are assumed to be
continuations of the previous occurrence of the same job) of 180 days
° Assigning 10th percentile cost and cubic yards dredged estimates for jobs lacking these
data
° Excluding jobs with only one occurrence over the 12 years from 1995 to 2006
D Discounting dredging occurrences from the end of each occurrence interval.
> Midpoint estimate:
° Assuming a minimum recurrence interval of 90 days to include jobs with more closely
spaced recurrences
D Using 50th percentile (median) cost and cubic yards dredged estimates for jobs lacking
these data
° Including jobs with only one occurrence over the 12 years from 1995 to 2006 under the
assumption that these jobs or similar jobs will occur once over the next 12 years
D Discounting dredging occurrences from the end of each period as in the low estimate.
> High estimate:
° Assuming a minimum recurrence interval of 30 days,
° Using 90th percentile cost and cubic yards dredged estimates for jobs lacking these data,
° Including jobs with only one occurrence between 1995 and 2006,
D Discounting dredging occurrence at the beginning of each period.
November 2008 7-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
This analysis also presents costs for all estimates using both 3 and 7 percent discount rates.
7.4.4 Annualizing Future Dredging Costs
Though dredging costs will not be incurred at a regular annual rate, EPA annualized its 12-year cost
projection in order to facilitate the comparison of these costs to other costs calculated for the analysis of
the C&D regulation that are measured on an annual timeframe. This annualized cost was calculated using
the same interest rates used to discount future dredging costs. Table 7-5 presents estimated total
annualized dredging (cubic yards removed) and the estimated costs for navigable waterway dredging for
the baseline scenario, including low, midpoint, and high estimates.
Under the baseline scenario, midpoint dredging costs projected over the period of 2009 to 2021 are
expected to range between $516.8 and $530.5 million per year (annualized using a 7 and a 3 percent
discount rate, respectively). The high and low estimates produce an overall range between $342.4 million
and $1.1 billion. Regions 4 and 6 are expected to incur the highest costs, as they have the largest
percentage of dredging activity and amounts of sediment dredged. In the low estimate, Region 8 incurs no
dredging costs because the sole dredging job recorded in Region 8 in the USAGE Dredging Information
System occurs only once and was thus excluded. Regions 1 and 7 also have considerably lower dredging
costs than other regions, due to the small number of recorded dredging jobs in these areas.
November 2008 7-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-5: Annualized Dredging Costs Under the Baseline Scenario (millions of 2008$)
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Sediment Dredged
(millions of cubic yards)
Low
0.1
3.7
6.4
28.1
5.9
66.3
0.1
0.0
4.1
3.3
118.0
Mid
0.2
6.9
8.7
37.4
7.9
83.1
0.8
0.0
4.8
5.7
155.5
High
0.4
13.1
12.4
56.1
9.1
115.2
0.9
0.0
7.1
10.9
225.2
Cost using 3% discount rate
(millions of 2008$)
Low
$0.4
$28.0
$45.0
$115.2
$22.6
$94.3
$0.1
$0.0
$26.3
$10.5
$342.4
Mid
$3.0
$52.2
$59.7
$191.1
$31.0
$141.9
$1.7
$0.2
$31.4
$18.4
$530.5
High
$13.0
$107.1
$96.5
$442.1
$47.7
$239.7
$2.5
$0.2
$47.5
$38.6
$1,034.8
Cost using 7% discount rate
(millions of 2008$)
Low
$0.4
$27.4
$44.3
$113.9
$22.2
$93.4
$0.1
$0.0
$26.0
$10.4
$338.1
Mid
$2.8
$50.5
$58.1
$186.1
$30.1
$138.9
$1.6
$0.2
$30.7
$17.9
$516.8
High
$15.3
$120.2
$108.8
$485.5
$53.3
$263.0
$3.1
$0.3
$52.3
$43.0
$1,144.8
Source: USEPA analysis
November 2008
7-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
7.5 Estimating the Cost Savings from Decreased Dredging of Navigable
Waterways
The cost savings for each post-compliance scenario are calculated as the difference in total annualized
dredging costs between the baseline and each post-compliance scenario, and are considered the benefits to
navigation resulting from the C&D regulation. Table 7-6, Table 7-7, and Table 7-8 present annualized
avoided cost estimates for navigable waterway dredging for each of the Agency's policy options,
including low, midpoint, and high estimates for cost reductions under each policy option. Each of these
estimates was calculated using both 3 and 7 percent rates to discount and annualize costs.
Annualized savings from reduced dredging activity range from $690,600 to $61.5 million, with Option 2
representing a savings of $12.6 to $12.9 million in the midpoint estimate. EPA estimates that Region 4
will benefit from the most substantial reductions in dredging costs under all policy options. This is due to
a large amount of dredging activity in this region, and a large percentage reduction in sediment runoff
expected as a result of the C&D regulation. Region 6, though it does account for a large portion of
dredging activity, does not represent as large a portion of the cost savings due to its substantially lower
cost per unit of dredging ($1.79, compared to $5.39 in Region 4) and larger number of construction sites
exempted by rainfall and soil type waivers. Due to the lack of significant dredging activity in Region 8,
no benefits are expected in this region.
Option 1, a construction general permit requiring sedimentation basins on all sites larger than 10 acres
and the implementation of best management practices on smaller sites, is EPA's least stringent policy
option. It is predicted to produce a range of cost savings between $690,000 and $1.8 million. Under this
policy option, Region 4 will benefit from the most significant savings, accounting for almost all of the
benefits under this option. This option does not predict any savings in Regions 7, 8, or 9.
Option 2 imposes a turbidity standard on construction sites larger than 30 acres in areas that have an "R-
factor" (see Chapter 5 for more details) of greater than 50 and with soils containing greater than 10
percent small particles, and requires a construction general permit for all other sites over one acre. The
turbidity standard would require on-site treatment of stormwater, which would reduce TSS and turbidity
in discharges from sites meeting the criteria for the turbidity standard. This option will prevent an
estimated 3.3 million cubic yards of sediment from entering navigable water bodies and requiring
dredging. The midpoint estimate for cost savings under this option is between $12.6 and $12.9 million per
year, ranging from $8.9 million to $23.8 million between the low and high estimates. EPA expects more
than 99 percent of these benefits to accrue to Regions 1 through 6, as many areas in Regions 7 through 10
are exempt through soil type or rainfall waivers provided by this option. Region 4 again represents a
significant portion of savings, though Region 6 is also expected to benefit from a reduction of more than
$1.6 million per year in dredging costs.
November 2008 7-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-6: Annualized Reductions in Dredging and Costs Under Option 1
EPA
Region
I1
2
3
4
5
6
71
8
91
10
Total
Reduction in Sediment Dredged
(thousands of cubic yards)
Low
0.0
0.2
0.2
169.3
0.2
1.1
0.0
0.0
0.0
0.0
171.0
Mid
0.0
0.3
0.3
204.0
0.2
1.3
0.0
0.0
0.0
0.0
206.1
High
0.0
0.5
0.4
302.3
0.3
1.8
0.0
0.0
0.0
0.0
305.4
Avoided Costs Using 3% Discount Rate
(thousands of 2008$)
Low
$0.1
$1.2
$0.7
$686.6
$0.7
$1.3
$0.0
$0.0
$0.0
$0.0
$690.6
Mid I
$0.4
$2.2 :
$1.3 :
$962.8
$0.9 :
$2.0 ;
$0.0
$0.0
$0.0 !
$0.0 :
$969.8
High
$1.8
$4.5
$2.5
$1,615.8
$1.5
$3.6
$0.0
$0.0
$0.0
$0.0
$1,629.7
Avoided Costs Using 7% Discount Rate
(thousands of 2008$)
Low
$0.1
$1.1
$0.7
$676.7
$0.7
$1.3
$0.0
$0.0
$0.0
$0.0
$680.7
I Mid I
$0.4
: $2.2 :
$1.3 :
$937.2
: $0.9 :
i $2.0 ;
$0.0
$0.0
; $0.0 1
: $0.0 :
$943.9
High
$2.0
$4.8
$2.7
$1,741.7
$1.6
$3.8
$0.0
$0.0
$0.0
$0.0
$1,756.6
1 Reductions in dredged sediment and costs in these regions are not zero, but not sufficiently large to show at this level of significant digits
Source: USEPA analysis
Table 7-7: Annualized Reductions in Dredging and Costs Under Option 2
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Reduction in Sediment Dredged
(thousands of cubic yards)
Low Mid High
0.2 0.8 1.2
29.6 53.6 98.9
79.8 112.6 162.1
1,744.6 2,125.3 3,136.1
30.0 39.9 47.2
744.7 951.3 1,314.2
0.7 4.0 4.8
0.0 0.0 0.0
2.7 3.2 4.8
35.4 52.9 84.5
2,667.8 3,343.6 4,853.9
Avoided Costs Using 3% Discount Rate
(thousands of 2008$)
Low Mid High
$1.3 $9.3 $39.0
$214.1 $408.1 $816.1
$519.5 $720.5 $1,172.9
$6,851.2 $9,759.3 $16,889.0
$109.6 $145.2 $206.9
$1,104.8 $1,669.5 $2,682.8
$0.5 $7.9 $11.7
$0.0 $0.0 $0.0
$17.8 $21.2 $31.1
$95.5 $161.9 $283.8
$8,914.4 $12,903.0 $22,133.3
Avoided Costs Using 7% Discount Rate
(thousands of 2008$)
Low Mid High
$1.3 $8.8 $44.5
$210.2 $395.0 $878.6
$512.0 $700.7 $1,276.4
$6,758.1 $9,505.3 $18,161.5
$107.7 $141.0 $222.9
$1,093.2 $1,632.0 $2,846.0
$0.5 $7.3 $13.9
$0.0 $0.0 $0.0
$17.5 $20.7 $33.0
$94.7 $157.2 $304.5
$8,795.3 $12,568.1 $23,781.3
Source: USEPA analysis
November 2008
7-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 7-8: Annualized Reductions in Dredging and Costs Under Option 3
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Reduction in Sediment Dredged
(millions of cubic yards)
Low
0.6
45.5
118.7
3,586.9
49.7
1,100.1
1.0
0.0
22.7
137.9
5,063.0
1 Mid
1.9
I 82.5
167.9
4,401.9
65.5
1,411.8
5.9
! o.o
26.7
203.8
6,368.0
1 High
3.0
I 153.1
241.6
6,741.2
77.4
1,949.1
7.2
i 0.0
39.6
319.6
9,531.7
Avoided
Low
$3.2
$330.5
$764.8
$15,463.9
$215.9
$1,649.5
$0.8
$0.0
$147.2
$367.3
$18,943.2
Costs Using 3% Discount Rate
(thousands of 2008$)
I Mid
1 $22.9
I $628.1
: $1,066.6
$21,927.8
$280.1
$2,494.1
! $11.7
1 $0.0
! $175.3
$620.1
$27,226.8
I High
1 $97.7
I $1,255.1
: $1,740.5
$36,849.6
$396.2
$3,985.9
i $17.4
; $0.0
; $257.5
$1,069.6
$45,669.6
Avoided
Low
$3.2
$324.5
$753.8
$15,248.8
$212.5
$1,631.9
$0.8
$0.0
$145.1
$364.2
$18,684.7
Costs Using 7% Discount Rate
(thousands of 2008$)
1 Mid
$21.7
I $607.8
$1,037.0
$21,345.9
$272.7
$2,437.2
$10.9
; $0.0
$171.7
$602.2
$26,507.1
High
$111.1
$1,352.0
$1,894.3
$39,776.9
$425.4
$4,232.8
$20.7
$0.1
$273.1
$1,147.5
$49,233.9
Source: USEPA analysis
November 2008
7-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Option 3, which imposes a turbidity standard on all sites larger than 10 acres (in addition to the
requirements of Option 1) and requires a construction general permit for others, is EPA's most stringent
policy option. This option would require on-site treatment of stormwater, substantially reducing its
contribution to in-stream TSS and turbidity. Cost savings from this action range between $18.6 and $49.2
million, with a midpoint estimate of $26.5 to $27.2 million. Region 4 is again the largest beneficiary from
this regulation, with $21.3 to $21.9 million in savings for the midpoint estimate, and Regions 3 and 6 also
benefit from $1.0 to $1.1 million and $2.4 to $2.5 million in cost savings, respectively, under this option.
The largest changes from Option 2 come in Region 9, which had many areas exempt due to rainfall under
Option 2. Region 10 also sees a substantial increase in its benefits under this option, with the midpoint
estimates rising from just over $157,200 and $161,900 (7 percent and 3 percent discount rates,
respectively) under Option 2 to $602,200 and $620,100, respectively, under Option 3.
Overall, Regions 4 and 6 are the largest beneficiaries of reduced dredging costs from this action, both in
terms of cost savings and in the amount of sediment prevented from entering and settling in navigable
waterways. This is likely an effect of the large portion of coastline located in these regions and their large
proportion of dredging activity.
7.6 Sources of Uncertainty and Limitations
> The USAGE dredging database identifies dredging jobs by name, which is usually the name of
the water body dredged. However, the data lack standardized naming conventions, so it is
possible that the same water body is dredged under different job names. This may result in the
exclusion of dredging job names that only appear once in the database, but in fact were carried
out in the same water bodies as a differently named job, which would result in a downward bias
in EPA's dredging frequency calculations and the project costs.
> The navigable waterway data provide latitude/longitude information for some dredging jobs,
which are used to link dredging jobs to RF1 reaches, but these data are incomplete. In cases
where latitude/longitude information was not available for a particular job, EPA matched it to an
RF1 reach using the job name. This is a potential source of inaccuracy, as the job name is often
the waterway name, and may not be very specific (in cases such as the Mississippi or Colorado
rivers). It is unclear whether this would lead to an over- or underestimate of benefits.
> The cost per cubic yard and interval data obtained from the dredging database vary significantly,
even for different occurrences of dredging in the same water body or within regions. Aggregating
such highly varied data and using regional averages may bias the cost and interval estimates. The
direction of this bias is, however, uncertain.
> The analysis assumes a fixed cost per unit for dredging within each waterway, and that the rate of
dredging (i.e., years between each dredging event) remains unchanged as a result of the C&D
regulation. That is, the only change is assumed to be the quantity of sediment dredged during
each dredging event. It is, however, possible that dredging frequency may be reduced in response
to reduced sediment loads, rather than maintaining historical schedules with reduced quantities
dredged during each event. If dredging frequency is reduced or if per-unit costs of dredging are
nonconstant (e.g., there are fixed costs or nonconstant marginal costs of dredging), cost
reductions may diverge from estimates presented above.
November 2008 7-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
8 Benefits to Water Storage
Reservoirs are water impoundments, often manmade, that serve many functions, including providing
drinking water, flood control, hydropower supply, and recreational opportunities. Sediment in streams can
be carried into reservoirs, where it may settle and build up layers of silt overtime. An increase in
sedimentation rates will reduce the useful life of a reservoir, unless measures are taken to reclaim some of
its capacity. Historically, the United States Geological Survey (USGS) has recorded an average of 1.2
billion kilograms of sediment deposition per reservoir each year (USGS 2007b).
The C&D regulation is expected to reduce the amount of sediment entering reservoirs and, as a result, the
cost of reservoir maintenance. Though there are multiple options for sediment mitigation, based on a
literature review EPA used the avoided cost of reservoir dredging as a measure of the benefits of the C&D
regulation to water storage. This approach represents an appropriate measure of social benefits for cases
in which a policy change reduces costs to producers (in this case producers of water supplies for human
use), but in which price effects are minimal (cf Boardman et al. 2001, p. 70-74). A description of other
amelioration measures can be found in Section 8.1: Review of Literature on Reservoir Sedimentation.
The analysis of these benefits includes the following steps:
> Estimating the unit cost of sediment removal from reservoirs
> Estimating the sediment accumulation in reservoirs under the baseline scenario and the post-
compliance EPA policy options
> Estimating the cost that would be incurred to dredge this accumulated sediment under each
scenario and the avoided cost of reservoir dredging expected from this regulation.
The remainder of this chapter provides details of this analysis.
8.1 Review of Literature on Reservoir Sedimentation
Research into reservoir sedimentation was conducted in the 1980s as part of a larger research effort into
the economic effects of sediment runoff from agricultural land use. As sediment has similar effects
regardless of its source, many of the insights from these studies will be applicable to sediment discharge
from construction activities.
Crowder (1987) estimated that the United States was losing about 0.22 percent of its reservoir capacity
each year due to sedimentation. Clark et al. (1985) notes that total U.S. reservoir capacity is filling up
slowly and has enough excess capacity dedicated to hold sediment build-up over hundreds of years.
However, the study goes on to conclude that while total reservoir sedimentation is manageable,
sedimentation is far from uniform and that in about 15 percent of U.S. reservoirs, sedimentation rates
exceeded 3 percent of capacity annually, and in the more extreme cases, 10 percent per year (Clark et al.
1985). The Reservoir Sedimentation Handbook (Morris and Fan 1997), an engineering guide, details
several methods of counteracting the effects of sedimentation to maintain reservoir functionality:
> Sediment routing encompasses a group of techniques that seek to allow sediment-laden water to
pass around or through the reservoir, not allowing the sediment to settle. All of these techniques
involve timing sediment-laden flows from events such as storms and floods and operating the
pass-through or bypass mechanism optimally. Pass-through techniques may be implemented at
November 2008 8-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
existing reservoirs; however, the bypass mechanisms are most cost-effective if incorporated into
the initial design of the reservoir. SPARROW does not take into account any of these techniques
in its assumptions about sediment removed from reaches by reservoirs, EPA was not able to find
any information on either the prevalence or effectiveness of these techniques.
> Flushing involves scouring accumulated sediment from the reservoir by partially or fully draining
the reservoir and allowing the erosive force of the water to carry the sediment through the dam
and downstream of the reservoir. This does not appear to be a very common practice in the
United States, as this creates extraordinarily high sediment concentrations downstream of the
flushed reservoir, sometimes in excess of 1,000,000 mg/L and thus may require special
environmental permission according to Morris and Fan (1997). Flushing is not one-hundred
percent effective at removing settled sediment, and the reservoir may still require dredging. The
process also requires interrupting the service of the reservoir.
> Dredging is a practical and common approach to sediment removal once a reservoir has been
built. Morris and Fan (1997) note that with the exception of draining a reservoir and excavating
settled sediment (which is often more expensive and less common than dredging), dredging may
be the only feasible option for sediment removal in many reservoirs. Dredging may be limited by
available funds, as well as by the amount of space available to dispose of the dredged material.
Due to the cost of transporting dredged sediment, a lack of nearby disposal sites may limit the
amount of sediment that can be removed using dredging.
In a study focused on the economic costs of reservoir sedimentation due to agricultural runoff, Crowder
(1987) describes several options for mitigating reservoir sedimentation. These options are:
> Including sediment pools in the initial construction of the reservoir to collect sediment entering
the reservoir and increase its useful life
> Dredging the reservoir
> Replacing the lost capacity by expanding or replacing the reservoir.
According to Crowder (1987), constructing sediment pools at the outset is more than five times less costly
than dredging the reservoir. This study estimates that reservoir capacity costs between $564 and $1,316
(2008$) per acre-foot to construct (regardless of whether it is built as excess capacity for sediment or to
replace capacity lost from sedimentation), and that dredging costs $4,700 (2008$) per acre-foot of
material dredged. He estimates that only 20,000 (32.3 million cubic yards) acre-feet of sediment are
removed from reservoirs each year by dredging, while sediment pools in reservoirs accumulate 1 million
acre-feet (1.6 billion cubic yards) of sediment per year, and that new capacity would have to account for
the remaining 620,000 acre-feet (1 billion cubic yards) lost to sediment each year (Crowder 1987). His
estimate of total economic damages from reservoir sedimentation is $1.3 billion (2008$) per year, based
on an assumed $94 million (2008$) in annual dredging expenditures and the discounted cost of replacing
all remaining lost capacity in 10 to 20 years.
The analysis used by Crowder (1987) assumes that all reservoir sediment that is not dredged or
accumulated by sediment pools is accounted for with the construction of new capacity. This assumption
implies that additional reservoir capacity can be built unlimitedly to maintain the useful capacity of
reservoirs, an assumption that the author acknowledges is unrealistic. Although Crowder (1987) finds that
building excess storage capacity may be more cost effective than dredging reservoirs, it may not be a
feasible option in many cases. Moreover, in comparing costs of reservoir dredging and building new
reservoir capacity, Crowder (1987) does not take into account ecological effects of new reservoir capacity
November 2008 8-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
construction. The value of ecological services lost due to new reservoir construction may outweigh cost
savings compared to reservoir dredging. The study does also not take into account any opportunity costs
of increasing the amount of land used for water storage. The objective of Crowder (1987) is to quantify
the impact of sedimentation on water storage reservoirs by assuming that all capacity lost to
sedimentation that is not currently recovered through dredging will need to be replaced by new capacity.
While this approach is appropriate for a theoretical estimation of the cost impact of reservoir
sedimentation, EPA's objective is to quantify the avoided costs of reduced reservoir sedimentation due to
reduced construction site stormwater discharges, and thus only takes into account actual measures being
taken to reduce reservoir sedimentation (i.e., dredging). Therefore, this analysis assumes that excess
sediment is removed from reservoirs by dredging rather than building new reservoir capacity.
8.2 Data Sources
Information on reservoir locations, uses, and historic sedimentation rates is available from the Reservoir
Sedimentation Survey Information System (RESIS) database (USGS 2007b). Formally known as RESIS-
II, this database was initially developed by the U.S. Department of Agriculture Natural Resources
Conservation Service, and was provided to EPA by USGS.
The database includes 4,227 sedimentation survey results from 1,819 reservoirs located across the
conterminous United States. Results from 1775 to 1992 are represented, with the majority of
measurements covering the mid-20th century and on. RESIS does not include all U.S. reservoirs, nor does
it contain information about dredging activity in reservoirs or about any other methods for recapturing
lost water storage capacity.
Table 8-1 details sediment deposition per year recorded by RESIS in each EPA region. More than 1.2
billion kilograms of sediment are deposited in the average U.S. reservoir each year. The most sediment
deposition is recorded in Region 6 with 6.6 billion kilograms per reservoir per year, almost four times
more than the national average. Region 9 also reports large masses of sediment deposition each year of
1.8 billion kilograms per reservoir, which is still greater than the national average. Other regions have
average annual deposition per reservoir ranging from 800,000 kilograms in Region 1 to 440 million in
Region 4.
Table 8-1: Average Annual Sedimentation Deposition
Recorded by RESIS
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
Total
Number of Average Annual Sediment Deposition
Reservoirs per Reservoir (million kg/year)
9
18
81
168
272
338
329
208
285
110
1,818
0.8
25.1
71.5
440.4
25.9
6,568.9
57.2
436.0
1,817.6
67.2
1,234.0
Source: RESIS (USGS 2007b)
November 2008 8-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Changes in sediment attenuation in reservoirs will be predicted by the SPARROW model, the details of
which can be found in Chapter 6: Water Quality Modeling.
8.3 Estimating the Unit Cost of Sediment Removal from Reservoirs
Because RESIS does not provide information on reservoir dredging, EPA assumed for the purpose of this
analysis that the average unit cost of dredging sediment from a navigable waterway is approximately
equal to the average unit cost of dredging sediment from a reservoir. Crowder (1987) does provide an
estimate of reservoir dredging cost per unit of sediment dredged, but gives no empirical basis for this
estimate. Because the USAGE dredging database provides costs and quantity dredged for more than 3,300
dredging jobs in the past 12 years, this unit cost of dredging has more robust empirical foundations. EPA
does not expect there to be any significant difference between the costs of reservoir dredging and the
costs of navigable waterway dredging, as all of the variable costs for dredging are related to the quantity
of sediment dredged. Morris and Fan (1997) cite a unit cost of $4.45 (2008$) per cubic yard for sediment
dredging at Lake Springfield, Illinois, which is quite close to the $4.41 average cost per cubic yard for
EPA Region 5 in the navigational dredging data. Chapter 7, Section 7.3: Estimating the Navigational
Maintenance Cost per Cubic Yard of Sediment Removed provides details on estimating the unit cost of
navigable waterway dredging that will be used as a surrogate for the unit cost of reservoir dredging. This
analysis will use average costs at the regional level.
8.4 Estimating the Total Cost of Reservoir Dredging Under Different Policy
Options
8.4.1 Estimating Sediment Accumulation in Reservoirs and the Amount of Sediment
Expected To Be Dredged
EPA estimated the annual sediment attenuation rate in each reservoir under the baseline and the three
policy options using output data from the SPARROW model. Chapter 6: Water Quality Modeling
provides a description of the model. This analysis uses data output from SPARROW that predicts the
amount of sediment (in kilograms) settling in reservoirs in a reach. To estimate the amount of sediment
flowing into and settling in reservoirs in each region, EPA summed the values in this field under the
baseline scenario and each policy option.8 The amount of sediment settling in reservoirs under each
scenario forms the basis for the estimation of the quantity of sediment dredged from reservoirs and the
costs of this dredging.
For the purposes of this analysis, EPA assumed, consistent with Crowder (1987), that all sediment
entering reservoirs must be removed in order to maintain the current water storage capacity in the United
States. The frequency of dredging is highly site-specific, depending on many factors including the
average sediment concentration of the river or stream, the size of the reservoir and excess storage
capacity, and any sediment routing practices. For this analysis, EPA chose a general frequency of
reservoir dredging based on information presented by the USAGE in a Final Dredged Material
Management Plan and Environmental Impact Statement for reservoirs in Washington (USAGE 2002).
This report states that "dredging cycles may vary from 2 to 10 years" (USAGE 2002, p. 66). EPA used
EPA conducted a preliminary analysis of baseline costs including and excluding outliers, and found that the exclusion of
outliers does not have a significant effect on the total cost of reservoir dredging. The values used for this analysis include
reservoir sedimentation for all reservoirs on reaches modeled by SPARROW.
November 2008 8-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
these frequencies as high and low estimates, and employed their arithmetic mean of 6 years for a midpoint
estimate. This approach provides a range of benefits estimates to account for uncertainty in the frequency
of reservoir dredging.
8.4.2 Estimating the Total Cost of Reservoir Dredging Under the Baseline Scenario and
Policy Options
The cost of dredging the sediment settling out in reservoirs is estimated by using the average dredging
cost per region as described in Section 8.3: Estimating the Unit Cost of Sediment Removal from
Reservoirs. Because this cost is given per cubic yard, the sediment attenuation in reservoirs given by
SPARROW will be converted from kilograms to cubic yards using a sediment density of 1.5 g/cm3
(Hargrove 2007). This translates to a conversion of 1,147 kilograms per cubic yard. The equation below
summarizes the calculation of costs for one cycle of dredging before discounting and annualization.
10
TC . =y
reservoirs / j
R=\
(Eq. 8-1)
Where:
TCreservoirs= total cost of dredging all sediment settling in reservoirs in all regions
R = region number
/ = the assumed interval in years of reservoir dredging; varied between 2, 6, and 10
SR = sum of all reservoir sediment settlement predicted by SPARROW in a region in
kilograms
1,147 = kilograms in a cubic yard
CR = regional average of historic dredging job cost per cubic yard, 1995-2006
The resulting costs were discounted and annualized over the assumed interval using both 3 and 7 percent
discount rates, in accordance with OMB Circular A-4 (OMB 2003).
Table 8-2 presents the total amount of sediment that is estimated to be dredged per year and the estimated
cost of this dredging under the baseline scenario, including low, midpoint, and high estimates. The 442
million cubic yards predicted by SPARROW to settle annually in reservoirs is expected to cost between
$1.5 and $1.7 billion to dredge under the baseline scenario. Region 8 has the highest dredging costs, and
though less sediment is predicted to accumulate in Region 8 reservoirs than in those in Region 6, the
former has a much higher unit cost of dredging. Regions 1, 2, and 3 are all predicted to accumulate less
than 6 million cubic yards of sediment per year, and accordingly are estimated to have lower overall
dredging costs. For other regions, midpoint estimates of dredging costs range between $50 and $255
million.
November 2008 8-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 8-2: Reservoir Dredging Under the Baseline Scenario
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Sediment Dredged
(yd3)
1,294,949
2,720,891
5,361,722
45,545,101
35,464,668
128,120,478
41,412,208
125,520,719
41,840,133
14,657,907
441,938,775
Estimated Cost (millions of 2008$)
Low
$12.5
$16.8
$31.7
$214.2
$138.6
$199.7
$86.5
$491.4
$239.6
$47.3
$1,478.3
Mid
$13.3
$17.9
$33.7
$227.7
$147.4
$212.4
$92.0
$522.6
$254.8
$50.3
$1,572.0
High
$14.2
$19.0
$35.8
$241.9
$156.6
$225.6
$97.7
$555.0
$270.6
$53.4
$1,669.7
Source: USEPA analysis
8.5 Estimating the Cost Savings from Reduced Reservoir Dredging
The difference between the anticipated dredging costs under the baseline and a particular policy option
represents the cost savings of that policy option. Table 8-3, Table 8-4, and Table 8-5 present reductions in
sedimentation and subsequent avoided costs from reduced reservoir dredging for each policy option,
including low, midpoint, and high estimates under these options.
Cost savings from a reduction in reservoir sedimentation range from $470,000 to $32 million, depending
on the policy options, the assumed frequency of reservoir dredging, and the discount rate. EPA's Option 2
represents a savings of $17.6 million for the midpoint estimate at a 3 percent discount rate and $15.9
million at a 7 percent rate. The largest savings are predicted in Region 4 under all options, as SPARROW
predicts the largest overall reductions in sediment accumulation in this region. Region 4 accounts for
nearly 98 percent of all savings under Option 1, and more than 50 percent in both Options 2 and 3.
Region 6 is also anticipated to benefit from savings between $4.3 and $4.8 million under Option 2, and by
between $6.5 and $7.4 million under Option 3 (at 3 and 7 percent discount rates, respectively).
Option 1, a construction general permit requiring sedimentation basins on all sites larger than 10 acres
and the implementation of best management practices on smaller sites, is EPA's least stringent policy
option. As noted above, Region 4 is the only region to benefit significantly from this policy option, and
accounts for $593,000 of the $606,000 in cost savings for the midpoint estimate of this option at a 3
percent discount rate, and $536,000 out of $548,000 at a 7 percent rate.
Option 2 imposes a turbidity standard on construction sites larger than 30 acres in areas that have an "R-
factor" (see Chapter 3 and USEPA (2008b) for more details) of greater than 50 and have soils containing
greater than 10 percent small particles, and requires a construction general permit for all other sites. The
turbidity standard would require on-site treatment of stormwater, which would reduce TSS and turbidity
in discharges from sites meeting the criteria for the turbidity standard. Cost savings under this option
range between $13.7 and $18.7 million, with an expected value of $17.6 million at a 3 percent discount
rate and $15.9 million at a 7 percent rate. This option greatly increases expected savings for all regions,
reducing reservoir sedimentation rates by more than 100,000 cubic yards per year in five regions. Region
4 is predicted to realize a large portion of the cost savings, though Region 6 is also expected to benefit
from $3.6 to $4.8 million dollars of reduced dredging costs each year. The next largest portion of savings
November 2008
8-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
is predicted to accrue to Region 7, which will benefit by at least $1.2 million in the midpoint estimate.
The Agency expects the smallest reductions in reservoir sedimentation in Regions 8 and 9, where many
sites will be exempt due to rainfall or soil type waivers. Both of these regions will experience a decline of
10,000 cubic yards or less per year in the midpoint estimate of reservoir sedimentation.
Option 3, which imposes a turbidity standard on all sites larger than 10 acres (in addition to the
requirements of Option 1) and requires a construction general permit for others, is EPA's most stringent
policy option. This option would require on-site treatment of stormwater, substantially reducing its
contribution to in-stream TSS and turbidity. Expected cost savings from this action range between $23.8
and $32.5 million, with midpoint estimates of $30.6 million at a 3 percent discount rate and $27.6 million
at a 7 percent rate. Region 4 is again the largest beneficiary from this regulation, with approximately
$16.4 million in savings for the midpoint estimate at a 3 percent discount rate and $14.8 million at a 7
percent discount rate. Region 6 is also estimated to benefit substantially from this regulation, with a
reduction between $5.4 and $7.4 million. The drier Regions 8 and 9, which have many areas exempt
under Option 2, are expected to benefit much more substantially under Option 3, with expected savings in
Region 9 more than ten times greater under Option 3 than under Option 2 and nearly seven times greater
in Region 8. Overall cost savings under this policy option nearly double for all estimate ranges.
Overall, Regions 4 and 6 are the largest beneficiaries of reduced dredging costs from this action, with
substantial cost and sedimentation reductions also occurring in Regions 7 and 10 under Options 2 and 3.
These two options are expected to result in significantly higher savings than Option 1, and are also
anticipated to provide a more even distribution of these avoided cost benefits among regions.
Table 8-3: Reduction in Reservoir Dredging and Costs Under Option 1
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Annual
Reduction in
Sediment yd3)
81
122
236
118,525
1,343
175
256
641
120
5
121,504
Avoided Costs (thousands of 2008$)
3%
Low
$0.8
$0.8
$1.4
$557.3
$5.3
$0.3
$0.5
$2.5
$0.7
$0.0
$569.5
Discount Rate
Mid
$0.8
$0.8
$1.5
$592.6
$5.6
$0.3
$0.6
$2.7
$0.7
$0.0
$605.6
High
$0.9
$0.9
$1.6
$629.4
$5.9
$0.3
$0.6
$2.8
$0.8
$0.0
$643.2
7% Discount Rate
Low
$0.6
$0.6
$1.2
$462.4
$4.4
$0.2
$0.4
$2.1
$0.6
$0.0
$472.5
Mid
$0.8
$0.7
$1.3
$535.9
$5.0
$0.3
$0.5
$2.4
$0.7
$0.0
$547.6
High
$0.9
$0.8
$1.5
$617.3
$5.8
$0.3
$0.6
$2.8
$0.8
$0.0
$630.8
Source: USEPA analysis
November 2008
8-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 8-4: Reduction in Reservoir Dredging and Costs Under Option 2
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Annual
Reduction in
Sediment (yd3)
9,405
11,408
77,481
1,894,111
151,379
2,754,706
701,561
8,869
3,541
182,545
5,795,007
Avoided Costs (thousand 2008$)
3% Discount Rate
Low
$91.1
$70.4
$458.2
$8,906.1
$591.7
$4,293.8
$1,465.4
$34.7
$20.3
$589.0
$16,520.7
Mid
$96.9
$74.8
$487.3
$9,470.5
$629.2
$4,565.9
$1,558.3
$36.9
$21.6
$626.3
$17,567.7
High
$102.9
$79.5
$517.6
$10,059.0
$668.3
$4,849.6
$1,655.1
$39.2
$22.9
$665.2
$18,659.3
7% Discount Rate
Low
$75.6
$58.4
$380.2
$7,389.6
$491.0
$3,562.7
$1,215.9
$28.8
$16.8
$488.7
$13,707.7
Mid
$87.6
$67.7
$440.6
$8,563.8
$569.0
$4,128.7
$1,409.1
$33.4
$19.5
$566.3
$15,885.7
High
$101.0
$78.0
$507.6
$9,864.6
$655.4
$4,755.9
$1,623.1
$38.5
$22.5
$652.3
$18,298.7
Source: USEPA analysis
Table 8-5: Reduction in Reservoir Dredging and Costs Under Option 3
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Annual
Reduction in
Sediment (yd3)
19,967
20,009
113,358
3,282,768
236,195
4,197,561
980,419
64,591
44,456
710,277
9,669,601
Avoided Costs (thousand 2008$)
3% Discount Rate
Low
$193.5
$123.4
$670.4
$15,435.5
$923.3
$6,542.8
$2,047.9
$252.9
$254.6
$2,291.6
$28,735.9
Mid
$205.8
$131.3
$712.9
$16,413.7
$981.8
$6,957.4
$2,177.6
$268.9
$270.7
$2,436.8
$30,556.9
High
$218.5
$139.4
$757.2
$17,433.6
$1,042.8
$7,389.7
$2,313.0
$285.6
$287.5
$2,588.3
$32,455.6
7% Discount Rate
Low
$160.5
$102.4
$556.3
$12,807.3
$766.1
$5,428.7
$1,699.2
$209.8
$211.2
$1,901.4
$23,842.9
Mid
$186.1
$118.7
$644.7
$14,842.2
$887.8
$6,291.3
$1,969.1
$243.2
$244.8
$2,203.5
$27,631.3
High
$214.3
$136.7
$742.6
$17,096.7
$1,022.6
$7,246.9
$2,268.3
$280.1
$282.0
$2,538.3
$31,828.5
Source: USEPA analysis
8.6 Sources of Uncertainty and Limitations
> The reliance of the SPARROW model on the RF1 network is a significant limitation because the
reservoirs located off the RF1 network are omitted from this analysis. This omission is likely to
bias the estimated savings from reduced reservoir dredging downward.
> There is uncertainty as to the uniformity of sediment density because it is related to the type of
soil in the area. Using a single density to convert volume to weight for all sediment may reduce
the accuracy of the resulting cost estimates. However, the direction of this potential bias is
uncertain.
> The lack of data on reservoir dredging results in significant uncertainty as to the types of
reservoirs that are dredged and the cost of this dredging. The appropriateness of the avoided cost
approach is conditional upon the estimates representing actual dredging costs that are no longer
November 2008
8-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
imposed on reservoirs. If typical reservoirs do not reduce dredging, and hence costs, but rather
undertake other actions in response to reduced sediment loads, then actual benefits may diverge
from estimates provided above.
> Neither RESIS nor the SPARROW model take into account sediment build-up in natural water
storage facilities such as glacial lakes and ponds, so any activity to mitigate sedimentation of
these areas is not included in this analysis, resulting in a potential downward bias of this estimate.
Though sediment pools are built to accumulate sediment and preserve the useful capacity of the reservoir,
they may also fill up more rapidly than anticipated at their initial construction, increasing the sediment
build-up in a reservoir and increasing the cost of dredging it. It is also possible that these sediment pools
themselves may be dredged. This analysis assumes that to maintain the current water storage capacity in
the United States, all influent sediment will have to be removed in some manner, or replaced. Building
replacement capacity is environmentally disruptive and may be more costly than sediment removal by
dredging; therefore, this analysis assumes that dredging will be used to maintain reservoir capacity.
November 2008 8-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
9 Benefits to Drinking Water Treatment
Drinking water treatment plants provide the essential services of removing contaminants from water taken
from surface and ground sources, making the water potable. Sediment is one of the contaminants that
must be removed from drinking water before it can be consumed, so a reduction in sediment levels in
water taken up by drinking water treatment plants is expected to result in lower treatment costs for these
plants. This section describes the methodology EPA used to estimate the reduction in drinking water
treatment costs that should result from a reduction of sediment discharge from construction sites. This
approach represents an appropriate measure of social benefits for cases in which a policy change reduces
costs to producers (in this case producers of drinking water), but in which price effects are minimal (cf
Boardman et al. 2001, p. 70-74).
9.1 Sediment Effects on Drinking Water Treatment Costs
The primary concern of sediment in terms of drinking water treatment is the turbidity that results from
suspended sediment in the water column. As turbidity increases in drinking water influent, the variable
costs of water treatment to mitigate this turbidity increase.
The variable costs associated with treating drinking water for turbidity are twofold:
> Adding chemicals coagulants that bond to the sediment particles and help them sink out of the
water column
> Disposing of the sludge which results from this treatment.
Several attempts have been made in the past two decades to isolate these elements of drinking water
treatment costs:
> Forster et al. (1987) estimated $172 (2008$) per million gallons for drinking water sediment
treatment, while Holmes (1988) performed several estimates, with ranges between $7.45 and
$212 (2008$) per million gallons. It is unclear in these studies what elements were included in
these estimates.
> Moore and McCarl (1987) estimated a total of $41 (2008$) per million gallons to treat water and
dispose of the sludge for drinking water treatment plants in Oregon.
> Dearmont et al. (1998) put the chemical costs of turbidity treatment alone at $96 (2008$) per
million gallons for Texas drinking water.
This broad range of cost estimates for sediment and turbidity treatment may be due to the fact that the
studies measure costs per million gallons of water, without respect to the specific sediment levels in
drinking water influent.
9.2 Avoided Costs of Drinking Water Treatment from Reducing Sediment
Discharge from Construction Sites
This section describes the estimation of the total expenditures to remove sediments from drinking water
and the cost savings expected with the reduction of sediment discharge anticipated from the regulation. It
involves the following steps:
November 2008
9-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
> Identifying RF1 reaches modeled by SPARROW that are sources for drinking water treatment
plants
> Determining TSS reductions in these reaches
> Estimating the chemical cost of treating the turbidity caused by TSS in these reaches
> Estimating the cost of disposing of the sludge generated from this turbidity treatment
> Estimating the total costs of drinking water treatment
> Estimating the total cost savings between the baseline and post-compliance scenarios.
Figure 9-8 outlines these steps in more detail. The ovals at the top represent primary data sources. The
gray boxes contain calculations or conversions, and the white boxes contain the results of these
calculations or conversions that serve as input for the next step in the calculation. The left-hand side of
the diagram shows the calculation of the chemical cost of treating turbidity, while the right-hand side
shows the calculation of the cost of sludge disposal. Finally, the pale yellow boxes at the bottom of the
figure represent the costs under baseline and post-compliance scenarios and the difference between them.
Sediment concentrations and drinking water influent volumes were estimated for each surface drinking
water intake in the United States. EPA calculated the costs of treatment chemicals and sludge disposal by
surface drinking water intake; the specific steps are described in more detail in the following sections.
Costs under the baseline and post-compliance scenarios were compared, and the difference represents the
national cost savings. To address uncertainty in its assumptions, EPA conducted a sensitivity analysis that
varies the assumptions described in the following sections, and the results of the following analyses
present low, midpoint, and high benefits estimates (see Section 9.8: Sensitivity Analysis for details).
November 2008 9-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Figure 9-8: Steps in Calculating the Cost of Treating Turbidity in Drinking Water
SPARROW results: TSS
concentrations by RF1
reach under baseline and
post-compliance scenarios
I
SDWIS: Drinking water
influent volumes by RF1
reach
TSS—Turbidity conversion
I
Influent turbidity under baseline and post
compliance scenarios
I
Alum dose—Turbidity relationship
I
Chemical usage under baseline and post-
compliance scenarios
I
Sludge generation equation
Alum unit cost
I
I
Sludge generation under baseline and post-
compliance scenarios
Chemical costs under baseline and post-
compliance scenarios
I
Probability of off-site disposal
I
Sludge disposal under baseline and post-
compliance scenarios
I
Distance to disposal facility
Sludge disposal costs under baseline and
post-compliance scenarios
_T
1
r
Total treatment costs by RF 1 reach under
baseline scenario
1 r
Total treatment costs by
post-compliance
RF 1 reach under
scenario
T
J
Reduction in total drinking water treatment costs by RF1
reach from baseline to post-compliance scenarios
November 2008
9-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
9.3 Data Sources
Information on drinking water intake points and volumes were provided by EPA's Safe Drinking Water
Information System (SDWIS), Federal Version. The federal version of SDWIS stores the information on
approximately 160,000 public water systems used to monitor the quality and safety of drinking water
(EPA 2006a). For this analysis, EPA was only interested in surface water intakes, as these are the only
type that would be affected by discharge from construction sites. Location information and average daily
flow were estimated from the latitude-longitude and population data provided by SDWIS, respectively.
EPA obtained data on the chemicals used to treat turbidity based on data from the American Society of
Civil Engineers (ASCE) and the American Water Works Association (AWWA). These data also contain a
formula for the calculation of sludge disposal costs (Baruth 2005).
Estimations of coagulant dosing for various turbidity levels were taken from research reports by Cornell
University's AguaClara Project. This project focuses on improving drinking water quality through
engineering research and offers calculations and other engineering information for both researchers and
professionals in the field (Menendez et al. 2007).
The costs of alum and alternative coagulant inputs were obtained by EPA for the Long Term 2 Enhanced
Surface Water Treatment Rule. The Technologies & Costs Document for this rule included research on
the expenses of public water systems and potential changes resulting from the proposed rule (EPA
2000c). For this purpose, many details of the costs of drinking water treatment were obtained. This
document reports cost data in 1998 dollars, so its data are be adjusted to 2008 dollars.
The equation used to estimate the volume of sludge generated from turbidity and TSS treatment comes
from the EPA document Identification and Characterization of Typical Drinking Water Treatment
Residuals (EPA 2007a), which examines the byproducts of drinking water purification from different
methods of treatment.
Calculations of the percentage of facilities that dispose of any of their residual sludge off-site and of the
percentage of total residual sludge generated that is disposed of off-site are obtained from the Drinking
Water Treatment Plant Master Survey Response Database (henceforth the Drinking Water Survey
Database) prepared by EPA for the Drinking Water Effluent Limitation Guidelines rule (EPA 2007b).
9.4 Modeling Sediment Concentrations and Reductions
EPA worked in conjunction with USGS to model sediment concentrations in RF1 stream reaches under
both the current and proposed scenarios. A detailed description of this model, SPARROW, can be found
in Chapter 6: Water Quality Modeling. SPARROW modeled the sediment concentrations in each RF1
reach according to many factors, including the amount of stormwater discharge from construction and
development activities affecting that reach. SPARROW predicts sediment loads by suspended sediment
concentration (SSC), which EPA converted into TSS by dividing the SPARROW-generated values by
1.3. This factor compensates for a 30 percent negative bias in TSS concentrations (Gray et al. 2000).
EPA assumed, as per Ngo and Nichols (1998), that facilities with baseline TSS concentrations exceeding
1,000 mg/L would employ plain sedimentation, during which influent water is allowed to settle for
extended periods of days or weeks. The effectiveness of this process depends on the size of the
sedimentation basin and the amount of time the water is left to settle. Studies by De Zuane (1997) and
November 2008 9-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Yao (1975) suggest that the effectiveness of plain sedimentation ranges from 30 and 90 percent with a
midpoint estimate of 60 percent. Thus, EPA varied the effectiveness of plain sedimentation basins
between 90, 60, and 30 percent TSS removal for the low, mid, and high benefit estimates, respectively.9
This reduction is applied to influent water TSS concentrations at facilities that are estimated to have
influent water with TSS concentrations exceeding 1,000 mg/L in the baseline. EPA deems this approach
reasonable, as it brings treatment costs within the range of values published in existing literature
presented in Section 9.1: Sediment Effects on Drinking Water Treatment Costs.
9.5 Identifying Reaches with Drinking Water Intakes and Their Intake Volumes
To estimate the avoided costs of drinking water treatment resulting from the regulation, EPA used the
drinking water intake database compiled by USGS. The database provides information on public water
systems, including intakes, treatment plants, sampling sites, distribution systems, wells, and consecutive
connections from EPA's SDWIS database. Each drinking water intake is matched to the RF1 reach and
has the corresponding "USGS River Reach Number." The database, however, does not provide
information on the intake volume. For this analysis, EPA approximated each intake volume based on the
population served by each drinking water system and the number of intakes corresponding to that system.
The population served by a particular drinking water system was divided by the number of intakes in that
system to obtain the average population served per intake. From these data the daily flow of each intake
was estimated using the following equation determined by EPA for the Revised Arsenic Rule (USEPA
2000a):
F = 0.067 *PIM2 (Eq. 9-1)
Where:
F = average daily flow in thousands of gallons per day
P = population served
Table 9-1 summarizes public surface drinking water intakes by EPA region and the total estimated daily
flow for all intakes in that region. Overall, public water systems in the United States withdraw nearly 18
billion gallons of water per day from surface water sources, an average of 2.8 million gallons per day per
intake.
The more effective the plain sedimentation process, the lower the treatment costs to the affected facilities, and consequently,
the lower their potential savings from reduced influent turbidity.
November 2008 9-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 9-1: Public Water Intakes and Estimated
Average Daily Flow by EPA Region
EPA Region
1
2
3
4
5
6
7
8
9
10
Total
Number of Public
Water Intakes
639
478
881
868
525
821
353
549
748
487
6,349
Total Estimated Daily
Flow (million gallons/day)
1,105
2,286
2,301
2,913
2,373
2,309
715
701
2,543
560
17,809
Source: SDWIS, Federal Version (2007)
9.6 Estimating the Cost of Chemical Turbidity Treatment
Reducing sediment concentrations in drinking water influent is expected to reduce the turbidity level of
the water, requiring smaller doses of chemical coagulants to treat. EPA estimated the amount of
coagulants needed to treat a given level of turbidity according to the following steps:
> Converting sediment concentrations (given in mg/L) into nephelometric turbidity units (NTU)
using an equation determined by Baruth (2005) where TSS (mg/L) = b • turbidity (in NTU) where
b has a value between 0.8 and 2.2. The value values assigned to b in this analysis are 2.2 for the
low estimate, 1.5 for the midpoint estimate, and 0.8 for the high estimate (as the value of b
decreases, a given level of TSS generates more turbidity, leading to higher treatment costs).
> Determining the doses of alum (the primary coagulant used to treat turbidity) required to treat this
turbidity level using relationships determined by researchers of the AguaClara Project (Menendez
et al. 2007).
> Multiplying this coagulant dose (given per volume of water treated) by the total volume of a
drinking water intake to obtain the total coagulant usage for water from that intake.
> Multiplying this amount of coagulant by the unit cost of the alum.
For this analysis, EPA assumed that alum is the coagulant used to treat turbidity, since ferrous sulfate is
likely to be used only on water with low turbidity (high doses of iron may have negative effects on water
quality). No cost information was available for ferrous sulfate, but ferric chloride was quoted at $497 per
ton (2008$); additionally, dosing information for iron compounds could not be determined. For the Long
Term 2 Enhanced Surface Water Treatment Rule, EPA determined that alum costs between $286 and
$373 per ton (2008$) (EPA 2005b). EPA's assumption that all turbidity will be treated with alum may
understate treatment costs and savings. EPA estimated cost savings using the cost of both dry stock and
liquid stock alum to present a range of savings estimates in the sensitivity analysis. The lower cost of
liquid alum was used in the low estimate, but it should be noted that the sludge generation equation in
Section 9.7.1: Estimating the Amount of Sludge Generated, calculates sludge volume based on dry alum
as a coagulant. Therefore, the cost of dry alum is used for the midpoint and high estimates. The
November 2008 9-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
approximate dose of alum used to treat water of a given turbidity can be estimated using the following
equation:
Al = 331og(r) - 28 (Eq. 9-2)
Where:
Al = alum dose in milligrams per liter
T= influent water turbidity in NTU (after any reduction from plain sedimentation).
After calculating the dose of alum necessary to treat the turbidity in the influent water at a given intake,
this dose was multiplied by the total daily influent volume of the intake (converted from gallons to liters
using a conversion factor of 3.8 liters to one gallon) to obtain the total amount of alum needed to treat
turbidity at the intake per day. This total daily amount of alum was multiplied by 365 to obtain an annual
estimate of the amount of alum needed to treat turbidity from the intake. EPA then used a conversion
factor of 907x106 milligrams per ton to estimate the amount of alum required for turbidity treatment in
tons per year. Finally, the estimated annual amount of alum was multiplied by the unit cost of alum to
arrive at a final annual cost of alum for each drinking water intake. The following equation presents this
calculation of the annual cost of alum.
TCM = Q *
907*106
*365*CAl (Eq. 9-3)
Where:
TCAl =
Q =
Al =
3.8 =
total annual alum cost in 2008$
daily plant flow in gallons
alum dose in milligrams per liter
liters in one gallon
907* 106 = milligrams in one ton
365 = days in one year
CAi= cost of one ton of alum in 2008$
The estimated annual cost of turbidity treatment was calculated under the baseline and post-compliance
scenarios for each surface drinking water intake in the SDWIS database where TSS concentration
information was available, atotal of 6,132 intakes.
9.7 Estimating the Cost of Sludge Disposal
The alum added to the water to treat turbidity binds to the particles of sediment, making heavier lumps of
coagulated sediment that settle out of the water column. This residual material, known as sludge, is
proportional to the amount of sediment in the water and the amount of coagulant added. Drinking water
November 2008
9-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
treatment facilities must regularly dispose of this sludge, either by discharging it to a receiving water
body or wastewater treatment plant, or by having it hauled away for off-site disposal. The cost of sludge
disposal can be significant.
The estimation of the cost of sludge disposal has several steps:
> Estimating the amount of sludge generated
> Calculating the probability that a generated quantity of sludge will be disposed of off-site (as
opposed to discharged to a receiving water body)
> Estimating the distance to the disposal location
> Calculating the cost of sludge disposal
9.7.1 Estimating the Amount of Sludge Generated
A reduction in the turbidity of the water will result in a reduction in the amount of sludge generated from
the chemicals added to treat turbidity, and thus a reduction in the cost of sludge disposal. Sludge
quantities are a function of the amount of chemical coagulants added to treat the water and the TSS levels
of the water. EPA (2007a) derived the following relationship between sludge generation and chemical
coagulants:
S = 8.34*Q*[2.0CaCH + 2.6MgCH + CaNCH +1.6MgNCH + CO2 +0.44Al + 2.9Fe + SS + A] (Eq. 9-4)
Where:
S = sludge production (Ib/day)
Q = plant flow (mgd)
CaCH = calcium hardness removed as CaCO3 (mg/L)
MgCH = magnesium hardness removed as CaCO3 (mg/L)
CaNCH = noncarbonated calcium hardness removal as CaCO3 (mg/L)
MgNCH = noncarbonated magnesium hardness removed as CaCO2 (mg/L)
CO2 = carbon dioxide removed by lime addition, as CaCO3
Al = dry alum dose (mg/L) (as 17.1 percent A12O3)
Fe = iron dose as Fe (mg/L)
SS = raw water suspended solids (mg/L)
A = other additives (mg/L)
EPA used this equation in conjunction with the estimated sediment concentrations in the source water, the
quantity of water taken up by the drinking water intake, and the alum doses required to treat turbidity to
estimate the amount of sludge generated from turbidity treatment.
November 2008 9-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
EPA recognizes that the estimated volume of sludge takes into account the residuals generated from alum
treatment of TSS, and thus does not represent the actual volumes of sludge generated from the entire
water treatment process. Leaving out the other chemical additives does not affect the difference in sludge
generation from turbidity treatment under the baseline and post-compliance scenarios, as they would be
held constant under both scenarios.
9.7.2 Calculating the Probability That the Sludge Will Require Off-Site Disposal
Public water systems may use different sludge disposal methods. Some facilities dispose of their sludge
off-site; some discharge to water bodies under an NPDES permit or discharge to a wastewater treatment
plant. Facilities may also discharge a certain amount of their residual sludge and dispose of the remainder
off-site. Facilities discharging sludge into receiving waters or to a wastewater treatment plant do not incur
sludge disposal costs. To calculate the amount of sludge disposed of off-site and resulting in disposal
costs to a drinking water treatment facility, EPA calculated the probability of off-site disposal.
The Agency used facility data from the drinking water survey database to calculate the number of
facilities reporting any off-site disposal and the percent of sludge being disposed of off-site at these
facilities. The survey found 207 out of 511 facilities (41 percent) reporting some off-site disposal. At
these facilities, on average 85 percent of all residual solid sludge was disposed of off-site. To determine
the likelihood that a generated quantity of sludge will result in off-site disposal costs for a facility, EPA
multiplied the probability that the sludge was generated at a facility performing any off-site disposal (41
percent) by the probability that a unit of sludge from this facility will be disposed of off-site (85 percent).
The estimated probability (Pr) that a given volume of sludge generated will be disposed of off-site, thus
incurring sludge disposal costs, is 35 percent.
9.7.3 Estimating the Distance to the Disposal Location
The cost of sludge disposal is also a function of the distance to the disposal location. A sample of 10
facilities and their disposal locations was taken randomly from the Drinking Water Survey Database.
Because location information was available for both the facilities and their disposal locations, EPA
calculated the driving distance between the facility and its discharge locations for each of the 10 facilities.
If a facility had more than one listed discharge location, an average distance was used. Based on this
sample of 10 facilities, the distance to a disposal location can vary between 1 and 200 miles, and the
estimated average distance (Z) to disposal locations is 21.6 miles.
9.7.4 Calculating the Cost of Sludge Disposal
EPA estimated the annual cost of sludge disposal as a function of sludge volumes and distance to the
disposal facility using the following equation from EPA (1996a):
7 = [54.12(S * 365 * .0005 * Pr)] + [(3.83 + 0.25Z)(S * 365 * .0005 * Pr)] (Eq. 9-5)
Where:
Y = annual cost of sludge disposal ($/year)
54.12= disposal cost per ton of sludge ($)
S = total sludge production from treating turbidity in drinking water (Ibs/day)
November 2008 9-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
365= days per year
0.0005 = tons per Ib
Pr = the estimated probability of off-site sludge disposal (35 percent)
3.83 + 0.25 = transportation cost ($)
Z = average transportation distance of 21.6 miles.
The cost of sludge disposal was estimated under both the baseline and post-compliance scenarios for all
surface drinking water intakes in SDWIS for which TSS concentrations were available.
9.8 Sensitivity Analysis
To account for uncertainty in drinking water treatment costs EPA varied several of the parameters of this
analysis, including the effectiveness of pre-sedimentation, the amount of turbidity generated by a unit of
TSS, and the cost of alum, to provide a range of benefits estimates. The values of these inputs in each
estimate range are as follows:
> Low Estimate:
D TSS levels exceeding 1,000 mg/L in the baseline are reduced by 90 percent for all options
TSS = 2.2 * Turbidity
CM = $250
> Midpoint Estimate:
° TSS levels exceeding 1,000 mg/L in the baseline are reduced by 60% for all options
TSS = 1.5 * Turbidity
CM = $327
> High Estimate:
D TSS levels at exceeding 1,000 mg/L in the baseline are reduced by 30% for all options
TSS = 0.8 * Turbidity
CM = $327
9.9 Estimating the Total Costs of Drinking Water Treatment Under the Baseline
and Post-Compliance Scenarios
EPA estimated the total national costs of treating drinking water for turbidity by summing the estimated
annual costs of chemical coagulants added to reduce turbidity and the estimated annual cost of disposing
of the sludge resulting from this chemical treatment. Table 9-2 presents annual estimated total drinking
water treatment costs for the baseline scenario, including low, midpoint, and high estimates for avoided
costs. National drinking water treatment costs for TSS and turbidity range from around $377 to about
$742 million between the low and high estimates for the baseline scenario, with an expected value of
November 2008 9-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
$585 million. Regional turbidity treatment costs range from $8.4 million in Region 10 to between $85 and
$110 million in Regions 4, 5, 6, and 9 for the midpoint estimate.
Table 9-2: Drinking Water Turbidity Treatment Costs Under the
Baseline Scenario
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Average Treated Turbidity (NTU)
Low
43.1
53.3 ;
73.3
109.3 1
139.2
190.6 I
167.1
123.6 I
136.1
45.7 \
111.0
Mid
63.9
90.5
107.9
161.2
274.8
610.7
424.8
267.1
323.4
83.9
247.9
High
121.1
.: 192.8
203.3
1 303.7
647.7
I 1,765.9
1,133.6
! 661.7
838.4
s 188.9
624.5
Cost per million gallons1
Treatment Cost
(millions of 2008$)
Low
$10.9
$28.3
$48.7
$61.4
$54.3
$64.8
$18.3
$18.3
$65.9
$5.5
$376.5
$59.4
Mid
$16.5
; $44.6 .:
$65.8
1 $84-5 1
$88.2
I $107.1 I
$33.7
I $25.6 !
$110.5
s $8.4 s
$584.9
$92.2
High
$22.2
$58.5
$77.6
$100.8
$113.1
$139.9
$44.3
$30.6
$143.9
$11.1
$742.0
$117.0
Calculated for the estimated 17,373 million gallons per day of intake for public water systems in
SDWIS.
Source: USEPA analysis
It is important to note that these avoided cost estimates do not take into account any treatment or disposal
cost savings to public wastewater treatment facilities to which drinking water treatment facilities may
route their sludge, nor any environmental benefits from reduced volumes of sludge discharged directly
into surface waters by direct drinking water treatment facilities. Therefore, the total benefits of reducing
turbidity in drinking water treatment are understated.
9.10 Estimating Cost Savings from Lower Sediment and Turbidity in Drinking
Water Influent
The total cost savings from lowered turbidity resulting from lower TSS concentrations in drinking water
influent was estimated as the difference between drinking water turbidity treatment under the baseline and
post-compliance scenarios. Table 9-3 ,Table 9-4 and Table 9-5 present estimated savings in drinking
water treatment costs from lowered turbidity for the three policy options, as well as low, midpoint, and
high estimates for savings under each of these options.
The anticipated savings from reduced TSS and turbidity treatment at drinking water facilities are between
$181,900 and $14.5 million, varying substantially between the least and most stringent policy options
though less dramatically between the low and high estimates. Option 2 is expected to reduce TSS and
turbidity treatment costs for drinking water facilities by between $5.8 and $8.0 million. As is the case
with navigable waterway and reservoir dredging, Region 4 benefits most significantly and consistently
from the TSS reductions expected from this regulatory action. The avoided costs in Region 4 under Policy
Option 1 account for a significant portion of the total national savings. Other regions receive a larger
portion of benefits under Policy Options 2 and 3, though Region 4 still accounts for the greatest
proportion of the cost reductions. Regions 6 and 10 also show significant savings under Options 2 and 3,
with savings in Region 10 between $1.2 and $3.5 million in all estimates for these two options.
November 2008
9-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Option 1, a construction general permit requiring sedimentation basins on all sites larger than 10 acres
and the implementation of best management practices on smaller sites, is EPA's least stringent policy
option. Region 4 receives the only considerable cost savings under this option, between $170,500 and
$200,500. Savings in Regions 2, 5, and 8 are between $1,00 and $10,000, and in all other regions,
expected cost reductions are less than $1,000.
Option 2 imposes a turbidity standard on construction sites larger than 30 acres in areas that have an "R-
factor" (see Chapter 5 for more details) of greater than 50 and with soils containing greater than 10
percent small particles, and requires a construction general permit for all other sites. The turbidity
standard would require on-site treatment of stormwater, which would reduce TSS and turbidity in
discharges from sites meeting the criteria for the turbidity standard. For the midpoint estimate of Option
2, the national average reduction in treated turbidity is 1.7 NTU, though in Region 4 (where many of the
benefits are expected) the reduction is more than 5 NTU. Region 10 is expected to benefit from an
average reduction of 4.3 NTU in influent water turbidity. The expected value of avoided costs for this
estimate is $7.4 million. Estimated savings exceed $1 million in Regions 4, 6, and 10. These savings are
much greater than the near-zero savings expected under Option 1 for Regions 6 and 10. Regions 8 and 9
do not benefit as substantially as other regions, likely due to a large number of sites being exempted from
the turbidity standard.
Option 3, which imposes a turbidity standard on all sites larger than 10 acres (in addition to the
requirements of Option 1) and requires a construction general permit for others, is EPA's most stringent
policy option. This option would require on-site treatment of storm water, substantially reducing its
contribution to in-stream TSS and turbidity. Total avoided costs for this option are between $10.4 and
$14.5 million, with a midpoint estimate around $13.1 million. The drier Regions 8 and 9, which have
many areas exempt under Option 2, are expected to benefit much more substantially under Option 3.
Estimated avoided costs are seven to ten times larger under this option compared to Options 1 and 2,
though still much lower than in other regions.
As construction site discharge is more likely to contain smaller particles that contribute less to TSS and
more to turbidity, the high estimates for Options 2 and 3 may be more relevant because EPA uses a
conversion factor between TSS and turbidity that takes this into account. The high estimate for turbidity
reductions under Option 2 is 7.2 NTU nationwide, with a reduction of nearly 13 NTU in Region 4 and of
nearly 12 NTU in Region 7. Under Option 3, the high estimate is on average around 11.2 NTU
nationwide, 22.2 NTU in Region 4, 13.1 NTU in Region 6, and nearly 17 NTU in Region 7.
Overall, Regions 4, 6, and 10 receive the largest proportion of the cost reductions from reduced turbidity
in drinking water. Options 2 and 3 produce significantly larger cost savings than Option 1, which is
expected to reduce treatment costs by only about $211,400 nationally (midpoint estimate). Midpoint
estimates of avoided costs under Options 2 and 3 are approximately $7.4 and $13.1 million, respectively.
Under Options 2 and 3, cost savings are distributed more evenly among the 10 regions, though the
aforementioned regions benefit most due to a more substantial effect of the regulation in these regions.
November 2008 9-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 9-3: Reduction in Drinking Water Treatment Costs Under Option 1
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total
Average Reduction
Low
0.0
0.0
0.0
0.5
0.0
o.o ;
0.0 I
o.o ;
0.0
0.0
0.1
in Treated
Mid
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
0.0
0.0
0.1
Turbidity (NTU) 1
High
0.0
0.0
0.0
1.4
0.1
; o.o
i o.o
: o.o
0.0
0.0
0.2
Cost Savings
Low
$0.2
$4.2
$0.2
$170.5
$5.6
$0.1 ;
$0.1 !
$1.1 ;
$0.0
$0.0
$181.9
(thousands of 2008$)
Mid
$0.2
$5.0
$0.2
$196.7
$7.8
$0.1 ;
$0.1
$1.4 ;
$0.0
$0.0
$211.4
High
$0.3
$5.0
$0.2
$200.5
$9.3
$0.1
$0.2
$1.5
$0.0
$0.0
$217.1
1 Average turbidity reductions shown as 0.0 are not actually zero, but not sufficiently large to show at this level of significant digits
Source: USEPA analysis
Table 9-4: Reduction in Drinking Water Treatment Costs Under Option 2
EPA
Region
1
2
3
4
5
6
7
8
9
102
Total
Average Reduction
Low
0.1
0.3
1.4
4.6
0.6
2.1
2.9
0.0
0.1
4.9
1.7
in Treated
Mid
0.1
0.5
2.0
6.7
1.0
3.7
5.3
0.1
0.1
15.9
3.1
Turbidity (NTU) 1
High
0.2
1.0
3.8
12.7
2.3
8.0
11.8
0.1
0.2
46.2
7.2
Cost Savings
Low
$11.5
$137.1
$567.0
$2,070.0
$192.4
$1,369.0
$322.0
$4.9
$7.3
$1,165.7
$5,846.9
(thousands of 2008$)
Mid
$14.4
$179.4
$652.7
$2,474.5
$258.5
$1,662.9
$411.8
$6.1
$10.1
$1,750.6
$7,421.0
High
$15.9
$197.9
$653.3
$2,583.8
$301.6
$1,738.3
$468.7
$6.7
$12.0
$2,034.4
$8,012.6
Average turbidity reductions shown as 0.0 are not actually zero, but not sufficiently large to show at this level of significant digits
2 The estimated turbidity reductions for Region 10 are heavily influenced by six drinking water treatment facilities on a reach with the
estimated TSS concentration greater than 5,000 mg/L in the baseline. The estimated TSS concentration is reduced significantly post-
compliance.
Source: USEPA analysis
November 2008
9-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 9-5: Reduction in Drinking Water Treatment Costs Under Option 3
EPA
Region
1
2
3
4
5
6
7
8
9
101
Total
Average Reduction
Low
0.4
0.6
2.0
8.0
1.0
3.1
4.1
0.2
0.6 '
7.7
2.8 |
in Treated
Mid
0.6
0.9
2.9
11.8
1.8
5.8
7.5
0.4
1.0
21.8
5.0
Turbidity (NTU)
High
1.1
2.0
5.5
22.2
4.0
13.1
16.7
0.8
! 2.1
60.6
| 11.2
Cost Savings
Low
$76.7
$279.0
$837.6
$4,398.0
$302.5
$2,161.7
$450.9
$29.9
$81.0 !
$1,751.7
(thousands of 2008$)
Mid
$93.1
$361.1
$964.8
$5,272.3
$407.6
$2,644.6
$581.7
$36.5
$112.2
$2,667.5
$10,369.1 | $13,141.5
High
$101.8
$394.5
$965.7
$5,422.2
$475.5
$2,772.5
$665.4
$39.1
! $134.4
$3,488.6
| $14,459.9
1 The estimated turbidity reductions for Region 10 are heavily influenced by six drinking water treatment facilities on a reach with the
estimated TSS concentration greater than 5,000 mg/L in the baseline. The estimated TSS concentration is reduced significantly post-
compliance.
Source: USEPA analysis
9.11 Sources of Uncertainty/Limitations
> Sediment filtration systems and pre-sedimentation (allowing water to sit and sediment to filter out
before treatment) at drinking water treatment facilities reduce the sediment concentration of the
water before it enters chemical treatment, so that the turbidity level of the water entering the
facility is not the turbidity level that is eventually treated with coagulants. Assuming that the
differential between pre- and post- compliance sediment concentration is proportional to the
differential between pre- and post-compliance turbidity treatment introduces uncertainty, as the
lower sediment levels may be more or less affected by the pre-sedimentation and filtration
processes. EPA's analysis attempts to account for this uncertainty by varying the maximum
amount of TSS and turbidity treated by a given drinking water treatment facility.
> If a drinking water treatment facility produces sludge that is toxic (due to other pollutants in the
water besides sediment), its disposal costs may be significantly higher because toxic sludge
disposal is more restricted and costly. If the facility cannot separate the sludge generated by
sediment treatment from the sludge generated by treatment of toxics (which is likely the case),
then all of its sludge will be characterized as toxic. This analysis may understate the cost of
disposal (and thus the cost savings of smaller quantities of sludge to be disposed of) for facilities
that generate toxic sludge.
> The analysis presumes that reduced costs of drinking water treatment will not result in
appreciable changes in prices (costs) of drinking water paid by households. Given the marginal
costs of reduced turbidity treatment relative to other costs of drinking water production, this
assumption seems valid. However, if substantial price effects do occur, additional benefits will
accrue at the consumer level.
November 2008
9-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
10 Benefits from Water Quality Improvements
As discussed in the preceding chapters of this document, sediments and other pollutants from construction
sites may have a wide range of effects on water resources located in the vicinity of the construction sites.
These environmental changes affect environmental services valued by humans, including recreation,
commercial fishing, public and private property ownership, navigation, water supply and use, and
existence services such as aquatic life, wildlife, and habitat designated uses. These nonmarket benefits
(Freeman 2003) are in addition to market benefits (e.g., avoided costs of producing various market goods
and services) estimated in prior chapters.
This chapter describes the use of meta-analysis of surface water valuation studies for estimating benefits
of water quality improvements resulting from the regulation. The technical details involved in the
estimation of original meta-analyses are presented in Appendix A as well as in sources such as Johnston
et al. (2005; 2006), Bateman and Jones (2003), Shrestha et al. (2007), Rosenberger and Phipps (2007),
and USEPA (2004c).
10.1 Water Quality Index
To link water quality changes from reduced sediment discharge to effects on human uses and support for
aquatic and terrestrial species habitat, this analysis utilizes a water quality index (WQI). The index
translates water quality measurements, gathered for multiple parameters that are indicative of various
aspects of water quality, into a single numerical indicator that can be more easily interpreted by policy
makers and the general public. The parameters used in formulating the WQI are determined based on
water body type, the scientific understanding of ecosystem response to varying conditions, and available
data.
Most importantly for the present analysis, the WQI provides the link between specific pollutant levels, as
reflected in individual index parameters (e.g., dissolved oxygen concentrations), and the presence of
aquatic species and suitability for particular recreational uses. The WQI value, which is measured on a
scale from 0 to 100, reflects varying water quality, with 0 for poor quality and 100 for excellent.
The WQI used in this analysis is based on a methodology originally developed by McClelland (1974),
which EPA subsequently modified to rely on fewer parameters (USEPA 2002, USEPA 2004c). For this
analysis, EPA further modified the WQI methodology to apply to estuarine reaches, building on prior
applications of WQI concepts to saltwater environments (Harrison et al. 2000; CCME 2001; Carruthers
and Wazniak 2003; Gupta et al. 2003; and USEPA 2007c).
The WQI methodology used for rivers and streams and for estuaries are described in Section 10.1.1 and
Section 10.1.2, respectively.
10.1.1 WQI Methodology for Rivers and Streams
McClelland (1974) relied on a survey of water quality experts to identify water quality parameters, the
structure and the relationship between the parameters, and the weights given to the parameters in the WQI
equation. This is a procedure known as the Delphi Method (Dalkey 1968). Following McClelland's
approach, EPA estimated the WQI for a given reach using a weighted geometric mean function as
follows:
November 2008
10-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
WQIr=YlQ,W' (Eq. 10-1)
Where:
WQIr = the multiplicative water quality index (from 0 to 100) for reach r
Qi = the water quality subindex measure for parameter /'
Wj = the weight of the i-th parameter
n = the number of parameters
Table 10-1 provides the parameters and their weights for inland rivers and streams (freshwater).
Table 10-1: Original and Revised Weights for Freshwater WQI Parameters
Original Weights Revised Weights
Parameters1 (McClelland 1974) (USEPA 2002)
Dissolved oxygen (% saturation and mg/L)
Fecal coliform (colonies/100 mL)
Biochemical oxygen demand5 (mg/L)
Total nitrogen (mg/L)
Total phosphorus (mg/L)
Total suspended solids (TSS(mg/L))
pH (standard)
Temperature (C)
Total Solids (mg/L)
Total
0.17
0.16
0.11
0.10
0.10
0.08
0.11
0.10
0.07
1.00
0.24
0.22
0.15
0.14
0.14
0.11
—
—
—
1.00
See Table 10-2 for detail on developing parameter quality values.
Source: McClelland 1974; USEPA 2002
To calculate the WQI, EPA first scaled the data parameters to a standard (0-100) range. The parameter
quality values are based on parameter concentrations in the affected water bodies and quality curves
developed by McClelland (1974) and updated by the Oregon Department of Environmental Quality for
the Oregon Water Quality Index (OWQI) (Cude 2001; Dunnette 1979). Table 10-2 presents parameter-
specific functions used in developing parameter quality values (hereafter referred to as water quality
subindexes).
November 2008 10-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-2: Freshwater Water Quality Subindex
Parameter
Concentrations
Concentration
Unit
Subindex
Dissolved Oxygen (DO)
DO saturation < 100%
DO
DO
DO
100% < DO saturation < 275%
DO
275% < DO saturation
DO
<3.3
3.3 1,600
lbs/100 mL
lbs/100 mL
lbs/100 mL
98
98 * exp((FC - 50) * -9.9178 E-4
10
Total Nitrogen (TN)
TN
TN
<3
>3
mg/L
mg/L
100 * exp(TN * -0.4605)
10
Total Phosphorus (TP)
TP
TP
<0.25
>0.25
mg/L
mg/L
100 - 299.5 * TP - 0.1384 * TPA2
10
Total Suspended Solids (TSS)
TSS
TSS
TSS
<40
40 < TSS < 300
>300
mg/L
mg/L
mg/L
100
124.7 * exp(TSS * -5.552 E-3)
10
Biochemical Oxygen Demand, 5-day (BOD)
BOD
BOD
<8
>8
mg/L
mg/L
100*exp(BOD*-0.1993)
10
Source: McClelland 1974; Cude 2001
10.1.2 WQI Methodology for Estuaries
The methodology used in calculating a WQI score for estuaries and coastal water bodies adapts the
approach outlined above for rivers and streams to account for the different characteristics and response of
saltwater systems. EPA reviewed several applications of WQI concepts to saltwater systems (Harrison et
al. 2000; Carruthers and Wazniak 2003; Gupta et al. 2003; and USEPA 2007c) to identify water quality
parameters pertinent to estuaries and their relative importance in characterizing overall water quality.
The relevant parameters include parameters that are common to both freshwater and saltwater systems
(dissolved oxygen, nitrogen, phosphorus, fecal coliform, and suspended solids). They also include
chlorophyll-a (ChA). Chlorophyll-a assesses the amount of phytoplankton present in the water body.
While phytoplankton is a component of the water body's food web, changes in its abundance (e.g., high
ChA), species composition, and productivity are commonly the first biological response to nutrient
enrichment (Chesapeake Bay Program 2008). Subsequent decomposition of the phytoplankton may
reduce available dissolved oxygen and induce hypoxia and anoxia.
November 2008
10-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
The weighting scheme used to combine the parameters is a modification of the one used for the
freshwater WQI. Table 10-3 provides the parameters and their weights.
Table 10-3: Weights for Estuarine WQI Parameters
Parameters
Estuary WQI Weights1
Dissolved oxygen (% saturation and mg/L)
0.26
Fecal coliform (colonies/1 OOmL)
0.25
Total nitrogen (mg/L)
0.15
Total phosphorus (mg/L)
0.15
Total suspended solids (SSC(mg/L))
0.11
Chlorophyll-a (mg/L)
0.08
Total
1.00
Weights used for rivers and streams were redistributed to account for the removal of BOD and
the addition of ChA.
Similarly to the WQI for rivers and streams, the data for each parameter are mapped to a standard (0-100)
range using subindex curves. Table 10-4 presents parameter-specific functions used in developing water
quality subindexes for estuaries. The transformation curves are based on the ones used for freshwater
systems, with the exception of TSS, for which the curve is adjusted to account for the range of
concentrations characteristic of estuarine conditions. The curve used for ChA is derived from the curves
used for TN, since both parameters indicate trophic status.
Table 10-4: Estuarine Water Quality Subindex
Parameter
Concentrations
Concentration
Unit
Subindex
Dissolved Oxygen (DO)
DO saturation <100%
DO
DO
DO
100% < DO saturation < 275%
DO
275% < DO saturation
DO
<3.3
3.3 1,600
lbs/100 mL
lbs/100 mL
lbs/100 mL
98
98 *exp((FC- 50)* -9.9178
10
E-4
Total Nitrogen (TN)
TN
TN
<3
>3
mg/L
mg/L
100 * exp(TN * -0.4605)
10
Total Phosphorus (TP)
TP
TP
<0.25
>0.25
mg/L
mg/L
100- 299.5 *TP- 0.1384 * TP
10
A2
November 2008
10-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-4: Estuarine Water Quality Subindex
Parameter
Concentrations
Concentration
Unit
Subindex
Total Suspended Solids (TSS)
TSS
TSS
TSS
<40
40 < TSS < 550
>550
mg/L
mg/L
mg/L
100
1 19 *exp(TSS* -0.0045)
10
Chlorophyll-a (ChA)
ChA
ChA
<40
>40
ug/L
ug/L
100 *exp(ChA * -0.05605)0
10
Source: McClelland 1974; Cude 2001
A similar correspondence between WQI and suitability for potential uses is then applied to relate the
estuarine WQI score to use attainability.
10.1.3 Relation between WQI and Suitability for Human Uses
Vaughan (1986) used a reduced set of parameters in a reweighted equation to develop a water quality
ladder (WQL) that can be used to indicate whether water quality is suitable for various human uses (i.e.,
boating, rough fishing, game fishing, swimming, and drinking without treatment). Vaughan identified
"minimally acceptable parameter concentration levels" for each of the five potential uses. Water quality is
deemed acceptable for each use if none of the six parameters exceeds the threshold concentration levels.
Vaughan used a scale of zero to 10 to classify water quality based on its suitability for potential uses.
Table 10-5 presents water use classifications and the corresponding WQL and WQI values.
Table 10-5: Water Quality Classifications
Water Quality Classification
WQL Value
WQI Value1
... drinking without treatment
9.5
95
swimming
7.0
70
game fishing
5.0
50
... rough fishing
4.5
45
... boating
2.5
25
1 The WQI value corresponding to a given classification of water quality equals the WQL value multiplied by 10.
Source: Vaughan (1986)
10.1.4 Sources of Data on Ambient Water Quality
For river and streams, EPA used the following data sources to obtain ambient concentrations for the six
parameters included in the reduced form of the WQI:
> The USGS National Water Information System (NWIS) provided concentration data for three
parameters: (1) fecal coliform, (2) dissolved oxygen, and (3) biochemical oxygen demand.
> The USGS model outputs provided data for nitrogen and phosphorus concentrations, as well as
suspended sediment concentrations. The suspended sediment concentrations were estimated for
the baseline scenario and each policy option. EPA converted suspended sediments measurements
November 2008
10-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
to TSS by dividing the SPARROW-generated values by 1.3. This factor compensates for a
30 percent negative bias in TSS concentrations (Gray et al. 2000).
> EPA's Storage and Retrieval (STORET) data warehouse provided additional data on fecal
coliform counts and biochemical oxygen demand (EPA 2008e). The NWIS database was not
available for 88.6 percent and 91.3 percent of the 9,444 RF1 reaches, respectively. To address
these data gaps, EPA augmented fecal coliform and biochemical oxygen demand data by adding
observations from the STORET data warehouse. The addition of these observations increased the
fecal coliform and biochemical oxygen demand data, so that 71.9 percent and 33.8 percent,
respectively, of the 9,444 reaches were covered.
The following data sources provided data for estuaries and coastal water bodies:
> EPA's Environmental Monitoring & Assessment Program, National Coastal Assessment
Monitoring Data (EMAP-NCA) database provided data for approximately 3,000 sampling
stations in the United States, for all pertinent parameters except fecal coliform.
> The National Estuarine Research Reserve System database (NERRS) provided data for 25
estuarine reserves, each containing multiple water monitoring stations, for all relevant parameters
except fecal coliform.
> EPA's STORET data warehouse provided additional data on all relevant parameters, including
fecal coliform counts, but excluding chlorophyll-a (EPA 2008e).
The analysis of baseline WQI is based on 57,531 reaches in the conterminous United States.10 EPA
removed from the analysis all reaches with TSS concentration in the 95th percentile or above, considering
them to be outliers. Ambient concentrations of WQI parameters were available for a total of 9,444 reaches
(9,122 of which were included in the analysis after removing all outliers). EPA used a successive average
approach to address the data gaps in the remaining freshwater reaches. If ambient concentrations for a
WQI parameter were available for a reach in a given HUC, the data were extrapolated to all reaches in
that HUC under the assumption that reaches in the same watershed share similar characteristics. A
successive average was calculated at the HUC8, HUC6, and HUC2 level for all reaches that did not fall
into a smaller HUC level.n Using this estimation approach, EPA was able to compile water quality data
for a total of 57,206 freshwater reaches (including 48,084 reaches estimated from HUC-based averages).
In the case of estuaries and coastal reaches, EPA used a similar approach to estimate missing average
concentrations for reaches based on the average concentration at the HUC level. Thus, for RF1 reaches
where no empirical measurement was found, the parameter value was estimated by averaging the
corresponding values for adjacent estuarine/coastal reaches within the same HUC8 watershed. Due to the
scale of estuarine water bodies, EPA used HUC8 to estimate average parameter concentrations and did
not estimate values based on HUC 6 or HUC2. Using this approach, EPA compiled ambient water quality
data for 2,068 coastal reaches in the conterminous United States; 325 of these reaches had data empirical
data available at the HUC8 level and were included in the analysis.
No ambient concentration data were available for Alaska and Hawaii.
HUCs are catalogue numbers with 8 to 14 digits, with each set of 2 digits giving more specific information about the water.
The first pair of values designate the region (of which there are 21), the next pair the subregion (total of 222), the third pair
the basin or cataloguing unit (total of 352), and the fourth pair the subbasin, or accounting unit (total of7,262) (USGS
2007a). Digits after the first eight offer more detailed information, but are not always available for all waters. A HUC level
refers to a set of waters that have that number of HUC digits in common. For example, the HUC6 level includes all waters
for which the first six digits of their HUC are the same.
November 2008 10-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
EPA then calculated baseline WQI for all reaches based on measured or estimated parameters, and
baseline TSS concentrations estimated in SPARROW. The WQI was calculated using the methodology
outlined above according to the type of reach (either an inland stream or an estuarine reach). SPARROW
identifies estuarine and coastal reaches using a specific flag.n Table 10-6 shows the distribution of river
miles by WQI values and EPA Region for the baseline scenario (current conditions).
Table 10-6: Percentage of River Miles in Conterminous 48 States
by WQI Classification for EPA Regions: Baseline Scenario
EPA Region
1
2
3
4
5
6
1
8
9
10
National Average
Source: U.S. EPA analysis
WQK25
0.18%
0.33%
4.70%
1.21%
8.95%
2.55%
28.03%
5.05%
19.69%
0.08%
7.05%
25
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
0.5< AWQI. For each combination of the baseline water quality category and the improvement range, the
Agency estimated average AWQI and the corresponding percentage of total river miles in the state.
Table 10-7, Table 10-8, and Table 10-9 summarize changes in ambient water quality resulting from
Options 1, 2, and 3, respectively. Appendix D provides more detail on water quality improvements by the
baseline water quality category.
EPA estimated that Option 1 will not result in significant water quality improvements; this option
improves water quality in only 9,325 RF1 miles or 1.6 percent of the 585,368 RF1 miles receiving
construction discharges. EPA Region 4 is essentially the only geographic region that shows any
significant water quality improvements, with 9,177 RF1 miles (10.5 percent of RF1 miles receiving
construction discharges in this region) being affected. Region 5 is expected to have the next largest water
quality improvement with 96 RF1 miles being affected under the post-compliance scenario.
EPA's Option 2 is estimated to improve ambient water quality in 11,418 (19.0 percent) RF1 miles that
receive construction discharges nationwide. EPA's Regions 2, 3, and 4 are each estimated to experience
improved water quality on 35 percent or greater of RF1 miles included in the analysis. EPA's water
quality analysis predicts that Regions 3 and 4 will experience the most improvements in terms of both
RF1 miles (18,844 and 48,229) and percentage of total regional reach length (64.6 percent and 55.2
percent). EPA's analysis also indicates that Region 4 would benefit the most in terms of the estimated
magnitude of water quality improvements with 5.5 percent of RF1 miles that receive construction
discharges estimated to improve by greater than 0.5 WQI units. Conversely, EPA Region 8 is estimated to
improve the least, with 288 river miles benefiting from higher water quality.
Option 3 yields the most significant results overall with 136,488 RF1 miles expected to improve under the
post-compliance scenario. The estimated scale of improvements ranges from 3.6 percent to 71.4 percent
of RF1 miles in EPA's Region 8 and 3, respectively. EPA estimated that Region 4 will experience the
largest water quality improvements, with 54,727 RF1 miles (62.7 percent of the RF1 miles that receive
Construction discharges) affected. EPA's analysis also shows that Region 4 will have the most RF1 miles
(8,486) improving by greater than .5 WQI units, while Region 9 will have the lowest water quality
improvements with 2,193 miles showing any improvement under the post-compliance scenario.
November 2008 10-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-7: Estimated Water Quality Improvements Under Option 1
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Total
Baseline Scenario
RFI Miles
Receiving
Construction
Discharges
16,952
15,888
29,183
87,312
71,508
82,855
54,018
113,710
47,474
66,468
585,368
Miles of
River in
RFI
Network
18,293
16,110
33,617
94,464
71,550
98,669
60,909
130,311
56,259
69,495
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
River
Miles
0
20
0
8,207
96
0
0
32
0
0
8,355
% of River
Miles
Receiving
Construction
Discharges
0.0%
0.1%
0.0%
9.4%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
1.4%
%of
Total
River
Miles
0.0%
0.1%
0.0%
8.7%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
1.3%
0.1 < AWQI < 0.5
River
Miles
0
0
0
738
0
0
0
0
0
0
738
% of River
Miles
Receiving
Construction
Discharges
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.1%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.1%
0.5 < AWQI
River
Miles
0
0
0
232
0
0
0
0
0
0
232
% of River
Miles
Receiving
Construction
Discharges
0.0%
0.0%
0.0%
0.3%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
<0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.2%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Total Improved Reaches
River
Miles
0
20
0
9,177
96
0
0
32
0
0
9,325
% of River
Miles
Receiving
Construction
Discharges
0.0%
0.1%
0.0%
10.5%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
1.6%
%of
Total
River
Miles
0.0%
0.1%
0.0%
9.7%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
1.4%
Source: U.S. EPA analysis
November 2008
10-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-8: Estimated Water Quality Improvements Under Option 2
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
National
Total
Baseline Scenario
RFI Miles
Receiving
Construction
Discharges
16,952
15,888
29,183
87,312
71,508
82,855
54,018
113,710
47,474
66,468
585,368
Miles of
River in
RFI
Network
18,293
16,110
33,617
94,464
71,550
98,669
60,909
130,311
56,259
69,495
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
River
Miles
1,510
5,345
16,244
30,169
7,156
5,734
2,310
278
342
4,545
73,632
% of River
Miles % of
Receiving Total
Construction River
Discharges Miles
8.9% 8.3%
33.6% 33.2%
55.7% 48.3%
34.6% 31.9%
10.0% \ 10.0%
6.9% 5.8%
4.3% 1 3.8%
0.2% 0.2%
0.7% I 0.6%
6.8% 6.5%
12.6% : 11.3%
O.K AWQI < 0.5
River
Miles
38
196
2,356
13,243
990
6,850
1,004
0
46
2,370
27,093
% of River
Miles
Receiving
Construction
Discharges
0.2%
1.2%
8.1%
15.2%
1.4%
8.3%
1.9%
0.0%
0.1%
3.6%
4.6%
%of
Total
River
Miles
0.2%
1.2%
7.0%
14.0%
1.4%
6.9%
1.6%
0.0%
0.1%
3.4%
4.2%
0.5 < AWQI
River
Miles
0
7
244
4,817
138
3,924
477
9
0
1,076
10,693
% of River
Miles % of
Receiving Total
Construction River
Discharges Miles
0.0% 0.0%
0.0% 0.0%
0.8% 0.7%
5.5% 5.1%
0.2% I 0.2%
4.7% 4.0%
0.9% 1 0.8%
0.0% 0.0%
0.0% 1 0.0%
1.6% 1.5%
1.8% : 1.6%
Total Improved Reaches
River
Miles
1,548
5,548
18,844
48,229
8,283
16,507
3,791
288
388
7,991
111,418
% of River \
Miles i %of
Receiving '-. Total
Construction \ River
Discharges ; Miles
9.1% ; 8.5%
34.9% 34.4%
64.6% : 56.1%
55.2% 51.1%
11.6% 1 11.6%
19.9% 16.7%
7.0% 1 6.2%
0.3% 0.2%
0.8% I 0.7%
12.0% 11.5%
19.0% : 17.1%
Source: U.S. EPA analysis
November 2008
10-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-9: Estimated Water Quality Improvements Under Option 3
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
National
Total
Baseline Scenario
RFI Miles
Receiving
Construction
Discharges
16,952
15,888
29,183
87,312
71,508
82,855
54,018
113,710
47,474
66,468
585,368
Miles of
River in
RFI
Network
18,293
16,110
33,617
94,464
71,550
98,669
60,909
130,311
56,259
69,495
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
River
Miles
2,754
6,864
16,843
26,844
9,310
5,212
2,973
3,603
1,786
4,489
80,678
% of River
Miles
Receiving
Construction
Discharges
16.2%
43.2%
57.7%
30.7%
13.0%
6.3%
5.5%
3.2%
3.8%
6.8%
13.8%
%of
Total
River
Miles
15.0%
42.6%
50.1%
28.4%
13.0%
5.3%
4.9%
2.8%
3.2%
6.5%
12.4%
0.1 < AWQI < 0.5
River
Miles
192
621
3,631
19,397
1,883
7,095
1,041
362
306
3,367
37,895
% of River
Miles %of
Receiving Total
Construction River
Discharges Miles
1.1% 1.0%
3.9% ; 3.9%
12.4% 10.8%
22.2% 1 20.5%
2.6% 2.6%
8.6% I 7.2%
1.9% 1.7%
0.3% I 0.3%
0.6% 0.5%
5.1% i 4.8%
6.5% 1 5.8%
0.5 < AWQI
River
Miles
9
30
355
8,486
414
5,395
735
128
101
2,261
17,915
% of River
Miles
Receiving
Construction
Discharges
0.1%
0.2%
1.2%
9.7%
0.6%
6.5%
1.4%
0.1%
0.2%
3.4%
3.1%
%of
Total
River
Miles
0.0%
0.2%
1.1%
9.0%
0.6%
5.5%
1.2%
0.1%
0.2%
3.3%
2.8%
Total
River
Miles
2,956
7,514
20,830
54,727
11,606
17,702
4,749
4,092
2,193
10,118
136,488
Improved Reaches
% of River
Miles
Receiving
Construction
Discharges
17.4%
47.3%
71.4%
62.7%
16.2%
21.4%
8.8%
3.6%
4.6%
15.2%
23.3%
%of
Total
River
Miles
16.1%
46.6%
62.0%
57.9%
16.2%
17.9%
7.8%
3.1%
3.9%
14.6%
21.0%
Source: U.S. EPA analysis
November 2008
10-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
10.2 Total Benefits of Water Quality Improvements
To estimate nonmarket benefits of water quality improvements resulting from the regulation, EPA used a
benefits transfer function based on meta-analysis results presented in Appendix E of this report. The
general approach follows standard methods illustrated by Johnston et al. (2005) and Shrestha et al. (2007),
among many others (see Rosenberger and Phipps 2007). This function allows the Agency to forecast
WTP based on assigned values for model variables, chosen to represent a resource change in the C&D
policy context. EPA's meta-analysis results imply a simple benefit function of the following general
form:
In(WTP) = intercept + ^(coefficient; )(Independent Variable Values;) (Eq. 10-2)
Here, \n(WTP) is the dependent variable in the meta-analysis—the natural log of WTP for water quality
improvements. The metadata include independent variables characterizing specific details of the
resource(s) valued, such as the baseline resource conditions; the extent of resource improvements and
whether they occur in estuarine or freshwater; the geographic region and scale of resource improvements
(e.g., the number of water bodies); resource characteristics (e.g., baseline conditions, the extent of water
quality change, and ecological services affected by resource improvements); characteristics of surveyed
populations (e.g., users, nonusers); and other specific details of each study. Table 10-10 provides the
estimated regression equation intercept (5.58), variable coefficients (coefficient?), and the corresponding
independent variable names. Appendix E provides detail on the metadata, model specification, and
justification for the functional form.
EPA assigned a value to each model variable corresponding with theory, characteristics of the water
resources, and sites affected by the regulation and the policy context. Table 10-10 presents a complete list
of assigned variable values.
EPA followed Johnston et al. (2006) in assigning values for methodological attributes (i.e., variables
characterizing the study methodology used in the original source studies), which are set at mean values
from the metadata except in cases where theoretical considerations dictate alternative specifications. This
follows general guidance from Bergstrom and Taylor (2006) that meta-analysis benefit transfer should
incorporate theoretical expectations and structures, at least in a weak form. In this instance, three of the
methodology variables, discrete, WQI study, and outlier_bids are all included with an assigned value of
one. Year_index is given the value of 9.49, which corresponds to the mean year that the studies were
conducted, 2002. Nonparam is set to zero because most studies included in metadata used parametric
methods to estimate WTP values. Other study and methodology variables (volunt, mail, lump_sum,
non_reviewed, median_WTP} are assigned a zero value.
EPA used state-specific median household income, as provided by the U.S. Census Bureau's 2006
American Communities Survey (U.S. Census Bureau 2006a), to assign a value for the income variable
(income). To be consistent with the WTP estimates, which are all estimated in 2008 dollars, EPA used the
Consumer Price Index to adjust the value from the 2006 survey to 2008 dollars. The variable nonusers
was set to zero because water quality improvements resulting from reduced sediment discharge from
construction sites would benefit both users and nonusers of the affected resources.
The regulation is expected to affect water quality at a regional level because construction sites are located
throughout the entire United States. The Agency set a dummy variable denoting multiple regions (mr) to
zero because the expected magnitude of water quality improvement is relatively modest and, as a result,
November 2008 10-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
EPA's analysis focuses primarily on in-state or local resource improvements. The Mountain Plain
regional dummy variable (mp) is set to zero because the magnitude of the regional effect suggests that
spurious or otherwise unexplained effects (e.g., the effect of specific researchers who appear more than
once in the data) may drive their overall magnitude. Appendix E provides more detail on regression
results.
To account for the regional scale of the water quality effect in freshwater bodies resulting from the
regulation, the variable regional Afresh is assigned a value of one. Other variables relating to water body
type (i.e., single_lake, single_river, salt_pond, multiple_river, num_riv_pond) are set to zero. In
estimating WTP for water quality improvements in estuaries and coastal reaches, EPA set all water body
type variables (including regionalAfresh) to zero because the default value is a resource change taking
place in estuaries.13
Water quality improvements resulting from the proposed regulation are likely to enhance a variety of
water resource uses, including fishing, swimming, and boating. Therefore, variables denoting multiple
uses (allmulf) and recreational fishing (ftsh_use) are assigned a value of one, while the variable denoting
nonspecified uses (nonspec) is set to zero. However, the variable fishplus is given a value of zero because
it is unlikely that the regulation will cause more than a 50 percent increase in the fish population. Baseline
water quality (baseline) and change in water quality (quality_ch) are assigned state-specific values as
described in Section 10.1: Water Quality Index. For a broader discussion of issues involved in the
specification of variable levels for meta-analysis benefit transfer, see Johnston et al. (2005, 2006) among
others.
The water quality model used in this analysis (SPARROW) does not go beyond the terminal reach to estimated water quality
in large water bodies, and these water bodies are not fully characterized. However, EPA did calculate change in WQI for
several more complex systems such as the Great Lakes and estuaries using ambient water quality data from other sources
discussed in Section 11.1.4. EPA used mixing zone dilution factors to estimate the expected change in WQI in complex
water bodies under the post-compliance scenario (see Chapter 6 for detail).
November 2008 10-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-10: Independent Variable Assignments
Variable
Type
Study and
Methodology
Variables
Surveyed
Population
Water Body
Type
Variable
intercept
year index
discrete
volunt
mail
lump sum
WQI
nonparam
non reviewed
median WTP
outlier bids
income
nonusers
single river
single lake
Coefficient
5.5802
-0.0774
-0.1198
-1.3255
-0.2198
0.5468
-0.3238
-0.6732
-0.2757
-0.514
-0.8736
0.00000247
-0.4093
-0.3971
-0.0575
Assigned
Value
1
9.49
1
0
0
0
1
0
0
0
1
Varies
0
0
0
Explanation
Set to one as a default value.
Set to 9.49 to reflect the mean year that the
studies in the dataset were conducted.
Set to one to reflect survey efforts that
employed discrete choice elicitation methods,
which are preferred over other approaches,
such as open-ended and payment card
methods.
Set to zero because hypothetical voluntary
payment mechanisms are not even potentially
incentive compatible (Mitchell and Carson
1989)
Set to zero because mail surveys are believed
to be of less quality than in-person interviews.
Set to zero because the policy option will be
paid for over a period of years.
Set to one because of the methodological use
of the WQI in the meta-analysis.
Set to zero because most studies used in the
meta analysis used regression analysis to
calculate willingness to pay values
Set to zero because studies published in peer-
reviewed journals are preferred.
Set to zero because only average or mean
WTP values in combination with the number
of affected households will mathematically
yield total benefits if the distribution of WTP
is not perfectly symmetrical.
Set to one because survey data that exclude
such responses are preferable; outlier bids are
often excluded from the analysis of stated
preference data because these bids (often
identified as greater than a certain percentage
of a respondent's income) may indicate that
a respondent did not consider his or her
budget constraints and or supplementary
goods.
Median annual household income data from
the 2006 American Communities Survey;
median household income values assigned
separately for each state (U.S. Census Bureau
2006a)
Set to zero in order to estimate the total value
for aquatic habitat improvements, including
both use and nonuse values; for nonuser
population, the total value of water resource
improvements includes nonuse values only
(Freeman 2003).
For valuing improvement in freshwater
bodies, regional 'Jresh is set to one because
November 2008
10-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-10: Independent Variable Assignments
Variable
Type
Variables
Geographic
Region and
Scale
Variables
Resource
Improvement
Variables
Variable
multiple river
regional Afresh
salt_pond
num riv
pond
mr
mp
allmult
nonspec
Inquality ch
fish use
fishplus
Inbase
Coefficient
-1.4554
0.1773
1.0251
0.1155
-0.8784
1.5114
-0.3755
-0.3849
0.4359
-0.3478
0.4343
0.0317
Assigned
Value
0
Varies
0
0
0
0
1
0
Varies
1
0
Varies
Explanation
the expectation is that multiple freshwater
bodies within a state would be affected by the
regulation. For saltwater reaches, all water
body type variables are set to zero because the
default value in the model is a resource
change taking place in estuaries.
Indicates the number of rivers or salt ponds
affected by a policy, and is set to zero because
this analysis assumes that the effluent
guidelines will affect the entire
watershed/region; this variable assignment is
constant across study regions.
Regional variables are omitted from the
predictive portion of the analysis (i.e., set to
zero) because the regression results suggest
that these variables may be picking spurious
or other unexplained effects (e.g., author's
effect) .
Set to one because it is assumed that multiple
species would benefit from water quality
improvements
Set to zero because the number of water
bodies affected is known.
Set to the natural log of average change in the
WQI for a given state.
Set to one because the regulation is expected
to benefit a variety of aquatic species and
therefore enhance recreational fishing
opportunities.
Set to zero because the regulation is not
expected to result in a fish population change
of 50 percent or greater.
Set to the natural log of the average base
water quality for a given state.
Source: U.S. EPA Analysis
Economic values of water quality improvements are calculated at the state level and organized by analytic
scenario. EPA calculated the state-level average of baseline and post-compliance WQI changes
corresponding to 12 analytic scenarios corresponding to four levels of baseline water quality conditions
(WQI< 25, 25
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
WTP = exp(ln_WFP + oe2/2) (Eq. 10-3)
Where:
exp(-) = the exponential operator
In WTP = the predicted natural log of WTP for a given analytic scenario
oe2 = the model residual variance (0.1883) taken from Table E-3 in Appendix E.
The total WTP regression model presented above can be used to predict WTP for each of the studies in
the database; however, estimates derived from regression models are subject to some degree of error and
uncertainty. To better characterize the uncertainty or error bounds around predicted WTP, EPA used a
procedure described by Krinsky and Robb in their 1986 Review of Economics and Statistics paper
"Approximating the Statistical Property of Elasticities." The procedure involves sampling the variance-
covariance matrix of the estimated coefficients, which is a standard output of the statistical package used
to estimate the meta-analysis model. WTP values are then calculated for each drawing from the variance-
covariance matrix and an empirical distribution of WTP values is constructed. By varying the number of
drawings, it is possible to generate an empirical distribution with a desired degree of accuracy (Krinsky
and Robb 1986). The lower or upper bound of WTP values is then identified based on the 10th and 90th
percentile of WTP values from the empirical distribution. These bounds may help decision-makers
understand the uncertainty associated with the benefit results.
EPA used the Krinsky and Robb (1986) procedure to estimate the confidence intervals of total WTP for
each EPA region, based on the results of the total WTP regression model. Table 10-11 presents the results
of these calculations. The Agency notes that this analysis provides confidence limits for WTP estimates
related to the covariance matrix of meta-analysis parameter estimates. It does not, however, assess the
sensitivity of results to changes in meta-regression model assumptions or specifications (cf Johnston et
al. 2005; 2006) or assumptions implied in benefit aggregation (cf. Loomis 1996; Loomis et al. 2000;
Bateman et al. 2006). As noted above, however, the Agency has to the extent practicable made
assumptions and specifications that lead to conservative benefit estimates.
Table 10-11 presents low, mid, and high estimates of household WTP for water quality improvements
resulting form reduced sediment discharge from construction sites by EPA region and policy option.
November 2008 10-16
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 10-11: Average Household Willingness to Pay for Water Quality Improvement by
Region (2008$)
EPA
Region
1
2
o
J
4
5
6
7
8
9
10
National
Lower
10%
Bound
$0.00
$0.00
$0.00
$0.21
$0.00
$0.00
$0.00
$0.00
$0.00
$0.00
$0.04
Option 1
Mean
$0.00
; $0.02
i $0.00 i
: $0.70 ;
$0.01
$0.00
$0.00
i $0.00
I $0.00 !
: $0.00 ;
$0.15
Upper
90%
Bound
$0.00
$0.04
$0.00
$1.54
$0.03
$0.00
$0.00
$0.01
$0.00
$0.00
$0.32
Lower
10%
Bound
$0.06
$0.65
$1.47
$2.06
$0.32
$1.43
$0.54
$0.01
$0.03
$0.91
$0.92
Option 2
Mean
$0.31
"' $3.11
: $5.90
: $5.12
$1.28
$2.28
$1.03
": $0.04
i $0.11
: $2.37 ;
$2.59
Upper
90%
Bound
$0.67
$6.27
$12.24
$11.03
$2.64
$5.91
$2.40
$0.08
$0.23
$4.65
$5.58
Lower
10%
Bound
$0.36
$1.02
$1.66
$3.19
$0.46
$1.74
$0.72
$0.16
$0.16
$1.42
$1.33
Option 3
Mean
$1.84
$4.61
$6.59 ;
$7.09
$1.68
$2.44
$1.22
$0.57
$0.62 1
$3.36 ;
$3.50
Upper
90%
Bound
$3.67
$9.07
$13.68
$15.51
$3.63
$6.71
$2.97
$1.15
$1.25
$6.50
$7.59
Source: U.S. EPA analysis
As shown in Table 10-11 the estimated national average household WTP for water quality improvements
resulting from the regulation range from $0.15 to $3.50 per household. The estimated WTP values vary
greatly across EPA regions depending on the policy option and the level of construction activity in a
given region.
EPA estimates that Option 1 is not expected to result in significant water quality improvements across all
EPA regions. Region 4 is the only geographic region that shows any significant water quality
improvements. The estimated WTP per household in Region 4 has a 10 percent lower bound of $0.21 per
household and a 90 percent upper bound of $1.54 per household, with an average WTP of $0.70. The
estimated WTP for water quality improvements in Regions 1,3,6, 7, 9, and 10 is negligible. Nationwide,
household WTP has a mean value of $0.15 and 90 percent confidence interval bounds ranging from $0.04
to $0.32.
The proposed policy option (Option 2) is estimated to improve ambient water quality in 19.0 percent of
RF1 river miles included in the analysis. The estimated national average WTP for water quality
improvements resulting from the regulation is $2.59 per household per year. Regions 3 and 4 are
estimated to see improvements in water quality in more than 40 percent of their total RF1 river miles
(Section 10.1.5). EPA's analysis suggests that Region 3 households would be willing to pay the most
($5.90 per household per year) for water quality improvements resulting from the C&D regulation.
Region 4 has the second largest household WTP of $5.12. Conversely, households located in Region 8 are
estimated to have the lowest WTP for water quality improvements from the regulation, $0.04.
Policy Option 3 yields the most significant results overall in terms of RF1 miles expected to improve
under the post-compliance scenario. The estimated scale of improvements ranges from 3.6 percent to 71.4
percent of total RF1 river miles in Region 8 and 3, respectively (Section 10.1.5). Nationwide, the 90
percent confidence interval for the estimated per-household WTP has a 10 percent lower bound of $1.33
and a 90 percent upper bound of $7.59, with a mean value of $3.50.
November 2008
10-17
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
10.3 Estimating Total WTP for Water Quality Improvements
For each policy option, EPA calculated reach-level WTP as follows. First EPA estimated mean state-level
per-household WTP for each combination of the baseline water quality category (WQI baseiine) and the
expected change in WQI (AWQPj). Then, the Agency assigned each reach in the analysis a mean
household WTP value based on reach location, baseline water quality, and change in water quality. The
WTP was then multiplied by the number of households in a given state in 2006 and the percentage of
river miles in that state that comprise a given reach. The number of households per state was calculated
by taking U.S. Census Bureau population estimates for 2006 for each state and dividing by the average
number of people per household for a given state as reported in U.S. Census Bureau (2006a, 2006b). The
total WTP equation for each reach is provided below (Eq. 10-4):
TWTPreach = WTP(WQIbaselme, AWQI) x StateHHx PercentRivrMiles)
Where:
(Eq. 10-4)
TWTPreach = the reach-level welfare change from improved water quality .
WTP = the estimated state-level per-household WTP for water quality
improvement for a given combination of the baseline water quality
category (WQI baseline) and the expected change in water quality under the
post-compliance scenario (AWQI)
StateHH = the number of households in a given state
PercentRiverMiles = the percentage of total river miles in a given state that are comprised of a
given reach.
Finally, EPA aggregated reach-level benefits to the regional level. The regional benefits for the ten EPA
regions were then combined to calculate the national benefit of the regulation. Table 10-12 presents
estimated benefits of the regulation by EPA region and policy option.
Table 10-12 :Regional Willingness to Pay for Water Quality Improvement (Millions 2008$)
EPA
Region
1
2
3
4
5
6
7
8
9
10
National
Option 1
Lower
10%
Bound
$0.00
$0.02
$0.00
$4.86
$0.04
$0.00
$0.00
<$0.00
$0.00
$0.00
$4.93
I Mean |
$0.00
$0.18
; $0.00 ;
$16.10
1 $0.28
$0.00
$0.00
$0.02
"• $0.00 ''
$0.00
i $16.57 i
Upper
90%
Bound
$0.00
$0.39
$0.00
$35.60
$0.61
$0.00
$0.00
$0.04
$0.00
$0.00
$36.63
Option 2
Lower
10%
Bound
$0.32
$6.81
$16.92
$47.57
$6.53
$18.89
$2.96
$0.05
$0.39
$4.16
$104.60
i Mean -,
$1.76
$32.66
; $67.86
$118.34
1 $25.98
$30.08
$5.62
$0.17
1 $1.74
$10.82
$295.01
Upper
90%
Bound
$3.77
$65.72
$140.83
$254.72
$53.67
$77.78
$13.11
$0.33
$3.59
$21.19
$634.72
Option 3
Lower
10%
Bound
$2.04
$10.71
$19.14
$73.77
$9.37
$22.96
$3.92
$0.64
$2.49
$6.49
$151.53
i Mean -,
$10.44
$48.32
; $75.90
$163.74
1 $34.09
$32.16
$6.67
$2.25
1 $9.63
$15.32
$398.53
Upper
90%
Bound
$20.76
$95.11
$157.42
$358.19
$73.78
$88.40
$16.22
$4.53
$19.46
$29.62
$863.49
Source: U.S. EPA analysis
November 2008
10-18
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
EPA estimates that total annual benefits of water quality improvements resulting from reduced sediment
discharge from construction sites range from $16.57 million under Option 1 to $398.53 million under
Option 3. The estimated mean regional benefits vary from $0.00 to $163.74 million per year, depending
on the level of construction activity and average rainfall in a given region and stringency of the policy
option.
As shown in Table 10-12, Option 1 is not expected to result in significant water quality improvements in
most EPA regions. Thus, this option yields the smallest benefits at the regional and national levels. The
national benefits of water quality improvements under this option have a 10 percent lower bound estimate
of $4.93 million, a 90 percent upper bound estimate of $36.63 million, and a national average of $16.57
million. Region 4 gains the most benefit from water quality improvement, with a total value of $16.10
million (97 percent of the total national benefits). Region 5 has the second largest benefits ($0.28
million), which account for 1.7 percent of the total national benefits.
Under Option 2, the average national benefits are $295.01 million. The 90 percent confidence interval has
a 10 percent lower bound of $104.60 million and a 90 percent upper bound value of $634.72 million.
Region 4 gains the most from water quality improvements resulting from the regulation ($118.34
million). EPA Region 8 receives the smallest benefits, $0.17 million.
Under Option 3, the estimated mean national benefits of water quality improvement from the regulation
are $398.53 million with a 10 percent lower bound of $151.53 and a 90 percent upper bound of $863.49
million. As with the other policy options, Region 4 receives the largest benefits from water quality
improvements, accounting for 41 percent ($163.74 million) of the total national benefits. Region 8 is
anticipated to gain least under this Option 3, with the total regional benefits estimated at $2.25 million per
year.
10.4 Uncertainty and Limitations
A number of issues are common to all benefit transfers. Benefit transfer involves adapting research
conducted for another purpose in the available literature to address the policy questions at hand. Because
benefits analysis of environmental regulations rarely affords enough time to develop original stated
preference surveys that are specific to the policy effects, benefit transfer is often the only option to inform
a policy decision. As a result, they are nearly universal in benefit-cost analyses (Smith et al. 2002).
> Benefit transfers are by definition characterized by a difference between the context in which
resource values are estimated and that in which benefit estimates are desired (Rosenberger and
Phipps 2007). The ability of meta-analysis to adjust for the influence of study, economic, and
resource characteristics on WTP can minimize, but not eliminate, potential biases (Rosenberger
and Stanley 2006; Rosenberger and Phipps 2007; Smith et al. 2002). As is typical in applied
benefit transfers, the meta-analysis model used in this analysis provides a close, but not perfect,
match to the context in which values are desired. Some related and additional limitations inherent
to the meta-analysis model and the subsequent benefit transfer include:
> The Agency notes, as detailed by Loomis (1996; Loomis et al. 2000) and Bateman et al. (2006),
among others, that there are numerous uncertainties and associated assumptions required to
aggregate WTP across spatial jurisdictions. While these uncertainties are well known, the
literature does not agree on appropriate, standardized guidance for benefit aggregations, and
applied benefit-cost analysis almost universally requires simplifying assumptions in order to
generate defensible welfare aggregations. In an ideal context, analysts would have information
November 2008 10-19
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
necessary to estimate spatially referenced distance decay relationships for all changes resulting
from policies under consideration (cf. Bateman et al. 2006). However, the Agency notes that even
the most advanced literature provides only simple illustrations of such issues, and none
methodologically sufficient to support regulatory analysis. As a result, the population considered
in the benefits analysis of the regulation does not represent all the households that are likely to
hold values for water resources in a given state. Residents of other states may hold values for
water resources outside of their home state, in particular if such resources have personal, regional,
or national significance.
^ Some resource valuation studies have found that respondents in the typical contingent market
situation may overstate their WTP compared to their likely behavior in a real-world situation.
However, the magnitude of hypothetical bias on the estimated WTP is uncertain. Following
standard benefit transfer approaches, including meta-analytic transfers, this analysis proceeds
under the assumption that each source study provides a valid, unbiased estimate of the welfare
measure under consideration (cf. Moeltner et al. 2006; Rosenberger and Phipps 2007). To
minimize potential hypothetical bias, EPA set independent variable values to reflect best benefit
transfer practices.
> The estimation of WTP may be sensitive to differences in the environmental water quality
measures. Studies that did not use the WQI were mapped to the WQI so a comparison could be
made across studies. The dummy variable (WQI) captures the effect of a study using (WQI=\) or
not using the WQI (WQI=0). It was found that studies that did not use the WQI had lower WTP
values. This may indicate that there may have been some systematic biases in the mapping of
studies that did not use the WQI. In analyzing the benefits of this regulation, EPA set the WQI to
one to reduce uncertainty in WTP estimates associated with studies that did not include WQI as a
native survey instrument. See Appendix C for a detailed discussion of water quality measures
used in the original studies included in the meta-analysis.
> Transfer error may occur when benefit estimates from a study site are adopted to forecast the
benefits of a policy site. Rosenberger and Stanley (2006) define transfer error as the difference
between the transferred and actual, generally unknown, value. While meta-analysis is fairly
accurate when estimating benefit function, transfer error may be a problem in cases where the
sample size is small. While meta-analyses have been shown to outperform other function-based
transfer methods in many cases, this result is not universal (Shrestha et al. 2007). This
notwithstanding, results reviewed by Rosenberger and Phipps (2007) are "very promising" for the
performance of meta-analytic benefit transfers relative to alternative transfer methods.
Additional limitations and uncertainties associated with the use of SPARROW are discussed in Chapter
11 of this report.
November 2008 10-20
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
11 Total Estimated Benefits
This chapter summarizes findings from EPA's analysis of the expected benefits from the proposed
regulation. EPA considered a wide range of policy options in developing this regulation and presents here
the three policy options for which a full cost-benefit analysis was performed. These options are:
> Option 1: Sediment basins of a certain capacity required on all sites larger than 10 acres draining
to one location; construction general permit for other sites.
> Option 2 (in addition to the requirements of Option 1): Turbidity standard for sites with 30 or
more acres and located in areas with an "R-factor"14 greater than 50 and that have soils that
contain more than 10 percent (by mass) particles less than 2 microns in diameter. Sites larger than
30 acres not meeting both of these conditions may obtain a waiver from the turbidity standard.
> Option 3 (in addition to the requirements of Option 1): Turbidity standard for all sites larger than
10 acres (no waivers available).
Chapter 5: Pollutant Load Modeling describes these policy options in more detail.
11.1 Summary of the Estimated Benefits
The Agency analyzed benefits of the regulation in four benefits categories: navigation, water storage,
drinking water treatment, and nonmarket benefits of water quality improvement. The previous chapters of
this report provide the details of the methods and data used in the analysis of the four monetized benefit
categories (i.e., list benefit categories). See Chapters 7, 8, and 9 for a discussion of the avoided cost
methods used to estimate benefits to navigation, water storage, and treatment. Chapter 10 and Appendix E
provide a discussion of the methods used to estimate the willingness to pay (WTP) for water quality
improvements resulting from the policy options considered for the proposed regulation.
EPA was unable to quantify benefits stemming from:
> Enhanced commercial fishing
> Industrial and agricultural use of water resources
> Reduced risk of flooding, and an increase in property values.
In addition, EPA's analysis of monetized benefits is limited to reaches included in the RF1 network.
Therefore, the monetized benefits presented in this section are underestimates because they (1) exclude
several benefit categories and (2) do not reflect benefits that will be incurred by complex water bodies
(i.e., estuaries and Great Lakes), or by streams outside of the RF1 network.
Table 11-1 presents low, midpoint, and high estimates of benefits under each policy option, consisting of
benefits to navigation, water storage, drinking water treatment, and WTP. Table 11-2, Table 11-3, and
Table 11-4 detail total benefits by EPA region, policy option, and estimate range. It should be noted that
these tables incorporate the confidence intervals of 10, 50, and 90 percent from the WTP analysis into the
low, mid, and high sensitivity analyses performed for the avoided cost estimates. Though these are
The "R-factor" refers to the rainfall-runoff erosivity factor in RUSLE, which represents the amount of rainfall and its
potential to contribute to soil erosion based on rainfall patterns. More detail is available in Chapter 5 and in the Technical
Development Document (EPA 2008b) for this regulation.
November 2008
11-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
conceptually different, they are both intended to present a range of values to account for some of the
uncertainty inherent in these estimates. The sensitivity analyses create a range by varying EPA's
assumptions underlying the analysis, while the confidence interval presents high and low bounds from the
meta-analysis regression.
All tables present benefits for navigable waterway and reservoir dredging calculated using both 3 and 7
percent discount rates. Because the discount rate only applies to two of the four monetized benefits
categories, which represent between 6 and 20 percent of total benefits (depending on estimate range and
regulatory option), varying it has little effect on the total benefits estimate. The remaining discussion
presents the benefits estimates assuming a 3 percent discount rate; the associated tables present results for
both discount rates. EPA calculated benefits for drinking water treatment and WTP using a single-year
timeframe, which did not require discounting or annualizing. All benefits presented reflect annual values.
Total national benefits vary significantly among the three regulatory options: under Option 1, the
estimated benefits range from approximately $6.4 million to approximately $39.1 million, with a
midpoint estimate of $18.4 million. Estimated avoided costs range from $1.4 million to $2.5 million, with
a midpoint of $1.8 million, and WTP varies between $4.9 and $36.6 million at the 5 and 95 percent
confidence intervals, with a mean estimate of $16.6 million.
As shown in Table 11-1, under Option 2, the estimated benefits range from $135.9 million to $683.5
million, with a midpoint estimate of $332.9 million. Nonmarket benefits estimated based on household
WTP for surface water quality improvements account for 77, 89, and 93 percent of total benefits from the
regulation in the low, mid, and high estimates, respectively. The estimated WTP for water quality
improvements from reduced sediment discharges from construction sites under Option 2 ranges from
$104.6 to $634.7 million, with a mean value of $295.0 million. The estimated cost savings to industry and
government through reduced costs of navigable waterway maintenance, reservoir dredging, and drinking
water treatment range from $31.3 million to $48.8 million per year, with a midpoint estimate of $37.9
million. Under Option 2, avoided cost benefits account for 23, 11, and 7 percent of total benefits in the
low, mid, and high estimates, respectively.
Under Option 3, total benefits are estimated to be between $209.6 and $956.1 million, with a midpoint
estimate of $469.5 million. The avoided costs are estimated to be between $58.0 and $92.6 million per
year, with a midpoint estimate of $70.9 million. WTP under Option 3 ranges between $151.5 million and
$863.5 million at the 5 and 95 percent confidence intervals, with amean value of $398.5 million.
Table 11-1: Annual Total National Benefits by Benefits Category (millions of 2008$)
Benefit Category
3% Discount Rate
Low
I Mid I
High
7% Discount Rate
Low
I Mid I
High
Option 1
Navigation
Water Storage
Drinking Water1
Avoided Costs
WTP1
Total2
$0.7
$0.6
$0.2
$1.4
$4.9
$6.4
$1.0
$0.6
$0.2
$1.8
$16.6
$18.4
$1.6
$0.6
$0.2
$2.5
$36.6
$39.1
$0.7
$0.5
$0.2
$1.3
$4.9
$6.3
$0.9
$0.5
$0.2
$1.7
$16.6
$18.3
$1.8
$0.6
$0.2
$2.6
$36.6
$39.2
Option 2
Navigation
Water Storage1
Drinking Water
Avoided Costs
WTP1
Total2
$8.9
$16.5
$5.8
$31.3
$104.6
$135.9
$12.9
$17.6
$7.4
$37.9
$295.0
$332.9
$22.1
$18.7
$8.0
$48.8
$634.7
$683.5
$8.8
$13.7
$5.8
$28.3
$104.6
$132.9
$12.6
$15.9
$7.4
$35.9
$295.0
$330.9
$23.8
$18.3
$8.0
$50.1
$634.7
$684.8
November 2008
11-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 11-1: Annual Total National Benefits by Benefits Category (millions of 2008$)
Benefit Category
3% Discount Rate
Low I Mid I High
7% Discount Rate
Low I Mid I High
Option 3
Navigation
Water Storage
Drinking Water1
Avoided Costs
WTP1
Total2
$18.9
$28.7
$10.4
$58.0
$151.5
$209.6
$27.2
1 $30.6
$13.1
1 $70.9
, $398.5
$469.5
$45.7
$32.5
$14.5
$92.6
$863.5
$956.1
$18.7
$23.8
$10.4
$52.9
$151.5
$204.4
$26.5
$27.6
$13.1
$67.3
$398.5
$465.8
$49.2
$31.8
$14.5
$95.5
$863.5
$959.0
These savings were calculated for a one-year timeframe, did not require discounting, and are equal under both discount rates.
2 Totals may not equal the sum of categories due to rounding.
Source: USEPA analysis
Table 11-2, Table 11-3, and Table 11-4 detail total monetized benefits (including benefits to navigation,
water storage, drinking water, and water quality) by region. Region 4 benefits the most from this
regulation under all policy options, as it experiences the most widespread changes in terms of improved
river miles and the most significant reductions in sediment concentrations in these river miles. Together,
the magnitude and extent of these changes produce larger cost savings and water quality improvements in
this region than in any other.
Under Policy Option 1, a construction general permit for all sites and sediment basin requirements for
large sites, total benefits range between approximately $6.4 million in the low estimate and $39.1 million
in the high estimate, with a midpoint estimate of $18.4 million. Region 4 is the sole region with expected
benefits totaling more than $1 million, with estimates ranging from $6.2 to $38.0 million. The estimated
benefits for other regions are modest at best, with midpoint estimates ranging from less than $500 in
Region 10 to $291,000 in Region 5.
Policy Option 2, which adds a turbidity standard to the regulation for construction sites meeting certain
conditions, is expected to result in significantly larger benefits in all regions. EPA expects total benefits
under this option to range between $135.9 and $683.5 million, with a midpoint estimate of $332.9
million. Under this regulatory option, Regions 2, 3, 4, 5, 6, and 10 benefit most substantially, with
midpoint estimates exceeding $25 million in each of Regions 2 through 6. Region 4 alone accounts for 41
to 48 percent of the total national benefits anticipated under this option, ranging from $65.4 million to
$284.3 million, with a midpoint estimate of $140.0 million. In Region 3, total benefits are estimated to
range between $18.5 million and $143.2 million, with a midpoint estimate of $69.7 million. In Regions 2,
5, 6, and 10, midpoint benefits estimates range between $13.4 and $38.0 million. The Agency expects
benefits for this option to be substantially larger than those under Option 1, because a turbidity standard
targeting stormwater discharge from construction sites is likely to produce higher benefits in terms of TSS
and turbidity reductions. However, areas of the country with low rainfall or low percentage of clay in
their soils, such as Regions 8 and 9, do not benefit as significantly because many sites in these areas are
likely to qualify for waivers of the turbidity standard. Conversely, construction sites are not likely to
affect sediment or turbidity pollution on a large scale in these regions.
Option 3, under which all large sites are required to meet an effluent turbidity standard, increases the
midpoint estimate of total national benefits by 41 percent compared to Option 2. This option is expected
to generate benefits of $469.5 million, with a range of $209.6 to $956.1 million between the low and high
estimates. Regions 2, 3, 4, 5, 6, and 10 are again estimated to benefit most under this option, though
Option 3 is also estimated to substantially increase benefits in Regions 8 and 9, where many areas are
November 2008 11-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
exempt from the turbidity standard under Option 2. The Agency's midpoint estimates for benefits in these
regions under Option 3 are $2.6 million and $10.2 million, respectively.
Table 11-2: Annual
EPA
Region
1
2
3
4
5
6
7
8
9
101
Total2
Benefits to
Low
$1
$31
$2
$6,273
$52
$2
$1
$7
$1
$0
$6,370
Total National
Benefits Under Option 1 (thousands of 2008$)
Navigation Discounted at 3%
! Mid
$1
$183
$3
$17,851
: $291
$2
I $1
$23
$1
$0
$18,357
High
$3
$396
$4
$38,044
$626
$4
$1
$46
$1
$0
$39,124
Benefits to Navigation Discounted at 7%
Low
$1
$31
$2
$6,169
$51
$2
$0
$6
$1
$0
$6,263
Mid
$1
$183
$3
$17,769
$290
$2
$1
$23
$1
$0
$18,273
High
$3
$396
$4
$38,157
$626
$4
$1
$46
$1
$0
$39,239
Benefits estimates in this region are not zero, but less than $500 annually.
2 Totals may not be equal to the sum of regional data because the WTP model estimates the national total rather than summing
regional totals.
Source: USEPA analysis
Table 11-3: Annual
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total1
Benefits to
Low
$425
$7,230
$18,463
$65,399
$7,421
$25,659
$4,745
$88
$440
$6,007
$135,877
Total National
Benefits Under Option 2 (thousands of 2008$)
Navigation Discounted at 3%
Mid
$1,877
'. $33,322
1 $69,720
'•• $140,044
1 $27,010
! $37,976
$7,598
i $209
1 $1,790
$13,357
$332,903
High
$3,932
$66,814
$143,177
$284,250
$54,847
$87,051
$15,247
$373
$3,661
$24,178
$683,530
Benefits to Navigation Discounted at 7%
Low
$410
$7,214
$18,377
$63,790
$7,319
$24,916
$4,496
$82
$436
$5,906
$132,945
Mid High
$1,867 $3,935
$33,302
$69,654
$138,884
$26,946
$37,501
$7,448
$66,875
$143,270
$285,328
$54,850
$87,120
$15,217
$205 $372
$1,788 $3,662
$13,292 $24,186
$330,887 $684,817
Totals may not be equal to the sum of regional data because the WTP model estimates the national total rather than summing
regional totals.
Source: USEPA analysis
November 2008
11-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table 11-4: Annual
EPA
Region
1
2
3
4
5
6
7
8
9
10
Total1
Benefits to
Low
$2,315
$11,442
$21,410
$109,071
$10,812
$33,319
$6,415
$924
$2,968
$10,901
$209,577
Total National
Benefits Under Option 3 (thousands of
Navigation Discounted at 3%
I Mid
; $10,758
$49,440
$78,646
$207,357
$35,759
: $44,255
^ $9,444
: $2,560
: $10,187
$21,048
$469,455
High
$21,178
$96,897
$160,880
$417,892
$75,696
$102,551
$19,215
$4,852
$20,143
$36,769
$956,073
2008$)
Benefits to Navigation Discounted at 7%
Low 1 Mid
$2,282 $10,737
$11,415 $49,407
$21,285
$106,228
$10,652
$78,549
$205,203
$35,657
$32,187 $43,532
$6,066
$881
$2,923
$9,234
$2,534
$10,158
$10,507 $20,797
$204,425 $465,809
High
$21,187
$96,991
$161,020
$420,483
$75,705
$102,655
$19,173
$4,846
$20,153
$36,797
$959,010
Totals may not be equal to the sum of regional data because the WTP model estimates the national total rather than summing
regional totals.
Source: USEPA analysis
11.2 Sources of Uncertainty and Limitations
EPA notes that quantifying and monetizing impacts of reducing sediment discharges from construction
sites is challenging. As a result, total national benefit estimates of the regulation are subject to the
limitations and uncertainties inherent in the valuation approaches used for assessing benefits to
navigation, water storage, drinking water treatment, and nonmarket benefits of water quality
improvement. Because the combined effect of these limitations and uncertainties is likely to
underestimate the national level of benefits of this regulation, the estimated benefits should be interpreted
in the context of these limitations.
The preceding sections of this report discuss specific limitations and uncertainties associated with
estimating benefits to navigation, water storage, drinking water treatment, and nonmarket benefits of
water quality improvement. This section summarizes the limitations inherent in EPA's analysis of
reducing discharges from the construction sites.
11.2.1 Water Quality Model Limitations
To estimate benefits of reduced sediment loadings to surface water, EPA relied on SPARROW. The
SPARROW model for suspended sediments has a number of limitations, some of which are inherent to
the methodology and some the result of the particular model application. The key model limitations are:
> Reliance on the Reach File 1 network. While the RF1 network provides reasonably
comprehensive national coverage of major rivers, streams, and other surface water bodies,
coverage is limited in certain important respects. First, RF1 network coverage is limited to the
conterminous United States, thus excluding Alaska and Hawaii. In addition, while RF1
l:500,000-scale network reaches have associated data or estimates of stream discharge and
velocity that are required to specify the SPARROW model, the network excludes the majority of
the nation's total stream mileage, and smaller streams in particular. The linear coverage of the
RF1 network is approximately 700,000 miles (USEPA 2007e). By contrast, coverage of the
USGS National Hydrographic Dataset, at 1:24,000 - 1:100,000 scale, is currently over 7 million
November 2008
11-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
miles (USGS 2007). Given that RF1 accounts only for 10 percent of the total river miles, the
impacts of construction-related sediment on smaller stream reaches are likely to be significantly
understated. As construction activities may be concentrated along lower-order streams not
included in the RF1 network, the relative share of total sediments contributed by construction
activities may be high on these reaches during active construction phases. By contrast, the
specific impacts of construction activities may diminish in importance relative to contributions
from spatially extensive and diffuse land uses, including agriculture, at the level of RF1 reaches.
> Omission of coastal waters from the analysis of monetized benefits. The SPARROW model does
not allow prediction of water quality changes in coastal waters. Therefore, all coastal waters in
the United States (10,630 shoreline miles) are omitted from the analysis of monetized benefits.15
Because the estimated willingness to pay for water quality improvements is a function of the total
river and shoreline miles that are expected to improve from reduced sediment discharges, this
omission is likely to lead to understatement of the estimated willingness to pay for water quality
improvements resulting from the proposed regulation.
> Omission of all ponds and lakes and reservoirs located off the RF1 network from the water
quality analysis. All lakes, ponds, and reservoirs located off the RF1 network are not included in
the SPARROW model and thus are excluded from estimation of monetized benefits. The National
Water Quality Inventory Report to Congress (USEPA 2007d) reports 40.6 million acres of lakes
and reservoirs in the conterminous United States. The RF1 network includes approximately 3.9
million acres or 9.5 percent of the total lakes and reservoir acres in the United States (USEPA
2007d).16 Omission of these water body types from the analysis of monetized benefits is likely to
lead to understatement of benefits in two benefit categories: (1) nonmarket benefits of water
quality improvements resulting from the regulation and (2) reservoir dredging.
> Restriction of the water quality analysis to the description of long-term mean water quality
conditions. Construction activities are, by contrast, transient in nature, extending over weeks or
months. Construction activities (unlike agricultural activities) are spatially compact, so they are a
sub-grid phenomenon with respect to the specification of the national scale of the SPARROW
model. The restriction to mean water quality conditions precludes an analysis of the frequency
with which conditions of extreme sediment transport conditions (e.g., during an active
construction period) occur. Although the predicted changes in average water quality conditions
may be small, the expected changes in sediment concentrations under extreme sediment transport
conditions may be significant.
11.2.2 Focus on One Pollutant of Concern (Sediment)
Existing case studies of environmental impacts associated with construction activities demonstrated that a
number of pollutants are found in construction site discharges, including turbidity, nitrogen, phosphorus,
BOD, metals, toxic organics, trash and debris, and other miscellaneous pollutants. However, EPA's
analysis of benefits from reduced construction site discharges focuses on total suspended solids only. This
is likely to result in underestimation of the expected water quality changes resulting from the proposed
15 EPA was able to estimate ambient concentrations of sediment (Total Suspended Solids (TSS)) within estuaries using
SPARROW output in conjunction with the Dissolved Concentration Potential (DCP) approach (Versar, Inc./USEPA 1997).
However, estuarine reaches were not included in estimation of monetized benefits due to time constraints.
16 The estimated total lake/reservoir acres do not include the Great Lakes.
November 2008 11-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
regulation because the combined impact of several pollutants on ambient water quality conditions is likely
to be greater than a single pollutant's impact.
11.2.3 Omission of Several Benefit Categories from the Analysis of Monetized Benefits
Due to data limitations, EPA did not estimate benefits in several benefit categories. Although the
magnitude of benefits in the omitted categories is uncertain, they may not be trivial. Chapter 4 of this
report provides a qualitative discussion of the omitted benefit categories. A brief summary of the omitted
benefit categories is provided below:
> Market values of properties located near water bodies. Reducing sediment discharges from
construction sites is likely to increase market values of properties located in the vicinity of
construction sites by enhancing the aesthetic quality of the affected land and water resources
(e.g., reducing erosion of river banks and improving water clarity).17
> Flood damages. Reducing sediment discharges from construction sites is expected to reduce
flooding damages by decreasing sedimentation of river beds and improving river capacity. Clark
et al. (1985) estimated flooding damages attributable to sediment discharges to be $1.5 billion
(2008$), annually. Therefore, even a small reduction in the frequency and severity of flooding is
likely to generate significant benefits.
> Ditch maintenance. The proposed regulation is expected to reduce the costs of ditch maintenance
by reducing the amount of sediment deposited in ditches.
> Industrial water use. The proposed regulation is expected to benefit industrial water users by
reducing sediment concentrations in source waters and thus increasing the useful life of industrial
equipment.
> Agricultural water use. The proposed regulation is expected to benefit agricultural producers by
reducing sediment discharges and, as a result, sediment deposition on farm land; this would lead
to improvements in land productivity and enhanced marketability of agricultural products.
The nonmarket component (i.e., increased satisfaction with the property) is implicitly accounted for in WTP for
improvements in environmental services provided by surface waters affected by construction site discharges (see Chapter 10
for detail).
November 2008 11-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
12 References
Aiken, R.A. 1985. Public Benefits of Environmental Protection in Colorado. Masters thesis, Colorado
State University.
Aldridge, D. W., B. S. Payne, and A. C. Miller. 1987. "The effects of intermittent exposure to suspended
solids and turbulence on three species of freshwater mussels." As cited in Waters 1995.
Alexander R.B., A.H. Elliott, U. Shankar U., and McBride G.B. 2002a. Estimating the sources and
transport of nutrients in the Waikato River basin. New Zealand: Water Resources Research 38:1268-
1290. As cited in Schwarz et al. 2006.
Alexander R.B., R.A. Smith, and G.E. Schwarz. 2000. Effect of stream channel size on the delivery of
nitrogen to the Gulf of Mexico. Nature 403:758-761. As cited in Schwarz et al. 2006.
Alexander, R.B., J.W. Brakebill, R.E. Brew, and RA. Smith, R.A. 1999, ERF1 - Enhanced river reach
file 1.2: U.S. Geological Survey Open-File Report 99-457. As cited in Schwarz et al. 2006.
Alexander, R.B., P.J. Johnes, E.A. Boyer, and R.A. Smith. 2002b. A comparison of methods for
estimating the riverine export of nitrogen from large watersheds. Biogeochemistry 57/58:295-339. As
cited in Schwarz et al. 2006.
Alexander, R.B., R.A. Smith, and G.E. Schwarz. 2004. Estimates of diffuse phosphorus sources in surface
waters of the United States using a spatially referenced watershed model. Water Science and
Technology 49:1-10. As cited in Schwarz et al. 2006.
Alexander, R.B., RA. Smith, G.E. Schwarz, E.W. Boyer, J.V. Nolan, and J.W. Brakebill, J.W. 2008.
Differences in phosphorus and nitrogen delivery to the Gulf of Mexico from the Mississippi River
Basin. Environ. Sci. Technol. 42(3):822-830.
Alexander, R.B., RA. Smith, G.E. Schwarz, S.D. Preston, J.W. Brakebill, R Srinivasan, and P.A.
Pacheco. 2001. Atmospheric nitrogen flux from the watersheds of major estuaries of the United States
- An application of the SPARROW watershed model, in Valigura, R.A. As cited in Schwarz et al.
2006.
Baruth, Edward E. 2005., Water Treatment Plant Design, 4th Edition. American Society of Civil
Engineers (ASCE)/American Water Works Association (AWWA). New York: McGraw-Hill.
Anderson, G.D. and S.F. Edwards. 1986. Protecting Rhode Island's Coastal Salt Ponds: An Economic
Assessment of Downzoning. Coastal Zone Management 14( i/z): 67-91.
Anderson, P. G., B. R. Taylor, and G. C. Balch. 1996. Quantifying the Effects of Sediment Release on
Fish and their Habitats. As cited in EPA 2004.
Appleby, J. P., and D. J. Scarratt. 1989. "Physical effects of suspended solids on marine and estuarine fish
and shellfish with special reference to ocean dumping: A literature review." Canadian Technical
Report of Fisheries and Aquatic Sciences No. 1681.^s citedin Kerr 1995.
Arnold, J.G., J.R. Williams, and D.R. Maidment. 1995. Continuous time water and sediment routing
model for large basins. Journal of Hydraulic Engineering 121(2): 171-183. As cited in Neitsch et al.
2005.
November 2008 12-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998. Large Area Hydrologic Modeling and
Assessment: I. Model Development. J. American Water Resources Association 34:73-89.
Arrow, K., R. Solow, E. Learner, P. Portney, R. Rander, and H. Schuman. 1993. Report of the NOAA
Panel on Contingent Valuation, Federal Register 58 (Jan.): 4602-4614.
Azevedo, C., J.A. Herriges, and C.L. Kling. 2001. Valuing Preservation and Improvements of Water
Quality in Clear Lake. Center for Agricultural and Rural Development (CARD), Iowa State
University, Staff Report 01-SR 94.
Barrett, Michael E., Joseph F. Malina, Jr., Randall J. Charbeneau, and George Ward. 1995. "Effects of
Highway Construction and Operation on Water Quality and Quantity in an Ephemeral Stream in the
Austin, Texas Area." Center for Research in Water Resources, University of Texas at Austin.
Barton, Bruce A. 1977. "Short-term effects of highway construction on the limnology of a small stream in
southern Ontario" Freshwater Biology Vol. 7: pp. 99-108.
Bateman, I.J. and A.P. Jones. 2003. Contrasting Conventional with Multi-Level Modeling Approaches to
Meta-Analysis: Expectation Consistency in U.K. Woodland Recreation Values. Land Economics
79(2): 235-258.
Bateman, I.J., B.H. Day, S. Georgiou and I. Lake. 2006. The Aggregation of Environmental Benefit
Values: Welfare Measures, Distance Decay and Total WTP. Ecological Economics 60(2): 450-460.
Bay Institute of San Francisco (BISF). 2003. San Francisco Bay Water Quality Index: Indicator Analysis
and Evaluation. The Bay Institute Ecological Scorecard. The Bay Institute of San Francisco, Novato,
CA. October.
Beaulac, M.N. and K.H. Reckhow. 1982. "An Examination of Land Use - Nutrient Export
Relationships." Water Resources Bulletin 18:1013-1024. As cited in Schwarz et al. 2006.
Bejranonda, Somskaow; Fred J. Hitzhusen; and Diane Kite. 1999. "Agricultural Sedimentation Impacts
on Lakeside Property Values" Agricultural and Resource Economics Review 28(2): 208-18.
Bell, M. C. 1973. Fisheries handbook of engineering requirements and biological criteria. As cited in
Waters 1995.
Bergstrom, J.C. and L.O. Taylor. 2006. "Using Meta-Analysis for Benefits Transfer: Theory and
Practice." Ecological Economics 60(2):351-360.
Bergstrom, J.C. and P. De Civita. 1999. Status of Benefits Transfer in the United States and Canada: A
Review. Canadian Journal of Agricultural Economics 47(1): 79-87.
Berkman, H. E., and C. F. Rabeni. 1987. "Effects of siltation on stream fish communities. "As cited in
Wood and Armitage 1997.
Berrens, R.P., P. Ganderston, and C.L. Silva. 1996. Valuing the protection of minimum instream flows in
New Mexico. Journal of Agricultural and Resource Economics 21(2):294-309.
Beven, K. 2002. Towards an alternative blueprint for a physically based digitally simulated hydrologic
response modeling system. Hydrological Processes 16:189-206. As cited in Schwarz et al. 2006.
Beven, K. and M. Kirkby. 1979. A physically-based variable contributing area model of basin hydrology.
Hydrologic Science Bulletin 24:43-69.
November 2008 12-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Bicknell, B.R., J.C. Imhoff, J.L. Kittle, T.H. Jobes Jr., and A.S. Donigian Jr. 2001. Hydrological
simulation program-Fortran, HSPF version 12 user's manual: Mountain View, CA, AQUA TERRA
Consultants, 873 p. As cited in Schwarz et al. 2006.
Bjornn, T. C., M. A. Brusven, M. P. Molnau, J. H. Milligan, R. A. Klamt, E. Chacho, and C. Schaye.
1977. "Transport of granitic sediment in streams and its effects on insects and fish." Completion
Report Project B-036-IDA, Bulletin 17, Idaho Cooperative Fishery Research Unit, University of
Idaho, Moscow, Idaho. 43 p. As cited in Kerr 1995.
Boardman, A.E., D.H. Greenberg, A.R. Vining, and D.L. Weimer. 2001. Cost-Benefit Analysis: Concepts
and Practice, 2nd edition. Upper Saddle River, NJ: Prentice Hall.
Bockstael, N.E., K.E. McConnell and I.E. Strand. 1989. Measuring the Benefits of Improvements in
Water Quality: The Chesapeake Bay. Marine Resource Economics 6: 1-18.
Boyd, W. 1990. "Effects of poultry waste on ground and surface water." In: Proceedings of the 1990
National Poultry Waste Management Symposium, Blake, J., and R.M. Hulet (eds.).
Boyle, K.J. and J.C. Bergstrom. 1992. "Benefit Transfer Studies: Myths, Pragmatism and Idealism."
Water Resources Research 28(3):657-663.
Brakebill, J.W. and S.E. Preston. 1999. Digital data used to relate nutrient inputs to water quality in the
Chesapeake Bay watershed, version 1.0: U.S. Geological Survey Water Open-File Report 99-60.
Available: http://md.usgs.gov/publications/ofr-99-60/. As cited in Schwarz et al. 2006.
Breffle, William S., Morey, E., Rowe, R., Waldman, D. and Wytinck, S. 1999. Recreational Fishing
Damages from Fish Consumption Advisories in the Waters of Green Bay. Prepared by Stratus
Consulting Inc. for U.S. Fish and Wildlife Service, U.S. Department of Interior and U.S. Department
of Justice, 1999.
Buckner, M. M., Wendy Smith, and Judy A. Takats (World Wildlife Fund). 2002. Tennessee,
Cumberland, and Moblie River Basins at Risk: A Biological Assessment and Vision for the World
Wildlife Fund's Southeast Rivers and Streams Project.
Burkholder J. M. 1998. "Implications of harmful marine microalgae and heterotrophic dinoflagellates in
management of sustainable fisheries." Ecological Applications (Supplement) Vol. 8: pp. S37-S62.
California Department of Fish and Game. 2004. "Recovery Strategy for California Coho Salmon."
Species Recover Strategy 2004-1. February, 2004.
Cameron, T.A. and D.D. Huppert. 1989. OLS versus ML Estimation of Non-market Resource Values
with Payment Card Interval Data. Journal of Environmental Economics and Management 17, 230-
246.
Cameron, Trudy Ann. 1988. Using the Basic "Auto-validation" Model to Assess the Effect of
Environmental Quality on Texas Recreational Fishing Demand: Welfare Estimates. EPA National
Center for Environmental Economics, 1988.
Canadian Council of Ministers of the Environment (CCME). 2001. Canadian water quality guidelines for
the protection of aquatic life: CCME Water Quality Index 1.0, Technical Report. In: Canadian
environmental quality guidelines, 1999, Canadian Council of Ministers of the Environment,
Winnipeg.
November 2008 12-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Canadian Department of Fisheries and Oceans. 2000. Effects of sediment on fish and their habitat. DFO
Pacific Region Habitat Status Report 2000/01.
Caraco, N.F., J.J. Cole, G.E. Likens, G.M. Lovett, and K.C. Weathers, K.C. 2003. Variation in NO3
export from flowing waters of vastly different sizes - Does one model fit all? Ecosystems 6:344-352.
As cited in Schwarz et al. 2006.
Carline, Robert F., Adam Smith, Curtis Gill, and Kirk Patten. 2003. "Effects of Bridge and Road
Construction on Spring Creek: Route 26 Transportation Improvements Project." Pennsylvania
Department of Transportation, Pennsylvania Fish and Boat Commission.
Carruthers, T. and Wazniak, C. 2003. Development of a Water Quality Index for the Maryland Coastal
Bays. Maryland's Coastal Bays: Ecosystem Health Assessment, Chapter 4.4. 2003.
Carson, R.T. and R.C. Mitchell. 1993. "The value of clean water: the public's willingness to pay for
beatable, fishable, and swimmable quality water." Water Resources Research 29(7): 2445-2454.
Carson, R.T., T. Groves, and M.J. Machina. 2000. Incentive and Informational Properties of Preference
Questions. Draft Paper. University of California, La Jolla.
Carson, R.T., W.M. Hanemann, R.J. Kopp, J.A. Krosnick, RC. Mitchell, S. Presser, P.A. Ruud, and V.K.
Smith. 1994. Prospective Interim Lost Use Value due to DDT and PCB Contamination in the
Southern California Bight. Volume 2. Report to the National Oceanic and Atmospheric
Administration, Produced by Natural Resources Damage Assessment Inc., LA Jolla, CA.
Chapman, D. W. 1988. "Critical review of variables used to define effects of fines in redds of large
salmonids."^4s cited in Canadian Department of Fisheries and Oceans 2000.
Chapra, S.C. 1997. Surface water-quality modeling. McGraw-Hill, New York. 844 p. As cited in Schwarz
et al. 2006.
Chesapeake Bay Program. 2008. "Chlorophyll A: Annual Assessment." Updated 04/03/2008. Accessed
10/30/2008. Available at http://www.chesapeakebay.net/status_chlorophyla.aspx?menuitem=19649.
Chisholm, James L., and Sanford C. Downs (USGS). 1978. Stress and Recovery of Aquatic Organisms as
Related to Highway construction along Turtle Creek, Boone County, West Virginia. Geological
Survey Water Supply Paper 2055.
Clark, Edwin, Jennifer A. Haverkamp, and William Chapman. 1985. "Eroding Soils: The Off-Farm
Impacts." Washington, DC: The Conservation Foundation.
Cleveland, Theodore G., and Adebola Fashokun. 2006. "Construction-Associated Solids Loads With a
Temporary Sediment Control BMP." Journal of Construction Engineering and Management Vol.
132(10): pp. 1122-1125.
Cline, Leo D., Robert A. Short, and James V. Ward. 1982. "The influence of highway construction on the
macroinvertebrates and epilithic algae of a high mountain stream." Hydrobiologia Vol. 96: pp. 149-
159.
Clonts, H.A., and J.W. Malone. 1990. Preservation Attitudes and Consumer Surplus in Free Flowing
Rivers. In: Social Science and Natural Resource Recreation Management, Joanne Vining, editor.
Westview Press, Boulder, CO. pp. 301-317.
Cloud, Thomas J. Jr, U.S. FWS, to Allen Ryan, Department of the Army. February 2, 2004. Letter.
Available at:
November 2008 12-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
http://www.fws.gov/southwest/es/Documents/R2ES/Ark_shiner_BNSF_railroad_BO_COE.pdf.
Accessed April 2, 2008.
Colin, T.A. 2005. Estimating contaminant loads in rivers: An application of adjusted maximum likelihood
to type 1 censored data. Water Resour. Res. 41:W07003, doi:10.1029/2004WR003833.
Conn, T.A., D.L. Caulder, E.J. Gilroy, L.D. Zynzuk, and R.M. Summers. 1992a. The validity of a simple
statistical model for estimating fluvial constituent loads: an empirical study involving nutrient loads
entering Chesapeake Bay. Water Resources Research 28(9):2353-2363. As cited in Schwarz et al.
2006.
Cohn, T.A., E.J. Gilroy, and W.G. Baier. 1992b. Estimating fluvial transport of trace constituents using a
regression model with data subject to censoring. Proceedings of the Section on Statistics and the
Environment of the American Statistical Association, p. 1-9. As cited in Schwarz et al. 2006.
Construction Cost Index. 2008. "Construction Economics." Engineering News-Record March 2008.
Cooper, C. M. 1987. "Benthos in Bear Creek, Mississippi: Effects of habitat variation and agricultural
sediments." Journal of Freshwater Ecology, Vol. 4:pp 101-114. As cited in Kerr 1995.
Cordone, A. J., and S. Pennoyer. 1960. "Notes on silt pollution in the Truckee River Drainage, Nevada
and Placer counties." As cited in Waters 1995.
Croke, K., R.G. Fabian and G. Brenniman. 1986-1987. Estimating the Value of Improved Water Quality
in an
Crowder, Bradley M. 1987. "Economic Costs of Reservoir Sedimentation: A Regional Approach to
Estimating Cropland Erosion Damages." Journal of Soil and Water Conservation 42(3): 194-197.
Cude, C. G. 2001. Oregon Water Quality Index: A Tool for Evaluating Water Quality Management
Effectiveness. Journal of the American Water Resources Association 37(1): 125-137.
Dalhuisen, J.M., R.J.G.M. Florax, H.L.F. de Groot, and P. Nijkamp. 2003. Price and income elasticities of
residential water demand: A meta analysis. Land Economics 79(2):292-308.
Dalkey, N.C. 1968. DELPHI. The Rand Corporation.
Daniel, T. C., P. E. McGuire, D. Stoffel and B. Miller. 1979. "Sediment and Nutrient Yield from
Residential Construction Sites." Journal of Environmental Quality Vol. 8(3): pp. 304-308.
De Robertis, Alex, Clifford H. Ryer, Adriana Veloza and Richard D. Brodeur. 2003. "Differential effects
of turbidity on prey consumption of piscivorous and planktivorous fish." Canadian Journal of
Fisheries and Aquatic Science Vol. 60: pp. 1517-1526.
De Zoysa, A.D.N. 1995. A Benefit Evaluation of Programs to Enhance Groundwater Quality, Surface
Water Quality and Wetland Habitat in Northwest Ohio. Dissertation, Ohio State University.
De Zuane, John. 1997. Handbook of Drinking Water Quality, 2nd Ed. New York: John Wiley & Sons.
Dearmont, David, Bruce A. McCarl, and Debora A. Tolman. 1998. "Cost of water treatment due to
diminished water quality: A case study in Texas." Water Resources Research Vol. 54(4): pp 849-853.
Desvousges, W.H., F.R. Johnson, and H.S. Banzhaf. 1998. Environmental Policy Analysis with Limited
Information: Principles and Applications of the Transfer Method. Northampton, MA: Edward Elgar
Publishers.
November 2008 12-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Desvousges,W.H., V.K. Smith and A. Fisher. 1987. Option Price Estimates for Water Quality
Improvements: A Contingent Valuation Study for the Monongahela River. Journal of Environmental
Economics and Management. Vol (14), pp. 248-267.
Downs, Sanford C., and David H. Appel (USGS). 1986. Progress Report on the Effects of Highway
Construction on Suspended-Sediment Discharge in the Coal River and Trace Fork, West Virginia,
1975-81. Water Resources Investigations Report 84-4275.
Duck, Robert W. 1984. "Short Communications: The Effect of Road Construction on Sediment
Deposition in loch Earn, Scotland." Earth Surface Processes andLandforms Vol. 10: pp. 401-406.
Dunnette, D. A. 1979. A Geographically Variable Water Quality Index Used in Oregon. As referenced in
Cude2001.
Edwards, D. 1969. "Some effects of siltation upon aquatic macrophyte vegetation in rivers."
Hydrobiologia, Vol. 34: pp 29-36. As cited in Kerr 1995.
Elliott, A.H., R.B. Alexander, G.E. Schwarz, U. Shankar, J.P.S. Sukias, and G.B. McBride. 2005.
Estimation of nutrient sources and transport for New Zealand using the hybrid physical-statistical
model SPARROW. Journal of Hydrology (New Zealand) 44:1-27. As cited in Schwarz et al. 2006.
Ellis, M. M. 1936. "Erosion silt as a factor in aquatic environments. "As cited in Waters 1995.
Forbes, A. M., J. J. Magnuson, and D. M. Harrell. 1981. Effects of habitat modifications from coal ash
effluent on stream macrobenthos: A synthesis. As cited in Kerr 1995.
Forster, D. L., C. P. Bardos, and D. D. Southgate. 1987. "Soil Erosion and Water Treatment Costs."
Journal of Soil and Water Conservation Vol. 42: pp 349-352. As cited in Dearmont et al. 1998.
Fossati, O., J. G. Wasson, C. Hery, R. Marin, and G. Salinas. 2001. "Impact of sediment releases on water
chemistry and macroinvertebrate communities in clear water Andean streams (Bolivia)." Archivfur
Hydrobiologie Vol. 151(1): pp. 33-50.
Freeman III, A.M. 2003. The Measurement of Environmental and Resource Values: Theory and Methods.
Washington, DC: Resources for the Future.
Gammon, J. R. 1970. "The effect of inorganic sediment on stream biota." Water Pollution Control
Research Series 18050 DWC 12/70, U.S. Environmental Protection Agency, Washington, DC. 141
pp. As cited in Kerr 1995.
Gassman, P.W., E. Osei, A. Saleh, and L.M. Hauck. 2002. Application of an Environmental and
Economic Modeling System for Watershed Assessments. J. American Water Resources Association
38(2):423-438. As cited in Gassman et al. 2005.
Gassman, P.W., J.R. Williams, V.W. Benson, R. Cesar Izaurralde, L.M. Hauck, C.A. Jones, J.D. Atwood,
J.R. Kiniry, and J.D. Flowers. 2005. Historical Development and Applications of the EPIC and APEX
Models. Iowa State University, Center for Agricultural and Rural Development, Working Paper 05-
WP 397.
Gassman, P.W., M. Jha, S. Secchi, and J. Arnold. 2003. Initial calibration and validation of the SWAT
Model for the Upper Mississippi River Basin. Paper 10B GIS, Diffuse Pollution Conference Dublin.
pp. 10-35-10-40.
Glass, G.V. 1976. Primary, Secondary, and Meta-Analysis of Research. Educational Researcher 5(10):
3-8.
November 2008 12-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Graham, A. A. 1990. "Siltation of stone-surface periphyton in rivers by clay-size particles from low
concentrations in suspension. "As cited in Waters 1995.
Graves, David S., and James W. Burns (The Resources Agency of California, Department of Fish and
Game). 1970. "Comparison of the Yields of Downstream Migrant Salmonids Before and After
Logging Road Construction on the South Fork Caspar Creek, Mendocino County. Inland Fisheries
Administrative Report No. 70-3.
Gray, J.R., Glysson, G.D., Turcios, L.M., and Schwarz, G.E. 2000. Comparability of Suspended-
Sediment Concentration and Total Suspended Solids Data. U.S. Geological Survey Water-Resources
Investigations Report 00-4191.
Green, W.H. and G.A. Ampt. 1911. Studies in soil physics, I. The flow of air and water through soils.
Journal of Agricultural Sciences 4:11-24. As cited in Neitsch et al. 2005.
Griffiths, C. undated. The Use of Benefit-Cost Analysis in Environmental Policy Making. Working
Paper. National Center for Environmental Economics, U.S. Environmental Protection Agency,
Washington, DC.
Guenther, Mariana, and Reinaldo Bozelli. 2004. Effects of inorganic turbidity on the phytoplankton of an
Amazonian Lake impacted by bauxite tailings. Hydrobiologia Vol. 511: pp. 151-159.
Gupta, A.K., Gupta, S.K., and Patil, R.S. 2003. A Comparison of Water Quality Indices for Coastal
Water. Journal of Environmental Science and Health 38(11):2711-2725. 2003.
Gupta, V.K. and S. Sorooshian. 1985. The relationship between data and the precision of parameter
estimates of hydrologic models. Journal of Hydrology 81:57-77. As cited in Schwarz et al. 2006.
Hallock, D. 2002. A Water Quality Index for Ecology's Stream Monitoring Program. Washington
Department of Ecology, Environmental Assessment Program, Olympia, WA. Publication No. 02-03-
052.
Harbor, Jonathan M., John Snyder, and Jim Storer. 1995. "Reducing nonpoint source pollution from
construction sites using rapid seeding and mulching." Physical Geography Vol. 16(5): pp. 371-378.
Hargrove, William W. 2007. "Brashear Island Volumetric Interpolation." Oak Ridge National Library.
Accessed April 4, 2008. Available at
http://research.esd.ornl.gov/CRERP/SEDIMENT/BRASHEAR.HTM.
Harrison, T.D., Cooper, J.A.G., and Ramm, A.E.L., 2000. Geomorphology, Ichthyofauna, Water Quality
and Aesthetics of South African Estuaries. South Africa Department of Environmental Affairs &
Tourism, Pretoria, South Africa, ENV-DC 2000-01.
Hayes, K.M., T.J. Tyrell and G. Anderson. 1992. Estimating the Benefits of Water Quality Improvements
in the Upper Narragansett Bay. Marine Resource Economics 7: 75-85.
Hedrick, Kevin P., R. Bruce Robinson, Bruce Tschantz, and Stephen E. Moore. 2006. "Impact of Tunnel
Reconstruction on Stream Water Quality in Great Smoky Mountains National Park." Journal of
Hydrologic Engineering Vol. 11(6): pp. 570-577.
Helsel, Dennis R. (USGS). 1985. Contributions of Suspended Sediment From Highway Construction and
Other Land Uses to the Olentangy River, Columbus, Ohio. Water Resources Investigations Report
84-4336.
November 2008 12-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Henley, W. E., M. A. Patterson, R. J. Neves, and A. Dennis Lemly. 2000. "Effects of Sedimentation and
Turbidity on Lotic Food Webs: A Concise Review for Natural Resource Managers." Review in
Fisheries Science Vol. 8(2): pp 125-139.
Herriges, J.A. and J.F. Shogren. 1996. Starting point bias in dichotomous choice valuation with follow-up
questioning. Journal of Environmental Economics and Management 30(1): 112-131.
Kite, D. 2002. Willingness to Pay for Water Quality Improvements: The Case of Precision Application
Technology. Deaprtment of Agricultural Economics and Rural Sociology, Auburn University,
Auburn, AL; 8 August 2002.
Holmes, Thomas P. 1988. "The Offsite Impact of Soil Erosion on the Water Treatment Industry." Land
Economics Vol. 64(4): pp 356-366.
Hoos, A.B. 2005. Evaluation of the SPARROW model for estimating transport of nitrogen and
phosphorus in streams of the Interior Low Plateau Ecoregion, Tennessee, Kentucky, and Alabama,
during 1992-2002. In Tennessee Water Resources Symposium, 15th, Burns, Tenn., 2005,
Proceedings. Tennessee Section of the American Water Resources Association, p. 2A-47. As cited in
Schwarz et al. 2006.
Howarth, R.W., G. Billen, D. Swaney, A. Townsend, N. Jaworski, K. Lajtha, J.A. Downing, R. Elmgren,
N. Caraco, T. Jordan, F. Berendse, J. Freney, V. Kudeyarov, P. Murdoch, and Z. Zhao-Liang. 1996.
Regional nitrogen budgets and riverine N and P fluxes for the drainages to the North Atlantic Ocean -
natural and human influences. Biogeochemistry 35:75-139. As cited in Schwarz et al. 2006.
Huang, J.C., T.C. Haab and J.C. Whitehead. 1997. Willingness to Pay for Quality Improvements: Should
Revealed and Stated Preference Data Be Combined? Journal of Environmental Economics and
Management 34(3): 240-255.
Hunt, C.L. and J.K. Grove. 2001. Impacts of Sedimentation to a Stream by a Non-Compliant
Construction Site. American Society of Civil Engineers Conference Proceedings, World Water
Congress 2001.
Hushak, L. and M. Bielen. 1999. Valuing the Ottawa River: The Economic Values and Impacts of
Recreational Boating. Ohio Sea Grant College Program Technical Bulletin Series Publication OSHU-
TB-042 2000, 2002.
Illinois Environmental Protection Agency. 2006. Illinois Integrated Water Quality Report and Section
303(d) List- 2006. Bureau of Water. Available at: http://www.epa.state.il.us/water/water-
quality/report-2006/2006-report.pdf
Jakeman, A. and G. Hornberger. 1993. How much complexity is warranted in a rainfall-runoff model?
Water Resources Research 29:2637-2649. As cited in Schwarz et al. 2006.
Jobson, H.E. 1996. Prediction of traveltime and longitudinal dispersion in rivers and streams.
U.S. Geological Survey Water-Resources Investigations Report 96-4013, 69 p. As cited in Schwarz et
al. 2006.
Johnes, P.J., B. Moss, and G.L. Phillips, G.L. 1996. The determination of water quality by land use,
livestock numbers and population data - testing of a model for use in conservation and water quality
management. Freshwater Biology 36:951-473. As cited in Schwarz et al. 2006.
Johnston, R.J., E.Y. Besedin and R.F. Wardwell. 2003. Modeling Relationships Between Use and Nonuse
Values for Surface Water Quality: A Meta-Analysis. Water Resources Research 39(12): 1363-1372.
November 2008 12-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Johnston, R.J., E.Y. Besedin, R. lovanna, C.J. Miller, R.F. Wardwell and M.H. Ranson. 2005.
"Systematic Variation in Willingness to Pay for Aquatic Resource Improvements and Implications for
Benefit Transfer: A Meta-Analysis." Canadian Journal of Agricultural Economics 53(2-3):221-248.
Johnston, R.J., J.J. Opaluch, T.A. Grigalunas, and M.J. Mazzotta. 2001. Estimating Amenity Benefits of
Coastal Farmland. Growth and Change 32(summer): 305-325.
Johnston, R.J., M.H. Ranson, E.Y. Besedin and E.G. Helm. 2006. "What Determines Willingness To Pay
Per Fish? A Meta-Analysis of Recreational Fishing Values." Marine Resource Economics 21:1-32.
Jowett, I.G. 1998. Hydraulic geometry of New Zealand rivers and its use as a preliminary method of
habitat assessment. Regulated Rivers - Research and Management 14:451-466. As cited in Schwarz
et al. 2006.
Julien, Pierre. 1995. Erosion and Sedimentation. Cambridge University Press.
Just, R.E., D.L. Hueth and A. Schmitz. 2004. The Welfare Economics of Public Policy: A Practical
Approach to Public Policy. Cheltenham, UK: Edward Elgar.
Kansas Department of Health and Environment. 2004 Surface water nutrient reduction plan. Available:
http://www.kdhe.state.ks.us/water/download/ks nutrient reduction_plan 12 29 final.pdf. Accessed
3/1/2005). As cited in Schwarz et al. 2006.
Kaoru, Y. 1993. Differentiating Use and Nonuse Values for Coastal Pond Water Quality Improvements.
Environmental and Resource Economics 3: 487-494.
Kelly, C.A., J.W.M. Rudd., RH. Hesslein, D.W. Schindler, P.J. Dillon, C.T. Driscoll, S.A. Gherini, and
R.E. Hecky. 1987. Prediction of biological acid neutralization in acid-sensitive lakes. Part 1: A
Theoretical and Practical Introduction to SPARROW 119. Biogeochemistry 3:129-140. As cited in
Schwarz et al. 2006.
Kent, Robert, and Kenneth Belitz (USGS). 2004. Concentrations of Dissolved Solids and Nutrients in
Water Sources and Selected Streams of the Santa Ana Basin, California, October 1998-September
2001. Water-Resources Investigations Report 03-4326.
Kerr, S. J. 1995. "Silt, Turbidity and Suspended Sediments in the Aquatic Environment: An Annotated
Bibliography and Literature Review." Ontario Ministry of Natural Resources, Southern Region
Science & Technology Transfer Unit Technical Report TR-008.
Kirk, K. L. 1992. "Effects of suspended clay on Daphnia body growth and fitness." Freshwater Biology,
Vol. 28: pp 103-109. A? cited in Kerr 1995.
Knisel, W.G. 1980. CREAMS, A Field-Scale Model for Chemicals, Runoff and Erosion from
Agricultural Management Systems. USDA Conservation Research Report No. 26. As cited in Neitsch
etal. 2005.
Kriesel, Warren; Randall, Alan; Lichtkoppler, Frank; 1993. "Estimating the Benefits of Shore Erosion
Protection in Ohio's Lake Erie Housing Market." Water Resources Research. 29(4). 795-801.
Krinsky, I. and A.L. Robb. 1986. Approximating the statistical properties of elasticities. Review of
Economics and Statistics 68(4):715-719.
Lant, C.L., and G.A. Tobin. 1989. The Economic Value of Riparian Corridors in Cornbelt Floodplains: A
Research Framework. Professional Geographer Vol (41), pp. 337-349.
November 2008 12-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Lant, C.L., and R. S. Roberts. 1990. Greenbelts in the Cornbelt: Riparian Wetlands, Intrinsic Values, and
Market Failure. Environment and Planning 22: 1375-1388.
LaPerriere, J. D., D. M. Bjerklie, R. C. Simmons, E. Van Nieuwenhuyse, S. M. Wagener and J. B.
Reynolds. 1983. "Effects of gold placer mining on the interior Alaska stream ecosystems. As cited in
Waters 1995.
Leonard, R.A., W.G. Knisel, and D.A. Still. 1987. GLEAMS: Groundwater loading effects on agricultural
management systems. Trans. ASAE 30(5): 1403-1428. As cited in Neitsch et al. 2005.
Levine, Jay F., Arthur E. Bogan, Kenneth H. Pollock, Hugh a. Devine, Lori L. Gustafson, Christopher B.
Eads, Periann P. Russell and Elizabeth F. Anderson. 2003. Distribution of Freshwater Mussel
Populations in Relationship to Crossing Structures. North Carolina Department of
Transportation/U.S. Department of Transportation. Report no. FIWY-2003-02.
Levine, Jay F., W. G. Cope, A. E. Bogan, M. Stoskopf, L. L. Gustafson, B. Showers, D. Shea, C. B. Eads,
P. Lazaro, W. Thorsen, D. Forestier and E. F. Anderson. 2005. Assessment of the Impact of Highway
Runoff on Freshwater Mussels in North Carolina Streams. North Carolina Department of
Transportation/U.S. Department of Transportation. Report no. FFIWA/NO/2004-03.
Lewis, K. 1973. "The effect of suspended coal particles on the life forms of the aquatic moss,
Eurhynchium ripariodus." Freshwater Biology, Vol. 3(73): pp 251-251. As citedin Kerr 1995.
Lichtkoppler, F.R. and T.W. Elaine. 1999. Environmental Awareness and Attitudes of Ashtabula County
Voters Concerning the Ashtabula River Area of Concern: 1996-1997. J. Great Lakes Resources. 25:
500-514.
Lindsey, G. 1994. Market Models, Protest Bids, and Outliers in Contingent Valuation. Journal of Water
Resources Planning and Management, Vol (12), pp. 121-129.
Line, D.E., and N.M. White. 2007. Effects of Development on Runoff and Pollutant Export. Water
Environment Research, Vol.79(2): 185-190.
Lipton, D. 2004. The Value of Improved Water Quality to Chesapeake Bay Boaters. Marine Resource
Economics. Vol(19):265-270
Lloyd, D. S. 1987. "Turbidity as a water quality standard for salmonid habitats in Alaska." North
American Journal of Fisheries Management Vol. 7: 18-33. As1 citedin Waters 1995.
Loomis, J., P. Kent, L. Strange, K. Fausch and A. Covich. 2000. Measuring the total economic value of
restoring ecosystem services in an impaired river basin: results from a contingent valuation survey.
Ecological Economics, Vol (33), pp. 103-117.
Loomis, J.B. 1996. How Large is the Extent of the Market for Public Goods: Evidence from a Nation
Wide Contingent Valuation Survey. Applied Economics 28(7): 779-782.
Lubliner, Brandi, and Steven Golding (Washington State Department of Ecology). 2005. Stormwater
Quality Survey of Western Washington Construction Sites, 2003-2005. Publication No. 05-03-028.
Lyke, A.J. 1993. Discrete Choice Models to Value Changes in Environmental Quality: A Great Lakes
Case Study. Dissertation submitted to the Graduate School of the University of Wisconsin, Madison.
Magat, W.A., J. Huber, K.W. Viscusi and J. Bell. 2000. An Iterative Choice Approach to Valuing Clean
Lakes, Rivers, and Streams. Journal of Risk and Uncertainty 21(1): 7-43.
November 2008 12-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Malins, D.C., McCain, B.B., Landahl, J.T., Myer, M.S., Krahn, M.M., Brown, D.W., Chan, S. and
Roubal, W.T. 1988. "Neoplastic and other diseases in fish in relation to toxic chemicals: an
overview." As cited in UK SACs Project (2008).
Marking, L. L., and T. D. Bills. 1980. "Acute effects of silt and sand sedimentation on freshwater
mussels."^ cited in Waters 1995.
Maryland Department of Environmental Protection (Maryland DEP), Montgomery County. 2000. County
Investigators Cite Contractor for Toxic Spill in Turkey Branch. Press Release, July 28, 2000.
Matthews, L.G., F.R. Romans, and K.W. Easter. 1999. Reducing Phosphorous Pollution in the Minnesota
River: How Much is it Worth? Staff Paper. Department of Applied Economics, University of
Minnesota.
McClelland, N. I. 1974. Water Quality Index Application in the Kansas River Basin. U.S. Environmental
Protection Agency Region 7, Kansas City, Missouri.
McEwan, Daniel and Terry Jackson. 1996. "Steelhead Restoration and Management Plan for California."
California Department of Fish and Game. February, 1996.
McMahon G., R.B. Alexander, and S. Qian S. 2003. Support of TMDL programs using spatially
referenced regression models. ASCE Journal of Water Resources Planning and Management
129:315-329. As cited in Schwarz et al. 2006.
McMahon, G. and C. Roessler. 2002. A regression-based approach to understand baseline total nitrogen
loading for TMDL planning. In Proceedings of Water Environment Federation National TMDL
Science and Policy Conference, Phoenix, Arizona, November 13-16, 2002. Phoenix, AZ, Water
Environment Federation, 28 p. As cited in Schwarz et al. 2006.
Menendez, Daniel, et al. 2007. "AguaClara: Ojojona Sub-team, Final Report." Department of
Environmental and Civil Engineering, Cornell University. Available at
http://eswserver.cee.cornell.edu/aguaclara/documents/reports.htm.
Meyer, B. C., and A. Heritage. 1941. "Effect of turbidity and depth of immersion on apparent
photosynthesis in Ceratophyllum demersum" Ecology Vol. 22: pp 17-22. As cited in Kerr 1995.
Michigan Department of Environmental Quality (Michigan DEQ). 2006. Water Quality and Pollution
Control in Michigan-2006 Sections 303(d), 305(b), and 314 Integrated Report. MI/DEQ/EB-06/019.
Missouri Department of Natural Resources (Missouri DNR). 2004. Stream Survey Sampling Report:
Hinkson Creek Stream Study, Columbia Missouri, Boone County.
Missouri Department of Natural Resources (Missouri DNR). 2005. Stream Survey Sampling Report:
Phase II, Hinkson Creek Stream Study, Columbia Missouri, Boone County.
Mitchell, R.C. and R.T. Carson. 1989. Using Surveys to Value Public Goods: The Contingent Valuation
Method. Resources for the Future, Washington, DC.
Moeltner, K., K.J. Boyle, and R.W. Paterson. 2007. "Meta-Analysis and Benefit Transfer for Resource
Valuation—Addressing Classical Challenges with Bayesian Modeling." Journal of Environmental
Economics and Management 53(2):250-269.
Molot, L.A. and P.J. Dillon. 1993. Nitrogen mass balances and denitrification rates in central Ontario
Lakes. Biogeochemistry 20:195-212. As cited in Schwarz et al. 2006.
November 2008 12-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Moore, R.B., C.M. Johnston, K.W. Robinson, and J.R. Deacon. 2004. Estimation of total nitrogen and
phosphorus in New England streams using spatially referenced regression models. U.S. Geological
Survey Scientific Investigations Report 2004-5012, 50 p. Available:
http://water.usgs.gov/pubs/sir/2004/5012/. As cited in Schwarz et al. 2006.
Moore, Walter B., and Bruce A. McCarl. 1987. "Off-site Costs of Soil Erosion: A Case Study in the
Willamette Valley." Western Journal of Agricultural Economics Vol. 12(1): pp 42-49.
Morris, Gregory L., and Jiahua Fan. 1997. Reservoir Sedimentation Handbook: Design and Management
of Dams, Reservoirs, and Watersheds for Sustainable Use. New York: McGraw-Hill.
Multi-Resolution Land Characteristics Consortium (MRLC). 2001. National Land Cover Database 2001
(NLCD 2001). Available at http://www.mrlc.gov/mrlc2k nlcd.asp. As cited in Schwarz 2008a.
Natural Resources Defense Council (NRDC) and State of New York v. U.S. Environmental Protection
Agency (EPA) and Associated General Contractors of America. 2005a. "Declaration of Anna E.
Slawsky." U.S. District Court, Central District of California. Case No. CV 04-8307 GHK (RCx).
Natural Resources Defense Council (NRDC) and State of New York v. U.S. Environmental Protection
Agency (EPA) and Associated General Contractors of America. 2005b. "Declaration of Betty
Rushton." U.S. District Court, Central District of California. Case No. CV 04-8307 GHK (RCx).
Natural Resources Defense Council (NRDC) and State of New York v. U.S. Environmental Protection
Agency (EPA) and Associated General Contractors of America. 2005c. "Declaration of Don
McEnhill." U.S. District Court, Central District of California. Case No. CV 04-8307 GHK (RCx).
Natural Resources Defense Council (NRDC) and State of New York v. U.S. Environmental Protection
Agency (EPA) and Associated General Contractors of America. 2005d. "Declaration of Kathleen
Krasenics." U.S. District Court, Central District of California. Case No. CV 04-8307 GHK (RCx).
Natural Resources Defense Council (NRDC) and State of New York v. U.S. Environmental Protection
Agency (EPA) and Associated General Contractors of America. 2005e. "Declaration of Michael
Mullen." U.S. District Court, Central District of California. Case No. CV 04-8307 GHK (RCx).
National Oceanic and Atmospheric Administration (NOAA). 2004. Sediment Removal from Freshwater
Salmonid Habitat: Guidelines to NOAA Fisheries Staff for the Evaluation of Sediment Removal
Actions from California Streams. Available at: http://swr.nmfs.noaa.gov/hcd/policies/Aprill9-
2004.pdf Accessed April 1, 2008.
National Oceanic and Atmospheric Administration (NOAA). 2008. National Estuarine Research Reserve
System (NERRS). Accessed 07/11/08. Available at http://cdmo.baruch.sc.edu/.
National Oceanic and Atmospheric Administration (NOAA) and U.S. Environmental Protection Agency
(EPA). 1989a. National Oceanic and Atmospheric Administration and U.S. Environmental Protection
Agency. Strategic assessment of near coastal waters. Susceptibility of east coast estuaries to nutrient
discharges: Albemarle/Pamlico Sound to Biscayne Bay. Rockville, MD: Strategic Assessment
Branch, NOAA. As cited in USEPA 1997.
National Oceanic and Atmospheric Administration (NOAA) and U.S. Environmental Protection Agency
(EPA). 1989b. National Oceanic and Atmospheric Administration and U.S. Environmental Protection
Agency. Strategic assessment of near coastal waters. Susceptibility of east coast estuaries to nutrient
discharges: Passamaquoddy Bay to Chesapeake Bay. Rockville, MD: Strategic Assessment Branch,
NOAA. As cited in USEPA 1997.
November 2008 12-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
National Oceanic and Atmospheric Administration (NOAA) and U.S. Environmental Protection Agency
(EPA). 1989c. National Oceanic and Atmospheric Administration and U.S. Environmental Protection
Agency. Strategic assessment of near coastal waters. Susceptibility and status of Gulf of Mexico
estuaries to nutrient discharge. Rockville, MD: Strategic Assessment Branch, NOAA. As cited in
USEPA 1997.
National Oceanic and Atmospheric Administration (NOAA) and U.S. Environmental Protection Agency
(EPA). 1991. National Oceanic and Atmospheric Administration and U.S. Environmental Protection
Agency. Strategic assessment of near coastal waters. Susceptibility and status of west coast estuaries
to nutrient discharges: San Diego Bay to Puget Sound. Rockville, MD: Strategic Assessment Branch,
NOAA. As cited in USEPA 1997.
National Research Council (NRC). 2000. Clean coastal waters - understanding and reducing the effects of
nutrient pollution. As cited in Schwarz et al. 2006.
National Research Council (NRC). 200la. Assessing the TMDL approach to water quality management.
As cited in Schwarz et al. 2006.
National Research Council (NRC). 200 Ib. Ecological indicators for the nation. As cited in Schwarz et al.
2006.
National Research Council (NRC). 2002a. Review of the USGS National Water-Quality Assessment
Program. As cited in Schwarz et al. 2006.
National Research Council (NRC). 2002b. Estimating water use in the United States - a new paradigm for
the National Water-Use Information Program. As cited in Schwarz et al. 2006.
National Research Council (NRC). 2008. Urban Stormwater Management in the United States.
Washington, D.C.: National Academies Press. 553 pp.
Natural Resources Conservation Service, United States Department of Agriculture. 2005. National
Resources Inventory—a statistical survey of land use and natural resource conditions and trends on
U.S. non-Federal lands. As cited in Schwarz et al. 2006.
NatureServe. 2008. NatureServe Explorer database. Available at: http://www.natureserve.org/explorer/.
Accessed April 1,2008.
Navrud, S. and R. Ready. Review of Methods for Value Transfer. In S. Navrud and R. Ready (eds.),
Environmental Value Transfer: Issues and Methods. Dordrecht, The Netherlands: Springer, pp. 1-10.
Neitsch, S.L., J.G. Arnold, J.R. Kiniry, and J.R. Williams. 2005. Soil and Water Assessment Tool
Theoretical Documentation, Version 2005. Agricultural Research Service, Grassland, Soil and Water
Research Laboratory and Texas Agricultural Experiment Station, Blackland Research Center,
Temple, TX.
Neitsch, S.L., J.G. Arnold, J.R. Kiniry, J.R. Williams, and K.W. King. 2002. Soil and Water Assessment
Tool Theoretical Documentation, Version 2000. Agricultural Research Service, Grassland, Soil and
Water Research Laboratory and Texas Agricultural Experiment Station, Blackland Research Center,
Temple, TX.
Nelson, Erin J., and Derek B. Booth. 2002. "Sediment Sources in an urbanizing, mixed land-use
watershed." Journal of Hydrology Vol. 264: pp. 51-68.
November 2008 12-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Nevada Division of Environmental Protection. 2003. Bryant Creek: Total Maximum Daily Loads -
Arsenic (total), Iron (total), Nickel (total), Turbidity, and Total Suspended Solids. Bureau of Water
Quality Planning. Available at: http://ndep.nv.gov/BWQP/file/bryantl21503.pdf.
New England Interstate Water Pollution Control Commission (NEIWPCC). 2004. New England
SPARROW water quality model. Interstate Water Report l(3):6-7. As cited in Schwarz et al. 2006.
Newcombe, C. P. 1994. Suspended Sediment in Aquatic Ecosystems: 111 Effects as a Function of
Concentration and Exposure. Victoria, BC: Habitat Protection Branch, Ministry of Environment,
Lands and Parks. As cited in Henley et al. 2000.
Newcombe, C. P., and J. O. Jensen. 1996. "Channel suspended sediment and fisheries: a synthesis for
quantitative assessment of risk and impact." North. American Journal of Fisheries Management Vol.
16: pp. 693-727. As cited in Henley et al. 2000.
Ngo, Due and Jessica Nichols. 1998. "Design of Sedimentation Basins." (Webpage). Water Treatment
Primer: CE 4124: Environmental Information Management. Civil Engineering Department, Virginia
Tech. Accessed September 29, 2008. Available online at
http://www.cee.vt.edu/ewr/environmental/teach/wtprimer/sedimen/sedimen.html.
Nolan, J.V., J.W. Brakebill, R.B. Alexander, and G.E. Schwarz. 2002. ERF1_2 - Enhanced River Reach
File 2.0: U.S. Geological Survey Open-File Report 02-40. Available:
http://water.usgs.gov/GIS/metadata/usgswrd/XML/ erfl_2 .xml.
Novotny, Vladimir, and Gordon Chesters. 1989. "Delivery of sediment and pollutants from nonpoint
sources: A water quality perspective." Journal of Soil and Water Conservation Vol. 44(6): pp. 568-
576.
Office of Management and Budget (OMB). 2003. OMB Circular A4. p. 33-34.
Ohio Environmental Protection Agency (Ohio EPA). 1994. Biological and Water Quality Study of the
Cuyahoga River. Ohio EPA Technical Report EAS/1992-12-11.
Ohio Environmental Protection Agency (Ohio EPA). 1995. Biological and Water Quality Study of the
Middle and Lower Great Miami River and Selected Tributaries, 1995. Ohio EPA Technical Report
MAS/1996-12-8.
Ohio Environmental Protection Agency (Ohio EPA). 1996a. Biological and Water Quality Study of Mill
Creek - 1995. Ohio EPA Technical Report MAS/1996-7-4.
Ohio Environmental Protection Agency (Ohio EPA). 1996b. Biological and Water Quality Study of the
Mahoning River Basin. Ohio EPA Technical Report MAS/1995-12-14.
Ohio Environmental Protection Agency (Ohio EPA). 1996c. Biological and Water Quality Study of the
Upper Great Miami River and Selected Tributaries. Ohio EPA Technical Report MAS/1995-12-13.
Ohio Environmental Protection Agency (Ohio EPA). 1997a. 1995-96 Biological and Water Quality Study
of the Chagrin River and Selected Tributaries. Ohio EPA Technical Report MAS/1996-12-6.
Ohio Environmental Protection Agency (Ohio EPA). 1997b. Biological and Water Quality Study of Big
Walnut Creek. Ohio EPA Technical Report MAS/1997-10-2.
Ohio Environmental Protection Agency (Ohio EPA). 1997c. Biological and Water Quality Study of Mill
Creek (Scioto River Basin) and Selected Tributaries. Ohio EPA Technical Report MAS/1996-12-11.
November 2008 12-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Ohio Environmental Protection Agency (Ohio EPA). 1997d. Biological and Water Quality Study of
Raccoon Creek. Ohio EPA Technical Report MAS/1996-12-14.
Ohio Environmental Protection Agency (Ohio EPA). 1997e. Biological and Water Quality Study of the
Upper Hocking River and Selected Tributaries. Ohio EPA Technical Report MAS/1996-12-10.
Ohio Environmental Protection Agency (Ohio EPA). 1997 f. Biological and Water Quality Study of Twin
Creek and Selected Tributaries 1995. Ohio EPA Technical Report MAS/1996-12-9.
Ohio Environmental Protection Agency (Ohio EPA). 1998a. Biological and Water Quality Study of
Blacklick Creek and Selected Tributaries. Ohio EPA Technical Report MAS/1997-12-7.
Ohio Environmental Protection Agency (Ohio EPA). 1998b. Biological and Water Quality Study of Little
Beaver Creek and Big Beaver Creek - 1997. Ohio EPA Technical Report MAS/1998-5-1.
Ohio Environmental Protection Agency (Ohio EPA). 1998c. Biological and Water Quality Study of
Lower Big Walnut Creek and Walnut Creek Tributaries 1996. Ohio EPA Technical Report
MAS/1997-12-10.
Ohio Environmental Protection Agency (Ohio EPA). 1998d. Biological and Water Quality Study of the
Little Cuyahoga River and Tributaries. Ohio EPA Technical Report MAS/1997-12-9.
Ohio Environmental Protection Agency (Ohio EPA). 1998e. Biological and Water Quality Study of the
Little Miami River Basin, 1998. Ohio EPA Technical Report MAS/1999-12-3.
Ohio Environmental Protection Agency (Ohio EPA). 1999a. Biological and Water Quality Study of the
Cuyahoga River and Selected Tributaries. Ohio EPA Technical Report MAS/1997-12-4.
Ohio Environmental Protection Agency (Ohio EPA). 1999b. Biological and Water Quality Study of the
Middle Scioto River and Alum Creek. Ohio EPA Technical Report MAS/1997-12-12.
Ohio Environmental Protection Agency (Ohio EPA). 1999c. Biological and Water Quality Study of the
Rocky River and Selected Tributaries. Ohio EPA Technical Report MAS/1998-12-3.
Ohio Environmental Protection Agency (Ohio EPA). 1999d. Biological and Water Quality Study of
Walnut Creek and Selected Tributaries 1996. Ohio EPA Technical Report MAS/1997-12-5.
Ohio Environmental Protection Agency (Ohio EPA). 2000. Biological and Water Quality Study of
Marion Area Streams 1998. Ohio EPA Technical Report MAS/1999-12-6.
Ohio Environmental Protection Agency (Ohio EPA). 200 la. Biological and Water Quality Study of the
Olentangy River and Selected Tributaries 1999. Ohio EPA Technical Report MAS/2000-12-6.
Ohio Environmental Protection Agency (Ohio EPA). 200 Ib. Biological and Water Quality Study of the
Stillwater River Watershed. Ohio EPA Technical Report MAS/2001-12-8.
Ohio Environmental Protection Agency (Ohio EPA). 2003a. Biological and Water Quality Study of the
Big Walnut Creek Basin, 2000. Ohio EPA Technical Report EAS/2003-11-10.
Ohio Environmental Protection Agency (Ohio EPA). 2003b. Biological and Water Quality Study of the
Sandusky River and Selected Tributaries 2001. Ohio EPA Technical Report EAS/2003-4-6.
Ohio Environmental Protection Agency (Ohio EPA). 2004. Biological and Water Quality Study of the Big
Darby Creek Watershed, 2001/2002. Ohio EPA Technical Report EAS/2004-6-3.
November 2008 12-15
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Ohio Environmental Protection Agency (Ohio EPA). 2005. Biological and Water Quality Study of the
Olentangy River, Whetstone Creek and Select Tributaries, 2003-2004. Ohio EPA Technical Report
EAS/2005-12-6.
Ohio Environmental Protection Agency (Ohio EPA). 2006a. Biological and Water Quality Study of the
Chagrin River and Selected Tributaries 2003-04. Ohio EPA Technical Report EAS/2006-12-7.
Ohio Environmental Protection Agency (Ohio EPA). 2006b. Biological and Water Quality Study of the
Grand River Basin 2003-2004. Ohio EPA Technical Report EAS/2006-11-6.
Ohio Environmental Protection Agency (Ohio EPA). 2006c. Recreational Use Water Quality Survey of
the Sugar Creek Watershed, 2005. Ohio EPA Technical Report NEDO/2006-02-01.
Ohio Environmental Protection Agency (Ohio EPA). 2006d. Biological and Water Quality Study of
Walnut Creek and Tributaries. Ohio EPA Technical Report EAS/2006-12-8.
Ohio Environmental Protection Agency (Ohio EPA). 2007a. Biological and Water Quality Study of
Furnace Run Watershed. Ohio EPA Technical Report NEDO/2007-02-01.
Ohio Environmental Protection Agency (Ohio EPA). 2007b. Biological and Water Quality Study of the
BlanchardRiver. Ohio EPA Technical Report EAS/2007-6-2.
Ohio Environmental Protection Agency (Ohio EPA). 2007c. Biological and Water Quality Study of Twin
Creek and Select Tributaries 2005. Ohio EPA Technical Report EAS/2007-10-03.
Ohio Environmental Protection Agency. 2000. Total Maximum Daily Loads for the Middle Cuyahoga
River - Final Report. Division of Surface Water. Available at:
http://www.epa.state.oh.us/dsw/tmdl/MidCuyFinalTMDL.pdf
Ohio Environmental Protection Agency. 2005. Total Maximum Daily Loads for the Little Beaver Creek
Watershed - Final Report Division of Surface Water. Available at:
http://www.epa.state.oh.us/dsw/tmdl/Little%20Beaver_fmal.pdf
Olsen, D., Richards, Jack, and Scott, R. D. 1991. Existence and Sport Values for Doubling the Size of
Columbia River Basin Salmon and Steelhead Runs. Rivers. Vol. 2, No. 1. pp. 44-56.
Opaluch, J.J., T. Grigalunas, M.J. Mazzotta, J. Diamantides, and R. Johnston. 1998. Resource and
Recreational Economic Values for the Peconic Estuary. Report prepared for Peconic Estuary
Program, Suffolk County Department of Health Services, Riverhead, NY, by Economic Analysis,
Inc., Peace Dale, Rhode Island.
Oregon State University (OSU). 2007. PRISM Group. Available: http://www.prism.oregonstate.edu/.
Osei, E., P.W. Gassman, R. Jones, S. Pratt, L. Hauck, L. Beran, W. Rosenthal, and J.R. Williams. 2000a.
Economic and Environmental Impacts of Alternative Practices on Dairy Farms in an Agricultural
Watershed. J. Soil & Water Conservation 55(4):466-472. As cited in Gassman et al. 2005.
Osei. E., P. Gassman, and A. Saleh. 2000b. Livestock and the Environment: A National Pilot Project:
Economic and Environmental Modeling Using CEEOT. Project Report PR0002. Texas Institute for
Applied Environmental Research, Temple, Tarleton State University. As cited in Gassman et al. 2005.
Owens, David W., Peter Jopke, David W. Hall, Jeremy Balousek and Aicardo Roa. (USGS). 2000. Soil
Erosion from Two Small Construction Sites, Dane County, Wisconsin. USGS Fact Sheet FS-109-00.
November 2008 12-16
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Paffenhofer, G. A. 1972. "The effects of suspended 'red mud' on mortality, body weight and growth of
the marine planktonic copepod Calanus helgolandius. Water, Air and Soil Pollution, Vol. 1: pp 314-
321. As cited in Ken 1995.
Pain, S. 1987. "After the gold rush." As cited in Waters 1995.
Peierls, B.L., N.F. Caraco, M.L. Pace, and J.J. Cole. 1991. Human influence on river nitrogen. Nature
350:386-387. As cited in Schwarz et al. 2006.
Poe, G.L., K.J. Boyle, and J.C. Bergstrom. 2001. A Preliminary Meta Analysis of Contingent Values for
Ground Water Quality Revisited, in J.C. Bergstrom, K.J. Boyle and G.L. Poe, eds. The Economic
Value of Water Quality. Northampton, MA: Edward Elgar Publishers.
Potter, K.N., J.R. Williams, F.J. Larney, and M.S. Bullock. 1998. Evaluation of EPIC's Wind Erosion
Submodel using Data from Southern Alberta. Canadian Journal of Soil Science 78:485-492. As cited
in Gassman et al. 2005.
Preston, S.D. and J.W. Brakebill. 1999. Application of spatially referenced regression modeling for the
evaluation of total nitrogen loading in the Chesapeake Bay watershed: Baltimore, MD, U.S.
Geological Survey Water Resources Investigations Report 99-4054, lip. Available:
http://md.water.usgs.gov/publications/wrir-99-4054/. As cited in Schwarz et al. 2006.
Reckhow, K.H. and S.C. Chapra. 1983. Engineering approaches for lakes management, vol. 1 - data
analysis and empirical modeling. Butterworth publishers, Boston, MA. 340 p. As cited in Schwarz et
al. 2006.
Reed, Lloyd A. (USGS). 1980. Suspended-Sediment Discharge, in Five Streams near Harrisburg,
Pennsylvania, Before, During, and After Highway Construction. Geological Survey Water-Supply
Paper 2072.
Reid, Scott M., Glenn Isaac, Serge Metikosh and Jim Evans. 2003. "Physiological Response of Rainbow
Trout to Sediment Released during Open-Cut Pipeline Water Crossing Construction." Water Quality
Resources Journal of Canada Vol. 38(3): pp 473-481.
Renard, K.G. 1997. Predicting Soil Erosion by Water: A Guide to Conservation Planning with the
Revised Universal Soil Loss Equation (RUSLE). Washington, D.C.: U.S. Department of Agriculture,
Agricultural Research Service. As cited in Gassman et al. 2005.
Reynolds, J. B., R. C. Simmons, and A. R. Burkholder. 1989. "Effects of placer mining discharge on
health and food of Arctic grayling." Water Resources Bulletin Vol. 25: pp. 625-635. As cited in
Henley et al. 2000.
Ribaudo, Marc O. 1989. "Water Quality Benefits from the Conservation Reserve Program." Agricultural
Economic Report No. 606. Resources and Technology Division, Economic Research Service,
U.S.D.A., Washington, D.C.
Ribaudo, Marc O. and C. Edwin Young. 1989. "Estimating the Water Quality Benefits from Soil Erosion
Control." Water Resources Bulletin 25(1): 71-78.
Ribaudo, Marc. O., D. Colacicco, A. Barbarika, and C.E. Young. 1989. "The Economic Efficiency of
Voluntary Soil Conservation Programs." Journal of Soil and Water Conservation 1989(Jan.-Feb).
Rivier, B., and J. Seguier. 1985. "Physical and biological effects of gravel extraction in river beds." As
cited in Waters 1995.
November 2008 12-17
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Roberts, L.A., and J.A. Leitch. 1997. Economic Valuation of Some Wetland Outputs of Mud Lake.
Agricultural Economics Report No. 381, Department of Agricultural Economics, North Dakota
Agricultural Experiment Station, North Dakota State University.
Rosenberger, R.S. and J.B. Loomis. 2000a. Using Meta-Analysis for Benefit Transfer: In-Sample
Convergent Validity Tests of an Outdoor Recreation Database. Water Resources Research 36(4):
1097-1107.
Rosenberger, R.S. and J.B. Loomis. 2000b. Panel Stratification in Meta-Analysis of Economic Studies:
An Investigation of Its Effects in the Recreation Valuation Literature. Journal of Agricultural and
Applied Economics 32(3): 459-470.
Rosenberger, R.S. and R.J. Johnston. 2007. Selection Effects in Meta-Valuation Function Transfers.
Meta-Analysis in Economics Conference, September 27-30, 2007, Sandbjerg Manor, S0nderborg,
Denmark.
Rosenberger, R.S. and T.D. Stanley. 2006. "Measurement, Generalization, and Publication: Sources of
Error in Benefit Transfers and Their Management." Ecological Economics 60(2):372-378.
Rosenberger, R.S. and T.T. Phipps. 2007. "Correspondence and Convergence in Benefit Transfer
Accuracy: Meta-Analytic Review of the Literature." In S. Navrud and R. Ready (eds.),
Environmental Value Transfer: Issues and Methods. Dordrecht, The Netherlands: Springer, pp.23-43.
Rowe, D., E. Graynoth, G. James, M. Taylor and L. Hawke. 2003. "Influence of turbidity and fluctuating
water levels on the abundance and depth distribution of small, benthic fish in New Zealand alpine
lakes." Ecology of Freshwater Fish Vol. 12: pp. 216-227.
Rowe, R.D., W.D. Schulze, B. Hurd and D. Orr. 1985. Economic Assessment of Damage Related to the
Eagle Mine Facility. Energy and Resource Consultants, Inc., Boulder, CO.
Ruddy, B.C. and K. J. Hitt. 1990. Summary of selected characteristics of large reservoirs in the United
States and Puerto Rico, 1988. U.S. Geological Survey, Open File Report 90-163.
Runkle, R.L., C.G. Crawford, and T.A. Cohn. 2004. Load estimator (LOADEST) - A FORTRAN
program for estimating constituent loads in streams and rivers: U.S. Geological Survey Techniques
and Methods 4-A5, 75 p.. As cited in Schwarz et al. 2006.
Ryan, Paddy A. 1991. "Environmental effects of sediment on New Zealand streams: a review. "New
Zealand Journal of Marine and Freshwater Research Vol. 25: pp. 207-221.
Samsel Jr., G. L.. 1973. "Effects of sedimentation on the algal flora of a small recreational
impoundment." Water Resources Bulletin, Vol. 9: pp 1145-1152. As cited in Kerr 1995.
Sanders, L. B., R.G. Walsh, and J.B. Loomis. 1990. Toward Empirical Estimation of the Total Value of
Protecting Rivers. Water Resources Research. Vol. 26, No. 7. pp. 1345-1357.
Santhi, C., N. Kannan, M. Di Luzio, S. R. Potter, Jeffrey G. Arnold, Jay D. Atwood and R. L. Kellog.
2005. An Approach for Estimating Water Quality Benefits of Conservation Practices at the National
Level. ASABE Paper No. 052043, 2005 ASAE Annual Meeting.
Santos, J.M.L. 1998. The Economic Valuation of Landscape Change. Cheltenham, UK: Edward Elgar.
SAS. 1993. SAS/ETS user's guide, version 6 (2nded.). Gary, NC, SAS Institute Inc., 1022 p. As citedin
Schwarz et al. 2006.
November 2008 12-18
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
SAS. 1999. SAS OnlineDoc, version 8. Available: http://v8doc.sas.com/sashtml/. Accessed 4/29/2005.
SAS Inc. As cited in Schwarz et al. 2006.
Saunders, J. W., and M. W. Smith. 1965. "Changes in a stream population of trout associated with
increase silt." Journal of Fisheries Research Board Canada, Vol. 22: pp 395-404. As cited in Kerr
1995.
Schulze, W.D., R.D. Rowe, W.S. Breffle, R.R. Boyce, and G.H. McClelland. 1995. Contingent Valuation
of Natural Resource Damages Due to Injuries to the Upper Clark Fork River Basin. State of Montana,
Natural Resource Damage Litigation Program. Prepared by: RCG/Hagler Bailly, Boulder, CO.
Schwarz, G. E (USGS). 2008a. A Preliminary SPARROW model of suspended sediment for the
conterminous United States: U.S. Geological Survey Open-File Report 2008-1205, 8 p., available
only online at http://pubs.usgs.gov/of/2008/1205
Schwarz, G. E. (USGS). 2008b. Personal Communication. June 4, 2008.
Schwarz, G. E., R. A. Smith, R. B. Alexander, and J. R. Gray. 2001. "A Spatially Referenced Regression
Model (SPARROW) for Suspended Sediment in Streams of the Coterminous U.S." U.S.
Subcommittee on Sedimentation, Proceedings of the Seventh Federal Interagency Sedimentation
Conference, March 25-29 2001, Reno, Nevada USA. p. VII-80-7. As cited in Schwarz et al. 2006.
Schwarz, G.E., A.B. Hoos, R.B. Alexander, and R.A Smith. 2006. The Sparrow Surface Water Quality
Model: Theory, Application and User Documentation. U.S. Geological Survey Techniques and
Methods Book 6, Section B, Chapter 3.
Seitzinger, S., RV. Styles, E.W. Boyer, R.B. Alexander, G. Billen, RW. Howarth, B. Mayer, andN.
vanBreemen. 2002. Nitrogen retention in rivers - model development and application to watersheds
in the Eastern U.S. Biogeochemistry 57:199-237. As cited in Schwarz et al. 2006.
Selbig, William R, Peter L. Jopke, David W. Marshall and Michael J. Sorge. (USGS). 2004. Hydrologic,
Ecologic, and Geomorphic Responses of Brewery Creek to Construction of a Residential Subdivision,
Dane County, Wisconsin, 1999-2002. Scientific Investigations Report 2004-5156.
Servizi, J. A., and D. W. Martens. 1987. "Some effects of suspended Fraser River sediment on sockeye
salmon, Oncorhynchus nerka. " p. 254-264. As cited in Reid et al. 2003.
Shen, Hseih Wen, and Pierre Y. Julien. 1993. Erosion and Sediment Transport. Chapter 12 in David
Maidment, Ed., Handbook of Hydrology. McGraw-Hill, Inc.
Shrestha, R. K., J.R.R. Alavalapati. 2004. Valuing Environmental Benefits of Silvopasture Practice: A
Case Study of the Lake Okeechobee Watershed in Florida. Ecological Economics, 49: 349-359.
Shrestha, R., R. Rosenberger, and J. Loomis. 2007. Benefit Transfer Using Meta-Analysis in Recreation
Economic Valuation, in Benefit Transfer Accuracy: Meta-Analytic Review of the Literature." In S.
Navrud and R. Ready (eds.), Environmental Value Transfer: Issues and Methods. Dordrecht, The
Netherlands: Springer, pp. 161-177.
Simons, D. B., R. M. Li and W. Fullerton. Theoretically derived sediment transport equations for Pima
County, Arizona. Prepared for Pima County DOT and Flood Control District, Tucson, Arizona. Ft.
Collins, Colorado: Simons, Li and Assoc., 1981. As citedin Julien 1995.
Singer, J.D. 1998. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and
individual growth models. Journal of Educational and Behavioral Statistics 23(4):323-355.
November 2008 12-19
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Singh, V.P. 1995. Computer models of watershed hydrology: Highlands Ranch, CO, Water Resources
Publications, 1,144 p. As cited in Schwarz et al. 2006.
Smiley, Jeff (USGS). 2007. "Statistics on the Nationwide High Resolution National Hydrography
Dataset" USGS National Hydrography Dataset Newsletter'Vol. 6, No. 11, September 2007.
Smith, R.A., G.E. Schwarz, and R.B. Alexander. 1997. Regional interpretation of water-quality
monitoring data. Water Resources Research 33:2781-2798. As cited in Schwarz et al. 2006.
Smith, R.A., RB. Alexander, G.D. Tasker, C.V. Price, C.V., K.W. Robinson, and D.A. White. 1994.
Statistical modeling of water quality in regional watersheds, in U.S. Environmental Protection
Agency, Proceedings of watershed '93 - a national conference on watershed management,
Alexandria, Virginia, March 21-24, 1993: Washington, DC, U.S. Environmental Protection Agency,
EPA 840-R-94-002, p. 751-754. As cited in Schwarz et al. 2006.
Smith, V.K., and L. Osborne. 1996. Do Contingent Valuation Estimates Pass the Scope Test? A Meta
Analysis. Journal of Environmental Economics and Management 31(3): 287-301.
Smith, V.K., and S.K. Pattanayak. 2002. Is Meta-Analysis a Noah's Ark for Non-Market Valuation?
Environmental and Resource Economics 22: 271-296.
Smith, V.K., and Y. Kaoru. 1990. "Signals or Noise? Explaining the Variation in Recreation Benefit
Estimates." American Journal of Agricultural Economics 42(2):419-433.
Smith, V.K., G. Van Houtven, and S.K. Pattanayak. 2002. Benefit Transfer via Preference Calibration:
"Prudential Algebra" for Policy. Land Economics 78(1): 132-152.
Sohngen, Brent, and Jon Rausch. 1998. "Soil Erosion in the Maumee River Basin: A Case Study Using
Market Methods to Value Environmental Externalities." Draft. Department of Agricultural
Economics, Ohio State University. June 5.
Soil Conservation Service (SCS). 1972. Section 4: Hydrology. In National Engineering Handbook. Soil
Conservation Service. As cited in Neitsch et al. 2005.
Spash, C.L. and A. Vatn. 2006. "Transferring Environmental Value Estimates: Issues and Alternatives."
Ecological Economics 60(2):379-388.
Srinivasan, R., J.G. Arnold, R.S. Muttiah, D. Walker, and P.T. Dyke. 1993. Hydrologic unit modeling of
the United States (HUMUS), in Yang, S. (ed.), Advances in hydro-science and engineering, vol I, part
A: University, MS, Center for Computational Hydroscience and Engineering, School of Engineering,
University of Mississippi, p. 451-456. As cited in Schwarz et al. 2006.
STATSGO. 1995. U.S. General Soil Map (STATSGO). Available:
http://www.ncgc.nrcs.usda.gov/products/datasets/statsgo/.
Stephens, Doyle W., Kendall R. Thompson, and James B. Wangsgard (USGS). 1996. Potential Effects of
Coal Mining and Road Construction on the Water Quality of Scofield Reservoir and its Drainage
Area, Central Utah, October 1982 to October 1984. Water Resources Investigations Report 96-4020.
Stream Solute Workshop. 1990. Concepts and methods for assessing solute dynamics in stream
ecosystems: Journal of the North American Benthological Society, v. 9, p. 95-119. As cited in
Schwarz et al. 2006.
Streiner, Carol F.; Loomis, John B. 1996. "Estimating the Benefits of Urban Stream Restoration Using
the Hedonic Price Method." Rivers. 5(4). 267-278.
November 2008 12-20
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Stumborg, B.E., K.A. Baerenklau, and R.C. Bishop. 2001. Nonpoint Source Pollution and Present Values:
A Contingent Valuation of Lake Mendota. Review of Agricultural Economics. 23(1): 120-132.
Sutherland, R. J., and R.G. Walsh. 1985. Effect of Distance on the Preservation Value of Water Quality.
Land Economics. Vol. 61, No. 3. pp. 282-291.
Trimble, S.W. and P. Crosson. 2000. U.S. soil erosion rates - myth and reality: Science, v. 289, p. 248-
250. As cited in Schwarz et al. 2006.
U.S. Army Corps of Engineers (USAGE). 2002. "Dredged Material Management Plan and Environmental
Impact Statement: McNary Reservoir and Lower Snake River Reservoirs." July 2002.
U.S. Army Corps of Engineers (USAGE). 2008. "Dredging Information System." Navigational Data
Center. Accessed 7 February 2008. Available at:
http://www.iwr.usace.army.mil/ndc/data/datadrgsel.htm
U.S. Army Corps of Engineers (USAGE). 1996. USAGE National Inventory of Dams. Available at
http://crunch.tec.army.mil/nidpublic/webpages/nid.cfm. As cited in Schwarz 2008a.
U.S. Census Bureau. 2006a. The 2006 American Community Survey.
http://www.census.gov/acs/www/index.html (Accessed April 1, 2008).
U.S. Census Bureau. 2006b. US Population Estimates by State., Population Estimates.
http://www.census.gov/popest/states/tables/NST-EST2007-01 .xls (Accessed April 1, 2008)
U.S. Department of Agriculture, 1994, State Soil Survey Geographic Data Base: U.S. Department of
Agriculture, Soil Conservation Service, National Cooperative Soil Survey, Miscellaneous Publication
Number 1492. As cited in Schwarz 2008a.
U.S. Department of Labor. 2008. Consumer Price Index (CPI): All Urban Consumers. Bureau of Labor
Statistics. Available at ftp://ftp.bis.gov/pub/special.requests/cpi/cpiai.txt. Accessed April 1, 2008.
U.S. Environmental Protection Agency (EPA) Science Advisory Board. 2007. Advisory on EPA's Issues
in Valuing Mortality Risk Reduction (EPA-SAB-08-001). U.S. EPA, Office of the Administrator,
Washington, DC, October 2007.
U.S. Environmental Protection Agency (EPA), 1996a. Technology Transfer Handbook: Management of
Water Treatment Plant Residuals (EPA 625-R-95-008), Office or Research and Development, April
1996.
U.S. Environmental Protection Agency (EPA). 1996b. The USEPA Reach File Version 1.0 (RF1) for the
Conterminous United States (CONUS): http://www.epa.gov/waterscience/BASINS/metadata/rfl.htm.
As cited in Schwarz et al. 2006.
U.S. Environmental Protection Agency (EPA). 1997. Development of Mixing Zone Dilution Factors.
Final report, EPA Contract No. 68-D3-0013 (Work Assignment No. 2-12) and EPA Contract No. 68-
W6-0023 (Work Assignment No. 1-2). Prepared by Versar, Inc. for the U.S. Environmental
Protection Agency, Economics, Exposure and Technology Division, Office of Pollution Prevention
and Toxics, Washington, DC.
U.S. Environmental Protection Agency (EPA). 2000a. Arsenic in Drinking Water Rule- Economic
Analysis (EPA 815-R-00-026). U.S. EPA, Office of Water, Washington, DC, December, 2000.
U.S. Environmental Protection Agency (EPA). 2000b. Guidelines for Preparing Economic Analyses.
(EPA 240-R-00-003). U.S. EPA, Office of the Administrator, Washington, DC, September 2000.
November 2008 12-21
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
U.S. Environmental Protection Agency (EPA). 2000c. Technologies and Costs for Removal of Arsenic
from Drinking Water (EPA 815-R-00-028). U.S. EPA, Office of Water, Washington, DC, December,
2000.
U.S. Environmental Protection Agency (EPA). 2002. Estimation of National Economic Benefits Using
the National Water Pollution Control Assessment Model to Evaluate Regulatory Options for
Concentrated Animal Feeding Operations (EPA-821-R-03-009). U.S. EPA, Office of Water,
Washington, DC, December 2002.
U.S. Environmental Protection Agency (EPA). 2004a. Biological Opinion for Threatened and Endangered
Species USEPA Superfund Program Revised Proposed Clean-up Plan for the Milltown Reservoir
Sediments Operable Unit of the Milltown Reservoir/Clark Fork River Superfund Site. Endangered
Species Act - Section 7 Consultation.
U.S. Environmental Protection Agency (EPA). 2004b. Development Document for Final Action for
Effluent Guidelines and Standards for the Construction and Development Category. EPA-821-B-04-
001.
U.S. Environmental Protection Agency (EPA). 2004c. Regional Analysis for the Final Section 316(b)
Phase II Existing Facilities Rule (EPA-821-R-04-006). U.S. EPA, Office of Water, Washington, DC,
February 2004.
U.S. Environmental Protection Agency (EPA). 2005a. Guidance for 2006 Assessment, Listing and
Reporting Requirements Pursuant to Sections 303(d), 305(b) and 314 of the Clean Water Act. U.S.
EPA, Office of Water, Washington, DC. July 29, 2005.
U.S. Environmental Protection Agency (EPA). 2005b. Technologies and Costs Document for the Final
Long Term 2 Enhanced Surface Water Treatment Rule and Final Stage 2 Disinfectants and
Disinfection Byproducts Rule. EPA 815-R-05-013. U.S. EPA, Office of Water, Washington, DC.
December 2005.
U.S. Environmental Protection Agency (EPA). 2006a. "Safe Drinking Water Information System /
Federal Version: About SDWIS/FED." Office of Ground Water and Drinking Water. Updated 22
Feb. 2006. http://www.epa.gov/ogwdw/sdwisfed/about_fed.html.
U.S. Environmental Protection Agency (EPA). 2006b. Framework for Developing Suspended and Bedded
Sediments (SABS) Water Quality Criteria. EPA-822-R-06-001.Office of Water, Office of Research &
Development.
U.S. Environmental Protection Agency (EPA). 2007a. "Identification and Characterization of Typical
Drinking Water Treatment Residuals." Peer Involvement for the Drinking Water Treatment Effluent
Guidelines. September 2007.
U.S. Environmental Protection Agency (EPA). 2007b. Master Survey Response Database for the Water
Treatment Plant Questionnaire (EPA Form No. 2040-0269).
U.S. Environmental Protection Agency (EPA). 2007c. National Estuary Program Coastal Condition
Report. U.S. EPA, National Coastal Assessment, Office of Wetlands, Oceans, and Watersheds, and
Office of Research and Development, Washington, DC, June 2007.
U.S. Environmental Protection Agency (EPA). 2007d. National Water Quality Inventory: Report to
Congress: 2002 Reporting Cycle. EPA 841-R-07-001.
http://www.epa.gov/305b/2002report/report2002305b.pdf
November 2008 12-22
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
U.S. Environmental Protection Agency (EPA). 2007e. U.S. EPA Reach File 1 (RF1) forthe
Conterminous United States in BASINSRF1. Available:
http://www.epa.gov/waterscience/BASINS/metadata/rfl.htm - Metadata Reference Information. As
cited in Schwarz 2008a.
U.S. Environmental Protection Agency (EPA). 2008a. BASINS (Better Assessment Science Integrating
Point & Non-point Sources). Available: http://www.epa.gov/waterscience/basins/bsnsdocs.html.
U.S. Environmental Protection Agency (EPA). 2008b. Development Document for Proposed Effluent
Guidelines and Standards forthe Construction and Development Category. EPA-821-R-08-007.
U.S. Environmental Protection Agency (EPA). 2008c. Economic Analysis of Proposed Effluent
Guidelines and Standards forthe Construction & Development Industry. EPA-821-R-08-008.
U.S. Environmental Protection Agency (EPA). 2008d. Environmental Monitoring and Assessment
Program: National Coastal Assessment (EMAP-NCA). Accessed October 1, 2008. Available at
http://www.epa.gov/emap/nca/html/data/index.html.
U.S. Environmental Protection Agency (EPA). 2008e. Polycyclic Aromatic Hydrocarbons (PAHs).
Factsheet.
U.S. Environmental Protection Agency (EPA). 2008f Welcome to STORET, EPA's largest computerized
environmental data system. Available: http://www.epa.gov/storet/.
U.S. Environmental Protection Agency (EPA). 2008g. "Water Assessment, Tracking & Environmental
Results (WATERS)." Office of Wetlands, Oceans, and Watersheds. Available at
http://www.epa.gov/waters/data/index.html.
U.S. Fish and Wildlife Service (USFWS). 1996. The Endangered Species Act of 1973. A Summary of the
ESA and Implementation Activities. Updated October 1996. U.S. Fish and Wildlife Service.
http://endangered.fws.gov/esasum.html. Accessed April 2, 2008.
U.S. Fish and Wildlife Service (USFWS). 2001. 2001 National Survey of Fishing, Hunting, and Wildlife-
Associated Recreation. Available at http://fa.r9.fws.gov/surveys/surveys.html#survey_reports.
U.S. Fish and Wildlife Service (USFWS). 2007. Listing a Species as Threatened or Endangered: Section
4 of the Endangered Species Act. Available at: http://www.fws.gov/endangered/factsheets/listing.pdf
Accessed April 2, 2008.
U.S. Fish and Wildlife Service (USFWS). 2008a. The Endangered Species Act and What We Do.
Available at: http://www.fws.gov/endangered/whatwedo.html. Accessed April 2, 2008.
U.S. Fish and Wildlife Service (USFWS). 2008b. Endangered Species Program: Species Information.
Available at: http://www.fws.gov/endangered/wildlife.html. Accessed April 2, 2008.
U.S. Geological Survey (USGS). 1999. The national hydrography dataset: USGS Fact Sheet 106-99.
Available: http://erg.usgs.gov/isb/pubs/factsheets/fs 10699.html.
U.S. Geological Survey (USGS). 2001. USGS National Water Quality Assessment Program (NAWQA).
Available at http://water.usgs.gov/nawqa/data/. As cited in Schwarz 2008a.
U.S. Geological Survey (USGS). 2003. Federal lands of the United States, available online at
http://nationalatlas.gov/mld/fedlanp.html. (Accessed June 5, 2008.) As cited in Schwarz 2008a.
November 2008 12-23
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
U.S. Geological Survey. 2006a, HYDROlk Documentation, available online at
http://edc.usgs.gov/products/elevation/gtopo30/hydro/readme.html. Accessed June 5, 2008. As cited
in Schwarz 2008a.
U.S. Geological Survey (USGS). 2006b. National Stream Quality Accounting Network (NASQAN)
Published Data, available online at http://water.usgs.gov/nasqan/data/finaldata.html. (Accessed
January 30, 2008.) As cited in Schwarz 2008a.
U.S. Geological Survey (USGS). 2007a. "Hydrologic Unit Codes (HUCs) Explained." Accessed
November 5, 2008. Available at http://nas.er.usgs.gov/hucs.asp.
U.S. Geological Survey (USGS). 2007c. Reservoir Sedimentation Information System (RESIS).
U.S. Geological Survey (USGS). 2008. National Water Information System - Web Interface. USGS
Surface Water Data for the Nation. (http://waterdata.usgs.gov/nwis/sw). Accessed June 5, 2008.
UK Special Areas of Conservation (SACs) Project. "PAHs (in general)." Accessed March 28, 2008.
Available at http://www.ukmarinesac.org.uk/activities/water-quality/wq8_40.htm#a4.
Van Metre, P.C., Barbara J .Mahler, Mateo Scoggins, and Pixie A. Hamilton. 2006. Parking Lot Sealcoat:
A Major Source of Fob/cyclic Aromatic Hydrocarbons (PAHs) in Urban and Suburban Environments.
U.S. Geological Survey (USGS) Fact Sheet 2005-3147. January 2006.
Van Nieuwenhuyse, E. E., and J. D. La Perriere. 1986. "Effects of placer gold mining on primary
production in subarctic streams of Alaska. "As cited in Waters 1995.
VandenBerg, T.P., G.L. Poe, and J.R. Powell. 2001. Assessing the accuracy of benefits transfers:
Evidence from a multi-site contingent valuation study of ground water quality. In The Economic
Value of Water Quality, J.C. Bergstrom, K.J. Boyle, and G.L. Poe (eds.). Edward Elgar Publishers,
Northampton, MA.
Vaughan, W.J. 1986. The RFF Water Quality Ladder. Appendix B in R.C. Mitchell and R.T. Carson, The
Use of Contingent Valuation Data for Benefit/Cost Analysis in Water Pollution Control, Final Report.
Washington: Resources for the Future.
Vecchia, A.V. 2000. Water-quality trend analysis and sampling design for the Souris River,
Saskatchewan, North Dakota, and Manitoba: U.S. Geological Survey Water- Resources
Investigations Report 00-4019, 77 p. As cited in Schwarz et al. 2006.
Vethaak, A.D. and AP Rheinhallt, T. 1992. "Fish disease as a monitor for marine pollution: the case of
the North Sea." As cited in UK SACs Project (2008).
Vice, R. B., H. P. Guy, and G. E. Ferguson (USGS). 1969. Sediment Movement in an Area of Suburban
Highway Construction, Scott Run Basin, Fairfax County, Virginia, 1961-64. Geological Survey
Water-Supply Paper 1591-E. U.S. Government Printing Office.
Vollenweider, R.A. 1976. Advances in defining critical loading levels for phosphorus lake eutrophication.
As cited in Schwarz et al. 2006. Memorie dell 'Institute Italiano di Idrobiologia 33:53-83.
Ward, Stephen M., and D. H. Appel. 1988. Suspended-Sediment Yields in the Coal River and Trace Fork
Basins, West Virginia, 1975-84. Water-Resources Investigations Report 87-4171.
Waters, T. F. 1995. Sediment in Streams: Sources, Biological Effects, and Control. Bethesda, MD;
American Fisheries Society Monograph 7.
November 2008 12-24
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Welle, P.O. 1986. Potential Economic Impacts of Acid Deposition: A Contingent Valuation Study of
Minnesota. Dissertation, University of Wisconsin-Madison.
Wey, K.A. 1990. Social Welfare Analysis of Congestion and Water Quality of Great Salt Pond, Block
Island, Rhode Island. Dissertation, University of Rhode Island.
Wheeler, Andrew P., Amanda E. Keiter Rosenberger, and Paul L. Angermeier. 2003. "Potential Impacts
of 1-73 on Stream Habitat and Biota,with Emphasis on the Federally Endangered Roanoke Logperch"
Department of Fisheries and Wildlife services, Virginia Polytechnic Institute and State University.
White, D. S., and J. R. Gammon. 1977. "The effect of suspended solids on macroinvertebrate drift in an
Indiana creek." As cited in Waters 1995.
Whitehead, J.C. and G.C. Blomquist. 1991a. A Link Between Behavior, Information, and Existence
Value. Leisure Sciences 13:97-109.
Whitehead, J.C. and G.C. Blomquist. 1991b. Measuring contingent values for wetlands: Effects of
information about related environmental goods. Water Resources Research 27(10):2523-2531.
Whitehead, J.C., and P.A. Groothuis. 1992. Economic Benefits of Improved Water Quality: a case study
of North Carolina's Tar-Pamlico River. Rivers. Vol. 3. pp. 170-178.
Whitehead, J.C., G.C. Blomquist, T.J. Hoban and W.B. Clifford. 1995. Assessing the Validity and
Reliability of Contingent Values: A Comparison of On-Site Users, Off-Site Users, and Non-users.
Journal of Environmental Economics and Management. Vol. 29. pp. 238-251.
Whitehead, J.C., T. Hoban, and W. Clifford. 2002. Landowners' Willingness to Pay for Water Quality
Improvements: Jointly Estimating Contingent Valuation and Behavior with Limited Information.
White paper developed in part by U.S. EPA, NCDENR and the College of Agriculture and Life
Sciences at NSCU.
Whittington, D., G. Cassidy, D. Amaral, E. McClelland, H. Wang and C. Poulos. 1994. The Economic
Value of Improving the Environmental Quality of Galveston Bay. Department of Environmental
Sciences and Engineering, University of North Carolina at Chapel Hill. GBNEP-38, 6/94.
Williams, J.R. 1995. The EPIC model. P. 909-1000. Chapter 25 in Computer Models of Watershed
Hydrology, V.P. Singh (ed.). Water Resources Publications, Highlands Ranch, CO.
Williams, J.R., C.A. Jones, and P.T. Dyke. 1984. A modeling approach to determining the relationship
between erosion and soil productivity. Trans. ASAE 27(1): 129-144. As cited in Neitsch et al. 2005.
Williams, J.R., R. C. Izaurralde, and E.M. Steglich. 2008. "Agricultural Policy/Environmental extender
Model: Theoretical Documentation. Version 0604". BREC Report #2008-17. Blacklands Research
and Extension Center, Texas A&M University, Temple, TX.
Wischmeier, W.H. andD.D. Smith. 1978. Predicting Rainfall Losses: A Guide to Conservation Planning.
USDA Agricultural Handbook No. 537. U.S. Government Printing Office, Washington, DC.
Wolman and Miller. 1960. Magnitude and frequency offerees in geomorphic processes. J. Geology 68:
54-74.
Wolman, M. Cordon, and Asher P. Schick. 1967. "Effects of Construction on Fluvial Sediment, Urban
and Suburban Areas of Maryland." Water Resources Research Vol. 3(2): pp. 451-464.
November 2008 12-25
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Wong, Michael F. (USGS). 2005. Water Quality in the Halawa, Haiku, and Kaneohe Drainage Basins
Before, During, and After H-3 Highway Construction, Oahu, Hawaii, 1983-99. Scientific
Investigations Report 2004-5002.
Wood, Paul J., and Patrick D. Armitage. 1997. "Biological Effects of Fine Sediment in the Lotic
Environment." Environmental Management Vol. 21(2): pp 203-217.
Woodward, R.T., and Y.S. Wui. 2001. The economic value of wetland services: a meta-analysis.
Ecological Economics Vol. 37, No. 2. pp. 257-270.
Yao, Kuan M. (ASCE). 1975. "Extended Plain Sedimentation." (Abstract). Journal of the Environmental
Engineering Division. Vol. 101, No. 3: May-June 1975, pp. 413-423.
Yapo, P.O., H.V. Gupta, and S. Sorooshian. 1996. Calibration of conceptual rainfall-runoff models -
sensitivity to calibration data. Journal of Hydrology 181:23-48. As cited in Schwarz et al. 2006.
Yoder, C.O. 1997. Important concepts and elements of an adequate state watershed monitoring and
assessment program, ASIWPCA Standards and Monitoring Task Force: prepared for USEPA, Office
of Water (Cooperative agreement CX 825484-01-0), 38 p. As cited in Schwarz et al. 2006.
Yorke, Thomas H., and William J. Herb (USGS). 1978. Effects of Urbanization on Streamflow and
Sediment Transport in the Rock Creek and Anacostia River Basins, Montgomery County, Maryland,
1962-74. Geological Survey Professional Paper 1003.
Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1995. AGNPS user's guide, version 5.00,
1995. Morris, MN, Agricultural Research Service, U.S. Department of Agriculture. As cited in
Schwarz et al. 2006.
Young, Robert J. and Gerald L. Mackie. 1991. "Effect of oil pipeline construction on the benthic
invertebrate community structure of Hodgson Creek, Northwest Territories." Canadian Journal of
Zoology Vol. 69: pp. 2154-2160.
November 2008 12-26
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Appendix A - Threatened and Endangered Aquatic Species
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Scientific Name
Threatened/Endangered
Status
Fish
Catfish, Yaqui
Cavefish, Alabama
Cavefish, Ozark
Chub, bonytail
Chub, Borax Lake
Chub, Chihuahua
Chub, Gila
Chub, humpback
Chub, Hutton tui
Chub, Mohave tui
Chub, Oregon
Chub, Owens tui
Chub, Pahranagat roundtail
Chub, slender
Chub, Sonora
Chub, spotfin
Chub, Virgin River
Chub, Yaqui
Cui-ui
Dace, Ash Meadows speckled
Dace, blackside
Dace, Clover Valley speckled
Dace, desert
Dace, Foskett speckled
Dace, Independence Valley speckled
Dace, Kendall Warm Springs
Dace, Moapa
Darter, amber
Darter, bayou
Darter, bluemask (=jewel)
Darter, boulder
Darter, Cherokee
Darter, duskytail
Darter, Etowah
Darter, fountain
Darter, goldline
Darter, leopard
Darter, Maryland
Darter, Niangua
Darter, Okaloosa
Darter, relict
Ictalurus pricei
Speoplatyrhinus poulsoni
Amblyopsis rosae
Gila elegans
Gila boraxobius
Gila nigrescens
Gila intermedia
Gila cypha
Gila bicolor ssp.
Gila bicolor mohavensis
Oregonichthys crameri
Gila bicolor snyderi
Gila robustajordani
Erimystax cahni
Gila ditaenia
Erimonax monachus
Gila seminuda
Gila purpurea
Chasmistes cujus
Rhinichthys osculus nevadensis
Phoxinus cumberlandensis
Rhinichthys osculus oligoporus
Eremichthys acros
Rhinichthys osculus ssp.
Rhinichthys osculus lethoporus
Rhinichthys osculus thermalis
Moapa coriacea
Percina antesella
Etheostoma rubrum
Etheostoma sp.
Etheostoma wapiti
Etheostoma scotti
Etheostoma percnurum
Etheostoma etowahae
Etheostoma fonticola
Percina aurolineata
Percina pantherina
Etheostoma sellare
Etheostoma nianguae
Etheostoma okaloosae
Etheostoma chienense
Threatened
Endangered
Threatened
Endangered
Endangered
Threatened
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Threatened
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Threatened
Threatened
Endangered
Threatened
Endangered
Endangered
November 2008
A-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Darter, slackwater
Darter, snail
Darter, vermilion
Darter, watercress
Gambusia, Big Bend
Gambusia, Clear Creek
Gambusia, Pecos
Gambusia, San Marcos
Goby, tidewater
Logperch, Conasauga
Logperch, Roanoke
Madtom, Neosho
Madtom, pygmy
Madtom, Scioto
Madtom, smoky
Madtom, yellowfin
Minnow, Devils River
Minnow, loach
Minnow, Rio Grande silvery
Pikeminnow, Colorado
Poolfish, Pahrump
Pupfish, Ash Meadows Amargosa
Pupfish, Comanche Springs
Pupfish, desert
Pupfish, Devils Hole
Pupfish, Leon Springs
Pupfish, Owens
Pupfish, Warm Springs
Salmon, Atlantic
Salmon, chinook
Salmon, chum
Salmon, coho
Salmon, sockeye
Sawfish, smalltooth
Sculpin, pygmy
Shiner, Arkansas River
Shiner, beautiful
Shiner, blue
Shiner, Cahaba
Shiner, Cape Fear
Shiner, palezone
Shiner, Pecos bluntnose
Shiner, Topeka
Silverside, Waccamaw
Smelt, delta
Spikedace
Scientific Name
Etheostoma boschungi
Percina tanasi
Etheostoma chermocki
Etheostoma nuchale
Gambusia gaigei
Gambusia heterochir
Gambusia nobilis
Gambusia georgei
Eucyclogobius newberryi
Percina jenkinsi
Percina rex
Noturus placidus
Noturus stanauli
Noturus trautmani
Noturus baileyi
Noturus flavipinnis
Dionda diaboli
Tiaroga cobitis
Hybognathus amarus
Ptychocheilus lucius
Empetrichthys latos
Cyprinodon nevadensis mionectes
Cyprinodon elegans
Cyprinodon macularius
Cyprinodon diabolis
Cyprinodon bovinus
Cyprinodon radiosus
Cyprinodon nevadensis pectoralis
Salmo salar
Oncorhynchus tshawytscha
Oncorhynchus keta
Oncorhynchus Msutch
Oncorhynchus nerka
Pristis pectinata
Cottus paulus
Notropis girardi
Cyprinella formosa
Cyprinella caerulea
Notropis cahabae
Notropis mekistocholas
Notropis albizonatus
Notropis simus pecosensis
Notropis topeka
Menidia extensa
Hypomesus transpacificus
Medafulgida
Threatened/Endangered
Status
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Threatened
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened/Endangered
Threatened
Threatened/Endangered
Threatened/Endangered
Endangered
Threatened
Threatened
Threatened
Threatened
Endangered
Endangered
Endangered
Threatened
Endangered
Threatened
Threatened
Threatened
November 2008
A-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Spinedace, Big Spring
Spinedace, Little Colorado
Spinedace, White River
Springfish, Hiko White River
Springfish, Railroad Valley
Springfish, White River
Steelhead
Stickleback, unarmored threespine
Sturgeon, Alabama
Sturgeon, gulf
Sturgeon, North American green
Sturgeon, pallid
Sturgeon, shortnose
Sturgeon, white
Sucker, June
Sucker, Lost River
Sucker, Modoc
Sucker, razorback
Sucker, Santa Ana
Sucker, shortnose
Sucker, Warner
Topminnow, Gila (incl. Yaqui)
Trout, Apache
Trout, bull
Trout, Gila
Trout, greenback cutthroat
Trout, Lahontan cutthroat
Trout, Little Kern golden
Trout, Paiute cutthroat
Woundfin
Scientific Name
Lepidomeda mollispinis pratensis
Lepidomeda vittata
Lepidomeda albivallis
Crenichthys baileyi grandis
Crenichthys nevadae
Crenichthys baileyi baileyi
Oncorhynchus mykiss
Gasterosteus aculeatus williamsoni
Scaphirhynchus suttkusi
Acipenser oxyrinchus desotoi
Acipenser medirostris
Scaphirhynchus albus
Acipenser brevirostrum
Acipenser transmontanus
Chasmistes liorus
Deltistes luxatus
Catostomus microps
Xyrauchen texanus
Catostomus santaanae
Chasmistes brevirostris
Catostomus warnerensis
Poeciliopsis occidentalis
Oncorhynchus apache
Salvelinus confluentus
Oncorhynchus gilae
Oncorhynchus clarki stomias
Oncorhynchus clarki henshawi
Oncorhynchus aguabonita whitei
Oncorhynchus clarki seleniris
Plagopterus argentissimus
Threatened/Endangered
Status
Threatened
Threatened
Endangered
Endangered
Threatened
Endangered
Threatened/Endangered
Endangered
Endangered
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Threatened
Endangered
Threatened
Threatened
Threatened
Threatened
Threatened
Threatened
Threatened
Endangered
Amphibians
Coqui, golden
Frog, California red-legged
Frog, Chiricahua leopard
Frog, Mississippi gopher
Frog, mountain yellow-legged
Guajon
Salamander, Barton Springs
Salamander, California tiger
Salamander, Cheat Mountain
Salamander, desert slender
Salamander, flatwoods
Salamander, Red Hills
Salamander, San Marcos
Salamander, Santa Cruz long-toed
Salamander, Shenandoah
Eleutherodactylusjasperi
Rana aurora draytonii
Rana chiricahuensis
Rana capita sevosa
Rana muscosa
Eleutherodactylus cooki
Eurycea sosorum
Ambystoma californiense
Plethodon nettingi
Batrachoseps aridus
Ambystoma cingulatum
Phaeognathus hubrichti
Eurycea nana
Ambystoma macrodactylum croceum
Plethodon shenandoah
Threatened
Threatened
Threatened
Endangered
Endangered
Threatened
Endangered
Threatened/Endangered
Threatened
Endangered
Threatened
Threatened
Threatened
Endangered
Endangered
November 2008
A-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Salamander, Sonora tiger
Salamander, Texas blind
Toad, arroyo (=arroyo southwestern)
Toad, Houston
Toad, Puerto Rican crested
Toad, Wyoming
Scientific Name
Ambystoma tigrinum stebbinsi
Typhlomolge rathbuni
Bufo californicus (=microscaphus)
Bufo houstonensis
Peltophryne lemur
Bufo baxteri (=hemiophrys)
Threatened/Endangered
Status
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Mollusks
Acornshell, southern
Bankclimber, purple (mussel)
Bean, Cumberland (pearlymussel)
Bean, purple
Blossom, green (pearlymussel)
Blossom, tubercled (pearlymussel)
Blossom, turgid (pearlymussel)
Blossom, yellow (pearlymussel)
Catspaw (=purple cat's paw
pearlymussel)
Catspaw, white (pearlymussel)
Clubshell
Clubshell, black
Clubshell, ovate
Clubshell, southern
Combshell, Cumberlandian
Combshell, southern
Combshell, upland
Elktoe, Appalachian
Elktoe, Cumberland
Fanshell
Fatmucket, Arkansas
Heelsplitter, Alabama (=inflated)
Heelsplitter, Carolina
Higgins eye (pearlymussel)
Kidneyshell, triangular
Lampmussel, Alabama
Lilliput, pale (pearlymussel)
Mapleleaf, winged
Moccasinshell, Alabama
Moccasinshell, Coosa
Moccasinshell, Gulf
Moccasinshell, Ochlockonee
Monkeyface, Appalachian
(pearlymussel)
Monkeyface, Cumberland
(pearlymussel)
Mucket, orangenacre
Mucket, pink (pearlymussel)
Epioblasma othcaloogensis
Elliptoideus sloatianus
Villosa trabalis
Villosa perpurpurea
Epioblasma torulosa gubernaculum
Epioblasma torulosa torulosa
Epioblasma turgidula
Epioblasma florentina florentina
Epioblasma obliquata obliquata
Epioblasma obliquata perobliqua
Pleurobema clava
Pleurobema curtum
Pleurobema perovatum
Pleurobema decisum
Epioblasma brevidens
Epioblasma penita
Epioblasma metastriata
Alasmidonta raveneliana
Alasmidonta atropurpurea
Cyprogenia stegaria
Lampsilis powelli
Potamilus inflatus
Lasmigona decorata
Lampsilis higginsii
Ptychobranchus greenii
Lampsilis virescens
Toxolasma cylindrellus
Quadrula fragosa
Medionidus acutissimus
Medionidus parvulus
Medionidus penicillatus
Medionidus simpsonianus
Quadrula sparsa
Quadrula intermedia
Lampsilis perovalis
Lampsilis abrupta
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
November 2008
A-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Mussel, oyster
Mussel, scaleshell
Pearlshell, Louisiana
Pearlymussel, birdwing
Pearlymussel, cracking
Pearlymussel, Curtis
Pearlymussel, dromedary
Pearlymussel, littlewing
Pigtoe, Cumberland
Pigtoe, dark
Pigtoe, finerayed
Pigtoe, flat
Pigtoe, heavy
Pigtoe, oval
Pigtoe, rough
Pigtoe, shiny
Pigtoe, southern
Pimpleback, orangefoot
(pearlymussel)
Pocketbook, fat
Pocketbook, finelined
Pocketbook, Ouachita rock
Pocketbook, shinyrayed
Pocketbook, speckled
Rabbitsfoot, rough
Riffleshell, northern
Riffleshell, tan
Ring pink (mussel)
Slabshell, Chipola
Spinymussel, James
Spinymussel, Tar River
Stirrupshell
Three-ridge, fat (mussel)
Wartyback, white (pearlymussel)
Wedgemussel, dwarf
Scientific Name
Epioblasma capsaeformis
Leptodea leptodon
Margaritifera hembeli
Conradilla caelata
Hemistena lata
Epioblasma florentina curtisii
Dromus dramas
Pegiasfabula
Pleurobema gibberum
Pleurobemafurvum
Fusconaia cuneolus
Pleurobema marshalli
Pleurobema taitianum
Pleurobema pyriforme
Pleurobema plenum
Fusconaia cor
Pleurobema georgianum
Plethobasus cooperianus
Potamilus capax
Lampsilis altilis
Arkansia wheeleri
Lampsilis subangulata
Lampsilis streckeri
Quadrula cylindrica strigillata
Epioblasma torulosa rangiana
Epioblasma florentina walkeri (=E.
walker i)
Obovaria retusa
Elliptic chipolaensis
Pleurobema collina
Elliptic steinstansana
Quadrula stapes
Amblema neislerii
Plethobasus cicatricosus
Alasmidonta heterodon
Threatened/Endangered
Status
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Crustaceans
Amphipod, Hay's Spring
Amphipod, Illinois cave
Amphipod, Kauai cave
Amphipod, Noel's
Amphipod, Peck's cave
Crayfish, cave
Crayfish, cave
Crayfish, Nashville
Crayfish, Shasta
Stygobromus hayi
Gammarus acherondytes
Spelaeorchestia koloana
Gammarus desperatus
Stygobromus (=Stygonectes) pecki
Cambarus aculabrum
Cambarus zophonastes
Orconectes shoupi
Pacifastacusfortis
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
Endangered
November 2008
A-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table A-1: List of Federal Threatened and Endangered Aquatic Species Potentially
Impacted by Sediment
Common name
Fairy shrimp, Conservancy
Fairy shrimp, longhorn
Fairy shrimp, Riverside
Fairy shrimp, San Diego
Fairy shrimp, vernal pool
Isopod, Lee County cave
Isopod, Madison Cave
Isopod, Socorro
Shrimp, Alabama cave
Shrimp, California freshwater
Shrimp, Kentucky cave
Shrimp, Squirrel Chimney Cave
Tadpole shrimp, vernal pool
Scientific Name
Branchinecta conservatio
Branchinecta longiantenna
Streptocephalus woottoni
Branchinecta sandiegonensis
Branchinecta lynchi
Lirceus usdagalun
Antrolana lira
Thermosphaeroma thermophilus
Palaemonias alabamae
Syncaris pacifica
Palaemonias ganteri
Palaemonetes cummingi
Lepidurus packardi
Threatened/Endangered
Status
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Threatened
Endangered
Endangered
Endangered
Endangered
Threatened
Endangered
Corals
Coral, elkhorn
Coral, staghorn
Acropora palmata
Acropora cervicornis
Threatened
Threatened
November 2008
A-6
-------
Environmental Assessment and Benefits Document for the C&D Regulation
Appendix B - SPARROW Model Documentation
B.1 General Overview
SPARROW (SPAtially Referenced Regressions On Watershed attributes) is a watershed modeling
technique for estimating contaminant source contributions and transport in watersheds and surface waters.
SPARROW employs a statistically estimated nonlinear regression model with contaminant supply and
process components, including surface-water flow paths, non-conservative transport processes, and mass-
balance constraints. Regression equation parameters are estimated by correlating stream water-quality
records with GIS (Geographic Information System) data on pollutant sources (atmospheric, point,
nonpoint) and climatic and hydrogeologic properties (e.g., precipitation, topography, vegetation, soils,
water routing) that affect contaminant transport. The procedure for estimating SPARROW parameters
also provides measures of uncertainty in model coefficients and in water-quality predictions. The
SPARROW software is written in SAS (Statistical Analysis System) IML (Interactive Matrix Language).
SPARROW model infrastructure consists of a detailed watershed - stream reach network to which all
monitoring stations and GIS data on watershed properties are spatially referenced. Digital elevation
models (DEMs) are used to delineate watershed boundaries and to identify overland flowpaths. The
spatially distributed model structure allows separate statistical estimation of land- and water-related
parameters that quantify rates of pollutant delivery from sources to streams, and transport of pollutants to
downstream locations within the stream network (i.e., reaches, reservoirs and estuaries). Mechanistic
separation of terrestrial and aquatic features of large watersheds, and improved parameter estimation
techniques represent significant advances in water-quality modeling to objectively evaluate alternative
hypotheses about important contaminant sources and watershed properties that control transport over
large spatial scales.
SPARROW has been applied in the analysis of suspended sediment, surface-water nutrients, pesticides,
organic carbon, and fecal bacteria, and is potentially applicable to other measures of water quality. Recent
applications of SPARROW have provided estimates of nutrient sources and long-term rates of nutrient
removal in surface waters (e.g., Smith et al., 1997; Alexander et al., 2000; 2001; 2002a; 2002b). The
model has demonstrated particular utility for quantifying long-distance transport and delivery of nutrients
to sensitive downstream locations (e.g., estuaries, reservoirs, drinking water intakes). The earliest version
of the SPARROW model was developed to describe contaminant transport in surface waters of the State
of New Jersey (Smith et al., 1994). Federal and State environmental managers are currently using
SPARROW to assess the sources of nutrient loadings in streams, including targeting of nutrient reduction
strategies in the Chesapeake Bay watershed (Preston and Brakebill, 1999) and in waters of the State of
Kansas (Kansas Dept. Health and Environment, 2004) as well as for developing TMDLs (Total
Maximum Daily Loads) in the Connecticut River Basin (NEIWPCC, 2004). Other applications
encompass New England watersheds (Moore et al., 2004), New Zealand river basins (Alexander et al.,
2002a; Elliott et al., 2005), North Carolina coastal watersheds (McMahon et al., 2003), and watersheds in
Tennessee and Kentucky (Hoos, 2005). SPARROW models are currently under development for the
Delaware River Basin and are being planned for selected regions of the U.S. under the U.S. Geological
Survey (USGS) National Water-Quality Assessment (NAWQA) Program.
November 2008
B-1
-------
Environmental Assessment and Benefits Document for the C&D Regulation
B.2 Modeling Concept
B.2.1 Objectives of SPARROW Modeling
The broad objective of SPARROW modeling is to establish a mathematical relation between water-
quality measurements made at a network of monitoring stations and attributes of the watersheds
containing the stations. Once constructed, the model may be used to satisfy a variety of water-quality
information objectives. One common modeling objective is to describe past or present water-quality
conditions for a state or region on the basis of monitoring data. The underlying challenge is to extrapolate
from a sample of water-quality measurements made at a finite number of stream and river locations to an
area containing un-sampled locations. The usual limitations in attempting this are: (1) sparse sampling,
reflecting the high cost of monitoring; and/or (2) unrepresentative (nonrandom or targeted) sampling
undertaken to characterize water quality at specific locations, especially those suspected of having water-
quality problems. In the absence of an interpretive model such as SPARROW, a single monitoring design
cannot be optimal for both of these distinct objectives. Because the Federal Clean Water Act requires
state governments to collect and report both types of information, the distinction between the two types of
monitoring is of great practical importance. "Probabilistic" monitoring has been promoted by the
U.S. Environmental Protection Agency (EPA) (Yoder, 1997) to obtain a spatially unbiased, broad
overview of water-quality conditions in State waters. An important limitation of this approach is that the
monitoring data alone do not provide detailed information on the geography of water-quality conditions
and give little understanding of the factors (sources and processes) that explain those conditions. Targeted
monitoring has been used extensively by the States to identify specific streams with water-quality
problems and to gage compliance with State water-quality regulatory standards and criteria. These data,
however, provide a spatially biased description of water-quality conditions in watersheds. SPARROW is
effective in integrating samples from these different monitoring approaches to provide both a
geographically representative description of water-quality conditions as well as insight into the sources
and watershed processes that control water quality.
A second objective of SPARROW modeling is to identify and quantify the sources of pollution that give
rise to in-stream water-quality conditions. SPARROW distinguishes between source categories, such as
point sources, atmospheric sources, and agriculture; and individual sources, defined as the rate of supply
of contaminant of a particular category originating in the watershed and draining to a specific stream
reach. The ability to develop quantitative information on pollution sources in SPARROW models stems
from the ability to trace, for each contaminant category, the predicted in-stream flux through a given
stream reach to the individual sources in each of the upstream reach watersheds contributing
contamination to that reach. Sources may be quantified either in mass units or in terms of their percent
contribution to the total contaminant flux to the reach. An example application of SPARROW in
quantifying pollution sources is in TMDL analyses (e.g., McMahon and Roessler, 2002). In general, the
Clean Water Act requires TMDL analyses for any stream reach in which the concentration of a
contaminant exceeds the applicable water-quality standard when all pollution discharge limits are met.
The ultimate objective of TMDL-related modeling is to establish a hypothetical waste-load allocation for
all individual sources affecting the reach in question that would meet the standard. In theory, an infinite
number of hypothetical load allocations will satisfy the standard, and choosing the official allocation
requires comparing many possible solutions, utilizing simulation (see below) in search of a least-cost or
other optimum solution. Prior to conducting the hypothetical analyses, however, a great deal of
November 2008 B-2
-------
Environmental Assessment and Benefits Document for the C&D Regulation
preliminary quantitative information on the actual or baseline relations between watershed sources and in-
stream conditions is useful. The percentage contributions (shares) of individual point sources are of
particular interest because point sources are usually the only sources subject to direct regulation. The
share contributions of individual nonpoint sources are also of interest as a means of identifying the
important stakeholders to include in discussions of voluntary pollution reductions. This describes the
current application of SPARROW: to evaluate the effectiveness of a range of proposed regulatory options
for controlling and mitigating sediments associated with construction activities, with explicit focus on the
impacts of such options on in-stream water quality as measured by Total Suspended Solids (TSS).
A third objective is water quality simulation. Simulation refers to the use of a calibrated model to predict
conditions on the basis of a set of altered, typically hypothetical model inputs. The ability to portray
counterfactual conditions for specified inputs is one of the most powerful justifications for models; there
are often no alternative methods for conducting controlled experiments on complex systems. Water-
quality model simulations can depict the in-stream effects of changes in contaminant sources associated
with alternative future pollution-control strategies, providing a critical step in the analysis of the costs and
benefits associated with specific Best Management Practices (BMP) or regulatory approaches. Simulating
alternative waste-load allocations in a TMDL analysis (see above) is a prime example of a water-quality
simulation application. In the present application, SPARROW simulation is used to conduct experiments
focused on specific regulatory options (BMPs) for managing construction site sediment that would not be
feasible to conduct physically outside a limited number of test sites.
A fourth objective of SPARROW modeling is hypothesis testing. One common feature of the preceding
modeling objectives is that they each make use of predictions of the dependent variable by a calibrated
model. Another class of modeling objectives focuses on the calibration process and its results directly.
The SPARROW estimation process explores the predictive value of a set of potential explanatory
variables and may also compare alternative mathematical forms. The selection of a final set of predictors
and mathematical form usually has the primary objective of maximizing the accuracy of model
predictions of the dependent variable, but an alternative modeling objective may be to test one or more
hypotheses about the nature and importance of factors and processes that may have influenced water
quality at the locations where samples were collected. Hypothesis tests are performed for each of the
model parameters estimated in the calibration, and these serve as indicators of an empirical relation
between the independent variables associated with each parameter and the dependent variable of the
model. Because the coefficients in a SPARROW model are specified to conform to physical processes,
and the potential explanatory variables are selected on the basis of theoretical or logical connections to the
dependent variable, a statistically strong parameter provides evidence of the physical relationship. For
example, if multiple categories of potential sources of the contaminant are being evaluated, the model
estimation process provides useful hypothesis tests on the importance of each category. "Importance" is
measured by the correlation between contaminant inputs from the source category and downstream
monitored loads of the contaminant. That correlation will tend to be stronger when the mass contribution
of the category to the total mass of contaminant flowing past many of the monitoring stations is large, but
model estimation may also indicate a source category is important even when the mass contribution is
small, provided the amount contributed to stream loads by each unit of source is consistent from place to
place.
The importance of factors and processes potentially related to the transport of contaminants from sources
to stream channels and within stream channels may be tested through calibration of the "land to water"
and "in-stream decay" terms in the model. These terms dictate the fraction of the contaminant mass that
completes the terrestrial and aquatic phases of transport within the watershed draining to each stream
November 2008 B-3
-------
Environmental Assessment and Benefits Document for the C&D Regulation
reach. The land-to-water terms describe the land-surface characteristics that influence both overland and
subsurface transport from sources to stream channels. Similarly, the in-stream decay terms describe the
effects of channel characteristics on downstream transport. In sum, an important objective in building and
calibrating SPARROW models is to gain insight and to test hypotheses concerning the role of specific
contaminant sources and hydrologic processes in supplying and transporting contaminants in watersheds.
B.2.2 SPARROW Mass Balance Approach
SPARROW models constructed to date, like most water-quality models, are expressed in the form of a
mass balance. Such models describe the movement of mass in space and/or the change of mass in time.
The law of conservation of mass implies that certain basic accounting rules must apply to a mass balance,
water-quality model, such as: (1) the sum effluxes entering the confluence of two streams equals the flux
leaving the confluence; (2) the sum of the fluxes attributable to each contaminant source must equal total
flux; and (3) a doubling of all sources in the model results in an exact doubling of the predicted flux at
each location. Because the dependent variable in SPARROW models (the mass of contaminant that
passes a specific stream location per unit time) is, in mathematical terms, linearly related to all sources of
contaminant mass in the model, all accounting rules relating to the conservation of mass will apply. Mass
accounting in SPARROW models is also supported by the explicit spatial structure defined by the stream
network.
There are a number of advantages to the mass balance approach. Because of the linear relation between
flux and sources, there is an expectation that the estimation of flux over spatial scales smaller and larger
than that covered by the model's sample data will yield reasonably accurate results. The imposition of
mass balance greatly improves the interpretability of model coefficients. For example, assuming mass
balance, the coefficient associated with the reach time-of-travel variable is interpreted as a first-order
decay rate and, because point-source loadings are delivered directly to the stream network, a reasonable
null hypothesis for coefficients associated with point sources is that they equal 1.0 (Smith et al., 1997).
The enhanced interpretability of the model coefficients in turn facilitates the comparison of coefficient
estimates from the model with other estimates described in the literature. These comparisons have been
generally favorable, especially for the model components that quantify in-stream nitrogen decay rates
(Alexander et al., 2000; Alexander et al., 2002a; McMahon et al., 2003), nutrient and suspended sediment
removal rates in reservoirs (Alexander et al., 2002a; Schwarz et al., 2001), and the nutrient export
associated with various land uses and pollutant sources (e.g., Alexander et al., 2001, 2004; Alexander,
Elliott et al., 2002; McMahon et al., 2003). Mass balance provides a basis for flux accounting, whereby
flux can be allocated to its various sources, both spatially and topically. For example, mass balance makes
it possible to attribute nutrients discharged to the Gulf of Mexico to specific sources within the
Mississippi basin (Alexander et al., 2000), thereby providing guidance in managing the reduction of this
discharge. This approach is also sensitive to the effects of natural and human-related processes that supply
and remove contaminants from watersheds over long periods. Thus, this approach de-emphasizes the
quantification of short-term cycling and transformation processes, which are often central to the
functioning of many dynamic mechanistic models, in favor of processes that have a long-term impact on
elemental budgets in aquatic ecosystems.
B.2.3 Time and Space Scales of the Model
SPARROW models are structurally designed to explain spatial variability in the long-term mean-annual
or mean-seasonal flux of contaminants in streams. Spatial variability is modeled as a function of natural
and human-related properties of watersheds that influence the supply and transport of contaminants.
November 2008 B-4
-------
Environmental Assessment and Benefits Document for the C&D Regulation
Estimates of the long-term mean flux are developed from water-quality and streamflow monitoring data
that are regularly collected at fixed locations on streams and rivers. The basic form of SPARROW models
is structured to describe the long-term, steady-state water-quality and flow conditions in streams.
Contaminant source inputs are assumed to be in balance with the estimated sinks and measured in-stream
water-quality load such that there is conservation of mass among the model components. The principal
objective supported by this model structure is the quantification of the location and rates (and statistical
uncertainties) of the supply, transport, and fate of contaminants within the terrestrial and aquatic
ecosystems of watersheds. In the current specifications of SPARROW models, temporal variability in
contaminant loads, including intra- and inter-annual variations in water quality and streamflow reflected
in the monitoring data, are explicitly modeled and accounted for in a step prior to modeling spatial
variability in loads with SPARROW.
The computation of mean-annual or mean-seasonal flux (the SPARROW response variable) requires the
prior application of a water-quality flux-estimation model constructed on the basis of streamflow and
water-quality records from regularly monitored stream locations. The flux-estimation model explicitly
accounts for temporal variability in contaminant loads related to streamflow, season of the year, and
trends (either continuous or abrupt) with time. A base-year load estimate ensures that the stream water-
quality loads and the contaminant source data (which are commonly reported only periodically, e.g., the
U.S. Agricultural Census reports every five years) are contemporaneous. Therefore, the mean-annual
loads used to calibrate SPARROW models describe the mean load that would be expected to occur during
a particular base year under long-term mean streamflow conditions.
The steady-state mass-balance structure of the basic SPARROW model quantifies the long-term net
effects of biogeochemical and hydrologic processes on contaminant transport in terrestrial and aquatic
ecosystems. Modeling the effects of these processes is typically of greatest interest for non-conservative
chemical, physical, and biological properties of water, such as nutrients, pesticides, fecal coliform
bacteria, organic carbon, and suspended sediment. Many of these constituents are subject to chemical
transformations or degradation during transport and may be stored over short or long periods. In
SPARROW models that are estimated using long-term water-quality records, biogeochemical cycling and
storage processes that temporarily immobilize or remove contaminants from flow paths are generally in
steady state with those processes that mobilize or release contaminants from storage. Hence, the effects of
transformation and removal processes that operate on relatively short intra-annual time scales (e.g., daily,
weekly, seasonal), or over multi-year time scales that are less frequent than the period of model
estimation, are not likely to be detected as contaminant losses in the steady-state form of SPARROW
models. Limitations in our knowledge of temporal lags in chemical transport and their causes also create
uncertainties in the periods over which steady-state conditions apply. For example, the time scales for the
transport of sediment in streams reflect erosion, storage, and transport processes that operate from
seasonal to decadal or even longer (e.g., century) periods (Trimble and Crosson, 2000).
Temporal changes in mean conditions also can be explicitly modeled in SPARROW using the current
model structure and software. One approach is to estimate multiple steady-state models that explain
mean-annual or mean-seasonal stream contaminant loads for separate multi-year periods. Alternatively,
the current software permits the estimation of a single SPARROW model having time-dimensioned
dependent and independent variables - each assumed to pertain to different steady-state conditions. In this
approach, temporal changes in mean-annual stream contaminant flux may be modeled as a function of
temporal changes in contaminant sources, land use, and climatic factors (e.g., precipitation, streamflow,
and temperature) over similar multi-year periods as those used to estimate stream loads at each
monitoring station. The coefficients in this model can be time dependent or restricted to take common
November 2008 B-5
-------
Environmental Assessment and Benefits Document for the C&D Regulation
values over the full period of the analysis. This type of model structure has considerable data demands
that require the development of historical data on contaminant sources and climatic/hydrologic variability
in the watersheds and reaches (including stream velocity, which is sensitive to average streamflow and
thus varies in response to changes in mean streamflow across different periods). Such a model allows
users to explicitly test for temporal changes in model parameters to determine whether changes are
evident in process rates or contaminant export coefficients with time. As an alternative to the explicit
estimation of a SPARROW model with time-dimensioned coefficients, existing SPARROW models can
be used to simulate changes in mean-annual stream water-quality loads as a function of changes in source
inputs. This approach assumes that the process rates and contaminant export coefficients of the model do
not change with time.
B.2.4 Accuracy and Complexity of SPARROW Models
It is generally the case for SPARROW, as for any model with statistically estimated parameters, that
model accuracy (bias and precision) and complexity (number of statistically significant or sensitive
parameters) are dependent on the information content of the water-resources data used in model
calibration. Investigations of hydrologic models have demonstrated that both the quantity and quality of
calibration data define the information content and have important effects on parameter estimation and
precision (e.g., Gupta and Sorooshian, 1985; Yapo et al., 1996). Increasing the quantity of data can
improve the precision provided the data give new, independent information about the values of the model
parameters. Data quality, as defined by Gupta and Sorooshian (1985), generally increases as the data
become more "representative" of the range of watershed properties that affect transport and the range of
conditions present in the sampled watersheds. An independent set of measurements are preferred for
estimating parameter values, such that the data reflect the most extreme combinations of watershed
conditions for the various properties. These general statistical guidelines have implications for the time
and space scales required to develop SPARROW data sets and accurate models. First, a sufficiently large
number of water-quality monitoring stations are required. In SPARROW models, the monitoring-station
loads serve as the response variable observations in the nonlinear spatial regressions. The number of
stations has a demonstrable effect on the statistical power of the regression - i.e., the capacity of the
model to detect the effect of an explanatory factor on stream loads. Models with more stations generally
have greater power, which typically supports more complex models - i.e., models with a larger number of
statistically significant parameters and functional components. Second, the amount of spatial variability in
the stream monitoring data and explanatory factors should reflect as broad and representative a range of
watershed conditions as possible. The most complex SPARROW models typically have been developed
for regions that have relatively large spatial variability (greater than one order of magnitude) in the
watershed properties that affect contaminant transport. Watershed properties that vary over a wide range
within a modeled region generally provide more information about the response of stream loads to
different levels of a given watershed property and are more likely to be statistically significant in
SPARROW models.
The spatial variability of a given variable is most readily increased in SPARROW models by expanding
the spatial domain of the model to include larger drainage areas. This, of course, has the effect of
increasing the number of monitoring stations, which also contributes to the potential for greater model
complexity. In selecting water-quality monitoring sites for modeling, it is especially important to obtain
sites that are located on a wide range of stream sizes (especially small streams) and are inclusive of
impoundments of varying size. Sediment and nutrient removal rates are highly responsive to the hydraulic
characteristics of streams and reservoirs, such as flow volume and velocity (Alexander et al., 2000;
November 2008 B-6
-------
Environmental Assessment and Benefits Document for the C&D Regulation
Alexander et al., 2002a). Accounting for the wide variation in these properties can provide more accurate
estimates of nutrient removal and improve the separation of land and water effects on transport in the
model. Estimation errors are much larger for constituents that are most affected by high flows, such as
suspended sediment, total phosphorus, and fecal coliform bacteria. These constituents are generally more
difficult to measure and exhibit larger variability in concentrations in streams, which can produce less
precise estimates of the mean-annual flux. By contrast, dissolved substances, such as nitrate, sulfate, and
total dissolved solids exhibit less variability with flow, and their fluxes can generally be more accurately
estimated.
B.2.5 Comparison of SPARROW with Other Watershed Models
A wide variety of hydrologic and water-quality models have been used to describe contaminant sources
and transport in watersheds and surface waters. These models can be characterized on the basis of their
process complexity and temporal and spatial scales (e.g., Singh, 1995). The level of complexity or process
detail represented by model descriptions of hydrologic and biogeochemical processes commonly varies
with the extent to which deterministic (mechanistic) and statistical/empirical methods are used to
describe and estimate these processes (Figure B-l; Alexander et al., 2002b). All models reflect some
blend of these methods, but most place greater emphasis on one or the other type of model structure and
process specification.
Figure B-1 : A Simple Continuum of Model Types Based on the Level of Statistical and Mechanistic
Descriptions of Contaminant Sources and Biogeochemical Processes
Statistical Deterministic
In general, purely statistical models tend to reflect more simplistic model constructs. These models have a
simple correlative mathematical structure and typically assume limited a priori knowledge of various
processes. Conventional versions of these models are expressed as simple linear (or log linear)
correlations of stream measurements with watershed sources and landscape properties (e.g., Caraco et al.,
2003; Peierls et al., 1991; Howarth et al., 1996). The methods have the advantage of being readily applied
in large watersheds (often relying on generally available stream monitoring records) and can readily
quantify the errors in model parameters and predictions. Simple correlative approaches, however, offer
little mechanistic explanation of contaminant sources and transport. They generally lack spatial detail on
the distribution of sources and sinks within watersheds, rarely account for nonlinear interactions between
sources and loss processes, and do not impose mass-balance constraints on contaminant transport. The
most purely statistical approaches are found in artificial neural network and kriging techniques. These
November 2008 B-7
-------
Environmental Assessment and Benefits Document for the C&D Regulation
models commonly provide an excellent fit to the observations, but provide little understanding of the
processes that affect contaminant transport.
By contrast, mechanistic water-quality models have complex mass-balance structures that simulate
hydrologic and contaminant transport processes, often according to relatively fine temporal scales
[e.g., Hydrologic Simulation Program-Fortran (HSPF), Bicknell et al., 2001; Soil Water Assessment Tool
(SWAT), Srinivasan et al., 1993; Agricultural Nonpoint Source Model (AGNPS), Young et al., 1995].
The components of these models frequently provide a highly detailed temporal description (e.g., daily,
hourly) of the response of stream contaminants to climatic variability, and the effects of coarser temporal
variations in land use and management activities are often superimposed on the more detailed climatic
variations. The mathematical descriptions of these responses are frequently based on a priori assumptions
about the dominant processes and their reaction rates. The complexity of mechanistic simulation models
creates intensive data and calibration requirements, which generally limits their application to relatively
small watersheds. Because mechanistic water-quality models are frequently calibrated manually, robust
measures of uncertainty in model parameters and predictions cannot be quantified. Despite the common
use of mechanistic models, there are growing concerns about whether sufficient water-resources data and
knowledge of biogeochemical processes exist to reliably support the general use of such highly complex
descriptions of processes (Beven, 2002; Jakeman and Hornberger, 1993). Without sufficient data, there is
limited ability to apply formal parameter-estimation techniques required to quantify model uncertainties
and to identify unique models having sensitive and uncorrelated parameters.
By comparison to other types of water-quality models, SPARROW may be best characterized as a hybrid
process-based and statistical modeling approach for estimating pollutant sources and contaminant
transport in surface waters. The mechanistic mass transport components of SPARROW include surface-
water flow paths (channel time of travel, reservoirs), non-conservative transport processes (i.e., first-order
in-stream and reservoir decay), and mass-balance constraints on model inputs (sources), losses (terrestrial
and aquatic losses/storage), and outputs (riverine nutrient export). The statistical features of the model
involve the use of nonlinear parameter-estimation for spatially correlating stream water-quality records
with geographic data on pollutant sources (e.g., atmospheric, point-, nonpoint) and climatic and
hydrogeologic properties (e.g., precipitation, topography, vegetation, soils, water routing). Parameter
estimation ensures that the calibrated model will not be more complex than can be supported by the data.
This provides an objective statistical approach for evaluating alternative hypotheses about important
contaminant sources and controlling transport processes over large spatial scales in watersheds.
SPARROW has been shown to improve the accuracy and interpretability of model parameters and the
predictions of nutrient loadings and sources in streams as compared with conventional statistical
modeling approaches (Smith et al., 1997; Alexander et al., 2000; Alexander et al., 2002a; 2002b).
A number of inter-model comparisons have been previously conducted that compare the performance and
properties of SPARROW to those of other water-quality models. These include national comparisons
with SWAT and comparisons in the Chesapeake Bay watershed with HSPF (see Alexander et al., 2001).
Inter-model comparisons also have been conducted with statistical and quasi-deterministic models in
watersheds of the northeastern U.S. (see Alexander et al., 2002b; Seitzinger et al., 2002). Finally,
evaluations of the SPARROW technique also are included as part of a number of recent National
Research Council (NRC) reports (2000, 200 la, 200 Ib, 2002a, 2002b). At least two of the NRC reports
(2000, 200la) have noted the advantages of statistical modeling approaches, such as SPARROW, for
general water-quality assessment and use in TMDL assessments.
November 2008 B-8
-------
Environmental Assessment and Benefits Document for the C&D Regulation
B.2.6 Model Infrastructure
The parameters of SPARROW models are statistically estimated with nonlinear regression techniques by
spatially correlating water-quality flux estimates at monitoring stations with watershed data on sources,
and landscape and surface-water properties that affect transport. The calibrated models are then used to
predict flux, total and disaggregated by contributing source, for stream reaches throughout a river
network. A flow diagram is provided in Figure B-2 to illustrate the functional linkages between the major
spatial components of SPARROW models. Pre-processing steps are required to develop reach-level
information for the major components of the SPARROW model infrastructure as shown in Figure B-3.
Monitoring station flux estimation refers to the estimates of long-term flux used as the response variable
in the model. Flux estimates at monitoring stations are derived from station-specific models that relate
contaminant concentrations from individual water-quality samples to continuous records of streamflow
and time. The stream reach, inclusive of its incremental contributing drainage basin, is the most elemental
spatial unit of the SPARROW model infrastructure. Stream reaches typically define the length of stream
channel that extends from one stream tributary junction to another. Explanatory data (e.g., climate,
topography, land use) are frequently compiled according to geographic units that are not coincident with
the drainage basin boundaries of river reaches. These data may be collected at different spatial scales and
according to spatial units that reflect political (e.g., counties) or other non-hydrologic features of the
landscape.
Figure B-2: Schematic of the Major SPARROW Model Components
Diffuse Sources
Industrial / Municipal
Point Sources
Landscape
Transport
SPARROW
Model Components
Aquatic Transport
Streams
Reservoirs
In-Stream Flux Prediction
Calibration minimizes differences
between predicted and calculated
mean-annual loads at the
monitoring stations
Water-Quality and
Flow Data
Periodic measurements
al monitoring stations
Monitoring
Station Flux
Estimation
Rating Curve Model
of Pollutant Flux
Station calibration to
monitoring dafa
Mean-Annual Pollutant
Flux Estimation
Evaluation of Model
Parameters and Predictions
Modified from Alexander, Elliott, et a/. (2002).
November 2008
B-9
-------
Environmental Assessment and Benefits Document for the C&D Regulation
Figure B-3: Location of 1,828 Water Quality Monitoring Stations Used in the SPARROW Sediment Model,
in Relation to the Reach File 1 (RF1) Reach Network
B.2.7 Use of Monitoring Data
The estimation of a SPARROW model requires estimates of long-term mean flux from a spatially
distributed set of monitoring stations within the study area, having sufficiently long periods of record. In
addition, it is necessary to include data from a diverse set of monitoring stations, inclusive of a wide range
of spatial scales and expressing considerable variation in predictor variable conditions. Often, monitoring
stations have different periods of record. Long-term variations in hydrologic conditions, combined with
long-term trends in water quality, imply mean fluxes computed over different periods may not be directly
comparable. If there are trends in water quality, the estimate of long-term mean water-quality flux will
depend on the period or window through which the water-quality data are acquired. Additionally, water
quality is rarely measured with the frequency necessary to directly estimate long-term mean flux, so that
indirect methods are needed to estimate flux from available measurements to account for hydrologic
conditions during periods when water-quality measurements are not made. The estimation procedure must
account for the fact that water-quality measurements are not always collected in a random manner,
implying the arithmetic average of flux is not a reliable estimate of mean flux. Rather, the measurements
of flux must be related in some way to other information, collected on a continuous basis, which exhibits
a close relation to flux, typically streamflow. The limitation, therefore, in constructing a mass-balance
model using long-term mean flux is that only water-quality measurements that can be associated with a
continuous record of streamflow will be suitable for model estimation.
Trends in water quality also create problems for the mass balance approach in relating water quality to its
predictor variables, principally the source variables. Ideally, the source variables will consist of a time
series of estimates, and these source variables would be included in SPARROW as long-term averages in
the same way as flux estimates. However, source information is rarely available for multiple periods.
Even if source information is available over time, the period over which it is compiled will rarely match
the period of the flux information; for example, point-source and land-use data are infrequently available
due to the high cost of compiling this information. To address the problem of incompatible periods of
record, a SPARROW model is typically specified for a single base year; with all water-quality and source
November 2008
B-10
-------
Environmental Assessment and Benefits Document for the C&D Regulation
information assumed to pertain to a given point in time. The severest constraint on the period of the
analysis is imposed by source information that is available only for a single year. Consequently, the base
year is generally selected to be in the middle of the range of available years for this 'one-time'
information.
B.2.8 Model Specification for Monitoring Station Flux Estimation
The extrapolation of infrequently sampled water-quality data and removal of trend require the
specification of a model of flux. The model must relate infrequently measured data to variables that are
measured continuously over time; and to accommodate de-trending, the model must include a function of
time. These requirements suggest a model that relates infrequently measured concentration to the
variables streamflow and time. The inclusion of streamflow as an explanatory factor serves another
purpose in the analysis. Because flux is typically positively related to streamflow, water-quality sampling
is commonly biased towards high flow events. The inclusion of flow in the water-quality model
effectively conditions the estimation of flux so as to remove the effects of high-flow sampling bias. The
estimation of mean flux by a station-specific model need not account for all processes affecting flux
within a basin; as this task is assigned to the SPARROW model. All that is required is that the estimated
mean flux at a station be reflective of long-term average processes within the basin. Station-specific flux
models need not be structurally accurate; they need only be predictively accurate. A causal explanation of
flux is ultimately obtained through application of a SPARROW model, whereby variations in mean flux
conditions across stations, estimated either explicitly or implicitly, are correlated with variations in basin
attributes across space.
Cohn et al. (1992a) have suggested a simple seven-parameter model in which the logarithm of
contaminant concentration (c) is related via a linear model to an intercept, the logarithm of flow, the
square of the logarithm of flow, decimal time T, decimal time squared, and a seasonal harmonic
consisting of two trigonometric terms - the sine and cosine of 2 times decimal time. Vecchia (2000) has
argued that there are important long-term lags affecting the relation between water quality and flow. He
suggests a specification that relates the log of contaminant concentration to a set of compound flow terms
consisting of moving averages of flows, of various lengths, in addition to time trend terms. A
specification that generalizes both the seven-parameter and the lag flow models takes the form
ct = M(Qt )j3Q + h(Tt )j3T + Xtj3x + et (B-l)
Where Qt is a/>-elenient row vector consisting of current and lagged logarithms of flow, M(Q) is a vector
function that transforms the />-element logged flows into a Kq element row vector; @Q is a Kq element
vector consisting of coefficients associated with the transformed flow terms; h(Tt) is a KT element row
vector function of decimal time; fir is a KT element vector of coefficients associated with the transformed
decimal time terms; Xt is a Kx element row vector of other exogenous variables affecting water quality; fix
is a KX element vector of coefficients associated with the other exogenous variables; and et is the normally
distributed error term, independent over time and uncorrelated with each of the predictor variables.
B.2.9 Tools for Flux Estimation
The water-quality models described above can be estimated using ordinary least squares, if the water-
quality data do not include any censored observations. For cases in observations are censored, Cohn et al.
(1992b) suggest estimation via adjusted maximum likelihood. The standard maximum likelihood method
for type I censored data (data for which the censoring threshold is known) is the Tobit model. The method
November 2008 B-11
-------
Environmental Assessment and Benefits Document for the C&D Regulation
of adjusted maximum likelihood combines the Tobit model with an adjustment to correct for first-order
bias in the coefficient estimates caused by estimation using a small sample. The method of adjusted
maximum likelihood is implemented within the USGS program Load Estimator (LOADEST) 2000
(Runkle et al., 2004). In addition to estimates of the parameters and their covariance matrix, the program
uses retransformation methods to produce unbiased estimates of daily and annual flux. A simple
averaging of the daily or annual estimates over all days or years yields an estimate of long-term mean
flux. The estimation of the model used to detrend flow requires a maximum likelihood method capable of
correcting for serial correlation in the errors. This capability is included in the PARMA model developed
by Vecchia (2000), and in standard statistical packages such as SAS (SAS, 1993). The more recently
developed program Fluxmaster (Schwarz, et al., 2006) includes methods to estimate the time-series flow
model using maximum likelihood, detrend flow, and estimate the water-quality model via adjusted
maximum likelihood. The program also computes unbiased, detrended estimates of long-term mean flux,
and provides an estimate of the associated standard error. The exact methods used in Fluxmaster differ
from those used in LOADEST 2000, but they are a close approximation, and identical if there are no
censored observations. Fluxmaster (Schwarz, et al., 2006) is used in the present study.
B.2.10 Guidance for Specifying Monitoring Station Flux Models
The ideal water-quality record has sufficient observations to reliably estimate the coefficients in the
model. Models that include both flow and time require a fairly long record, one that includes the base date
for detrending, and encompassing a wide range of hydrologic conditions. One consideration that places
limits on the length of the record is the need for flux estimates to be representative of base-year
conditions. This does not imply that the record must be representative of the hydrologic conditions in the
base year: the SPARROW model is estimated using mean estimates of water quality with the intention of
removing variations due to hydrologic conditions. Rather, long-term patterns in water quality should be
adequately captured by the model specification. This implies the period of record should not be so long as
to violate the assumption that water-quality trend is reasonably approximated by a simple linear function
of time. It is generally true that longer water-quality records display more complex patterns of trend. A
record that is too long runs the risk of misrepresenting the trend in water quality for any given year,
adding noise to the detrended flux estimate. For national analyses, Smith et al. (1997) and Alexander et al.
(2000) have based mean flux on water-quality data that span a 15 to 20 year period. Regional analyses by
Preston and Brakebill (1999) and Moore et al. (2004) have used water-quality data that span 20 to
25 years. Shorter periods of about six years were used by Alexander, Elliott, et al. (2002). Shorter records
may certainly be used to obtain an estimate of mean flux, although a shorter record will generally result in
a larger standard error in the mean flux estimate. For very short records, however, the specification of
trend becomes unreliable. It is recommended that trend terms be excluded from the model if record spans
less than three years.
One of the principal determinants of accuracy in flux estimation is the number of observations input to the
water-quality model. Generally, it is recommended that an estimate of flux depend on no fewer than
15 uncensored water-quality observations. However, it may be necessary to deviate from this threshold in
situations in which a site is subject to high variability in streamflow. The final consideration for inclusion
of data for a station in the SPARROW analysis is that the standard error of the mean flux estimate not be
too large. SPARROW has the capability of weighting observations according to their standard errors;
therefore wholesale exclusion of high-error flux estimates is unnecessary. If, however, a high-error flux
estimate enters the model as an upstream source for some downstream flux observation (the nested basin
arrangement) the potential for biasing estimates increases. High-error flux estimates also present
November 2008 B-12
-------
Environmental Assessment and Benefits Document for the C&D Regulation
problems in assessing the error of SPARROW predictions if the predictions are conditioned on observed,
upstream flux estimates. In national analyses, we have frequently excluded mean flux estimates that have
a standard error greater than about 20 to 30 percent.
B.2.11 Stream Network Topology
A vector- or raster-based digital representation of the stream and river network topology is the most
fundamental component of the spatial infrastructure that supports the SPARROW model (Figure B-4).
Vector representations are based on point and line (arc) GIS features, whereas raster representations are
based on a cellular (areal) structure. In either instance, a stream reach network explicitly defines surface-
water flow paths that spatially connect contaminant sources and landscape features with water quality
observations at downstream monitoring stations. In a vector-reach topology, a stream reach represents the
length of stream channel that extends from one tributary junction to another. Reach nodes are point
features that are associated with the location of tributary junctions, hence reach boundaries. Reach nodes
will also occur at the locations where reaches overlay with the shorelines of impoundments (reservoirs,
lakes), and with stream water-quality monitoring sites. In a raster representation of streams, nodes may
also define various intermediate locations along the stream reach between tributary junctions. Whether
digitally represented in vector or raster form, the reach topology must define either a set of reach nodes or
raster cells that are linked hydrologically to indicate the direction of water flow. The node topology must
be defined according to an upstream (from-) and downstream (to-) node attribute table (Figure B-4). This
tabulation of surface-water flow paths is required for the routing of water and contaminants through the
river network by SPARROW navigation software during estimation and application of the model. Thus,
the reach-node table defines the fundamental data infrastructure of the model.
Figure B-4: Schematic Illustrating a Vector Stream Reach Network with Node Topology and
Water/Contaminant Reach-Node Routing Table
The reach-type indicator has possible values of "0" (Stream Reach), "1 "(Impoundment Reach), and "2"
Outlet Reach for Impoundment
12
13
.13
12
(7
1
Reach-Node Attribute Table
Reach Up- Down- Diversion Reach-Type
Number stream stream Fraction Indicator
Node Node
Reservoir
"shorelines
Divergent
Reach
Reach Node
'Stream Reach
13
10
12
11
9
8
7
5
6
4
2
3
1
13
10
12
11
9
8
7
6
6
5
4
3
2
10
9
11
9
8
4
6
5 (
5 (
4
2
2
1
.0
.0
.0
.0
.0
.0
.0
).7
13
.0
.0
.0
.0
0
1
0
1
2
0
0
0
0
0
0
0
0
The SPARROW model structure supports distributary and diversion reaches in the stream network;
including braided channels or reaches where water is diverted to canals or other water bodies. The
November 2008
B-13
-------
Environmental Assessment and Benefits Document for the C&D Regulation
SPARROW model assumes that contaminants are diverted in proportion to volume of flow. Therefore, an
estimate is required for each reach of the diversion fraction - a measure of the fraction of streamflow that
is diverted in distributary reaches. A diversion fraction of 1.0 is assigned to reaches without water
diversions. In the example distributary reaches shown in Figure B-4, the fraction of the flow in reach 7
that is diverted to the two downstream reaches is 0.7 for reach 5 and 0.3 for reach 6. The SPARROW
reach topology also supports the separate designation of impoundments (e.g., lakes, reservoirs) associated
with stream reaches. This designation is used in SPARROW to separately estimate the sediment and
contaminant attenuation in reservoirs and lakes. In Figure B-4, reaches 9-11 are associated with a
reservoir. A "reach-type" indicator is used to identify reaches associated with impoundments separately
from conventional stream reaches. The outlet reach of an impoundment is coded separately from other
interior impoundment reaches to facilitate the pollutant attenuation calculations.
In addition to the reach properties listed in Figure B-4, the reach length and an estimate of the mean-
annual streamflow of the reach is also required. Estimates of mean water velocity are required to estimate
in-stream contaminant attenuation as a function of the water time of travel (alternatively, in-stream
attenuation can be estimated as a function the reach length). Measures of the areal water load are needed
for impoundments for use in estimating the contaminant attenuation in these water bodies; the areal water
load is computed as the quotient of the outflow to surface area of the water body (see Alexander et al.,
2002a). Digital representations of the drainage basin boundaries associated with river reaches are also
needed to estimate the incremental and total drainage area of the reaches and to support the digital overlay
of drainage boundaries with polygonal boundaries (vector or raster) that define the locations of
contaminant sources (point and diffuse) and various landscape properties.
The assessment of pollutant loadings to coastal estuaries requires an expanded reach network that
includes shoreline features and the identification of reaches that terminate at estuaries. The node points of
shoreline reaches include the downstream nodes of the terminating reaches. Shoreline reaches are used to
define coastal drainage areas - areas that discharge runoff directly to the estuary without transport
through a stream reach. Because the discharge for a shoreline reach does not accumulate from any
upstream location, the diversion fraction for these features is set to zero. Shoreline reaches also have no
stream attenuation so travel time is set to zero.
Spatial referencing is a critical step in SPARROW modeling that is necessary to construct the watershed
attribute data used as explanatory variables and to verify the accuracy of the hydrologic connectivity of
the stream reaches, which governs routing and accumulation of mass in the model. Spatial referencing
entails the use of GIS techniques to digitally establish the geographic relation between stream reaches in
the river network and the various watershed attributes that are used to specify the model. Watershed
attribute data sets (e.g., topography, land use, climate) are frequently compiled and reported according to
geographic units that are not coincident with the drainage basin boundaries of river reaches; for example
at finer spatial scales or according to spatial units that reflect cultural (e.g., counties, states) or other non-
hydrologic features of the landscape. Watershed attributes may also be geographically located according
to precise locations coded as latitude and longitude, which must be digitally linked to a watershed or
nearby river reach. These may include the locations of stream monitoring gages or municipal wastewater
treatment outfalls. Verification of automated GIS operations is often required to ensure accurate spatial
referencing; including, for example, comparisons of the river name of a gage with that of a river reach or
the comparison of the reported drainage area of a gage with that estimated for a river reach. Finally, the
SPARROW node and routing architecture also can fully support the modeling of contaminant transport
along "off-reach" (landscape) flow paths according to flow directions defined by landscape topography as
reflected, for example, in DEMs, although this capability is not currently implemented in the sediment
November 2008 B-14
-------
Environmental Assessment and Benefits Document for the C&D Regulation
model. Such architecture facilitates the incorporation of high-resolution spatial data sets that delineate
sources and other explanatory variables at scales finer than an incremental drainage area for the reach.
B.2.12 Watershed Sources and Explanatory Variables
The explanatory variables evaluated in SPARROW models reflect current knowledge of natural and
human-related sources and the important physical, chemical, or biological properties of the terrestrial and
aquatic ecosystems that affect supply and transport of contaminants in watersheds, along with practical
considerations of availability. Point- and diffuse-source variables may include direct measures of the
introduction or supply of contaminant mass to the landscape and streams and reservoirs (e.g., municipal
and industrial wastewater discharge, fertilizer application). Alternatively, source variables may serve as
surrogate indicators of the contaminant mass supplied by point and diffuse sources in watersheds, such as
land-use/land-cover data or census data on human and livestock populations. Watershed data on sources
and other properties (e.g., climate, topography, land-use) must be spatially referenced to the drainage
basins and stream reaches of the SPARROW river network. Figure B-5 shows the relation between areas
of source polygons and the incremental drainage basin of a hypothetical stream reach. Estimates of the
diffuse sources associated with the drainage basins of stream reaches can be obtained using GIS
operations to digitally overlay the drainage basin boundaries of stream reaches with the polygonal areas
associated with the diffuse source data. Once quantitative measures of the overlap in watershed areas are
obtained (Figure B-5), estimates of the area-weighted sum of source characteristics (S ) for reach/ and
source type n can be calculated as:
(B-2)
where P(j) is the set of all source-related polygons that intersect the incremental drainage polygon for
stream reach j, Sn>k is the quantity of source-type n associated with polygon k, Ajik is the sub-area of reach
/s incremental drainage that intersects the source-related polygon k, and A k is the total area of the source-
related polygon k.
Figure B-5: Schematic Illustrating Digital Overlay of Stream Reach Drainage Area and
Polygonal Areas Associated with Diffuse Sources
Area//c=3
Stream Reachy
Area,A'=2
Reach Drainage
Boundary
Area/r=f
Diffuse Source
Polygon k
If it is known that a particular contaminant source is associated with a specific land use (or some grouping
of land uses) - for example, fertilizer is associated with cultivated land - and if the spatial scale of land-
use information is similar to the scale at which watersheds are delineated, then the method described
above can be modified to obtain a more refined estimate of sources within a reach incremental drainage
area. The enhanced method requires only that the area term Ajik in Equation B-2 be redefined to represent
November 2008
B-15
-------
Environmental Assessment and Benefits Document for the C&D Regulation
the area of the associated land use that intersects the reach j incremental drainage and source polygon k,
and that the area term A "k be redefined to equal the total area of the associated land use within the source
polygon k.
Climatic and landscape properties that affect contaminant transport may include measures of water-
balance terms (e.g., solar radiation, precipitation, evaporation, evapotranspiration), soil characteristics
(e.g., permeability, moisture content), water-flow path properties (e.g., slope, hydraulic roughness,
topographic index), or management practices and activities (e.g., tile drains, conservation tillage, BMPs).
Estimates of various climatic and landscape characteristics or properties are often calculated as an area-
weighted mean estimate (Z!>;) for stream reach j and landscape property /' according to:
(B-3)
where P(j) is the set of all land-characteristics polygons that intersect the incremental drainage polygon
for stream reach j, Zik is the landscape property /' associated with polygon k, Ajik is the sub-area of reach
fs incremental drainage that intersects the landscape property's polygon k, and A ; is the total drainage
area of the reach watershed/. Figure B-6 shows the relation between areas of landscape polygons and the
incremental drainage basin of a hypothetical stream reach that is described in Equation B-3. Note that the
area-weighted mean estimate defined in Equation B-3 differs from the area-weighted sum given by
Equation B-2 in that the area ratio terms sum to 1.0 in Equation B-3 but not in Equation B-2.
Figure B-6: Schematic Illustrating Digital Overlay of Stream Reach Drainage Area and
Polygonal Areas Associated with Landscape Properties
Stream
Area
Are
Reach Drainage
Boundary
Landscape
k
The basic SPARROW models described are designed to model long-term mean contaminant loadings in
streams, implying explanatory variables in the model should be computed to reflect long-term conditions.
Variables that describe contaminant sources and landscape characteristics may be averaged over multiple
years, corresponding to the available period of record for water-quality monitoring data, or may reflect
the conditions during a specified base year used to estimate stream water-quality loadings, as in the
sediment study.
B.3 Model Specificatior
The specification of a SPARROW model consists of identifying the explanatory variables and functional
forms for the associated processes the model is to include. The process is described here in generic terms,
supplemented by specific descriptions of the specification of the SPARROW sediment model.
November 2008
B-16
-------
Environmental Assessment and Benefits Document for the C&D Regulation
B.3.1 Model Equation and Specification of Terms
Conceptually, the contaminant load or flux leaving a reach is the sum of two components. The first
component is the load generated within upstream reaches that is delivered to the reach via the stream
network. Losses of flux from the stream network may occur at points where flow is diverted.
Additionally, in moving through the reach, flux will generally be attenuated by stream or reservoir
processes. The second component consists of source flux that is generated within the reach's incremental
watershed and delivered to the stream network somewhere along the reach segment. A number of source-
dependent processes, in addition to stream attenuation processes, affect the amount of source flux
reaching the stream network and transported to the reach's downstream outlet node. For flux originating
on the landscape, the processes affecting delivery to the stream network are called land-to-water delivery
processes, and may include both surface and sub-surface elements. A conceptual illustration of the
pertinent spatial relations is given in Figure B-7. A connecting reach is generally defined as a stream
segment that connects the confluence of two stream segments; a headwater reach is a reach that is defined
without an upstream confluence. Figure B-7 shows five complete reaches, two of which are headwater
reaches, with one of the headwater reaches classified as a reservoir. Each reach is embedded within a
Figure B-7: Conceptual Illustration of a Reach Network for Five Incremental Watersheds. Model Equation B-4
Describes the Supply and Transport of Load within an Individual Reach and its Incremental Watershed.
Upstream
monitoring
station, Y
Stream
reach
segment
Downstream
monitoring
station, X
Reservoir
Reach
contributing
area
Point source
Source: McMahon et a/., (2003)
color-coded area representing the reach's incremental drainage - the area that drains directly to the reach
without passing through another reach. Because there are five reaches, there are five incremental drainage
areas. Land-to-water delivery processes determine the amount of contaminant generated within an
incremental drainage area, excluding contaminant generated directly on a reach, which is then delivered
to the area's corresponding reach. In the figure, two monitoring stations, X and Y, form the boundaries of
a nested basin, defined as all reaches above a monitored reach /', containing monitoring station X, but
exclusive of reaches above and including all upstream monitoring stations - in this case the single
monitoring station Y. In-stream attenuation processes associated with headwater reaches affect only the
transport of contaminants from their incremental drainage; in-stream attenuation processes for all non-
headwater reaches affect the transport of contaminants from their own incremental drainage and also from
the incremental drainage of all upstream reaches (including reaches in upstream nested basins).
This conceptual model can be formalized through a mathematical equation. Let F*t be the model-
estimated flux for contaminant leaving reach /'. This flux is related to the flux leaving adjacent reaches
November 2008
B-17
-------
Environmental Assessment and Benefits Document for the C&D Regulation
upstream of reach /', denoted by Fj where/ indexes the set J(i) of adjacent reaches upstream of reach /',
plus additional flux that is generated within the incremental reach segment /'. In most cases, the set of
adjacent upstream reaches J(i) will consist of either two reaches, if reach / is the result of a confluence, or
no reaches if reach / is a headwater reach (see Figure 5-7). The functional relations determining reach /
flux are given by:
F,=
(B-4)
The first summation term represents the amount of flux that leaves upstream reaches and is delivered
downstream to reach /', where F} equals measured flux, F}M, if upstream reach 7 is monitored or, if it is
not, is given by the model-estimated flux F*,-. St is the fraction of upstream flux delivered to reach /'. If
there are no diversions, then $ is set to 1. Otherwise, this fraction is defined by the fraction of streamflow
leaving upstream reaches that is delivered to reach /'. A(.) is the stream delivery function representing
attenuation processes acting on flux as it travels along the reach pathway. This function defines the
fraction of flux entering reach /' at the upstream node that is delivered to the reach's downstream node.
The factor is a function of measured stream and reservoir characteristics, denoted by the vectors 7? and
2s, with corresponding coefficient vectors 9$ and OR. If reach /' is a stream, then only the Z? and 9S terms
determine the value of A(); conversely, if reach /' is a reservoir then the terms that determine A(.) consist
ofZ* and OR,
The second summation term represents the amount of flux introduced to the stream network at reach /'.
This term is composed of the flux originating in specific sources, indexed by n = 1,... JVS. Associated with
each source is a source variable, denoted Sn. Depending on the nature of the source, this variable could
represent the mass of the source available for transport to streams, or the area of a particular land use. The
variable an is a source-specific coefficient. This coefficient retains the units that convert source variable
units to flux units. The function /)„(.) represents the land-to-water delivery factor. For sources associated
with the landscape, this function along with the source-specific coefficient determines the amount of
contaminant delivered to streams. The land-to-water delivery factor is a source-specific function of a
vector of delivery variables, denoted by Z,D, and an associated vector of coefficients 6D. For point sources
that are described by a measured (same units as flux) discharge of mass directly to the stream channel
-i
(e.g., municipal wastewater effluent measured in kilograms year ), the delivery factor should be 1.0, with
no underlying factors acting as determinants, and the source-specific coefficient should be close to 1.0.
The last term in the equation, the function A '(.), represents the fraction of flux originating in and delivered
to reach / that is transported to the reach's downstream node. This function is similar in form to the stream
delivery factor defined previously; however, the default assumption in SPARROW models is that if reach
/ is classified as a stream (as opposed to a reservoir reach), the contaminants introduced to the reach from
its incremental drainage area receive the square root of the reach's full in-stream delivery. This
assumption is consistent with the notion that contaminants are introduced to the reach network at the
midpoint of reach / and thus experience only half of the reach's time of travel. For reaches classified as
reservoirs, the default assumption is that the contaminant receives the full attenuation defined for the
reach.
The nonlinear model structure in Equation B-4 contains several key features. The additive contaminant
source components and multiplicative land and water transport terms are conceptually consistent with the
physical mechanisms that explain the supply and movement of contaminants in watersheds. Total
modeled flux for a reach is shown to be decomposed into its individual sources. Because this same
November 2008 B-18
-------
Environmental Assessment and Benefits Document for the C&D Regulation
decomposition is done for all upstream reaches, it is possible in this framework to perform flux
accounting, whereby total flux is attributed to its source components. All processes are spatially
referenced with respect to the stream network according to the reach in which they operate. This means,
for example, that a reservoir at reach / affects the transport of all contaminants entering the reach network
upstream (but not downstream) of reach /'. The additive source components also provide a mathematical
structure in the model that preserves mass. This can be seen by noting that a doubling of each of the
source variables S^, along with a doubling of all upstream sources, as represented by a doubling of F'j
results in an exact doubling of modeled flux F*t. Finally, the modeled flux at any reach / is conditioned on
monitored fluxes entering the stream network anywhere upstream of reach /'. This approach to nested
basins serves to isolate errors introduced in any upstream basin from incremental errors that arise in a
downstream basin, making it defensible to treat nested basins as independent observations.
B.3.2 Contaminant Sources
The selection of potential contaminant source variables in SPARROW models initially depends upon a
user's particular knowledge of a watershed as well as inferences that can be derived from the research
literature about the major sources that contribute pollutants to watersheds Knowledge of the geography of
these sources, based on direct measures or surrogate indicators of the contaminant mass supplied to the
land surface and surface waters, is also critical to provide an effective use of the spatially distributed
structure in SPARROW. The compilation and availability of large digital spatial data sets has become
much more common (e.g., Brakebill and Preston, 1999) and the advances in many types of digital
topographic and stream network data [e.g., National Hydrography Dataset (NHD), National Land Cover
Database (NLCD)] have made spatially distributed modeling more feasible. The inclusion of detailed
information in the SPARROW model on the geographic locations where contaminants are released in
watersheds has particular relevance to the potential policy and management applications of the model.
The source terms used in the model can be generally classified as intensive and extensive measures of
contaminant mass. The former measures are typically descriptive of direct measures of pollutant mass,
such as fertilizer application, livestock waste, atmospheric deposition, or sewage-effluent loadings. In
these cases, the source-specific parameter (a) is expressed as a dimensionless coefficient that, together
with standardized expressions of the land-to-water delivery factor, describes the proportion or fraction of
the source input that is delivered to streams. This fraction would be expected to be less than 1.0 but
greater than zero, reflecting the removal of contaminants in soils and ground water.
Extensive measures of contaminant mass also may be used in SPARROW models. These are surrogate
indicators of contaminant mass and include measures of watershed properties such as specific land-use
area and sewered population that are considered to be proportional to the actual mass loadings generated
by a general type of a contaminant source. The empirical estimates of the source coefficients in the model
provide a quantitative measure of the proportion of the mass loading that is associated with a specified
source. If extensive measures are used, the associated model coefficients are expressed as the contaminant
mass generated per unit of the source type (e.g., kilograms kilometer"2 year"1; kilograms person"1 year"1). If
combined with the land-to-water delivery factor and expressed as a standardized source coefficient, the
coefficient indicates the mean quantity of contaminant mass per unit of the surrogate source measure that
is delivered to streams. For land-use terms, the standardized coefficient gives what is frequently cited as
an export coefficient (Johnes et al., 1996; Beaulac and Reckhow, 1982). Observed values of export
coefficients have been reported in the literature for various land-use/land-cover types, such as crop, urban,
and forested lands (Beaulac and Reckhow, 1982); and often compare favorably with corresponding
estimates by SPARROW (e.g., Alexander et al., 2004). Other sources of guidance include sediment
November 2008 B-19
-------
Environmental Assessment and Benefits Document for the C&D Regulation
erosion estimates generated from the Universal Soil Loss Equation (USLE) as part of the
U.S. Department of Agriculture (USDA) National Resources Inventory (NRI; Natural Resources
Conservation Service, 2005)
B.3.3 Landscape Variables
Landscape variables in SPARROW describe properties of the landscape that relate to climatic, natural- or
human-related terrestrial processes affecting contaminant transport. These include properties that are
assumed on some conceptual or empirical basis to control the rates of contaminant processing and
transport, and which are widely available. The model structure allows tests of hypotheses about the
influence of specific features of the landscape on contaminant transport. Landscape variables may include
water-balance terms (e.g., precipitation, evapotranspiration) related to climate and vegetation, soil
properties (e.g., organic content, permeability, moisture content), topographic water flow-paths variables
(e.g., TOPMODEL overland flow, topographic index, and slope) (Beven and Kirkby, 1979), or
management practices and activities, including tile drainage, conservation tillage practices, and BMPs
related to stream riparian properties. Particular types of land-use classes, such as wetlands or impervious
cover, may also be potentially used to describe transport properties of the landscape.
The land-to-water delivery factor in Equation B-4, Z)n(Z!D;^D), is a source-specific function of a vector of
delivery variables, denoted by Z®, and an associated vector of coefficients 0D. In basic SPARROW
models, the land-to-water delivery factor has been expressed in exponential functional form. For source n,
the fraction of contaminant mass generated in the incremental reach drainage area and delivered to the
reach (excluding source coefficient term) is estimated as
/ \ {M° ^
D (ZD;0n)=exp\yct} Z D9n \ (B-5)
n \ i ' D / r \ / j nm m,i Dm v /
where Zm® represents delivery variable m for the incremental drainage of reach /', 6Dm is its corresponding
coefficient, conm is an indicator variable that is 1.0 if delivery variable m affects source n and zero
otherwise, andMD is the number of delivery variables. Log-transformed delivery variables, such as the
logarithm of soil permeability, may provide an improved fit to the data. Under the log-transformation, the
land-to-water coefficient is interpreted as the percent change in flux delivered to streams, derived from all
sources to which the land-to-water variable is applied, from a one-percent increase in the land-to-water
delivery variable. The full effect of landscape variables on the delivery of source n to streams is
determined by the product of the delivery factor in Equation B-5 and the source coefficient otn. Therefore,
modification of delivery factor specification can be expected to change the mean of the delivery factor,
resulting in a corresponding counter-adjustment of the source coefficient otn. In order to improve the
interpretability of the source coefficient, and to provide some stability in its value across alternative
specifications of the land-to-water delivery factor, it is recommended that the delivery variables Zmf in
Equation B-5 be expressed as differences from their mean value over all reaches.
B.3.4 Stream Transport
Stream attenuation processes that act on contaminant flux as it travels along stream reaches are frequently
modeled as first-order reaction rate processes (Chapra, 1997). A first-order decay process implies that the
rate of removal of the contaminant from the water column per unit of time is proportional to the
concentration or mass that is present in a given volume of water (a zero-order process corresponds to a
constant rate of removal per unit of time). In a first-order decay process, the fraction of contaminant
November 2008 B-20
-------
Environmental Assessment and Benefits Document for the C&D Regulation
removed over a given stream distance is estimated as an exponential function of a first-order reaction rate
coefficient (in reciprocal time units) and the cumulative water time of travel over this distance. A reaction
rate is estimated on a volumetric basis, and therefore is expected to depend on properties of the water
column that are proportional to water volume, such as streamflow and water-column depth (Stream Solute
Workshop, 1990). Accordingly, in basic forms of the SPARROW model, the fraction of the contaminant
mass originating from the upstream node and transported along reach / to its downstream node is
estimated as a function of the mean water time of travel (T7/,; units of time) in reach / and stream class c
defined according to discrete intervals of mean streamflow or depth (in this case, Zf = {Tcst}, c= 1,. . . ,CS,
where T/t is nonzero only for the streamflow class corresponding to reach /'), and a stream-size dependent
loss rate coefficient (0Sc; units of time) such that:
Mean water time of travel is estimated as the quotient of the reach channel length and mean water
velocity. The most accurate estimates of water velocity are obtained from time-of-travel dye studies
(Jobson, 1996), but in some cases may be obtained from instantaneous measurements of water velocity
taken during flow gage site visits. Empirical geomorphic relations, which use regression methods to relate
time-of-travel measurements to channel and basin properties (e.g., streamflow, slope), also are available
for regions of the U.S. and other countries (e.g., USEPA, 1996b; Alexander et al., 1999; Jobson, 1996;
Jowett, 1998) and can be used to estimate the time-of-travel of stream reaches for a given river network.
Alternatively, channel length may be used in Equation B-6 where water velocity estimates are not
available. This assumes that the water time of travel is proportional to the channel length, and estimates of
the removal rate are expressed as reciprocal length (e.g., see McMahon et al., 2003; Alexander et al.,
2002a).
B.3.5 Reservoir and Lake Transport
Attenuation processes that act on contaminant mass as it travels through a lake or reservoir are often
modeled as a net removal process, with the loss coefficient expressed as either a first-order reaction rate
or a mass-transfer coefficient (also referred to as an apparent settling velocity) (Chapra, 1997). Both of
these loss expressions have been used in empirical mass-balance lake models for phosphorus (e.g.,
Vollenweider, 1976; Reckhow and Chapra, 1983) and nitrogen (e.g., Kelly et al., 1987; Molot and Dillon,
1993), although the use of the mass-transfer rate is more common. These mass-balance models typically
assume steady-state and uniformly mixed conditions in the waterbody. The reaction rate is expressed in
reciprocal time units and its estimation and use are dependent on knowledge of the water depth and
surface area of the waterbody (volume). The apparent settling velocity is expressed in units of length per
time and is estimated as a function of the ratio of the reservoir outflow and the surface area of the
reservoir sediments (assumed equal to the surface area of the waterbody); this ratio, denoted qR for reach
/', is termed the areal hydraulic load and is a measure of the water displacement or velocity in the
reservoir. The term "apparent" indicates that the settling velocity measures the net effect of various
processes that remove the contaminant from the water column and deliver it to the sediments, and
processes that may add contaminant to the water column. The function that relates the areal hydraulic
load to the fraction of flux attenuated in a reservoir is
Xz?,zf;es,efi)= - - (B-T)
November 2008 B-21
-------
Environmental Assessment and Benefits Document for the C&D Regulation
where ORO is a parameter, to be estimated in the model, representing the apparent settling velocity.
B.4 Model Estimation
The SPARROW model equation, given in Equation B-4, is a nonlinear function of its parameters, and the
model must be estimated using nonlinear techniques. The model errors are assumed to be independent
across observations with zero mean, while the variance of each observation may be observation specific.
SPARROW utilizes a nonlinear weighted least squares (NWLS) estimation method, which does not
assume the precise distribution of the residuals. Several algorithms exist for obtaining NWLS estimates,
and the Levenberg-Marquardt Least-Squares method, implemented in SAS, is used to estimate the
SPARROW model (SAS, 1999). Unlike linear models, the statistical properties of the estimated
parameters for nonlinear models are not precisely known in finite samples. For this reason, much of the
theory of nonlinear estimation has focused on characterizing asymptotic properties - the statistical
properties of the coefficient estimates as the sample size goes to infinity. Emphasis in the following
sections is on interpretation of estimated model parameters.
B.4.1 Evaluation of Model Parameters
The objective of parameter evaluation in SPARROW modeling is to determine whether a converged
model provides statistically sound and physically interpretable coefficient values. The first objective
entails the appraisal of model parameters for statistical significance and the quantification of uncertainty.
This provides important information for identifying unique model specifications (parameters and values
for which the model predictions are sensitive) and determining the level of model complexity (number
and types of explanatory variables and model functions) that can be empirically supported by the stream
monitoring data. The emphasis on parameter estimation in SPARROW models reflects the objective of
identifying the important contaminant sources and factors affecting mean-annual contaminant transport
over large spatial scales in soils and in ground and surface waters. The key parameter statistics include the
estimated mean coefficient values, estimated variance of these coefficient estimators, and measures of
statistical significance based on the t statistics (ratio of the coefficient value to its standard error). These
statistics are biased in finite samples but consistent as sample size goes to infinity; the t statistics are
asymptotically distributed as standard normal. The />-values are based on a two-tailed probability from a
Student's t distribution. The/>-values can be used to identify statistically significant model coefficients -
those that are statistically distinguishable from zero - and can be used to refine the parameter set to
identify parsimonious SPARROW models.
The second complementary objective is the evaluation of the parameters for physical interpretability. This
objective entails the evaluation of the sign and magnitude of model coefficients to test hypotheses about
the importance of different contaminant sources and the hydrologic and biogeochemical processes that are
represented by the explanatory variables of the model. SPARROW model parameters reflect the net
effects over large spatial scales of an aggregate set of hydrologic and biogeochemical processes and
human-related activities. The interpretability of the parameters and their relation to specific processes is
enhanced in SPARROW by the use of a mass balance, mechanistic structure that explicitly separates the
terrestrial and aquatic properties of watersheds and accounts for nonlinear interactions among watershed
properties, together with an emphasis on the statistical estimation of parameter values. The sign of
SPARROW model coefficients can be evaluated to determine the direction of influence of any
explanatory variable on in-stream estimates of mean-annual flux. The direction of influence is assessed
for consistency with the anticipated response based on available theoretical or empirical information
November 2008 B-22
-------
Environmental Assessment and Benefits Document for the C&D Regulation
about processes related to individual explanatory factors. For example, a negative sign on the soil
permeability coefficient indicates that total loads in streams are inversely related to permeability - i.e., in-
stream loads of contaminants are generally lower in watersheds with highly permeable soils. The sign of
the coefficient is also important in estimating physically meaningful contaminant source terms in
SPARROW, which are generally expected to contribute positive contaminant mass to the watershed
system.
Coefficients associated with source inputs expressed in areal units describe the mass per unit area
delivered to streams from these land areas. These areal expressions of contaminant transport (export) can
be directly compared with ranges of export coefficients reported in the literature (e.g., Beaulac and
Reckhow, 1982). Estimated SPARROW coefficients associated with specific land uses generally compare
favorably with published export coefficients. Other source coefficients that are expressed in
dimensionless units provide a measure of the fraction of the contaminant that is delivered from each
source to streams, rivers, and reservoirs. These coefficients can be evaluated to determine how reasonably
they reflect the net mean rates of contaminant removal by a source as part of the delivery to aquatic
systems.
B.5 Model Prediction
SPARROW output contains prediction results paired with measures of accuracy. A number of technical
issues can arise in the derivation of these statistics; most of these caused by the nonlinear nature of the
SPARROW model. The prediction equation is similar to the calibration Equation B-4. Because
predictions are generated for specific sources, however, it is necessary to decompose flux into source-
specific components. Let F*in denote the reach / model-estimated flux associated with source n, and let
J:H be the source-w flux from upstream reach/. If reach7 is monitored (that is,/ £/, where / is the set of
monitored reaches), and predictions are conditioned on measured flux, then F jiH is apportioned from the
measured flux F}M according to
(B-8)
Otherwise, if reach7 is not a monitored reach (that is,/ £7) or predictions are not conditioned on
measurements, then F,-,n is set equal to F itn. The equation defining the model-estimated flux for source n,
F*in, is given by
F',., =
(B-9)
November 2008 B-23
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Appendix C - A Preliminary SPARROW Model of Suspended Sediment
for the Conterminous United States (Schwarz 2008a)
This appendix reprints in full the USGS Open File Report detailing the specific application of the
SPARROW model to predict sediment concentrations in the RF1 stream network.
November 2008
C-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Appendix D - Water Quality Index Tables
This appendix presents detailed versions of Table 10-5 through Table 10-7, showing WQI river mile and
percentage improvements by baseline water quality index range, improvement range, and EPA region for
each policy option.
November 2008
D-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-1: Estimated Water Quality Improvements Under Option 1
EPA
Region
1
2
3
4
5
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
30
916
12,343
3,663
16,952
52
2,394
4,824
8,617
15,887
1,371
13,438
12,695
1,679
29,183
1,053
37,102
34,667
14,490
87,312
6,402
44,892
17,591
2,624
71,509
Miles of
River in
RFI
Network
30
1,130
12,558
4,575
18,293
52
2,579
4,846
8,633
16,110
1,371
15,450
13,321
3,475
33,617
1,065
37,828
37,287
18,284
94,464
6,402
44,929
17,596
2,624
71,550
Water Quality Improvements
0.01 < AWQI < 0.1
River
Miles
0
0
0
0
0
0
20
0
0
20
0
0
0
0
0
0
1,793
4,102
2,313
8,207
0
46
51
0
96
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
4.8%
11.8%
16.0%
9.4%
0.0%
0.1%
0.3%
0.0%
0.1%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.1%
0.0%
00%
0.0%
0.0%
0.0%
0.0%
4.7%
11.0%
12.6%
8.7%
0.0%
0.1%
0.3%
0.0%
0.1%
0.
River
Miles
0
o
0
o
0
0
0
0
0
0
0
0
o
0
0
0
356
214
168
738
0
0
0
0
0
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
1.0%
0.6%
1.2%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.9%
0.6%
0.9%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
by WQI Change
0.5 < AWQI
River
Miles
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
123
31
232
386
0
0
0
0
0
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0^0%
0.0%
0.0%
0.0%
0.0%
0.0%
0^0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.3%
0.1%
1.6%
0.4%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.3%
0.1%
1.3%
0.4%
0.0%
0.0%
0.0%
0.0%
0.0%
Total Improved Reaches
River
Miles
0
0
0
0
0
0
20
0
0
20
0
0
0
0
0
0
2,272
4,347
2,712
9,331
0
46
51
0
96
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
6.1%
12.5%
18.7%
10.7%
0.0%
0.1%
0.3%
0.0%
0.1%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.8%
0.0%
0.0%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
6.0%
11.7%
14.8%
9.9%
0.0%
0.1%
0.3%
0.0%
0.1%
November 2008
D-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-1: Estimated Water Quality Improvements Under Option 1
EPA
Region
6
7
8
9
10
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
2,111
60,261
16,527
3,957
82,856
15,139
35,697
2,965
217
54,018
5,741
62,368
28,974
16,627
113,710
9,345
35,035
2,833
261
47,474
50
13,754
20,140
32,525
66,469
Miles of
River in
RFI
Network
2,332
73,189
17,718
5,430
98,669
16,103
41,624
2,965
217
60,909
8,426
76,218
29,040
16,627
130,311
13,040
38,566
3,468
1,185
56,259
50
13,862
20,533
35,050
69,495
Water Quality Improvements
0.01 < AWQI < 0.1
River
Miles
0
0
0
0
0
0
0
0
0
0
0
4
28
0
32
0
0
0
0
0
0
0
0
0
0
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
<0.0%
0.1%
0.0%
<0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
<0.0%
0.1%
0.0%
<0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.
River
Miles
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
by WQI Change
0.5 < AWQI
River
Miles
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Total Improved Reaches
River
Miles
0
0
0
0
0
0
0
0
0
0
0
4
28
0
32
0
0
0
0
0
0
0
0
0
0
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.1%
0.0%
<0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
<0.0%
0.1%
0.0%
<0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
November 2008
D-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-1: Estimated Water Quality Improvements Under Option 1
EPA
Region
•
Total
Baseline
Water
Quality
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
41,294
305,857
153,559
84,660
585,370
Miles of
River in
RFI
Network
48,871
345,376
159,331
96,099
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
0 0.0% 0.0%
1,862 r 0.6% r 0.5%
4,181 ; 2.7% ; 2.6%
2,313 2.7% 2.4%
8,356 | 1.4% | 1.3%
0.1 < AWQI < 0.5
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
0 0.0% 0.0%
356 ' 0.1% 0.1%
214 ; 0.1% ; 0.1%
168 0.2% 0.2%
738 | 0.1% | 0.1%
0.5 < AWQI
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
0 0.0% 0.0%
123 r 0.0% ' 0.0%
31 ; 0.0% ; 0.0%
232 0.3% 0.2%
386 | 0.1% | 0.1%
Total Improved Reaches
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
0 0.0% 0.0%
2,341 ' 0.8% : 0.7%
4,426 ; 2.9% ; 2.8%
2,712 3.2% 2.8%
9,479 | 1.6% | 1.5%
November 2008
D-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-2: Estimated Water Quality Improvements Under Option 2
EPA
Region
1
2
3
4
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
30
916
12,343
3,663
16,952
52
2,394
4,824
8,617
15,887
1,371
i"3,438
12,695
1,679
29,183
1,053
37,102
34,667
14,490
87,312
Miles of
River in
RFI
Network
30
1,130
12,558
4,575
18,293
52
2,579
4,846
8,633
16,110
1,371
15,450
13,321
3,475
33,617
1,065
37,828
37,287
18,284
94,464
Water Quality Improvements by WQI Change
0.01 < AWQI <
River
Miles
0
59
794
658
1,510
0
698
2,210
2,437
5,345
283
7,438
7,517
1,006
16,244
57
8,247
15,799
6,066
30,169
%of
River
Miles
Receiving
C&D
Discharges
0.0%
6.4%
6.4%
18.0%
8.9%
0.0%
29.2%
45.8%
28.3%
33.6%
20.6%
55.3%
59.2%
59.9%
55.7%
5.4%
22.2%
45.6%
41.9%
34.6%
0.1
%of
Total
River
Miles
0.0%
5.2%
6.3%
14.4%
8.3%
0.0%
27.1%
45.6%
28.2%
33.2%
20.6%
48.1%
56.4%
29.0%
48.3%
5.3%
21.8%
42.4%
33.2%
31.9%
0.
River
Miles
0
0
32
6
38
0
30
78
88
196
65
774
1,390
126
2,356
0
2,840
7,201
3,202
13,243
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
03%
0.2%
0.2%
0.0%
1.2%
L6%
L0%
1.2%
4.8%
5^8%
11.0%
7.5%
8.1%
0.0%
7/7%
20.8%
22.1%,
15.2%
%of
Total
River
Miles
0.0%
0.0%
0.3%
0.1%
0.2%
0.0%
1.2%
1.6%
1.0%
1.2%
4.8%
5.0%
10.4%
3.6%
7.0%
0.0%
7.5%
19.3%
17.5%
14.0%
0.5 < AWQI
River
Miles
0
o
0
0
0
0
7
0
0
7
11
116
117
0
244
0
1,506
2,412
899
4,817
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0^0%
0^0%
0.0%
0.0%
03%
0^0%
0^0%
0.0%
0.8%
0^9%
0.9%
0.0%
0.8%
0.0%
4.T%
7.0%
6^2%
5.5%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.3%
0.0%
0.0%
0.0%
0.8%
0.8%
0.9%
0.0%
0.7%
0.0%
4.0%
6.5%
4.9%
5.1%
Total
River
Miles
0
59
826
664
1,548
0
735
2,288
2,525
5,548
359
8,328
9,024
1,133
18,844
57
12,593
25,412
10,168
48,229
Improved Reaches
%of
River
Miles
Receiving
C&D
Discharges
0.0%
6.4%
6.7%
18.1%
9.1%
0.0%
30.7%
47.4%
29.3%
34.9%
26.2%
62.0%
71.1%
67.5%
64.6%
5.4%
33.9%
73.3%
70.2%
55.2%
%of
Total
River
Miles
0.0%
5.2%
6.6%
14.5%
8.5%
0.0%
28.5%
47.2% '
29.3%
34.4%
26.2%
53.9% '
67.7% '
32.6%
56.1%
5.3%
33.3% '
68.2%
55.6% '
51.1%
November 2008
D-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-2: Estimated Water Quality Improvements Under Option 2
EPA
Region
5
6
7
8
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
6,402
44,892
17,591
2,624
71,509
2,111
60,261
16,527
3,957
82,856
15,139
35,697
2,965
217
54,018
5,741
62,368
28,974
16,627
113,710
Miles of
River in
RFI
Network
6,402
44,929
17,596
2,624
71,550
2,332
73,189
17,718
5,430
98,669
16,103
41,624
2,965
217
60,909
8,426
76,218
29,040
16,627
130,311
Water Quality Improvements by WQI Change
0.01 < AWQI <
River
Miles
8
3,328
3,326
494
7,156
89
1,978
3,067
599
5,734
145
1,636
500
29
2,310
0
32
140
107
278
%of
River
Miles
Receiving
C&D
Discharges
0.1%
7.4%
18.9%
18.8%
10.0%
4.2%
3.3%
18.6%
15.1%
6.9%
1.0%
4.6%
16.9%
13.5%
4.3%
0.0%
0.1%
0.5%
0.6%
0.2%
0.1
%of
Total
River
Miles
0.1%
7.4%
18.9%
18.8%
10.0%
3.8%
2.7%
17.3%
11.0%
5.8%
0.9%
3.9%
16.9%
13.4%
3.8%
0.0%
0.0%
0.5%
0.6%
0.2%
0.
River
Miles
0
158
743
89
990
0
1,107
4,476
1,267
6,850
0
187
791
26
1,004
0
0
0
0
0
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.4%
4.2%
3.4%
1.4%
0.0%
1.8%
27.1%
32.0%
8.3%
0.0%
0.5%
26.7%
12.2%
1.9%
0.0%
0.0%
0.0%
0.0%
0.0%
%of
Total
River
Miles
0.0%
0.4%
4.2%
3.4%
1.4%
0.0%
1.5%
25.3%
23.3%
6.9%
0.0%
0.4%
26.7%
12.1%
1.6%
0.0%
0.0%
0.0%
0.0%
0.0%
0.5 < AWQI
%of
River
Miles
Receiving
River C&D
Miles Discharges
0 0.0%
57 0.1%
81 0.5%
o I 0.0%
138 I 0.2%
27 ' 1.3%
733 1.2%
2,694 16.3%
471 11.9%
3,924 4.7%
36 0.2%
157 , 0.4%
285 I 9.6%
0 I 0.0%
477 ! 0.9%
0 0.0%
9 0.0%
0 0.0%
0 0.0%
9 | 0.0%
%of
Total
River
Miles
0.0%
0.1%
0.5%
0.0%
0.2%
1.2%
1.0%
15.2%
8.7%
4.0%
0.2%
0.4%
9.6%
0.0%
0.8%
0.0%
0.0%
0.0%
0.0%
0.0%
Total
River
Miles
8
3,542
4,150
582
8,283
116
3,818
10,237
2,337
16,507
180
1,980
1,576
56
3,791
0
41
140
107
288
Improved Reaches
%of
River
Miles
Receiving
C&D
Discharges
0.1%
7.9%
23.6%
22.2%
11.6%
5.5%
6.3%
61.9%
59.0%
19.9%
1.2%
5.5%
53.1%
25.6%
7.0%
0.0%
0.1%
0.5%
0.6%
0.3%
%of
Total
River
Miles
0.1%
7.9%
23.6%
22.2%
11.6%
5.0%
5.2%
57.8%
43.0%
16.7%
1.1%
4.8%
53.2%
25.6%
6.2%
0.0%
0.1%
0.5%
0.6%
0.2%
November 2008
D-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-2: Estimated Water Quality Improvements Under Option 2
EPA
Region
9
10
National
Total
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
9,345
35,035
2,833
261
47,474
50
13,754
20,140
32,525
66,469
41,294
305,857
153,559
84,660
585,370
Miles of
River in
RFI
Network
13,040
38,566
3,468
1,185
56,259
50
13,862
20,533
35,050
69,495
48,871
345,376
159,331
96,099
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI <
River
Miles
0
331
10
0
342
167
1,514
2,864
0
4,545
748
25,260
36,228
11,396
73,632
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.9%
0.4%
0.0%
0.7%
334.0%
11.0%
14.2%
0.0%
6.8%
1.8%
8.3%
23.6%
13.5%
12.6%
0.1
%of
Total
River
Miles
0.0%
0.9%
0.3%
0.0%
0.6%
335.0%
10.9%
13.9%
0.0%
6.5%
1.5%
7.3%
22.7%
11.9%
11.3%
0.
River
Miles
0
40
0
6
46
0
0
0
0
0
65
5,135
14,711
4,811
24,723
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.1%
0.0%
2.4%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.2%
1.7%
9.6%
5.7%
4.2%
%of
Total
River
Miles
0.0%
0.1%
0.0%
0.5%
0.1%
0.0%
0.0%
0.0%
0.0%
0.0%
0.1%
1.5%
9.2%
5.0%
3.8%
0.5 < AWQI
%of
River
Miles
Receiving
River C&D
Miles Discharges
0 0.0%
0 0.0%
o ; 0.0%
0 | 0.0%
0 I 0.0%
0 ' 0.0%
0 0.0%
0 0.0%
0 0.0%
0 0.0%
74 0.2%
2,584 . 0.8%
5,589 I 3.6%
1,370 I 1.6%
9,616 | 1.6%
%of
Total
River
Miles
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.2%
0.7%
3.5%
1.4%
1.5%
Total
River
Miles
0
371
10
6
388
167
1,514
2,864
0
4,545
888
32,979
56,528
17,578
107,972
Improved Reaches
%of
River
Miles
Receiving
C&D
Discharges
0.0%
1.1%
0.4%
2.4%
0.8%
334.0%
11.0%
14.2%
0.0%
6.8%
2.1%
10.8%
36.8%
20.8%
18.4%
%of
Total
River
Miles
0.0%
1.0%
0.3%
0.5%
0.7%
335.0%
10.9%
13.9%
0.0%
6.5%
1.8%
9.5%
35.5%
18.3%
16.6%
November 2008
D-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-3: Estimated Water Quality Improvements Under Option 3
EPA
Region
1
2
3
4
5
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
30
916
12,343
3,663
16,952
52
2,394
4,824
8,617
15,887
1,371
13,438
12,695
1,679
29,183
1,053
37,102
34,667
14,490
87,312
6,402
44,892
17,591
2,624
71,509
Miles of
River in
RFI
Network
30
1,130
12,558
4^575
18,293
52
2,579
4,846
8,633
16,110
1,371
15,450
13",321
3,475
33,617
1,065
37,828
37,287
18,284
94,464
6,402
44,929
17,596
2,624
71,550
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
River
Miles
0
210
1,627
887
2,725
0
1,013
2,684
3,167
6,864
255
7,951
7,566
1,071
16,843
55
8,192
13,599
4,999
26,844
222
4,880
3,657
551
9,310
%of
River
Miles
Receiving
C&D
Discharges
0.0%
23.0%
13.2%
24.2%
16.1%
0.0%
423%
55.6%
36.8%
43.2%
18.6%
59.2%
59.6%
63.8%
57.7%
5.2%
22.1%
39.2%
34.5%
30.7%
3.5%
10.9%
20.8%
21.0%
13.0%
%of
Total
River
Miles
0.0%
18.6%
13.0%
19.4%
14.9%
0.0%
39.3%
55.4%
36.7%
42.6%
18.6%
51.5%
56.8%
30.8%
50.1%
5.1%
21.7%
36.5%
27.3%
28.4%
3.5%
10.9%
20.8%
21.0%
13.0%
0.
River
Miles
0
o
112
81
192
0
107
307 "
207
621
93
^358""""
1,980
199
3,631
31
4,318 "
10,446
4,603
19,397
0
458
1,214
210
1,883
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.9%
2.2%
1.1%
0.0%
4~5%
6.4%
2.4%
3.9%
6.8%
10.1%
15^6%
11.9%
12.4%
2.9%
lT.6%
30J%
31.8%
22.2%
0.0%
1.0%
6.9%
8.0%
2.6%
%of
Total
River
Miles
0.0%
0.0%
0.9%
1.8%
1.1%
0.0%
4.1%
6.3%
2.4%
3.9%
6.8%
8.8%
14.9%
5.7%
10.8%
2.9%
11.4%
28.0%
25.2%
20.5%
0.0%
1.0%
6.9%
8.0%
2.6%
0.5 < AWQI
River
Miles
0
0
9
o
9
0
21
9
0
30
20
186
142
7
355
6
2,665
4,110
1,705
8,486
0
82
332
0
414
%of
River
Miles
Receiving
C&D
Discharges
0.0%
0.0%
0.1%
0.0%
0.1%
0.0%
0.9%
0.2%
0.0%
0.2%
1.5%
1.4%
1.1%
0.4%
1.2%
0.6%
7.2%
11.9%
11.8%
9.7%
0.0%
0.2%
1.9%
0.0%
0.6%
%of
Total
River
Miles
0.0%
0.0%
0.1%
0.0%
0.0%
0.0%
0.8%
0.2%
0.0%
0.2%
1.5%
1.2%
1.1%
0.2%
1.1%
0.6%
7.0%
11.0%
9.3%
9.0%
0.0%
0.2%
1.9%
0.0%
0.6%
Total Improved Reaches
River
Miles
0
210
1,748
968
2,926
0
1,140
3,001
3,373
7,514
369
9,495
9^688
1,278
20,830
91
15,175
28454
11,307
54,727
222
5,420
5,203
761
11,606
%of
River
Miles
Receiving
C&D
Discharges
0.0%
23.0%
14k2%
26.4%
17.3%
0.0%
47.6%
62.2%
39.1%
47.3%
26.9%
70.7%
76.3%
76.1%
71.4%
8.7%
40.9%
81.2%
78.0%
62.7%
3.5%
12.1%
29.6%
29.0%
16.2%
%of
Total
River
Miles
0.0%
18.6%
119%
21.2%,
16.0%
0.0%
44.2%
61.9%
39.1%
46.6%
26.9%
613%
72.7%,
36.8%
62.0%
8.6%
40.1%
75.5%
61.8%
57.9%
3.5%
12.1%
29.6%
29.0%
16.2%
November 2008
D-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-3: Estimated Water Quality Improvements Under Option 3
EPA
Region
6
7
8
9
10
Baseline
Water
Quality
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
2,111
60,261
16,527
3,957
82,856
15,139
35,697
2,965
217
54,018
5,741
62,368
28,974
16,627
113,710
9,345
35,035
2,833
261
47,474
50
13,754
20,140
32,525
66,469
Miles of
River in
RFI
Network
2,332
73,189
17,718
5,430
98,669
16,103
41,624
2,965
217
60,909
8,426
76,218
29,040
16,627
130,311
13,040
38,566
3,468
1,185
56,259
50
13,862
20,533
35,050
69,495
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
River
Miles
92
1,775
2,829
517
5,212
207
2,138
599
29
2,973
22
520
2,116
945
3,603
138
1,518
130
0
1,786
0
282
1,623
2,584
4,489
%of
River
Miles
Receiving
C&D
Discharges
4.3%
2.9%
: 17.1%
13.1%
6.3%
1.4%
6.0%
20.2%
13.5%
5.5%
0.4%
0.8%
7.3%
5.7%
3.2%
1.5%
4.3%
4.6%
0.0%
3.8%
0.0%
2.1%
8.1%
7.9%
6.8%
%of
Total
River
Miles
3.9%
2.4%
16.0%
9.5%
5.3%
1.3%
5.1%
20.2%
13.4%
4.9%
0.3%
0.7%
7.3%
5.7%
2.8%
1.1%
3.9%
3.7%
0.0%
3.2%
0.0%
2.0%
7.9%
7.4%
6.5%
0.
River
Miles
34
1,349
4,423
1,288
7,095
19
209
788
26
1,041
0
96
127
138
362
0
306
0
0
306
133
886
2,349
3,367
1 < AWQI < 0.5
%of
River
Miles
Receiving
C&D
Discharges
1.6%
2.2%
26.8%
32.6%
8.6%
0.1%
0.6%
26.6%
12.2%
1.9%
0.0%
0.2%
0.4%
0.8%
0.3%
0.0%
0.9%
0.0%
0.0%
0.6%
0.0%
1.0%
4.4%
7.2%
5.1%
%of
Total
River
Miles
1.5%
1.8%
25.0%
23.7%
7.2%
0.1%
0.5%
26.6%
12.1%
1.7%
0.0%
0.1%
0.4%
0.8%
0.3%
0.0%
0.8%
0.0%
0.0%
0.5%
0.0%
1.0%
4.3%
6.7%
4.8%
0.5 < AWQI
River
Miles
62
1,148
3,560
625
5,395
56
256
423
0
735
0
57
11
60
128
0
95
0
6
101
0
68
308
1,886
2,261
%of
River
Miles
Receiving
C&D
Discharges
3.0%
1.9%
21.5%
15.8%
6.5%
0.4%
0.7%
14.3%
0.0%
1.4%
0.0%
0.1%
0.0%
0.4%
0.1%
0.0%
0.3%
0.0%
2.4%
0.2%
0.0%
0.5%
1.5%
5.8%
3.4%
%of
Total
River
Miles
2.7%
1.6%
20.1%
11.5%
5.5%
0.3%
0.6%
14.3%
0.0%
1.2%
0.0%
0.1%
0.0%
0.4%
0.1%
0.0%
0.2%
0.0%
0.5%
0.2%
0.0%
0.5%
1.5%
5.4%
3.3%
Total Improved Reaches
River
Miles
188
4,273
10,811
2,430
17,702
281
2,603
1,810
56
4,749
22
673
2,254
1,142
4,092
138
1,919
130
6
2,193
0
483
2,818
6,818
10,118
%of
River
Miles
Receiving
C&D
Discharges
8.9%
7.1%
65.4%
61.4%
21.4%
1.9%
7.3%
61.0%
25.6%
8.8%
0.4%
1.1%
7.8%
6.9%
3.6%
1.5%
5.5%
4.6%
2.4%
4.6%
0.0%
3.5%
14.0%
21.0%
15.2%
%of
Total
River
Miles
8.1%
5.8%
61.0%
44.8%
17.9%
1.7%
6.3%
61.0%
25.6%
7.8%
0.3%
0.9%
7.8%
6.9%
3.1%
1.1%
5.0%
3.7%
0.5%
3.9%
0.0%
3.5%
13.7%
19.5%
14.6%
November 2008
D-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table D-3: Estimated Water Quality Improvements Under Option 3
EPA
Region
National
Total
Baseline
Water
Quality
WQK26
26WQI
Total
Baseline Scenario
RFI Miles
Receiving
C&D
Discharges
41,294
305,857
153,559
84,660
585,370
Miles of
River in
RFI
Network
48,871
345,376
159,331
96,099
649,676
Water Quality Improvements by WQI Change
0.01 < AWQI < 0.1
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
990 2.4% 2.0%
28,480 ! 9.3% r 8.2%
36,429 : 23.7% ; 22.9%
14,750 17.4% 15.3%
80,648 | 13.8% | 12.4%
0.1 < AWQI < 0.5
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
177 0.4% 0.4%
8,333 r 2.7% r 2.4%
20,284 ; 13.2% ; 12.7%
9,102 10.8% 9.5%
37,895 | 6.5% | 5.8%
0.5 < AWQI
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
144 0.3% 0.3%
4,578 r 1.5% 1.3%
8,904 ; 5.8% ; 5.6%
4,288 5.1% 4.5%
17,915 | 3.1% | 2.8%
Total Improved Reaches
%of
River
Miles % of
Receiving Total
River C&D River
Miles Discharges Miles
1,311 3.2% 2.7%
41,391 r 13.5% r 12.0%
65,617 ; 42.7% ; 41.2%
28,139 33.2% 29.3%
136,458 | 23.3% | 21.0%
November 2008
D-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Appendix E - Meta-Analysis Results
EPA used function-based benefit transfer to estimate benefits of surface water quality improvements due
to reductions in sediment runoff from construction sites under the regulatory analysis options considered
for the analysis of the proposed C&D regulation. The benefit function was derived using meta-analysis,
following the general approach of Johnston et al. (2005), Shrestha et al. (2007), and others, following
conceptual methods outlined by Bergstrom and Taylor (2006). The recent literature has given increasing
emphasis to the potential use of meta-analysis to conduct and inform function-based benefit transfer
(Bergstrom and Taylor 2006; Johnston et al. 2005; Rosenberger and Stanley 2006; Shrestha et al. 2007).
For the present analysis, the meta-regression model was based on a model specification and data
developed originally for the 316(b) Phase II Cooling Water Intake rule, and extended to generate a benefit
function more suitable for assessing changes in ambient water quality from reducing sediment discharges.
Chapter A12, "Methods for Estimating Non-use Benefits," in the Regional Analysis document for the
final Phase II rule provides details on the original meta-analysis (USEPA 2004c); revisions are detailed
below. These revisions included adding new studies to the metadata, re-estimating the willingness-to-pay
(WTP) function to better account for ecological services potentially affected by sediments and nutrients,
and testing additional functional forms and statistical approaches.
As stated by Rosenberger and Johnston (2007, p. 1-2):
One of the primary advantages of meta-analysis as a benefit transfer tool relates to its capacity to allow more
appropriate adjustments of welfare measures based on patterns observed in the literature. Within a benefit transfer
context, transfer error is often inversely related to the correspondence between a study site and a policy site among
various dimensions (Rosenberger and Phipps 2007). The probability of finding a good fit between a single (or multiple)
study site and a policy site, however, is usually low (Boyle and Bergstrom 1992; Spash and Vatn 2006). If, on the other
hand, empirical studies contribute to a body of WTP estimates (i.e., metadata), and if empirical value estimates are
systematically related to variations in resource, study, and site characteristics, then meta-regression analysis may
provide a viable tool for estimating a more universal transfer function with distinct advantages over unit value or other
function-based transfer methods (Johnston et al. 2003; Rosenberger and Loomis 2000a; Rosenberger and Stanley
2006). More specifically, Rosenberger and Phipps (2007) posit a meta-valuation function as the envelope of a set of
empirically-defined valuation functions reported in the literature.
In the present case, EPA identified 45 valuation studies that use stated preference techniques to elicit
benefit values for water quality improvements. To examine the relative influence of study, economic, and
resource characteristics on WTP for improving surface water quality, the Agency conducted a regression-
based meta-analysis of 115 estimates of WTP for water quality improvements, provided by the 45 original
studies. Analytic methods and model specifications follow established methods in the published literature
(e.g., Johnston et al. 2005; Bergstrom and Taylor 2006). The estimated econometric model is used as the
basis of a function-based benefit transfer, to calculate WTP for improving water quality in water bodies
affected by sediment runoff from construction sites.
The following discussion summarizes the results of EPA's meta-analysis of surface water valuation
studies and the use of the resulting benefit function for transfer.
E.1 Literature Review of Water Resource Valuation Studies
As outlined in the introduction, EPA conducted a meta-analysis of water resource valuation studies to
examine the relative influence of study, economic, and resource characteristics on total WTP for water
quality improvements. The Agency analyzed 45 studies, published between 1985 and 2004, that applied
generally accepted, standard valuation methods to determine total (including use and non-use) values
November 2008
E-1
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
associated with aquatic habitat improvements.18 These 45 studies all used stated preference techniques to
assess WTP, but varied in other respects, including the survey administration methods used, the specific
environmental change valued, and the geographic region affected by the environmental changes. Data
from the 45 studies result in a total of 115 observations for the meta-analysis because 30 studies provide
more than one usable estimate of total WTP for aquatic habitat improvements.
When constructing metadata for subsequent meta-analysis, analysts must determine the optimal scope of
the metadata (Rosenberger and Johnston 2007), interpreted as the exact definition of the dependent
variable in the meta-regression model, which, in turn, defines the set of source studies to be considered
for inclusion in the metadata. The primary tradeoff is often between maintaining close similarity among
dependent variables versus including additional information (i.e., observations) in the metadata.
Similarity in dependent variable definition and study attributes within the metadata can be important for
two reasons. First, theory may dictate that certain types of estimated values are not strictly comparable
(e.g., Hicksian compensating surplus from a stated preference model versus Marshallian consumer
surplus from a travel cost model). Second, model fit may be improved by narrowing the metadata, for
example to include only valuation studies that use a particular valuation approach (e.g., stated preference
methods, as in Johnston et al. (2005); or travel cost model estimates, as in Smith and Kaoru (1990)). Such
study selection issues may be framed in terms of a requirement that at a minimum, studies included in
metadata satisfy both commodity consistency and welfare consistency (Bergstrom and Taylor 2006). The
former implies that "the commodity (Q) being valued should be approximately the same within and
across studies" (Bergstrom and Taylor 2006, p. 353). The latter implies that "measures of WTP within
and across studies ... should represent the same ... welfare change measure, or ex-post calibrations [are]
made to account for theoretical differences between welfare change measures" (Bergstrom and Taylor
2006, p. 355).
The requirement of welfare consistency implies that—outside of preference calibration approaches that
explicitly account for theoretical differences between welfare constructs (e.g., Smith et al. 2002)—meta-
analyses should not combine data representing theoretically distinct welfare measures. For example, the
ad hoc combination of stated and revealed data for meta-analysis would generally violate the condition of
welfare consistency. Although one may introduce independent variables in regression models (typically
dummy variable intercept shifters) to account for such theoretical differences, Smith and Pattanayak
(2002) argue that such methods are unlikely to represent appropriate adjustments for fundamental
differences in theoretical welfare measures. Results of preliminary models estimated by EPA support this
conclusion; preliminary models that attempted to combine both stated and revealed preference studies
generated poorly fitting, counter-intuitive, and often nonsensical results. For this reason, all final models
impose welfare consistency and limit analyses to metadata including only stated preference studies.
E.1.1 Identifying Water Resource Valuation Studies
EPA identified surface water valuation studies used in the total WTP meta-analysis by conducting an in-
depth search of the economic literature. EPA used a variety of sources and search methods to identify
relevant studies:
18 Data were collected from 66 studies. Nine studies using revealed preference techniques were dropped from the model since
combination of stated and revealed preference studies violates the requirement of welfare consistency. Additional studies
were dropped from the analysis because they utilized nonstandard methodologies or did not provide information for key
regressors.
November 2008 E-2
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
> Review of EPA's research and bibliographies dealing with non-market benefits associated with
water quality changes
> Systematic review of recent issues of resource economics journals (e.g., Land Economics, Marine
Resource Economics, Journal of Environmental Economics and Management)
> Searches of online reference and abstract databases (e.g., Environmental Valuation Resource
Inventory (EVRI), Benefits Use Valuation Database (BUVD), AgEcon Search)
> Visits to home pages of authors known to have published contingent valuation studies and or
water quality research
> Searches of Web sites of agricultural and resource economics departments at several colleges and
universities
> Searches of Web sites of organizations and agencies known to publish environmental and
resource economics valuation research (e.g., Resources for the Future (RFF), National Center for
Environmental Economics (NCEE), Natural Resource Conservation Service (NRCS), National
Bureau of Economic Research (NBER)).
From this review, EPA identified approximately 300 surface water valuation studies that were potentially
relevant for this analysis and compiled a bibliographic database to organize the literature review process.
Sixty-six of these studies met the criteria identified for inclusion in the meta-analysis.19 These criteria
were designed to ensure both commodity and welfare consistency as noted above, and include:
> Specific amenity valued: Selected studies were limited to those in which the environmental
quality change being valued affects ecological services provided by surface water bodies,
including aquatic life support, recreational activities (such as fishing, boating, and swimming),
and non-use value
> Values estimated: Selected studies were limited to those that elicited total household WTP (the
total of use and non-use components)
> Study location: Selected studies were limited to those that surveyed U.S. and Canadian
populations to value resources
> Research methods: Selected studies were limited to those that applied research methods
supported by journal literature.
The Agency compiled extensive information from the 66 selected studies. Of these studies, 45 were
utilized in the model estimation. Reasons for the difference between the total number of studies in the
final metadata (66) and studies represented in the final model (45) include unavailability of information
for certain key regressors for all studies.
The tradeoff between the number of regressors or independent variables that may be included in a meta-
regression analysis (K) and the number of studies that are included in the metadata (TV) is a fundamental
tradeoff in most meta-analyses in the valuation literature (Moeltner et al. 2007). That is, if a study
considered for inclusion in the metadata does not provide information for a certain regressor that analysts
might wish to include in the meta-regression, analysts must generally choose between omitting the
19 The remaining studies were either earlier unpublished versions or slightly modified versions of the included studies, focused
on water resources outside of the United States and Canada, used secondary research methods, or valued environmental
quality changes that were not directly linked to changes in water quality (e.g., change in recreational catch rates).
November 2008 E-3
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
regressor from the meta-regression or omitting the study from the metadata. As a result, researchers
wishing to increase the number of explanatory variables in meta-regression models (increasing K) often
do so at the cost of reductions in the number of studies or observations in the metadata (reducing N).
Conversely, increases in the number of observations in meta-regression models are sometimes only
possible if one reduces the number of independent variables in the model. Hence, there is a tradeoff
between the quantity of information in the metadata (i.e., the number of observations or studies) and the
possible risk of omitted variables bias due to the omission of influential regressors. See Moeltner et al.
(2007) for additional discussion of this "TV versus K" tradeoff in meta-analysis.
The complete data set used in the meta-analysis is provided in the docket for the proposed C&D
regulation (DCN 6-1900), and includes the following information:
> Full study citation
> Study location
> Sample data and description (e.g., size, response rate, income)
> Resource characteristics (e.g., affected water body type, recreational uses, baseline quality)
> Description of environmental quality change, including geographic scale, affected species, and
affected recreational uses (e.g., water quality change from fishable to beatable)
> Quantitative measure of environmental quality change measured in terms of improvements in
Water Quality Index (WQI)20 and/or percentage reduction in pollutant concentration
> Study WTP values updated to 2007 dollars
> WTP estimation characteristics (i.e., parametric versus non-parametric, inclusion of protest bids
and outlier bids, WTP description).
E.1.2 Description of Studies Selected for Total WTP Meta-Analysis
The 45 studies that EPA used in the total WTP meta-analysis were conducted between 1981 and 2002,
and applied standard, generally accepted stated preference valuation methods to assess WTP. Studies
were excluded if they did not conform to general tenets of economic theory, or if they applied methods
not generally accepted in the literature.
All selected studies focus on environmental quality changes that affect surface water resources in the
United States and Canada. Beyond this general similarity, the studies vary in several respects. Differences
include the specific environmental change valued, scale of environmental improvement, geographic
region affected by environmental changes, types of values estimated, survey administration methods,
demographics of the survey sample, and statistical methods employed. The 45 studies include 25 journal
articles, 6 reports, 5 Ph.D. dissertations, 7 academic or staff papers, 1 book, and 1 master's thesis.
The 45 studies selected for the meta-analysis provided 115 observations in the final data set because
multiple estimates of WTP were available from 30 studies. The availability of multiple observations from
Additional details on the WQI and the use of the WQI in survey instruments are provided by McClelland (1974), Mitchell
and Carson (1989, p. 342), and Vaughan (1986). This index is linked to specific pollutant levels, which in turn are linked to
presence of aquatic species and suitability for particular recreational uses. The WQI allows the use of objective water quality
parameters (e.g., dissolved oxygen concentrations) to characterize ecosystem services or uses provided by a given water
body.
November 2008 E-4
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
single studies is common in meta-analyses of this type (e.g., Bateman and Jones 2003; Johnston et al.
2005). Some of the characteristics that allowed multiple observations to be derived from a single study
include the extent of the amenity change, the respondent population type, elicitation method(s), water
body type, number of water bodies affected, recreational activities affected by the quality change, and
species affected by the quality change. Table E-l lists key study and resource characteristics and indicates
the number of observations derived from each study.
Surveys in 26 studies were administered by mail; 10 studies collected information through personal
interviews in the home, onsite, or in a centralized location, and 9 surveys were conducted by telephone.
Study sample sizes range from 96 to 2,907 responses.
Table E-1: Selected Summary Information for Studies
Author(s) and Year
Aiken(1985)
Anderson and
Edwards (1986)
Azevedoetal. (2001)
Bockstael et al.
(1989a)
Bockstael et al.
(1989b)
Breffleetal. (1999)
Cameron (1988)
Cameron and
Huppert(1989)
Carson and Mitchell
(1993)
Carson etal. (1994)
Clonts and Malone
(1990)
Croke etal. (1987)
DeZoysa(1995)
Desvousges et al.
(1987)
Hayes etal. (1992)
Herriges and Shogren
(1996)
Kite (2002)
Huang etal. (1997)
Hushak and Bielen
(1999)
Kaoru(1993)
Lant and Roberts
(1990)
Lant and Tobin
(1989)
Lichtkoppler and
Elaine (1999)
Lindsey(1994)
Lipton (2004)
Loomis(1996)
Loomis et al. (2000)
Lyke(1993)
Magat et al. (2000)
Observations
1
1
5
1
2
2
1
1
4
2
3
9
2
12
2
2
2
2
2
1
3
9
1
8
1
1
2
2
4
State
CO
RI
IA
DC,MD,
VA
MD
WI
TX
CA
National
CA
AL
IL
OH
PA
RI
IA
MS
NC
OH, MI
MA
IA,IL
IA,IL
OH
MD
MD
WA
CO
WI
CO,NC
Water Body
Type
all freshwater
salt
pond/marshes
lake
estuary
estuary
estuary
Estuary
river/stream
multiple
estuary
river/stream
river/stream
river and lake
river/stream
estuary
lake
river/stream
estuary
river/stream
salt
pond/marshes
river/stream
river/stream
multiple
estuary
estuary
river/stream
river/stream
lake
all freshwater
Type of Water Quality
Improvement
general water quality
general water quality
nutrients
general water quality;
phosphorus and nitrogen
general water quality
general water quality
general water quality
wildlife habitat
general water quality
DDTandPCBs
general water quality
general water quality
sediment and nutrients;
wildlife habitat
general water quality
general water quality
general water quality
general water quality
general water quality
general water quality
fecal coliform
sediment
general water quality
PCBs and general water
quality
nutrients
general water quality
general water quality
general water quality
general water quality
general water quality
Affected
Recreational Uses
fishing
fishing and swimming
fishing and swimming
swimming, beach,
boating, fishing,
outings
swimming
fishing
fishing
game fishing
boating; fishing;
swimming
fishing
multiple uses
multiple uses
multiple uses
multiple uses
swimming; fishing
boating and fishing
multiple uses
fishing
multiple uses
fishing
boating, fishing, and
swimming
boating, fishing
all recreational uses
multiple uses
boating
fishing
multiple uses
fishing
fishing and swimming;
fishing
November 2008
E-5
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table E-1: Selected Summary Information for Studies
Author(s) and Year
Matthews et al.
(1999)
Opaluchetal. (1998)
Olsenetal. (1991)
Roberts and Leitch
(1997)
Roweetal. (1985)
Sanders etal. (1990)
Schulzeetal. (1995)
Shrestha and
Alavalapati (2004)
Stumborg et al.
(2001)
Sutherland and
Walsh (1985)
Welle (1986)
Wey(1990)
Whitehead et al.
(2002)
Whitehead and
Groothuis(1992)
Whitehead et al.
(1995)
Whittington et al.
(1994)
Observations
2
1
3
1
1
4
2
2
2
1
6
2
1
3
2
1
State
MN
NY
ID, MT,
OR,WA
MN, SD
CO
CO
MT
FL
WI
MT
MN
RI
NC
NC
NC
TX
Water Body
Type
river/stream
estuary
river/stream
lake
river/stream
river/stream
river and lake
multiple
lake
river and lake
all freshwater
salt pond/
marshes
river/stream
river/stream
estuary
estuary
Type of Water Quality
Improvement
phosphorus
general water quality
wildlife habitat
general water quality
general water quality
general water quality
hazardous pollutants
phosphorus and wildlife
habitat
phosphorus
general water quality
acid rain
general water quality
general water quality
sediment and nutrients
general water quality
heavy metals and pesticides
Affected
Recreational Uses
boating and fishing
shellfishing
fishing
multiple uses
boating, fishing, and
swimming
swimming
boating, fishing, and
swimming
multiple uses
multiple uses
swimming
game fishing and
wildlife viewing
Other
fishing, boating,
swimming
multiple uses
boating, fishing, and
swimming
multiple uses
The Agency's review of the relevant economic literature showed that available surface water valuation
studies focus primarily on general water quality rather than specific pollutants or changes. Even in cases
in which specific pollutants are the primary policy issue, the stated preference surveys from which
welfare estimates are derived often characterize water quality changes only in general (i.e., non-pollutant
specific) terms. Hence, the associated welfare measures are conditioned on this general description. Of
the 45 studies, 28 presented only WTP values for changes in general water quality (approximately 62
percent). Of the studies that did address specific changes, nine specified nutrients and/or sediment, four
addressed hazardous pollutants including heavy metals and pesticides, four addressed wildlife habitat, one
addressed acid deposition, and one addressed fecal coliform bacteria. Preliminary model estimates
showed no evidence that the type of pollutant considered had a statistically significant influence on WTP
across and within studies. For this reason, EPA used a standardized scale to define both the baseline water
quality and the water quality change valued in the original study. Additional details are provided below.
From these 45 studies, the Agency compiled a data set for the meta-analysis of total WTP values. EPA
specified a regression model based on these data to estimate a range of total household benefits for
surface water and aquatic habitat improvements. General empirical methods follow those outlined by
Johnston et al. (2005), following standard approaches in the meta-analysis literature. The model and
results are described in the next section.
November 2008
E-6
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
E.2 Total WTP Meta-Analysis Regression Model and Results
EPA estimated both trans-log and semi-log meta-regression models based on 115 WTP estimates for
improvements in water resources, derived from 45 original studies.21 These metadata, the model
specification, model results, and interpretation of the results are described in the following sections. EPA,
however, notes that only the trans-log model is used in the analysis of benefits from reduced sediment
discharges. The alternative specification (semi-log) is presented for illustrative purposes only.
In a frequently cited work, Glass (1976) characterizes meta-analysis as "the statistical analysis of a large
collection of results for individual studies for the purposes of integrating the findings. It provides a
rigorous alternative to the casual, narrative discussion of research studies which is commonly used to
make some sense of the rapidly expanding research literature" (p. 3; cited in Poe et al. 2001, p. 138).
Meta-analysis is being increasingly explored as a potential means to estimate resource values in cases
where original targeted research is impractical, or as a means to reveal systematic components of WTP
(e.g., Smith and Osborne 1996; Santos 1998; Rosenberger and Loomis 2000a; Poe et al. 2001; Woodward
and Wui 2001; Bateman and Jones 2003). While the literature often urges caution in the use and
interpretation of benefit transfers for direct policy application (e.g., Desvousges et al. 1998; Poe et al.
2001; Navrud and Ready 2007), such methods are "widely used in the United States by government
agencies to facilitate benefit-cost analysis of public policies and projects affecting natural resources"
(Bergstrom and De Civita 1999). Transfers based on meta-analysis are likewise common in both the
United States and Canada (Bergstrom and De Civita 1999; Bergstrom and Taylor 2006).
Depending on the suitability of available data, a meta-analysis can provide a superior alternative to the
calculation and use of a simple arithmetic mean WTP over the available observations, as it allows
estimation of the systematic influence of study, economic, and natural resource attributes on WTP
(Shrestha et al. 2007; Rosenberger and Phipps 2007; USEPA 2000b The primary advantage of a
regression-based (statistical) approach is that it accounts for differences among study characteristics that
may contribute to changes in WTP, to the extent permitted by available data (Rosenberger and Phipps
2007; Johnston et al. 2005). An additional advantage is that meta-analysis can reveal systematic factors
influencing WTP, allowing analysts to assess whether, for example, WTP estimates are (on average)
sensitive to scope (Smith and Osborne 1996).
There is, however, some controversy regarding whether regression-based meta-analyses should be used
for direct benefit transfer. Many contemporary sources in the literature note the potential ability of
regression-based meta-analyses to generate benefit functions better able to adjust and forecast benefits at
policy sites in question, and either explicitly or implicitly favor the use of meta-analysis over alternative
benefit transfer approaches (e.g., Bergstrom and Taylor 2006; Moeltner et al. 2007; Johnston et al. 2005;
Rosenberger and Johnston 2007; Shrestha et al. 2007; Rosenberger and Phipps 2007). USEPA (2000b)
characterizes meta-analysis as "the most rigorous" benefit transfer method. In contrast, the USEPA
Science Advisory Board's Advisory on EPA's Issues in Valuing Mortality Risk Reduction (2007)
recommends against the use of regression-based meta-analysis for VSL (value of statistical life) transfers,
and other authors advise caution in such uses (Navrud and Ready 2007). The primary disagreement is
whether it is appropriate to use meta-analysis results as a reduced form model to estimate benefits, and
In its analysis of non-use benefits for the final Phase II rule, EPA also specified trans- and semi-log regression models
similar to the models estimated for Phase II discussed in this section. See Chapter A12, "Methods for Estimating Non-use
Benefits," in the Regional Analysis document for the final Phase II rule for further details regarding both the log-log and
semi-log regression models estimated in EPA's analysis of non-use benefits for the final Phase II rule (USEPA, 2004c;
http://www.epa.gov/ost/316b/casestudy/final.htm).
November 2008 E-7
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
whether the empirical ability of meta-analysis in many cases to generate benefit transfers with reduced
transfer errors offsets the lack of an underlying theoretical model to "calibrate" benefit estimates across
studies (cf. Smith et al. 2002; Smith and Pattanayak 2002). While the Agency recognizes this ongoing
controversy, it notes that there are a large number of practitioners and publications supporting the use of
regression-based meta-analysis for benefit transfer. Hence, the following model is presented as a means to
provide a benefit function that capitalizes on the substantial information available for existing water
quality valuation studies, notwithstanding potential concerns voiced by some regarding the use of meta-
analyses for such purposes.
E.2.1 Metadata Total WTP Regression Model
Meta-analysis is largely an empirical, data-driven process, but one in which variable and model selection
is guided by theory (Bergstrom and Taylor 2006). Given a reliance on information available from the
underlying studies that comprise the metadata, meta-analysis models most often represent a middle
ground between model specifications that would be most theoretically appropriate and those
specifications that are possible given available data. Smith and Osborne (1996), Rosenberger and Loomis
(2000a), Poe et al. (2001), Bateman and Jones (2003), Dalhuisen et al. (2003), Moeltner et al. (2007),
Bergstrom and Taylor (2006), Johnston et al. (2005, 2006), and others provide insight into the mechanics
of specifying and estimating meta-equations in resource economics applications.
Past meta-analyses have incorporated a range of different statistical methods, with none universally
accepted as superior (Johnston et al. 2005). EPA followed recent work of Bateman and Jones (2003) and
Johnston et al. (2005) in applying a multilevel model specification to the metadata, to address potential
correlation among observations gathered from single studies. Also following prior work (e.g., Smith and
Osborne 1996; Poe et al. 2001), EPA applied the Huber-White robust variance estimation. As described
by Smith and Osborne (1996, p. 293), "this approach treats each study as the equivalent of a sample
cluster with the potential for heteroskedasticity.. .across clusters." Weighted models are avoided
following the arguments of Bateman and Jones (2003).22
To guide development of the model and variable specifications, EPA relied upon a set of general
principles. These principles are designed to help prevent excessive data manipulations and other factors
that may lead to misleading model results. The general principles include, all else being equal:
> Fewer and simpler data transformations are preferred to more extensive ones.
> In the absence of overriding theoretical considerations, continuous variables are generally
preferred to discrete variables derived from underlying continuous distributions.
> Models should attempt to capture elements of the scope and scale of resource changes.
> Models should distinguish WTP associated with different types of resources and resource uses,
particularly where relevant to the policy question at hand.
> Following the "weak structural utility theoretic" (WSUT) approach of Bergstrom and Taylor
(2006, p. 352), exogenous structural constraints are avoided to afford the flexibility necessary to
For comparison, models were also estimated using ordinary least squares (OLS) with robust variance estimation, weighted
least squares with robust variance estimation, and multilevel models with standard (non-robust) variance estimation. None
of these models outperformed the illustrated model in terms of overall model significance and fit, and statistical significance
of individual coefficients (see Section 12E. 2.2: Total WTP Regression Model and Results for further details concerning the
specification of the model).
November 2008 E-8
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
appropriately model empirical patterns that may not necessarily flow from an underlying
theoretical modeling structure. The dependent variable in the meta-analysis is the natural
logarithm of estimated household WTP for water quality improvements in aquatic habitat, as
reported in each original study. For this analysis, original study values were adjusted to 2007$
based on the relative change in Consumer Price Index (CPI) from the study year to 2007. Total
WTP over the sample ranged from $8.35 to $503.36, with a mean value of $115.37.
As noted above, two model specifications are estimated (cf Johnston et al. 2005). For the first
specification, all right-hand-side variables are linear, resulting in a standard semi-log functional form. The
second specification is identical to the semi-log model, except for the specification of the explanatory
variable measuring water quality change and baseline as natural logs. This results in a trans-log functional
form, also common in empirical applications (Johnston et al. 2001, 2005). Both the semi-log and trans-log
models have advantages related to (1) their fit to the data, (2) the intuitive results that are provided, and
(3) their common use in the empirical valuation and meta-analysis literature (e.g., Smith and Osborne
1996; Santos 1998; Johnston et al. 2001, 2005, 2006). The trans-log model, however, has the additional
structural advantage that estimated WTP is necessarily zero when water quality change is also zero (cf.
Johnston et al. 2001)—a property suggested by theory that analysts may wish to weight against model fit
considerations when choosing a model for benefit estimation. While linear forms are also common in this
literature (Rosenberger and Loomis 2000a, 2000b; Poe et al. 2001; Bateman and Jones 2003),
specifications requiring more intensive data transformations (e.g., Box-Cox, log-log) are less common.
As noted in the preceding section, the metadata include independent variables characterizing specific
details of the resource(s) valued such as the baseline resource conditions; the extent of resource
improvements and whether they occur in estuarine or freshwater; the geographic region and scale of
resource improvements (e.g., the number of water bodies); resource characteristics (e.g., baseline
conditions, the extent of water quality change, and ecological services affected by resource
improvements); characteristics of surveyed populations (e.g., users, non-users); and other specific details
of each study. For ease of exposition, these variables are categorized into those characterizing (1) study
and methodology, (2) surveyed populations, (3) geographic region and scale, and (4) resource
improvements. Attributes included within each category are summarized below.
Study and methodology variables characterize such features as:
> The year in which a study was conducted
> The payment vehicle and elicitation format (e.g., discrete choice versus open-ended, voluntary
versus non-voluntary, interview versus mail versus phone)
> WTP estimation methods and conventions (e.g., approaches to protest and outlier bids, use of
parametric versus non-parametric statistical methods, estimation of mean or median WTP, the use
of annual or lump-sum payments)
> Whether the original survey represented water quality changes using the WQI.
Surveyed populations variables characterize such features as:
> The average income of respondents
> Whether the survey specifically targeted non-users.
Geographic region and scale variables characterize such features as:
> The number of water bodies affected by the policy
November 2008 E-9
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
> Whether the study considered water quality improvements in all water bodies in a region
> The geographic area of the country in which the study was conducted.
Resource improvement variables characterize such features as:
> The extent of water quality change estimated as a difference between the baseline and post-
change water quality index
> Baseline water quality index
> Those studies for which recreational uses such as fishing are specifically noted in the survey
> Aquatic species affected by resource improvements (e.g., game fish and shellfish)
> Those studies identifying large increases in fish populations (i.e., greater than 50 percent).
Although the interpretation and calculation of most independent variables requires little explanation, a
few variables require additional detail. These include the variables characterizing surface water quality
and its measurement. Many (23) observations in the metadata characterize quality changes using variants
of the WQI (e.g., Mitchell and Carson 1989). This scale is linked to specific pollutant levels, which, in
turn, are linked to the presence of aquatic species and recreational uses. However, some observations
provide water quality measures using other, primarily descriptive, means that differ from the WQI.
To allow consistent comparisons of water quality change using a single scale, EPA mapped all water
quality measures to the original WQI developed by McClelland (1974). WQI values were therefore
developed for those studies that did not originally use this index. This scale was chosen for two reasons:
> WQI values are linked to specific pollutant levels including sediment concentrations, which, in
turn, are linked to the presence of aquatic species and suitability for particular recreational uses.
Therefore, the WQI can be used to link water quality changes from reduced sediment runoff to
effects on human uses and support for aquatic species habitat. Chapter 10 of this report provides
detail on application of the WQI to estimating resource improvements from reduced sediment
runoff from construction sites.
> A large number of the original studies in the metadata included WQI measures as "native"
components of the original surveys. Hence, for these studies, no additional transformations were
required.
While not all studies in the metadata included the WQI as a native survey component, in most cases the
descriptions of water quality (present in the studies that did not apply the WQI) rendered mapping of
water quality measures to the WQI straightforward. In cases where baseline and improved (or declined)
water quality was not defined by suitability for recreational activities (e.g., boating, fishing, and
swimming) or corresponding qualitative measures (e.g., poor, fair, good), EPA used descriptive
information available from studies (e.g., amount/indication of the presence of specific pollutants,
historical decline of the quality of the resource) to approximate the baseline level of water quality and the
magnitude of the change.23 For studies that valued discrete changes in the size of species populations,
EPA characterized the baseline quality based on the current presence and prevalence of the species at
hand, and assumed population increases to correspond to modest increases in water quality in order to be
23 For example, a study by Huang et al. (1997) described current water quality as degraded from 1981 levels in terms of
reduced fish catches (60 percent) and reduced number of open shellfish beds (25 percent). However, because the water
resource was still supporting recreational fishery, the baseline water quality was set to "fishable" on the WQI.
November 2008 E-10
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
conservative. To account for the uncertainty involved in mapping those studies that are not based on the
WQI, EPA introduced the binary variable WQI, which indicates those studies in which WQI
measurements were an original component of the survey instrument. This approach is based largely on the
published methods of Johnston et al. (2005), drawn from prior Agency work for the 316(b) Phase II
Cooling Water Intake rule.
Variables incorporated in the final model are listed and described in Table E-2.
Table E-2: Variables and Descriptive Statistics for the Total WTP Regression Model
Variable
ln_WTP
year indx
Discrete
Volunt
Mail
lump sum
Nonparam
quality_ch
Inquality ch1
WQI
non-reviewed
outlier_bids
median_WTP
Income
Nonusers
Description
Natural log of WTP for specified resource improvements.
Year in which the study was conducted, converted to an index by
subtracting 1980.
Binary (dummy) variable indicating that WTP was estimated
using a discrete choice survey instrument.
Binary (dummy) variable indicating that WTP was estimated
using a payment vehicle described as voluntary as opposed to, for
example, property taxes.
Binary (dummy) variable indicating that the survey was
conducted through mail (default value for this dummy is a phone
survey).
Binary (dummy) variable indicating that payments were to occur
on something other than an annual basis over a long period of
time, such as property taxes. For example, some studies specified
that payments would occur over a five-year period.
Binary (dummy) variable indicating that WTP was estimated
using non-parametric methods.
The change in mean water quality, specified on the WQI
(McClelland 1974; Mitchell and Carson 1989). Defined as the
difference between baseline and post-compliance quality. Where
the original study (survey) did not use the WQI, EPA mapped
water quality descriptions to analogous levels on the WQI to
derive water quality change (see text).
Natural log of the change in mean water quality (quality ch),
specified on the WQI (McClelland 1 974; Mitchell and Carson
1989).
Binary (dummy) variable indicating that the original survey
reported resource changes using a standard WQI.
Binary (dummy) variable indicating that the study was not
published in a peer-reviewed journal.
Binary (dummy) variable indicating that outlier bids were
excluded when estimating WTP.
Binary (dummy) variable indicating that the study reported
median, not mean, WTP.
Mean income of survey respondents, either as reported by the
original survey or calculated by EPA based on U.S. Census
Bureau averages for the original surveyed region.
Binary (dummy) variable indicating that the survey was
implemented over a population of non-users (default category for
this dummy is a survey of any population that includes users).
Units and
Measurement
Natural log of dollars
(Range: 1.98 to 5. 93)
Year index (Range:
1 1 to 22)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
WQI units
(Range: 0.25 to 5.75)
WQI units
(Range: 0.92 to 4. 17)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Dollars
(Range: 34,955 to
158,347)
Binary
(Range: 0 or 1)
Mean
(Std. Dev.)
4.45
(0.78)
9.49
(6.04)
0.28
(0.45)
0.05
(0.22)
0.48
(0.50)
0.13
(0.34)
0.50
(0.50)
2.17
(1.08)
2.90
(0.60)
0.34
(0.48)
0.32
(0.47)
0.94
(0.24)
0.04
(0.20)
56,806.24
(13,903.11)
0.09
(0.29)
For example, a study by Lyke (1993) describes the baseline conditions as follows: (1) "there are no naturally reproducing
lake trout in Lake Michigan; all lake trout found there are from hatcheries." (2) "Lake Superior stocks of self-reproducing
lake trout were much reduced, but not wiped out, and both natural and hatchery-raised lake trout are found there." These
baseline conditions correspond to the "game-fishable" level on the WQI. The study estimates WTP for restoring natural
populations of lake trout to the Wisconsin Great Lakes. Therefore, the expected change that will occur within the "game-
fishable" category is likely to be small.
November 2008
E-11
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table E-2: Variables and Descriptive Statistics for the Total WTP Regression Model
Variable
single^ver2
single lake
regional fresh
multiple river
salt_pond
num_riv_pond
Mr
Mp
Allmult
Nonspec
fish_use
Fishplus
Baseline
Lnbase
Description
Binary (dummy) variable indicating that resource change
explicitly took place over a single river (default is a change in an
estuary or that takes place on a national scale).
Binary (dummy) variable indicating that resource change
explicitly took place over a single lake (default is a change in an
estuary or that takes place on a national scale).
Binary (dummy) variable indicating that resource change
explicitly took place over an entire region such as a state (default
is a change in an estuary or a change that takes place on a national
scale).
Binary (dummy) variable indicating that resource change
explicitly took place over multiple rivers (default is a change in an
estuary or that takes place on a national scale).
Binary (dummy) variable indicating that resource change
explicitly took place over multiple salt ponds (default is a change
in an estuary or that takes place on a national scale).
Number of rivers or salt ponds affected by policy; if unspecified
num riv_pond = 0. (In the present data, only studies addressing
rivers and lakes specified >1 number of water bodies. All others
specified either 1 water body, or the number was unspecified.)
Binary (dummy) variable indicating that the survey included
respondents from more than one of the EPA regions.
Binary (dummy) variable indicating that the survey included
respondents from the Mountain Plain region.4
Binary (dummy) variable indicating that either all or multiple
aquatic species are affected by the resource change.
Binary (dummy) variable indicating that the study did not specify
what species would be affected by water quality improvements.
Binary (dummy) variable identifying studies in which changes in
fishing use are specifically noted in the survey.
Binary (dummy) variable identifying studies in which a fish
population or harvest change of 50 percent or greater is reported
in the survey.
Baseline water quality, specified on the WQI.
Natural log of baseline water quality, specified on the WQI.
Units and
Measurement
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Number of specified
rivers or ponds
(Range: 0 to 15)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
Binary
(Range: 0 or 1)
WQI units (Range: 5
to 70)
WQI units (Range: 5
to 70)
Mean
(Std. Dev.)
0.22
(0.42)
0.08
(0.28)
0.36
(0.48)
0.06
(0.24)
0.03
(0.18)
1.00
(2.98)
0.08
(0.26)
0.03
(0.18)
0.20
(0.40)
0.40
(0.49)
0.56
(0.50)
0.08
(0.28)
40.33
(20.07)
3.48
(0.80)
Source: U.S. EPA analysis.
1 The variable quality_ch is defined earlier in this table as the difference between baseline and post-compliance quality, specified on the WQI
(Mitchell and Carson 1989).
2 Examples of rivers and streams considered in the studies include the Columbia, Potomac, Elwha, Eagle, and Tar-Pamlico rivers.
3 Includes one study that focused on a segment of the Lake Erie shoreline.
4 The Mountain Plain region includes the following states: Colorado, Iowa, Kansas, Missouri, Montana, Nebraska, North Dakota, South
Dakota, Utah, Wyoming.
E.2.2 Total WTP Regression Model and Results
As noted above, EPA estimated the meta-analysis regression using a multilevel, random-effects
specification. This model follows the general approaches of Bateman and Jones (2003) and Johnston et al.
(2005), among others. Multilevel (or hierarchical) models may be estimated as either random-effects or
random-coefficients models, and are described in detail elsewhere (Singer 1998). The fundamental
distinction between these and classical linear models is the two-part modeling of the equation error to
account for hierarchical data. Here, the metadata are comprised of multiple observations per study, and
November 2008
E-12
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
there is a corresponding possibility of correlated errors among observations that share a common study or
author.
The common approach to modeling such potential correlation is to divide the residual variance of
estimates into two parts, a random error that is independently and identically distributed across all studies
and for each observation, and a random effect that represents systematic variation related to each study.
The model is estimated as a two-level hierarchy, with level one corresponding to WTP estimates
(individual observations), and level two corresponding to individual studies. The random effect may be
interpreted as a deviation from the mean equation intercept associated with individual studies (Bateman
and Jones 2003). The model is estimated using a maximum likelihood estimator, assuming that random
effects show a multivariate normal distribution. Following Bateman and Jones (2003), observations are
unweighted. Covariances are obtained using the Huber-White covariance estimator (Smith and Osborne
1996). Random-effects models such as the multilevel model applied here are becoming increasingly
standard in resource economics applications, and are estimable using a variety of readily available
software packages.
A Note on Model Specification
As noted above, EPA considered two functional forms in this analysis: semi-log and trans-log. In both
cases, the dependent variable is the natural log of estimated household WTP for surface water quality
improvement, as shown in Table E-2. For model one, all right-hand-side variables are linear, resulting in a
semi-log functional form common in meta-analysis (e.g., Smith and Osborne 1996; Johnston et al. 2003).
While linear forms are also common (Bateman and Jones 2003; Poe et al. 2001; Rosenberger and Loomis
2000a, 2000b), the semi-log and trans-log forms were chosen based on statistical performance and ability
to capture curvature in the valuation function, and because they allow independent variables to influence
WTP in a multiplicative rather than additive manner.
Model two is a trans-log model, identical to the semi-log specification save for the inclusion of water
quality measures (baseline and quality_ch) as natural logarithms. This form—common in the hedonic
modeling literature (Johnston et al. 2001) and illustrated by Johnston et al. (2005) within the meta-
analysis literature—shares many advantages of the semi-log functional form, but also incorporates the
desirable quality that WTP is constrained to zero when quality change is also equal to zero.
Following standard econometric practice and the WSUT approach to meta-analysis summarized above
(Bergstrom and Taylor 2006), the final models are specified based on guidance from theory and prior
literature. For example, Arrow et al. (1993) made a fundamental distinction between discrete choice and
open-ended payment mechanisms (such as iterative bidding and payment cards). Hence, this distinction is
made in the final model (i.e., including the variable discrete_ch). Similarly, other "survey methodology"
variables in the model were chosen based on theoretical considerations and prior findings in the literature
(e.g., voluntary versus mandatory payment vehicles, parametric versus non-parametric, treatment of
protest and outlier bids, use of mean versus median WTP). Also included are variables characterizing the
scope and scale of the resource change, based on theoretical expectations that such factors should be
relevant to welfare estimates.
Few variables were excluded solely because of lack of statistical significance. Individual variables were
only excluded if they could not be shown to be statistically significant in any version of the model
(restricted or unrestricted), and there was no overriding rationale for retaining the variable in the model.
For example, variables distinguishing different types of discrete choice instruments (e.g., conjoint versus
dichotomous choice) added no significant explanatory power to the model (p = 0.44).
November 2008 E-13
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
It is important to note that although empirical considerations play a role in model development, certain
variables were retained in the model for theoretical reasons, even if significance levels were low. Such
specification of meta-analysis models using a combination of theoretical guidance and empirical
considerations is standard in modeling efforts (Bergstrom and Taylor 2006).
Table E-3 presents results of the total WTP regression model.
E.2.3 Interpretation of Total WTP Regression Analysis Results
Regression results reveal strong systematic elements influencing WTP. The analysis finds both
statistically significant and intuitive patterns that influence WTP for surface water quality improvements.
In general, the statistical fit of the three estimated equations is good; model results suggest a considerable
systematic component of WTP variation that allows forecasting of WTP based on site and study
characteristics. Likelihood ratio tests (Table E-3) show that model variables are jointly significant at
p<0.0001 in both cases. In both models, the majority of independent variables are statistically significant
at p<0.05, with most statistically significant at p<0.01. Signs of significant parameter estimates generally
correspond with intuition, where prior expectations exist. As shown in Table E-3, the random effect is
statistically insignificant (i.e., study level covariance factors are zero). Considering these factors, the
statistical performance of both models compare favorably to prior meta-analyses in the valuation
literature.
Table E-3: Estimated Multilevel Model Results for the Trans-log and Semi-log Total WTP
Regression Models: WTP for Aquatic Habitat Improvements
Variable
Intercept
year indx
Discrete
Volunt
Nonparam
Income
WQI
outlier bids
single river
single lake
multiple river
regional fresh
salt_pond
num riv pond
Mr
Mp
Nonusers
Allmult
Nonspec
Quality ch (Inquality ch
in trans-log model)
fish use
Fishplus
baseline (Inbase in trans-
log model)
Mail
Trans-log Model
Parameter Estimate
5.58023
-0.077413
-0.1198
-1.32553
-0.67323
2.47 x Id'6
-0.3238
-0.87363
-0.39713
-0.0575
-1.45543
0.1773
1.02513
0.11553
-0.87843
1.51143
-0.40933
-0.37552
-0.38491
0.43593
-0.34783
0.43432
0.0317
-0.2198
Standard Error
1.0452
0.02074
0.225
0.1611
0.1386
3.9 xlO'6
0.2693
0.3028
0.1286
0.2333
0.3768
0.1587
0.3718
0.02859
0.1783
0.2968
0.1303
0.1597
0.2012
0.1471
0.1165
0.1671
0.1177
0.1559
Semi-log Model
Parameter Estimate
5.9S923
-0.058993
-0.153
-1.29373
-0.69193
-6.10X10'7
-0.37091
-0.71083
-0.2251
-0.1391
-1.44853
0.26982
1.21633
0.1 1043
-0.80143
1.35343
-0.45123
-0.40733
-0.2419
0.034053
-0.45053
0.37813
0.00577
-0.3453
Standard Error
0.5566
0.01755
0.1673
0.155
0.1122
3.35 xlO'6
0.2076
0.2133
0.1162
0.2161
0.3086
0.129
0.3181
0.02346
0.1226
0.2042
0.1278
0.1189
0.155
0.006313
0.08377
0.1108
0.003655
0.1199
November 2008
E-14
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
Table E-3: Estimated Multilevel Model Results for the Trans-log and Semi-log Total WTP
Regression Models: WTP for Aquatic Habitat Improvements
Variable
lump sum
non reviewed
Median WTP
-2 Log Likelihood
Trans-log Model
Parameter Estimate
0.54682
-0.27571
-0.5143
134.3
Standard Error
0.2364
0.1653
0.188
Semi-log Model
Parameter Estimate
0.70393
-0.40073
-0.43992
115.4
Standard Error
0.1728
0.1205
0.1986
Covariance Factors:
Study Level (ou) 0
Residua]
Source
(ae) 0.1883
USEPA analysis
0
0.1597
2p>|t| <0.05
3 p> 1 1| <
0.01
Despite differences in the functional form of the two models, statistical results are robust across models.
In most cases, coefficient magnitudes and standard errors vary to only a small degree. Measures of
equation fit are similar, and both models are significant atp<0.0001. Such results mirror those of
Johnston et al. (2005), whose earlier meta-analysis of WTP for water quality improvements finds a high
degree of robustness to changes in model specification.
The initial discussion emphasizes results of the trans-log model. Although policy implications of both
model specifications are nearly identical for moderate to relatively large water quality improvements
(e.g., more than 5 percent increase in WQI), the trans-log model provides more accurate WTP estimates
when the expected water quality change is very small (e.g., less than 1 percent increase in WQI).
One of the primary means to assess the validity of benefit transfers—and the only one that may be applied
in cases wherein the true value for the study site is unknown—is value surface tests (Bergstrom and De
Civita 1999). These tests involve assessments of ways in which "different factors may cause values to
vary across sites, providing guidance for adjustments needed to make a valid transfer of value estimates
from the study site(s) to the policy site" (Bergstrom and De Civita 1999). Following general approaches
for such value surface assessments—which generally involve comparisons of empirical patterns found via
meta-analysis to theoretical expectations or norms—the Agency concluded that most results of the
estimated value surface suggest an appropriate benefit function. Results of these value surface
assessments are detailed in the following sections.
Resource Improvement Effects
Seven variables characterize resource improvements and uses; most are of the expected sign. The
coefficient on the quality_ch variable is positive and statistically significant (p<0.01), indicating that
larger water quality improvements generate larger WTP. This is an important result, and indicates that
WTP is sensitive to the scope of water quality improvements. The estimated model showed that WTP
values are not sensitive to the baseline water quality from which water quality change would occur. The
estimated parameter on the variable baseline representing the baseline water quality from which water
quality change would occur is not statistically significant (p>0.1). This finding differs from that of
Johnston et al. (2005), which shows that WTP for marginal water quality improvements declines as
baseline water quality improves. Here, the value is positive but is only significant at p<0.12 in the semi-
log model. In the trans-log model, this parameter is not statistically significant. The reason for this result
is unknown, but may be related to highly valued uses that are associated with larger values on the WQI.
For example, increases beginning at higher levels on the WQI may cross thresholds allowing such highly
November 2008
E-15
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
valued uses as swimming and drinking, such that increases at these high-quality levels may be valued
more highly than otherwise similar changes at lower baselines, which may not allow such uses. Such
thresholds, or other non-convex preference patterns, may lead to unexpected results for the baseline water
quality variable (baseline).
Both models reveal that water quality changes associated with recreational fishing uses lead to a
significant decrease in total WTP values, compared to improvements that do not affect fishing. The
variable fish_use identifies those studies that specified effects of water quality improvements on
recreational fishing (e.g., increase in catch rates). The associated parameter estimate is significant
(p<0.01) and has the negative sign. In contrast, the variable fishplus identifies those studies for which the
associated survey identified particularly large gains in fish populations or harvest rates (>50 percent). The
positive and statistically significant result (p<0.01) indicates that large gains in fish populations or
harvests are associated with statistically significant increases in total WTP. Results suggest that while
water quality improvements targeting fishing uses may not be valued particularly highly on average, very
large resultant improvements in fish populations or harvests are associated with increases in WTP, ceteris
paribus.
The variables alljnult and nonspec indicate that water quality improvements affect multiple species
(alljnulf) or unspecified species (nonspec), respectively. The default category from which these variables
allow systematic variations in WTP is a focus on particular aquatic species affected by water quality (e.g.,
shellfish or game fish). The associated coefficients are negative, indicating that WTP is lower when a
survey instrument does not specify what aquatic species would be affected (nonspec) or when all or
multiple species are affected (alljnult). The latter finding seems to be counterintuitive at first. However,
when a survey instrument focuses on the effect of water quality on particular species, it is a likely
indication that these effects are of significant concern to the affected communities, which typically leads
to a higher WTP. That is, this result suggests that WTP is higher when water improvements can be shown
to offer targeted benefits to specific, and often high-profile, species groups—as opposed to cases in which
improvements benefit an often poorly characterized group of species.
Geographic Region and Scale Effects
Ten binary variables characterize geographic region and scale; seven are statistically significant at
p<0.10. The default category from which these variables allow systematic variations in WTP is an
estuarine water body. Also included in this default are a small number of observations addressing national
level improvements. Compared to this baseline, WTP associated with rivers is lower (single_river and
multiple _river both have negative and significant values). Single Jake has a negative value, but it is not
significant. WTP for water quality gains in salt ponds (salt_pond) is higher than for estuaries (p<0.05).
This is not surprising since water quality gains in salt ponds correspond to an increase in the number of
acres of shellfish beds.
Of particular importance for the general validity of empirical findings, the model results further suggest
that WTP is sensitive to the number of water bodies under consideration and geographic scale of
improvement (regional_Jresh). Of the water body categories distinguished above, both rivers and salt
ponds allowed variation in numbers of affected water bodies explicitly described by the survey. This
variation is captured by the variable num_riv_pond (see Table E-3).25 The associated parameter estimate
Technically, this variable is the sum of two interaction variables: (1) an interaction between multiple_river and the
number of water bodies noted in the survey (0 if unspecified) and (2) an interaction between salt_pond and the number of water
bodies noted in the survey (0 if unspecified).
November 2008 E-16
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
is statistically significant (p<0.01) and indicates that WTP increases with the number of water bodies
considered. The parameter estimate on the regionaljresh variable is positive and significant (p<0.05) in
the semi-log model, indicating that large-scale regional water quality improvements lead to an increase in
WTP. These results, combined with the statistical significance of the water quality change variables noted
above, suggest that WTP values (in this case for water quality improvements) are strongly sensitive to
scope, both in terms of the number of water bodies considered, geographic scale of improvement, and the
magnitude of water quality change. In the trans-log model, the regionaljresh variable is positive but not
statistically significant (p>0.26).
Finally, the regional indicator variables mp and mult_reg are statistically significant at p<0.01, suggesting
that there are significant differences among WTP estimates from surveys in different geographical regions
of the United States. The parameter estimate on the mult_reg variable is negative, indicating that WTP for
non-local water quality improvements (e.g., out of state) are lower compared to in-state or local resource
improvements. This is consistent with prior findings that WTP for water quality improvements declines
with the distance from the resource (Bateman et al. 2006). The magnitude of the Pacific Mountain (mp)
regional effect suggests that spurious or otherwise unexplained effects (e.g., the effect of specific
researchers who appear more than once in the data) may drive their overall magnitude. For example, the
size of the positive parameter estimate associated with WTP in the Pacific Mountain dummy (mp) leads in
many cases to relatively large increases in WTP for Pacific Northwest policies. Hence, EPA believes that
particular, spurious, or unexplained aspects of studies from this region may have caused the associated
parameter estimate to have a larger-than-expected influence on WTP. Although effects of regional
dummy variables often escape simple, intuitive characterization, EPA notes that they are often
statistically significant in meta-analysis found in the valuation literature. Similar issues are found by
Johnston et al. (2005), for example.
Surveyed Populations Effects
Only two variables, nonusers and income, are used to characterize surveyed populations. In particular, the
nonusers variable is of substantial policy relevance. The negative and strongly significant (p<.0001)
parameter estimate indicates that surveys of non-users only, who by definition have only non-use values
for the resource improvements in question (cf Freeman 2003, p. 142), generate lower WTP values than
surveys that include users, who may have both use and non-use values. Based on this statistically
significant result, it is possible to use this model to estimate non-use values, interpreted as the mean WTP
values estimated by surveys of non-users only. Such methods, however, may underestimate non-use
values of the general population, if the non-use values of users exceed those of non-users (Whitehead and
Blomquist 1991a,b).
The income parameter estimate is positive in the trans-log model, as expected, but is not statistically
significant. Such lack of statistical significance for income parameters is not uncommon in meta-analyses
found in the literature (e.g., Johnston et al. 2005).
Study and Methodology Effects
As often found in meta-analyses within the valuation and benefit transfer literature (Navrud and Ready
2007), a variety of study and methodology effects can be shown to influence WTP for water quality
improvements. While expected, this does indicate that the methodological approach influences WTP, as
argued by Arrow et al. (1993). Of nine variables characterizing study and methodological effects, eight
are statistically significant at p<0.10. Among these is the year in which a study was conducted (yearjndx,
a continuous variable), with later studies associated with lower WTP. This is the expected result, as the
November 2008 E-17
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
focus of survey design overtime has often been on the reduction of survey biases that would otherwise
result in an overstatement of WTP (Arrow et al. 1993).
Model results reveal that voluntary (voluntary=\) payment vehicles (i.e., surveys that describe
hypothetical payments as voluntary) are associated with reduced WTP estimates. This result counters
common intuition and empirical findings that voluntary payment vehicles are associated with
overstatements of true WTP (Carson et al. 2000). The reason for this counter-intuitive finding is
unknown, but may reflect an unwillingness among respondents to offer large voluntary payments, given
the fear that others will free-ride (Johnston et al. 2005). Reduced WTP estimates are also associated with
studies applying non-parametric methods to WTP estimation (nonparani). Survey elicitation method does
not have a strong effect in this model; studies using discrete choice formats have lower WTP values, but
this difference is not statistically significant.
Smaller WTP estimates are associated with studies that eliminate or trim outlier bids when estimating
WTP (outlier_bids=\; p<0.01). Studies that report median WTP (median WTP;p<0.05) have lower WTP
values.
Lower WTP is associated with the use of the WQI in the original survey (WQI=\). This parameter is,
however, significant in the semi-log model only (p<0.1). As is the case with a variety of study design
variables, there is no necessary expectation with respect to the direction of this effect. Nonetheless, this
finding might suggest the capacity of such scales to clarify the specific magnitude and implications of
water quality change, and hence (perhaps) reduce methodological misspecification or symbolic biases that
might act to systematically inflate estimated WTP.
Survey format variables also have an effect on WTP, as might be expected. Mail has a negative and
statistically significant coefficient (p<0.01) in the semi-log model, compared to the default of telephone
surveys or interviews. This parameter is negative, but not statistically significant (p>0.16) in the trans-log
model. It may be possible that the interview and telephone survey format results in larger WTP values
either because the respondents are better able to understand the valuation scenario, or because
respondents may feel pressure from interviewers to bias their WTP estimates upward. Finally, studies that
ask respondents to report an annual payment (as opposed to a lump_sum payment) have higher WTP
estimates (p<0.05). This likely to reflect the fact that annual payments are regarded as an infinite
contribution and may reflect the respondent's uncertainly regarding his future income and budget
constraints.
E.2.4 Model Selection
To select the model for estimating benefits of water quality improvements from the C&D regulation, EPA
calculated WTP values for a range of WQI changes using both semi-log and trans-log models. In all cases
the baseline WQI is set to 50, which approximates the average WQI value across all RF1 reaches in the
United States. EPA assigned values to other independent variables corresponding with theory,
characteristics of the water resource, and the policy context. Table 10-8 in Chapter 10 provides a
complete list of values assigned to the remaining independent regressors.
As shown in Table E-4, both the semi-log and trans-log models yield similar WTP for water quality
changes greater than five points as measured by WQI. However, the semi-log model is not sensitive to
very small water quality changes (i.e., changes less than one point on the WQI index). Because the
expected water quality changes from the C&D regulation are relatively small, the Agency selected the
November 2008 E-18
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
trans-log specification for estimating benefits from reducing sediment runoff from construction sites, as a
more conservative option.
Table E-4: Comparison of WTP for Different Changes in WQI Based on Semi-log and
Trans-log Models
Model
Semi-log
Trans-log
20 Point Change
in WQI
20.6584129
22.8032699
10 Point Change
in WQI
14.6967
16.8814684
5 Point Change in
WQI
12.39599
12.50388
1 Point Change in
WQI
10.81758
6.247071
0.1 Point Change
in WQI
10.4911
2.349331
Semi-log Error Term = 0.07985
Trans-log Error Term = 0.09415
E.3 Model Limitations
The validity and reliability of benefit transfer—including that based on meta-analysis—depends on a
variety of factors. While benefit transfer can provide valid measures of use and non-use benefits, tests of
its performance have provided mixed results (e.g., Desvousges et al. 1998; Vandenberg et al. 2001; Smith
et al. 2002; Shrestha et al. 2007). Nonetheless, benefit transfers are increasingly applied as a core
component of benefit cost analyses conducted by EPA and other government agencies (Bergstrom and De
Civita 1999; Rosenberger and Phipps 2007). Moreover, Smith et al. (2002, p. 134) argue that "nearly all
benefit cost analyses rely on benefit transfers, whether they acknowledge it or not." Given the increasing
[or as Smith et al. (2002) might argue, universal] use of benefit transfers, an increasing focus is on the
empirical properties of applied transfer methods and models.
Although the statistical performance of the model is good, EPA notes several limitations of the model.
These limitations stem largely from information available from the original studies, as well as degrees of
freedom and statistical significance. An important factor in any benefit transfer is the ability of the study
site or estimated valuation equation to approximate the resource and context under which benefit
estimates are desired. As is common, the meta-analysis model presented here provides a close but not
perfect match to the context in which values are desired. Although all of the studies used in the meta-
analysis valued changes in water quality improvements, many studies did not specify the cause of water
quality impairment in the baseline or focused on causes (i.e., nutrients) that are different from the
pollutant of concern in the C&D regulation (i.e., sediment). Preliminary models, however, suggest no
systematic patterns in WTP associated with such factors, at least in the present metadata.
Additional limitations relate to the paucity of demographic variables available for inclusion in the model.
The only demographic variable incorporated in the analysis (income) was not statistically significant.
Moreover, other demographic variables are unavailable.
The estimated model is statistically significant and allows estimation of WTP based on study and site
characteristics. However, strictly speaking, model findings are relative to the specific case studies
considered, and must be viewed within the context of the 115-observation data set, with all the
appropriate caveats. Although this represents a fairly standard-to-large sample size for a meta-analysis in
this context, it is relatively small relative to other statistical applications in resource and environmental
economics. Model results are also subject to choices regarding functional form and statistical approach,
although many of the primary model effects are robust to reasonable changes in functional form and/or
statistical methods. The rationale for the specific functional form chosen here (the semi-log form) is
detailed above.
November 2008
E-19
-------
Environmental Impact and Benefits Assessment for the C&D Regulation
As in all cases, results of the meta-analysis are dependent on the sample of studies available for the given
resource change (Navrud and Ready 2007), and may be subject to various selection biases if the available
literature does not provide a representative, unbiased perspective on welfare estimates associated with
resource changes (Rosenberger and Johnston 2007). In this case, however, the Agency took various steps
to ameliorate such potential biases, including the incorporation of both peer-reviewed and gray literature
to avoid possible publication biases (Rosenberger and Johnston 2007), and the use of a comprehensive
literature review in the attempt to avoid—as much as possible—other types of selection biases.
The relatively large (positive) magnitude of the parameter estimate for the Pacific Mountain U.S. regional
dummy variable (mp) leads EPA to question the appropriate interpretation of this effect. While it is
theoretically possible that WTP for water quality changes is substantially higher in the Pacific Northwest,
the magnitude of the effect suggested by the model seems unlikely from an intuitive perspective. As
suggested above, it is possible that spurious, unexplained factors influence the magnitude of this
parameter in the present model. However, assessments of preliminary model runs suggest that this effect
is relatively robust given the present data and selection of variables available. Nonetheless, EPA
recommends that the magnitude of the predicted shift in WTP associated with the Pacific Mountain
region should be viewed with caution.
Finally, as noted above, there is some controversy over the appropriateness of meta-analysis for benefit
transfer. While recognizing this controversy, the Agency emphasizes that the broader literature provides
support for the use of various types of meta-analysis for benefit transfer (cf. Bergstrom and Taylor 2006;
Shrestha et al. 2007; USEPA 2000b).
Based on the results presented in Table E-3, EPA estimates WTP for water resource changes as a function
of resource, regional, and study design attributes (see Chapter 10). This, in general, provides a superior
alternative to the calculation and use of a simple arithmetic mean over the 115 observations, as it allows
WTP to be adjusted to account for the characteristics of the transfer site. The ability of the model to
appropriately adjust WTP is suggested by the many systematic (statistically significant) patterns revealed
by the meta-analysis regression model. Nonetheless, the use (and interpretation) of such WTP estimates
for benefit transfer is subject to the constraints and concerns expressed elsewhere in the literature (e.g.,
Rosenberger and Johnston 2007; Desvousges et al. 1998; Poe et al. 2001; Vandenberg et al. 2001; Navrud
and Ready 2007).
November 2008 E-20
------- |