United States
Environmental Protection
Agency
Estimation of National Economic Benefits
Using the National Water Pollution Control
Assessment Model to Evaluate Regulatory
Options for Concentrated Animal Feeding
Operations
December 2002
-------
U.S. Environmental Protection Agency
Office of Water (4303T)
1200 Pennsylvania Avenue, NW
Washington, DC 20460
EPA-821 -R-03 -009
-------
Estimation of National Economic Benefits
Using the National Water Pollution Control Assessment
Model to Evaluate Regulatory Options for Concentrated
Animal Feeding Operations
Christine Todd Whitman
Administrator
G. Tracy Mehan III
Assistant Administrator, Office of Water
Sheila E. Frace
Director, Engineering and Analysis Division
Linda Chappell
Economist
Lisa McGuire
Environmental Scientist
Christopher Miller
Economist
Engineering and Analysis Division
Office of Science and Technology
U.S. Environmental Protection Agency
Washington, D.C. 20460
December 2002
-------
ACKNOWLEDGMENTS AND DISCLAIMER
This report was prepared by RTI International under the direction and review of the
Office of Science and Technology.
Neither the United States government nor any of its employees, contractors,
subcontractors, or other 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 report,
or represents that its use by such a third party would not infringe on privately owned
rights.
-------
Table of Contents
Table of Contents
Section Page
Executive Summary ES-1
1.0 Introduction 1-1
1.1 Background 1-1
1.2 Focus of Report 1-1
1.3 Report Overview 1-2
2.0 NWPCAM System 2-1
2.1 Spatial and Environmental Databases 2-4
2.1.1 Hydrologic Routing File 2-5
2.1.2 Land-Use/Land-Cover File 2-6
2.1.3 RF3 Hydrologic Data 2-8
2.1.4 Overland Transport Hydrologic Data 2-14
2.1.5 PS Loadings Data Set 2-16
2.1.6 NPSs (Non-AFO) Loadings Data Set 2-16
2.2 Kinetics 2-20
2.2.1 Carbonaceous Biochemical Oxygen Demand 2-21
2.2.2 Nitrogen Species 2-22
2.2.3 Phosphorus Species 2-23
2.2.4 Dissolved Oxygen 2-23
2.2.5 Total Suspended Solids (TSS) 2-28
2.2.6 Pathogens 2-29
2.2.7 Other Processes 2-29
3.0 AFO/CAFO Modeling Process 3-1
3.1 Changes to NWPCAM Since the Proposed Rulemaking 3-1
3.2 AFO/CAFO Input Files 3-3
3.3 Methodology 3-4
3.3.1 Method for Distributing AFO/CAFO Loadings 3-5
3.3.2 Routing AFO/CAFO Loads to RF3 Reaches 3-7
3.3.3 Routing AFO/CAFO Loads to RF3Lite Reaches 3-7
3.3.4 In-Stream Modeling in the RF3Lite network 3-7
3.3.5 Water Quality Assessment Ladder 3-7
3.3.6 Economic Benefits Calculations Using the WQL 3-8
3.3.7 Water Quality Index 3-9
3.3.8 Economic Benefits Analysis Using the WQI6 3-10
4.0 Results of AFO/CAFO Analyses 4-1
4.1 AFO/CAFO Loadings 4-1
4.2 Economic Benefits 4-3
4.3 Discussion of Benefit Results 4-4
in
-------
Table of Contents
Table of Contents (continued)
Section Page
5.0 Quality Assurance 5-1
5.1 Reviewing Hydrologic Inputs 5-1
5.2 Model Robustness 5-1
5.3 Data Processing 5-2
5.4 Modeling Results 5-2
6.0 References 6-1
List of Figures
Figure
1 Overview of AFO/CAFO benefits analysis process 2-2
2 Comparison of drainage area estimates between NWPCAM and HCDN 2-10
3 Map of the hydroregions from USGS 2-11
4 Average annual runoff by cataloging unit 2-12
5 USGS runoff contour map showing average annual runoff for the conterminous
48 states 2-13
6 Predominant ecoregions in cataloging units 2-18
7 Mosaic Composite of Spatial Data at the Watershed (HUC) Level 3-5
8 Total nitrogen loadings (g/s) by county FIPS code for baseline 5-3
List of Tables
Table
1 NWPCAM 1.6 Modules 2-2
2 Data Elements of NWPCAM 1.6 2-4
3 Key Fields of the RF3 Routing Data File 2-6
4 Key Fields of the Land-Use/Land-Cover Data File 2-7
5 Anderson and NWPCAM Land-Cover Classes 2-7
6 Key Fields of NWPCAM 1.6 Hydrologic Data 2-9
7 Refined Hydrology in the Western Hydroregions 2-11
8 Total Loadings for PSs 2-16
9 Speciation Factors for Nitrogen and Phosphorus 2-17
10 Total Loadings for PSs at the RF3Lite Scale 2-18
11 Total Loadings on Land-Use/Land-Cover Cells for NPSs 2-20
12 Total Loadings for NPSs at the RF3 and RF3Lite Scale 2-21
13 CBODU:BOD5 Ratios for PSs 2-23
14 SPARROW Coefficients 2-23
15 Kinetic Parameters for Nitrogen Species 2-24
iv
-------
Table of Contents
List of Tables (continued)
Table
16 Depth (H) and Velocities (U) Ranges, Reaeration Formulations, and Coefficients
for Owens et al., Churchill, and O'Connor-Dobbins 2-28
17 Summary of Differences in NWPCAM Versions Used for the Proposed
Rulemaking and Final Rulemaking 3-1
18 Water Quality Ladder Threshold Concentrations 3-7
19 Original and Revised Weights for WQI Parameters 3-10
20 National AFO/CAFO Loadings on Agricultural Cells 4-1
21 AFO/CAFO Nutrient/Pollutant Loadings to RF3 Rivers/Streams 4-2
22 AFO/CAFO Delivery Ratios to the RF3 Network 4-2
23 AFO/CAFO Nutrient/Pollutant Loadings to RF3Lite Network 4-2
24 AFO/CAFO Delivery Ratios to the RF3Lite Network 4-2
25 Annual Economic Benefits Using the WQL 4-3
26 Annual Economic Benefits Using the WQI 4-3
27 Verification of Loads Distribution Module 5-2
-------
Executive Summary
Executive Summary
One goal of the Clean Water Act (CWA) is to improve water quality conditions of the
nation's waters to attain "fishable and swimmable" status nationwide. In support of this goal, the
U.S. Environmental Protection Agency (EPA) is revising the National Pollutant Discharge
Elimination System (NPDES) program regulations and the effluent limitation guidelines (ELGs)
for concentrated animal feeding operations (CAFOs). Changes to the NPDES regulations affect
which animal feeding operations (AFOs) are considered CAFOs and are therefore subject to the
NPDES permit program. Changes to the ELG determine what technology-based requirements
apply to CAFOs.
RTI applied the National Water Pollution Control Assessment Model (NWPCAM) to
estimate national economic benefits to surface water quality resulting from implementation of
new regulations for CAFOs (including revision of NPDES permit regulations and the ELGs for
CAFOs). NWPCAM is a national-scale water quality model for simulating the water quality and
economic benefits that result from various water pollution control policies. NWPCAM is
designed to characterize water quality for the nation's network of rivers and streams and, to a
more limited extent, its lakes. NWPCAM is able to translate spatially varying water quality
changes resulting from different pollution control policies into terms that reflect the value
individuals place on water quality improvements. In this way, NWPCAM is capable of deriving
economic benefit estimates for scenarios for regulating CAFOs.
Economic benefits associated with the various AFO/CAFO scenarios are calculated using
two estimation methods. The first is based on the Vaughn Water Quality Ladder (WQL), which
calculates changes in water quality use-support (i.e., boatable, fishable, swimmable) and the
population benefitting from the changes. The second method uses a six-parameter Water Quality
Index (WQI6), which represents a composite measure of water quality. Benefits are calculated
for each state at the local and nonlocal scales. Local benefits represent the value that a state
population is willing to pay for improvements to waters within the state. Nonlocal benefits
represent the value that a state population is willing to pay for improvements to waters in all
other states in the conterminous 48 states.
Based on the WQL estimation method, the sum of local and nonlocal benefits represented
a total willingness to pay (WTP) of $102 to $166 million (2001 dollars). Using the WQI6
estimation method, the sum of local and nonlocal benefits represented a total WTP of $182 to
$298 million (2001 dollars).
ES-1
-------
Section 1.0
Introduction
1.0 Introduction
1.1 Background
Enactment of PL 92-500 in 1972, known as the Clean Water Act (CWA), established a
national water pollution control policy based on technology-driven effluent standards for
industrial wastewaters and a minimum level of secondary treatment for municipal wastewaters
discharged to surface waters. The goal of the CWA was to improve water quality conditions of
the nation's waters to attain "fishable and swimmable" status nationwide. The CWA requires
that all point sources (PSs) discharging pollutants into U.S. waters obtain a permit under the
National Pollutant Discharge Elimination System (NPDES) program. The purpose of the
NPDES program is to protect human health and the environment by controlling the types and
amounts of pollutants that can be discharged into U.S. waters. NPDES permits implement a
multifaceted approach to protecting water quality. At the core of these permits is a two-pronged
pollution control strategy that incorporates both technology-based effluent limitation guidelines
(ELGs) and more stringent site-specific limits based on water quality considerations.
The U.S. Environmental Protection Agency (EPA) is revising the NPDES regulations for
concentrated animal feeding operations (CAFOs) and the ELGs for feedlots. Although similar
changes are being considered regarding both regulations, the effects of such changes are
different under each. Proposed changes to the NPDES regulations for CAFOs affect which
animal feeding operations (AFOs) are considered CAFOs and are therefore subject to the
NPDES permit program. Changes to the ELG regulations for feedlots determine the technology-
based requirements that apply to CAFOs.
1.2 Focus of Report
This report presents the findings of modeling efforts conducted by RTI that were
designed to estimate national economic benefits to surface water quality resulting from
implementation of various rulemaking scenarios for regulating CAFOs. These scenarios include
revision of both NPDES permit regulations and ELG regulations for CAFOs. Regulatory
scenarios assessed include the following:
Baseline - Current regulations (AFOs are considered CAFOs if certain
criteria are met.)
RTI Scenario 1 - ELG P-based, >1,000 animal units (AU) (AFOs with > 1000 AU
are considered CAFOs, whereas AFOs with < 1000 AU are not
affected. CAFOs are subject to 100 ft setback for manure
application and phosphorus-based requirements, as necessary.)
1-1
-------
Section 1.0
Introduction
RTI Scenario 2 - ELG N-based, > 1,000 AU (AFOs with > 1000 AU are considered
CAFOs, whereas AFOs with < 1000 AU are not affected. CAFOs
are subject to 100 ft setback for manure application and nitrogen-
based requirements, as necessary.)
The National Water Pollution Control Assessment Model (NWPCAM) version 1.6 was
used to conduct the water quality and economic benefits analyses. NWPCAM version 1.5 was
used during the CAFO proposed rulemaking process. Modifications made to NWPCAM 1.5 for
the CAFO proposed rule application are described in EPA (2000). Changes made to NWPCAM
1.6 since the proposed rulemaking process are described in Section 2 of this report. Changes to
the model since proposal include
¦ Revisions to the methodology used to distribute AFO/CAFO loadings to
agricultural land use cells;
¦ New reach network for in-stream modeling of pollutants;
¦ New methodologies for estimating hydraulic parameters;
¦ Updated inventories of point source and nonpoint source loadings; and
¦ Enhanced water quality kinetics.
1.3 Report Overview
Section 2.0 contains a detailed description of the NWPCAM 1.6 modeling system.
Section 3.0 outlines the processes associated with the AFO/CAFO modeling process. Section
4.0 presents results from the AFO/CAFO modeling process. Section 5 describes measures taken
to reduce the errors and uncertainties in the AFO/CAFO analysis. Section 5 describes measures
taken to reduce the errors and uncertainties in the AFO/CAFO analysis.
1-2
-------
Section 2.0
NWPCAM System
2.0 NWPCAM System
The NWPCAM is a national-level water quality model for simulating the water quality
and economic benefits that result from various water pollution control policies. NWPCAM is
designed to characterize water quality for the nation's network of rivers and streams and, to a
more limited extent, its lakes. NWPCAM incorporates a national-scale water quality model into
a system designed for conducting policy simulations and benefits assessments. NWPCAM is
able to translate spatially varying water quality changes into terms that reflect the value that
individuals place on water quality improvements. In this way, NWPCAM is capable of deriving
benefit estimates for a wide variety of water pollution control policies.
NWPCAM's water quality modeling system is suitable for developing water quality
estimates for virtually the entire inland portion of the country. Its national-scale framework
allows hydraulic transport, routing, and connectivity of surface waters to be simulated in the 48
conterminous states. The model can be used to characterize source loadings (e.g.,
AFOs/CAFOs) under a number of alternative policy scenarios (e.g., loadings with controls).
These loadings are processed through the NWPCAM water quality modeling system to estimate
in-stream pollutant concentrations on a detailed spatial scale and to provide estimates of policy-
induced changes in water quality. The model incorporates routines to translate water quality
concentration estimates into measures of "beneficial use attainment"—categories including
boating, fishing, and swimming—which are commonly used to characterize water quality for
policy purposes. The model also calculates a six-parameter water quality index (WQI6) that
provides a composite measure of overall water quality. Both the beneficial use attainment
categories and the WQI6 estimates allow for the calculation of economic benefits associated with
the estimated water quality improvements. NWPCAM can be used to assess both the water
quality impacts and the social welfare implications of alternative policy scenarios.
NWPCAM is an evolving system developed for EPA's Office of Water (OW) by RTI and
has been used in several applications to estimate the benefits of pollution control policies. An
adaptation of version 1.0 was used by OW's Office of Waste Management (OWM) to evaluate
the potential benefits of the Stormwater Phase II rulemaking (Bondelid, Ali, et al., 1999).
Version 1.1 (RTI, 2000b), the version developed in response to external peer review on version
1.0, is a complete system oriented toward evaluating the effects of PS controls. NWPCAM
version 1.1 was used in the recent Meat Processing Effluent Guidelines
(http://www.epa.gov/ost/guide/mpp/). Version 1.5 was used in the proposed AFO/CAFO rule
(RTI, 2000a). Version 1.6 was used in developing the final AFO/CAFO rule.
The foundation of NWPCAM 1.6 is the stream flow, transport, and flow-routing data
obtained from the EPA Reach File 3 (RF3) database and the U.S. Geological Survey (USGS)
Hydro-Climatic Data Network (HCDN). The RF3 database contains information about the
nation's network of rivers and streams in the United States. As a national-scale model,
NWPCAM's framework is limited to readily available national databases that can be accessed
2-1
-------
Section 2.0
NWPCAM System
and processed using automated input and output file management procedures. Waterbody types
currently included in NWPCAM 1.6 include free-flowing streams and rivers, lakes characterized
by inflows and outflows from streams and rivers run-of-river reservoirs, and tidal rivers. Large,
open water systems of estuaries (e.g., Chesapeake Bay), embayments (e.g., Waquoit Bay),
coastal waters (e.g., New York Bight, Southern California Bight), the Great Lakes, and other
large lakes (e.g., Lake Champlain) are not incorporated in the current framework of NWPCAM
1.6.
NWPCAM 1.6 consists of an Oralcle database that is manipulated through a series of
Visual Basic modules, listed in Table 1. These modules perform analytical or simulation
routines required for the overall modeling process. The modeling results are kept in Oracle tables
and may be exported to an Oracle export file or other formats, including Microsoft Access.
Figure 1 presents a simplified flowchart of the NWPCAM 1.6 process employed for
estimating the benefits of AFO/CAFO regulations using the Vaughn Water Quality Ladder
(WQL) approach. The left-hand column of Figure 1 represents the main processes, and the right-
hand columns represent integration of data and analytical modeling modules.
Table 1. NWPCAM 1.6 Modules
Preprocessing Routines
• Route and sequence RF3
• Define the RF3Lite subset for in-stream modeling
• Generate land-cover data set with routed and sequenced RF3
• Calculate slopes and sinuosity for RF3 reaches and land-use cells
• Route PS and combined sewer overflow (CSO) loads to the RF3Lite network
• Route NPS loads to the RF3 network
• Route NPS loads to the RF3Lite network
Modeling & Analysis Modules
• Distribute AFO/CAFO loads to agricultural cells using farm area and a random
distribution method
• Route AFO/CAFO loads to the RF3 network
• Route AFO/CAFO loads to the RF3Lite network
• Model transport, decay, and transformation of pollutants in RF3Lite
• Calculate WQI6 and overall use support for each RF3Lite reach
• Calculate national use support summary
• Calculate local and nonlocal economic benefits for each state using the Vaughn WQL
• Calculate local and nonlocal economic benefits for each state using the WQI6
2-2
-------
Section 2.0
NWPCAM System
RF3/RF3Lite
Reach File
Database
E
Route and Sequence
RF3/RF3Lite
1
'
Overlay Landcover
and Drainage Area
1
'
J Land-Cover /
/ Database /
Establish Stream/
Reach Discharges
and Flow and
Velocity Model
I
SPARROW
Develop Fate and
Transport Model
I
Tune Export
Coefficients
Export
Coefficient
Database
Add in NPS Nutrient
Loads
/ Nonmanure NPS /
/ Loads Database/
Add in Non-CAFO PS
Loads
/ PS Loads ~~7
/ Database /
Distribute AFO/ CAFOs
to Agricultural Cells
AFO/CAFO
Counts Data
Add in AFO/CAFO
Loads
I
Compute Agricultural
Cell Loadings for Each
Scenario
Deliver Loadings from
Agricultural Cell to
Reach/Stream Network
Edge-of-Field
Nutrient and
Pollutant Loads
and CAFO
Percentages Data
by Scenario
Slope/Distance
and Sinuosity Data
Route Loads
through RF3/RF3
Lite Network with
Fate/Transport
Models
Determine Water
Quality Use-Support
for each Scenario
State Populations
Determine Water
Quality Benefits
I
Calculate
Economic Benefits
at State/National
Level
Figure 1. Overview of AFO/CAFO benefits analysis process.
2-3
-------
Section 2.0
NWPCAM System
2.1 Spatial and Environmental Databases
NWPCAM 1.6 relies on several extensive data sets to support the analytical routines
developed to represent physical and chemical processes occurring within a watershed and along
river reaches. Primary databases include (1) RF3 hydrologic and reach routing information;
(2) land-use and land-cover information; (3) watershed and stream discharge information;
(4) NPS nutrient export coefficients and other loading data; (5) PS pollutant loading information;
and (6) AFO/CAFO loading information. Table 2 presents a listing of the principal data
requirements for NWPCAM 1.6.
Table 2. Data Elements of NWPCAM 1.6
Tsihlcs Report Section Source
RF3 routing data (RF3RCHID, level, sequence number,
stream order, routing parameters, open waters data, etc.)
2.2.1
U.S. EPA (2002a)
Land-use/land-cover data (land-use type, county Federal
Information Processing Standards (FIPS) code, nearest
reach and distance to nearest reach, sinuosity for each
reach)
2.2.2
USGS (2002a)
8-digit hydrologic unit code (HUCs) (elevation, slope,
discharge per km2 per HUC based on USGS data
drainage areas and discharges for watersheds)
2.2.3
ESRI (2000a)
USGS (2002b)
AFO facility counts by county FIPS code for each animal
operation type and size
3.2
U.S. EPA (2002b)
Percentage of AFOs considered as CAFOs by
rulemaking scenario and state FIPS code
3.2
U.S. EPA (2002b)
AFO/CAFO edge-of-field nitrogen and phosphorus
loadings by animal operation type and size
3.2
U.S. EPA (2002b)
AFO/CAFO edge-of-field pathogen indicator and
sediment loadings by animal operation type and size
3.2
U.S. EPA (2002b)
Speciation factors for AFO/CAFO nitrogen and
phosphorus loads
3.2
U.S. EPA (2002b)
Locations and loading data for industrial and municipal
PSs
2.2.5
U.S. EPA (2002c,
2002d)
Combined sewer overflow (CSO) locations and loading
data
2.2.5
U.S. EPA (1993)
NPs nutrient loadings based on land-use types and
SPAtially Referenced Regression On Watershed
(SPARROW) results
2.2.6
Smith et al. (1997)
Export coefficients for NPS of 5-day biochemical
oxygen demand and sediment based on land use
Appendix A
Table 1
Novotony and Olem
(1994)
State population and RF3Lite segment length
Appendix A
Table 2
ESRI (2000a; 2000b)
(continued)
2-4
-------
Section 2.0
NWPCAM System
Table 2. (continued)
Tables
Report Section
Source
Speciation factors for PS and NPS loadings
2.2.5
Metcalf & Eddy et al.
(1991)
Modeling coefficients
Appendix A
Table 3
U.S. EPA (1985)
Contribution of intermittent streams by hydroregion
RTI, 2001
RTI (2001)
Ratio of agricultural land slope to average slope by
hydroregion
Appendix A
Table 4
Calculated
Vaughn WQL threshold values
3.3.5
Vaughan (1986)
McClelland Water Quality Index (WQI)
3.3.7
McClelland (1974)
2.1.1 Hydrologic Routing File
The EPA Reach Files are a series of hydrologic databases that contain information on the
U.S. surface waters. The Reach File databases were created to establish hydrologic ordering, to
perform hydrologic navigation for modeling applications, and to provide a unique identifier for
each surface water feature (i.e., the reach code). Reach codes uniquely identify the individual
components of the nation's rivers and lakes. A reach represents a segment of a river or stream.
Several segments may be linked together to characterize the total length and properties of a
waterbody. The longer the river or stream, the more reaches are used to represent the full length
of the waterbody. RF3 contains tabular data tables for routing, as well as full GIS coverages for
mapping and data overlays. Currently, the best source for RF3 data is the BASINS Model
developed by OW and the EPA Office of Science and Technology (OST) (U.S. EPA, 2002e).
NWPCAM 1.6 uses the RF3 network to move water and pollutants from a point of origin
within the conterminous 48 states toward the major rivers and ultimately toward the discharge of
these waters, which usually is to the oceans.
The RF3 network in NWPCAM 1.6 includes 1,817,988 reaches totaling 2,655,437 miles
within the conterminous 48 states. The routing framework for Hydroregions 8 and 17 is only
available from the first version of the Reach File (RF1) and includes 11,937 reaches totaling
90,253 miles.
A subset of the RF3 network (referred to as RF3Lite) was developed for in-stream
modeling in NWPCAM 1.6. RF3Lite was defined as
¦ Streams >10 miles in length, and
¦ Small streams that are needed to connect streams >10 miles into a complete
network.
RF3Lite maximizes the information available from RF3, while limiting the computational
burden imposed on the full system. The RF3Lite network includes 577,068 reaches totaling
840,835 miles.
2-5
-------
Section 2.0
NWPCAM System
Hydrologic sequence numbers were assigned to RF3 reaches starting at the most
upstream reach of a watershed and moving down the stream network. A small percentage of
RF3 reaches were unable to be networked because they were (1) isolated reaches; (2) missed
because of a discontinuity in the Reach File; or (3) located in an area with artificial reaches,
channel, or swamp land. In the cases where a sequence number was not assigned to a reach, the
reach was considered to have no connectivity with the network and was removed from the
NWPCAM 1.6 database. Table 3 lists the key fields and field description of the RF3 routing data
file.
Table 3. Key Fields of the RF3 Routing Data File
l-iclri
Description
RF3RCHID
RF3 Reach ID
RUNSEQUENCE
Hydrologic sequence number
STRMORDER
Stream order
CU
8-digit catalog unit
JLEVEL
(Networked) stream junction level
NLEVEL
(Networked) stream level
SEGLENGTH
Segment length of the reach
SINU
Sinuosity
2.1.2 Land-Use/Land-Cover File
The USGS conterminous 48 states Land Cover Characteristics (LCC) Data Set version 2
forms the basis for the land-use/land-cover spatial coverage used by NWPCAM 1.6. The USGS
developed the LCC database by classifying 1990 National Oceanic and Atmospheric
Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) satellite time-
series images, with postclassification refinement based on other data sets, including topography,
climate, soils, and ecoregions (Eidenshink, 1992). The database is intended to offer flexibility in
tailoring data to specific requirements for regional land-cover information.
The raster image has a pixel size of 8-bit, representing an area of 1 km2. The image
contains 2,889 lines and 4,587 samples covering the conterminous 48 states. The projection of
the images is Lambert Azimuthal Equal Area. Based on this information, it was possible to
extract a specific area from the image into an ASCII file using a routine written in C. This
approach allowed portions of the image to be imported, reducing loading and processing time
compared to a full-image import into a commercial geographic information systems (GIS)
package. The ASCII file was used to generate a point coverage in Arc/Info, which was
converted to geographic coordinates in order to process it with existing RF3 coverages.
Each land-use cell was assigned to the nearest routed RF3 reach for subsequent drainage
area, stream discharge, and hydrologic routing purposes. Table 4 lists the key fields and field
description for the land-use/land-cover data file.
2-6
-------
Section 2.0
NWPCAM System
Table 4. Key Fields of the Land-Use/Land-Cover Data File
Field
Description
CELLID
Identification number assigned to land-use/land-cover cell for CAFO
NWPCAM study
ANDERSONLEVEL
Code describing type of land-use/land-cover for cell
AGCELLFLAG
Marker to designate agricultural land-use/land-cover cells
STCOFIPS
County FIPS code
DISTFT
Distance from cell centroid to nearest RF3 reach (feet)
RF3RCHID
Nearest RF3 reach ID
RNDID
Random number generated for agricultural cells
BOD5EXPORT
BOD5 export coefficient
TSSEXPORT
TSS export coefficient
The LCC originally contained 27 Anderson Land-Cover Classes. These were compressed
into eight land-use categories in the NWPCAM 1.6 system (see Table 5).
Table 5. Anderson and NWPCAM Land-Cover Classes
NWPCAM l.iind-
Co\er (hiss ( ode
(deri\ed)
NWPCAM
l.iind- Co\er
( iileiion
(deri\ed)
Anderson l.;ind-Co\er
< "hiss ( ode
Anderson l.iind-Co\er Class
( iileiion
1
Agriculture
1
Dryland Cropland and Pasture
1
Agriculture
2
Irrigated Cropland and Pasture
1
Agriculture
3
Mixed Dryland/Irrigated Cropland and
Pasture
2
Agriculture/
herbaceous
4
Grassland/Cropland Mosaic
3
Agriculture/
woodland
5
Woodland/Cropland Mosaic
4
Herbaceous
6
Grassland
4
Herbaceous
7
Desert Shrubland
4
Herbaceous
8
Mixed Shrubland/Grassland
4
Herbaceous
9
Chaparral
4
Herbaceous
10
Savanna
5
Forest
11
Northern Deciduous Forest
5
Forest
12
Southeastern Deciduous Forest
5
Forest
13
Western Deciduous Forest
(continued)
2-7
-------
Section 2.0
NWPCAM System
Table 5. (continued)
NWPCAM l.iind-
( o\er ( hiss Cock-
(domed)
NWPCAM
l.iiml- Co\cr
( iileiion
(dcri\cd)
Anderson I.;iiuI-Co\it
< "hiss Cock*
Anderson l.;ind-Co\er (hiss
C.iU'fion
5
Forest
14
Northern Coniferous Forest
5
Forest
15
Southeastern Coniferous Forest
5
Forest
16
Western Coniferous Forest
5
Forest
17
Western Woodlands
5
Forest
18
Northern Mixed Forest
5
Forest
19
Southeastern Mixed Forest
5
Forest
20
Western Mixed Forest
6
Waterbodies
21
Waterbodies
4
Herbaceous
22
Herbaceous Coastal Wetlands
5
Forest
23
Forested Coastal Wetlands
6
Barren
24
Barren or Sparsely Vegetated
5
Forest
25
Subalpine Forest
7
Tundra
26
Alpine Tundra
8
Urban (derived)
30
Urban
2.1.3 RF3 Hydrologic Data
Stream drainage area, discharge data, and related hydrologic data at the RF3 reach level
(RF1 for Hydroregions 8 and 17) are required for hydrologic routing and in-stream nutrient
transport and decay processes simulated by NWPCAM 1.6. After the RF3 routing system is
established, information regarding the hydrologic (e.g., flow rate) and the hydrodynamic
characteristics (e.g., channel depth, channel width, velocity) of each reach are incorporated into
the model framework.
The fate of a water quality parameter along a hydrologic network is driven by the time of
travel from one reach to the next and kinetic parameters. Time of travel is dependent on reach
velocity and segment length. Velocity depends on the flow rate and the channel geometry of the
reach. Consequently, the hydraulic routing process of the water quality model largely becomes a
system of accounting for discharges, stream geometry, velocity, and travel distances to derive the
time of travel. Table 6 lists principal hydrologic data used in NWPCAM 1.6.
2.1.3.1 Stream Drainage Area. To develop drainage areas for each RF3 reach, the
land-use coverage was overlain on the RF3 routing framework to associate each land-use cell
with the nearest RF3 reach. The number of cells assigned to each reach provides the
approximate drainage area in square kilometers for the specific RF3 reach. This value represents
the land that contributes direct runoff to the reach versus the runoff received from the immediate
upstream reach (i.e., the hydrologically routed runoff).
2-8
-------
Section 2.0
NWPCAM System
Table 6. Key Fields of NWPCAM 1.6 Hydrologic Data
l-iclri
Description
RF3RCHID
RF3 Reach ID
CU
8-digit catalog unit
DRAINAGE
Drainage area (km2)
CUMDRAIN
Cumulative drainage (km2)
UNITQ
Weighted-average unit discharge for the CU (cfs/km2)
Q
Discharge (cfs)
N
Manning's n (min = 0.025, max = 0.040)
SLOPE
Average slope in 8-digit catalog unit
WIDTH
Channel width (ft)
DEPTH
Channel depth (ft)
VELOCITY
Reach velocity (ft/s)
TOT
Time of travel (days)
Cumulative drainage area for an RF3 reach represents the land area associated with the
reach itself plus the land area of upstream reaches. The cumulative drainage area for a given
RF3 reach is calculated by hydrologically routing all reaches in the RF3 file according to the
routing sequence number and summing the reach-specific drainage areas as they are routed
through the system. For example, the cumulative drainage area of the most headwater reach of a
stream simply would be calculated from the land-use cells that are directly associated with that
reach. As the routing algorithm moves downstream, the cumulative drainage area for a specific
reach would be calculated as the area of the land-use cells that are directly associated with that
reach plus the drainage areas of each reach that is hydrologically upstream of the specific reach.
Validation exercises were conducted to verify the drainage area methodology. The
drainage area estimates derived from the land-use cells were compared with estimates of
drainage area derived from USGS stream gages. USGS stream gages in the HCDN were selected
for data comparisons because they represent relatively natural hydrologic conditions and are not
influenced by controlled releases from reservoirs. Only gages with a drainage area less than the
drainage area of its cataloging unit were selected. This ensured that the discharge data from the
same set of HCDN gages could be used for future discharge comparisons.
Each of the HCDN gages was assigned to the nearest RF3 reach based on geographic
coordinate information. The drainage area estimate from the HCDN gage was compared with the
drainage area estimate for the RF3 reach derived from the land-use/land-cover coverage. Several
outliers were initially observed, although further review of the data sets showed that either the
HCDN gage had been assigned to the wrong RF3 reach or the RF3 reach had been removed from
the RF3 data set because of incomplete data. The analysis indicated close agreement between
the two drainage area estimates in the eastern hydroregions (see Figure 2 for a comparison of
estimates in Hydroregion 7; R2 = 0.995).
2-9
-------
Section 2.0
NWPCAM System
7,000
6,000
5,000
4,000
~
v 2,000
DC
«
U.
QC
1,000
Comparison of USGS vs. RF3 Reach-Calculated Drainage Area for Hgdroregion 7 (Corrected)
1,000
2,000
3,000
4,000
5,000
6,000
7,000
USGS Drainage Area (fc*~2)
Figure 2. Comparison of drainage area estimates between NWPCAM and HCDN.
2.1.3.2 Discharge Data. Initial comparisons of drainage area were not as favorable in
the western hydroregions as in other areas. In the western areas, estimates of routed discharge
(i.e., streamflow) in NWPCAM 1.6 were generally greater than the HCDN gage values.
Consequently, an attempt was made to match NWPCAM discharge estimates with the HCDN
gage data by incorporating only a percentage of the runoff for intermittent stream reaches. A best
fit was selected based on the correlation between NWPCAM estimates and HCDN values (see
Table 7).
2.1.3.3 Runoff. Data from the full HCDN network of l,391gages were used to derive a
mean annual unit runoff (ft3/sec/km2) for each cataloging unit. HCDN gages are selected by
USGS as gages that best reflect natural runoff, as opposed to gages that include significant
human-influenced streamflows. In general, the HCDN gages require at least 10 years of records
to establish mean annual and mean summer flows. A 200-mi maximum search radius from the
centroid of the cataloging unit was used to identify the five nearest HCDN gages. In some cases,
less than five gages were available within the 200-mi search radius. Mean annual unit runoffs
were calculated using a weighted-average technique based on the distance of the HCDN gage
from the centroid of the cataloging unit. For each cataloging unit, a mean annual unit runoff was
calculated based on mean annual discharge for the HCDN gages. Total discharge for a reach
equals the sum of the discharge for the associated land-use cells plus the discharge originating
2-10
-------
Section 2.0
NWPCAM System
18
r
J
21
Figure 3. Map of the hydroregions from USGS.
Table 7. Refined Hydrology in the Western Hydroregions
Hydroregion
Contribution from Intermittent
Streams to Discharge Estimate
10
10%
11
50%
12
100%
13
10%
14
1%
15
1%
16
1%
17
NA*
18
50%
*NA: Not applicable because routing is based on RF1, so perennial/
intermittent flag is not available.
from upstream reaches. The resulting unit runoffs for each cataloging unit were converted to
inches of runoff (see Figure 4) and compared with the USGS runoff contour map for the
2-11
-------
Section 2.0
NWPCAM System
Figure 4. Average annual runoff by cataloging unit.
conterminous 48 states (see Figure 5). A visual comparison of the two maps indicates close
agreement between the two sets of data.
2.1.3.4 Velocity. RF3 reach velocity estimates were developed using time-of-travel
analyses (Jobson, 1996). Regression equations were developed to relate peak velocity to
drainage area, a dimensionless drainage area, slope, discharge, and a dimensionless relative
discharge. Jobson presents two velocity estimation methods: one that considers reach slope and
one that does not. Based on a comparison between the velocity estimates and an RF1 velocity
data set, the RF3 velocity estimates derived without the slope were determined to provide the
best match. For most hydroregions, the comparisons were favorable (R2 = 0.6-0.95; RTI, 2001),
suggesting that the velocity estimates provide a reasonable basis for use in water quality
modeling. The final regression equation used to estimate velocity was
VelA_pnos = 0.02 + (0.051 x (D'aa821) x (Q'a"a465) x (Q/DJ) (1)
where
VelA_pnos = velocity estimate without considering slope (m/s)
D'a = dimensionless drainage area
2-12
-------
Section 2.0
NWPCAM System
Average Annual Runoff (inches) - USGS
Figure 5. USGS runoff contour map showing average annual runoff for the
conterminous 48 states.
Q'a = dimensionless relative discharge
Q = reach discharge (m3/s)
Da = drainage area (m2).
2.1.3.5 Channel Properties. Because of changes in the process used to determine RF3
reach velocity, a new method was needed for estimating channel geometry. For single-line RF3
reaches, the methodology of Leopold and Maddock (1953) was reviewed to estimate reach width
and depth. Channel geometry and mean annual discharge data for more than 100 rivers/streams
reported by Leopold and Maddock were used to derive regression equations relating discharge to
stream width. The evaluations included consideration of data sets only for rivers with extended
periods of data (e.g., 5 years, 10 years, and 18 years). Analysis of the differences between
reported widths from the Leopold and Maddock data set and predicted widths from the
regression equations were completed. The analysis suggested that the following regression
equation provides the best estimates of stream width:
W = 5.25 x (Qa/V)0'5465 (2)
2-13
-------
Section 2.0
NWPCAM System
where
W = width of channel (ft)
Qa = mean annual discharge (cfs)
V = velocity (ft/s).
For open waters in RF3Lite (i.e., lakes and wide rivers), stream widths were estimated by
dividing the area by one-half of the total circumference. Both area and perimeter data are in the
RF3 data set for double-wide reaches. This method approximates the average width along the
open water lake or wide-river channel. This approach can understate or overstate average widths
based on the shape of the waterbody and the sinuosity of the shoreline. The width estimates in
the Upper Mississippi hydroregion (Hydroregion 7) were manually compared with maps and
were found to be in close agreement except for one outlier. However, the width estimate, even
for the outlier, was within acceptable bounds for modeling purposes.
The method for determining channel depth for all reaches is shown in Equation 3.
D= Q / (W x V) (3)
where
D = depth (ft)
W = channel width (ft)
Q = stream flow (ft3/s)
V = stream velocity (ft/s).
2.1.4 Overland Transport Hydrologic Data
The operational foundation for delivering NPS and AFO/CAFO loadings from land cover
cells to surface waters is the assumption that these source areas are hydrologically connected to
surface waters by surface flow in small channelized systems. NWPCAM 1.6 includes routines to
route loadings from land-cover cells to the RF3 network. These modules account for overland
transport, decay, and transformation as a function of travel time. Travel time is based on
distance from the center of the land-cover cell to the associated RF3 reach, and velocity, which is
calculated using standard open-channel flow assumptions. There is assumed to be one open
channel from the centroid of each land-cover cell to its nearest RF3 reach.
The flow rates in the open channels were approximated using the unit runoff values
estimated from the HCDN network. In addition to estimating flow volume, channel
characteristics were estimated as a prerequisite for estimating velocity and time of travel. A log-
log relationship between stream flow and channel width was developed based on Keup's
methodology (1985):
W = 5.27 x Q 0459 (4)
2-14
-------
Section 2.0
NWPCAM System
where
W = channel width (ft)
Q = discharge (stream flow in cubic feet per second [cfs]).
Channel depths were calculated based on the classic Manning's n formulation for channel
resistance analysis. Assuming a rectangular channel cross-section, the following formula was
used to calculate stream depth:
Yo = 0.79 (Q x n/(W x (S0) ) (5)
where
y0 = channel depth (ft)
Q = discharge (stream flow in cfs)
n = Manning's n roughness coefficient
W = channel width (ft) calculated above
S0 = channel slope (ft/ft) (for RF3Lite reaches).
A Manning's n of 0.10 was selected to represent weedy, windy, overgrown channels,
such as might be found on agricultural lands. This value for Manning's n is specific to cell-to-
RF3 routing.
Velocity in the channels was calculated as
V=Q/(Wxy0)xC (6)
where
V = velocity (ft/sec)
Q = the discharge (per km2) for the HUC
y0 = channel depth (ft)
W = channel width (ft)
C = conversion factor to reconcile units.
Time of travel to the nearest RF3 reach was calculated as
TOT= SL/(V x C) (7)
where
TOT = overland time of travel (days)
S = sinuosity (ft/ft)
L = distance from cell center to nearest reach (ft)
V = velocity (ft/sec)
C = conversion factor = 86,400.
2-15
-------
Section 2.0
NWPCAM System
Sinuosity varied on a hydroregion basis and was calculated as the 75th percentile value of
the sinuosities for the first-order stream RF3 reaches.
2.1.5 PS Loadings Data Set
Loading data from industrial facilities, municipal facilities, and combined sewer
overflows are included in NWPCAM 1.6. These sources are collectively referred to as PS of
pollution. Loading data for these sources were obtained from EPA's 1997 Permit Compliance
System (PCS), Clean Water Needs Survey (CWNS), and the Industrial Facilities Database (IFD).
The databases supplied information on effluent nitrogen, phosphorus, total suspended solids
(TSS), 5-day biochemical oxygen demand (BOD), and fecal coliform. CSOs were assigned
default values for total nitrogen (TN) and total phosphorus (TP) loadings (U.S. EPA, 2001).
Loads from 24,231 industrial facilities, 10,501 municipal facilities, and 617 combined sewer
overflows were included in the NWPCAM 1.6 system. Table 8 lists total loads included in
NWPCAM 1.6 for the PS category.
Table 8. Total Loadings for PSs*
I'CCit I
(olil'orni
Source Type
li()l)5 loitd
(»/s)
I N l()illl
(li/s)
Tl' Io\/s)
Municipal and
Industrial Facilities
58,829
23,460
5,628
214,964
1.96 x 1012
Combined Sewer
Overflows
3,707
643
248
13,613
1.87 x 1014
Total
62,536
24,103
5,876
228,577
1.89 x 1014
* i.e., industrial facilities, municipal facilities, and combined sewer overflows
All of the nutrient loadings from the PSs were speciated into inorganic and organic forms
based on treatment level (see Table 9; Metcalf & Eddy et al., 1991). CSOs that were not linked
to a municipal facility were speciated using the "Raw" factors.
In preparation for the AFO/CAFO modeling process, the PS loads were routed directly to
the RF3Lite network in preparation for further in-stream modeling. Table 10 lists the total PS
loads at the RF3Lite scale, along with delivery ratios for each component.
2.1.6 NPSs (Non-AFO) Loadings Data Set
AFO loadings to river reaches have typically been included in NWPCAM as part of NPS
loadings. In order to evaluate AFO/CAFO regulatory options, they must be a separate source
category. An approach was developed to estimate non-AFO/CAFO NPS loadings (Bondelid,
Dodd, et al., 1999).
Within each watershed, a simple export coefficient loading model was used to deliver
nutrients from NPSs to a reach. Export coefficients are empirical values that describe the loading
2-16
-------
Section 2.0
NWPCAM System
Table 9. Speciation Factors for Nitrogen and Phosphorus*
TlTilllllCIll l.l'M'l
AihiiiuTil Aihiinml
I'olliiliinl U.ilio Ksiw I'rinisirv I'rimsirv Sccondiirv Sccondiirv Tcrlisirv
TKN:TN
1
1
1
0.9
0.7
0.25
NH3:TN
0.6
0.615
0.615
0.67
0.186
0.138
TON:TN
0.4
0.385
0.385
0.23
0.514
0.112
(N02 + N03):TN
0
0
0
0.1
0.3
0.75
P04:TP
0.625
0.6
0.6
0.75
0.75
0.925
TOP: TP
0.375
0.4
0.4
0.25
0.25
0.075
*Speciation factors are on a mass basis (e.g., mg/L:mg/L)
Table 10. Total Loadings for PSs at the RF3Lite Scale
I'simmclcr
UO 1)5
IN
1 l>
TSS
1 IX
RF3Lite Load
(g/s or MPN/s)
30,149
14,979
3,600
128,885.9
4.32 x 1013
Delivery Ratio
0.48
0.62
0.61
0.56
0.23
of a given nutrient in terms of mass per unit time per unit area. The analytical specification for
export coefficients requires estimates of both the unit loading and the land area within a
catchment described in terms of different types or classes of land use or land cover. The
analytical model can be summarized as
L= E (ECn.An) (8)
where
L = loading to a reach (kg/yr)
ECn = export coefficient for category n (kg/ha/yr)
A„ = area draining to reach in land use category n (ha)
n = land-cover or -use category.
The principal data sources for this model are (1) the LCC data set; (2) empirically based
estimates of export coefficients derived from a national study (Reckhow et al., 1980); and
(3) model output from a national study of nutrient sources, transport, and in-stream flux (Smith
et al., 1997).
Nutrient loads for NPSs were computed by land-use type by ecoregion based on
SPAtially Referenced Regression On Watershed attributes (SPARROW), which is a statistical
modeling approach for estimating major nutrient source loadings at a reach scale based on
spatially referenced watershed attribute data (Smith et al., 1997). An optimization algorithm was
2-17
-------
Section 2.0
NWPCAM System
developed to estimate nonmanure loadings by comparing SPARROW nonmanure NPS estimates
for cataloging units with NWPCAM modeled outputs.
The first step in regional export coefficient estimation was to determine appropriate
ranges for different land-cover classes. The basis of export coefficient estimation was defined as
all cataloging units sharing the same predominant ecoregion (see Figure 6). For each ecoregion
within a hydroregion, export coefficients were estimated using a genetic optimization routine to
find a set of optimal coefficients. The criterion for optimization was minimizing the sum of
squared error between predicted (coefficient) and empirically based (SPARROW) cataloging
unit level data.
Predominant Ecoregions
in Cataloging Units
nsas Valley
I Blue Ridge Mountains
Cascades
Central Appalachian Ridges and Valleys
Centra I Ap pal ac hi ans
Central California Valley
¦aI Com Belt Plains
¦aI Great Plains
"aI Irregular Plains
"al 01 ka ho ma/Texas Plains
t Range
¦ado Plateaus
^ Driftless Area
st Central Texa s P lai ns
tern Cascades Slopes and Foothills
tern Corn Belt Plains
wards Plateau
3 Erie/Ontario Lake Hills and Plain
| Flint Mills
iron/Erie Lake Plains
enor Plateau
erior River Lowland
I Klamath Mountains
ic Coastal Plain
Mississippi Valley Loess Plains
Montana Valley and Foothill Prairie
Nebraska Sand H ills
North Cascades
North Central Appalachians
North Central Hardwood Forests
Northeastern Coastal Zone
Northeastern Highlands
Northern Appalachian Plateau and
Northern Basin and Range
Northern Glaciated Plains
a Glaciated Plain:
a Drift PI
I Sierra Ne
Snake River BasirVHigh Desert
South Central Plains
Southeastern Plains
Southea ste rn W is co ns in Til IP la in:
Southern Basin and Range
Southern Coastal Plain
| Southern Deserts
| Southern Florida Coastal Plain
1 Southern Rockies
I Southern Texas Plains
I Southern and Central California PI
1 Southwestern Appalachians
rn Allegheny Plateau
rn Com Belt Plains
rn Gulf Coastal Plain
Figure 6. Predominant ecoregions in cataloging units.
Export coefficients for BOD were assigned based on three general land-use categories:
agricultural, forest, and urban areas (Thomann and Mueller, 1987; Novotny and Olem, 1994).
Typical values for TSS export coefficients were also obtained for the three land-use categories
mentioned above (Thomann and Mueller, 1987; Novotny and Olem, 1994). TSS export
coefficients on agricultural cells were amended using a generalized revised universal soil loss
equation (RUSLE; USD A, 1997). This method allowed spatial variability to be introduced into
the TSS NPS loadings. The RUSLE equation estimates average soil loss in the following
manner:
2-18
-------
Section 2.0
NWPCAM System
TSS =R x K x L x S x C x p x 2241.7 (9)
where
TSS =
average soil (TSS) loss (kg/ha/yr)
R
rainfall-runoff erosivity factor
K
soil erodibility factor
L
slope length factor
S
slope steepness factor
c
cover management factor
p
support practice factor (assumed to be 1)
The factors used in the RUSLE vary by climate, soil type, and other physiographic factors.
Fifteen representative cities were selected to represent various climates across the United States.
The major land resource areas (MLRAs) were then overlaid with RF3 and related to a
representative city using latitude/longitude coordinates. The values for R, K, and C were
obtained from literature and vary by representative city (U.S. ACE, 1998). L and S are
calculated values that vary with each stream reach, but vary in NWPCAM 1.6 by cataloging unit
(USD A, 1997):
L = (kH2.6)m and S = 10.8 sin0 + 0.03 (10)
where
X = 1,640 ft (the distance from centroid to edge of a land-use cell)
m = p/(l + P)
P = (sin0/O.O896) / (3sin008 + 0.56)
0 = arctan (slope).
All of the NPS loadings that discharge to intermittent streams in the western
hydroregions were multiplied by the percentage of flow included into the discharge estimates
during the validation exercise.
NPS data for fecal coliform and fecal streptococci were not readily available at the
national scale. Table 11 shows the total loadings from all NPSs. Data from the STOrage and
RETrieval (STORET) database were used to develop default NPS fecal coliform loads by
hydroregion. The median fecal coliform concentration was derived from each hydroregion. In
areas of high AFO activity, one-half of the median fecal concentration was used to avoid double
counting those loads.
Table 11. Total Loadings on Land-Use/Land-Cover Cells for NPSs
li()l)5 1 oil cl
I N l()illl
I P 1 Oil cl
TSS liisul
1X'II load
Source Type
(••/s)
(»ls)
(»/s)
(»/s)
(Ml>\/s)
NPS
226,526
165,572
12,382
15,033,394
2-19
-------
Section 2.0
NWPCAM System
In preparation for the in-stream modeling process, the NPS loads are first routed from
land-cover cells to the RF3 network by accounting for overland transport, decay, and
transformation. The NPS loads are then routed to the RF3Lite network using in-stream transport,
decay, and transformation processes. Table 12 lists the total NPS loads at the RF3 and RF3Lite
scales, along with delivery ratios for each component.
Table 12. Total Loadings for NPSs at the RF3 and RF3Lite Scale
UO 1)5
IN
11>
TSS
i c i; loiui
I'simmeler
(li/s)
(»ls)
(••/s)
(li/s)
(MI'N/s)
RF3 Load
170,227
134,388
9,555
11,076,412
Delivery Ratio
0.75
0.81
0.77
0.74
RF3Lite Load
163,237
131,775
9,366
10,105,756
5.343 x 1010
Delivery Ratio
0.72
0.80
0.76
0.67
2.2 Kinetics
Overland transport modeling occurs for every land-cover cell to its associated RF3 reach.
In-stream modeling is conducted on a reach-by-reach basis. The pollutant concentration at the
upmost point of a reach or open channel is a function of the mixing and dilution of the mass load
from an upstream reach plus any pollution sources. The upstream boundary concentration (C0) is
obtained from a steady-state mass balance dilution calculation in Equation 11:
Cn
[(Qu Cu) + (Qe Ce) + (Qt Ct)] / [(Qu + Qe + Qt)]
(11)
where
C0
Qu
cu
Qe
ce
Qt
c,
Concentration at beginning of the reach
upstream stream flow entering reach
concentration from upstream reach
effluent flow of PS
effluent concentration of PS constituent
tributary flow of PS
tributary concentration of constituent.
The fate of nutrients and pollutants during overland transport and in-stream transport is
driven by first-order decay kinetics based on the following equation:
dc
dt
= -K*c
(12)
where
dc/dt = the instantaneous change in pollutant concentration
2-20
-------
Section 2.0
NWPCAM System
K = decay rate (1/d)
c = pollutant concentration (mg/L).
The closed-form solution of this simple differential equation is
Cf = Q x e("Kt) (13)
where
Q = concentration (mg/L) at time 0
Cf = concentration (mg/L) at time t (d).
Extensive experience from a large number of studies has shown that the first-order decay
process can be adequate for modeling many of the complex physical and biological processes
that occur in water. The decay rate is generally based on field measurements, other modeling
studies, or calibration of the model. Decay rates for biological processes tend to be temperature
dependent. For NWPCAM 1.6, temperature adjustments to decay rates were adopted from U.S.
EPA (1985; see Equation 13).
K(T) = K(20) x 0 (T"2°) (14)
where
K(T) = the decay coefficient at the average water temperature
K(20) = the decay coefficient at 20° Celsius
0 = temperature correction factor = 1.07 for FCB
T = average water temperature.
2.2.1 Carbonaceous Biochemical Oxygen Demand
Although BOD5 is typically reported as the sole indicator of oxygen-demanding material
in wastewater, streams, and rivers, the biochemical reactions that determine the actual
consumption of dissolved oxygen in these waters are not completed within a 5-day incubation
period. The true measure of the long-term oxygen demand of the carbonaceous and nitrogenous
materials in wastewater and natural waters can be determined only by continuing the incubation
period for longer than 5 days.
Ultimate carbonaceous biochemical oxygen demand (CBODU) is defined as the oxygen
equivalent needed for complete stabilization of organic carbon in water and wastewater.
Depending on the type of PS or NPS load, ratios of CBODU to BOD5 are used to convert
effluent loading data compiled as BOD5 to CBODU required for the model. NWPCAM 1.6
assumes a ratio of 3 to convert NPSs of BOD5 into CBODU. Ratios for PSs vary by treatment
level (see Table 13).
The decay rate for CBODU was set to 0.075 day"1.
2-21
-------
Section 2.0
NWPCAM System
Table 13. CBODU:BOD5 Ratios for PSs
Treatment l.e\el
C IJOIM :li()l)5 R 1,000 ft3/s and
< 10,000 ft3/s
0.1227
> 1,000 ft3/s
0.0956
> 10,000 ft3/s
0.0408
Lake
0.3586
The following transformations between nitrogen species are considered: hydrolysis of
total organic nitrogen to ammonia, and oxidation of ammonia to nitrate. Total fluxes of nitrogen
species at a given time (^) therefore become
N03 (/,) = N03 (/„) + N03_NH3 (15)
NH3 (/,) = NH3 (/„) + NH3_TON - NO3 NH3
TON ft) = TON(/0) - NH3 TON
2-22
-------
Section 2.0
NWPCAM System
where
N03 (t0) = nitrate concentration at time t0
NH3 (70) = ammonia concentration at time t0
TON(/0) = total organic nitrogen concentration at time t0
NO3 NH3 = nitrate formed from ammonia during t0 to /,
NH3 TON = ammonia formed from total organic nitrogen during t0 to
The kinetic coefficients used to transform the nitrogen species are shown in Table 15 (U.S. EPA,
1985).
Table 15. Kinetic Parameters for Nitrogen Species
Kinetic Coefficient
Temperature
Species
(day1: 20°C)
Correction I'actor
Nitrate
0.100
1.045
Ammonia
0.120
1.080
Total Organic Nitrogen
0.075
1.080
2.2.3 Phosphorus Species
TP decay coefficients are also based on the SPARROW model. The model was
calibrated using ambient phosphorus flux data. The SPARROW study suggests three values for
decay coefficients that vary for different flow ranges (see Table 14).
The NWPCAM 1.6 model accounts for the transformation of total organic phosphorus to
phosphate. The fluxes of phosphorus species at a given time (^) therefore become:
P04(0 = P04 (t0) + P04_T0P (16)
TOP (tx) = TOP(f„) - P04 TOP
where
P04 (t0) = phosphate concentration at time t0
TOP(/0) = total organic phosphorus concentration at time t0
P04_T0P = phosphate formed from total organic phosphorus during t0 to /,
The kinetic coefficient used to transform total organic phosphorus into phosphate was set at 0.03
day1 (U.S. EPA, 1985).
2.2.4 Dissolved Oxygen
Dissolved oxygen (DO) is included in the model framework as a key indicator of water
quality for the protection of aquatic biota. DO levels are also directly related to policy scenarios
that drive municipal and industrial effluent loading rates of carbonaceous (CBODU) and
nitrogenous (TKN) oxygen-demanding materials. Sources of DO that add oxygen to surface
2-23
-------
Section 2.0
NWPCAM System
waters include atmospheric reaeration and photosynthetic oxygen production from algae,
macrophytes, and periphyton. DO is lost from surface waters by respiration of algae,
macrophytes, and periphyton; biochemical decomposition of organic carbon (i.e., CBODU);
nitrification of ammonia; and consumption of oxygen in the sediment bed. The photosynthetic
gains (P) and respiratory losses (R) from aquatic plants are assumed to balance (i.e., P - R = 0 or
P = R).
DO concentrations are dependent on CBODU and ammonia concentrations, because the
latter account for the carbonaceous and nitrogenous oxygen demands. The solution for DO may
also be given in terms of the DO deficit, or departure from the oxygen saturation concentration.
The solution for the spatial distribution of oxygen deficit, D(x), is taken from Thomann
and Mueller (1987) and given in Equation 17 for oxygen balance:
The components of the oxygen balance equation (17) are as follows:
(a) the initial value of the oxygen deficit
(b) PS of CBODU
(c) PS of oxidizable nitrogen
(d) distributed source of ammonia load with no significant addition to river flow
(e) input due to distributed source from algal gross photosynthesis
(f) deficit due to distributed sink from algal respiration
(g) deficit due to distributed sink from sediment oxygen demand
D(x) = Do exp
, ( K.
1 x
- K —¦
aU
K - K
v a r
exp
K
x
ruJ
exp
K
\Ti
au
(17a)
(17b)
( K_
VKa-Kn
exp
( x
-*s,-
K
vKaKn
x
1 " 6XP " Ka^
/ J
exp
(
( x\Ti
- K„—
'U
/ J
Na
o n
(Ka - Kn)Kn)
exp
1 x
nU
exp
' x ^
"Ka-
a u
/ J
(17c)
Saman (17d)
1 - exp
(
- K —
a U
/ J
'P.
lKay
(17e)
1 - exp - K„
1 - exp
x
U
(
K
x
^auJ
'K
w
( s "i
B
K H,
V a y
(17 f)
(17g)
2-24
-------
Section 2.0
NWPCAM System
where
D(x) =
oxygen deficit along longitudinal distance of river
Do =
initial oxygen deficit at upstream end of a segment
Ka =
atmospheric reaeration coefficient
x =
longitudinal distance in direction of flow
U
freshwater stream velocity
Kd =
CBOD decomposition rate
Kr =
CBOD removal rate
L0 =
initial CBODU concentration at upstream end of segment
N0 =
initial TKN concentration at upstream end of segment
Kn =
nitrification rate
c =
^am
distributed source of ammonia from sediments
Pa =
daily average gross photosynthetic oxygen production (Pa = Ra)
Ra =
algal respiration rate (Ra = Pa)
SB =
sediment oxygen demand
H
depth of river segment
an
oxygen to nitrogen stoichiometric ratio (4.57 g 02/g N).
All reaction rates are computed for the ambient water temperature (T, °C).
After computation of the oxygen deficit, D(x), the DO concentration is computed using
Equation 18:
DO(x) = [Cs - D(x)] (18)
where
Cs = dissolved oxygen saturation concentration
D(x) = oxygen deficit along longitudinal distance of rivers.
The DO saturation concentration (Cs [S,T, EmsJ) depends on water temperature, salt
concentration, and elevation above mean sea level and is computed from relationships given by
Thomann and Mueller (1987) and Chapra (1997).
The effect of water temperature on oxygen saturation (Oj is computed with Equation 19:
InO ~ 13.3/I/111 1-5757-01 x 10s 6.642308 x 1Q7 (19)
Sf rp rp2
a a
1.243800 x 1010 86.621949 x 1011
+ 3 " 3
where
Ta = absolute temperature (degrees K)
T = temperature (°C)
2-25
-------
Section 2.0
NWPCAM System
where Ta is computed from Equation 20:
Ta = T +273.15. (20)
The effect of salt on oxygen saturation (OJ is computed using Equation 21:
In 0sf = In 0sf-S (1.7674 x 10"2 - 1-0754 x 101 + 2.1407 x 103 j ^
Ta Ta
where
S = salinity (g L"1).
Using data extracted from STORET, the spatial distribution of chlorides is represented as
a mean summer forcing function with summary statistics of chlorides assigned to RF3 reaches as
catalog unit mean values. Chloride levels (as mg/L) are converted to salinity (S, as g/L) to
estimate oxygen saturation using Equation 22:
S = 0.03 + 1.80655 x 10"3 [CI"] (22)
The effect of elevation on the temperature (T) and salt-dependent DO saturation (Osp) is
computed from a formulation given by Chapra (1997) using Equation 23:
Osp = (Osf+OJ [1-114.8 Emsl)] (23)
where
Osf = temperature-dependent oxygen saturation (mg/L)
Oss = salt-dependent oxygen saturation (mg/L)
EMsl = mean elevation above sea level (m).
For headwater start reaches, 100 percent oxygen saturation is assumed so that the initial
deficit is zero. For inflows across the upstream boundary and tributary inflows, the oxygen
deficit is computed, stored, and assigned from upstream solutions of the model. For PSs,
characteristic oxygen concentrations are assigned.
Loadings for groundwater pollution are not considered in NWPCAM because of a lack of
nationally available data. Groundwater flows are implicitly taken into account in the HCDN-
derived streamflow estimates.
Oxygen transfer from the air to the surface layer of a waterbody depends on water
temperature and turbulence due to velocity in the river, turbulence due to wind mixing, and any
turbulence contributed by water falling over waterfalls and dams. For this simplified model, the
atmospheric contributions from wind mixing, waterfalls, and dams are not considered. The
atmospheric reaeration coefficient (KJ is determined using the method of Covar (1976)
presented in Bowie et al. (1985) and adopted for the Wasp5-Eutro5 model (Ambrose et al.,
2-26
-------
Section 2.0
NWPCAM System
1993). The method computes reaeration as a function of velocity and depth using formulations
developed by Owens et al. (1964), Churchill et al. (1962), and O'Connor and Dobbins (1958) for
different categories of streams and rivers. The selection of the specific formulation is governed
by depth and velocity assigned to the RF3 reach. The computation of Ka is given in Equation 24:
Ka = a Ub Hc (24)
where
a, b, c = coefficients for depth and velocity
U = velocity (m/s or ft/s)
H = depth (m or ft).
The lower and upper ranges for depth (H) and velocity (U) and the numerical values of
the coefficients {a, b, and c) for the three formulations are given for both metric and English
units in Table 16.
Table 16. Depth (H) and Velocities (U) Ranges, Reaeration Formulations, and Coefficients
for Owens et al., Churchill, and O'Connor-Dobbins (Chapra, 1997; Ambrose et
al., 1993)
Nlelric I nils
l-ln^lish I nils
(I ;is in s II ;is in)
(I
-------
Section 2.0
NWPCAM System
The atmospheric reaeration rate is adjusted for ambient water temperature according to
the following relationship using a temperature correction factor of 1.024.
The importance of the decomposition of organic matter deposited in the sediment bed has
been understood since oxygen balance models were developed during the 1960s. Water quality
models typically define spatially dependent rates of sediment oxygen demand (SOD) as a zero-
order, external forcing function specified as input data to a model (e.g., Qual2E, Brown and
Barnwell, 1987; Wasp5-Eutro5, Ambrose et al., 1993). In using NWPCAM to simulate water
quality under reduced loading conditions as a result of control alternatives, the effects of
loadings reductions on SOD conditions was problematic since no reliable methods were
available to link changes in organic matter deposition to changes in SOD. Where the control
alternatives were not expected to greatly alter the loading of particulate organic matter to the
sediments, an assumption of no change in the SOD was reasonable.
Incorporating a full SOD model is far beyond the scope of NWPCAM. Rather, a
simplified SOD model was adopted based on the analysis of Di Toro et al. (1990). This work
showed that the SOD exerted by decomposition of particulate organic carbon in the sediments is
dependent on the square root of the loading of particulate organic carbon to the sediments. This
theoretical result of the SOD model has been confirmed in analyses of published data sets and
contemporary field measurements (Di Toro et al., 1990). SOD rates at 20 °C are assigned to
RF3 reaches as follows:
¦ RF3 Reaches Not Affected by PS: 0.5 g 02 m'May"1
¦ RF3 Reaches Assigned PS Loads: 1.5 g 02 m"2day"\
2.2.5 Total Suspended Solids (TSS)
Suspended solids are included in the model framework as an indicator of water clarity.
Solids are introduced into surface waters by naturally occurring geomorphological processes and
anthropogenic loading from PSs and land-use-influenced NPSs. In streams and rivers, the
distribution of solids suspended in the water column is determined by the particle size
characteristics of cohesive and noncohesive solids, hydrodynamics, and the particle size-
dependent balance between deposition and bottom shear-induced resuspension.
The representation of suspended solids in NWPCAM is highly simplified. A single size
class of solids is used to define both the inorganic and organic components of TSS with no
distinction made between cohesive and noncohesive solids. No attempt was made to account for
the solids content of a sediment bed that can be resuspended back into the water column under
high-flow conditions of erosion because (1) national-scale data are not available to characterize
the spatial distribution of solids in the sediment bed or to distinguish between cohesive and
noncohesive size classes either in the water column or the bed; and (2) any representation of
resuspension based on bottom shear stresses and velocities computed from the simplified flow
balance would introduce an enormous amount of uncertainty into the model framework. The
simplified model for TSS, based on no interaction of solids between the water column and the
sediment bed, assumes a "one-way loss of solids to the bed" (Chapra, 1997).
2-28
-------
Section 2.0
NWPCAM System
Sources of suspended solids in the model are derived from external inputs from PSs and
NPSs. The balance between deposition and resuspension is represented in the model as a simple,
first-order loss term governed by the settling velocity assigned to the single size class of solids
and the depth of the water column. As phosphorus is generally bound to sediments, the decay
rate for phosphorus is related to the deposition rate of sediments. Settling velocities for TSS were
based on the national SPARROW model (Smith et al., 1997).
2.2.6 Pathogens
Fecal coliform bacteria (FCB), used as an indicator for the public health risk of exposure
to waterborne pathogens, are present in surface waters primarily from sources accounted for by
direct discharges from municipal and industrial wastewater facilities, CSOs, and watershed
runoff from urban and rural land uses. Bacteria are lost from the water column primarily by
mortality. Settling and/or resuspension of bacteria sorbed onto particles are also processes that
can influence the density of bacteria. The loss of FCB is represented in the model as a simple,
first-order lumped mortality term. The FCB decay coefficient at 20° Celsius was set to 0.8 day"1.
A temperature correction factor of 1.07 is employed by NWPCAM 1.6.
Fecal streptococci (FS) are also modeled by NWPCAM 1.6 for the AFO/CAFO modeling
process. The FS decay coefficient was set at 0.168 day"1.
2.2.7 Other Processes
NWPCAM does not simulate the effects of nutrients on primary production, and the
subsequent effects of primary production on turbidity, dissolved oxygen, and BOD. Thus, it is
fair to say that the water quality benefits of reductions in nutrient loadings may be
underestimated. EPA seeks to improve representation of these processes in future development
of the NWPCAM model.
2-29
-------
Section 3.0
AFO/CAFO Modeling Process
3.0 AFO/CAFO Modeling Process
3.1 Changes to NWPCAM Since the Proposed Rulemaking
Table 17 contains a summary of major differences in the versions of the NWPCAM
system used for the AFO/CAFO proposed rulemaking process and the final rulemaking process.
These changes are described in detail elsewhere (RTI, 2002).
Table 17. Summary of Differences in NWPCAM Versions Used for the Proposed
Rulemaking and Final Rulemaking
Component
Proposed Ruleiiiitkiii"
(NWPCAM 1.5)
l-iiiiil Ruleiiiitkiii"
(N\VI»CAM 1.6)
K fleet
Database Platform
Microsoft Access
Oracle
Automated model
runs
Streamlined quality
control process
Simplified analysis
of inputs or delivery
ratios
Reach Network
RF3 used to route
loadings
RF3Lite used for
chlorophyll-a
modeling in lakes
RF1 used for in-
stream modeling
RF3 used to route
loadings
RF3Lite used for
in-stream modeling
Improved network
connectivity
Better coverage of
open waters
Stream Flow
RF3 stream flows
based on average
annual runoff by
cataloging unit
RF1 stream flows
based on RF1
characteristics data
set
RF3 stream flows
calibrated using
USGS gaging
station data
Stream flows in
western
hydroregions
adjusted for
intermittent stream
contribution
• Improved RF3
stream-flow
estimates
Improved modeling
accuracy
(continued)
3-1
-------
Section 3.0
AFO/CAFO Modeling Process
Table 17. (continued)
Proposed Ruleiiiitkiii" l iiiitl Riilem:tkiii»
Component (WYI'C AM 1.5) (WYIH AM 1.6) EITecl
Slope by Cataloging
Unit
Used one-half of average
slope of first-order
streams in the cataloging
unit
Slope estimates based
on Digital Elevation
Model (DEM)
More accurate slope
estimates
• Higher channel
velocities and
delivery ratios from
land cells to RF3
Stream Velocity
Velocity estimates
based on Keup
(1985)
• Used RF1
characteristics
database for in-
stream modeling
All velocity estimates
based on Jobson (1996)
Improved velocity
estimates
PS Inventory
Used PS inventory from
NWPCAM 1.1
Used PS inventory
from NWPCAM 2.1
More comprehensive
account of PS loadings
PS Delivery
PS loads routed directly
toRFl
PS loads routed to
RF3Lite with decay
and transformation
Capitalizes on PS
location information
Conventional NPS
Loads
Based on county-level
loadings apportioned to
reaches
Based on land-
cover export
coefficients
Incorporated the
RUSLE for TSS
loads on
agricultural cells
Improved spatial
resolution
Improved
consistency with
nutrient approach
More accurate DO
modeling
Nutrient NPS Loads
Export coefficients by
hydroregion/ecoregion/
land-cover type
calibrated to SPARROW
nutrient fluxes
Same loadings, but
speciated by land-cover
type
Allows use of a water
quality index that
incorporates nutrient
measures
Non-AFO/CAFO
NPS Delivery
NPS loads routed to
RF3Lite and RF1 without
decay and transformation
NPS loads routed to
RF3 and RF3Lite with
decay and
transformation
Improved consistency
with PS load approach
AFO/CAFO Load
Distribution
Loads assigned
randomly to
agricultural cells
within a county
Limited loads on
land-use cells based
on Beaulac and
Reckhow(1982)
Loads assigned
randomly to
agricultural cells using
model farms areas and
without considering
nutrient load caps
See Section 5.1
(continued)
3-2
-------
Section 3.0
AFO/CAFO Modeling Process
Table 17. (continued)
Proposed Ruleiiiitkiii" l iiiitl Riilem:tkiii»
Component (WYI'C AM 1.5) (WYIH AM 1.6) EITecl
In-Stream Modeling
of Nutrients
BATHTUB used to
model chlorophyll-a at
RF3Lite scale
NWPCAM 1.6
includes nutrient
modeling at RF3Lite
scale
Permits use of a
water quality index
that includes nitrates
and phosphates
Oxygen demand
from non-
AFO/CAFO
nutrients will be
modeled
In-Stream Modeling
of Conventional
Pollutants
Based onNWPCAM 1.1
kinetics with some peer-
review comments
incorporated
Based onNWPCAM
1.1 with all peer-
review comments
incorporated
Changes in kinetics are
expected to have
minimal impact on
results
In-Stream Modeling
of Bacteria
Included in-stream
kinetics for FCB, but not
FS
Includes in-stream
kinetics for FCB and
FS
Although not used in
use support
determination or
benefits analysis,
provides in-stream
concentrations of FS
Benefits Metric
• WQL
Regionalized
chlorophyll-a water
quality ladder used to
estimate economic
benefits
Adds capability to
calculate six-parameter
WQI that includes
nitrates and phosphates
Integrated economics
benefits approach
3.2 AFO/CAFO Input Files
Four Microsoft Excel workbooks were supplied by EPA to characterize AFO/CAFO
loadings to surface water:
1. AFO facility counts by county, animal type, and operation size1
2. Percentage of AFOs considered CAFOs in each state under various regulatory
scenarios2
'File name = Distributionll.xls; Date received = 8/12/02
2Filename = StatePct(102502)corr.xls; Date Received = 11/15/02
3-3
-------
Section 3.0
AFO/CAFO Modeling Process
3. Nutrient, sediment, BOD, and pathogen loadings for unregulated AFOs and
regulated CAFOs under various technology options by U.S. region, animal type,
and operation size. Model farm areas were also included in this data set.3
4. Speciation factors to break TN and TP loadings into nutrient species under
various technology options by U.S. region, animal type, and operation size.4
Loadings were supplied for TN, TP, TSS, BOD, FCB, and FS. Loadings for each parameter
were separated into manure, commercial fertilizer, and feedlot load categories.
3.3 Methodology
All preprocessing steps (i.e., the parts of NWPCAM 1.6 unaffected by the AFO modeling
process) were outlined in Section 2.0 of this report. For each AFO/CAFO regulatory option
modeled by NWPCAM 1.6, several analytical and data management processes were conducted:
¦ Distribute AFO/CAFOs and associated edge-of-field loadings to agricultural land-
cover cells within each county.
¦ Route AFO/CAFO loadings from the land-cover cells to the nearest RF3 reach
using an overland transport, loss, and transformation routine.
¦ Route AFO/CAFO loadings from the RF3 network to the RF3Lite subset using an
in-stream transport, loss, and transformation routine.
¦ Simulate dilution, transport, and kinetics of the nutrients/pollutants in the RF3Lite
network.
¦ Relate the nutrient and pollutant concentrations in the RF3Lite reach to beneficial
use attainment criteria and goals.
¦ Calculate the overall WQI6.
¦ Compute economic benefits based on changes in water quality use-support.
¦ Compute economic benefits based on changes in WQI6.
See Section 2.1 for a flowchart of the NWPCAM 1.6 process employed for estimating the
benefits of AFO/CAFO regulations using the Vaughn WQL approach.
3Nutrients: File name = Output (11132002) adjusted Opt5.xls; Date received = 11/13/02
4File name = Speciation(1108).xls; Date received = 11/8/02
3-4
-------
Section 3.0
AFO/CAFO Modeling Process
3.3.1 Method for Distributing AFO/CAFO Loadings
AFO/CAFO farm locations were supplied at the county level via county FIPS codes. In
order to associate farms with RF3 reaches, the loadings were randomly distributed onto
agricultural land-cover cells within each county. This process involved (1) identifying
agriculture land-cover cells within each county using the Anderson Land-Cover Class Code;
(2) assigning random identification numbers to each agricultural cell; and (3) writing a module to
integrate the AFO/CAFO data sets and distribute loadings to the agricultural land-cover cells.
Figure 7 shows the RF3 network, land-use/land-cover data, and county overlay for one 8-
digit HUC. It is onto this mosaic that AFO/CAFO counts by county and associated edge-of-field
loadings are distributed. The load distribution module operated according to the following steps:
1. Selected a county using the county FIPS code
2. Generated a list of agricultural cells in that county, ordered by the random
identification number.
3. Selected the facility counts for the county. Each county was associated with
facility counts for 42 animal type and farm size combinations.
Land Use Cells and RF3 Hydrography in CU 07010201
Wate rsh e d (H U C) b o un d a ry
County boundary
Y\/ HUC 07010201 hydrography
Land Use Cell
^2 Dryland Crop/Pasture
~ Mixed Dry1and/lrr. Cropland
"2 Grassland/Crop Mosaic
g] Woodland/Crop Mosaic
~ Grassland
~ Northern Deciduous Forest
~ Southeastern Deciduous Forest
~ Northern Mixed Forest
~ Water Bodies
~ Urb a n
Figure 7. Mosaic Composite of Spatial Data at the Watershed (HUC) Level.
3-5
-------
Section 3.0
AFO/CAFO Modeling Process
4. Obtain the percentage of AFOs considered CAFOs for each animal type and farm
size combination under the regulatory scenario being simulated.
5. Obtained loadings data for regulated and unregulated facilities for each animal
type and farm size combination under the regulatory scenario being simulated.
6. Calculated total loadings according to the following equation:
Species Load = (%reg - reg_load x reg_species_fraction}\
(%unregx unreg_loadx unreg_species_fraction)
(25)
7. Calculated nutrient species loads by the following equation:
Total Load = (% regulated x regulated _load) +
(% unregulated x unregulated_load}
(26)
Calculated the size of the farm in square kilometers using the model farm area.
This is equivalent to the number of agricultural land-cover cells onto which the
loads will be distributed.
9. Assigned loads to a land-cover cell with sufficient area available. If necessary,
distributed loads to additional land-cover cells until the farm area was depleted.
10. Looped through the 42 animal type and operation size combinations until all
AFO/CAFO loads in the county were distributed to agricultural land-cover cells.
11. Looped through the 3,045 counties in the conterminous 48 states.
Several rules were applied during the load distribution process to reflect regulatory
options or assumptions made during the load development process. These include the following:
¦ Feedlot loads were not distributed using model farm areas, because of the
modeling assumption that these loads have a direct hydrological connection to
surface water. Feedlot loads were uniformly distributed to land-cover cells that
were less than 100 ft from the nearest RF3 reach, and they did not undergo
overland loss or transformation.
Under RTI Scenarios 1 and 2, a 100 ft. setback was used for land application of
manure. Under the setback, regulated manure loadings were not distributed to
land-cover cells that were less than 100 ft from the nearest RF3 reach.
Unregulated manure loadings were not affected by the 100 ft. setback.
Commercial fertilizer load distribution was not affected by the 100 ft. setback.
3-6
-------
Section 3.0
AFO/CAFO Modeling Process
3.3.2 Routing AFO/CAFO Loads to RF3 Reaches
In order to be hydrologically routed through the stream network, the manure and
commercial fertilizer loads are routed from the agriculture land-cover cells to the nearest RF3
reach using a routine that simulates overland transport, loss, and transformation. The feedlot
loads are routed to the nearest RF3 reach without any loss or transformation of loads. Overland
travel times are based on flow in a natural channel such as may be found on agricultural lands
(see Section 2.2.4). A unit runoff (ft3/sec/km2) was derived for each HUC based on data from
USGS stream gages in the HCDN network. Travel distances were calculated from the center of
the agricultural cell to the nearest RF3 reach.
3.3.3 Routing AFO/CAFO Loads to RF3Lite Reaches
AFO/CAFO loads were routed to the RF3Lite network using an in-stream transport,
decay, and transformation module.
3.3.4 In-Stream Modeling in the RF3Lite network
The final stage of in-stream modeling is conducted in the RF3Lite network. At this stage,
a module routes through the RF3Lite network according to the hydrologic sequence order. For
each reach, NPS, PS, and AFO/CAFO loads at the RF3Lite scale are combined. The loads are
decayed and transformed to the middle of the reach to produce an estimate of average in-stream
concentration that is inserted into the results table. The loads are then decayed and transformed
to the end of the reach to continue routing down the RF3Lite network.
3.3.5 Water Quality Assessment Ladder
NWPCAM 1.6 uses the water quality ladder described in Table 18 to translate in-stream
concentration estimates for BOD5, TSS, DO, and FCB into corresponding use-support categories
using an approach developed by Vaughn for Resources for the Future (Mitchell and Carson,
1986). This approach assigns maximum pollutant levels for BOD, TSS, and FCB that
correspond to boatable, fishable, and swimmable waters. Minimum threshold values are also
established for DO. A water resource that fails to meet the boating criteria is classified as a
"nonsupport" resource. Vaughn's original water quality ladder included BOD5, turbidity, DO,
pH, and FC. In NWPCAM, TSS is used as a surrogate for turbidity.
Table 18. Water Quality Ladder Threshold Concentrations
lionoliciiil I so
liiolo^iciil
()\v«en Dcniiincl
(niii/l.)
l oliil Suspended
Solids (m»/l.)
Dissohcd
()\v«en
(% s;iUir:iled)
I Vcnl (olilorms
ill.)
Swimmable
1.5
10
0.83
200
Fishable
3
50
0.64
1000
Boatable
4
100
0.45
2000
3-7
-------
Section 3.0
AFO/CAFO Modeling Process
The Vaughn WQL model categorizes reaches as boatable, fishable, swimmable according
to the worst pollutant. Benefits are assigned only for reaches that are move across the designated
use categories. Thus, for a reach that is classified as "boatable" prior to regulation, even if there
is significant improvement in one or more water quality parameters (e.g., TSS and DO)
following regulation, no monetized benefit is assigned to that reach if any other parameter
remains in the boatable category.
3.3.6 Economic Benefits Calculations Using the WQL
Each RF3Lite reach is categorized using the WQL for each AFO/CAFO regulatory
scenario. The difference in the miles for each use category between baseline conditions and a
given rulemaking scenario is a measure of the improvement in water quality attributable to the
scenario. These differences in miles are converted into economic benefits (dollars) based on the
population and their willingness to pay (WTP) for improvement in water quality.
Benefits are calculated state-by-state and are broken down into local and nonlocal
benefits. Local benefits correspond to the amount a population is willing to pay for water quality
improvements within their own state. Nonlocal benefits correspond to the amount a population
is willing to pay for water quality improvements outside of their own state. Local benefits are
calculated as follows:
BOATWTP = (BOAT SCN - BOATBASE) / segjength x p0p / 2.62 x 206 x C
FISHWTP = (FISHSCN - FISHBASE) / segjength x pop / 2.62 x 155 x C (27)
SWIMWTP = (SWIMSCN - SWIM_BASE) / segjength x pop / 2.62 x 173 x C
where
BOAT SCN
BOATBASE
FISHSCN
FISH BASE :
SWIMSCN =
SWIM BASE ^
seg_length
pop
2.62
206
155
173
C
miles of the state's boatable waters for rulemaking scenario
miles of the state's boatable waters for baseline conditions
miles of the state's fishable waters for rulemaking scenario
miles of the state's fishable waters for baseline conditions
miles of the state's swimmable waters for rulemaking scenario
miles of the state's swimmable waters for baseline conditions
total miles of RF3Lite reaches in the state
population of the state (from 2000)
average number of people per household in the United States
average annual household WTP (in 2000 dollars) to increase U.S.
waters from nonboatable to boatable levels
average annual household WTP (in 2000 dollars) to increase U.S.
waters from boatable to fishable levels
average annual household WTP (in 2000 dollars) to increase U.S.
waters from fishable to swimmable levels
fraction of WTP applied to local benefits = 2/3.
Nonlocal benefits can also be calculated using Equation 27, with the following variable
definitions:
3-8
-------
Section 3.0
AFO/CAFO Modeling Process
BOAT SCN
BOATBASE
FISHSCN
FISHBASE :
SWIMSCN =
SWIM BASE :
seg_length
pop
2.62
206
155
173
C
miles of the nation's boatable waters for rulemaking scenario
miles of the nation's boatable waters for baseline conditions
miles of the nation's fishable waters for rulemaking scenario
miles of the nation's fishable waters for baseline conditions
miles of the nation's swimmable waters for rulemaking scenario
miles of the nation's swimmable waters for baseline conditions
total miles of RF3Lite reaches outside of the state
population of the state (from 2000)
average number of people per household in the United States
average annual household WTP (in 2000 dollars) to increase U.S.
waters from nonboatable to boatable levels
average annual household WTP (in 2000 dollars) to increase U.S.
waters from boatable to fishable levels
average annual household WTP (in 2000 dollars) to increase U.S.
waters from fishable to swimmable levels
fraction of WTP applied to local benefits = 1/3.
3.3.7 Water Quality Index
McClelland (1974) suggested using a Water Quality Index (WQI) that falls on a
continuous scale from 0 to 100, rather than four discrete beneficial-use attainment indicators
used in the WQL. The original WQI included nine water quality characteristics: BOD5, DO,
FCB, TSS, N03, P04, temperature, turbidity, and pH. NWPCAM 1.6 uses a six-parameter Water
Quality Index (WQI6) that incorporates BOD5, DO, FCB, TSS, N03, and P04. The remaining
three parameters are not modeled in NWPCAM 1.6 and are factored out of the WQI.
The WQI is derived by converting concentrations of each water quality characteristic into
a corresponding score (g,) between 0 and 100. As documented in McClelland (1974), this was
accomplished by averaging the judgments from 142 water quality experts regarding the
appropriate functional relationship between conventional measures and a 0-100 scale.
Appendix B includes charts of the six functional relationships used by NWPCAM 1.6. Weights
for each of the scores (w,) were derived, again based on the summary judgments of the expert
panel. These weights were designed to sum to 1 for the nine water quality characteristics. The q,
and wt values were combined into a composite multiplicative index of the following form:
IIViq> (28)
The z subscript refers to the z'-th parameter, and zz is the number of parameters (in this
case, n = 9). By design, WQI varies between and is bounded by 0 and 100.
To apply McClelland's index to output from NWPCAM 1.6, it must be modified to
account for the three characteristics (i.e., temperature, turbidity, and pH) that are not modeled.
To accomplish this, new weights are calculated for the remaining six parameters so that the
ratios of the six weights are retained and the weights sum to 1. Table presents the original and
revised parameter weights for the nine pollutants. Under the revised index, n = 6, and the w- s are
specified using the revised weights in Table 19.
3-9
-------
Section 3.0
AFO/CAFO Modeling Process
Table 19. Original and Revised Weights for WQI Parameters
I'.ini meters
Origin;il Weights
Re\ iseil Weights
BOD5
0.11
0.15
DO
0.17
0.24
FCB
0.16
0.23
TSS
0.07
0.1
no3
0.1
0.14
P04
0.1
0.14
Temperature
0.1
--
Turbidity
0.08
--
pH
0.11
--
Total
1
1
3.3.8 Economic Benefits Analysis Using the WQI6
The following WTP function is used to derive economic benefits using the WQI6
approach. This equation was estimated and reported by Mitchell and Carson using WTP
responses from their survey sample.
TOTWTP = exp [0.413 + 0.819 x ln(WQI/10) + 0.959 x ln(Y/1000) + 0.207 x W + 0.46 x A] (29)
where
TOTWTP = each household's total annual WTP (in 1983 dollars) for increasing
water quality up to each of the three WQI values
Y
W
A
= annual household income (sample average = $24,220 in 1983 dollars)
= dummy variable indicating whether the household engaged in water-
based recreation in the previous year (sample average = 0.59)
= dummy variable indicating whether the respondent regarded the nation
goal of protecting nature and controlling pollution as very important
(sample average = 0.65).
In solving this equation, Mitchell and Carson used Vaughn's WQL to map each
beneficial-use category to a corresponding WQI value (boatable = 25, fishable = 50, and
swimmable = 70). These values were applied in the right-hand side of Equation 28.
Equation 28 can also be used as a benefit-transfer function, to assess the value of
increasing water quality along the continuous WQI scale. In other words, assuming that W and A
3-10
-------
Section 3.0
AFO/CAFO Modeling Process
are representative of the current population, the incremental value associated with increasing
WQI from WQI0 to WQI, can be calculated as
aTOTWTP = exp[0.8341 + 0.819 x logCWQIj/lO) + 0.959 x log(Y)]
- exp[0.8341 + 0.819 x log(WQI0/10) + 0.959 x log(Y)] (30)
7, in this case, would be selected to correspond to average (or median) household income
in the year of the water quality change (expressed in 1983 dollars). The resulting value estimates
can be inflated to 2000 dollars using the growth rate in the consumer price index (CPI) of 1.72
since 1983.
Benefits are calculated state-by-state and are broken down into local and nonlocal
benefits. Local benefits correspond to the amount a population is willing to pay for water quality
improvements within their own state. Nonlocal benefits correspond to the amount a population
is willing to pay for water quality improvements outside of their own state.
3-11
-------
Section 4.0
Results of AFO/CAFO Analyses
4.0 Results of AFO/CAFO Analyses
This section summarizes the results of the NWPCAM 1.6 analyses for the AFO/CAFO
rulemaking scenarios.
4.1 AFO/CAFO Loadings
Total national AFO/CAFO loadings are key inputs estimated by EPA and used to drive
the NWPCAM 1.6 model simulations. The AFO/CAFO nutrient and pollutant loadings to
agricultural cells and their production area loads input directly into RF3 reaches for baseline
conditions and rulemaking scenarios are summarized in Table 20. These represent the total
national AFO/CAFO loadings actually distributed to agricultural cells and production area loads
input directly into RF3 reaches. According to EPA estimates, nutrients and sediments decline
moderately under both scenarios. FCB loads decline slightly under all scenarios. FS loads
decline slightly under both scenarios. For further discussion of the AFO/CAFO loading
calculations, please see the document entitled "Pollutant Loading Reductions for the Revised
Effluent Limitation Guidelines for Concentrated Animal Feeding Operations" in the rulemaking
record.
Table 20. National AFO/CAFO Loadings on Agricultural Cells*
liiochemical
Rulemaking
Scenario
TN (»/s)
\'\y (»/s)
TSS
(••/s)
()xv«en
Demand (*»/s)
1 ( li (clu/s)
rs (ci'u/s)
Baseline
3316
5076
958,380
1012
3.29 x 1014
4.30 x 1015
RTI Scenario 1
2977
4332
931,008
779
2.91 x 1014
3.50 x 1015
RTI Scenario 2
3166
4682
936,444
927
3.22 x 1014
4.16 x 1015
* Note: To calculate a loading rate per unit area, the values in this table should be divided by the agricultural area in
the country.
The AFO/CAFO nutrient and pollutant loadings in the RF3 network for baseline
conditions and rulemaking scenarios are summarized in Table 21. These represent the total
national loadings delivered to the RF3 reaches after overland transport from the agricultural cells
to the nearest reach occurs (manure and commercial fertilizer loads only). Table 22 lists the
delivery ratios to RF3 for baseline and scenarios. Between 75 and 90 percent of the total
national loads are delivered to the RF3 reaches for TN, TP, TSS, BOD, FCB and FS.
4-1
-------
Section 4.0
Results of AFO/CAFO Analyses
Table 21. AFO/CAFO Nutrient/Pollutant Loadings to RF3 Rivers/Streams
Rulemaking
IN
11>
'I'SS
liOl)
1 ( li
IS
Scenario
(»/s)
(«i/s)
(••/s)
(«i/s)
(cfu/s)
(cfu/s)
Baseline
2674
3832
748,148
907
2.52 x 1014
3.75 x 1015
RTI Scenario 1
2414
3296
733,841
686
2.20 x 1014
3.02 x 1015
RTI Scenario 2
2571
3563
738,741
826
2.48 x 1014
3.62 x 1015
Table 22. AFO/CAFO Delivery Ratios to the RF3 Network
Rulemaking
IN
11>
TSS
liOl)
1 C li
IS
Scenario
(••/s)
(»/s)
(»/s)
(»/s)
(clu/s)
(cfu/s)
Baseline
0.81
0.75
0.78
0.90
0.77
0.87
RTI Scenario 1
0.81
0.76
0.79
0.88
0.76
0.86
RTI Scenario 2
0.81
0.76
0.79
0.89
0.77
0.87
AFO/CAFO loadings to the RF3Lite subset of RF3 reaches for baseline conditions and
rulemaking scenarios are summarized in Table 23. These represent the total national AFO/
CAFO loadings delivered to the RF3Lite subset of RF3 reaches after transport down the RF3
network to the first RF3Lite reach segment encountered. Table 24 lists the delivery ratios to
RF3Lite for baseline and the scenarios. Between 69 and 86 percent of the total national loading
are delivered to the RF3Lite reaches for TN, TP, TSS, BOD, and FS. FCB have a high die-off
rate, which translates into a smaller delivery ratio at around 62 percent.
Table 23. AFO/CAFO Nutrient/Pollutant Loadings to RF3Lite Network
Rulemaking
IN
1 l>
TSS
liOl)
1 C li
IS
Scenario
(ii/s)
(ii/s)
(»/s)
(••/s)
(cfu/s)
(cfu/s)
Baseline
2383
3502
683,817
875
2.05 x 1014
3.53 x 1015
RTI Scenario 1
2149
3007
670,391
663
1.80 x 1014
2.84 x 1015
RTI Scenario 2
2290
3252
674,921
798
2.02 x 1014
3.40 x 1015
Table 24. AFO/CAFO Delivery Ratios to the RF3Lite Network
Rulemaking
IN
11>
TSS
liOl)
1 C li
IS
Scenario
(»/s)
(Ii/s)
(ii/s)
(li/s)
(cfu/s)
(cfu/s)
Baseline
0.72
0.69
0.71
0.86
0.62
0.82
RTI Scenario 1
0.72
0.69
0.72
0.85
0.62
0.81
RTI Scenario 2
0.72
0.69
0.72
0.86
0.63
0.82
4-2
-------
Section 4.0
Results of AFO/CAFO Analyses
4.2 Economic Benefits
Table 25 provides a summary of the annual economic benefits for each scenario using the
WQL. This summary was computed by summing the local and nonlocal benefits for each state.
Scenario 1 exhibited a higher benefit because of its layer reduction of all constituents.
Table 25. Annual Economic Benefits Using the WQL (2001 dollars, thousands)
Rulemaking
lioalahle
l-'ishahle
Swim in a hie
Total
Scenario
Waters-
Waters-
Waters-
licncl'il
RTI Scenario 1
114,051
38,811
13,322
166,184
RTI Scenario 2
73,065
23,202
6,122
102,389
* Boatable benefits include only those benefits attributable to improvements from non-boatable to boatable.
Benefits from improvements to other beneficial use categories appear in the other columns. For a reach that
improved from nonbotable to fishable, for example, a portion of the benefits appear in the boatable column, and
the remainder appears in the fishable column. Similarly, fishable and swimmable benefits include only those
benefits attributable to improvements from boatable to fishable and from fishable to swimmable, respectively.
Benefits from improvements to other use categories appear in the other columns as described above.
Table 26 provides a summary of the annual economic benefits for each scenario using the
WQI. This summary was computed by summing the local and nonlocal benefits for each state.
Scenario 1 exhibited a higher benefit because of its layer reduction of all constituents.
Using the WQL, the bulk of monetary benefits occur in the boatable waters category.
Using the WQI, the majority of the benefits occur in the middle (i.e., 26 26 < WQI < 70 WQI > 70 - Total ISenelil
RTI Scenario 1
10,088
24,154
46,950
298,552
RTI Scenario 2
7,187
135,266
40,105
182,558
* This category includes only the benefits attributable to improvements between 26 and 70. For example, for a
reach that improved from 24 to 30, the portion of benefits attributable to the increase from 24 to 26 appears in
the WQI<26 category; the remainder appears in the 2670. For a reach that
improved from 24 to 80, for example, a portion of the benefits is allocated to each of the WQI<26, the
2670 categories.
4-3
-------
Section 4.0
Results of AFO/CAFO Analyses
4.3 Discussion of Benefit Results
Both estimation methods rely on WTP values derived by Carson and Mitchell (1993).
The WQL captures the benefits of discrete changes in the type of uses or amenities provided by
waterbodies and, in doing so, reflects the principles of water quality standards where
determinants of beneficial use attainment are based on water quality criteria. Carson and
Mitchell (1993) indicate that amenities such as boatable, fishable, and swimmable water quality
are "concepts that are widely understood."
However, the pollutant criteria for making use determinations in the discrete ladder
include criteria for which federal guidance has not been developed. Criteria for TSS and BOD
are not typically adopted for the boatable, fishable, and swimmable amenities, and inclusion of
criteria for these pollutants implies lower probability of beneficial use attainment under the
ladder than might be indicated by other methods for determining use attainment in the nation's
waters.
In contrast, the WQI approach adopted for this final rule characterizes changes in water
quality using an aggregate index derived from six individual pollutant concentrations. Carson
and Mitchell (1993) state that the use of this type of index greatly facilitates the task of
communicating the several quality levels (i.e., amenities) to the (survey) respondents. This
observation accentuates the fact that different respondents are likely to rely on different
measures of water quality to make value judgments. The minimum index values (25 for
boatable, 50 for fishable, and 70 for swimmable) adopted by Carson and Mitchell help explain
why the magnitude and distribution of benefits differ between the discrete ladder and the
continuous WQI approaches.
Differences in magnitude are due in part to the likelihood that the distribution of
predicted changes in some parameters is not sufficiently large to meet criteria necessary for an
amenity change, including the boatable category. As a consequence, changes in beneficial use
are unlikely to occur, and corresponding benefits are lower under the discrete ladder. Under the
continuous WQI, benefit estimates are not constrained by "limiting parameter" distributions, and
the benefits from all changes in water quality are captured, regardless of changes in amenity
support. The relative difference in magnitude of benefits is a function of the baseline
distribution of water quality parameters; in some special cases, the benefits under the ladder
could approximate or even exceed those under the continuous index (when baseline measures of
central tendency (e.g., median) are approximately equal to the threshold criteria for supporting
amenities).
Apparent inconsistencies in the distribution of benefits between the two methods arise
because many waterbodies fail to meet beneficial use criteria in the ladder, yet most of these
same waterbodies have WQI values that exceed the minimum index thresholds specified in the
ladder. For example, most of the benefits realized under the ladder occur when waters improve
from nonboatable to boatable because, as noted above, a majority of waterbodies are not capable
of meeting the criteria for higher uses. However, in the case of the continuous index, a majority
of benefits are due to changes in water quality within an index range of 26 to 70; this range
reflects a boatable and/or fishable attainment, based on index thresholds in the ladder (boatable =
25, fishable =50, swimmable = 70). The discrepancy occurs because many nonboatable reaches
4-4
-------
Section 4.0
Results of AFO/CAFO Analyses
under the ladder actually have index values that are far higher than the minimum threshold for
boating. Approximately 80 percent of reach segments designated as nonboatable under the
ladder under baseline conditions have WQI values that range from 29 to 79 based on NWPCAM
output for a four-parameter index, implying that many waterbodies deemed nonboatable under
the ladder would be considered boatable, fishable, or even swimmable under the continuous
index. It is felt that many people would be willing to boat or fish in waters that are deemed
unboatable under the ladder. As a final note regarding the distribution of benefits, it is also
possible that a particular regulation, such as the final CAFO rule, may affect specific geographic
areas where nonboatable waters predominate, thus implying that a majority of benefits are
attributable to improvements from nonboatable to boatable.
A comparison of the two valuation methods is most easily understood within the context
of the original Carson and Mitchell survey. Recall that the Carson and Mitchell survey presents
(1) explicit relationships between beneficial use categories and numeric values of the WQI, and
(2) baseline water quality conditions for the nation that are similar in some respect to the results
in the NWQI (2000). The results from the survey are used to estimate (1) mean WTP values for
water quality levels supporting different amenity categories, and (2) a valuation function that
predicts WTP as a function of water quality index values. The ladder approach to estimating
benefits maintains consistency with the explicit correlation between WTP and beneficial use
categories specified by Carson and Mitchell (e.g., a change in water quality and WTP can be
related to changes in amenities), but is not consistent with baseline water quality conditions. The
WQI approach maintains consistency with baseline water quality conditions but is less capable
of maintaining consistent relationships between WTP and changes in beneficial use categories.
Other advantages of the continuous index approach include (1) use of a decreasing marginal
benefits curve with respect to the WQI (consistent with economic theory), (2) the ability to
capture benefits of marginal changes in individual water quality parameters without triggering
changes in amenities, and (3) the ability to capture benefits associated with changes in other
parameters (i.e., nitrate and phosphate) that are not included in the ladder.
4-5
-------
Section 5.0
Quality Assurance
5.0 Quality Assurance
Potential sources of error and uncertainty in the analysis include model inputs (e.g.,
hydrologic inputs from RF3), data processing, model parameters (e.g., decay rates), and benefits
monetization methods. This section describes measures taken to reduce these errors and
uncertainties for the AFO/CAFO analysis, including (1) reviewing hydrologic inputs for
reasonableness, (2) evaluating the robustness of model predictions to changes in model
parameters, (3) performing quality assurance on all data processing steps, including the
computational modules, (4) evaluating modeling results for reasonableness, and (5) evaluating
the sensitivity of estimated benefits to the monetization method selected.
5.1 Reviewing Hydrologic Inputs
RTI has performed extensive quality assurance on the flow and velocity estimates
included in NWPCAM 1.6. Comparisons were made between NWPCAM 1.6 and observational
values of flow and velocity obtained from the USGS HCDN network. The results of this work,
including the methodology used to develop the NWPCAM 1.6 estimates, are contained
elsewhere (RTI, 2001).
5.2 Model Robustness
A full calibration exercise on NWPCAM 1.6 has not been conducted. However, a
sensitivity analysis was performed in Hydroregion 5 to evaluate changes in predicted water
quality due to changes in modeling inputs (RTI, 2002). Ten parameters (i.e., flow, velocity,
depth, PS loads, non-AFO NPS loads, BOD oxidation rate, TSS settling rate, FCB die-off rate,
sediment oxygen demand, and CBODU:BOD5 ratio) were varied by a factor of 1.5 to 2.
Because NWPCAM is a screening-level model, the sensitivity analysis was aimed at evaluating
whether changes in water quality from baseline to scenario were robust, as opposed to absolute
water quality. Four model runs were conducted for each of the 10 parameters: (1) low parameter
value, baseline AFO/CAFO loads; (2) low parameter value, scenario AFO/CAFO loads; (3) high
parameter value, baseline AFO/CAFO loads; and (4) high parameter value, scenario AFO/CAFO
loads. For these analyses, the baseline AFO/CAFO loadings were taken from the dummy
loadings files supplied by EPA on March 27, 2002. Scenario loadings were taken from Option 2
in the same dummy loadings file.
Flow, velocity, and non-AFO NPS loads had the greatest impact on absolute water
quality as assessed by the WQI6. However, changes in water quality were robust, with average
WQI improvements of approximately 1.3 for all runs. This indicates that uncertainty in model
coefficients and inputs may not have a significant impact on predicted water quality changes
under regulatory scenarios. However, absolute model results (e.g., DO concentrations by reach)
will be affected significantly by uncertainty in model coefficients and inputs.
5-1
-------
Section 5.0
Quality Assurance
5.3 Data Processing
The compatibility between AFO/CAFO loads distributed onto agriculture cells and
AFO/CAFO input files was checked using a hand calculation. Table 27 shows the AFO/CAFO
loads distributed in Vermont using the two methods.
Table 27. Verification of Loads Distribution Module
I N I.uskI
TSS Lo:ul
i c i; i-0;ui
rs i.o:ui
Modioli
(»/s)
'IT l.o:ul (*»/s)
(»/s)
(MI'N/s)
(Ml>\/s)
Manual
10.99
16.14
1.388
1.21 x 1010
6.62 x 1010
Module
11.02
16.17
1.391
1.21 x 1010
6.66 x 1010
A shapefile was created in Arc View 3.2 under baseline regulations to confirm that high
loadings occur in rural areas (see Figure 8). The shapefile was also used to identify counties with
zero AFO/CAFO loadings. Five of these counties were selected to confirm they did not have
associated animal farms in the input files.
Other quality assurance steps include the following:
¦ A check on the data import from Excel to Oracle by summing across columns and
comparing column totals.
¦ A qualitative comparison of loads reductions was conducted between scenarios to
ensure reasonableness based on technology options and/or percent of regulated
facilities. It confirmed that higher loads in NPDES scenarios agreed with smaller
state percentages in medium categories.
¦ A check on the loading distribution algorithm by summing the number of land-
use/land-cover cells receiving loads. To compare loads distribution between
baseline and scenario, confirmed that fewer Layer A land-use/land-cover cells
(i.e. < 100 ft from the nearest RF3 reach) received loads for the scenario (due to
the 100 ft setback).
¦ A check on the delivery ratios calculated for each run. Delivery ratios from land-
cover cells to the RF3 network were compared to literature values and found to be
within acceptable ranges (SCS, 1983).
5.4 Modeling Results
Checks have been made on the distributions of predicted water quality for FC and TSS.
FC water quality standards are typically expressed in terms of geometric means. For example,
EPA water quality criteria suggest a geometric mean of 200 MPN/lOOmL as a guideline for
swimmable waters (U.S. EPA, 1986). Mean values of 100 to 200 MPN/100 mL were calculated
in Hydroregion 1 after a log transform was applied to the FC concentrations.
5-2
-------
Section 5.0
Quality Assurance
0
0.01 -2.858
2.858 - 6.042
6.042 -13.444
13.444-29.461
Figure 8. Total nitrogen loadings (g/s) by county FIPS code for baseline.
General comparisons of baseline results to National Stream Quality Accounting Network
(NASQAN) data for TSS, TN, and TP developed by EPA staff show similar ranges of values and
patterns of high and low values.
Other quality assurance steps include the following:
1. An examination of individual use-support for each run to assess reasonableness of
overall use-support and differences between scenarios (i.e., examined that
changes in individual use-support agreed with changes in loadings).
2. An examination of estimated benefits per mile of improvement. For RTI
Scenario 7, a geographic analysis was conducted to justify the large estimated
economic benefit.
3. A hand calculation of miles affected under Scenarios 6 and 7 using the Vaughan
WQL.
5-3
-------
Section 6.0
References
6.0 References
Ambrose, R.B., T.A. Wool, and J.L. Martin. 1993. The Water Quality Analysis Simulation
Program, WASP5, Parts A andB, Version 5.10. Environmental Research Laboratory,
U.S. Environmental Protection Agency, Athens, GA.
Beaulac, M.N. and K.H. Rechkow. 1982. An examination of land use-nutrient export
relationships. Water Res. Bulletin. 18:(6) 1013-1024
Bondelid, T.R., G. Ali, and G. Van Houtven. The National Water Pollution Control Assessment
Model Benefits Assessment of Stormwater Phase II Program. Draft. Prepared for the
United States Environmental Protection Agency, Office of Water, Washington, DC.
Research Triangle Institute. Research Triangle Park, NC. June 1999.
Bondelid, T., R. Dodd, C. Spoerri, and A. Stoddard. 1999. The Nutrients Version of the National
Water Pollution Control Assessment Model. Draft. Prepared for the U.S. Environmental
Protection Agency, Office of Water, Washington, DC. Research Triangle Institute.
Research Triangle Park, NC. December.
Bowie, G.L., Mills, W.B., Porcella, D.B., Campbell, C.L., Pagenkopf, J.R., Rupp, G.L., Johnson,
K.M., Chan, P.W.H., Gherini, S.A. and C.E. Chamberlin. 1985. Rates, Constants, and
Kinetics Formulations in Surface Water Quality Modeling (SecondEdition), EPA/600/3-
85/040, Environmental Protection Agency, Athens, GA, June. [Available in Adobe
Acrobat format at: www.epa.gov//ordntrnt/ORD/WebPubs/surfaceH20/surface.html
Brown, L., and T. Barnwell. 1987. The Enhanced Stream Water Quality Model Qual2E and
Qual2e-UNCAS: Documentation and User's Manual. Athens, GA: U.S. Environmental
Protection Agency, Environmental Research Laboratory. EPA/600/3-87/007.
Carson, R.T., and R.C. Mitchell. 1993. The Value of Clean Water: The Public's Willingness to
Pay for Boatable, Fishable, and Swimmable Quality Water. Water Resources Research
29(7): 2445-2454. July.
Chapra, S.C. 1997. Surface Water Quality Modeling. New York: McGraw Hill Publishing.
Churchill, M.A., H.L. Elmore and R.A. Buckingham. 1962. The prediction of stream reaeration
rates. American Society of Civil Engineers Journal of Sanitary Engineering Division.
88(SA4):l-46.
Covar, A.P. 1976. Selecting the proper reaeration coefficient for use in water quality models.
Presented at U.S. Environmental Protection Agency Conference on Environmental
Simulation and Modeling, April 19-22, Cincinnati, OH. EPA-600/9-76-016.
6-1
-------
Section 6.0
References
Di Toro, D.M., P.R. Paquin, K. Subburamu, and D.A. Gruber. 1990. Sediment Oxygen Demand
Model: Methane and Ammonia Oxidation. Jour. EED, ASCE, 116(5):945-987.
Eidenshink, J.E. 1992. The 1990 conterminous United States AVHRR data set. In
Photogrammetric Engineering and Remote Sensing 58(6): p. 809-813.
ESRI (Environmental Systems Research Institute). 2000a. Data & Maps Media Kit, CD 6:
North America Digital Elevation Model (grid).
ESRI (Environmental Systems Research Institute). 2000b. Data Maps Media Kit, CD5: Zip
Code and Population Data.
Jobson, H.E. 1996. Prediction of Traveltime and Longitudinal Dispersion in Rivers and
Streams. USGS Water Resources Investigations Report 96-4013.
Keup, L.E. 1985. Flowing Water Resources. Prepared for Water Resources Bulletin 21(2),
American Water Resources Association. April.
Leopold, L.B., and T. Maddock, Jr. 1953. The Hydraulic Geo9metry of Stream Channels and
Some Physiographic Implications. USGS Geological Survey Professional Paper 252.
McClelland, Nina I. 1974. Water Quality Index Application in the Kansas River Basin.
Prepared for the U.S. Environmental Protection Agency - Region 7. EPA-907/9-74-001.
Metcalf & Eddy, Inc., G.T. Tchobanoglous, and F.L. Burton. 1991. Wastewater Engineering:
Treatment, Disposal, and Reuse (Third Edition). Boston, MA: Irwin McGraw-Hill
(Series in Water Resources and Environmental Engineering).
Mitchell, R.C., and R.T. Carson. 1986. The Use of Contingent Valuation Data for Benefit/Cost
Analysis in Water Pollution Control. CR-810224-02. Prepared for U.S. EPA, Office of
Policy Planning and Evaluation. Washington, DC.
Novotny, V., and H. Olem. 1994. Water Quality Prevention, Identification and Management of
Diffuse Pollution. New York: Van Nostrand Reinhold.
O'Connor, D.J., and W.E. Dobbins. 1956. "Mechanism of Reaeration in Natural Streams,"
Trans. ASCE 123:641-666.
Owens, M., R. Edwards, and J. Gibbs. 1964. Some reaeration studies in streams. International
Journal of Air Water Pollution. 8:469-486.
Reckhow, K.H., M.N. Beaulac, and J.T. Simpson. 1980. Modeling Phosphorus Loading and
Lake Response Under Uncertainty: A Manual and Compilation of Export Coefficients.
U.S. Environmental Protection Agency, Office of Water Regulations and Standards,
Criteria and Standards Division, Washington, DC. EPA 440/5-80-011.
6-2
-------
Section 6.0
References
RTI (Research Triangle Institute). 2000a. Estimation of National Surface Water Quality
Benefits of Regulating Concentrated Animal Feeding Operations (CAFOs) Using the
National Water Pollution Control Assessment Model (NWPCAM). Prepared for the
United States Environmental Protection Agency, Office of Water, Washington, DC.
Research Triangle Institute. Research Triangle Park, NC. August.
RTI (Research Triangle Institute). 2000b. National Water Pollution Control Assessment Model
(NWPCAM) Version 1.1. Prepared for the United States Environmental Protection
Agency, Office of Water, Washington, DC. Research Triangle Institute. Research
Triangle Park, NC. November.
RTI (Research Triangle Institute). 2001. The National Water Pollution Control Assessment
Model (NWPCAM v2): Quality Review Process Report for NWPCAM2 RF3 - Reach
Routing, Hydrology, and Hydraulics Datasets. Draft Report.
RTI (Research Triangle Institute). 2002. A Review of Changes and Advances Related to the
National Water Pollution Control Assessment Model Since the Proposed AFO/CAFO
Rulemaking Process. Prepared for the U.S. Environmental Protection Agency. Research
Triangle Park, NC: Research Triangle Institute. April.
SCS (Soil Conservation Service). 1983. National Engineering Handbook by Soil Conservation
Service.
Smith, R.A, G.E. Schwarz, and R.B. Alexander. 1997. Regional Interpretation of Water Quality
Monitoring Data. Water Resources Research 33(12):2781-2798. December.
Thomann, R.V., and J.A. Mueller. 1987. Principles of Surface Water Quality Modeling and
Control. New York: Harper & Row.
U.S. ACE (Army Corp of Engineers). 1998. Analysis of Best Management Practices for Small
Construction Sites. Work performed for the U.S. Environmental Protection Agency,
Office of Wastewater Management. June.
U.S. EPA (Environmental Protection Agency). 1985. Rates, Constants, and Kinetics
Formulations in Surface Water Quality Modeling (Second Edition). Environmental
Research Laboratory, Office of Research and Development, Athens, GA.
U.S. EPA (Environmental Protection Agency). 1986. Quality Criteria for Water. U.S.
Environmental Protection Agency, Office of Water Regulations and Standards, Criteria
and Standards Division, Washington, DC. EPA 440/5-86/001.
U.S. EPA (Environmental Protection Agency). 1993. Support to the 1992 Needs Survey CSO
Cost Assessment, CSO Water Quality Modeling. EPA Contract No. 68-C9-0013.
Prepared for the U.S. Environmental Protection Agency, Washington, DC: Tetra Tech.
U.S. EPA (Environmental Protection Agency). 2000. Estimation of National Surface Water
Quality Benefits of Regulating Concentrated Animal Feeding Operations (CAFOs) Using
6-3
-------
Section 6.0
References
the National Water Pollution Control Assessment Model (NWPCAM). Prepared for the
United States Environmental Protection Agency, Office of Water, Washington, DC.
Research Triangle Institute. Research Triangle Park, NC. February.
U.S. EPA (Environmental Protection Agency). 2002a. RF3 Technical Documentation.
www.epa.gov/owow/monitoring/rf/techref.html.
USDA (Department of Agriculture). 1997. Predicting Soil Erosion by Water: A Guide to
Conservation Planning with the Revised Universal Soil Loss Equation (RUSLE).
Agricultural Handbook No. 703.
U.S. EPA (Environmental Protection Agency). 2002b. EPA Loadings Report.
U.S. EPA (Environmental Protection Agency). 2002c. Permit Compliance System.
www.epa.gov/compliance/planning/data/water/pcssys.html.
U.S. EPA (Environmental Protection Agency). 2002d. Industrial Facilities Discharge Database.
http://www.epa.gov/waterscience/BASINS/metadata/ifd.htm.
U.S. EPA (Environmental Protection Agency). 2002e. http://www.epa.gov/ost/basins.
USGS(U.S. Geological Survey). 2002a. http://www.nationalatlas.gov/landcvm.html.
USGS (U.S. Geological Survey). 2002b. Http://water.usgs.gov/pubs/wri/wri934076/.
Vaughan, William J. 1986. The Water Quality Ladder. Included as Appendix B in Mitchell,
R.C., and R.T. Carson. 1986. The Use of Contingent Valuation Data for Benefit/Cost
Analysis in Water Pollution Control. CR-810224-02. Prepared for the U.S.
Environmental Protection Agency, Office of Policy, Planning, and Evaluation.
Washington, DC.
6-4
-------
Appendix A
-------
Appendix A
Table A-l. NPS Export Coefficients
Parameter
I I'hiin
Agriculture
lores!
BOD5
34-90 (average 62)
26
5
TSS
360-672 (average 466)
1600*
256
*Note: TSS EC on agriculture land is modified using the RUSLE equation.
Table A-2. Population and RF3Lite Segment Lengths by State
Stale
Population
Total KTiLite Segment Length
Alabama
4,395,481
17,428.075
Alaska
624,523
0
Arizona
4,894,006
25,308.12
Arkansas
2,566,938
14,330.275
California
33,603,430
27,235.235
Colorado
4,139,027
29,715.86
Connecticut
3,289,062
1,615.165
Delaware
762,227
565.31
District of Columbia
513,618
30.085
Florida
15,341,185
8,184.08
Georgia
7,950,119
18,219.925
Hawaii
1,184,688
0
Idaho
1,273,309
20,762.9
Illinois
12,187,552
19,116.84
Indiana
5,979,311
11,045.41
Iowa
2,877,060
18,809.845
Kansas
2,672,387
29,638.32
Kentucky
3,988,695
14,032.635
Louisiana
4,386,033
9,859.67501
Maine
1,257,219
10,555.36
Maryland
5,212,902
3,261.64
Massachusetts
6,206,482
2,340.99
Michigan
9,907,530
14,896.01
(continued)
A-3
-------
Appendix A
Table A-2. (continued)
Slule
Population
Total Kril.ite Segment Length
Minnesota
4,820,250
26,432.81
Mississippi
2,788,415
12,988.835
Missouri
5,502,243
23,655.775
Montana
885,795
54,897.135
Nebraska
1,672,199
22,222.055
Nevada
1,879,204
11,811.455
New Hampshire
1,215,100
2,535.74
New Jersey
8,192,386
2,366.83
New Mexico
1,750,921
22,290.475
New York
18,223,519
15,388.13
North Carolina
7,762,819
15,900.53
North Dakota
631,032
20,267.665
Ohio
11,281,851
13,537.26
Oklahoma
3,383,158
24,012.525
Oregon
3,356,108
24,845.13
Pennsylvania
11,986,139
13,536.59
Rhode Island
992,011
278.33
South Carolina
3,935,123
10,035.835
South Dakota
734,993
28,963.115
Tennessee
5,539,577
13,440.56
Texas
20,398,490
75,366.515
Utah
2,164,175
17,323.19
Vermont
596,714
2691.94
Virginia
6,945,067
14,427.955
Washington
5,835,089
16,318.4
West Virginia
1,804,812
7,614.28
Wisconsin
5277833
17,848.63
Wyoming
479673
32,885.195
A-4
-------
Appendix A
Table A-3. Model Coefficients for the NWPCAM Model
(ocllkiciH
Abbreviation
Coefficient
Description
I nils
Delimit
Value
KBOD
Decay coefficient for BOD
day-1
0.075
KFS
Decay coefficient for fecal streptococci
day-1
0.168
KFC
Decay coefficient for fecal coliform
day1
0.8
KTN LOWFLOW
Decay coefficient for total nitrogen where
stream flow < 1,000 ft3/s
day1
0.3842
KTN MEDFLOW
Decay coefficient for total nitrogen where
stream flow between 1,000 and 10,000 ft3/s
day"1
0.1227
KTN HIFLOW
Decay coefficient for total nitrogen where
stream flow > 10,000 ft3/s
day1
0.0408
KTP LOWFLOW
Decay coefficient for total phosphorus
where stream flow < 1,000 ft3/s
day"1
0.268
KTP MEDFLOW
Decay coefficient for total phosphorus
where stream flow > 1,000 ft3/s
day1
0.0956
KTP OW
Decay coefficient for total phosphorus
where reach is a lake
day"1
0.3586
KNH3
Rate coefficient for oxidation of NH3 to
N03
day1
0.12
TNH3
Temperature adjustment factor for KNH3
none
1.08
KTON
Rate coefficient for hydrolysis of TON to
NH3
day1
0.075
TTON
Temperature adjustment factor for KTON
none
1.08
KTOP
Rate coefficient for transformation of TOP
to P04
day"1
0.3
TTOP
Temperature adjustment factor for KTOP
none
1.08
NPS RATIO
Ratio of CBODU to BOD5 for nonpoint
sources
mg/L:mg/L
3
CSORATIO
Ratio of CBODU to BOD5 for combined
sewer overflows
mg/L:mg/L
1.4
SOD1
Sediment oxygen demand for a reach that is
not downstream of a point source
g/m2-d
0.5
SOD2
Sediment oxygen demand for a reach that is
downstream of a point source
g/m2-d
1.5
TBOD
Temperature adjustment factor for KBOD
none
1.047
(continued)
A-5
-------
Appendix A
Table A-3. (continued)
Coefficient
Abbreviation
Coefficient
Description
I nils
Dcl'mill
Value
TSOD
Temperature adjustment factor for SOD1
and SOD2
none
1.06
TK2
Temperature adjustment factor for K2
none
1.024
TFC
Temperature adjustment factor for KFC
none
1.07
Table A-4. Agricultural Slope Factor by Hydroregion
1 lydrorciiion
Agricultural Slope l-ador
1
0.47
2
0.37
3
0.50
4
0.68
5
0.37
6
1.00
7
0.97
8
0.48
9
1.07
10
0.50
11
061
12
0.77
13
1.00
14
1.00
15
0.37
16
1.16
17
0.72
18
0.15
A-6
-------
Appendix B
-------
Appendix B
Charts from http://pathfinderscience.net/stream/cp4wqi.cfm.
Chart 1: Dissolved Oxygen (DO] Test Results
DO: % saturation
Sot« 4 DC % viluraion > 1 *& 0. O—SO O
B-3
-------
Appendix B
CAIjCULATING the results
Chart 2: Fecal Coliform (FC) Test Results
FC: colonies/lOO ml
Note: if FC > 10s, Q=2.0
100
90
80
70
60
Q-value
50
40
30
20
10
0
SO 200 500 2,000 5,000
100 I.OOO 10
j 20,000 50,000
,ooo 100,000
B-4
-------
Appendix B
Chart 6: Total Phosphate (as PO^-P] Test Results
60
Q-value
50
40
30
20
10
P04-P: mg/1
Note it PQ,-P > i0.a o«2.o
B-5
-------
Appendix B
Chart 7: Nitrate (as NOJ Test Results
NO^mg/I
Not# if WjilQO.O Q*l,0
B-6
-------
Appendix B
Chart 9: Total Solids [TS] Test Results
Q-value
100
90
60
70
60
50
40
30
20
10
0
0 50 100 150 200 Z50 3OT 350 4G0 450 900
TS:mg/l
Nf>to if TS > 50C.Q, O-20.0
B-7
------- |