United States EPA/600/R-15/061 | March 2015
Environmental Protection www.epa.gov/ord
Agency
Ecosystem Services and Environmental
Markets in Chesapeake Bay Restoration
Final Report
Project Officers:
Naomi Detenbeck
Brenda Rashleigh
US EPA, Atlantic Ecology Division,
Office of Research and Development
27 Tarzwell Drive,
Narragansett, Rl 02882
Prepared by
George Van Houtven
Ross Loom is
Justin Baker
RTI International
3040 Cornwallis Road
Research Triangle Park, NC 27709
EPA Contract Number: EPA-11-029
Office of Research and Development
National Health and Environmental Effects Research Laboratory
-------
NOTICE
The U.S. Environmental Protection Agency through its Office of Research and Development
funded and managed the research described here under contract no. EPA-11-029 to RTI
International. It has been subjected to the Agency's peer and administrative review and has been
approved for publication as an EPA document. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
11
-------
CONTENTS
Section Page
Notice ii
List of Figures v
List of Tables vi
Abstract vii
Section 1 Introduction 1-1
Section 2 Optimization Framework 2-1
2.1 Model Overview 2-1
2.2 Significant Wastewater Treatment and Industrial Dischargers 2-3
2.2.1 Cost and Effectiveness of Nutrient Controls at Significant
Wastewater Treatment and Industrial Dischargers 2-3
2.2.2 Aggregate Nutrient Reduction Targets for Significant Wastewater
Treatment and Industrial Dischargers 2-3
2.3 Agricultural Nonpoint Sources 2-4
2.3.1 Costs, Effectiveness, and Co-benefits of Agricultural Nutrient
Controls 2-4
2.3.2 Eligibility of Agricultural Nutrient Controls 2-8
2.3.3 Nutrient Reduction Target for Agricultural Sources 2-8
Section 3 Including Co-Benefits from Freshwater quality improvements in the
optimization Framework for Chesapeake Bay Restoration 3-1
3.1 Introduction 3-1
3.2 Freshwater Quality Benefits of Reduced Nutrient Loads 3-1
3.3 Scenario Analysis Results 3-7
Section 4 Alternative Market-Based Incentive designs for
Chesapeake Bay Restoration 4-1
4.1 Introduction 4-1
4.2 Nutrient Credit Trading and Other Incentives to Meet Agricultural
Baseline Requirements 4-2
4.2.1 Nutrient Credit Trading and the Role of Trading Baselines 4-2
in
-------
4.2.2 Additional Incentives through Public Sector Subsidies or Credit
Purchases 4-5
4.2.3 Incentive Scenarios 4-6
4.2.4 Results 4-8
4.3 Impacts of Alternative Methodologies for Estimating Nutrient Reductions
from Nonpoint Source Controls 4-11
4.3.1 Estimating Nutrient Reductions from Nonpoint Source Controls 4-13
4.3.2 Alternative Crediting Scenarios 4-14
4.3.3 Results 4-15
Section 5 Conclusions 5-1
Appendix A. Co-Benefits from Agricultural Best Management Practices A-1
Appendix B. Data Quality Assurance B-l
References R-l
IV
-------
LIST OF FIGURES
Number Page
2-1. Overview of the Optimization Framework 2-2
3-1. Without-TMDL Water Quality Index (100-point Scale) Values 3-3
3-2. Change in Water Quality Index (100-point Scale) Values with the TMDL 3-4
3-3. Value ($) per Pound of Nitrogen Reduced by River Segment 3-7
3-4. Least-Cost Scenario Nitrogen Load Reductions in the Susquehanna-PA by River
Segment (Delivered Pounds per Year) 3-9
3-5. Least-Cost Scenario Phosphorus Reductions in the Susquehanna-PA by River
Segment (Delivered Pounds per Year) 3-10
3-6. Difference in Nitrogen Reductions from Least-Net-Cost Source Controls and Least-
Cost Source Controls in the Susquehanna-PA by River Segment (Delivered Pounds
per Year) 3-11
3-7. Difference in Phosphorus Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the Susquehanna-PA by River Segment (Delivered
Pounds per Year) 3-11
3-8. Least-Cost Scenario Nitrogen Reductions in the James-VA by River Segment
(Delivered Pounds per Year) 3-12
3-9. Least-Cost Scenario Phosphorus Reductions in the James-VA by River Segment
(Delivered Pounds per Year) 3-13
3-10. Difference in Nitrogen Reductions from Least-Net-Cost Source Controls and Least-
Cost Source Controls in the James-VA by River Segment (Delivered Pounds per
Year) 3-15
3-11. Difference in Phosphorus Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the James-VA by River Segment (Delivered Pounds
per Year) 3-16
-------
LIST OF TABLES
Number Page
2-1. Nutrient Load Reduction Targets by Major Basin, Jurisdiction, and Sector 2-5
2-2. Agricultural Land Uses in the Chesapeake Bay Watershed Model and Optimization
Framework 2-6
2-3. Agricultural Best Management Practices in the Chesapeake Bay Watershed Model
and Optimization Framework 2-7
3-1. Costs and Benefits of Significant Wastewater Treatment and Industrial Facilities
and Agricultural Nonpoint Source Nutrient Reductions in the Susquehanna-VA
(Million$) 3-9
3-2. Costs and Benefits of Significant Wastewater Treatment and Industrial Facilities
and Agricultural Nonpoint Source Nutrient Reductions in the James-VA (Million $).... 3-12
4-1. Policy Scenario Results for the Susquehanna River Basin in Pennsylvania 4-9
4-2. Policy Scenario Results for the James River Basin in Virginia 4-12
4-3. Nutrient Credits Generated by Best Management Practices in the James River Basin
in Virginia 4-14
4-4. Potential Impacts of Uniform Crediting on Environmental Outcomes and Cost of
Source Controls under Nutrient Trading 4-14
4-5. Actual vs. Credited AgNPS Load Reductions and Control Costs Under Alternative
Crediting Approaches 4-17
A-l. Assumed GHGEmission Factors for Selected Land Uses A-2
A-2. Per-Acre Value of Carbon Sequestration Services from Land-Use Conversion ($/ac) A-5
A-3. Duck Energy Days per Acre for Selected Land Cover Types A-7
A-4. Incremental Annual Value of Duck Hunting Services per Acre of Wetland Restoration
(2010$) A-7
A-5. Annual Value of Nonwaterfowl Hunting Services in the Chesapeake Bay
Watershed (2008 $) A-9
VI
-------
ABSTRACT
This report contains two separate analyses, both of which make use of an optimization
framework previously developed to evaluate trade-offs in alternative restoration strategies to
achieve the Chesapeake Bay Total Maximum Daily Load (TMDL). The first analysis expands on
model applications that examine how incorporating selected co-benefits of nutrient reductions
into the optimization framework alters the optimal distribution of nutrient reductions in the
watershed (U.S. EPA, 2011). In previous applications, the analyzed co-benefits included carbon
sequestration and recreational hunting benefits from certain agricultural best management
practices (BMPs). In this report we expand the optimization framework to also include benefits
from water quality improvements in freshwater river and streams. We find that these nontidal
water quality co-benefits are larger than the other co-benefits combined and would result in
greater nutrient control efforts in upstream portions of the watershed. Compared to cost-
minimization results that do not account for co-benefits, including all co-benefits in the
optimization would increase annual nutrient control costs by $16 million in the Susquehanna
River Basin in Pennsylvania; however, the co-benefits would increase by $31 million, for a net
gain of $15 million per year. In the James River Basin in Virginia, considering monetized co-
benefits results in an estimated increase in nutrient control costs of $17 million but an increase in
co-benefits of $42 million (net gain of $25 million per year).
The second analysis expands on previous applications of the optimization framework that
have focused on the potential cost savings from allowing nutrient trading in the Chesapeake Bay
watershed (Van Houtven et al., 2012). These applications do not include the co-benefit estimates.
Instead they examine how the costs of achieving TMDL goals could be reduced under alternative
trading scenarios. For this report, we apply the optimization framework to assess how nutrient
trading may interact with other incentives for agricultural nutrient reductions, as well as how
simplified crediting of nutrient reductions influences the nutrient control costs, load reductions,
and participation in a nutrient trading market. We estimate that nutrient trading can act as an
incentive for some agricultural entities to adopt nutrient controls and meet their load allocation
under the TMDL. However, we also find that the incentive of nutrient trading alone would only
support achieving 11 percent of the required agricultural nitrogen load reductions in the
Susquehanna River Basin in Pennsylvania and 4 percent of the required agricultural phosphorus
reductions. In the James River Basin in Virginia, we estimate nutrient trading would be a more
effective incentive to achieve the required agricultural nutrient reductions, with 35 percent of the
nitrogen reduction and 41 percent of the phosphorus reduction achieved through nutrient trading.
vn
-------
Finally, we estimate that simplified crediting of nutrient reductions results in higher costs
(by 8 percent across the watershed) for achieving significant wastewater and industrial discharge
nutrient reductions through nutrient trading because it discourages placement of nutrient controls
where they would be most effective. In addition, simplified crediting of nutrient trading is
estimated to result in failure to meet the load reduction requirements in 11 of the 14 basin-state
combinations in the Chesapeake Bay watershed due to practices in certain agricultural areas
receiving more credit for nutrient reductions than would be achieved.
Vlll
-------
SECTION 1
INTRODUCTION
In 2010, EPA established the Chesapeake Bay Total Maximum Daily Load (TMDL), which
limits the amount of nutrients and sediment that enter the largest estuary in the United States.
The goal of these nutrient and sediment limits is to achieve water quality standards for dissolved
oxygen (DO), water clarity, submerged aquatic vegetation (SAV) and chlorophyll-a. To meet the
TMDL, sufficient controls must be in place by 2025 to reduce nitrogen, phosphorus, and
sediment by 25 percent, 24 percent, and 20 percent respectively relative to 2009 conditions
(U.S. EPA, 2010).
In the development of the TMDL, five of the seven jurisdictions within the Bay watershed
agreed to a methodology to allocate needed load reductions based on their controllable load and
relative effectiveness of improving DO in the Bay's main channel. Jurisdictions then developed
watershed implementation plans (WIPs) describing what nutrient and sediment controls will be
implemented to achieve their load allocation. The WIPs also provide reasonable assurances that
controls for sources not regulated under the Clean Water Act, including much of the agricultural
sector, will be implemented through regulatory or voluntary programs.
While all of the source controls described in the WIPs will reduce nutrients and/or sediment,
some source controls provide additional environmental co-benefits. For example, in addition to
retaining nutrients and sediment, a restored wetland provides multiple ecosystem services
including water retention that reduces downstream flood risk, wildlife habitat that can improve
recreational experiences such as wildlife watching and waterfowl hunting, and carbon
sequestration to reduce the risk of damages from climate change.
In 2011, EPA issued a report evaluating how considering these co-benefits might change the
costs and benefits of implementing alternative combinations of source controls to achieve the
TMDL (U.S. EPA, 2011). To examine tradeoffs among alternative implementation strategies, an
optimization framework was developed to estimate the least-cost combination of source controls
needed to meet the TMDL and the least-net-cost combination of source controls, which was
defined as the difference between monetized co-benefits and costs.
More recently, the optimization framework has been applied by the Chesapeake Bay
Commission to analyze potential market-based approaches to support more cost-effective Bay
restoration, such as nutrient trading (Van Houtven et al., 2012). Nutrient trading is a mechanism
through which entities with high costs of nutrient reduction can purchase nutrient reduction
1-1
-------
credits from other entities that reduce nutrient loads beyond their required level. Setting aside the
co-benefit estimates, Van Houtven et al. (2012) used the cost-minimization features of the
optimization framework to simulate conditions under alternative nutrient trading scenarios and to
estimate potential cost savings from trading.
The purpose of this report is to build on these two previous applications of the optimization
framework through two separate analyses. In both cases, the purpose is to evaluate the economic
trade-offs and implications of alternative restoration strategies for achieving the Chesapeake Bay
TMDL; however, the first analysis examines the implications of expanding the types of co-
benefits included in the framework, and the second analysis focuses on the implications of
alternative incentive-based approaches for achieving the TMDL.
The main objective of the first analysis is to expand on the approach of U.S. EPA (2011) to
include co-benefits from water quality improvements upstream from the Bay itself. The earlier
analysis primarily valued co-benefits that were independent of the nutrient and sediment
reductions, and thus omitted the potentially important co-benefits derived from water quality
improvements in the tributaries and main stem of the Bay. To address this shortcoming, the
current analysis incorporates estimates of the nontidal (i.e., freshwater river and stream) water
quality improvement co-benefits associated with the TMDL. It incorporates these estimates into
the optimization framework to investigate how their inclusion influences the optimal selection
(based on economic considerations) of available projects to reduce nutrient and sediment loads to
the Chesapeake Bay.
The main objective of the second analysis is to expand on the application in Van Houtven et
al. (2012) by assessing (1) how nutrient trading may interact with other incentives for
agricultural nutrient reductions and (2) how simplified crediting of nutrient reductions influences
the control costs, load reductions, and participation in a nutrient trading market. The current
analysis explores how nutrient trading, which would allow agricultural entities to sell credits for
any load reductions that exceed their required reductions, can also provide an incentive for these
sources to meet their required reductions. In other words, to what extent can nutrient trading help
to achieve the dual objectives of reducing TMDL costs and providing incentives to meet TMDL
load reduction requirements? We also evaluate how other incentive-based approaches, such as
public funding for agricultural nutrient reductions, may interact with a nutrient trading market
to achieve these objectives. In addition, the analysis explores how alternative nutrient crediting
methodologies aimed at reducing transaction costs of offset and trading programs, such as
uniformly crediting practices throughout a river basin, may impact water quality and the
potential cost savings from nutrient trading.
1-2
-------
The document is organized as follows:
• Section 2 describes the general optimization framework and input data used for the
subsequent analyses of alternative policies and approaches for meeting the TMDL.
• Section 3 details how including the upstream water quality co-benefits of nutrient
reductions within the optimization framework changes the economically optimal
distribution of nutrient reductions in the watershed.
• Section 4 discusses the analysis of alternative designs for incentive-based systems
to meet TMDL goals.
• Section 5 summarizes the main findings and limitations of the analysis and discusses
implications.
1-3
-------
-------
SECTION 2
OPTIMIZATION FRAMEWORK
2.1 Model Overview
As noted in Section 1, the main purpose of the optimization framework is to identify the
combinations of available nutrient control projects in the Chesapeake Bay watershed that
together achieve the TMDL goals at lowest overall cost (or costs net of co-benefits) to society.
In other words, within this framework, the optimal solution is one that minimizes the cost (or net
costs) of nutrient controls, subject to the load reduction constraints defined by the TMDL.
Figure 2-1 provides a graphical representation of the main components of the optimization
framework and how they are connected.1 In this framework the first step is to define the total
load reduction targets, which represent the main model constraints. For each defined geographic
area (i.e., major basin and/or jurisdiction), total nutrient load reductions must be at least as large
as those required by the TMDL. As described in more detail below, we defined separate load
reduction targets for the two main source sector categories included in this analysis—significant
wastewater treatment and industrial facilities and agricultural nonpoint sources—across the main
basins and jurisdictions. These sources contribute more than half of the annual nutrient loads
to the Bay. As a simplification, the analysis does not include load reduction targets or nutrient
control projects for urban stormwater sources, septic systems, concentrated animal feeding
operations, or nurseries. The implications of including these sources are left for investigation in
future analyses.
The next step is to create an inventory of potential nutrient control projects. These projects
include discrete upgrades at significant wastewater treatment and industrial facilities, as well as
agricultural best management practices (BMPs) that can be adopted across the landscape. Next,
each project must be assigned estimates of its annual nutrient load reductions to the Bay and its
annual costs. As feasible, we also develop and assign ecosystem service co-benefit estimates
(in monetary terms) to the projects. In previous model applications, these monetized co-benefits
were primarily comprised of carbon sequestration and recreational hunting benefits, which are
described in more detail below and in Appendix A. Combining the previous steps, the net costs
of each project are calculated as the difference between its annual costs and its annual monetized
co-benefits.
A more detailed and technical discussion of the optimization model and framework is provided by Van Houtven et
al. (2012). Additional details, particularly regarding co-benefit estimates, are provided by U.S. EPA (2011).
2-1
-------
Inventory of Sources
& Control Projects
Total Load
Reduction Targets
Project Costs
& Load Reductions
Project Co-Benefits
v v
Project NET Costs
& Load Reductions
OPTIMIZATION
ANALYSIS
Least-Cost Solution
• Selected Projects
• Total Control Costs
•Total Co-Benefits
Least-NET-Cost Solution
• Selected Projects
• Total Control Costs
•Total Co-Benefits
•Total NET Costs
Figure 2-1. Overview of the Optimization Framework (Adapted from U.S.EPA, 2011)
The modeling framework for this analysis is built around and expands on the Chesapeake
Bay Program's Phase 5.3.2 Watershed Model (CBWM).1 CBWM subdivides the watershed into
a network of 2,468 "land-river segments" and simulates the movement of nutrients through the
network. Importantly, it serves as the main accounting framework for estimating compliance
with the TMDL goals for nutrient and sediment loads.
The optimization component of our analytical framework is defined as a mixed integer linear
programming (MTLP) optimization problem. Because agricultural nonpoint source controls can
be adopted continuously (i.e., on an acre-by-acre basis) across a landscape, these controls can be
modeled using a linear programming approach. In contrast, upgrade options at significant
wastewater treatment and industrial facilities are best characterized as discrete, all-or-nothing
source control options; therefore, a binary integer decision variable is required to reflect their
adoption. The MILP is solved using the CPLEX solver in the General Algebraic Modeling
System.
As shown in Figure 2-1, the optimization model can be applied to solve for two general types
of solutions—a least-cost solution or a least-net-cost solution. The only difference is that the
least-cost solution does not consider co-benefits as part of the optimization process. They can be
The 2011 analysis (U.S.EPA, 2011) was conducted using an earlier version of the CBWM.
2-2
-------
estimated for the projects in the solution set, but they do not affect which projects are selected.
In contrast, co-benefits are directly factored into the net costs which are minimized as part of the
second type of solution.
For each solution type, the optimization model identifies (1) the optimal control technology
for each point source and (2) the optimal number of additional acres of each BMP in each land-
river segment. The resulting total costs, load reductions, and co-benefits for each area can then
be calculated by summing across the selected projects.
Below we provide additional details about how the inputs for the optimization analysis are
defined for point sources and agricultural nonpoint sources.
2.2 Significant Wastewater Treatment and Industrial Dischargers
2.2.1 Cost and Effectiveness of Nutrient Controls at Significant Wastewater Treatment and
Industrial Dischargers
To represent point-source projects in our inventory, we developed estimates of nutrient
reductions and annual costs of upgrades at significant wastewater treatment and industrial
facilities (PS) within the Chesapeake Bay watershed based on data provided by EPA (CBPO,
2012). Available upgrades for each facility are represented by different combinations of nitrogen
and phosphorus effluent concentration targets. For nitrogen, the upgrade options are below 8
mg/L, 5 mg/L, and 3 mg/L, and for phosphorus they are below 1 mg/L, 0.5 mg/L, and 0.1 mg/L.
For each upgrade option at each facility, the annual end-of-pipe nutrient reductions were
calculated as the difference between the current and new nutrient concentrations multiplied by
the annual treated wastewater flow. These end-of-pipe nutrient reductions were further adjusted
to account for instream attenuation estimated in the CBWM between the facility and the tidal
waters of the Chesapeake Bay. In other words, using this attenuation adjustment, we estimated
the reductions in "delivered" loads to the Bay for each upgrade option. Annual costs for each
upgrade option were based on estimates of (1) annual operation and maintenance costs for the
technology and (2) the one-time capital costs of the technology, annualized using a 20-year
assumed lifetime and 7 percent discount rate.1
2.2.2 Aggregate Nutrient Reduction Targets for Significant Wastewater Treatment and
Industrial Dischargers
We estimated the aggregate nutrient reduction targets for the significant wastewater and
industrial sector consistent with EPA's 2012 draft cost analysis assumptions (CBPO, 2012).
A detailed description of the cost estimates are available in Appendix A of Van Houtven et al. (2012).
2-3
-------
Reduction targets are only included for facilities with 2010 nutrient discharges above their
TMDL-assigned waste load allocation. For these facilities, their required nutrient reduction is
equal to the difference between their 2010 and TMDL-assigned nutrient concentration
(a) multiplied by the facilities' 2010 treated wastewater flow to estimate annual end-of-pipe
loads, and (b) adjusted to reflect instream attenuation between the facility and the tidal waters
of the Chesapeake Bay (Table 2-1). The aggregate nutrient reduction targets were defined and
calculated as the sum of delivered load reductions across these facilities within a basin-state.
These aggregate nutrient reduction targets are included as constraints in the optimization
framework to ensure that the set of nutrient controls selected by the optimization routine
combine across both sectors to meet the required reductions for the significant point-source
sector. In other words the load reduction requirements for this sector can in part be met through
reductions in the agricultural sector.
2.3 Agricultural Nonpoint Sources
2.3.1 Costs, Effectiveness, and Co-benefits of Agricultural Nutrient Controls
The required data for representing agricultural nonpoint-source (AgNPS) controls were
developed primarily using data from the CBWM, Scenario Builder, cost data provided by EPA
(CBPO, 2012), and co-benefit estimates described in U.S. EPA (2011). Data from Scenario
Builder describe the effectiveness of BMPs as well as their estimated level of implementation in
the Bay watershed. Output data from the CBWM describe how 35 land uses and other sources
from 2,468 modeled land-river segments contribute nutrients to the tidal waters of the
Chesapeake Bay. Within the optimization framework, we aggregate the 14 agricultural land uses
specified in the CBWM into 5 land use categories (Table 2-2). We include 14 agricultural BMPs
in the optimization framework (Table 2-3). Ten of the BMPs and their possible combinations
(52 possibilities) are eligible on cropland and 10 BMPs and their possible combinations,
(57 possibilities) are eligible on pastureland. Nutrient reductions for some BMPs are based on
an efficiency estimate, which specifies a percentage reduction in edge-of-stream loads when
applied, while others are based on a conversion to a land use with lower nutrient contributions,
such as replacing cropland with forest.
2-4
-------
Table 2-1. Nutrient Load Reduction Targets by Major Basin, Jurisdiction, and Sector
Significant Industrial and
Wastewater Treatment Facilities Agricultural Nonpoint Sources
Major Basin
Eastern Shore
Eastern Shore
Eastern Shore
Eastern Shore
James
James
Patuxent
Potomac
Potomac
Potomac
Potomac
Potomac
Rappahannock
Susquehanna
Susquehanna
Susquehanna
Western Shore
York
Total
Jurisdiction
Delaware
Maryland
Pennsylvania
Virginia
Virginia
West Virginia
Maryland
Washington D.C.
Maryland
Pennsylvania
Virginia
West Virginia
Virginia
Maryland
New York
Pennsylvania
Maryland
Virginia
N
16,265
271,242
0
198,746
9,565,746
0
47,552
1,530,618
218,557
55,572
716,517
109,952
57,861
0
614,862
4,725,252
4,650,796
502,936
23,282,474
P
0
21,579
0
2,274
596,542
0
15,571
0
36,411
32,775
112,135
78,777
18,922
0
121,018
256,906
178,742
19,790
1,491,442
N
603,721
3,924,500
117,750
551,696
1,388,383
3,514
296,692
0
2,075,880
917,129
1,834,727
147,983
1,225,637
472,586
1,887,455
20,667,783
0
749,634
36,865,070
P
0
0
2,428
23,199
335,755
0
24,055
0
139,891
72,068
309,500
20,669
230,863
14,397
137,910
491,755
6,694
50,652
1,859,836
2-5
-------
Table 2-2. Agricultural Land Uses in the Chesapeake Bay Watershed Model and
Optimization Framework
Chesapeake Bay Watershed Model
Optimization Framework
Alfalfa
Alfalfa nutrient management
Hay without nutrients
Hay with nutrients
Hay with nutrients nutrient management
Hay
High-till without manure
High-till with manure
High-till with manure nutrient management
High-till without manure nutrient management
High-Till
Low-till with manure
Low-till with manure nutrient management
Low-Till
Degraded riparian pasture
Degraded Riparian Pasture
Pasture
Pasture nutrient management
Pasture
2-6
-------
Table 2-3. Agricultural Best Management Practices in the Chesapeake Bay Watershed
Model and Optimization Framework
Agricultural Land Use
Eligible Best Management Practices
Hay, Cropland, and Pastureland
Conservation Plans
Forest Buffers
Grass Buffers
Land Retirement
Tree Planting
Wetland Restoration
Hay and Cropland
Decision Agriculture
Enhanced Nutrient Management
Cropland
Continuous No Till
Cover Crops3
Pastureland
Pasture Alternative Watering
Prescribed Grazing
Precision Intensive Rotation Grazing
Degraded Riparian Pasture
Stream Access Control with Fencing
aWhile the CB WM includes multiple cover crop options, the optimization framework includes only Early Drilled
Rye, which is the least costly and most effective (according to the CB WM framework, which must be used for
TMDL accounting) cover crop option.
Two main categories of monetized co-benefits were included in the U.S. EPA (2011) report
for selected AgNPS BMPs. The first category—carbon co-benefits—includes unit value (dollars
per acre per year) estimates for changes in carbon sequestration and changes in greenhouse gas
(GHG) emissions. Carbon sequestration benefits apply to BMPs involving land use conversion
from crop or pastureland to forest, wetland, or grass cover. Changes in GHG emissions were also
associated with land conversion BMPs and "working land" BMPs that reduce fertilizer
application. The second category—recreational hunting co-benefits—includes unit value
estimates associated with increases in wetland cover (waterfowl hunting benefits) and increases
in forest cover (nonwaterfowl hunting benefits). In Section 3 of this report, we expand the list of
co-benefits to also include values for improving water quality in rivers and streams.
To estimate nutrient reductions associated with a BMP or combination of BMPs, we first use
data from CBWM to estimate the per-acre nutrient loads delivered to the Bay from each land use
and existing BMP combination within each land-river segment. We then determine which BMP
or BMP combinations could be added to these existing ones. For each potential option, we then
apply BMP-specific load removal efficiencies and attenuation factors from CBWM to estimate
reduced loads to the Bay. For example, if there are 100 acres of high till cropland in a land-river
segment, none of which currently have BMPs in place, and they deliver 1,500 Ibs of nitrogen to
2-7
-------
the Chesapeake Bay each year (15 Ib N/acre), we would estimate that applying a BMP with a
40 percent removal efficiency to these acres would reduce loads by 600 Ibs per year.
Alternatively, if the 100 acres deliver 1,500 Ibs per year, but half of those acres currently use a
BMP with a 50 percent efficiency, then the 50 acres with BMPs would be estimated to deliver
10 Ib N/acre, while the 50 acres without a BMP would deliver an estimated 20 Ib N/acre. Adding
the BMP with 40 percent efficiency to the area without BMPs would reduce loads by 400 Ibs per
year, whereas it would reduce loads by only 200 Ibs per year if added to the area that already
uses one BMP. In other words, nutrient reductions from implementing a specific BMP depend on
the current estimated delivered loads per acre, which depends on the currently implemented
BMPs.
2.3.2 Eligibility of Agricultural Nutrient Controls
Eligibility for BMP implementation on agricultural lands is determined by the type of land
use (e.g., crop or pasture) and the set of BMPs currently implemented. For example, if cover
crops are already implemented on a portion of cropland acres, cover crops would not be an
eligible BMP in that same area. For some BMPs, eligible acres are identified by the following
site-suitability criteria. Land retirement is restricted to occur only on highly erodible soils,
wetland restoration is restricted to only hydric soils, and forest and grass buffers are restricted
to only the riparian area remaining after subtracting the acres of forest and grass buffer
implemented in the CBWM. This combination of land-river segment, land use, existing BMP
combinations, and highly-erodible soils, hydric soils and riparian areas constitutes the unit of
analysis for agricultural source controls in the optimization framework.
2.3.3 Nutrient Reduction Target for Agricultural Sources
We estimate the aggregate nutrient reduction targets for the agricultural nonpoint sector by
taking the difference between the nutrient contribution of the included agricultural land uses in
2010 and with the TMDL1 (Table 2-1). These nutrient reduction targets are included in the
optimization framework as constraints to ensure that the set of selected nutrient controls combine
to meet the agricultural nonpoint sector's required reductions. In addition to including the
nutrient reduction target as a constraint, where specified, we impose a constraint on the
conversion of agricultural land to other land uses.
The CBWM TMDL scenario used is based on the jurisdictions' Phase I WIPs.
2-8
-------
SECTION 3
INCLUDING CO-BENEFITS FROM FRESHWATER QUALITY IMPROVEMENTS IN
THE OPTIMIZATION FRAMEWORK FOR CHESAPEAKE BAY RESTORATION
3.1 Introduction
Nutrient control projects being implemented throughout the watershed in order to comply with
the TMDL are addressing water quality concerns in the Chesapeake Bay estuary, but will
simultaneously generate ancillary benefits in nontidal rivers and streams by improving water quality.
Reduced nutrient loads to upstream catchments will improve conditions within those catchments, as
well as along the downstream river network to the Bay. These improvements will enhance freshwater
ecosystem services and benefit people living inside and outside the watershed.
In this section, we expand our optimization framework to include monetized estimates for these
freshwater co-benefits. This effort supplements prior work (U.S. EPA, 2011) in which we monetized
co-benefits from reductions in atmospheric GHGs and recreational hunting benefits from increased
forest cover and wetland restoration. We found that the GHG co-benefits generated the vast majority
of monetized co-benefits from nutrient source controls that achieve the TMDL. In this section, we
examine how including freshwater co-benefits into the optimization framework influences the
selection and geographic distribution of nutrient source controls.
3.2 Freshwater Quality Benefits of Reduced Nutrient Loads
To include these freshwater co-benefits in the optimization framework, we must estimate the
value of reducing nitrogen and phosphorus loads to streams and rivers throughout the watershed. In
particular, we must estimate average per-pound values for reduced edge-of-stream loads in each river
segment. We do this using the following four steps.
1. For each nontidal river segment in the Bay watershed, we acquire water quality
improvement estimates from the CBWM, based on a watershed-wide load reduction
scenario.
2. We use economic benefit transfer methods to estimate the total annual value of these
water quality improvements in each river segment.
3. We estimate the average per-pound value of reduced nutrient loads received by each
segment. To calculate this value we divide the total value of water quality improvements
in the segment (from the previous step) by the sum of reduced loads from nutrient sources
in the segment itself and from all upstream river segments (accounting for instream
attenuation of nutrients between segments).
4. We estimate the average per-pound value of reducing nutrient loads discharged to surface
water in each river segment. Using results from the previous step, we take the segment's
3-1
-------
per-pound value for received loads, and we add the per-pound values^br all downstream
segments (adjusting for instream attenuation between segments).1
Below, we describe each of these four steps in more detail.
For the first step, we use results from CBWM model runs to define water quality improvements
for a specific watershed-wide load reduction scenario. The selected scenario compares estimated
nutrient loads in 2009 (i.e., "without-TMDL") to load estimates based on implementation of the
Phase II WIPs (i.e., "with-TMDL"). For both with- and without-TMDL load conditions, the CBWM
provides daily average concentration estimates in each nontidal (i.e., freshwater) river segment for six
main water quality parameters: total nitrogen (TN), total phosphorus (TP), DO, chlorophyll-a, total
suspended solids (TSS), and 5-day biochemical oxygen demand (BODS). For our analysis, we use
these estimates to calculate average annual concentrations for each parameter within a river segment.2
For the second step, we apply a benefit transfer approach to approximate the total economic value
(i.e., use plus non-use benefits) of changes in these water quality parameters resulting from the load
reduction scenario. Our approach is similar to methods used by EPA's Office of Water to estimate the
benefits of proposed and promulgated effluent guidelines (e.g., U.S. EPA, 2013). The approach
requires several simplifications, recognizing that developing precise monetary estimates of the
benefits that humans receive from water quality improvements is difficult because the links between
changes in these water quality parameters and their effects on human well-being are complex and
multifaceted. The benefits of water quality improvements are likely to include in situ uses
(recreational activities and aesthetic appreciation), as well as non-use benefits for those who simply
derive value from knowing that water resources are being protected.
The benefit transfer approach has three main components. First, for each nontidal (freshwater)
river segment and load condition (i.e., with- or without-TMDL), we combine the multiple water
quality parameter values from the CBWM into a single composite water quality index (WQI) (Figure
3-1). The WQI is a nonlinear combination of subindices for each of the six water quality parameters
listed above. Each of the six subindices is derived from a separate nonlinear function, which
translates concentrations of the water quality parameter to a 100-point scale.3 The water quality
improvement for each river segment due to the TMDL is therefore expressed as a change in the WQI
1 The per-pound value for received loads approximates the value that each pound discharged in the segment has on water
in the segment. The per-pound values for downstream segments approximates the value that each pound discharged in
the segment has on water in the downstream segments.
2 Water quality parameters for the tidal river segments closest to the Bay itself are not simulated in the CBWM.
3 For some of these parameters, including TN and TP, the coefficients and upper and lower bounds of these functions vary
across ecoregions in the United States (including within the Chesapeake Bay watershed).
3-2
-------
Rivers
Major Basins
Minor Basins
Baseline WQI
14-29
30-38
39-47
46-55
56-61
^B °7 - '2
^B 78-94
^H 85 - 94
| | States
Not Simulated River Segments
Figure 3-1. Without-TMDL Water Quality Index (100-point Scale) Values
between the without-TMDL and with-TMDL load conditions (Figure 3-2).1 Second, to value these
water quality improvements, we apply a benefit transfer function that translates changes in WQI into
an average household-level willingness to pay (WTP) for the improvement. This benefit transfer
function is reported in Van Houtven et al (2007) and is based on a meta-analysis of 131 WTP
estimates from 18 water quality valuation studies conducted in the United States. Third, we apply
1 For this analysis, we assume that this load reduction scenario (combining the with- and without-TMDL load scenarios)
provides representative average per-pound values for load reductions in the watershed. Additional investigation will
be needed to determine how much these average values would vary using alternative scenarios.
3-2
-------
these WTP values to estimate aggregate state-level benefits for WQI improvements in each river
segment. Specifically, we multiply the WTP values for each segment by (1) the total number of
households residing in the state where the segment is located and (2) the percentage of each state's
total river miles that are located in the segment.
Legend
"— Rivers
Major Basins
Minor Basins
Change in WQI
Figure 3-2. Change in Water Quality Index (100-point Scale) Values with the TMDL
For the remaining steps, the objective is to transform the values for water quality improvements in
each segment to values for load reductions from each segment. That is, for each freshwater river
3-4
-------
segment, we need to estimate the average per-pound value of a load reduction, as it contributes to
water quality improvements in its own segment, as well as in all downstream segments.
In the third step, we estimate the total amount of nutrients entering the river segment both from
within the river segment and from all upstream river segments. To estimate the nutrient contribution
from an upstream segment to a connected downstream river segment, we account for instream
attenuation of nitrogen and phosphorus between the segments. These instream attenuation rates
between segments are derived from the CBWM's attenuation factors, which are expressed as delivery
ratios between each segment and the Bay tidal waters. We estimated attenuation between segments
by dividing the delivery ratio of the upstream river segment by the delivery ratio of the receiving
segment. For example, if every pound of nitrogen from an upstream river segment^ contributed half
a pound of nitrogen to the Bay (delivery ratio = 0.5) and every pound from a downstream segment B
contributed three-fourths of a pound to the Bay (delivery ratio = 0.75), then we estimated the
attenuation rate between river segment^ and B to be two-thirds (0.667 = 0.5/0.75).
In addition, we need to account for the separate contributions of nitrogen loads and phosphorus
loads to water quality. To simplify, we first created a nutrient load index (NLI) that combines both
nutrients in proportion to their expected impact on water quality and can be interpreted as the total
nutrient contribution in pounds of nitrogen equivalent.1 The index formula is sum of the nitrogen
load plus 12 times the phosphorus load (NLI = N +12*P). The selection of the number 12 to convert
pounds of phosphorus into equivalent pounds of nitrogen is based on two main considerations. First,
the TN and TP subindices of the WQI share the same functional form but have different coefficient
values, which also differ across the 12 ecoregions in the Chesapeake Bay watershed (U.S. EPA,
2013). For each ecoregion, we estimated the nitrogen and phosphorus concentrations (mg/L) that
would correspond with a subindex value of 75 (a midrange of modeled values for the watershed).
Across ecoregions, the ratio of these values ranged from 4:1 to 39:1 with an average of 12:1. Second,
after generating changes in the WQI for each river segment, we regressed these WQI changes on
changes in nitrogen and phosphorus loads in each segment. In this regression, the ratio of the
coefficients on nitrogen and phosphorus is 11.6, meaning that a 1 pound change in phosphorus will
have on average the same impact on the WQI as roughly a 12 pound change in nitrogen.
To complete step three, we divide the valued water quality benefits in each river segment by the
total nutrient load index delivered to the segment to estimate the value per pound of nitrogen
equivalent in that river segment. These per-pound nitrogen equivalent values are, by necessity, linear
approximations of complex and often nonlinear relationships between nutrient loads, instream
1 To estimate a separate average value for nitrogen and phosphorus loads in a way that avoids double counting, we would
need to divide and allocate the portion of WQI changes in each segment that are attributable to upstream loads of each
pollutant. Using a combined nutrient index provides a simplified alternative for including both pollutants.
3-5
-------
concentrations, and the WQI. Nutrient reductions in river segments selected by the optimization
framework may be much higher or much lower than those observed in the scenarios used to calculate
the per-pound value, and, as such, the per-pound value applied may be too high or too low. To reduce
the chance that an overestimate of per-pound values drives the results in the optimization framework,
we removed outliers by capping all values at the third quartile plus 1.5 times the interquartile range of
estimated values, or $9.53 per pound of nitrogen equivalent. Within a river segment, the value per
pound of nitrogen equivalent can now be used to estimate the value per pound of nitrogen and the
value per pound of phosphorus.
In the fourth step, we estimate the average per-pound value of reduced nutrient loads to surface
water in each river segment. This estimate is based on the value per pound of nitrogen equivalent
within the river segment, plus the value per pound of nitrogen equivalent in all downstream river
segments, after accounting for the attenuation rate between the river segment and all downstream
river segments. This attenuation rate can differ for nitrogen and phosphorus. Using the hypothetical
river segments A and B described above, if a pound of nitrogen reduced to river segment A is valued
at $4/lb and a pound of nitrogen reduced to the downstream river segment B is valued at $2/lb, the
pound reduced in river segment^ is valued at $4 plus $2 times the attenuation rate between^ and5,
or $5.33 ($4 + $2 x 2/3 = $5.33).
The general pattern observed in the geographic distribution of values per pound of nutrient
reductions is that the higher values per pound are for those farther upstream (Figure 3-3). This is due
to two complementary factors. First, upstream river segments have relatively lower flow than
downstream river segments. A pound of nutrients in a river with lower flow will have a larger impact
on the nutrient concentration in the river, which is the basis for the WQI. Second, a pound reduced in
an upstream river segment will improve water quality in all downstream river segments. Therefore,
load reductions in segments located farther from the Bay have the capacity to improve water quality
in a higher number of downstream segments, even when accounting for instream attenuation.
3-6
-------
Legend
Rivers
I Major Basins
^J Minor Basins
$/lb N Reduced
$000-5039
$0 40 - $081
$082-$1 29
j^H $1.30-$1.87
j^H $1,88 -$256
^H $2 57-$3.52
H $3.53 -$4,71
^H $4.72 -$639
^H $6.40 - $9 26
••
| | States
Figure 3-3. Value ($) per Pound of Nitrogen Reduced by River Segment
3.3 Scenario Analysis Results
The next steps in the analysis are to (1) incorporate these water quality co-benefit estimates into
the optimization framework, (2) estimate the least-net-cost solution including these co-benefits, and
(3) compare this solution to other load reduction scenarios that meet the TMDL requirements. The
first load reduction scenario, which we refer to as "TMDL," represents the load reduction practices
specified in the jurisdictions' Phase I WIPs. This scenario does not make use of the optimization
framework.
3-7
-------
The second scenario—the least-cost scenario—applies the optimization framework to achieve the
same overall load reductions as the TMDL scenario. In both cases, the load reduction targets are
defined as the combined targets for significant wastewater and industrial facilities (PS) and
agricultural nonpoint sources (AgNPS) (see Table 2-1). This scenario optimizes over the costs and
delivered load reductions from available nutrient source projects; however, it does not include co-
benefits in the optimization.
The third scenario—the least-net-cost scenario—includes co-benefits in the optimization
framework. These co-benefits include the monetized carbon and hunting benefits estimated in the
(U.S. EPA, 2011) analysis, which only accrue to agricultural acres applying selected BMPs (see
Appendix A). The co-benefits also include the freshwater quality benefits associated with edge-of-
stream load reductions in each nontidal land-river segment. These water quality benefits accrue both
to point sources and to agricultural nonpoint sources. The net-costs for each potential project are
calculated by deducting all of the monetized co-benefits from the costs of the control project. The
optimization solves for the approach that minimizes these net costs while still achieving the total load
reduction targets.
To analyze the differences among these scenarios, we focused on two "basin-state" areas — the
Susquehanna River basin in Pennsylvania (Susquehanna-PA) and the James River basin in Virginia
(James-VA). In all scenarios, agricultural land conversion is restricted to 25 percent. The results are
shown in Table 3-1.
In the Susquehanna-PA, the TMDL scenario results in total annual costs of approximately $280
million to meet their required load reductions for PS and AgNPS. These costs are from the specific
PS and AgNPS source controls included in the Phase I WIPs, which are the basis for this TMDL
scenario. We estimate these nutrient controls generate $137 million in ecosystem services co-
benefits, including $29 million of upstream water quality benefits from PS source controls.
By solving for the least-cost set of source controls, we estimate that these reductions could be
achieved at 49 percent lower costs than specified by the Phase I WIPs; however, the co-benefits from
upstream water quality improvements would also be lower (Table 3-1). In the least-cost scenario,
much of the nutrient reduction occurs in southeastern Pennsylvania (Figures 3-4 and 3-5).
-------
Table 3-1. Costs and Benefits of Significant Wastewater Treatment and Industrial Facilities
(PS) and Agricultural Nonpoint Source (AgNPS) Nutrient Reductions in the
Susquehanna-PA (Million $)
Nutrient Load and
Cost-Benefit Categories
N Load Reductions
(mil. Ibs/yr)
P Load Reductions
(mil. Ibs/yr)
Control Costs
($ mil./yr)
Co-Benefits ($ mil./yr)
Freshwater Benefits
Freshwater Benefits
Carbon Benefits
Hunting Benefits
Net Costs
($ mil./yr)
Source
Total
PS
AgNPS
Total
PS
AgNPS
Total
PS
AgNPS
Total
PS
AgNPS
AgNPS
AgNPS
Total
PS
AgNPS
TMDL
25.4
4.7
20.7
0.7
0.3
0.5
$280.0
$60.0
$220.0
$137.2
$29.0
$86.9
$21.0
$0.3
$142.8
$31.0
$111.8
Scenario
Cost
Minimizing
25.4
3.6
21.8
0.7
0.3
0.5
$142.7
$24.0
$118.7
$103.9
$24.9
$62.1
$16.7
$0.2
$38.8
($0.9)
$39.6
Net Cost
Minimizing
25.4
3.2
22.2
0.7
0.3
0.5
$158.9
$21.1
$137.7
$135.3
$26.2
$82.1
$26.6
$0.4
$23.6
($5.0)
$28.6
.egend
• Rivers
] Major Basins
] M'inor Basins
Stales
D Reductions
4 - 4,224
4,226-13,361
13,362 - 28 235
| 28,236 - 52.931
| 52,932 - 83,363
| 83.384 -129,164
| 129,185- 185,493
| 185,494 - 292,443
| 292,444-430,310
| 430.311 - 590.659
I 590,660 - 1 460,419
Figure 3-4. Least-Cost Scenario Nitrogen Load Reductions in the Susquehanna-PA by River
Segment (Delivered Pounds per Year)
3-9
-------
Legend
| Rhwrs
Major Basins
Minor Basins
States
3hosphorus Reductions
1-139
140-161
462-962
953-1,621
1,622-2.324
2,325-4.032
4.033-6.469
6,470-10,952
10,953-18,175
18,176 - 35,535
35.536 - 63 758
7
Figure 3-5. Least-Cost Scenario Phosphorus Reductions in the Susquehanna-PA by River
Segment (Delivered Pounds per Year)
By including monetized ecosystem service co-benefits and solving for the least-net-cost set of
source controls, the cost of source controls increases by $16 million relative to the least-cost scenario,
and the monetized co-benefits increase by $31 million. Including the co-benefits encourages a shift
towards AgNPS source controls. Interestingly, despite generating fewer nutrient reductions than in
the least-cost solution, PS nutrient reductions generate greater upstream water quality benefits under
the least-net-cost solution. Overall, we observe that more nutrient reductions occur in upstream
tributaries to the main channel of the Susquehanna River under the least-net-cost scenario relative to
the least-cost scenario (Figures 3-6 and 3-7).
In the James River-VA, the TMDL scenario costs are estimated to be $188 million per year.
Compared to the Susquehanna case, a much larger portion of these costs (73%) and load reductions
are associated with PS controls. Despite this difference, about 75 percent of $116 million in annual
freshwater quality co-benefits come from AgNPS. The proximity of many PS sources to the Bay
accounts for their relatively low contribution to freshwater benefits in this scenario.
3-10
-------
Legend
- Rivers
] Major Bi
States
Nitrogen Difference
| -125.077
| -125,076--37,897
-37,896- -15,676
~-16.676--4,487
-4,486 - -1
~ 0
1-1579
1,530-5,966
| 5,987-14,905
| 14:606 - 30,566
30.567-47,998
Note: Positive values reflect river segments where more nutrient reductions occur in the least-net-cost scenario relative to
the least-cost scenario.
Figure 3-6.
Difference in Nitrogen Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the Susquehanna-PA by River Segment (Delivered
Pounds per Year)
Note: Positive values reflect river segments where more nutrient reductions occur in the least-net-cost scenario relative to
the least-cost scenario.
Figure 3-7. Difference in Phosphorus Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the Susquehanna-PA by River Segment (Delivered
Pounds per Year)
3-11
-------
As in the Susquehanna-PA, the results of the least-cost scenario in the James River basin indicate
that there could be substantial (46%) cost savings relative to the scenario based on the Phase I WIPs.
However, in the James River basin, the least-cost scenario also results in relatively larger freshwater
quality co-benefits from AgNPS compared to the TMDL scenario. In the least-cost scenario, nitrogen
reductions occur primarily around the tidal region of the James River (Figure 3-8), while phosphorus
reductions primarily occur farther upstream (Figure 3-9).
Going from the least-cost scenario to least-net-cost scenario costs increase by $17 million, while
benefits increase by $42 million. Therefore, the strategy that targets all benefits instead of just tidal
water quality improvements increases net benefit (reduces net costs) by $25 million. The majority of
the estimated co-benefits in the least-net-cost scenario are from freshwater quality improvements
attributable to agricultural nonpoint source controls (Table 3-2), which were valued using the
methods described in Section 3.2. Figures 3-10 and 3-11 show how the spatial pattern of nitrogen and
phosphorus load reductions change between the least-cost and least-net-cost scenarios. When
freshwater co-benefits are included in the optimization, the nutrient source control tends to shift away
from the main stem and tidal areas of the James-VA to river segments further upstream.
Rivers
States
Major Basins
Minor Basins
Jitrogen Reductions
0-5
599-1.596
1,597 -3,594
3,595 - 3,466
3,467-17,055
17,056-29.909
29,910-50,621
50,622-89,741
99,742-289,113
289,114 - 783,226
783.227 -2,326.491
Figure 3-8. Least-Cost Scenario Nitrogen Reductions in the James-VA by River Segment
(Delivered Pounds per Year)
3-12
-------
Rivers
SJates
Major Basins
Minor Basins
Phosphorus Redugtion
0-181
182-545
546 - 957
958-1,646
1,647-2,61
2,620 - 3,684
3,685 - 5,266
5,267-10,491
10,492-21,493
21,494-41,638
41,639-100,341
Figure 3-9. Least-Cost Scenario Phosphorus Reductions in the James-VA by River Segment
(Delivered Pounds per Year)
3-13
-------
Table 3-2. Costs and Benefits of Significant Wastewater Treatment and Industrial Facilities
and Agricultural Nonpoint Source Nutrient Reductions in the James-VA
(Million $)
Nutrient Load and
Cost-Benefit Categories
N Load Reductions
(mil. Ibs/yr)
P Load Reductions
(mil. Ibs/yr)
Control Costs
($ mil./yr)
Co-Benefits ($ mil./yr)
Freshwater Benefits
Freshwater Benefits
Carbon Benefits
Hunting Benefits
Net Costs
($ mil./yr)
Source
Total
PS
AgNPS
Total
PS
AgNPS
Total
PS
AgNPS
Total
PS
AgNPS
AgNPS
AgNPS
Total
PS
AgNPS
TMDL
11
9.6
1.4
0.9
0.6
0.3
$188.0
$138.0
$50.0
$41.8
$6.3
$24.4
$11.1
$0.1
$146.2
$131.7
$14.5
Scenario
Cost
Minimizing
11
9.2
1.8
0.9
0.4
0.5
$101.2
$84.6
$16.7
$39.0
$3.5
$26.4
$9.0
$0.1
$62.2
$81.1
($18.8)
Net Cost
Minimizing
11
8.8
2.1
1.1
0.3
0.7
$118.5
$75.7
$42.8
$81.1
$3.6
$45.4
$31.8
$0.2
$37.5
$72.1
($34.6)
In summary, through this analysis we demonstrate an approach for including co-benefits from
freshwater quality improvements into the optimization framework. In both of the river basins we
examined and across all scenarios, we estimate values for freshwater quality co-benefits that are
larger than the combined values for the other monetized co-benefits.
Applying these results in the optimization model, we are also able to improve our estimates of the
efficiency gains that can be achieved by accounting for co-benefits. Compared to cost-minimization
results that do not account for co-benefits, we find that including all co-benefits in the optimization
would increase annual nutrient control costs by $16 million in the Susquehanna-PA; however, the co-
benefits would increase by $31 million (net gain of $15 million per year). In the James-VA,
considering monetized co-benefits results in an estimated increase in nutrient control costs of
$17 million but an increase in co-benefits of $42 million (net gain of $25 million per year).
5-14
-------
Rivers
SJates
Major Basins
Minor Basins
Nitrogen Difference
-239,265--130,22
-130,222 --2,413
3,412--1
3
1-759
2,396
2,397 - 4,274
4,275 - 6,628
6,629-9,012
9,013-13,763
13,764-27,055
Note: Positive values reflect river segments where more nutrient reductions occur in the least-net-cost scenario relative to
the least-cost scenario.
Figure 3-10. Difference in Nitrogen Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the James-VA by River Segment (Delivered
Pounds per Year)
In addition, when including these freshwater co-benefits in the optimization analysis, we find that
the least-net-cost scenario tends to shift load reductions (1) towards river segments located farther
from the Bay and (2) from point sources to agricultural NFS (compared to the least-cost scenario).
The spatial shift is primarily due to the higher per-pound values of load reduction co-benefits in more
upstream areas. These higher per-pound values occur because there is less flow in the more upstream
receiving waters (i.e., less dilution of loads) and because they affect more downstream river miles.
The shift in load reduction away from point sources occurs in large part because point sources do not
provide carbon sequestration benefits. The tendency for point sources to be located in the more
downstream areas also contributes to this result.
3-15
-------
Rivers
SJates
Major Basins
Minor Basins
hosphorus Difference
• -35,332
-35,331 --18,100
-18,099--1
3
1 - 104
105-791
792-1,650
1,651 -2,
2,740 - 4,67
4,677 - 7,604
7,605-16944
Note: Positive values reflect river segments where more nutrient reductions occur in the least-net-cost scenario relative to
the least-cost scenario.
Figure 3-11. Difference in Phosphorus Reductions from Least-Net-Cost Source Controls and
Least-Cost Source Controls in the James-VA by River Segment (Delivered
Pounds per Year)
3-16
-------
SECTION 4
ALTERNATIVE MARKET-BASED INCENTIVE DESIGNS FOR
CHESAPEAKE BAY RESTORATION
4.1 Introduction
The Chesapeake Bay TMDL is designed to reduce nutrient loads delivered to the Bay by
roughly 25 percent compared to conditions in 2009 (U.S. EPA, 2010). To achieve this objective,
the TMDL defines maximum allowable annual load allocations (to be achieved by 2025) for
various point and nonpoint source categories. To achieve these load allocations, agricultural
nonpoint sources are expected to contribute over 60 percent of the total annual nutrient load
reductions.
How the TMDL load allocations for agricultural sources will be achieved remains a crucial
question, since they are not federally regulated sources and, therefore, their reductions have
historically been achieved largely through voluntary programs. Several states in the Bay
watershed have developed trading programs to support more cost-effective nutrient reductions
and to provide a potential additional incentive for agricultural sources to generate nutrient
reductions beyond their allocation, or baseline, to sell. In addition to trading programs, more
traditional agricultural cost-share and payment programs are being used to make progress
towards reduction goals. The incentives offered by these public sector programs can be used to
bring farms into compliance with baseline requirements for trading.
An efficient market offers the potential to achieve a more cost-effective implementation of
the TMDL by allocating nutrient reductions to those with the lowest costs. The optimization
framework provides insights into how an efficient nutrient trading market may function in the
face of multiple incentives and how changes in market rules, such as trading baseline
requirements, may influence the load reductions and potential cost savings of an efficient
nutrient trading market.
The remainder of this section is organized as follows:
• Section 4.2 assesses the role nutrient trading can play in encouraging agricultural
sources to meet their TMDL requirements, including interactions of markets with
other incentive programs.
• Section 4.3 analyzes how alternative methodologies for crediting agricultural BMPs
impact the load reductions and potential cost savings from nutrient trading.
4-1
-------
4.2 Nutrient Credit Trading and Other Incentives to Meet Agricultural Baseline
Requirements
In this section, we use the optimization framework to examine the potential implications of
different types and combinations of incentive-based approaches for agricultural nonpoint
sources. In particular, we address the following questions:
• To what extent can the incentives offered by nutrient credit trading encourage farmers
to achieve their TMDL load reduction requirements?
• How are these conclusions affected by including additional public sector payments to
farmers for nutrient controls?
• How does the interaction of trading, baseline requirements, and direct payments
affect the total costs of nutrient control implementation resource costs and public
sector budget requirements for funding nutrient controls?
Although our optimization approach provides a useful and tractable framework for
simulating behaviors and policy outcomes, it also requires simplifying assumptions that are
likely to overstate behavioral responses to a nutrient credit market and public sector incentives.
Consequently, the simulated results are best interpreted as upper bound estimates of policy-
induced changes rather than as predictions of actual outcomes In particular, by using a cost-
minimization approach, we are assuming that farmers and point-source operators are strictly
motivated by a desire to maximize their profits. In practice, their behaviors are more complex
and include several factors that we cannot formally observe or account for in our model. To
address some of the potential transaction and information-related costs, such as the time required
to understand the requirements for nutrient trading, finding a trading partner, and negotiating a
contract, that can interfere with these behaviors, we have augmented the unit costs of nonpoint-
source nutrient controls by a fixed percentage (38%) (McCann and Easter, 2000); however, this
approach only provides a partial adjustment.1 Although in practice there are differences in
trading program requirements across jurisdictions in the watershed (which may lead to
differences in transaction costs), for the purposes of this analysis we assume uniform trading
requirements across the entire watershed.
4.2.1 Nutrient Credit Trading and the Role of Trading Baselines
The purpose of nutrient trading programs is primarily to provide regulated (i.e., "capped")
sources with additional flexibility and the opportunity to incur lower costs for meeting their
regulatory requirements. In particular, trading offers point source dischargers the ability to
1 For example, some administrative costs of trading may be fixed and not vary directly with the number or size of
control practices credited.
4-2
-------
achieve compliance by purchasing load reductions (credits) from other sources, who are either
unregulated or who reduce loads beyond their own regulatory requirements.
Trading programs often include baseline requirements for credit generation by agricultural
nonpoint sources. This trading baseline is defined by a set of preconditions for controlling
nutrient runoff that a farm must achieve before it is eligible to generate credits for additional
(i.e., "beyond baseline") nutrient reductions. Generally speaking, there are two main approaches
for defining the trading baseline requirements:
• Practice-based approach, which typically defines the types or combinations of BMPs
that must be in place; and
• Performance-based approach, which typically defines the load reductions or
maximum allowable level of loads that must be achieved.
Currently, three Bay states—Maryland, Pennsylvania, and Virginia—have established credit
trading programs, and each uses a different approach for defining agricultural trading baselines
(Branosky et al., 2011). Virginia uses a practice-based approach, Maryland uses a performance-
based approach, and Pennsylvania uses a combination.
Despite the differences across state programs, it is important to note that the trading baselines
are all expected to be consistent with the TMDL load allocations. In other words, if all
agricultural nonpoint sources were to meet the trading baseline requirements of their state's
trading programs, then they are expected collectively to be in compliance with the TMDL load
allocations.
The inclusion of baseline requirements in these trading programs has potentially important
implications for farmers' incentives to generate credits and for the agricultural sector as a whole
to meets its TMDL load allocations. In effect, the baseline requirements create a hurdle for
farmers to participate in trading. If trading programs did not include this hurdle, farmers would
have a greater incentive to reduce their loads and generate credits; however, reductions sold as
credits cannot be counted as progress towards meeting agriculture's load allocation. Instead, all
of these load reduction must be transferred to the buyers (e.g., wastewater treatment plants) and
credited towards achieving the buyers' TMDL allocations.
In other words, without the baseline requirement, it would be easier and cheaper for farmers
to generate load reductions that they can sell as credits. At the same time, farmers would have
less of an incentive to generate load reductions that they would not sell and that would count
towards their own TMDL requirements. While credit buyers would benefit by being able to meet
4-3
-------
their load reduction requirements at a lower cost, there would be less incentives for the
agricultural sector to meet its TMDL requirements. Moreover, allowing farmers to sell their
relatively low-cost load reductions to other sectors would make it more costly for the agricultural
sector to achieve its own load allocation.
To examine the implications of baseline requirements, the first question then becomes:
• How large of a hurdle do the baseline requirements present for potential credit
sellers?
For trading to be profitable for a farmer, the revenue from selling credits must exceed the full
costs of reducing nutrient runoff. However, whereas costs must be incurred to both meet the
baseline and generate additional reductions, only the nutrient reductions that surpass the trading
baseline requirements are eligible to generate salable credits. Therefore, the profitability of credit
trading for a farmer will depend in part on the costs she must incur to meet her trading baseline
requirements.
If selling credits can in some cases be profitable despite the trading baseline hurdle, the next
questions are:
• How large will the resulting load reductions from the agricultural sources be?
• How much will the portion of reductions below the trading baseline contribute
towards agriculture's TMDL load reduction requirement?
• How will trading between point and nonpoint sources affect the overall costs of
achieving the TMDL?
• How would results change if public sector incentives for nutrient reductions were also
available?
• How do results change in response to more or less complex program rules?
To examine these questions, we restrict our analysis to the same two main geographic areas
("basin-states") that were are the focus of Section 3. The first is the portion of the Susquehanna-
PA, which includes 76 percent of the Susquehanna River basin area. The second is the James-
VA, which constitutes over 99 percent of the James River basin. Table 2-1 shows the total annual
TMDL nutrient load reduction targets for significant wastewater treatment and industrial
facilities (PS) and agricultural nonpoint sources (AgNPS) in each basin-state. These two basin-
states were selected in part because they provide an interesting contrast. In the Susquehanna-PA,
4-4
-------
the TMDL requires that a majority of reductions come from AgNPS, whereas in the James-VA,
the PS load-reduction requirements are larger.
Although our analysis focuses on these two basin-states, our intention is not to model the
existing trading programs in these two areas. Rather, we examine the implications of an
alternative policy framework applied in both areas. We apply the same optimization framework,
trading assumptions, and policy scenarios in the two areas, and we examine how the results
differ due to their other features and attributes, such as the relative contribution of AgNPS and
PS sources in the basin-state.
To examine the effects of trading baseline requirements, we assume and apply a
performance-based approach in both areas. Using estimates from CBWM, we define baseline
load reduction requirements for each agricultural land use category within each land-river
segment that are consistent with the aggregate TMDL load reduction targets shown in Table 2-1.
In each land-river segment and 2010 land use category (cropland, hay, and pasture), we compute
the baseline requirement as the per-acre reduction in delivered nutrient loads from 2010 to
TMDL conditions in the land use.1 For each eligible BMP or combinations of BMPs available to
be selected in the optimization framework, only the load reductions beyond this baseline are
eligible to generate credits. For areas that are estimated to already be in compliance with TMDL
conditions in 2010, all nutrient reductions for newly implemented BMPs in these units are
assumed to be eligible to generate credits.
4.2.2 Additional Incentives through Public Sector Subsidies or Credit Purchases
In addition to developing and supporting water quality trading programs, the public sector
can also provide direct incentives to agricultural nonpoint sources for reducing nutrient loads. A
number of incentive-based programs already exist for farmers, for example, through USDA
agricultural cost share programs. Expanded and more targeted programs by federal and state
agencies would be needed to make additional progress towards the TMDL goals.
For this analysis, we consider two types of government incentives:
• Direct dollar-per-pound payments to farmers for reductions in nitrogen and
phosphorus loads delivered to the Bay. All reductions in a basin-state receive the
same subsidy.
This calculation includes areas prescribed to be converted to other land uses in the TMDL.
4-5
-------
• Nutrient credit purchases from fanners through participation in an expanded credit
market that uses current (2010) practices as the trading baseline.
In the first case, the government is assumed to set the per-pound subsidy rate and then
compensate farmers for all of their nonpoint source reductions up to their basin-level load
allocation (from current conditions) at that rate. In the second case, the government would be
required to purchase credits from nonpoint sources, but at the same market-based price,
estimated endogenously in the optimization framework, as point source credit buyers. In the
second case, the total number of credits purchased by the government would have to be equal to
the total AgNPS load reduction targets for the basin. In both cases, the public sector is assumed
to pay a constant per-pound amount for load reductions, which does not vary across BMPs.
We also model programs that use a combination of incentive payments and trading. As a way
to deal with the trading baseline hurdle, government policies allow farmers to use payments to
meet the trading baseline but not to generate credits for sale.
4.2.3 Incentive Scenarios
To address the research questions posed above, we apply the optimization framework1 to
analyze the following scenarios:
4.2.3.1 Reference Scenario. PS-PS Trading + Fixed AgNPS Subsidy
The purpose of this scenario is to provide a reference point for examining how AgNPS
trading, baseline requirements and public sector payments affect total nutrient control costs, load
reductions, and public sector spending. In this case, the incentives for AgNPS and PS load
reductions are kept completely separate.
As a reference point for a larger trading market, we assume in this case that, to meet the total
PS load reduction targets, trading is only allowed between point sources. We simulate trading
between point sources with our optimization framework by solving for the least-cost
combination of point source control projects in each basin-state that together achieve the point
source reduction targets shown in Table 2-1. The marginal conditions (i.e., marginal nutrient
reduction costs) at the model solution are interpreted as the market price for credits.
As a reference point for AgNPS payments, we assume that the government uses dollar-per-
pound payments to farmers to meet the AgNPS load reductions targets. We assume that the
dollar-per-pound subsidy rate, which is the same for all AgNPS in the basin-state, is set by the
No agricultural land conversion constraints are applied to these scenarios.
4-6
-------
government to exactly achieve the load reduction target. We simulate this system with our
optimization framework by solving for the least-cost combination of available AgNPS BMPs in
2010 that together meet the relevant total AgNPS reduction target shown in Table 2-1. The
marginal conditions (i.e., marginal nutrient reduction costs) at the model solution are interpreted
as the subsidy rate required to exactly achieve the target.
4.2.3.2 Policy Scenario 1. PS-AgNPS Trading + No Subsidy
In this scenario, trading is allowed between PS and AgNPS to meet the PS load reduction
target. However, to participate in the trading market, AgNPS must meet their performance-based
load reduction baseline requirements—that is, they can only sell credits for reductions beyond
this baseline. No new subsidy payments are provided by the public sector to AgNPS.
For this scenario, we simulate trading between point and nonpoint sources by solving for the
least-cost combination of PS and AgNPS source control projects in each basin-state that together
achieve the point source reduction targets shown in Table 2-1. To account for baseline
requirements for agricultural nonpoint source, the cost minimization model only credits BMP
alternatives for load reductions beyond their baseline, but it includes all of the costs of
implementing the BMP alternative.
4.2.3.3 Policy Scenario 2. PS-AgNPS Trading + Subsidy ($2/lb N to $10/lb N)
This scenario is equivalent to Scenario 1 except that the public sector also provides fixed per-
pound subsidy payments to farmers for load reductions that are less than or equal to the baseline
required reductions. Three sub-scenarios are investigated, with payments of $2, $5, and $10 per
pound of nitrogen. There is no "double-dipping" of payments because only the government
purchases reductions to meet the baseline and only point sources purchase reductions (credits)
beyond the baseline.
We simulate this scenario using the same approach as for Scenario 1; however, we include
the subsidies to offset some of the costs of implementing the BMP alternative.
4.2.3.4 Policy Scenario 3. PS & Public Demand for AgNPS Credits
We include this scenario to some extent as an alternative point of reference. Like the
Reference Scenario, it includes incentives to ensure that both the PS and AgNPS targets are fully
achieved; however, rather than separating the sources into two separate incentive systems, they
are combined into one trading system. Under this scenario, point sources and the public sector
both participate in the nutrient market as demanders of AgNPS credits. The government's
objective is to achieve the AgNPS load allocation by purchasing credits from farmers. However,
4-7
-------
rather than setting a price for load reductions, as in the Reference Scenario and Scenario 2, the
public sector must compete with point sources for AgNPS credits and pay the market-determined
price. In this case, there is no baseline requirement for AgNPS credits because the public sector's
involvement in the credit market ensures that the AgNPS target is met. In other words, for
farmers, all load reductions from current conditions are eligible for generating credits. Requiring
the government to pay market prices is consistent with using reverse auctions in the sense that it
helps to ensure that credits purchased are cost-effective. However, market prices could be more
than traditional payments which often have a cost-share requirement.
We simulate this scenario by solving for the least-cost combination of point and nonpoint
source control projects in each basin-state that together achieve the combined PS and AgNPS
reduction targets shown in Table 2-1. Once again, the marginal conditions (i.e., marginal nutrient
reduction costs) at the model solution are interpreted as the market price for credits, which
applied to both PS and public sector credit buyers.
4.2.4 Results
Model simulation results for the three scenarios are shown in Table 4-1. To simulate the
Reference Scenario, we ran the optimization model separately for the AgNPS and PS load
reduction targets. For the PS-PS trading, we solve for the least-cost combination of installed PS
treatment technologies required to meet the PS load reduction targets (4.7 million pounds of
nitrogen and 0.3 million pounds of phosphorus in Susquehanna-PA). The resulting annual
control cost estimate for point sources is $40 million.
For AgNPS, we solve for the least-cost combination of new agricultural BMPs that are
needed to meet the AgNPS target, and we use the marginal conditions of this solution to provide
estimates of the subsidy rates (market prices) for nitrogen and phosphorus reductions. Under this
solution, the annual cost to farmers for implementing these BMPs is $92.3 million. The public
sector purchases all of the required AgNPS load reductions—20.7 million pounds of nitrogen
and 0.5 million pounds of phosphorus—for $208 million per year. Although not shown in Table
4-1, the model-estimated prices for nitrogen and phosphorus are $6.03/lb and $169.27/lb
respectively. Under this scenario, farmers earn annual profits of $115.6 million for reducing
nutrient loads1.
1 Farm profits are estimated by subtracting the cost of the nutrient control project from the revenue received for
nutrient reductions based on the estimated nutrient market prices.
4-8
-------
Table 4-1. Policy Scenario Results for the Susquehanna River Basin in Pennsylvania
Load Reduction (mil. Ibs/yr)
Total Control
Cost (mil. $)
AgNPS PS
REFERENCE
PS-PS Trading + Fixed AgNPS Subsidy 92.33 40.06
SCENARIO 1
PS-AgNPS Trading + No Subsidy 12.61 11.09
SCENARIO 2
PS-AgNPS Trading + $2/lb N Subsidy 20.00 8.18
PS-AgNPS Trading + $5/lb N Subsidy 48.24 7.50
PS-AgNPS Trading + $ 10/lb N Subsidy 62.96 6.42
SCENARIO 3
PS & Public Demand for AgNPS Credits 106.78 15.92
Nitrogen
AgNPS
Below
Trading
Baseline"
20.67
2.29
5.62
14.30
16.33
20.67
AgNPS
Sold to
PS PS
0.00 4.73
2.27 2.45
2.68 2.05
2.76 1.96
2.99 1.73
1.95 2.77
Phosphorus
AgNPS
Below
Trading
Baseline"
0.49
0.02
0.08
0.14
0.16
0.49
AgNPS
Sold to
PS PS
0.00 0.26
0.04 0.21
0.05 0.21
0.06 0.20
0.06 0.20
0.01 0.25
Revenue to AgNPS
(mil. $)
Subsidy Credits
207.95 0.00
0.00 18.37
11.24 16.58
71.49 13.64
163.30 13.04
235.43 15.93
1 AgNPS TMDL load reduction target and trading baseline is 20.67 million pounds of nitrogen and 0.49 million pounds of phosphorus. This column shows the
portion of the target that is achieved under each scenario (the remainder is unmet).
-------
To simulate Scenario 1, we solve for the least-cost combination of agricultural BMPs and PS
treatment technologies needed to meet the PS load reduction target; however, the AgNPS
controls only receive credit for nutrient reductions beyond their trading baseline. By applying
BMPs that cost $12.6 million per year, the AgNPS generate 2.3 million pounds of nitrogen
credits and 0.04 million pounds of phosphorus credits, which they sell to PS for $18.4 million.
Because these reductions are sold to PS, they do not contribute to the AgNPS load reduction
targets under the TMDL. The PS sector incurs $11 million in control costs and $18.4 million in
credit purchase costs each year, which translates to $10.6 million in savings for the PS sector
compared to the Reference Scenario.
In addition to the load reductions sold as credits to the PS sector, the agricultural BMPs in
Scenario 1 generate another 2.3 million pounds of nitrogen and 0.02 million pounds of
phosphorus reductions. These reductions cannot be sold to PS, but they do contribute to the
TMDL load reduction target for AgNPS. Unfortunately, these contributions are a relatively small
percentage of the reduction needed to meet nitrogen and phosphorus reduction targets—11
percent and 4 percent.
For Scenario 2, we expand Scenario 1 to include different fixed payment subsidies for
nitrogen reductions. While such a policy is not currently in place in the Chesapeake Bay, this is
analogous in some respects to Maryland's Chesapeake Bay Restoration Fee (also known as the
"flush tax"), which is partly used to fund a payment per acre of implemented cover crops. If the
state were to estimate the pounds of nutrients reduced from the cover crop implementation, the
farmer could be compensated for the estimated performance of the practice instead. These
payments to AgNPS only apply to nitrogen reductions below their trading baseline, and they are
treated in the optimization model as reductions in BMP costs. As expected, these subsidies lead
to larger AgNPS load reductions, which also increase as the subsidy is raised from $2/lb to
$10/lb of nitrogen. However, even with a $10/lb subsidy and an annual cost to the public sector
of $163 million, the portion of these load reductions that can be applied to baseline requirements
16.3 million pounds of nitrogen and 0.16 million pounds of phosphorus, which only achieves 79
percent of the nitrogen target and 33 percent of the phosphorus target. Meanwhile, the PS
purchasers also benefit indirectly from the subsidies through lower credit prices of $3.80/lb N
and $27.42/lb P compared to $17.95/lb N and $180.75/lb P under Scenario 1.
For Scenario 3, we simulate a market that combines public sector and PS demand for credits.
We remove the trading baseline requirement for AgNPS, and we solve for the least-cost
combination of agricultural BMPs and PS treatment technologies that meets the combined load
reduction targets for AgNPS and PS. To meet the combined targets, this scenario results in the
4-10
-------
highest annual public sector spending for load reductions ($235.4 million) and highest annual
cost for AgNPS controls ($106.8 million). However, compared to the Reference Scenario, which
also meets the combined targets, it results in lower total control costs. The combined control cost
for AgNPS and PS sources is $122.7 million for Scenario 3, compared to $132.4 million for the
Reference Scenario. Through nutrient credit trading, the benefits of this $9.7 million reduction in
total costs are spread between AgNPS and PS, with most (85%) going to PS.
Table 4-2 reports results of these same scenarios for the James River basin in Virginia.
Overall, the results are qualitatively very similar to those from the Susquehanna River Basin in
Pennsylvania. Without a subsidy (Scenario 1), trading between PS and AgNPS provides an
incentive for AgNPS to partially achieve their TMDL reduction targets. When subsidies are
added (Scenario 2), the AgNPS get closer but not completely to their targets, even with a $10 per
pound subsidy. When subsidies are replaced with a combined market for credits (Scenario 3), it
results in the higher spending by the public sector, but more overall savings in control costs
compared to the Reference Scenario.
One of the main differences between the results for the two basins is that nutrient trading in
the James-VA basin gets AgNPS closer to their load reduction targets than in the Susquehanna-
PA basin. For example, even without a subsidy (Scenario 1), AgNPS achieve 35 percent of their
nitrogen target and 41 percent of their phosphorus target in the James, compared to 11 percent
and 4 percent, respectively, in the Susquehanna. One reason for this difference is the higher
percentage of the total load reduction placed on PS in the James, which translates to a higher
demand for credits from AgNPS compared to the AgNPS required reductions.
4.3 Impacts of Alternative Methodologies for Estimating Nutrient Reductions from
Nonpoint Source Controls
A concern with trading is that nonpoint-source control practices are less reliable and
measurable than point-source controls. Thus, the development of a nutrient trading market
between point and nonpoint sources requires methods for measuring or estimating nutrient
reductions from nonpoint sources. Unfortunately, monitoring the actual nutrient reductions from
the implementation of a specific BMP through in field and/or in stream measurements is usually
costly. As a result, methods to calculate nutrient credits from nonpoint sources generally rely on
estimated nutrient reductions associated with specific BMP implementation. In this section, we
use the optimization framework to analyze how alternative methodologies influence both the
estimated environmental outcome and the potential cost savings from nutrient trading with
nonpoint sources.
4-11
-------
Table 4-2. Policy Scenario Results for the James River Basin in Virginia
Total Control
Cost (mil. $)
AgNPS PS
REFERENCE
PS-PS Trading + Fixed AgNPS Subsidy 7.80 109. 17
SCENARIO 1
PS-AgNPS Trading + No Subsidy 10.42 75.70
SCENARIO 2
PS-AgNPS Trading + $2/lb N Subsidy 10.54 75.70
PS-AgNPS Trading + $5/lb N Subsidy 15.45 71.79
PS-AgNPS Trading + $10/lb N Subsidy 16. 19 71.79
SCENARIO 3
PS & Public Demand for AgNPS Credits 20.28 75.57
Load Reduction (mil. Ibs/yr)
Nitrogen
AgNPS
Below
Trading
Baseline"
1.39
0.49
0.58
0.87
0.98
1.39
AgNPS
Sold to
PS PS
0.00 9.57
0.75 8.82
0.75 8.82
0.94 8.62
0.94 8.62
0.75 8.82
Phosphorus
AgNPS
Below
Trading
Baseline"
0.34
0.14
0.17
0.23
0.26
0.34
AgNPS
Sold to
PS PS
0.00 0.60
0.24 0.36
0.24 0.36
0.24 0.36
0.24 0.36
0.25 0.34
Revenue to
AgNPS (mil. $)
Subsidy Credits
14.21 0.00
0.00 18.10
1.16 17.03
4.35 23.50
9.76 22.22
33.57 18.50
'AgNPS TMDL load reduction target and trading baseline is 1.39 million pounds of nitrogen and 0.3 4 million pounds of phosphorus. This column shows the
portion of the target that is achieved under each scenario (the remainder is unmet).
-------
4.3.1 Estimating Nutrient Reductions from Nonpoint Source Controls
Nutrient reductions from implementing a BMP or set of BMPs on a farm depend on several
site-specific factors, such as the type of soil present, the slope of the land, the timing and
magnitude of rain events, and existing management practices (Sharpley et al, 2009; Simpson and
Weammart, 2009). Methodologies used to estimate nutrient reductions for a nutrient trading
market could take all relevant site-specific factors into consideration, allowing nutrient credits
generated by a BMP to vary farm by farm.1 Or, methodologies could rely on an average expected
performance across a wider region, where a BMP would generate the same number of nutrient
credits at every farm.
To estimate nutrient reductions towards meeting the TMDL, CBWM relies on average
expected performance values based on numbers for correctly implemented BMPs in each land-
river segment. Nutrient trading programs within the Chesapeake Bay watershed have adopted
methods designed to be consistent with CBWM; however, they vary in the level of detail
included to calculate the number of credits generated by a BMP. Crediting methodologies
developed for Maryland, Pennsylvania, and West Virginia rely on BMP performance efficiencies
from CBWM and also include site-specific information such as fertilizer and manure application
data (Branosky et al., 2011). In contrast, Virginia has adopted a uniform method of crediting
nutrient reductions. For every eligible BMP, such as early cover crops, the number of nutrient
credits it generates within a river basin only depends on whether it is installed east or west of I-
95 (Table 4-3), regardless of other local conditions that could be considered (VDEQ, 2008).
The methods used to calculate how many credits a BMP generates can influence both the
environmental outcome and the potential cost savings from nutrient trading (Table 4-4). For
example, if uniform regional crediting, as opposed to site-specific crediting, were to result in
placing the majority of BMPs on areas that generate below-average nutrient reductions, then
uniform crediting could result in a failure to achieve the required nutrient reductions.
Alternatively, if it would result in BMPs being distributed evenly across the landscape, such that
their average performance is equal to the uniform credited value, then nutrient targets would be
met. The major advantage of using regional averages is that simpler calculations would be
expected to lower costs of market transactions compared to systems that required lots of farm-
specific information.
Selective monitoring of farms would also help validate BMP performance more generally.
4-13
-------
Table 4-3. Nutrient Credits Generated by Best Management Practices in the James River
Basin in Virginia
Best Management Practice
Early Planted Cover Crops
Enhanced Nutrient Management
Continuous No -Till
Tree Planting on Cropland
Cropland Retirement
West
Nitrogen
Credited
(Ibs/acre)
0.54
1.75
1.05
5.48
3.44
of 1-95
Phosphorus
Credited
(Ibs/acre)
0.00
0.00
0.49
1.22
0.33
East
Nitrogen
Credited
(Ibs/acre)
0.91
3.70
1.13
9.34
3.08
of 1-95
Phosphorus
Credited
(Ibs/acre)
0.00
0.00
0.19
0.93
0.00
Source: VDEQ, 2008
Table 4-4. Potential Impacts of Uniform Crediting on Environmental Outcomes and Cost
of Source Controls under Nutrient Trading
BMP Placement under Uniform
Crediting
Environmental Outcome
Cost of Source Controls
Areas with Above Average Nutrient
Reductions
Areas with Below Average Nutrient
Reductions
Above Nutrient Target
Below Nutrient Target
Increase Relative to Site-Specific
Crediting
Decrease or Increase Relative to Site-
Specific Crediting
4.3.2 Alternative Crediting Scenarios
To estimate the impacts of alternative methodologies of estimating nutrient reductions, we
use the optimization framework to analyze the following scenarios:
4.3.2.1 Scenario A. PS-AgNPS Trading with Site-Specific Crediting
In this scenario, nutrient reductions from PS upgrades and AgNPS source controls beyond
baseline (the same AgNPS inputs described in Section 4.2) can contribute to meeting the PS
nutrient reduction requirement. We apply the optimization framework to identify the least-cost
solution for meeting the PS load reduction targets in each basin-state.1 To represent site-specific
crediting for agricultural BMPs, we use the nutrient load data and BMP performance
assumptions from CBWM. In other words, we use the same approach described in previous
1 In addition, agricultural land conversion is restricted to 25 percent to represent policies used to prevent loss of
agricultural lands. (Wainger et al, 2013).
4-14
-------
sections of this report to estimate load reductions for AgNPS projects. CBWM uses a modeling
approach to calculate load reductions; therefore, it involves simplifications and does not provide
the same level of site-specific information that might be achieved with direct monitoring.
Nevertheless, it does account for differing conditions across land use categories and land-river
segments, which contribute to variation in BMP load reductions.
4.3.2.2 Scenario B. PS-AgNPS Trading with Uniform Crediting
For this scenario, we again use the optimization framework to identify the least-cost solution
for meeting the PS load reduction targets in each basin-state. In contrast to Scenario A, nutrient
reductions for AgNPS BMPs are estimated to be uniform within a basin-state.1 This uniform
value is set at the mean value from CBWM for each basin-state. For example, each acre of
enhanced nutrient management is assigned the same nutrient reductions throughout the
Susquehanna River Basin in Pennsylvania.
4.3.3 Results
Applying the optimization framework across the 14 basin-states in the watershed, overall we
find that uniform crediting results in fewer AgNPS credits being generated compared to site-
specific crediting. As shown in Table 4-5, the uniform crediting scenario results in 3.1 million
pounds of nitrogen credits and 295 thousand pounds of phosphorus credits being sold to PS
buyers per year, compared to 4.1 million and 227 thousand pounds, respectively, under site-
specific crediting.
Table 4-5 also shows how, under a uniform crediting approach, credited and "actual" load
reductions may differ from each other. In this application, "actual" load reductions are estimated
using the CBWM method (i.e., the same approach used for site-specific crediting). With this
approach, we find that, across all basin-states, actual reductions are 8 percent lower than credited
reductions for nitrogen and 23 percent lower for phosphorus. However, these results vary across
basin-states. In several cases, especially for phosphorus, we find that actual reductions are higher
than credited, which means that in these areas BMPs are placed such that their average load
reductions are higher than the uniform rate for the basin-state.
Overall, we find that uniform crediting results in lower actual nutrient load reductions from
AgNPS than site-specific crediting (by 30 percent for nitrogen and 34 percent for phosphorus)
and higher costs for achieving these reductions (by 8%). This result occurs because site-specific
crediting encourages BMP placement for nutrient trading in areas where they produce relatively
1 The one exception to this is for practices, such as forest buffers, where effectiveness varies by hydrogeomorphic
region. For these BMPs, the effectiveness varied by hydrogeomorphic region.
4-15
-------
high nutrient reductions and are therefore more cost-effective (assuming costs are not positively
correlated with removal efficiencies). In contrast, uniform crediting does not provide this type of
incentive. It should also be noted that, although uniform crediting results in higher costs in every
basin-state, in a few cases it also results in greater load reductions; however, even in these cases
the overall cost-effectiveness (load reduction per dollar) is higher under site-specific crediting.
4-16
-------
Table 4-5. Actual" vs. Credited AgNPS Load Reductions and Control Costs under Alternative Crediting Approaches
Major Basin Jurisdiction
Eastern Shore Delaware
Eastern Shore Maryland
Eastern Shore Virginia
James Virginia
Patuxent Maryland
Potomac Maryland
Potomac Pennsylvania
Potomac Virginia
Potomac West Virginia
Rappahannock Virginia
Susquehanna New York
Susquehanna Pennsylvania
Western Shore Maryland
York Virginia
Total
Nitrogen (1,000 pounds)
Uniform Crediting
Credited Actual
Load Load
Reductions Reductions
16 17
74 87
15 14
379 359
1 1
33 23
29 23
571 467
110 77
55 47
203 204
1,514 1,456
1 0
132 100
3,132 2,876
Site-Specific
Crediting
Credited and
Actual
Reductions
16
271
15
618
0
33
56
710
110
55
263
1,815
1
132
4,094
Phosphorus (1,000 pounds)
Uniform Crediting
Credited Actual
Load Load
Reductions Reductions
1 1
2 3
2 2
126 69
0 0
5 6
2 1
82 67
11 9
10 13
20 21
22 25
0 0
13 11
295 227
Site-Specific
Crediting
Credited and
Actual
Reductions
1
15
2
164
0
6
2
64
11
12
20
32
0
12
342
Site-
Uniform Specific
Crediting Crediting
Scenario Scenario
Cost Cost
(million $) (million $)
$0.1 $0.1
$5.7 $2.8
$0.3 $0.3
$93.9 $90.1
$0.3 $0.3
$2.7 $2.1
$1.0 $0.8
$6.0 $4.2
$1.0 $0.9
$0.6 $0.4
$7.7 $6.7
$31.1 $27.3
$40.1 $40.0
$3.1 $2.9
$193.6 $179.0
1 "Actual" load reductions are modeled (based on the CBWM methods) rather than monitored values.
-------
-------
SECTION 5
CONCLUSIONS
In this report we adapt and apply an economic optimization framework to analyze strategies
for achieving the goals of the Chesapeake Bay TMDL. With this framework, we conduct two
main analyses.
The purpose of the first analysis is to expand the existing framework, which includes costs
and selected co-benefits (i.e., carbon sequestration and hunting recreation benefits) of nutrient
control practices, to include monetized benefit estimates for improvements in freshwater quality
in the watershed. Using a benefit transfer approach, we first develop estimates of the average
(per-pound) value of reducing edge-of-stream nitrogen and phosphorus loads in each nontidal
river segment of the watershed. These values represent approximations of households' average
willingness to pay for the resulting freshwater quality improvements in their own state. We find
that these per-pound values are generally higher in the more upstream sections of the watershed,
which reflect the relatively low water flow in these segment and the relatively high number of
downstream segments affected.
With this expanded framework, we then analyze and compare three scenarios for achieving
the TMDL load reduction scenarios in selected basins: (1) a TMDL scenario based on the states'
WIPSs (i.e., no optimization), (2) a least-cost optimization scenario, and (2) a least-net-cost
optimization scenario. In all cases, we find that the benefits from improving freshwater quality in
the watershed (separate from the water quality benefits for the Bay itself) are greater than the
carbon and hunting co-benefits combined. Comparing the least-cost and least-net-cost scenarios,
we also find the latter scenario results in greater nutrient control efforts in the more upstream
portions of the watershed, which is consistent with the higher per-pound values for freshwater
benefits.
The results from this first analysis indicate that, although the purpose of the TMDL is to
improve water quality in the Bay estuary, many measures to achieve this goal will also provide
significant upstream water quality benefits. Therefore, providing additional incentives for
delivered load reductions that originate farther upstream may improve the overall
efficiency (in a net-cost sense) of meeting TMDL goals. However, it must be emphasized that
the per-pound value estimates for upstream load reductions are based on a linear approximation
derived from a single watershed-wide load reduction scenario. Additional sensitivity analyses,
including the use of alternative load reduction scenarios to generate the per-pound values, will be
needed to determine the robustness of these estimates. In addition, this analysis does not include
5-1
-------
nutrient controls from other sectors, in particular urban stormwater BMPs. Although those
BMPs tend to be less cost-effective and agricultural BMPs, they also offer distinct co-benefits
(e.g., flood control and air quality improvements). Future analyses could examine how including
these sources and BMPs alters our findings.
For the second analysis, we use the optimization framework to analyze the implications of
different nutrient trading and incentive-based approaches. In particular, we investigate (1) how
nutrient trading may interact with other incentives for agricultural nutrient reductions and
(2) how simplified crediting of nutrient reductions influences the control costs, load reductions,
and participation in a nutrient trading market.
We find that, although nutrient trading can act as an incentive for some agricultural entities to
adopt nutrient controls and meet their load allocations (i.e., trading baseline) under the TMDL,
these incentives would only support a portion of the required agricultural load reductions. In
other words, these results indicate that nutrient trading is not a particularly effective
mechanism for encouraging the agricultural sector to meet its TMDL goals. In the
Susquehanna-PA, trading would only incentivize agricultural NPS to achieve 11 percent or less
of required reductions. In contrast, in the James-VA, we estimate nutrient trading would be more
effective, with 35 percent of the nitrogen reduction and 41 percent of the phosphorus reduction
achieved through nutrient trading. One reason for this difference is the higher percentage of the
total load reduction placed on PS in the James. This difference translates to a relatively high
demand for credits from AgNPS.
Given this gap between achieved and required load reductions with trading alone, we
examine how additional incentives from public subsidies could alter these outcomes. We find
that per-pound subsidies could help to narrow this gap, but at a relatively high budgetary cost for
the public sector. A "combined market," where the public sector competes with PS for credits,
would be the most economically efficient approach for achieving both PS and AgNPS targets,
but the budgetary costs of this approach are likely to be prohibitive.
Finally, we explore how simplified crediting approaches for nutrient reductions would affect
trading outcomes. In the case examined, we estimate that a simplified approach results in higher
costs (by 8 percent across the watershed) of achieving significant wastewater and industrial
discharges nutrient reductions. Unlike the site-specific approach the simplified uniform crediting
approach does not encourage placement of nutrient controls where they would be most cost-
effective for reducing nutrients. In addition, simplified crediting of nutrient trading is estimated
to result in failure to meet the load reduction requirements in 11 of the 14 basin-state
5-2
-------
combinations in the Chesapeake Bay watershed. The shortfall occurs because, in these cases, the
simplified approach results in certain agricultural areas receiving more credit for nutrient
reductions than are actually achieved.
While these findings provide potentially important insights for designing and evaluating
incentive-based approaches for achieving the TMDL, it is important to interpret them with
certain caveats in mind. Most importantly, even with adjustments for transaction costs, the
optimization framework offers a somewhat idealized representation of credit markets. Due to
uncertainties and real world market frictions, in practice credit buyers and sellers are unlikely to
take advantage of all the cost saving opportunities available. Therefore, the cost estimates
generated with the optimization framework should be interpreted as lower bound values. The
framework also provides a simplified representation of the load reduction options in the
watershed. For example, it does not include all of the possible agricultural and urban stormwater
BMPs that can be used to achieve the TMDL goals. Future analyses would benefit from an
expanded framework that includes a larger set of BMP options.
5-3
-------
-------
APPENDIX A.
CO-BENEFITS FROM AGRICULTURAL BEST MANAGEMENT PRACTICES
This appendix describes the methods used to estimate specific ancillary benefits resulting
from implementing agricultural best management practices (BMPs) to meet the Total Maximum
Daily Load (TMDL). The methods described in this section are based, to a large extent, on those
developed and applied by EPA's Office of Research and Development to quantify the ecosystem
services from these practices (U.S. EPA, 2011).
A.I Carbon Sequestration and Changes in Greenhouse Gas Emissions
To predict the carbon-related benefits of agricultural BMPs, it is necessary to calculate
both the change in the total amount of greenhouse gases (GHGs) emitted and the change in
amount of carbon sequestered.
A. 1.1 Estimation of GHG Emissions
We identified three main types of GHGs whose emissions can be estimated for selected
land use/land cover categories. Expressed in the common unit of carbon dioxide (CO2)
equivalents using the most recent Intergovernmental Panel on Climate Change (IPCC) (2013)
estimates of global warming potential, they are the following:
CO2 = 1 CO2 equivalent
CH4 = 28 CO2 equivalents (global warming potential for 100 years)
N2O = 265 CO2 equivalents (global warming potential for 100 years).
CO2 emissions occur as a result of decomposition and aerobic degradation and can be
temporarily accelerated following conversion of lands to wetlands. Methane (CH/t), a product of
anaerobic degradation, also commonly occurs in wetlands because of the low oxygen availability
with a high water table. Nitrous oxide (TSbO) emissions are most common with croplands, with
higher emissions associated with crops such as corn that require nitrogen fertilization, unlike
nitrogen-fixing crops such as soybeans.
Two main reference sources were used to identify GHG emission rates for this exercise.
First, the Forest and Agricultural Sector Optimization Model (FASOM) (Adams et al., 1996) was
used for crop and pasture N2O emission rates. FASOM was initially developed to evaluate
welfare and market impacts of alternative policies for sequestering carbon in trees, but also has
been applied to a wider range of forest and agricultural sector policy scenarios
(http://www.treesearch.fs.fed.us/pubs/viewpub.jsp?index=2876). N2O emission rates were
identified in FASOM's March 2010 version.
Second, the IPCC 2006 IPCC Guidelines for National Greenhouse Gas Inventories,
Volume 4—Agriculture, Forestry and Other Land Use (henceforth referred to as the IPCC 2006
A-l
-------
Guidelines) was referenced to identify CO2 and CH4 emission rates, where available, for
wetlands.
Table A-l lists the GHG emission rates used in this analysis.
Table A-l. Assumed GHG Emission Factors for Selected Land Uses
Land Use
Cropland
Pasture
Wetland
Forest1
C02
0 kg C/ha-yr3
Of
13.55 Ib/ac-day
(15.2 kg CO2/ha-dayd)
0
N20
b
g
0
0
CH4
oc
oh
0.54 Ib/ac-day
(0.061 kg CHVha-day
0
e)
a Assumes no crop burning (negligible; EPA GHG Inventory report [U.S. EPA, 2010] assumes only 3% of crops in
the United States are burned).
b Crop-specific N2O emission factors reported in U.S. EPA(2011).
0 Assumes no rice grown (only crop that emits CH4) and no crop burning (negligible; EPA GHG Inventory report
[U.S. EPA, 2010] assumes only 3% of crops in the United States are burned).
d Source: IPCC, 2006 v.4 , App 2, Table 2A.2.
e Source: IPCC, 2006 v.4, App 3, Table 3A.2.
f Source: IPCC, 2006 v.4.
g Crop-specific N2O emission factors reported in U.S. EPA(2011).
h Assumes zero CH4 emissions for this analysis (pasture emissions from enteric fermentation depend on herd size).
1 Assumes no thinning or harvesting for this analysis.
A. 1.1.1 Cropland and Pastur'eland Emissions
To estimate crop-based GHG emissions, county-based predominant crop types can be
identified using U.S. Department of Agriculture (USDA) National Agriculture Statistics Service
(NASS) data. These data can be combined with FASOM's N2O emission rates, which are
reported by crop and by state. These rates range from 0.002 to 0.009 ton N2O/ha-year.
The emission sources included in the N2O emission estimates include:
• nitrogen fertilizer application practices under managed soil categories under
AgSoilMgmt,
• emissions from nitrogen-fixing crops,
• emissions from crop residue retention,
• indirect soils volatilization, and
• indirect soils leaching runoff.
FASOM also includes N2O emission rates for pasture, which were used in this analysis.
A-2
-------
A. 1.1.2 Wetland Emissions
For this analysis, we used the IPCC emission estimates for flooded lands (IPCC, 2006) to
estimate CH4, CO2, and N2O emissions associated with wetland restoration or construction in the
Chesapeake Bay,1 as described below.
• Methane—A CH4 emission rate of 0.54 Ib/ac-day (0.061 kg CHVha-day) represents
the median diffusive emission rate of CH4 for flooded land located in a cold
temperate, moist climate (IPCC, 2006, Appendix 3). When using this emission rate,
expressed in kilograms of CFLt per hectare per day, annual emissions should exclude
days with ice cover, because CH4 emissions are reduced dramatically when wetland
waters are frozen. The number of ice-free days can be determined by the mean
number of days with minimum temperatures 32° F or less for cities within the
Chesapeake Bay watershed. This value is 257 days based on 37 to 73 years of data
from seven cities (NOAA, 2010).
• Carbon Dioxide—A CO2 emission rate of 13.55 Ib/ac-day (15.2 kg CCh/ha-day) was
selected by the IPCC to represent the median diffusive emission rate of CO2 for
flooded land located in a cold temperate, moist climate (IPCC, 2006, Appendix 2).
This daily emission rate should only be applied for the first 10 years after flooding
(i.e., the first 10 years following conversion to wetland), and the annual emission
estimate should exclude days of the year with ice cover.
A. 1.1.3 Changes in GHG Emissions from Agricultural BMPs
For agricultural BMPs involving land conversion away from cropland or pastureland, we
assumed that GHG emissions are reduced according to the rates reported in Table A-l. For the
wetland conversion BMP, we also estimated an offsetting increase in GHG emissions, based on
the wetland emission rates in Table A-l.
In addition, we estimated a reduction in GHG emission for "working land" BMPs that
reduce fertilizer application. For the decision agriculture and enhanced nutrient management
BMP, the Chesapeake Bay Watershed Model (CBWM) assumed a reduction in fertilizer
application of 7.5 percent and 15 percent, respectively. For these BMPs, we therefore also
assumed a reduction in N2O emissions by 7.5 percent and 15 percent, respectively.
A. 1.2 Estimation of Carbon Sequestration
For this analysis, we also estimated carbon sequestration for the BMPs involving land
conversion to forests, wetlands, or grasslands. Conversion to forests will result in accumulation
or sequestration of carbon in aboveground and belowground vegetation, as well as soil pools
1 The IPCC chose not to recommend emission rates specifically for wetlands because of a lack of wetlands research
at the time of publication. The IPCC's 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Volume
4—Agriculture, Forestry and Other Land Use reports "Some uses of wetlands are not covered in the report
because adequate methodologies are not available. These include 'rewetting of previously drained wetlands' and
'wetland restoration'" (Section 7.3.2.1). However, wetlands can be significant sources of GHGs; therefore, we
included them in the analysis as described in this section.
A-3
-------
during stand development. Conversion to wetlands will sequester carbon in vegetation and soils,
with a large amount of carbon accumulating in the soils because of higher water tables and
anoxic conditions, which slows decomposition. Conversion to grasslands also will result in
carbon sequestration, mostly below ground.
The following five-step process was used to estimate sequestered carbon:
• Step 1: Determine predominant forest type by ecoregion by county. County-level
forest cover within the Chesapeake Bay watershed was calculated with the U.S.
Forest Service (USFS) National Forest Type Dataset and Omernik ecoregions
(Omernik, 1987). A total of eight Omernik ecoregions overlap the CBW.
• Step 2: Select tree species. For crop or pastureland converted to forest, we assumed
that the land would be planted with the main tree species found in the dominant forest
type of each ecoregion. Conversion to wetlands was assumed to involve planting of
the wetland area with a bald cypress/water tupelo forest type (Neely, 2008). For land
retirement, land would naturally regenerate to an even mixture of all forest types
found within the ecoregion.
• Step 3: Obtain carbon sequestration rates by tree species and ecoregion. The
National Council for Air and Stream Improvement/USFS Carbon On-Line Estimate
(COLE) was used to calculate total carbon stocks. Estimates were made for the forest
types assigned in Step 2. The total no-soil carbon storage values reported for 5- to 10-
year increments during years 0 to 90 were combined with the total soil carbon values
to produce "total carbon sequestered."
• Steps 4 and 5: Create tables of sequestered carbon by county and land-use
categories, and apply estimates to modeled scenarios. Applying Steps 1 through 3
described above and assigning counties to their respective main ecoregions, we
calculated carbon sequestration rates by county for the land-use conversion from
cropland and pastureland to (1) forest, (2) wetlands, (3) natural revegetation, and
(4) grassland. The carbon estimates produced for each land-use conversion scenario
were compiled by county as 5-year sequestration rates (tons of carbon per acre per 5-
year period) over a 90-year term.
In addition to the agricultural BMPs involving land conversion described above, other
agricultural BMPs included in the model also have an effect on carbon sequestration. Below we
describe how these effects are included (or not included).
Management practices that implement varying levels of tillage are expected to impact soil carbon
pools. Full tillage reduces soil carbon, whereas the absence of tillage increases carbon
sequestration (Ogle et al., 2005). Therefore, changes in soil carbon were estimated using three
(low, high, and no) tillage levels and the methodology outlined in IPCC (2006). It was assumed
that all cropland in the modeled areas of the Chesapeake Bay watershed would be planted with
perennial crops; the Bay is subject to a moist, temperate climate regime; and the soils would
consist of high and low activity clays (equal amounts of each).
A-4
-------
Based on these conditions, the native soil carbon pools in the Bay soils were estimated to be 75.5
metric tons C/ha in the top 30 cm. It was also assumed that all modeled agricultural lands have
been subject to long-term cultivation. Full (1.0), reduced (1.08), and no-till (1.15) relative stock
change factors were used to determine the influence of different tillage levels on soil carbon over
a 20-year period. The impacts of different levels of residue return or input on soil tillage carbon
sequestration were not considered or included in the carbon sequestration estimates.
A. 1.3 Valuation of Carbon Sequestration and Reduced GHG Emissions
The ecosystem services associated with carbon sequestration and avoided GHG emissions can be
valued using estimates of the average avoided damages that would otherwise result from a
release of 1 metric ton of carbon to the atmosphere (also referred to as the social cost of carbon
[SCC]). For the benefits of the Chesapeake Bay TMDL analysis, we relied on a recommended
mean SCC for 2010 using a 3 percent discount rate (Interagency Working Group on Social Cost
of Carbon, 2013). From this, we assumed a value of $34.71 (2010$) per metric ton of CO2
equivalent emissions reduced, or $127.18 per metric ton of C.
Applying this estimate of SCC to the estimated time paths of carbon flux reported, we calculated
the present value of carbon storage associated with each land-use conversion category using a 3
percent discount rate and an annualized value of carbon storage for each acre of land conversion
determined. These estimates are reported in Table A-2.
Table A-2. Per-Acre Value of Carbon Sequestration Services from Land-Use Conversion
($/ac)
Present Value3
Ecoregion
To Forest
Per Acre Value of Carbon Sequestration
Central Appalachians
Middle Atlantic Coastal Plain
North Central Appalachians
Northern Appalachian Plateau
and Uplands
Northern Piedmont
Piedmont
Ridge and Valley
Southeastern Plains
$3,021
$3,700
$3,021
$3,259
$4,123
$4,101
$2,802
$4,185
To To Grass
Wetland Buffer
Services from
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
Cropland
$308
$308
$308
$308
$308
$308
$308
$308
Land
Retirement
Annualized Value3
To
Forest
Conversion (S/ac)
$2,539
$3,116
$2,539
$2,749
$2,950
$3,109
$2,386
$3,205
$97
$119
$97
$105
$133
$132
$90
$135
To
Wetland
$124
$124
$124
$124
$124
$124
$124
$124
To Grass
Buffer
$10
$10
$10
$10
$10
$10
$10
$10
Land
Retirement
$82
$101
$82
$89
$95
$100
$77
$103
a 90-year period; 3% discount rate.
A-5
-------
Table A-2. Per-Acre Value of Carbon Sequestration Services from Land-Use Conversion
($/ac) (continued)
Present Value3
Ecoregion
To Forest
Per Acre Value of Carbon Sequestration
Central Appalachians
Middle Atlantic Coastal Plain
North Central Appalachians
Northern Appalachian Plateau
and Uplands
Northern Piedmont
Piedmont
Ridge and Valley
Southeastern Plains
$2,747
$3,403
$2,747
$2,985
$3,836
$3,815
$2,601
$3,897
To
Wetland
To Grass
Buffer
Land
Retirement
Annualized Value3
To
Forest
To
Wetland
To Grass
Buffer
Land
Retirement
Services from Pastureland Conversion (S/ac)
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
$3,841
$0
$0
$0
$0
$0
$0
$0
$0
$2,698
$3,378
$2,698
$2,934
$3,200
$3,382
$2,529
$3,492
$89
$110
$89
$96
$124
$123
$84
$126
$124
$124
$124
$124
$124
$124
$124
$124
$0
$0
$0
$0
$0
$0
$0
$0
$87
$109
$87
$95
$103
$109
$82
$113
a 90-year period; 3% discount rate.
A.2 Waterfowl Hunting Services from Wetland Restoration
For this analysis, we used a methodology adapted from Murray et al. (2009), who estimated the
effects of wetland restoration in the Mississippi Alluvial Valley on waterfowl hunting services.
The first step is to develop a model for estimating energetic carrying capacity of the CBW for
ducks. To accomplish this, we applied a "duck energy day" (DED) model. DEDs are the number
of ducks that can meet their daily energy requirements from an area of foraging habitat for a
single day (Lower Mississippi Valley Joint Venture Management Board [LMVJV] Waterfowl
Working Group, 2007). The first step is to calculate DEDs per acre provided by a specific habitat
(i.e., land use) using the following equation:
(Food density kg/ac - 20.24 kg/ac)*(l,000 g/kg)*TME kcal/g (1.1)
DER(294.35kcal/day)
where
food density = The food available in kilograms per acre in a given foraging habitat; the value
20.24 kg/ac is subtracted from the food available because ducks do not forage
in habitats where finding food becomes difficult
TME = true metabolizable energy of waterfowl foods in kilocalories per gram
DER = daily energy requirement per duck, assumed to be 294.35 kilocalories per day
for a dabbling duck
Based on a review of the literature, we selected the parameter values reported in Table A-3 for
estimating DEDs from various land-cover types in the Chesapeake Bay watershed. Food density
A-6
-------
estimates for wetland habitats were derived from a study of American black duck carrying
capacity in the Chesapeake Bay of Virginia (Eichholz and Yerkes, 2008). The resulting average
DEDs per acre vary from 34 for soybean cropland to 1,098 for tidal wetlands.
Table A-3. Duck Energy Days per Acre for Selected Land Cover Types
Land Cover
Cropland
Corn
Soybean
Wetland
Freshwater
Tidal
Food Available
(kg/ac)
61a
24a
107b
194b
TMEa
(kcal/g)
3.67
2.65
2.47
2.47
DERa
(k/cal)
294.35
294.35
294.35
294.35
DEDs/ac
508
34
482
1,206
a LMVJV (2007).
b Eichholz and Yerkes (2008) combines food from seeds and invertebrates and, for tidal wetlands, is an average
value for the brackish water, salt marsh, and mudflat categories.
The second step is to estimate baseline DEDs in the Chesapeake Bay watershed. We
accomplished this step by multiplying the number of acres in each land-cover category by the
corresponding DED/acre estimates. The third step is to estimate the baseline value of duck
hunting services, multiplying the total number of duck hunting days by state (based on Richkus
et al., 2008) by the regional average consumer surplus value of a duck hunting day (Rosenberger
and Loomis, 2001).
The final step is to estimate the increase in the value of duck hunting services associated with
each acre converted from cropland or pastureland to freshwater or tidal wetland. For this step, we
assumed that the aggregate value of duck hunting in each state increases in direct proportion to
the increase in total DEDs. The results of this step are reported in Table A-4.
Table A-4. Incremental Annual Value of Duck Hunting Services per Acre of Wetland
Restoration (2010 $)
Type of Land-Use Conversion
State
DE
MD
NY
PA
VA
WV
Cropland to
Tidal Wetland
$7.19
$7.65
NA
NA
$3.83
NA
Cropland to
Freshwater Wetland
$3.07
$3.38
$3.16
$2.27
$1.71
$0.95
Pastureland to
Tidal Wetland
$8.28
$8.60
NA
NA
$4.26
NA
Pasture to
Freshwater Wetland
$4.16
$4.33
$6.85
$4.06
$2.14
$1.61
A-7
-------
A.3 Nonwaterfowl Hunting Services from Increases in Forest Cover
To estimate the effects of land-use/land-cover change on other hunting services, a hedonic price
study of hunting leases by Shrestha and Alavalapati (2004) was applied to the Bay. Although this
study was conducted in central Florida, it is geographically the closest study that has estimated
the effect of different types of land cover on hunting values. Using 2002 data for 74 parcels, the
study found that the percentage of land under forest (i.e., tree and vegetation) cover had a
positive and statistically significant effect on the value of leases. In particular, they estimated a
price elasticity of 0.132 with respect to forest cover.
To apply the Shrestha and Alavalapati (2004) results to the Bay watershed, their estimated price
elasticity was assumed to also reflect the incremental contribution of forest cover to
nonwaterfowl hunting values in the Bay watershed. Specifically, the following relationship was
used to estimate the increase in hunting services associated with increases in forest cover:
= DixVlxax(100xAFi/Li) (1 2)
where
AHSi = increase in the aggregate value of hunting services in the Chesapeake Bay
watershed from state /' in 2010 dollars
Di = annual number of nonwaterfowl hunting days in the watershed in state /' in
2008
Vi = average value (consumer surplus) of a nonwaterfowl hunting day in state / in
2010 dollars
a = elasticity of hunting value with respect to forest cover
AFi = increase in acres of forest land cover in the watershed in state /' due to land-use
conversion
Li = acres of land in the watershed in state/'
To estimate A, we used data from Ribaudo et al. (2008) and the National Survey of
Fishing, Hunting, and Wildlife-Associated Recreation (FHWAR) (U.S. Department of the
Interior, Fish and Wildlife Service and U.S. Department of Commerce, U.S. Census Bureau,
2002; 2007). The duck hunting day estimates were deducted from these estimates to get
nonwaterfowl hunting days by state.
Average nonwaterfowl hunting day values (Vi) are based on the average of estimates for small-
and big-game hunting reported in the Rosenberger and Loomis (2001) meta-analysis. Converted
to 2010 dollars using the Consumer Price Index, these estimates are $45. 19 per day in the
Southeast (Virginia and West Virginia) and $52.32 per day in the other states.
A-8
-------
The model summarized in equation 1.2 uses estimates of the percentage point change in forest
cover per state in the Chesapeake Bay watershed to estimate the increase in value of hunting land
(i.e., public and private land used for hunting). By setting AF equal to 1 acre in equation 1.2, the
incremental annual value of nonwaterfowl hunting per acre of additional forest cover can be
estimated. These resulting estimates are reported in Table A-5.
Table A-5. Annual Value of Nonwaterfowl Hunting Services in the Chesapeake Bay
Watershed (2008 $)
(Di) (Vi) Aggregate Annual Incremental Value of
Estimated Per-Day Value of Value of Nonwaterfowl Nonwaterfowl Hunting
Nonwaterfowl Hunting Nonwaterfowl Hunting Hunting Days per Additional Forest
State Days in 2008 (OOOs) (2010$) (2010$) Acre (2010 $)
DE
MD
NY
PA
VA
WV
212
1,825
3,114
6,835
3,698
649
$52.32
$52.32
$52.32
$52.32
$45.19
$45.19
$11,092,566
$95,502,031
$162,908,344
$357,579,328
$167,089,839
$29,348,078
$1.15
$1.98
$4.84
$3.40
$1.28
$1.94
A-9
-------
-------
APPENDIX B.
DATA QUALITY ASSURANCE
None of the analyses conducted for this report required or involved primary data
collection, either from environmental media (such as water quality sampling or monitoring) or
from human subjects (such as through household surveys). Instead, the analysis relied on
secondary data sources, with the main ones being datasets created by the EPA's Chesapeake Bay
Program Office (CBPO), either as output or input to Phase 5.3.2 Chesapeake Bay Watershed
Model (CBWM). These data have either been posted on the Program's public ftp site
(ftp://ftp.chesapeakebay.net/Modeling/) or provided by CBPO staff and have therefore been
developed in compliance with CBPO's data quality assurance procedures. These data include
CBWM estimates of nutrient loads by land-river segment and land use category and policy
scenario runs, water quality by land-river segment, nutrient attenuation factors, best management
practice (BMP) application rates, point source loads and treatment technology options and costs,
BMP effectiveness rates, and BMP unit costs. Other secondary sources of data are reported in
the reference sections of this report and include published federal government reports and peer-
reviewed journal publications.
The datasets created as part of the analyses discussed in this report were all generated
using established software programs - Microsoft Excel, SAS, and GAMS. Following QA
procedures established within RTI, the batch program files are created and documented using a
template format that requires the program author to specify (1) the filename and server location,
(2) the RTI project number, (3) the author name, (4) the dates of the initial program and most
recent update, (5) the purpose of the program, and (6) the names and server location for input
and output files. In addition, subroutines within the program are commented to describe each
step in plain language. Intermediate and all final datasets generated with these programs were
routinely verified and validated by reviewing summary statistics and conducting consistency
checks, and they were stored in a project file on a secure RTI server accessible to team members.
B-l
-------
-------
References
Adams, D.M., RJ. Alig, J.M. Callaway, B.A. McCarl, and S.M. Winnett. September 1996.
"The Forest and Agricultural Sector Optimization Model (FASOM): Model Structure
and Policy Applications 1996." Research Paper PNW-RP-495. Portland, OR: U.S.
Department of Agriculture, U.S. Forest Service, Pacific Northwest Research Station.
Branosky, E., C. Jones, and M. Selman. 2011. "Comparison Tables of State Nutrient Trading
Programs in the Chesapeake Bay Watershed." WRI Fact Sheet. Version 1. Available online
athttp://pdfwri.org/factsheets/comparison_tables_of_state_chesapeake_bay_nutrient_
trading_programs.pdf
Chesapeake Bay Program Office (CBPO). 2012, in progress. "Chesapeake Bay Cost Model
(Draft)."
Eichholz, M., and T. Yerkes. 2008. "Determining Food Resources and Estimating Habitat
Carrying Capacity for Wintering and Spring Staging American Black Ducks in the
Chesapeake Bay of VA." Working paper.
Intergovernmental Panel on Climate Change (IPCC). 2006. "2006 IPCC Guidelines for National
Greenhouse Gas Inventories. Volume 4—Agriculture, Forestry and Other Land Use.
Appendix 3 CFU Emissions from Flooded Land: Basis for Future Methodological
Development." Available at http://www.ipcc-nggip.iges.or.jp/public/2006gl/pdf/4_
Volume4/V4_p_Ap3_WetlandsCH4.pdf.
Intergovernmental Panel on Climate Change (IPCC). 2013. "Climate Change 2013: The Physical
Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change," T.F. Stocker, D. Qin, G.-K. Plattner,
M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P.M. Midgley (eds.).
Cambridge, UK: Cambridge University Press.
Interagency Working Group on Social Cost of Carbon (IWGSCC), United States Government.
2013. "Technical Support Document: Technical Update of the Social Cost of Carbon for
Regulatory Impact Analysis under Executive Order 12866." Available at
http://www.whitehouse.gov/sites/default/files/omb/inforeg/social_cost_of_carbon_for_ria_2
013_update.pdf
Lower Mississippi Valley Joint Venture (LMVJV). June 2007. "LMVJV Waterfowl Working
Group Report to the Management Board." Vicksburg, MS: LMVJV.
McCann, L., and K.W. Easter. 2000. "Estimates of public sector transaction costs in NRCS
programs." Journal of Agriculture Applied Economics, 32(3), 555-63.
Murray, B., A. Jenkins, R. Kramer, and S.P. Faulkner. 2009. "Valuing Ecosystem Services from
Wetlands Restoration in the Mississippi Alluvial Valley." Ecosystem Services Series NIR
09-02. Durham, NC: Nicholas Institute for Environmental Policy Solutions, Duke
University.
R-l
-------
National Oceanic and Atmospheric Administration (NOAA). 2010. "Comparative Climatic Data.
Mean Number of Days with Minimum Temperature 32 Degrees F or Less." NOAA National
Climatic Data Center. Available at
http://lwf.ncdc.noaa.gov/oa/climate/online/ccd/min32temp.html.
Neely, H. 2008. "Restoring Farmland to Wetlands: The Potential for Carbon Credits in Eastern
North Carolina." Master's thesis. Durham, NC: Nicholas School of the Environment and
Earth Sciences, Duke University.
Ogle, S.M., FJ. Breidt, and K. Paustian. 2005. "Agricultural management impacts on soil
organic carbon storage under moist and dry climatic conditions of temperate and tropical
regions." Biogeochemistry 72:87-121.
Omernik, J.M. 1987. "Ecoregions of the Conterminous United States. Map (scale 1:7,500,000)."
Annals of the Association of American Geographers 77(1): 118-125.
Ribaudo, M., L. Hansen, D. Hellerstein, and C. Greene. 2008. The Use of Markets to Increase
Private Investment in Environmental Stewardship. U.S. Department of Agriculture,
Economic Research Service, Economic Research Report. Number 64.
Richkus, K.D., K. A. Wilkins, R.V. Raftovich, S.S. Williams, and H.L. Spriggs. 2008.
"Migratory Bird Hunting Activity and Harvest during the 2006 and 2007 Hunting Seasons:
Preliminary Estimates." Laurel, MD: U.S. Fish and Wildlife Service.
Rosenberger, R.S. and J.B. Loomis. 2001. "Benefit Transfer of Outdoor Recreation Use Values:
A Technical Document Supporting the Forest Service Strategic Plan (2000 Revision)." Gen.
Tech. Rep. RMRS-GTR-72. Fort Collins, CO: U.S. Department of Agriculture, Forest
Service, Rocky Mountain Research Station.
Sharpley, A.N., P.J.A. Kleinman, P. Jordan, L. Bergstrom, and A.L. Allen. 2009. "Evaluating the
success of phosphorus management from field to watershed." Journal of Environmental
Quality ^(5), 1981-88.
Shrestha, R.K. and J.R.R. Alavalapati. 2004. "Effect of ranchland attributes on recreational
hunting in Florida: A hedonic price analysis." Journal of Agricultural and Applied
Economics 36(3):763-772.
Simpson, T., and S. Weammert. 2009. "Developing Nitrogen, Phosphorus and Sediment
Reduction Efficiencies for Tributary Strategies Practices. BMP Assessment: Final Report."
University of Maryland, Mid-Atlantic Water Program.
U.S. Department of the Interior, Fish and Wildlife Service and U.S. Department of Commerce,
U.S. Census Bureau. 2002. 2001 National Survey of Fishing, Hunting, and Wildlife-
Associated Recreation. Washington, DC.
U.S. Department of the Interior, Fish and Wildlife Service and U.S. Department of Commerce,
U.S. Census Bureau. 2007. 2006 National Survey of Fishing, Hunting, and Wildlife-
Associated Recreation. Washington, DC.
R-2
-------
U.S. Environmental Protection Agency (U.S. EPA). 2010. "Chesapeake Bay Total Maximum
Daily Load for Nitrogen, Phosphorus and Sediment." Available online at
http://www.epa.gov/chesapeakebaytmdl/
U.S. Environmental Protection Agency (U.S. EPA). 2010. Greenhouse Gas Equivalencies
Calculator website. Available at www.epa.gov/cleanenergy/energyresources/calculator.html.
U.S. Environmental Protection Agency (U.S. EPA). 2011. "An Optimization Approach to
Evaluate the Role of Ecosystem Services in Chesapeake Bay Restoration Strategies."
EPA/600/R-11/001. Available online at http://www.epa.gov/research/docs/chesapeake-bay-
pilot-report.pdf
U.S. Environmental Protection Agency (U.S. EPA). 2011. "An Optimization Approach to
Evaluate the Role of Ecosystem Services in Chesapeake Bay Restoration Strategies. "
Washington, DC, Office of Research and Development.
U.S. Environmental Protection Agency (U.S. EPA). 2013. "Benefit and Cost Analysis for the
Proposed Effluent Limitations Guidelines and Standards for the Steam Electric Power
Generating Point Source Category." EPA-821-R-13-004. Available online at
http://water.epa.gov/scitech/wastetech/guide/steam-electric/upload/Steam-
Electric_BC A_Proposed-rule_2013 .pdf
Virginia Department of Environmental Quality (VDEQ). 2008. "Trading Nutrient Reductions
from Nonpoint Source Best Management Practices in the Chesapeake Bay Watershed:
Guidance for Agricultural Landowners and Your Potential Trading Partners." Available
online at
http://www.deq.virginia.gOv/Portals/0/DEQ/Water/PollutionDischargeElimination/VANPS
TradingManual_2-5-08.pdf
Van Houtven, G., R. Loomis, J. Baker, R. Beach, S. Casey. 2012. "Nutrient Credit Trading for
the Chesapeake Bay: An Economic Study." Report prepared for the Chesapeake Bay
Commission. Available online at http://www.chesbay.us/nutrienttrading.htm
Van Houtven, G., J. Powers, and S.K. Pattanayak. 2007. "Valuing water quality improvements in
the United States using meta-analysis: Is the glass half-full or half-empty for national policy
analysis?" Resource and Energy Economics 29(3), 206-28.
Wainger, L., G.L. Van Houtven, RJ. Loomis, J. Messer, R.H. Beach, and M.E. Deerhake. 2013.
"Tradeoffs among ecosystem services, performance certainty, and cost-efficiency in the
implementation of the Chesapeake Bay TMDL." Agricultural and Resource Economics
Review, 42(1), 196-224.
------- |