Content uploaded by Arjumand Zaidi
Author content
All content in this area was uploaded by Arjumand Zaidi on Apr 18, 2016
Content may be subject to copyright.
A FRAMEWORK FOR POLLUTANT TRADING DURING THE TMDL
ALLOCATION PHASE
A. Z. Zaidi1, S. M. deMonsabert2, R. El-Farhan3, and S. Choudhury4
ABSTRACT
The Environmental Protection Agency (EPA) encourages pollutant trading programs that help
achieve Total Maximum Daily Load (TMDL) implementation goals. Such trades need to be
consistent with water quality standards. For an approved TMDL, EPA recommends that the point
and/or nonpoint source waste load allocations be used as the baseline for trading credits. The
complexity inherent in modeling the effects of pollutants on receiving bodies makes it difficult to
understand the implications associated with trading pollutant loads from different sources. The
reduction of one credit from one source does not equal the reduction of one credit at another
location in the watershed. Similarly, unit costs for load reductions vary considerably depending
on the control strategy and the level of reduction. Though trading ratios may account for some
uncertainties associated with estimates of nonpoint source loads and long-term performance of
the control measures, the selection of an effective trading ratio is not very straightforward and
does not fully address the environmental impacts. Although higher trading ratios may be used,
this alone cannot guarantee that the water quality standard will be met and may unnecessarily
increase mitigation costs. This paper proposes an alternative strategy for the TMDL trading
framework. Instead of explicitly determining trading ratios, a trading scenario selection method
is utilized. Water quality is simulated for the alternate trading scenarios based on various
pollutant loads obtained from accepted models such as HSPF. The costs associated with the
nonpoint reductions are compared for various load allocation strategies. This comparison
enables an efficient screening of watersheds to identify those well-suited for a pollutant trading
strategy. Watersheds with high cost variations and flexibility among allocation strategies are
ideal candidates for trading. The methodology is demonstrated using the results of the TMDL
allocation for the Muddy Creek WAR1 subwatershed in Rockingham County, Virginia.
KEYWORDS. TMDL Trading, Fecal Coliform TMDL, Economic Analysis, Optimization,
Mathematical Modeling
1 Graduate Research Assistant, George Mason University, Fairfax, VA. 22030-4444, azaidi@gmu.edu
2 Associate Professor of Civil, Environmental and Infrastructure Engineering, George Mason University, Fairfax,
VA. 22030-4444, sdemonsa@gmu.edu
3 Senior Water Resources Engineer, The Louis Berger Group, Washington, D.C., 20037, relfarhan@louisberger.com
4 Postdoctoral Fellow, George Mason University, George Mason University, Fairfax, VA, 22030-4444,
shc758@yahoo.com
1
INTRODUCTION
According to the United States Environmental Protection Agency (EPA) the market-based
approaches such as water quality trading provide greater flexibility and have potential to achieve
greater environmental benefits than would otherwise be achieved under more traditional
regulatory approaches (EPA, 2003). Given the inherent scientific uncertainty in the nature of
nonpoint pollutants there is a high risk involved in complying with environmental regulation
while trading nonpoint pollutants. The objective of this paper is to utilize a previous model
developed by Zaidi et al (2003) as a decision tool to evaluate and determine the cost-
effectiveness of various trading options during the TMDL allocation phase. The approach uses a
mathematical model to minimize the costs for the subwatershed load allocation to achieve the
desired water quality goal while minimizing the uncertainties associated with the nonpoint
pollutant trading. The costs associated with the subwatershed load allocations establish the basis
for pollutant trading.
TMDL TRADING FRAMEWORK DURING THE ALLOCATION PHASE
Use of the proposed cost optimization approach can help to determine at the allocation stage
whether a specific TMDL lends itself to a watershed-based pollutant trading approach.
Furthermore, the approach will provide an estimate of the potential savings that may be realized
by incorporating a pollutant trading approach in the TMDL load allocation phase. Water quality
based trading accomplishes time and economic efficiencies. Not only are the reductions more
cost-effective but also the process of achieving the reductions necessary to meet the water quality
standard is accelerated.
For an approved TMDL, EPA recommends the consideration of applicable point source waste
load allocations or non-point source load allocations to establish a baseline for trading credits. A
baseline is defined as the level below which a reduction is made to create a pollutant reduction
credit (EPA, 2003). Pollutant trading takes advantage of the control cost differentials and
economies of scale between various sources of pollutants. Under EPA guidelines, watershed
based trading may be considered if it results in an overall reduction of pollutant loads without
violation of the water quality standard. Commonly, “trading ratios” are used to account for the
expected differences in impact on water quality, for two different loadings in a watershed. These
factors, may also account for the uncertainties associated with estimates of nonpoint source loads
and reductions achieved through treatment options or other control measures. The relative
impact of one unit of pollutant discharged from different sources from varied locations is never
the same.
2
MUDDY CREEK WATERSHED
In this paper, nonpoint trading options are evaluated under the framework of an approved TMDL
for fecal coliform bacteria in a subwatershed (WAR1) of the Muddy Creek watershed. For
TMDL modeling purposes, the Muddy Creek watershed was delineated into eight interconnected
subwatersheds as shown in Figure 1. The Muddy Creek watershed is located in Rockingham
County, Virginia. Muddy Creek is on the Commonwealth of Virginia’s 1998 303(d) list of
impaired waterbodies because of fecal coliform bacteria violations. The Virginia Department of
Environmental Quality (VADEQ) ambient water quality monitoring stations recorded the
exceedance of the standard to indicate that the stream does not support primary contact
recreation (swimming). The Muddy Creek fecal coliform target was a geometric mean of 200
counts/100ml with 0% violation. The Muddy Creek loads are comprised of direct loads (in-
stream discharges due to animals, failed septic systems and uncontrolled releases) and indirect
loads (surface depositions resulting from land use).
Figure 1. Muddy Creek Subwatersheds
POTENTIAL TRADING SCENARIOS
The proposed methodology identifies viable load allocation tradeoffs such that implementation
costs are minimized while achieving an acceptable level of water quality in the stream. The
optimization model identifies the minimum cost mitigation approach for a given allocation
scenario. A comparison of the costs associated with the mitigation strategies for various
allocation scenarios, provides a starting point for discussion and evaluation of trading options.
For several reasons, including site characteristics (soil type, slope, etc.), fecal load deposition
(direct vs. indirect), diversity in control strategies, etc. the remediation costs may differ among
sources. Only scenarios that produce modeled results consistent with the TMDL standards are
3
considered. In the case of the Muddy Creek WAR1 subwatershed, the trading scenarios were
modeled using a calibrated HSPF model provided by The Virginia Department of Conservation
and Recreation (DCR). The fecal coliform bacteria sources for the Muddy Creek WAR1
subwatershed are: land-based loads, wildlife loads, failed septic systems, and direct deposit as
listed in Tables 1 and 2. In the final TMDL allocation report for Muddy Creek (VADEQ, 2000),
the allocated TMDLs are based on variable monthly reductions for each source. For the
purposes of this work, the maximum monthly reduction from each source was used in the
analysis.
Table 1: Baseline Scenario
Nonpoint Source Total Existing Load
(counts/yr) Load Reduction
(counts/yr) PV Cost1 ($)
Land-based Loads 3.51E+11 1.68E+11 3,085
Wildlife Loads 2.67E+10 0.00E+00 30,0562
Failed Septic System 1.04E+11 1.04E+11 17,286
Direct Deposit 1.95E+13 1.95E+13 164,682
Total 2.00E+13 1.98E+13 215,109
1. Total Present Value (PV) costs are calculated for a 7% interest rate and 15-year planning horizon.
2. This management option reflects the cost of maintaining the existing wildlife population (assuming a 10% growth rate). As such, it does
not constitute a load reduction.
Tables1 and 2 present the baseline scenario based on the TMDL allocation report (VADEQ,
2000), and a potential trading scenario respectively. In these tables the annual load reduction
from each source and the cost of achieving that reduction are given. These costs depend not only
upon the level of load reduction but also upon the choices and selection of control measures. In
this study the vegetative buffer strip for land-based load, wildlife management for wildlife load,
system repair and installation for failed septic systems, and streamside fencing for direct cattle
deposit are considered as control strategies.
Table 2: Potential Trading Scenario
Nonpoint Source Load Reduction
(counts/yr) PV Cost ($)
Land-based Loads 8.59E+10 1,094
Wildlife Loads 1.47E+10 31,751
Failed Septic System 0.0 0
Direct Deposit 1.95E+13 164,682
Total 1.96E+13 197,527
As shown in Table 1 the total load reduction cost for the baseline scenario is $215,109, whereas
for the alternate trading scenario (Table 2) the cost is $197,527. The alternate trading scenario
presents a cost savings of 8% from the baseline scenario. The baseline scenario represents a
slightly higher annual load reduction with higher cost as compared to the alternate trading
4
scenario. Despite a lower annual load reduction, the alternate scenario is also a viable scenario
since it complies with the state standard as modeled in Figure 2. Also the geometric means of
the fecal coliform concentrations for the trading scenario are generally less than for the baseline
scenario. In both cases, the direct deposit load reduction dominates the load allocation. The
direct deposit is several orders of magnitude greater; for this reason, reductions in the land-based,
wildlife and failed septic system loads are insignificant relative to the direct deposit reduction.
10
100
1000
12/18/1990
5/1/1992
9/13/1993
1/26/1995
6/9/1996
10/22/1997
Date
counts/100ml
Standard
Baseline Scenario
Trading Scenario
Figure 2. 30-Day Geometric Mean Fecal Coliform Loadings under Baseline and Trading Scenarios
A matrix of viable trading scenarios may be produced using the calibrated HSPF model. The
calibrated HSPF model for Muddy Creek watershed obtained from DCR is used for this analysis.
This matrix of options will help to identify potential tradeoffs in the reduction of pollutant loads
from different categories of nonpoint sources to levels below the standard and their associated
costs.
The existing load distribution in the watershed does not offer much flexibility to the Muddy
Creek subwatershed WAR1 for pollutant trading because of the dominance of the direct loads.
For the illustration purpose, a synthetic load distribution was used in the following section to
better illustrate the strengths of the model.
TRADING SCENARIO ANALYSIS
The existing loads for the WAR 1 subwatershed were multiplied with randomly selected
coefficients to provide a synthetic load distribution that would better illustrate the proposed
methodology. Table 4 shows the synthetic loads used in the analysis. Table 5 presents three
alternative trading scenarios. Each of these scenarios represents an option that will meet the
5
water quality standard. Individual and total load reduction costs for each source are shown in
Table 5.
Table 4: Synthetic Loads
Nonpoint Source Loads (counts/year)
Land-based Loads 2.66E+12
Wildlife Loads 2.02E+11
Failed Septic System 2.60E+10
Direct Deposit 9.31E+11
Total Synthetic Load 3.82E+12
Table 5: Trading Scenario PV Costs
Scenario I Scenario II Scenario III
Nonpoint
Source
Reduction1
(%) PV Cost2
($) Reduction1
(%) PV Cost2
($) Reduction1
(%) PV Cost2
($)
Land-based
Loads 57 (75) 7,409 0 0 73.7 (74.2) 4,899
Wildlife
Loads 67.8 249,113 87.13 253,618 49.4 244,854
Failed Septic
System 100 4,322 0 0 0 0
Direct
Deposit 80 (89.7) 94,706 79 (89.7) 94,706 92 (93.3) 127,942
Total 63.6 (77) 355,550 24 (26.5) 348,324 76.4 (77) 377,695
1. Values in parentheses are the actual reductions achieved after applying the control measures.
2. Total Present Value (PV) costs are calculated for a 7% interest rate and 15-year planning horizon.
6
10
100
1000
12/18/1990
5/1/1992
9/13/1993
1/26/1995
6/9/1996
10/22/1997
Date
counts/100ml
Standard
Scenario I
Scenario II
Scenario III
Figure 3. 30-Day Geometric Mean Fecal Coliform Loadings under Scenarios I, II and III
Discussion
Scenario II provides the best results from both an economic and environmental perspective. The
modeled geometric means of the fecal coliform concentrations are generally less than those for
Scenarios I and III (see Figure 3). It is interesting to note that the least cost option (Scenario II,
minimum annual load reduction) provides the “best” environmental solution. This result
supports the need for economic assessment during the load allocation phase of the TMDL
development. In a trading environment, Scenario II would be preferred over Scenarios I and III.
Also a comparison of Scenarios I and III supports a trading solution. Scenario III yields a
solution that is environmentally inferior to Scenario I, yet the total cost is greater. The total cost
is 5.86% greater for Scenario III than Scenario I despite a reduced environmental load. This
result suggests that trading could result in an improved solution (both environmentally and
economically). Under Scenario I, the land based, wildlife and failing septic systems loads would
sell the ‘polluting rights’ to the direct deposit loads for reduced total costs and improved water
quality. To better understand the concept, consider scenario III as the baseline scenario. A
mutually advantageous trade may be conducted between scenarios I and III. Greater reductions
in the pollutants from land-based, wildlife, and failed septic systems are needed in scenario I
than in the scenario III. Similarly the costs associated with these added reductions are greater.
The benefits associated with reducing the direct deposit load reductions ($33,236) far exceed the
net dollar loss ($11,090) from the other three sources. Therefore, an opportunity for trading
between direct deposit sources and the other sources exists. Direct deposit sources can purchase
‘pollutant rights’ from other sources. If the ‘pollutant rights’ cost more than what is required by
the three sources to achieve the level of pollutant reduction in scenario I (more that $11,090),
then the direct load sources will save money by purchasing these rights at a cost less than
$33,236 (their cost to reduce a comparable load).
7
CONCLUSIONS
The trading scenario selection method provides improved assurance of meeting the water quality
goal. Economic modeling during the allocation phase of TMDL development is vitally
important to the environmental and economic success. Cost data, coupled with modeled
geometric means of fecal coliform concentrations will enable improved allocations that
accomplish both environmental and economic goals. Without economic information for the
various proposed allocation scenarios, it is difficult to determine what solution would be
preferred. For example, in the case study presented above, the solution with a higher load
reduction produces an environmentally and economically inferior solution. This solution
presents an opportunity for improved water quality at a reduced cost through the use of trading
scenarios.
REFERENCES
1. USEPA. 2003. Final Water Quality Trading Policy. United States Environmental Protection
Agency, Office of Water, January 13, 2003.
http://www.epa.gov/owow/watershed/trading/finalpolicy2003.html
2. VADEQ. 2000. Revised Final Report: Fecal coliform TMDL Development For Muddy Creek,
Virginia. The Muddy Creek TMDL Establishment Workgroup.
http://www.deq.state.va.us/pdf/rivers/mdycrk.pdf
3. Zaidi, Arjumand Z., deMonsabert, S. M., El-Farhan, R. 2003. A Cost-Based Strategy for
TMDL Allocation. Proceedings of the Second Conference on Watershed Management to
Meet Emerging TMDL Environmental Regulations, ASAE, Albuquerque, New Mexico,
November 8-12, 2003.
8