Total Maximum Daily Load (TMDL) for
Phosphorus in Cossayuna Lake
Washington County, New York
September 2008
Prepared for:
U.S. Environmental Protection Agency
Region 2
290 Broadway
New York, NY 10007
New York State Department of
Environmental Conservation
625 Broadway, 4th Floor
Albany, NY 12233
Environmental
Prepared by:
THE-
CADMUS
GROUP, INC.
-------
TABLE OF CONTENTS
1.0 INTRODUCTION 3
1.1. Background 3
1.2. Problem Statement 3
2.0 WATERSHED AND LAKE CHARACTERIZATION 4
2.1. History of the Lake and Watershed 4
2.2. Watershed Characterization 5
2.3. Lake Morphometry 8
2.4. Water Quality 9
3.0 NUMERIC WATER QUALITY TARGET 10
4.0 SO URGE ASSESSMENT 11
4.1. Analysis of Phosphorus Contributions 11
4.2. Sources of Phosphorus Loading 11
5.0 DETERMINATION OF LOAD CAPACITY 15
5.1. Lake Modeling Using the BATHTUB Model 15
5.2. Linking Total Phosphorus Loading to the Numeric Water Quality Target 15
6.0 POLLUTANT LOAD ALLOCATIONS 17
6.1. Wasteload Allocation (WLA) 17
6.2. Load Allocation (LA) 17
6.3. Margin of Safety (MOS) 19
6.4. Critical Conditions 19
6.5. Seasonal Variations 19
7.0 IMPLEMENTATION 19
7.1. Reasonable Assurance for Implementation 20
7.2. Follow-up Monitoring 22
8.0 PUBLIC PARTICIPATION 23
9.0 REFERENCES 24
APPENDIX A. AVGWLF MODELING ANALYSIS 26
APPENDIX B. BATHTUB MODELING ANALYSIS 40
APPENDIX C. TOTAL EQUIVALENT DAILY PHOSPHORUS LOAD ALLOCATIONS 45
-------
1.0 INTRODUCTION
1.1. Background
In April of 1991, the United States Environmental Protection Agency (EPA) Office of Water's
Assessment and Protection Division published "Guidance for Water Quality-based Decisions: The
Total Maximum Daily Load (TMDL) Process." In July 1992, EPA published the final "Water
Quality Planning and Management Regulation" (40 CFR Part 130). Together, these documents
describe the roles and responsibilities of EPA and the states in meeting the requirements of Section
303(d) of the Federal Clean Water Act (CWA) as amended by the Water Quality Act of 1987, Public
Law 100-4. Section 303(d) of the CWA requires each state to identify those waters within its
boundaries not meeting water quality standards for any given pollutant applicable to the water's
designated uses.
Further, Section 303(d) requires EPA and states to develop TMDLs for all pollutants violating or
causing violation of applicable water quality standards for each impaired waterbody. A TMDL
determines the maximum amount of pollutant that a waterbody is capable of assimilating while
continuing to meet the existing water quality standards. Such loads are established for all the point
and nonpoint sources of pollution that cause the impairment at levels necessary to meet the
applicable standards with consideration given to seasonal variations and margin of safety. TMDLs
provide the framework that allows states to establish and implement pollution control and
management plans with the ultimate goal indicated in Section 101 (a) (2) of the CWA: "water quality
which provides for the protection and propagation of fish, shellfish, and wildlife, and recreation in
and on the water, wherever attainable" (USEPA, 1991).
1.2. Problem Statement
Cossayuna Lake (WI/PWL ID 1103-0002) is situated in the Town of Argyle, within Washington
County, northeast of the Capital District region of New York State. Over the past couple of
decades, the lake has experienced degraded water quality that has reduced the lake's recreational and
aesthetic value. Cossayuna Lake is presently among the lakes listed on the 1996 Upper Hudson
River Drainage Basin Priority Waterbody Listings (PWL), with bathing, fish survival, aesthetics, and
[listed as impaired due to excessive algae and weeds (NYS DEC, 2007).
The recreational suitability of Cossayuna Lake has been mostly unfavorable since first evaluated by
the New York State Department of Environmental Conservation (NYS DEC) in 1992. Recreational
conditions are most often described as "slightly" to "substantially" impaired for most uses, and the
lake has been described mostly as "not quite crystal clear" to having "definite algal greenness."
Recreational assessments are typical of other lakes with similar water quality and plant densities
(aquatic plants regularly grow to the lake surface, often densely, more so when clarity is high), and
these assessments have been less favorable in recent years, despite the rise in water transparency.
The recreational assessments degrade slightly during the summer, consistent with seasonal increases
in lake productivity, although aquatic plant populations are generally stable during the summer (NYS
DEC, 2007).
A variety of sources of phosphorus are contributing to the poor water quality in Cossayuna Lake.
The water quality of the lake is influenced by runoff events from the drainage basin, as well as
-------
loading from nearby residential septic tanks. In response to precipitation, nutrients, such as
phosphorus naturally found in New York soils drain into the lake from the surrounding drainage
basin by way of streams, overland flow, and subsurface flow. Nutrients are then deposited and
stored in the lake bottom sediments. Phosphorus is often the limiting nutrient in temperate lakes
and ponds and can be thought of as a fertilizer; a primary food for plants, including algae. When
lakes receive excess phosphorus, it "fertilizes" the lake by feeding the algae. Too much phosphorus
can result in algae blooms, which can damage the ecology/aesthetics of a lake, as well as the
economic well-being of the surrounding drainage basin community.
The results from state sampling efforts confirm eutrophic conditions in Cossayuna Lake, with the
concentration of phosphorus in the lake violating the state guidance value for phosphorus (20 ug/L
or 0.020 mg/L, applied as the mean summer, epilimnetic total phosphorus concentration), which
increases the potential for nuisance summertime algae blooms. In 2004, Cossayuna Lake was added
to the NYS DEC CWA Section 303(d) list of impaired waterbodies that do not meet water quality
standards due to phosphorus impairments, but not designated as a "high priority for TMDL
development" (NYS DEC, 2004). Based on this listing, a TMDL for phosphorus is being developed
for the lake to address the impairment.
2.0 WATERSHED AND LAKE CHARACTERIZATION
2.1. History of the Lake and Watershed
Cossayuna Lake has been sampled by NYS DEC as part of several major state monitoring programs,
including the NYS DEC Division of Water's Lake Classification and Inventory (LCI) survey in 1982
and 1987 and through a similar NYS DEC Division of Water program in 1976. The lake was first
sampled as part of the Citizens Statewide Lake Assessment Program (CSLAP) in 1992. Early lake
data suggest that conductivity was slightly lower and pH slightly higher; however, water clarity and
phosphorus readings from earlier studies are within the same range as recent data (NYS DEC,
2007).
Cossayuna Lake was also sampled as part of the Conservation Department (predecessor to NYS
DEC) Biological Survey of the Upper Hudson River basin in 1932. This monitoring program
focused primarily on the relationship between water quality and fisheries management, although
some of the water quality indicators evaluated through CSLAP were also monitored in 1932.
Cossayuna Lake was described as follows (NYS DEC, 2007):
"There is considerable agitation among sportsmen and cottage owners to raise the water level by the construction of a three-
foot dam at the foot of the lake. It is hoped by this method to increase the area of water which would be too deep to support
the rank growth of vegetation which makes fishing difficult. It is safe to predict that an increase of three feet in the depth of
Cossayuna would not materially affect the extent of the weed areas... It would improve the appearance of the lake by
preventing the lower end from becoming practically dry in late summer"
These data show that water transparency was much higher in 1932, although the same seasonal drop
seen in present clarity measurements in Cossayuna Lake was also found in this earlier study.
Although not measured through CSLAP, data collected in other monitoring programs have
indicated that Cossayuna Lake exhibits deepwater oxygen depletion during the summer; this was not
apparent in the 1932 study of the lake.
-------
2.2. Watershed Characterization
Cossayuna Lake is part of the Upper Hudson River drainage basin with a southward flow draining
into Whittaker Creek, Carter Creek, and the Battenkill River. The Battenkill River flows east into
the Hudson River between the towns of Greenwich and Easton (Town of Argyle, 2001). Cossayuna
Lake has a direct drainage basin area of 6,841 acres excluding the surface area of the lake (Figure 1).
Elevations in the lake's basin range from approximately 1,165 feet above mean sea level (AMSL) to
as low as 485 feet AMSL at the surface of Cossayuna Lake. Cossayuna Lake's watershed is
distributed among the following townships: Argyle, 85%; Greenwich, 10%, Salem, 4%; Hebron, 1%
(Town of Argyle, 2001).
Figure 1. Cossayuna Lake Direct Drainage Basin
CADMUS'
GROUP. INC.
Legend
Basin Boundary
Roads
Streams
\ V
Existing land use and land cover in the Cossayuna Lake drainage basin was determined from digital
aerial photography and geographic information system (GIS) datasets. Digital land use/land cover
data were obtained from the 2001 National Land Cover Dataset (Homer, 2004). The NLCD is a
consistent representation of land cover for the conterminous United States generated from classified
30-meter resolution Landsat thematic mapper satellite imagery data. High-resolution color
orthophotos were used to manually update and refine land use categories for portions of the
drainage basin to reflect current conditions in the drainage basin (Figure 2). Appendix A provides
-------
additional detail about the refinement of land use for the drainage basin. Land use categories
(including individual category acres and percent of total) in Cossayuna Lake's drainage basin are
listed in Table 1 and presented in Figures 3 and 4.
Figure 2. Aerial Image of Cossayuna Lake
Table 1. Land Use Acres and Percent in Cossayuna Lake Drainage Basin
Land Use Category
Open Water
Ariculture
Developed Land
Low Intensity
Forest
Wetlands
Acres % of Drainage Basin
TOTAL
61
1,593
1,409
184
411
406
5
4,506
270
6,841
1%
23%
20%
3%
6%
6%
0.1%
66%
4%
100%
-------
Figure 3. Percent Land Use in Cossayuna Lake Drainage Basin
Wetland Open Water
4% \ / 1%
Row Crops
3%
Developed Land
6%
Figure 4. Land Use in Cossayuna Lake Drainage Basin
Legend
I Water
] Low Development
| High Development
] Hay/Pasture
J Row Crops
| Conifer Forest
J Mixed Forest
| Deciclous Forest
] Wooded Wetland
] Emergent Wetland
| Quarry
I Transitional
IN
Miles
0.8
1.6
2.4
3.2
-------
2.3. Lake Morphometry
Cossayuna Lake is a 649 acre waterbody at an elevation of about 485 feet AMSL. Figure 5 shows a
bathymetric map for Cossayuna Lake based on a lake contour map developed by NYS DEC. Table
2 summarizes key morphometric characteristics for Cossayuna Lake.
Figure 5. Bathymetric Map of Cossayuna Lake
Cossayuna Lake
Legend
0-5 feet
5-10 feet
10-15 feet
15-20 feet
-20 feet
CADMUS
-GROUP. INC.
Mile
0.4
1 2
1 6
Table 2. Cossayuna Lake Characteristics
Surface Area (acres)
Elevation (ft AMSL)
Maximum Depth (ft)
Mean Depth (ft)
Length (ft)
Width at widest point (ft)
Shoreline perimeter (ft)
Direct Drainage Area (acres)
Watershed: Lake Ratio
Mass Residence Time (years)
Hydraulic Residence Time (years)
649
485
20
11
10,014
2,932
45,280
6,841
11:1
0.2
0.47
-------
2.4. Water Quality
CSLAP is a cooperative volunteer monitoring effort between NYS DEC and the New York
Federation of Lake Associations (FOLA). The goal of the program is to establish a volunteer lake
monitoring program that provides data for a variety of purposes, including establishment of a long-
term database for lakes in New York State, identification of water quality problems on individual
lakes, geographic and ecological groupings of lakes, and education for data collectors and users. The
data collected in CSLAP are fully integrated into the state database for lakes, have been used to assist
in local lake management and evaluation of trophic status, spread of invasive species, and other
problems seen in the state's lakes.
Volunteers undergo on-site initial training and follow-up quality assurance and quality control
sessions are conducted by NYS DEC and trained NYS FOLA staff. After training, equipment,
supplies, and preserved bottles are provided to the volunteers by NYS DEC for bi-weekly sampling
for a 15 week period between May and October. Water samples are analyzed for standard lake water
quality indicators, with a focus on evaluating eutrophication status-total phosphorus, nitrogen
(nitrate, ammonia, and total), chlorophyll a, pH, conductivity, color, and calcium. Field
measurements include water depth, water temperature, and Secchi disk transparency. Volunteers
also evaluate use impairments through the use of field observation forms, utilizing a methodology
developed in Minnesota and Vermont. Aquatic vegetation samples, deepwater samples, and
occasional tributary samples are also collected by sampling volunteers at some lakes. Data are sent
from the laboratory to NYS DEC and annual interpretive summary reports are developed and
provided to the participating lake associations and other interested parties.
As part of CSLAP, a limited number of water quality samples were collected in Cossayuna Lake
during the summers of 1992-2007. The results from these sampling efforts show eutrophic
conditions in Cossayuna Lake, with the concentration of phosphorus in the lake exceeding the state
guidance value for phosphorus (20 ug/L or 0.020 mg/L, applied as the mean summer, epilimnetic
total phosphorus concentration), which increases the potential for nuisance summertime algae
blooms. Figure 6 shows the summer mean epilimnetic phosphorus concentrations for phosphorus
data collected during all sampling seasons and years in which Cossayuna Lake was sampled as part of
CSLAP; the number annotations on the bars indicate the number of data points included in each
summer mean.
The CSLAP dataset indicates that Cossayuna Lake is a eutrophic, or highly productive, lake, although
the lake has become less productive in recent years. The productivity of Cossayuna Lake increases
steadily (clarity drops as nutrient and algae levels rise) during the summer, resulting in recreational
assessments that degrade slightly over the course of a typical sampling season (aquatic plant densities
and coverage appear to be relatively stable over this period). This appears to be related to deepwater
nutrient levels that are highly elevated relative to those measured at the lake surface. Deepwater
phosphorus levels are higher than those measured at the lake surface; this suggests that nutrient
release from bottom sediments is probably significant, and that deepwater oxygen levels probably
become depleted during the summer. This was not apparent, however, in the 1930s assessment of
the lake (NYS DEC, 2007).
Surface phosphorus readings decreased from the mid 1990s through 2006, resulting in a drop in
algae levels and increase in water clarity readings. Phosphorus levels were higher in 2007, more
closely resembling conditions from the mid 1990s, although algae levels and water clarity readings
-------
were similar to those from 2005 and most recent years. While these data continue to suggest that
water quality conditions have improved, the annual variability in phosphorus readings and continued
eutrophic conditions indicate a continuing susceptibility to algal blooms and recreational use
impairments (NYS DEC, 2007).
Figure 6. Summer Mean Epilimnetic Total Phosphorus Levels in Cossayuna Lake
40
35
30
25
Total
Phosphorus 20
(ug/D
15
»
,l A
10 H«
?.-
**
1992 1993 1994 1995 1996 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
^^Phosphorus Water Quality Target (20 ug/L)
3.0 NUMERIC WATER QUALITY TARGET
The TMDL target is a numeric endpoint specified to represent the level of acceptable water quality
that is to be achieved by implementing the TMDL. The water quality classification for Cossayuna
Lake is A., which means that the most appropriate usages are a source of water supply for drinking,
culinary, or food processing purposes; primary and secondary contract recreation; and fishing. New
York State has a narrative standard for nutrients none in amounts that will result in growths of
algae, weeds and slimes that will impair the waters for their best usages (6 NYSCRR Part 703.2). As
part of its Technical and Operational Guidance Series (TOGS 1.1.1 and accompanying fact sheet,
NYS, 1993), NYS DEC has suggested that for waters classified as ponded (i.e., lakes, reservoirs and
ponds, excluding Lakes Erie, Ontario, and Champlain), the epilimnetic summer mean total
phosphorus level shall not exceed 20 ug/L (or 0.02 mg/L), based on biweekly sampling, conducted
from June 1 to September 30. This guidance value of 20 ug/L is the TMDL target for Cossayuna Lake.
10
-------
4.0 SOURCE ASSESSMENT
4.1. Analysis of Phosphorus Contributions
The Arc View Generalized Watershed Loading Function (AVGWLF) watershed model was used in
combination with the BATHTUB lake response model to develop the Cossayuna Lake TMDL.
This approach consists of using AVGWLF to determine mean annual phosphorus loading to the
lake, and BATHTUB to define the extent to which this load must be reduced to meet the water
quality target. This approach required no additional data collection thereby expediting the modeling
efforts.
The GWLF model was developed by Haith and Shoemaker (1987). GWLF simulates runoff and
stream flow by a water-balance method based on measurements of daily precipitation and average
temperature. The complexity of GWLF falls between that of a detailed, process-based simulation
model and a simple export coefficient model that does not represent temporal variability. The
GWLF model was determined to be appropriate for this TMDL analysis because it simulates the
important processes of concern, but does not have onerous data requirements for calibration.
AVGWLF was developed to facilitate the use of the GWLF model via an Arc View interface (Evans,
2002). Appendix A discusses the setup, calibration, and use of the AVGWLF model for lake TMDL
assessments in New York.
4.2. Sources of Phosphorus Loading
AVGWLF was used to estimate long-term mean annual phosphorus (external) loading to Cossayuna
Lake for two different time periods 1990-2001 and 2002-2007. The estimated mean annual
external loads of 3,515 Ibs/yr (1990-2001) and 2,544 Ibs/yr (2002-2007) of total phosphorus to
Cossayuna Lake were estimated to originate from the sources listed in Table 3 and shown in Figure
7. The 1990-2001 run was performed using the National Land Cover Database (NLCD) 2001 as is.
For the 2002-2007 run, minor corrections were made to the NLCD 2001 based on analysis of
orthoimagery; further, model parameters were adjusted to account for the retirement of cropland in
the basin. The TMDL for Cossayuna Lake was developed using the 2002-2007 model run;
therefore, the detailed source loading results (discussed in the following sections) are only provided
for the 2002-2007 run. Appendix A provides the detailed simulation results from AVGWLF.
11
-------
Table 3. Estimated Sources of Phosphorus Loading to Cossayuna Lake
Source
Hay/Pasture
Cropland
Forest
Wetlands
Developed Land
Stream Bank
Groundwater
TOTAL
Mean Annual Total Phosphorus Load (Ibs/yr)
1990 - 2001 I 2002-2007
645
1,244
165
3
47
1
436
460
95
4
68
2
1,062
3,515
1,062
418
2,544
Figure 7. Estimated Sources of Annual Total Phosphorus Loading to Cossayuna Lake
1990 - 2001
2002 - 2007
10%
Septic
Systems
30%
Developed
Land
3%
4.2.1. Residential On-Site Septic Systems
Residential on-site septic systems contribute an estimated 1,062 Ibs/yr of phosphorus to Cossayuna
Lake, which is about 42% of the total loading to the lake. Residential septic systems contribute
dissolved phosphorus to nearby waterbodies due to system malfunctions. Septic systems treat
human waste using a collection system that discharges liquid waste into the soil through a series of
distribution lines that comprise the drain field. In properly functioning (normal) systems,
phosphates are adsorbed and retained by the soil as the effluent percolates through the soil to the
shallow saturated zone. Therefore, normal systems contribute very little phosphorus loads to nearby
waterbodies. A septic system (ponding) malfunction occurs when there is a discharge of waste to
the soil surface (where it is available for runoff); as a result, malfunctioning septic systems can
contribute high phosphorus loads to nearby waterbodies. Short-circuited systems (those systems in
12
-------
close proximity to surface waters where there is limited opportunity for phosphorus adsorption to
take place) also contribute significant phosphorus loads; septic systems within 250 feet of the lake
are subject to potential short-circuiting, with those closer to the lake more likely to contribute
greater loads. Additional details about the process for estimating the population served by normal
and malfunctioning systems within the lake drainage basin is provided in Appendix A.
GIS analysis of orthoimagery for the basin shows approximately 307 houses within 50 feet of the
shoreline and 102 houses between 50 and 250 feet of the shoreline; all of the houses are assumed to
have septic systems. Within 50 feet of the shorelines, 100% of septic systems were categorized as
short-circuiting systems. Between 50 and 250 feet of the shoreline, 25% of septic systems were
categorized as short-circuiting, 10% were categorized as ponding systems, and 65% were categorized
as normal systems. To convert the estimated number of septic systems to population served, an
average household size of 2.61 people per dwelling was used based on the circa 2000 U.S. Census
Bureau estimate for number of persons per household in New York State. To account for seasonal
variations in population, NYS DEC obtained data from the county on the approximate percentage
of seasonal homes surrounding the lake. Approximately 55% of the homes around the lake are
assumed to be year-round residences, while 45% are seasonally occupied (i.e., June through August
only). The estimated population in the Cossayuna Lake drainage basin served by normal and
malfunctioning systems is summarized in Table 4.
Table 4. Population Served by Septic Systems in the Cossayuna Lake Drainage Basin
95 15 477 587
173 27 867 1,067
4.2.2. Agricultural Runoff
Agricultural land encompasses 1,593 acres (23%) of the lake drainage basin and includes hay and
pasture land (20%) and row crops (3%). Overland runoff from agricultural land is estimated to
contribute 896 Ibs/yr of phosphorus loading to Cossayuna Lake during the period 2002-2007, which
is 35% of the total phosphorus loading to the lake. This is substantially less than the 1,889 Ibs/yr of
phosphorus loading (54% of total load) to Cossayuna Lake in the time period prior to the retirement
of cropland in the basin (1990-2001).
In addition to the contribution of phosphorus to the lake from overland agriculture runoff,
additional phosphorus originating from agricultural lands is leached in dissolved form from the
surface and transported to the lake through subsurface movement via groundwater. The process for
estimating subsurface delivery of phosphorus originating from agricultural land is discussed in the
Groundwater Seepage section (below). Phosphorus loading from agricultural land originates
primarily from soil erosion and the application of manure and fertilizers. Implementation plans for
agricultural sources will require voluntary controls applied on an incremental basis.
4.2.3. Urban and Residential Development Runoff
Developed land comprises 411 acres (6%) of the lake drainage basins. Stormwater runoff from
developed land contributes 68 Ibs/yr of phosphorus to Cossayuna Lake during the period 2002-
13
-------
2007, which is about 3% of the total phosphorus loading to the lake. This load does not account for
contributions from malfunctioning septic systems.
In addition to the contribution of phosphorus to the lake from overland urban runoff, additional
phosphorus originating from developed lands is leached in dissolved form from the surface and
transported to the lake through subsurface movement via groundwater. The process for estimating
subsurface delivery of phosphorus originating from developed land is discussed in the Groundwater
Seepage section (below).
Phosphorus runoff from developed areas originates primarily from human activities, such as
fertilizer applications to lawns. Shoreline development, in particular, can have a large phosphorus
loading impact to nearby waterbodies in comparison to its relatively small percentage of the total
land area in the drainage basin.
4.2.4. Forest Land Runoff
Forested land comprises 4,506 acres (66%) of the lake drainage basin. Runoff from forested land is
estimated to contribute 95 Ibs/yr of phosphorus loading to Cossayuna Lake during the period 2002-
2007, which is about 4% of the total phosphorus loading to the lake. Phosphorus contribution from
forested land is considered a component of background loading.
4.2.5. Groundwater Seepage
In addition to nonpoint sources of phosphorus delivered to the lake by surface runoff, a portion of
the phosphorus loading from nonpoint sources seeps into the ground and is transported to the lake
via groundwater. Groundwater is estimated to transport 418 Ibs/yr (16%) of the total phosphorus
load to Cossayuna Lake during the period 2002-2007. With respect to groundwater, there is typically
a small "background" concentration owing to various natural sources. In the Cossayuna Lake
drainage basin, the model-estimated groundwater phosphorus concentration is 0.013 mg/L. The
GWLF manual provides estimated background groundwater phosphorus concentrations for >90%
forested land in the eastern United States, which is 0.006 mg/L. Consequently, about 46% of the
groundwater load (193 Ibs/yr) can be attributed to natural sources, including forested land and soils.
The remaining amount of the groundwater phosphorus load (about 225 Ibs/yr) likely originates
from agricultural or developed land sources (i.e., leached in dissolved form from the surface). It is
estimated that 50% (112.5 Ibs/yr) of the remaining 225 Ibs/yr of phosphorus transported to the lake
through groundwater originates from developed land and 50% (112.5 Ibs/yr) from agricultural land.
Table 5 summarizes this information.
Table 5. Sources of Phosphorus Transported in the Subsurface via Groundwater
193 46%
112.5 27%
112.5 27%
418 100%
14
-------
4.2.6. Other Sources
Atmospheric deposition, wildlife, waterfowl, and domestic pets are also potential sources of
phosphorus loading to the lake. All of these small sources of phosphorus are incorporated into the
land use loadings as identified in the TMDL analysis (and therefore accounted for). Further, the
deposition of phosphorus from the atmosphere over the surface of the lake is accounted for in the
lake model, though it is small in comparison to the external loading to the lake.
5.0 DETERMINATION OF LOAD CAPACITY
5.1. Lake Modeling Using the BATHTUB Model
BATHTUB was used to define the relationship between phosphorus loading to the lake and the
resulting concentrations of total phosphorus in the lake. The U.S. Army Corps of Engineers'
BATHTUB model predicts eutrophication-related water quality conditions (e.g., phosphorus,
nitrogen, chlorophyll a, and transparency) using empirical relationships previously developed and
tested for reservoir applications (Walker, 1987). BATHTUB performs steady-state water and
nutrient balance calculations in a spatially segmented hydraulic network. Appendix B discusses the
setup, calibration, and use of the BATHTUB model.
5.2. Linking Total Phosphorus Loading to the Numeric Water Quality Target
In order to estimate the loading capacity of the lake, simulated phosphorus loads from AVGWLF
were used to drive the BATHTUB model to simulate water quality in Cossayuna Lake. AVGWLF
was used to derive mean annual phosphorus loadings to the lake for the period 1990-2007; the
individual year results were run through BATHTUB and compared against the observed summer
mean phosphorus concentration for years with observed in-lake data (Figure 8). The TMDL for
Cossayuna Lake was developed using the long-term mean annual phosphorus loading to the lake for
the period 2002-2007. Using this load as input, BATHTUB was used to simulate water quality in
the lake. The combined use of AVGWLF and BATHTUB provides a good fit to the observed data
for Cossayuna Lake (Figure 8).
15
-------
Figure 8. Observed vs. Simulated Summer Mean Epilimnetic Total Phosphorus
Concentrations (ug/L) in Cossayuna Lake
50
45
40
35
Total 3°
Phosphorus 25
(ug/D 20
15
10
5
0
PI PI nil
I Observed Simulated
^
The BATHTUB model was used as a "diagnostic" tool to derive the total phosphorus load
reduction required to achieve the phosphorus target of 20 ug/L. The loading capacity of Cossayuna
Lake was determined by running BATHTUB iteratively, reducing the concentration of the drainage
basin phosphorus load until model results demonstrated attainment of the water quality target. The
maximum concentration that results in compliance with the TMDL target for phosphorus is used as
the basis for determining the lake's loading capacity. This concentration is converted into a loading
rate using simulated flow from AVGWLF.
The maximum annual phosphorus load (i.e., the annual TMDL) that will maintain compliance with
the phosphorus water quality goal of 20 ug/L in Cossayuna Lake is a mean annual load of 1,425
Ibs/yr. The daily TMDL of 3.9 Ibs/day was calculated by dividing the annual load by the number of
days in a year. Lakes and reservoirs store phosphorus in the water column and sediment, therefore
water quality responses are generally related to the total nutrient loading occurring over a year or
season. For this reason, phosphorus TMDLs for lakes and reservoirs are generally calculated on an
annual or seasonal basis. The use of annual loads, versus daily loads, is an accepted method for
expressing nutrient loads in lakes and reservoirs. This is supported by EPA guidance such as The
Lake Restoration Guidance Manual (USEPA, 1990) and Technical Guidance Manual for Performing Waste
Load Allocations, Book IV, lakes and Impoundments, Chapter 2 Eutrophication (USEPA, 1986). While a
daily load has been calculated, it is recommended that the annual loading target be used to guide
implementation efforts since the annual load of total phosphorus as a TMDL target is more easily
aligned with the design of best management practices (BMPs) used to implement nonpoint source
and stormwater controls for lakes than daily loads. Ultimate compliance with water quality
standards for the TMDL will be determined by measuring the lake's water quality to determine when
the phosphorus guidance value is attained.
16
-------
6.0 POLLUTANT LOAD ALLOCATIONS
The objective of a TMDL is to provide a basis for allocating acceptable loads among all of the
known pollutant sources so that appropriate control measures can be implemented and water quality
standards achieved. Individual wasteload allocations (WLAs) are assigned to discharges regulated by
State Pollutant Discharge Elimination System (SPDES) permits (commonly called point sources)
and unregulated loads (commonly called nonpoint sources) are contained in load allocations (LAs).
A TMDL is expressed as the sum of all individual WLAs for point source loads, LAs for nonpoint
source loads, and an appropriate margin of safety (MOS), which takes into account uncertainty
(Equation 1).
Equation 1. Calculation of the TMDL
TMDL = YWLA + T.LA +MOS
6.1. Wasteload Allocation (WLA)
There are no permitted wastewater treatment plant dischargers in the Cossayuna Lake basin. There
are also no Municipal Separate Storm Sewer Systems (MS4s) in the basin. Therefore, the WLA is set
at 0 (zero), and all of the loading capacity is allocated as a gross allotment to the load allocation.
6.2. Load Allocation (LA)
The LA is set at 1,282 Ibs/yr. Nonpoint sources that contribute total phosphorus to Cossayuna
Lake on an annual basis include loads from developed land, agricultural land, and malfunctioning
septic systems. Table 6 lists the current loading for each source and the load allocation needed to
meet the TMDL; Figure 9 provides a graphical representation of this information. Phosphorus
originating from natural sources (including forested land, wetlands, and stream banks) is assumed to
be a minor source of loading that is unlikely to be reduced further and therefore the load allocation
is set at current loading.
17
-------
Table 6. Total Annual Phosphorus Load Allocations for Cossayuna Lake1
Source
Agriculture*
Developed Land*
Septic Systems
Forest, Wetland, Stream Bank, and
Natural Background
LOAD ALLOCATION
Point Sources
WASTELOAD ALLOCATION
LA +WLA
Margin of Safety
TOTAL
Total Phosphorus Load (Ibs/yr)
Current Allocated Reduction
1,008
181
1,062
293
2,544
0
0
2,544
2,544
845
145
0
163
36
1,062
/o Reduction
16%
20%
100%
293
0 0%
1,283
0
0
1,283
142
1,425
1,261
0
0
1,261
50%
0%
0%
50%
1 - Note: The values reported in Table 6 are the annually integrated values. The daily equivalent values are provided in
Appendix C of the TMDL support document.
* Includes phosphorus transported through surface runoff and subsurface (groundwater)
Figure 9. Total Phosphorus Load Allocations for Cossayuna Lake (Ibs/yr)
Developed Land
145 Ibs/yr
Septic Systems
0 Ibs/yr
Forest, Wetland,
Stream Bank, and
Natural
Background
293 Ibs/yr
Point Sources
0Ibs/yr
Margin of Safety
142 Ibs/yr
18
-------
6.3. Margin of Safety (MOS)
The margin of safety (MOS) can be implicit (incorporated into the TMDL analysis through
conservative assumptions) or explicit (expressed in the TMDL as a portion of the loadings) or a
combination of both. For the Cossayuna Lake TMDL, the MOS is explicitly accounted for during
the allocation of loadings. An implicit MOS could have been provided by making conservative
assumptions at various steps in the TMDL development process (e.g., by selecting conservative
model input parameters or a conservative TMDL target). However, making conservative
assumptions in the modeling analysis can lead to errors in projecting the benefits of BMPs and in
projecting lake responses. Therefore, the recommended method is to formulate the mass balance
using the best scientific estimates of the model input values and keep the margin of safety in the
"MOS" term. The TMDL contains an explicit margin of safety corresponding to 10% of the
loading capacity, or 142 Ibs/yr. The MOS can be reviewed in the future as new data become
available.
6.4. Critical Conditions
TMDLs must take into account critical environmental conditions to ensure that the water quality is
protected during times when it is most vulnerable. Critical conditions were taken into account in the
development of this TMDL. In terms of loading, spring runoff periods are considered critical
because wet weather events transport significant quantities of nonpoint source loads to lakes.
However, the water quality ramifications of these nutrient loads are most severe during middle or
late summer. Therefore, BATHTUB model simulations were compared against observed data for
the summer period only. Furthermore, AVGWLF takes into account loadings from all periods
throughout the year, including spring loads.
6.5. Seasonal Variations
Seasonal variation in nutrient load and response is captured within the models used for this TMDL.
In BATHTUB, seasonality is incorporated in terms of seasonal averages for summer. Seasonal
variation is also represented in the TMDL by taking 17 years of daily precipitation data when
calculating runoff through AVGWLF, as well as by estimating septic system loading inputs based on
residency (i.e., seasonal or year-round). This takes into account the seasonal effects the lake will
undergo during a given year.
7.0 IMPLEMENTATION
One of the critical factors in the successful development and implementation of TMDLs is the
identification of potential management alternatives, such as BMPs and screening and selection of
final alternatives in collaboration with the involved stakeholders. The ongoing watershed protection
efforts (e.g., watershed characterization, restoration, and volunteer monitoring) have already been
outlined in the 1999 Cossayuna Lake Watershed Management Plan. Coordination with state
agencies, federal agencies, local governments, and stakeholders such as Cossayuna Lake
Improvement Association (CLIA), Cossayuna Lake Watershed Management Team, the general
public, environmental interest groups, and representatives from the nonpoint pollution sources will
ensure that the proposed management alternatives are technically and financially feasible. NYS
DEC, in coordination with these local interests, will address the sources of impairment, using
19
-------
primarily non-regulatory tools in this watershed, matching management strategies with sources, and
aligning available resources to effect implementation.
NYS DEC recognizes that TMDL designated load reductions alone may not be sufficient to restore
eutrophic lakes. The TMDL establishes the required nutrient reduction targets and provides some
regulatory framework to effect those reductions. However, the nutrient load only affects the
eutrophication potential of a lake. The implementation plan therefore calls for the collection of
additional monitoring data, as discussed in Section 7.2, and consideration of the Cossayuna Lake
Watershed Management Plan, which discusses in-lake measures, such as water level control, that
may need to be taken to supplement the nutrient reduction measures required by the TMDL.
7.1. Reasonable Assurance for Implementation
This TMDL was written based upon the elimination of phosphorus loading from septic systems in
conjunction with slight reductions from agricultural and developed lands. Meeting the necessary
load reductions using this approach is the most technically achievable alternative, however it may
not be financially viable. Although other allocation alternatives exist, such as clustered treatment
and on-site upgrades, they may not completely eliminate this source of phosphorus, so greater
reductions from other sources would be needed, and reasonable assurance of meeting the TMDL
would be lower.
7.1.1. Recommended Phosphorus Management Strategies for Septic Systems
Due to the fact that septic systems are the primary source of loading in the Cossayuna Lake
Watershed, restoration depends on significant reductions from that source. A cooperative
systematic approach involving the Towns of Argyle and Greenwich, such as the formation of a
management district, will be essential to achieving the load reductions specified above. A study
should be undertaken to evaluate alternatives, such as sewering with centralized treatment and
discharge out of the watershed or clustered treatment and on-site upgrades. New York State has
begun to offer funding for the abatement of inadequate onsite wastewater systems through the
development and implementation of a septic system management program by a responsible
management entity.
In the interim, a surveying and testing program should be implemented to document the location of
septic systems and verify failing systems requiring replacement in accordance with the State Sanitary
Code. State funding is also available for a voluntary septic system inspection and maintenance
program or a septic system local law requiring inspection and repair. Property owners should be
educated on proper maintenance of their septic systems and encouraged to make preventative
repairs.
To further assist municipalities, NYS DEC is involved in the development of a statewide training
program for onsite wastewater treatment system professionals. A largely volunteer industry group
called the Onsite Wastewater Treatment Training Network (OTN) has been formed. NYS DEC has
provided financial and staff support to the OTN during the last five years.
20
-------
7.1.2. Recommended Phosphorus Management Strategies for Agricultural Runoff
The New York State Agricultural Environmental Management (AEM) Program was codified into
law in 2000. Its goal is to support farmers in their efforts to protect water quality and conserve
natural resources, while enhancing farm viability. AEM provides a forum to showcase the soil and
water conservation stewardship farmers provide. It also provides information to farmers about
Concentrated Animal Feeding Operation (CAFO) regulatory requirements, which helps to assure
compliance. Details of the AEM program can be found at the New York State Soil and Water
Conservation Committee (SWCC) website, http://www.nys-soilandwater.org/aem/index.html.
Using a voluntary approach to meet local, state, and national water quality objectives, AEM has
become the primary program for agricultural conservation in New York. It also has become the
umbrella program for integrating/coordinating all local, state, and federal agricultural programs. For
instance, farm eligibility for cost sharing under the SWCC Agricultural Nonpoint Source Abatement
and Control Grants Program is contingent upon AEM participation.
AEM core concepts include a voluntary and incentive-based approach, attending to specific farm
needs and reducing farmer liability by providing approved protocols to follow. AEM provides a
locally led, coordinated and confidential planning and assessment method that addresses watershed
needs. The assessment process increases farmer awareness of the impact farm activities have on the
environment and by design, it encourages farmer participation, which is an important overall goal of
this implementation plan.
The AEM Program relies on a five-tiered process:
Tier 1 Survey current activities, future plans and potential environmental concerns.
Tier 2 Document current land stewardship; identify and prioritize areas of concern.
Tier 3 Develop a conservation plan, by certified planners, addressing areas of concern tailored to
farm economic and environmental goals.
Tier 4 Implement the plan using available financial, educational and technical assistance.
Tier 5 Conduct evaluations to ensure the protection of the environment and farm viability.
Washington County Soil and Water Conservation District should continue to implement the AEM
program on farms in the watershed, focusing on identification of management practices that reduce
phosphorus loads. These practices would be eligible for state or federal funding and because they
address a water quality impairment associated with this TMDL, should score well.
Tier 1 could be used to identify farmers that for economic or personal reasons may be changing or
scaling back operations, or contemplating selling land. These farms would be candidates for
conservation easements, or conversion of cropland to hay, as would farms identified in Tier 2 with
highly-erodible soils and/or needing stream management. Tier 3 should include a Comprehensive
Nutrient Management Plan with phosphorus indexing. This action was also recommended in the
Cossayuna Lake Watershed Management Plan. Additional practices could be fully implemented in
Tier 4 to reduce phosphorus loads, such as conservation tillage, stream fencing, rotational grazing
and cover crops. Also, riparian buffers reduce losses from upland fields and stabilize stream banks
in addition to the reductions from taking the land in buffers out of production.
21
-------
7.1.3. Recommended Phosphorus Management Strategies for Urban Stormwater Runoff
In March 2002, NYS DEC issued SPDES general permits GP-02-01 for construction activities, and
GP-02-02 for stormwater discharges from municipal separate stormwater sewer system (MS4s) in
response to the Federal Phase II Stormwater rules. These permits were re-issued, effective May 1,
2008, as GP-0-08-001 and GP-0-08-002, respectively. GP-0-08-002 applies to urbanized areas of
New York State, so it does not cover the Cossayuna Lake Watershed.
Stormwater management in rural areas can be addressed through the Nonpoint Source Management
Program. There are several measures, which, if implemented in the watershed, could directly or
indirectly reduce phosphorus loads in stormwater discharges to the lake or watershed. Many of the
following measures are also recommended in the Cossayuna Lake Watershed Management Plan:
Public education regarding:
Lawn care, specifically reducing fertilizer use or using phosphorus-free products, now
commercially available;
Cleaning up pet waste; and
Discouraging waterfowl congregation by restoring natural shoreline vegetation.
Management practices to address any significant existing erosion sites. In particular there are
two unpaved roads with steep grades located in the watershed with significant roadside ditch
erosion that would benefit from stabilization measures.
Construction site and post construction stormwater runoff control ordinance and inspection and
enforcement programs.
Pollution prevention practices for road and ditch maintenance.
7.1.4. Additional Protection Measures
Measures to further protect water quality and limit growth of phosphorus load that would otherwise
offset load reduction efforts should be considered. The basic protections afforded by local zoning
ordinances could be enhanced to limit non-compatible development, preserve natural vegetation
along shorelines and promote smart growth. The Cossayuna Lake Watershed Management Plan
recommended the implementation of a shoreline protection "belt" with guidelines for shoreline
development and land use management tools that promote protection of water quality. The State of
the Lake Report identifies eight regulated and eleven unregulated wetlands in the watershed.
Identification of wildlife habitats, sensitive environmental areas, and key open spaces within the
watershed could lead to their preservation or protection by way of conservation easements or other
voluntary controls.
7.2. Follow-up Monitoring
A targeted post-assessment monitoring effort will determine the effectiveness of the
implementation plan associated with the TMDL. Cossayuna Lake will be sampled at its deepest
location (approximately 5-6 meters), during the warmer part of the year (May through September)
on 8 sampling dates. Grab samples will be collected at 1.5 meter and in the hypolimnion. The
22
-------
samples will be analyzed for the phosphorus series (total phosphorus, total soluble phosphorus, and
soluble reactive phosphorus), the nitrogen series (nitrate, ammonia, and total nitrogen), and
chloride. The epilimnetic samples will be analyzed for chlorophyll a and the Secchi disk depth will
be measured. A simple macrophyte survey will also be conducted one time during mid-summer.
In recent years, this monitoring has been done through CSLAP. If CSLAP is discontinued at this
lake, the sampling will be repeated at a regular interval. The initial plan will be to set the interval at
5 years, but could vary based on the speed and extent of implementation. In addition, as the
information on the NYS DEC GIS system is updated (land use, BMPs, etc.), these updates will be
applied to the input data for the AVGWLF and BATHTUB models. The information will be
incorporated into the New York State 30 5 (b) report as needed.
8.0 PUBLIC PARTICIPATION
NYS DEC met with local representatives from the Washington County Water Quality Coordinating
Committee and others from the Cossayuna Lake Watershed Management Team on November 5,
2007 to discuss TMDL fundamentals and development. A second meeting was held on February
11, 2008 to refine data and to receive local input on the draft TMDL. Notice of availability of the
draft TMDL was made to local government representatives and interested parties. This draft
TMDL was public noticed in the Environmental Notice Bulletin on July 23, 2008. A 30-day public
review period was established for soliciting written comments from stakeholders prior to the
finalization and submission of the TMDL for EPA approval. NYS DEC did not receive comments.
23
-------
9.0 REFERENCES
ASCE Task Committee on Definition of Criteria for Evaluation of Watershed Models of the
Watershed Management Committee, Irrigation and Drainage Division, 1993. Criteria for evaluation
of watershed models. Journal of Irrigation and Drainage Engineering, Vol. 199, No. 3.
Day, L.D., 2001. Phosphorus Impacts from Onsite Septic Systems to Surface Waters in the
Cannonsville Reservoir Basin, NY. Delaware County Soil and Water Conservation District, Walton,
NY, June, 2001.
Evans, B.M., D.W. Lehning, K.J. Corradini, Summary of Work Undertaken delated to Adaptation of
AVGWLFfor Use in New England andNew York. 2007.
Evans, B.M., D.W. Lehning, K.J. Corradini, G.W. Petersen, E. Nizeyimana, J.M. Hamlett, P.D.
Robillard, and R.L. Day, 2002. A Comprehensive GIS-Based Modeling Approach for Predicting
Nutrient Loads in Watersheds. Journal of Spatial Hydrology, Vol. 2, No. 2.
Haith, D.A. and L.L. Shoemaker, 1987. Generalized Watershed Loading Functions for Stream Flow
Nutrients. Water Resources Bulletin, 23(3), pp. 471-478.
Homer, C. C. Huang, L. Yang, B. Wylie and M. Coan. 2004. Development of a 2001 National Eandcover
Database for the United States. Photogrammetric Engineering and demote Sensing, Vol. 70, No. 7, July 2004,
pp. 829.
National Atmospheric Deposition Program (NRSP-3). 2007. NADP Program Office, Illinois State
Water Survey, 2204 Griffith Dr., Champaign, IL 61820.
New York State Department of Environmental Conservation and New York Federation of Lake
Associations, 2007. 2006 Interpretive Summary, New York Citizens Statewide Lake Assessment
Program, Cossayuna Lake.
New York State, 2004. New York State Water Quality Report 2004. NYS Department of
Environmental Conservation, Division of Water, Bureau of Watershed Assessment and
Management.
New York State, 1998. 6 NYS Codes Rules and Regulations, Part 703.2, Narrative Water Quality
Standards.
New York State, 1993. New York State Fact Sheet, Ambient Water Quality Value for Protection of
Recreational Uses, Substance: Phosphorus, Bureau of Technical Services and Research. NYS
Department of Environmental Conservation.
Sherwood, D.A., 2005, Water resources of Monroe County, New York, water years 2000-02
atmospheric deposition, ground water, streamflow, trends in water quality, and chemical loads in
streams: U.S. Geological Scientific Investigations Report 2005-5107, 55 p.
Town of Argyle, 2001. An Action Plan for the Long Term Management of Nuisance Aquatic
Vegetation in Cossayuna Lake.
24
-------
United States Army Corps of Engineers, Engineer Research and Development Center, 2004. Flux,
Profile, and BATHTUB: Simplified Procedures for Eutrophication Assessment and Prediction.
.
USEPA. 2002. Onrite Wastemater Treatment Systems Manual. EPA/625/R-00/008. February 2002.
USEPA, 1999. Protocol for Developing Sediment TMDLs (First Edition). EPA 841-B-99-004.
Office of Water (4503F), United States Environmental Protection Agency, Washington, DC.
USEPA. 1990. The Lake and Reservoir Restoration Guidance Manual. 2nd Ed. and Monitoring Lake and
Reservoir Restoration (Technical Supplement). Prepared by North American Lake Management Society.
EPA 440/4-90-006 and EPA 440/4-90-007.
USEPA. 1986. Technical Guidance Manual for Performing Wasteload Allocations, Book IV:
Lakes, Reservoirs and Impoundments, Chapter 2: Eutrophication. EPA 440/4-84-019, p. 3-8.
United States Census Bureau, Census 2000 Summary File 3 (SF 3) - Sample Data. H18.
AVERAGE HOUSEHOLD SIZE OF OCCUPIED HOUSING UNITS BY TENURE Average
Household Si%e of Occupied housing Units by Tenure. 2007, http://factfinder.census.gov/
Watts, S., B. Gharabaghi, R.P. Rudra, M. Palmer, T. Boston, B. Evans, and M. Walters, 2005.
Evaluation of the GIS-Based Nutrient Management Model CANWET in Ontario. In: Proc. 58th
Natl. Conf. Canadian Water Resources Assoc., June 2005, Banff, Canada.
25
-------
APPENDIX A. AVGWLF MODELING ANALYSIS
Northeast A VGWLF Model
The AVGWLF model was calibrated and validated for the northeast (Evans et al, 2007). AVGWLF
requires that calibration watersheds have long-term flow and water quality data. For the northeast
model, watershed simulations were performed for twenty-two (22) watersheds throughout New York
and New England for the period 1997-2004 (Figure 10). Flow data were obtained directly from the
water resource database maintained by the U.S. Geological Survey (USGS). Water quality data were
obtained from the New York and New England State agencies. These data sets included in-stream
concentrations of nitrogen, phosphorus, and sediment based on periodic sampling.
Figure 10. Location of Calibration and Verification Watersheds for the Northeast AVGWLF
Model
I | Counties
| | Calibration
~~| Verification
Initial model calibration was performed on half of the 22 watersheds for the period 1997-2004. During
this step, adjustments were iteratively made in various model parameters until a "best fit" was achieved
between simulated and observed stream flow, and sediment and nutrient loads. Based on the
calibration results, revisions were made in various AVGWLF routines to alter the manner in which
model input parameters were estimated. To check the reliability of these revised routines, follow-up
26
-------
verification runs were made on the remaining eleven watersheds for the same time period. Finally,
statistical evaluations of the accuracy of flow and load predictions were made.
To derive historical nutrient loads, standard mass balance techniques were used. First, the in-stream
nutrient concentration data and corresponding flow rate data were used to develop load (mass) versus
flow relationships for each watershed for the period in which historical water quality data were
obtained. Using the daily stream flow data obtained from USGS, daily nutrient loads for the 1997-2004
time period were subsequently computed for each watershed using the appropriate load versus flow
relationship (i.e., "rating curves"). Loads computed in this fashion were used as the "observed" loads
against which model-simulated loads were compared.
During this process, adjustments were made to various model input parameters for the purpose of
obtaining a "best fit" between the observed and simulated data. With respect to stream flow,
adjustments were made that increased or decreased the amount of the calculated evapotranspiration
and/or "lag time" (i.e., groundwater recession rate) for sub-surface flow. With respect to nutrient loads,
changes were made to the estimates for sub-surface nitrogen and phosphorus concentrations. In regard
to both sediment and nutrients, adjustments were made to the estimate for the "C" factor for cropland
in the USLE equation, as well as to the sediment "a" factor used to calculate sediment loss due to
stream bank erosion. Finally, revisions were also made to the default retention coefficients used by
AVGWLF for estimating sediment and nutrient retention in lakes and wetlands.
Based upon an evaluation of the changes made to the input files for each of the calibration watersheds,
revisions were made to routines within AVGWLF to modify the way in which selected model
parameters were automatically estimated. The AVGWLF software application was originally developed
for use in Pennsylvania, and based on the calibration results, it appeared that certain routines were
calculating values for some model parameters that were either too high or too low. Consequently, it
was necessary to make modifications to various algorithms in AVGWLF to better reflect conditions in
the Northeast. A summary of the algorithm changes made to AVGWLF is provided below.
ET: A revision was made to increase the amount of evapotranspiration calculated automatically by
AVGWLF by a factor of 1.54 (in the "Pennsylvania" version of AVGWLF, the adjustment factor
used is 1.16). This has the effect of decreasing simulated stream flow.
GWR: The default value for the groundwater recession rate was changed from 0.1 (as used in
Pennsylvania) to 0.03. This has the effect of "flattening" the hydrograph within a given area.
GWN: The algorithm used to estimate "groundwater" (sub-surface) nitrogen concentration was
changed to calculate a lower value than provided by the "Pennsylvania" version.
Sediment "a" Factor: The current algorithm was changed to reduce estimated stream bank-
derived sediment by a factor of 90%. The streambank routine in AVGWLF was originally
developed using Pennsylvania data and was consistently producing sediment estimates that were
too high based on the in-stream sample data for the calibration sites in the Northeast. While the
exact reason for this is not known, it's likely that the glaciated terrain in the Northeast is less
credible than the highly credible soils in Pennsylvania. Also, it is likely that the relative
abundance of lakes, ponds and wetlands in the Northeast have an effect on flow velocities and
sediment transport.
Lake/Wetland Retention Coefficients: The default retention coefficients for sediment, nitrogen
and phosphorus are set to 0.90, 0.12 and 0.25, respectively, and changed at the user's discretion.
27
-------
To assess the correlation between observed and predicted values, two different statistical measures
were utilized: 1) the Pearson product-moment correlation (R2) coefficient and 2) the Nash-Sutcliffe
coefficient. The R2 value is a measure of the degree of linear association between two variables, and
represents the amount of variability that is explained by another variable (in this case, the model-
simulated values). Depending on the strength of the linear relationship, the R2 can vary from 0 to 1,
with 1 indicating a perfect fit between observed and predicted values. Like the R2 measure, the Nash-
Sutcliffe coefficient is an indicator of "goodness of fit," and has been recommended by the American
Society of Civil Engineers for use in hydrological studies (ASCE, 1993). With this coefficient, values
equal to 1 indicate a perfect fit between observed and predicted data, and values equal to 0 indicate that
the model is predicting no better than using the average of the observed data. Therefore, any positive
value above 0 suggests that the model has some utility, with higher values indicating better model
performance. In practice, this coefficient tends to be lower than R2 for the same data being evaluated.
Adjustments were made to the various input parameters for the purpose of obtaining a "best fit"
between the observed and simulated data. One of the challenges in calibrating a model is to optimize
the results across all model outputs (in the case of AVGWLF, stream flows, as well as sediment,
nitrogen, and phosphorus loads). As with any watershed model like GWLF, it is possible to focus on a
single output measure (e.g., sediment or nitrogen) in order to improve the fit between observed and
simulated loads. Isolating on one model output, however, can sometimes lead to less acceptable results
for other measures. Consequently, it is sometimes difficult to achieve very high correlations (e.g., R
above 0.90) across all model outputs. Given this limitation, it was felt that very good results were
obtained for the calibration sites. In model calibration, initial emphasis is usually placed on getting the
hydrology correct. Therefore, adjustments to flow-related model parameters are usually finalized prior
to making adjustments to parameters specific to sediment and nutrient production. This typically
results in better statistical fits between stream flows than the other model outputs.
For the monthly comparisons, mean R2 values of 0.80, 0.48, 0.74, and 0.60 were obtained for the
calibration watersheds for flow, sediment, nitrogen and phosphorus, respectively. When considering
the inherent difficulty in achieving optimal results across all measures as discussed above (along with the
potential sources of error), these results are quite good. The sediment load predictions were less
satisfactory than those for the other outputs, and this is not entirely unexpected given that this
constituent is usually more difficult to simulate than nitrogen or phosphorus. An improvement in
sediment prediction could have been achieved by isolating on this particular output during the
calibration process; but this would have resulted in poorer performance in estimating the nutrient loads
for some of the watersheds. Phosphorus predictions were less accurate than those for nitrogen. This is
not unusual given that a significant portion of the phosphorus load for a watershed is highly related to
sediment transport processes. Nitrogen, on the other hand, is often linearly correlated to flow, which
typically results in accurate predictions of nitrogen loads if stream flows are being accurately simulated.
As expected, the monthly Nash-Sutcliffe coefficients were somewhat lower due to the nature of this
particular statistic. As described earlier, this statistic is used to iteratively compare simulated values
against the mean of the observed values, and values above zero indicate that the model predictions are
better than just using the mean of the observed data. In other words, any value above zero would
indicate that the model has some utility beyond using the mean of historical data in estimating the flows
or loads for any particular time period. As with R2 values, higher Nash-Sutcliffe values reflect higher
degrees of correlation than lower ones.
28
-------
Improvements in model accuracy for the calibration sites were typically obtained when comparisons
were made on a seasonal basis. This was expected since short-term variations in model output can
oftentimes be reduced by accumulating the results over longer time periods. In particular, month-to-
month discrepancies due to precipitation events that occur at the end of a month are often resolved by
aggregating output in this manner (the same is usually true when going from daily output to weekly or
monthly output). Similarly, further improvements were noted when comparisons were made on a
mean annual basis. What these particular results imply is that AVGWLF, when calibrated, can provide
very good estimates of mean annual sediment and nutrient loads.
Following the completion of the northeast AVGWLF model, there were a number of ideas on ways
to improve model accuracy. One of the ideas relates to the basic assumption upon which the work
undertaken in that project was based. This assumption is that a "regionalized" model can be
developed that works equally well (without the need for resource-intensive calibration) across all
watersheds within a large region in terms of producing reasonable estimates of sediment and
nutrient loads for different time periods. Similar regional model calibrations were previously
accomplished in earlier efforts undertaken in Pennsylvania (Evans et al., 2002) and later in southern
Ontario (Watts et al., 2005). In both cases this task was fairly daunting given the size of the areas
involved. In the northeast effort, this task was even more challenging given the fact that the
geographic area covered by the northeast is about three times the size of Pennsylvania, and arguably
is more diverse in terms of its physiographic and ecological composition.
As discussed, AVGWLF performed very well when calibrated for numerous watersheds throughout
the region. The regionalized version of AVGWLF, however, performed less well for the verification
watersheds for which additional adjustments were not made subsequent to the initial model runs.
This decline in model performance may be a result of the regionally-adapted model algorithms not
being rigorous enough to simulate spatially-varying landscape processes across such a vast
geographic region at a consistently high degree of accuracy. It is likely that un-calibrated model
performance can be enhanced by adapting the algorithms to reflect processes in smaller geographic
regions such as those depicted in the physiographic province map in Figure 11.
Fine-tuning & Re-Calibrating the Northeast A VGWLF for New York State
For the TMDL development work undertaken in New York, the original northeast AVGWLF
model was further refined by The Cadmus Group, Inc. and Dr. Barry Evans to reflect the
physiographic regions that exist in New York. Using data from some of the original northeast model
calibration and verification sites, as well as data for additional calibration sites in New York, three new
versions of AVGWLF were created for use in developing TMDLs in New York State. Information on
the fourteen (14) sites is summarized in Table 7. Two models were developed based on the following
two physiographic regions: Eastern Great Lakes/Hudson Lowlands area and the Northeastern
Highlands area. The model was calibrated for each of these regions to better reflect local conditions, as
well as ecological and hydrologic processes. In addition to developing the above mentioned
physiographic-based model calibrations, a third model calibration was also developed. This model
calibration represents a composite of the two physiographic regions and is suitable for use in other areas
of upstate New York.
29
-------
Figure 11. Location of Physiographic Provinces in New York and New England
| Atlantic Coastal Pine Barrens
j Eastern Great Lakes and Hudson Lowlands
j Erie Drift Plain
| Laurentian Plains and Hills
j North Central Appalachians
j Northeastern Coastal Zone
j Northeastern Highlands
j Northern Appalachian Plateau and Uplands
| Northern Piedmont
j Ridge and Valley
Table 7. AVGWLF Calibration Sites for use in the New York TMDL Assessments
Site
Owasco Lake
West Branch
Little Chazy River
Little Otter Creek
Poultney River
Farmington River
Saco River
Squannacook River
Ashuelot River
Laplatte River
Wild River
Salmon River
Norwalk River
Lewis Creek
Location
NY
NY
NY
VT
VT/NY
CT
ME/NH
MA
NH
VT
ME
CT
CT
VT
Physiographic Region
Eastern Great Lakes/Hudson Lowlands
Northeastern Highlands
Eastern Great Lakes /Hudson Lowlands
Eastern Great Lakes/Hudson Lowlands
Eastern Great Lakes/Hudson Lowlands & Northeastern
Highlands
Northeastern Highlands
Northeastern Highlands
Northeastern Highlands
Northeastern Highlands
Eastern Great Lakes /Hudson Lowlands
Northeastern Highlands
Northeastern Coastal Zone
Northeastern Coastal Zone
Eastern Great Lakes /Hudson Lowlands
30
-------
Set-up of the "New York State" AVGWLF Model
Using data for the time period 1990-2007, the calibrated AVGWLF model was used to estimate
mean annual phosphorus loading to the lake. Table 8 provides the sources of data used for the
AVGWLF modeling analysis. The various data preparation steps taken prior to running the final
calibrated AVGWLF Model for New York are discussed below the table.
Table 8. Information Sources for AVGWLF Model Parameterization
WEATHER.DAT file
Data
Source or Value
Historical weather data from Glen Falls, NY and
Whitehall, NY National Weather Services Stations
TRANSPORT.DAT file
Data
Basin size
Land use/cover distribution
Curve numbers by source area
USLE (KLSCP) factors by source area
ET cover coefficients
Erosivity coefficients
Daylight hrs. by month
Growing season months
Initial saturated storage
Initial unsaturated storage
Recession coefficient
Seepage coefficient
Initial snow amount (cm water)
Sediment delivery ratio
Soil water (available water capacity)
Source or Value
GIS/derived from basin boundaries
GIS/derived from land use/cover map
GIS/derived from land cover and soil maps
GIS/derived from soil, DEM, & land cover
GIS/derived from land cover
GIS/derived from physiographic map
Computed automatically for state
Input by user
Default value of 10 cm
Default value of 0 cm
Default value of 0.1
Default value of 0
Default value of 0
GIS/based on basin size
GIS/derived from soil map
NUTRIENT.DAT file
Data
Dissolved N in runoff by land cover type
Dissolved P in runoff by land cover type
N/P concentrations in manure runoff
N/P buildup in urban areas
N and P point source loads
Background N/P concentrations in GW
Background P concentrations in soil
Background N concentrations in soil
Months of manure spreading
Population on septic systems
Per capita septic system loads (N/P)
Source or Value
Default values /adjusted using GWLF Manual
Default values/adjusted using GWLF Manual
Default values/adjusted using AEU density
Default values (from GWLF Manual)
Derived from SPDES point coverage
Derived from new background N map
Derived from soil P loading map/adjusted using
GWLF Manual
Based on map in GWLF Manual
Input by user
Derived from census tract maps for 2000 and house
counts
Default values/adjusted using AEU density
31
-------
Land Use
The 2001 NLCD land use coverage was obtained, receded, and formatted specifically for use in
AVGWLF. The New York State High Resolution Digital Orthoimagery (for the time period 2000
2004) was used to perform updates and corrections to the 2001 NLCD land use coverage to more
accurately reflect current conditions. Each basin was reviewed independently for the potential need
for land use corrections; however individual raster errors associated with inherent imperfections in
the satellite imagery have a far greater impact on overall basin land use percentages when evaluating
smaller scale basins. As a result, for large basins, NLCD 2001 is generally considered adequate,
while in smaller basins, errors were more closely assessed and corrected. The following were the
most common types of corrections applied generally to smaller basins:
1) Areas of low intensity development that were coded in the 2001 NLCD as other land use types
were the most commonly corrected land use data in this analysis. Discretion was used when
applying corrections, as some overlap of land use pixels on the lake boundary are inevitable due
to the inherent variability in the aerial position of the sensor creating the image. If significant
new development was apparent (i.e., on the orthoimagery), but was not coded as such in the
2001 NLCD, than these areas were re-coded to low intensity development.
2) Areas of water that were coded as land (and vise-versa) were also corrected. Discretion was used
for reservoirs where water level fluctuation could account for errors between orthoimagery and
land use.
3) Forested areas that were coded as row crops/pasture areas (and vise-versa) were also corrected.
For this correction, 100% error in the pixel must exist (e.g., the supposed forest must be
completely pastured to make a change); otherwise, making changes would be too subjective.
Conversions between forest types (e.g., conifer to deciduous) are too subjective and therefore
not attempted; conversions between row crops and pasture are also too subjective due to the
practice of crop rotation. Correction of row crops to hay and pasture based on orthoimagery
were therefore not undertaken in this analysis.
Phosphorus retention in wetlands and open waters in the basin can be accounted for in AVGWLF.
AVGWLF recommends the following coefficients for wetlands and pond retention in the northeast:
nitrogen (0.12), phosphorus (0.25), and sediment (0.90). Wetland retention coefficients for large,
naturally occurring wetlands vary greatly in the available literature. Depending on the type, size and
quantity of wetland observed, the overall impact of the wetland retention routine on the original
watershed loading estimates, and local information regarding the impact of wetlands on watershed
loads, wetland retention coefficients defaults were adjusted accordingly. The percentage of the
drainage basin area that drains through a wetland area was calculated and used in conjunction with
nutrient retention coefficients in AVGWLF. To determine the percent wetland area, the total basin
land use area was derived using Arc View. Of this total basin area, the area that drains through
emergent and woody wetlands were delineated to yield an estimate of total watershed area draining
through wetland areas. If a basin displays large areas of surface water (ponds) aside from the water
body being modeled, then this open water area is calculated by subtracting the water body area from
the total surface water area.
32
-------
On-site Wastewater Treatment Systems ("septic tanks")
GWLF simulates nutrient loads from septic systems as a function of the percentage of the
unsewered population served by normally functioning vs. three types of malfunctioning systems:
ponded, short-circuited, and direct discharge (Haith et al., 1992).
Normal Systems are septic systems whose construction and operation conforms to
recommended procedures, such as those suggested by the EPA design manual for on-site
wastewater disposal systems. Effluent from normal systems infiltrates into the soil and enters
the shallow saturated zone. Phosphates in the effluent are adsorbed and retained by the soil and
hence normal systems provide no phosphorus loads to nearby waters.
Short-Circuited Systems are located close enough to surface water (~15 meters) so that
negligible adsorption of phosphorus takes place. The only nutrient removal mechanism is plant
uptake. Therefore, these systems are always contributing to nearby waters.
Ponded Systems exhibit hydraulic malfunctioning of the tank's absorption field and resulting
surfacing of the effluent. Unless the surfaced effluent freezes, ponding systems deliver their
nutrient loads to surface waters in the same month that they are generated through overland
flow. If the temperature is below freezing, the surfacing is assumed to freeze in a thin layer at
the ground surface. The accumulated frozen effluent melts when the snowpack disappears and
the temperature is above freezing.
Direct Discharge Systems illegally discharge septic tank effluent directly into surface waters.
GWLF requires an estimation of population served by septic systems to generate septic system
phosphorus loadings. In reviewing the orthoimagery for the lake, it became apparent that septic
system estimates from the 1990 census were not reflective of actual population in close proximity to
the shore. Shoreline dwellings immediately surrounding the lake account for a substantial portion of
the nutrient loading to the lake. Therefore, the estimated number of septic systems in the drainage
basin was refined using a combination of 1990 and 2000 census data and GIS analysis of
orthoimagery to account for the proximity of septic systems immediately surrounding the lake. If
available, local information about the number of houses within 250 feet of the lakes was obtained
and applied. Great attention was given to estimating septic systems within 250 feet of the lake (those
most likely to have an impact on the lake). To convert the estimated number of septic systems to
population served, an average household size of 2.61 people per dwelling was used based on the
circa 2000 USCB census estimate for number of persons per household in New York State.
GWLF also requires an estimate of the number of normal and malfunctioning septic systems. This
information was not readily available for the lake. Therefore, several assumptions were made to
categorize the systems according to their performance. These assumptions are based on data from
local and national studies (Day, 2001; USEPA, 2002) in combination with best professional
judgment. To account for seasonal variations in population, data from the 2000 census were used to
estimate the percentage of seasonal homes for the town(s) surrounding the lake. The failure rate for
septic systems closer to the lake (i.e., within 250 feet) were adjusted to account for increased loads
due to greater occupancy during the summer months. If available, local information about seasonal
occupancy was obtained and applied. For the purposes of this analysis, seasonal homes are
considered those occupied only during the month of June, July, and August.
33
-------
Groundwater Phosphorus
Phosphorus concentrations in groundwater discharge are derived by AVGWLF. Watersheds with a
high percentage of forested land will have low groundwater phosphorus concentrations while
watersheds with a high percentage of agricultural land will have high concentrations. The GWLF
manual provides estimated groundwater phosphorus concentrations according to land use for the
eastern United States. Completely forested watersheds have values of 0.006 mg/L. Primarily
agricultural watersheds have values of 0.104 mg/L. Intermediate values are also reported. The
AVGWLF-generated groundwater phosphorus concentration was evaluated to ensure groundwater
phosphorus values reasonably reflect the actual land use composition of the drainage basin and
modifications were made if deemed unnecessary.
Point Sources
If permitted point sources exist in the drainage basin, their location was identified and verified by
NYS DEC and an estimated monthly total phosphorus load and flow was determined using either
actual reported data (e.g., from discharge monitoring reports) or estimated based on expected
discharge/flow for the facility type.
Concentrated Animal Feeding Operations (CAFOs)
A state-wide Concentrated Animal Feeding Operation (CAFO) shapefile was provided by NYS
DEC. CAFOs are categorized as either large or medium. The CAFO point can represent either the
centroid of the farm or the entrance of the farm, therefore the CAFO point is more of a general
gauge as to where further information should be obtained regarding permitted information for the
CAFO. If a CAFO point is located in or around a basin, orthos and permit data were evaluated to
determine the part of the farm with the highest potential contribution of nutrient load. In Arc View,
the CAFO shapefile was positioned over the basin and clipped with a 2.5 mile buffer to preserve
those CAFOS that may have associated cropland in the basin. If a CAFO point is found to be
located within the boundaries of the drainage basin, every effort was made to obtain permit
information regarding nutrient management or other best management practices (BMPs) that may
be in place within the property boundary of a given CAFO. These data can be used to update the
nutrient file in AVGWLF and ultimately account for agricultural BMPs that may currently be in
place in the drainage basin.
Municipal Separate Storm Sewer Systems (MS4s)
Stormwater runoff within Phase II permitted Municipal Separate Storm Sewer Systems (MS4s) is
considered a point source of pollutants. Stormwater runoff outside of the MS4 is non-permitted
stormwater runoff and, therefore, considered nonpoint sources of pollutants. Permitted Stormwater
runoff is accounted for in the wasteload allocation of a TMDL, while non-permitted runoff is
accounted for in the load allocation of a TMDL. NYS DEC determined there are no MS4s in this
basin.
34
-------
AVGWLF Model Simulation Results (2002-2007^
Input Transport File
Edit Transport File
-I lxi
Rural LU
HAWPAST
CROPLAND
FOREST
WETLAND
Aiea (ha)
Baie Land
Urban LU
LO_INT_DEV
HI INT DEV
Month Ket
Season Eros Stream Ground
Coef Extract Extract
Antecedent Moisture Condition
Day 1 Day 2 Day 3 Day 4 Day 5
[ci " Ei " [01 ~ W~ ~ W~
[
transedit"! .dat
t3 avgwlf
Init Unsat Stor (cm) \f\
Init Sat Stor (cm) |Q
Recess Coef (1/dia) n ri5
Seepage Coef (1'dia:) n
Tile Drain Density IQ
Initial InitSnow (cm) g
Sed Delivery Ratio |g -|g
Sediment A Factor |7.8B57E-05
U nsat Avail Wat (cm) j g gggyi 3
Tile Drain Ratio [rfi
Load Transport File M
Save File
Close
35
-------
Input Nutrient File
3 Edit Nutrient File
Runoff Loads by Source
Rural Runoff Dis N mg/L Dis P mg/L
HAY/PAST 12.9
CROPLAND |2.9 |0.26
FOREST J0.19 10.006
WETLAND 10.19 |0.006
Manure
Urban Build-Up N kg*a/d P kgrtia/d
LO_INT_DEV 10.055 10.008
HI INT DEV 10.101 10.011
avgwlf
ta output
^Revisions
^^^f
J
Nitrogen and Phosphorus Loads from Point Sources and Septic Systems
Point Source Loads/Discharge
Month
KgN
Discharge
MGD
Septic System Loads
Normal Ponding Short Cine Direct
Systems Systems Systems Discharge
|95 [15 1477 [Q~
|95
fl73
|173
|T73
[95
|9~5
J95
195
m
j95
195
27
!27
;27
15
1477
;867~
[867~
[B67~
f477~
i477
|477~
f477~
;477
f477~
f477~
Per capita tank effluent
N (g/d) P(g/d)
Ry~ |2.5
Growing season N/P Uptake Sediment
N (g/d) P (g/d) N (mg/Kg) P (mg/Kg)
|1.6 jo". 4 13000.0 1267.0
Groundwater
N (mg/L)
10.94
P (mg/L)
|0.013|
Tile Drainage (mg/L)
N P Sed
lF~ IbT
Load Nutrient File
Save File
Close
36
-------
Simulated Hydrology Transport Summary
GWLF Transport Summary for cossayuna_revision2b
Period of analysis 6 years, from Apr 2002 to Mar 2008
Units in Centimeters
Month
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
JAN
FEB
MAR
Total
Prec ET
|10.58 1 3.72
)10.52 1 7.02
|11.63 |7.26
|11.92 |8.26
|9.87 |7.87
|10.03 1 5.37
|11.78 1 4.58
|10.60 1 2.46
|10.18 1 0.90
|7.1Q |0.27
|5.77 |0.25
|7.48 |1.79
|117.4 |49.74
Go Back |
Extraction
1 0.00
|o.oo
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
IO.OQ
1 0.00
IO.OQ
1 0.00
1 0.00
Runoff
1 0.91
|0.25
|0.31
|0.15
1 0.43
|1.02
1 0.50
]1.73
1 0.66
1 2. 47
|10.04
Loads by Month j
Subsurface Point Src
Flow Flow
1 6. 54
|5.04
1 4. 06
1 3. 73
1 2. 64
1 2. 80
1 4. 90
[6.34
1 6. 56
|6.11
1 3. 57
5.54
1 57. 82
Print |
1 0.00
|0.00
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
[0.00
1 0.00
(0.00
CEjyjiiMtMOf
Tile Drain
0.00
[bio
1 0.00
1 0.00
1 0.00
1 0.00
1 0.00
0.00
1 0.00
0.00
IO.QQ
1 0.00
1 0.00
^ia::::}|
Stream
Flow
1 7. 45
|5.29
1 4. 37
1 3. 88
1 2. 75
1 3. 23
1 5. 92
1 6. 83
1 8. 28
7.62
|4.23
1 8. 00
1 67. 86
Close
37
-------
Simulated Nutrient Transport Summary
GWLF Transport Summary for cossayuna_revision2b
Period of analysis 6 years, from Apr 2002 to Mar 2008
Kg X 1000
Month
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
JAN
FEE
MAR
Total
Erosion
1310.0
1143.9
1733.5
1596.0
1032.1
356.4
414.6
257.5
123.3
67.3
11.9
67.4
8118.9
Sediment Dis N
|15.7
|8.4
|15.6
|4.1
|7.4
1 39. 5
|72.2
1 36. 9
|204.5
|176.7
1 34. 7
1 250. 4
|866.1
Go Back
CH
1 1864. 5
1 1346. 8
|1161.6
1 1048.0
1 774. 6
|821.2
1 1458. 9
1 1709. 9
1 1998. 2
1 1890.1
1 1077. 7
1 1957. 4
|1 71 08.7
Loads by
HiMltiOPEGij|
Nutrient Loads (Kg)
Total N
1 1972. 9
1 1402.0
1 1277.1
1 1090.1
816.4
999.0
1 1785. 6
1 1868.1
1 2894. 8
1 2662. 5
1 1223. 4
1 3052. 5
(21044.2
Source
Print
DisP
(62.7
47.5
|66.7
1 65. 7
(61.8
1 47.1
1 75.0
|65.7
1 97. 7
J67.3
1 48. 7
1 74.1
(779.9
Close
J
Total P
(74.8
53.8
1 79.6
|71.0
|66.2
|62.9
|104.1
1 79. 7
|175.3
|134.1
|61.3
|168.5
(1131.2
38
-------
Simulated Total Loads by Source
GWLF Total Loads for cossayuna_revision2b
Period of analysis: 6 years, from Apr 2002 to Mar 2008
Source 'Hal lcml
HAY/PAST (5so [gl
CROPLAND |78~~ ~ ^~
FOREST |1807 ~ [al
WETLAND Til 24.8
LO_INT_DEV (iso [TIT"
HI_INT_DEV [2"" ~ [431
Tile Drainage
Stream Bank
Groundwater
Point Sources
Septic Systems
T°tals [2738 [TOO
Kg X 1000 |
Erosion
2484.2
[4373.3
J882.9
|2.5
|374.9
[1.1
1
1
1
1
1
1
: 81 18. 9
Sediment
1 258.0
1 454.1
191.7
|0.3
1 23. 9
1 0.0
r
i
i
!
I
I
I
|o.o
1 23.0
1 850. 9
DisN
1 1395. 5
1 335. 4
[267.5
]49.9
[0.0
[0.0
r
i
i
i
i
i
r
[14185.5
ID
[875.0
[17108.7
Total Loads (Kg]
Total N
.2532.1
[2336.3
[671.5
[51.0
|219.0
0.0
r
i
i
i
i
i
i
[0.0
|1.7
[14185.5
ID
[875.0
[20872.1
DisP
[100.0
|36.2
|8.2
P
[0.0
JO.O
r
I
I
I
I
I
r
[189.8
[0
[444.2
[779.9
Total P
[197.8
|208.5
;43.0
[1.6
[30.8
0.0
r
i
i
i
i
i
i
0.0
|0.7
183.8
l°
|444.2
1116.4
Go Back
! Export to JPEG M
Print
Close
39
-------
APPENDIX B. BATHTUB MODELING ANALYSIS
Model Overview
BATHTUB is a steady-state (Windows-based) water quality model developed by the U. S. Army
Corps of Engineers (USACOE) Waterways Experimental Station. BATHTUB performs steady-
state water and nutrient balance calculations for spatially segmented hydraulic networks in order to
simulate eutrophication-related water quality conditions in lakes and reservoirs. BATHTUB's
nutrient balance procedure assumes that the net accumulation of nutrients in a lake is the difference
between nutrient loadings into the lake (from various sources) and the nutrients carried out through
outflow and the losses of nutrients through whatever decay process occurs inside the lake. The net
accumulation (of phosphorus) in the lake is calculated using the following equation:
Net accumulation = Inflow Outflow Decay
The pollutant dynamics in the lake are assumed to be at a steady state, therefore, the net
accumulation of phosphorus in the lake equals zero. BATHTUB accounts for advective and
diffusive transport, as well as nutrient sedimentation. BATHTUB predicts eutrophication-related
water quality conditions (total phosphorus, total nitrogen, chlorophyll-a, transparency, and
hypolimnetic oxygen depletion) using empirical relationships derived from assessments of reservoir
data. Applications of BATHTUB are limited to steady-state evaluations of relations between
nutrient loading, transparency and hydrology, and eutrophication responses. Short-term responses
and effects related to structural modifications or responses to variables other than nutrients cannot
be explicitly evaluated.
Input data requirements for BATHTUB include: physical characteristics of the watershed lake
morphology (e.g., surface area, mean depth, length, mixed layer depth), flow and nutrient loading
from various pollutant sources, precipitation (from nearby weather station) and phosphorus
concentrations in precipitation (measured or estimated), and measured lake water quality data (e.g.,
total phosphorus concentrations).
The empirical models implemented in BATHTUB are mathematical generalizations about lake
behavior. When applied to data from a particular lake, actual observed lake water quality data may
differ from BATHTUB predictions by a factor of two or more. Such differences reflect data
limitations (measurement or estimation errors in the average inflow and outflow concentrations) or
the unique features of a particular lake (no two lakes are the same). BATHTUB's "calibration
factor" provides model users with a method to calibrate the magnitude of predicted lake response.
The model calibrated to current conditions (against measured data from the lakes) can be applied to
predict changes in lake conditions likely to result from specific management scenarios, under the
condition that the calibration factor remains constant for all prediction scenarios.
Model Set-up
Using descriptive information about Cossayuna Lake and its surrounding drainage area, as well as
output from AVGWLF, a BATHTUB model was set up for Cossayuna Lake. Mean annual
phosphorus loading to the lake was simulated using AVGWLF for the period 1990-2007. The
TMDL for Cossayuna Lake was developed using the long-term mean annual phosphorus loading to
the lake for the period 2002-2007. Using this load as input, BATHTUB was used to simulate water
40
-------
quality in the lake. After initial model development, NYS DEC sampling data were used to assess
the model's predictive capabilities and, if necessary, "fine tune" various input parameters and sub-
model selections within BATHTUB during a calibration process. Once calibrated, BATHTUB was
used to derive the total phosphorus load reduction needed in order to achieve the TMDL target.
Sources of input data for BATHTUB include:
Physical characteristics of the watershed and lake morphology (e.g., surface area, mean depth,
length, mixed layer depth) - Obtained from CSLAP and bathymetric maps provided by NYS
DEC or created by the Cadmus Group, Inc.
Flow and nutrient loading from various pollutant sources - Obtained from AVGWLF output.
Precipitation Obtained from nearby National Weather Services Stations.
Phosphorus concentrations in precipitation (measured or estimated), and measured lake water
quality data (e.g., total phosphorus concentrations) Obtained from NYS DEC or USGS.
Tables 9 12 summarize the primary model inputs for Cossayuna Lake, including the coefficient of
variation (CV), which reflects uncertainly in the input value. Default model choices are utilized
unless otherwise noted. Spatial variations (i.e., longitudinal dispersion) in phosphorus
concentrations are not a factor in the development of the TMDL for Cossayuna Lake. Therefore,
division of the lake into multiple segments was not necessary for this modeling effort. Modeling the
entire lake with one segment provides predictions of area-weighted mean concentrations, which are
adequate to support management decisions. Water inflow and nutrient loads from the lake's
drainage basin were treated as though they originated from one "tributary" (i.e., source) in
BATHTUB and derived from AVGWLF.
BATHTUB is a steady state model, whose predictions represent concentrations averaged over a
period of time. A key decision in the application of BATHTUB is the selection of the length of
time over which water and mass balance calculations are modeled (the "averaging period"). The
length of the appropriate averaging period for BATHTUB application depends upon what is called
the nutrient residence time, which is the average length of time that phosphorus spends in the water
column before settling or flushing out of the lake. Guidance for BATHTUB recommends that the
averaging period used for the analysis be at least twice as large as nutrient residence time for the lake.
The appropriate averaging period for water and mass balance calculations would be 1 year for lakes
with relatively long nutrient residence times or seasonal (6 months) for lakes with relatively short
nutrient residence times (e.g., on the order of 1 to 3 months). The turnover ratio can be used as a
guide for selecting the appropriate averaging period. A seasonal averaging period (April/May
through September) is usually appropriate if it results in a turnover ratio exceeding 2.0. An annual
averaging period may be used otherwise. Other considerations (such as comparisons of observed
and predicted nutrient levels) can also be used as a basis for selecting an appropriate averaging
period, particularly if the turnover ratio is near 2.0.
Precipitation inputs were taken from the observed long term mean daily total precipitation values
from the Glen Falls, NY and Whitehall, NY National Weather Services Stations for the 2002-2007
period. Evapotranspiration was derived from AVGWLF using daily weather data (2002-2007) and a
cover factor dependent upon land use/cover type. The values selected for precipitation and change
in lake storage have very little influence on model predictions. Atmospheric phosphorus loads were
41
-------
specified using data collected by NYS DEC from the Cedar Lane Atmospheric Deposition Station
located in Lake George Village, in Warren County. Atmospheric deposition is not a major source of
phosphorus loading to Cossayuna Lake and has little impact on simulations.
Lake surface area, mean depth, and length were derived using GIS analysis of bathymetric data.
Depth of the mixed layer was estimated using a multivariate regression equation developed by
Walker (1996). Existing water quality conditions in Cossayuna Lake were represented using an
average of the observed summer mean phosphorus concentrations for years 2002-2007. These data
were collected through NYS DEC's CSLAP. The concentration of phosphorus loading to the lake
was calculated using the average annual flow and phosphorus loads simulated by AVGWLF. To
obtain flow in units of volume per time, the depth of flow was multiplied by the drainage area and
divided by one year. To obtain phosphorus concentrations, the nutrient mass was divided by the
volume of flow.
Internal loading rates reflect nutrient recycling from bottom sediments. Internal loading rates are
normally set to zero in BATHTUB since the pre-calibrated nutrient retention models already
account for nutrient recycling that would normally occur (Walker, 1999). Walker warns that
nonzero values should be specified with caution and only if independent estimates or
measurements are available. In some studies, internal loading rates have been estimated from
measured phosphorus accumulation in the hypolimnion during the stratified period. Results from
this procedure should not be used for estimation of internal loading in BATHTUB unless there is
evidence the accumulated phosphorus is transported to the mixed layer during the growing season.
Specification of a fixed internal loading rate may be unrealistic for evaluating response to changes in
external load. Because they reflect recycling of phosphorus that originally entered the reservoir from
the watershed, internal loading rates would be expected to vary with external load. In situations
where monitoring data indicate relatively high internal recycling rates to the mixed layer during the
growing season, a preferred approach would generally be to calibrate the phosphorus sedimentation
rate (i.e., specify calibration factors < 1). However, there still remains some risk that apparent
internal loads actually reflect under-estimation of external loads.
42
-------
Table 9. BATHTUB Model Input Variables: Model Selections
Water Quality Indicator
Phosphorus Calibration
Error Analysis
Availability Factors
Mass Balance Tables
Default model choice
Description
2nd Order Available Phosphorus
01 Decay Rate*
01 Model and Data*
00 Ignore*
Use Estimated Concentrations
Table 10. BATHTUB Model Input: Global Variables
Model Input
Averaging Period (years)
Precipitation (meters)
Evaporation (meters)
Atmospheric Load (mg/m2-yr)- Total P
Atmospheric Load (mg/m2-yr)- Ortho P
Default model choice
Table 11. BATHTUB Model Input: Lake Variables
Morphometry
Surface Area (km2)
Mean Depth (m)
Length (km)
Estimated Mixed Depth (m)
Observed Water Quality
Mean
0.5
1.17
0.50
4.83
2.91
Mean
2.63
3.4
3.04
3.4
Mean
28.105
0.2*
0.3*
0.5*
0.5*
NA
NA
NA
0.12
0.5
Default model choice
Table 12. BATHTUB Model Input: Watershed "Tributary" Loading
Monitored Inputs
Total Watershed Area (km2)
Flow Rate (hm3/yr)
Organic P (nnb)
Mean
27.38
18.58
62.12
44.00
0.1
O.z
0.2
43
-------
Model Calibration
BATHTUB model calibration consists of:
1. Applying the model with all inputs specified as above
2. Comparing model results to observed phosphorus data
3. Adjusting model coefficients to provide the best comparison between model predictions and
observed phosphorus data (only if absolutely required and with extreme caution).
Several t-statistics calculated by BATHTUB provide statistical comparison of observed and
predicted concentrations and can be used to guide calibration of BATHTUB. Two statistics
supplied by the model, T2 and T3, aid in testing model applicability. T2 is based on error typical of
model development data set. T3 is based on observed and predicted error, taking into consideration
model inputs and inherent model error. These statistics indicate whether the means differ
significantly at the 95% confidence level. If their absolute values exceed 2, the model may not be
appropriately calibrated. The Tl statistic can be used to determine whether additional calibration is
desirable. The t-statistics for the BATHUB simulations for Cossayuna Lake are as follows:
Observed
Simulated
-0.31
-0.39
-0.31
-0.52
-0.47
-0.48
-0.51
-1.22
-0.13
-0.67
-0.42
-0.31
Average (02-0
0.12
0.53
-0.10
-U.OD
-0.58
-0.72
-0.58
-0.96
-0.88
-0.90
-0.96
-2.27
-0.25
-1.25
-0.78
-0.58
0.22
0.98
-0.19
-0.29
-0.36
-0.29
-0.48
-0.44
-0.45
-0.48
-1.13
-0.12
-0.62
-0.39
-0.29
0.11
0.49
-0.09
In cases where predicted and observed values differ significantly, calibration coefficients can be
adjusted to account for the site-specific application of the model. Calibration to account for model
error is often appropriate. However, Walker (1996) recommends a conservative approach to
calibration since differences can result from factors such as measurement error and random data
input errors. Error statistics calculated by BATHTUB indicate that the match between simulated
and observed mean annual water quality conditions in Cossayuna Lake is quite good. Therefore,
BATHTUB is sufficiently calibrated for use in estimating load reductions required to achieve the
phosphorus TMDL target in the lake.
44
-------
APPENDIX C. TOTAL EQUIVALENT DAILY PHOSPHORUS LOAD ALLOCATIONS
Source
Total Phosphorus Load (Ibs/day)
Current Allocated Reduction
2.8
0.5
2.9
0.8
2.3
0.4
0.0
0.8
7.0
0.0
0.0
7.0
___
7.0
3.5
0.0
0.0
3.5
0.4
3.9
0.5
0.1
2.9
Agriculture*
Developed Land*
Septic Systems
Forest, Wetland, Stream Bank, and
Natural Background
LOAD ALLOCATION
Point Sources
WASTELOAD ALLOCATION
LA +WLA
Margin of Safety
TOTAL
Includes phosphorus transported through surface runoff and subsurface (groundwater)
/o Reduction
18%
20%
100%
50%
0%
0%
50%
45
------- |