&EPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EFA/600/R-99/102
November 1999
A Landscape Analysis of
New York City's Water
Supply (1973-1998)
KSLEBOjCOV < 2/1/K
-------
A LANDSCAPE ANALYSIS OF NEW YORK CITY'S WATER SUPPLY (1973-1998)
by
M.H. Mehaffey, T.G. Wade, M.S. Nash and C.M. Edmonds
Environmental Sciences Division, U.S. Environmental Protection Agency,
Las Vegas, Nevada
Program Area: Sound Science, Improved Understanding
of Environmental Risk and Greater Innovation
to Address Environmental Problems
Program: Ecosystem Research
Task Area: Development of Landscape Indicators for Use
in Regional Ecological Risk Assessment
National Exposure Research Lab
Office of Research and Development
U.S. Environmental Protection Agency
Las Vegas, Nevada, 89119
-------
NOTICE
Notice: The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), funded and performed the research described here. This manuscript has been
subject to external and EPA peer review and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation by the EPA for use.
11
-------
PROJECT SUMMARY
Title: A Landscape Analysis of New York City's Water Supply (1973-1998).
Authors: M.H. Mehaffey, T.G. Wade, M.S. Nash and C.M. Edmonds
Organization: Environmental Sciences Division, U.S.Environmental Protection Agency
Core Task Summary
Currently the city of New York is trying save taxpayers the cost of a billion dollar
filtration system by protecting water quality through implementing a long-range watershed
protection program. A large part of the watershed protection plan focuses on the solicitation for
acquisition of 350 thousand acres of sensitive lands surrounding streams and rivers. The objective
of this study is to explore relationships between changes in landscape and water quality
degradation in the Catskill/Delaware basins. Together these basins supply 90% of New York
City's drinking water. The goals of this study are 1) improvements in environmental risk
assessment and 2) target watersheds and sub-watersheds influencing water quality in New York
City's water supply system.
hi this study, land use imagery will be acquired or generated for four dates, ranging from
1973 to 1998. Landsat TM imagery will be used to classify land use/ land cover for all dates
except 1973, which will require the use of multispectral scanner (MSS) images from the North
American Landscape Characterization (NALC) program. Landscape metrics believed to impact
drinking water quality, such as proportion of land use, proportion of land use in and near riparian
zones, length of roads near streams and crops on steep slopes. These landscape metrics will be
calculated from the land use coverages for all sub-basins in the Delaware/Catskill watershed. Like
many long term landscape analyses, this study will require comparisons between images having
different resolutions. In order to better determine landscape change over a 25 year time span
further evaluation of the multi-date and multi-resolution data sets will be conducted using paired
Landsat TM and MSS imagery for one of four selected image dates.
Physical, chemical and biological water data will be collected from various database
sources for the 70's 80's and 90's for streams and reservoirs. The main sources for the water quality
parameters will be the NYCDEP, NYSDEC, USGS and EPA's EMAP and storage and retrieval
data base STORET. Analyses may include 1) multi-variate and non-parametric procedures
analyses to examine yearly and seasonal change in water quality and 2) stepwise regression,
correlations and cluster analysis to find associations and relationships between land-use with water
quality parameters.
From these analyses an expanded knowledge of the impacts of landscape proportion on
water quality will be gained, and a set of landscape metrics important to surface water quality
conditions will be calculated. These landscape metrics can then be used to target watersheds and
sub-watersheds having landscape proportions strongly correlated with water quality degradation in
the Catskill/Delaware basins. In addition to the area of this study there is the potential for
application of this research in regions having similar landscape characteristics in the nearby states
of Pennsylvania and New Jersey. This study is consistent with the goals and objectives of
improved understanding of environmental risk, ecosystem research, assessment and restoration, as
articulated by the 10 year strategic plan for the landscape science program (LEB Strategic Plan).
111
-------
TABLE OF CONTENTS
Page
NOTICE ii
PROJECT SUMMARY iii
TABLE OF CONTENTS iv.
LIST OF TABLES AND FIGURES v
LIST OF ABBREVIATIONS vi
PROJECT NARRATIVE 1
GOALS AND OBJECTIVES 1
INTRODUCTION 2
New York City Water Supply - Problems 2
New York City Water Supply - Solutions 2
TECHNICAL APPROACH 4
Site Characteristics 5
Physiography and Land Use 5
Data Collection and Analytical Methods 5
Landscape Classification ' 5
Landscape Metrics 9
Water Quality 11
Statistical Analysis 13
POTENTIAL FOR REDUCING UNCERTAINTY 13
QUALITY ASSURANCE 14
ANTICIPATED PRODUCTS AND RESULTS 15
REFERENCES 16
iv
-------
LIST OF TABLES AND FIGURES
Page
TABLES
Aggregation of MRLC land use classes 9
Work Breakdown Structure 15
FIGURES
The New York City water supply system and, inset, the
Catskill/Delaware watersheds 3
The five reservoir basins and surrounding counties of the
Catskill/Delaware watersheds 3
Painted shaded relief showing elevational topography
of the Catskill/Delaware watershed 6
Surface Geology in the Catskill/Delaware Watersheds 6
Land use (1991) and Catskill Park boundary in the Catskill/Delaware watershed 7
Land use (1991) near town of Trout Creek in the Catskill/Delaware Watershed 7
STORET Water Quality Sites in the Catskill/Delaware Watershed 12
Benthic Sample Site Locations in the Catskill/Delaware Watershed 12
-------
LIST OF ABBREVIATIONS
CREP - Conservation Reserve Enhancement Program
CWA-Clean Water Act
DEM - Digital Elevation Model
EPA- Environmental Protection Agency
EPIC - Environmental Photographic Interpretation Center
EROS - Earth Resources Observation Systems
GIS - Geographic Information System
GPRA - Governmental Performance Results Act
LEB - Landscape Ecology Branch
MO A - Memorandum of Agreement
MRLC - Multi-Resolution Land Characteristics
MSS - Multispectral Scanner
NALC - North American Landscape Characterization
NERL - National Exposure Research Laboratory
NHAP - National High Aerial Photography
NYCDEP - New York City Department of Environmental Protection
NYSDEC - New York State Department of Environmental Conservation
ORD - Office of Research and Development
QA - Quality Assurance
STORET - STORage and RETrieval
. TM - Thematic Mapper
USDA - United States Department of Agriculture
USGS - United States Geological Survey
VI
-------
PROJECT NARRATIVE
GOALS AND OBJECTIVES
The primary goal of this study is to improve risk assessment through the development of methods
and tools for characterization of the relationship between landscape and water resource change.
The relationship between landscape proportion and ecological and biological response has been
established by a number of studies (Wiens and Milne, 1989; O'Neill et al., 1988; Graham et
al.,1991). However, the existing studies are limited to a narrow set of landscapes features and
tend to focus on percentages of near stream riparian land cover. The advent of geographic
information systems and greater access to remote satellite imagery has enhanced landscape
monitoring capabilities. Using land cover derived from satellite imagery with other spatial data
and a GIS, it is now possible to generate landscape metrics that reflect changes in ecological
processes (O'Neill et al.,1997).
Metrics such as proportion of land cover, topography and soils correlate well with stream and
lake condition, suggesting a relationship between landscape proportion and water quality and
quantity (Hunsaker and Levine, 1995). By further exploring this relationship an assessment of
the vulnerability of land and water resources to stresses in the environment can be measured.
Furthermore, once the information and techniques used in this study are validated through quality
assurance and peer reviews, there will be a potential of utilizing the above techniques in similar
types of watersheds in the region.
The main objectives of this study are to 1) generate a land cover database for the Catskill-
Delaware basins spanning two and a half decades (1973 to 1998), 2) produce a landscape metric
database for the watershed at the sub-basin level for each land cover image, 3) create water
quality databases, which have been verified and checked against original agency sources and 4)
relate water chemical (i.e. pH, NO3-N, TP, chloride, and silica), physical (i.e. temperature and
flow rate) and biological (i.e. microbial and macro-invertebrate) parameters to physiographic and
landscape indices at the basin, sub-basin and near stream (100 and 500 meter) scale.
Because of changes in technology from the 1970's to the 1980's the resolution of satellite images
has changed from 60 meters in the MSS images to 30 meters in the TM. Trying to evaluate
landscape change across a 25 year time span, as has been proposed by this study, requires using
both Landsat TM and MSS images. Therefore a secondary objective of this study will be to relate
the land useMand cover classification of the MSS and TM imagery using paired imagery for one
of the four selected image dates.
-------
INTRODUCTION
New York City Water Supply - Problems
Initially New York City's drinking water was supplied by public wells. The need for a greater
water supply first became an issue in 1776, when the population swelled to 22,000 residents,
resulting in construction of the city's first reservoir. Pollution of the well waters and increasing
demand eventually led to the impoundment of the Croton river in 1842 and the construction of an
elaborate aqueduct system to supply water from the Old Croton reservoir. The aqueduct could
supply over 90 million gallons per day to the city, but eventually even this amount of water was
not enough to supply the city's continued growth. New reservoirs and a second aqueduct system
were constructed in Boyds Corner and Middle Branch and were put into service in 1890.
In an effort to insure the water supply into the future, the Catskill and Delaware systems were
developed (Hecker, 1991, Weidner, 1974)(Figure 1). The construction of the impoundments,
aqueducts and tunnels needed to supply the water to the city spanned a total of six decades
(1905-1965). The six reservoirs are the Cannonsville, Pepacton, Schoharie, Neversink,
Rondout and Ashokan (Figure 2). Together, these reservoirs provide 90% of the 1.4 billion
gallons of drinking water consumed by New York City residence. Of the six reservoirs,
Ashokan Reservoir is the oldest (1907) and largest reservoir with a water surface area of 12.8
square miles. The Schoharie reservoir which is located farther north and on higher ground (1,130
ft.) was constructed between 1917 and 1927 and is considerably smaller than the Ashokan with a
water surface area of 1.8 square miles and a storage capacity of 22 billion gallons (Weidner,
1974). The first reservoir constructed as part of the Delaware water supply system was the
Neversink. The Neversink was completed in 1950 and is located to the southeast of the
Ashokan with a shoreline 17 miles long. The second reservoir the Rondout, was completed in
1951 and receives waters from the Neversink, Cannonsville and Pepacton, via the Delaware
aqueduct, making it the keystone to the Delaware supply system (Figure 2).
With the development of the Catskill/Delaware systems, New York City's drinking water
remains one of the cleanest in the nation. However, as a result of topographic constrains both
agricultural and urban land use is concentrated close to rivers and streams. While the human
population in the Catskill/Delaware watershed has increased only 17% in the past thirty years, the
continued effluent inputs from waste treatment plants, non-point agricultural, and urban runoff
increase the potential risk of contamination above acceptable water quality standards for drinking
and habitat survival. The economy of the area depends largely on seasonal tourism and dairy
farming (Stave, 1995), making fecal coliforms, nutrients and sedimentation levels a particular
concern (Wall et al., 1998; Hansan, 1996).
New York City Water Supply - Solutions
Currently the city is trying to decrease the threat to its quality of water through the development
and implementation of a long-range watershed protection program. The plan was signed in
1997 and unites efforts by the local communities, the city and state of New York, environmental
groups, and the EPA to preserve the high quality of New York City's drinking water supply
-------
50 0 50 Kilometers
Figure 1. The New York City water supply system including the Catskill/Delaware
and the Croton watersheds.
Basins
Counties
10 0 10 Kilometers
Figure 2. The six basins of the Catskill/Delaware watershed and surrounding counties.
-------
(Ashendorff et al., 1997, Okun et al., 1997). The plan includes upgrading current.sewage
treatment plants, implementing new watershed regulations, design and construction of a filtration
system for the 10% of water supplied by the Croton system, and acquisition of land deemed
critical to the preservation of high water quality in the Catskill/Delaware system.
Land acquisition was selected, partly to protect the environment, but also as a means to save the
taxpayers of New York state billions of dollars. Installing a filtration system for the city's water
supply would cost an estimated 2-8 billion dollars, versus the 250-300 million dollars set a side
for purchase of land (MOA, 1997, Featherstone, 1996). The majority (90%) of money spent for
land acquisition will be used in the Catskill/Delaware basins. The city has set as its goal, the
solicitation of 350 thousand acres of land (MOA, 1997). Under the agreement the State
Department of Environmental Conservation issued a permit to acquire land through outright
purchase and conservation easement. Priority is being given to purchase of undeveloped land
around reservoirs, streams and wetland areas. Secondary land acquisition will focus on the
purchase of sensitive lands surrounding streams and rivers. In conjunction with the watershed
protection program, the city of New York is working with the Watershed Agriculture Council
and the United States Department of Agriculture in developing a Conservation Reserve
Enhancement Program (CREP) to help protect, restore and manage these sensitive lands. The
program pays'farmers to remove sensitive lands from production and apply conservation
practices in place of agriculture. A total of 3,000 acres of highly credible land and 2,000 acres of
riparian buffer lands have been targeted for protection under CREP. The expected result of land
acquisition and conservation practices is the protection of over 165 stream miles, the preservation
of thousands of acres of natural areas, and continued high water quality without the cost of a $6
billion filtration system (Featherstone, 1996, Murphy etal., 1995).
TECHNICAL APPROACH
Human activity in a watershed can lead to a number of changes in the landscape including 1)
conversion of natural land cover to urban and agricultural, use 2) changes to soil infiltration,
compaction and erosion rates, and 3) alteration of hydrologic pathways. Forested riparian zones
act as a buffer, which help diminish excessive runoff of water and pollutants such as fertilizer,
pesticides, and other chemicals (Manlanson, 1993; Finlayson, 1991). When forests are
removed, the potential for flow of pollutants into streams is greatly increased. Fertilizers,
pesticides and other pollutants can be transported into streams that flow through or very close to
agricultural or urban land more easily than streams that flow through forest (Osborne and Wiley,
1988, Lowrance et al., 1984). Roads near streams Increase the amount of impervious surface
altering runoff rates (Forman and Alexander, 1998). Frequent driving can deposit small amounts
of gasoline, oil, antifreeze and other chemicals, which are washed away during precipitation
events. Salt applied during winter is another potential source of pollution where streams are in
close proximity to roads. Agricultural has the potential to increase soil erosion. When cropping
takes place on soils with a 3 % or greater slope there is an increased likelihood of erosion
(Lowrance, 1986). Studying the relationship between land use and water quality can help target
landscape proportions with a potential to either contribute or prevent undesired impacts. By
doing so the relationships between landscape proportion and water quality in the
Catskill/Delaware basins, the potential areas of risk to New York City's drinking-water supply
system can be evaluated. The databases generated from this study can then be used to prioritize
-------
and target basins, subbasins and near stream sites in need of preservation and restoration in the
Catskill/Delaware watersheds.
Site Characteristics
Physiography and Land Use
The 5000 square kilometer Catskill/Delaware watersheds are located in the southeast corner of
New York State, 160 kilometers northwest of New York City. The topography of the area is
diverse and except for the Adirondacks to the north has the greatest elevation in the State (Figure
3). The bedrock geology of the area is sedimentary, consisting of sandstone, shale and
conglomerates formed during the Devonian time period (Miller, 1970). The bedrock is exposed
or within one meter of the surface in about 30% of the watershed area. Of the remaining surface
geology almost 60% is glacial till (Figure 4). The watershed is bisected by numerous creeks,
streams and rivers including the Delaware, Schoharie, and the Neversink. The alluvial
deposition and out wash from these major rivers and their tributaries accounts for the remaining
10% of the watersheds surficial geology. The direction of hydrologic flow and drainage end
points (the six reservoirs) split the Catskill/Delaware watersheds into six basins .
Historically, the Catskill/Delaware watershed was dominated by northern hardwoods, including
maple, birch and beech trees. Other species present, but in smaller numbers were white/red
pine, hemlock, elm and ash (van Valkenburg, 1996). Much of the area was logged prior to the
mid-1800's. However, due to the need to eliminate a substantial debt owed by Ulster County,
lands currently owned by the county were conveyed back to the state in 1884. With the transfer
of ownership, 34,000 acres of forest was brought under state protection and the Catskill Park and
Preserve was born. In the decades since the Preserve's inception, the forest has rebounded from
its previous losses and is once again made up of a mixture of hardwoods, deciduous and
evergreens trees (van Valkenburg, 1996). Today, the Catskill Park covers 705,500 acres or more
than 40% of the Catskill/Delaware watershed. The watershed as a whole has benefitted by the
Park's presence and remains relatively undisturbed, having a forest cover near 90% in five of the
six basins as of 1991 (Figure 5). Together, the urban and agriculture land use makes up 10% of
the watershed. Nine percent can be classified as farming, pasture and recreational (i.e. golf and
lawns) and one percent as urban or commercial (Figure 6). The low urban and agricultural land
use in the watershed is related not only to presence of the Park and Forest Preserve, but also to
low population growth. From 1970 to 1995, population in the five counties containing the
Catskill/Delaware watershed increased by only 20%, shifting the population density from 150 to
180 people per square mile.
Data Collection and Analytical Methods
Landscape Classification
We will use land use data that is derived from satellite imagery for regional and larger areas .to
calculate most landscape metrics, and is generally derived from. In this study, land use will be
acquired or generated for four dates, ranging from 1973 to 1998. Landsat TM imagery will be
-------
Elevation (meters)
| |123-251
251 - 379
379 - 507
507 - 635
635 - 763
763 - 891
891 -1019
1019-1147
_J 1147-1276
Figure 3. Painted shaded relief showing elevational topography of the Catskill-Delaware watershed.
*** p* ;"T~:-^^
\J rf>:C"'
>y J^ w$ , ^
Surface Geology
~ Alluvium
I Water
[ Glacial till
Bedrock
Figure 4. Surface geology in the Catskill-Delaware watershed.
-------
Land Use Catagories
I water
urban
baren
! deciduous
evergreen\mixed
| pasture
f row crop 10
| urban grass
|^| wetland
I Catskill Park
Figure 5. Land use (1991) and Catskill Park boundary in the Catskill-Delaware watershed.
10 Kilometers
Land Use Catagories
IB water
I urban
~ barren
B forest
evergreenVnixed
pasture
iJ crop
~ urban grass
3 wetland
Streams
Jh
MM 0 500 M*Un
Figure 6. Land use (1991) near town of Trout Creek in Catskill-Delaware watershed.
-------
used for all dates except 1973, where multispectral scanner (MSS) from the North American
Landscape Characterization (NALC) program will be used. The study area is contained
completely within one scene, Landsat Worldwide Reference System 2 (WRS) path 14, row 31.
Like many long term landscape analyses, this study will require comparisons between images
having different resolutions. In order to better determine landscape change over a 25 year time
span further evaluation of the multi-date and multi-resolution data sets will be conducted using
paired Landsat TM and MSS imagery for one of the four selected image dates. Re-sampling
techniques such as nearest neighbor, bilinear and cubic convolution will be evaluated, as well as
re-sampling to 120, 60, and 30 meter resolutions. The effects of this re-sampling will be
demonstrated and evaluated to develop a standard method for measuring landscape change in a
multi-date, multi-resolution data set.
Coverages available from the New York Department of Environmental Protection for the study
area include basin and sub-basin boundaries, topography, soils, roads, streams (including lake
shorelines), and water sampling site locations. Land use data for 1989/91 from the Multi-
Resolution Land Characteristics program (MRLC), a consortium of four federal agencies, was
acquired from the EROS Data Center. This data set was generated using an unsupervised
classification technique. Clusters of spectrally similar areas were grouped and manually assigned
a land use class with the aid of National High Aerial Photography program (NHAP) aerial
photos. Areas that contained more than one land use class were labeled with the aid of ancillary
data such as population or elevation (EROS Data Center 1998). For use in this study the final
MRLC image classification will be aggregated to six classes for use in metric calculations (Table
1).
Land use will be derived from the three additional scenes (7/7/1973, 9/21/1984, and 7/26/98)
using supervised and unsupervised classification. Unsupervised analysis will include dividing
the image into 100 clusters of spectrally similar areas which will than be manually assigned a land
use class with the aid of National High Aerial Photography program (NHAP) aerial photos, and
yearly farm crop data collected by the Agricultural Farm Association of New York. Two ground
truthing site visits, one summer and one fall, will be conducted to verify classification. Areas
that contain more than one land use class will be labeled with the aid of ancillary data such as
population, elevation, and site survey. Training sites of known land use will be identified to
define spectral characteristics for each class. Training sites will be selected separately for each
image with the aid of aerial photography and other ancillary data. Each pixel in the image will
then be assigned to the class it most closely resembles using a maximum likelihood statistical
algorithm (Lee and Marsh, 1995; Lillesand and Kieffer, 1994). The classification will consist of
six categories: water, forest, barren, agriculture, urban (residential and commercial), and
cloud/shadow (Wickham and Norton, 1994). If possible, agriculture will be separated into row
crops and pasture.
-------
Table 1. Aggregation of MRLC land use classes.
MRLC Class
Open Water
Low Intensity Residential
High Intensity Residential
High Intensity Commercial/Industrial/Transportation
Pasture/Hay
Row Crops
Other Grasses (e.g. parks, lawns, golf courses)
Evergreen Forest
Deciduous Forest
Mixed Forest
Woody Wetlands
Emergent Herbaceous Wetlands
Quarries/Strip Mines/Gravel Pits
Transitional
Aggregated Class
Water
Urban
Agriculture
Forest
Barren
Landscape Metrics
Landscape metrics of characteristics believed to have a relationship to drinking water quality will
be calculated for six basins (average area 700 km2) and 79 sub-basins (average area 50 km2) in
the Delaware/Catskill watershed. Five metrics were chosen based on results of a previous study
in the Mid-Atlantic region; these metrics are proportion of the six land cover/use categories listed
above within the watershed, basins and sub-basins, proportion of land use in and near (100 meter
buffer) stream riparian zones, length of roads near streams, proportion of agriculture on steep
slopes, proportion of agriculture on highly erodible soils and Normalized Difference Vegetation
Index (NDVI) (Lyon et al. 1998, Jones et al. 1997).
The NDVI is a measure of relative greenness of an area (Lyon et al. 1998; Davenport and
Nicholson, 1993). NDVI will be calculated from the MSS and TM satellite spectral reflectance
data in the red and infrared wavelengths. NDVI values range between 0 and 1 with high values
indicating healthy vegetation and low values bare ground, pavement or water. Positive changes
in NDVI between dates will indicate a gain hi greenness, while negative changes will indicate a
loss. Difference of temporal satellite images usually have an approximately normal distribution.
We will use two standard deviations from the mean, which will capture 95% of the values
(McClave and Dietrich 1979), to detect change/no change in greenness.
Because the potential natural vegetation of the Catskill/Delaware watershed is almost entirely
forest (Kuchler, 1964), the proportion of other land use provides general information on the
-------
extent of disturbance in each sub-basin. This metric will be generated by overlaying the land use
with sub-basin boundaries. Proportions of each land use will then be calculated by sub-basin.
Water will be excluded from total sub-basin area when calculating proportions. Total human
use, or U-index (O'Neill, 1988), will be calculated by summing the proportions of agriculture,
urban, and non-natural barren (e.g., mines or gravel pits) land use by sub-basin.
Forested riparian zones provide a buffer, which help diminish excessive runoff of water and
pollutants such as fertilizer, pesticides, and other chemicals (Manlanson 1993, Finlayson 1991).
When forests are removed, the potential for flow of pollutants into streams is greatly increased.
Fertilizers, pesticides and other pollutants can be transported into streams that flow through or
very near agricultural or urban land uses more easily than streams that flow through forest
Osborne and Wiley, 1988; Lowrance et al., 1984). Two sets of metrics will be generated for this
group, the proportion of stream length adjacent to each land use type by sub-basin and the
proportion of each land use within 100 meters of streams by sub-basin. Both sets of metrics will
be calculated by overlaying land cover with streams (or buffered streams), and then overlaying
the result with sub-basin boundaries. Water will again be excluded when calculating
proportions.
Roads near streams can affect water quality in several ways (Forman and Alexander 1998). The
impervious surface increases the rate of water runoff, which can increase sediment loadings in
streams. Everyday driving can deposit small amounts of gasoline, oil, antifreeze and other
chemicals, which are washed away during precipitation events. Salt applied during winter is
another potential source of pollution where streams are in close proximity to roads. This metric
will be calculated by creating a 30-meter buffer around streams, and then overlaying with roads.
The metric will be reported as total length of roads in meters within the buffer per kilometer of
stream length by sub-basin.
Soil erosion can result in transport of nutrients and sediment to the surface water. Agricultural
practices have the potential to increase soil erosion. Erosion potential is related to the steepness
of slopes in agricultural use and soil type (Comeleo et al., 1996; Lowrance, 1986). The threshold
for increased erosion is a 3% slope (USDA 1951). Percent slope will be generated from 30-
meter Digital Elevation Model (DEM) data. This will be overlaid with land use to determine
overlap of steep slopes and agriculture to create an metric of the percentage of land with
agriculture (crop or pasture) on slopes greater than three percent by sub-basin.
The State Soil Geographic Soil Maps (STATSGO) and Data Base provides an credibility factor
(k-factor) which quantifies the susceptibility of soil particles to detachment and movement by
water under fallow field conditions (USDA 1994). Erodibility data will be aggregated following
the methods described in the STATSGO data use information report (USDA 1994) The
aggregated k-factors will then be overlaid with land use to determine overlap of highly erodible
soils and agriculture. The metric created will be a measure of the percentage of land with
agriculture (crop and pasture) on highly erodible soils.
10
-------
Water Quality
In order to describe the water quality in the Catskill/Delaware watershed data will be collected
from the 70's, 80's and 90's for streams and reservoirs (Figure 7). The main source for the
chemical and physical water quality parameters will come from the NYCDEP, NYSDEC, USGS
and EPA'sEMAPandthe storage and retrieval data base called STORET. STORETisa
repository for water parameter data for the contiguous United States. Many organizations
contribute to the data base including federal, state, interstate, universities, contractors, individuals
and water laboratories. The station operators include USGS , the regional and environmental
research laboratories from the EPA, and the NYSDEC. The NYSDEC will supply separate
databases containing 49 benthic sample locations with data, spanning a time period from 1984 to
1997 (Figure 8). Benthic data prior to 1984 will be collected from the NYSDEC 20-year trend
report published in 1993 (Bode et al., 1993). The NYCDEP has a large data base dating back to
1908 containing numerous water chemical, physical and biological parameter.
Compiling water quality data from a number of different sources can lead to problems of
consistency and comparability, due to differences in measurement and analysis methods. In the
course of our data compilation we anticipate specific problems will be encountered including:
1) differences in methodologies between agencies and with time, 2) changes in detection limits of
chemicals and metals over time, making comparison to earlier samples difficult, 3) sample sites
monitored for short time spans (1 to 3 years) rather than continuously, 4) some sites sampled
more frequently than others, 5) discontinuation of monitoring and initiation of new sites in
different location. Together these problems could result in insufficient temporal and spatial
sampling sites to fully characterize the relationships between changes in watershed metrics and
water quality.
For this study, we will use chemical, physical and biological water quality parameters from a
total of 248 water sample stations and 49 benthic sample stations located throughout the
Catskill/Delaware Watershed. Because climate influences both the physical and chemical
properties of surface water rainfall, snow pack and temperature data will be collected for the
surrounding area from stations monitored by the North East Regional Climate Center and
deposition data from five nearby National Atmospheric Deposition stations. The collection of
water quality data from these sample sites spans a 25 year time period (1973-1994). Most of the
data collected from the stations have already been published by the controlling agencies, either as
reports or journal articles.
Changes in the amounts of the chemical constituents of the surface water may indicate human
impacts or simply a shift in stream flow or water temperature (Gray, 1994). Nutrients such as
nitrogen and phosphorous are essential to the metabolic reactions of plants and animals and can
be growth limiting factors. However, when present in large quantities, nutrient inputs can result
in water eutrophication. While natural occurrences, such as fire, can cause an "algal bloom",
most eutrophication is the result of human induced enrichment, especially changes in
atmospheric deposition. When calcium availability from bedrock is limited and deposition is
high, nitrate concentrations, declining forests, and acidic surface water can be in-part related to
an overall low surface water calcium value. Similar to nitrogen and phosphorous, chloride and
silica both occur in nature but documented increases at a point location or across time may be
11
-------
10 0 10 Kilometers
Figure 7. STORET Water Quality Sites in the Catskill/Delaware watershed.
10 0 10 Kilometers
Figure 8. Benthic Sample Site Locations in the Catskill/Delaware watershed.
12
-------
tied to impacts from waste water treatment (Taras et al., 1971). Oxygen and pH levels are good
secondary metrics of human induced impacts. Most natural waters have pH levels between 4 and
9. Departure from normal pH levels could indicate acid or basic contributions by industrial
waste. Water physical parameters important to the evaluation of chemical changes are stream
flow, temperature and suspended sediments. These physical factors can influence oxygen levels,
nutrient loading, turbidity and habitat suitability. The primary concern of water pollution is its
effect on living organisms including humans. Aquatic organisms are often the first to experience
the impacts of changes in water quality and therefore can act as metrics of stress. Benthic
macro-invertebrates have proven especially valuable for this purpose (Abel, 1989; Hilsenhoff,
1982). Shifts in macro-invertebrate communities from pollution intolerant to pollution tolerant
species indicate a decrease in water quality (Rosenberg and Resh, 1993). Data to be used for
evaluating changes in water quality are species richness, diversity, ratios of sensitive to
nonsensitive species, total individual numbers and percent model affinity (compares known non-
impacted site communities to sample locations) (Novak and Bode, 1992). Other biological
parameters of concern are include changes in Fecal Coliforms counts resulting from the direct
discharge of human and animal sewage into the water. A high Fecal Coliforms count by itself is
not pathogenic but can be an indication of other disease causing organisms such as bacteria,
viruses and parasites.
Statistical Analysis
Water quality data will be divided into four time groups corresponding to available landscape
imagery, with four years of water quality data on either side of the imagery date. For example if
the land use image is from 1990, then water quality data from 1986 to 1994 will be used in the
statistical comparisons with the landscape metrics data. An analysis of variance (ANOVA) and
Tukey's HSD test of means will be conducted to determine significant differences (P^O.05)
between land cover-use and landscape metrics by basins (SAS 1990). Assuming that sufficient
data is available for analysis a 1) multi-variate and non-parametric analyses to examine yearly
and seasonal changes in water quality data; and 2) step-wise regression, correlations and cluster
analysis will be conducted to determine the extent of influence of the surrounding land-use and
landscape metrics on yearly and seasonal water quality parameters.
POTENTIAL FOR REDUCING UNCERTAINTY AND MEETING PROGRAM GOALS
Any task undertaken in the EPA has as one of its primary goals the reduction of uncertainty in
exposure assessment. The goals of this study align with the goals and objectives, of the 10 year
strategic plan for the landscape science program (LEB Strategic Plan), which in turn follow the
guidelines set forth by the NERL strategic Research Plan, the Strategic Plan for the Office of
Research and Development (ORD Strategic Plan) and the ORD Ecological Research Strategy.
All of these plans are driven by a need to respond to national goals such as the Governmental
Performance Results Act (GPRA) or to specific legislative mandates, like amendments to the
Clean Water Act (CWA). The sub-task set forth in this document complies with the goals of
GPRA, Objective 8.1.1, to improve the understanding of environmental risk and innovative
addressing of environmental problems, particularly in the area of ecosystem research, assessment
and restoration.
13
-------
To date, only limited analysis has been conducted in comparing landscape proportions to surface
water conditions in the Catskill/Delaware Watershed. The USGS recently published a report on
water quality in the Hudson River Basin (1992-1995) an area that overlaps part of this study site
(Wall et al. 1998). The study's main focus is water quality parameters obtained from throughout
the basin. Some attempt is made at discussing connections between results from the water
quality sampling and nearby land use. Similar comparisons between land use and water quality
were made in a 20 year study of benthic and chemical water parameters for the state of New York
by the NYSDEC (Hansen, 1996). The USGS is currently undertaking a study of the Delaware
River basin which may include a greater analysis of land use, but the project is only in its initial
stages and data has yet to be collected or analyzed.
The research done by the agencies collecting the water quality data is an important step in
understanding changes in the water quality with time. However, while landscape proportions are
often mentioned as a potential contributing factor to changes in water quality, only limited
attempts have been made at relating land use to water quality. One such study is being conducted
at Cornell University's School of Agriculture. This study focuses on modeling soil erosion
effects on phosphorous loadings within the Cannonsville Reservoir. The purpose of this sub-
task is to expand the on the previous set of studies by correlating changes in landscape proportion
and water quality over varying scales and over a 25 year time period.
Results will then be used to construct a set of landscape metrics, such as proportion of land
cover, topography and soils credibility, which correlate well with stream and lake condition in
the Catskill/Delaware basins. An evaluation of each metric can then be conducted at the basin,
subbasin and near stream scale. This diagnostic methodology can be used to develop necessary
protocols needed to maintain or restore and target sites within other New York watersheds and
in regions having similar landscape characteristics (i.e., Pennsylvania and New Jersey).
QUALITY ASSURANCE
The quality assurance (QA) for this project will include statistical and comparative procedures.
These will include a computer audit for all water quality and landscape data. Comparisons to
original data from contributing agencies will insure that no errors occurred during data transfer
either to STORET or to local computers stations. To test the validity of statistical inferences of
the fitted model we will check the model residuals for outliers, normality and homoscedasticity.
Methods such as Cook's D test and uni-variate analysis will be used to test whether the residuals
meets the basic assumptions (Madansky 1988). Statistical analysis of data will include a search
for outliers via descriptive statistics, uni-variate analysis, and residual comparisons with Cook's
D test. Accuracy assessment of the land cover will be performed at the Environmental
Photographic Interpretation Center (EPIC) in Reston, Virginia. High resolution imagery and
photographs of the study area will be used to generate a "truth" land cover. An error matrix will
be generated by comparing the "truth" to the derived land cover. In addition to these procedures,
an in-lab review of the project plan will be conducted as an initial screening for the project goals
and objectives. A second review by the Division Director will follow. Once the project plan has
obtained the necessary signatures it will be submitted to at least three outside EPA readers and
review by 05/99.
14
-------
Table 2. Work Breakdown Structure
Agency
USEPA/LEB
USEPA/LEB
USEPA/LEB
USEPA/LEB
Name
Megan
Mehaffey
Tim
Wade
Maliha
Nash
Curt
Edmonds
Work Assignment
Project Coordinator, Water Quality Data
Collection, QA, Statistical Analysis,
Writing
GIS Data Collection, Landscape
Coverages, Metrics, QA Writing
Statistical Analysis, QA,
Editing
Image Collection, Processing and QA
(supported by contract)
Time
Commitment
50%
20%
20%
20%
ANTICIPATED RESULTS AND PRODUCTS
1) This project will produce an expanded knowledge of the impacts of landscape and landscape
change on water quality. A set of landscape metrics will be evaluated for their importance to and
correlation with surface water conditions. These land cover-use and landscape metrics can then be
used to compare basins, subbasins and near stream sites relative to each other for the purpose of
targeting sites for maintenance or restoration in the Catskill/Delaware watersheds.
2) Water quality data will be compiled from various sources including STORET, USGS, EPA and
NYSDEC. The final database will contain site name, location, chemical, physical and benthic
values. The databases will be quality checked and then converted to a format for statistical and
ArcInfo/ArcView analysis.
3) The project will produce a land cover database spanning three decades with scenes from 1973,
1984,1987,1991, and!998. Geographic data to be collected includes NYDEP watershed and sub-
watershed boundaries, elevation (30 meter DEM), surface geology, aqueduct and tunnel locations,
streams (based on 1:24,000 scale USGS, improved by DEP), city and state owned lands, and roads.
The EPA, EROS data center and USGS will supply land cover/land use data derived from MRLC,
MSS (60 meter resolution)and landsat TM (30 meter resolution). In addition to the water quality and
coverage data bases, a set of four landscape metric groups will be developed (see above methods for
details).
4) Two journal articles are expected to be produced as a result of the project they are 1) Analysis of
Landscape and Water Quality in the New York Catskill-Delaware Watersheds and 2)Twenty-five
Years of Landscape Change in the Catskill/Delaware Basins.
15
-------
REFERENCES
Abel, P.D. 1989. Water Pollution Biology. Ellis Horwood, Chichester, England.
Ashendorff, A., M.A. Principe, J. Mantus 1997. Watershed protection for New York City's
supply. Journal of the American Water Works Association. 89:75
Bode, R.W., M.A. Novak, and L.E. Abele 1993. 20 Year Trends in Water Quality of Rivers and
Streams in New York State Based on Macroinvertebrate data 1972-1992. NYS Department of
Environmental Conservation, Division of Water.
Comeleo, R.L., J.F. Paul, P.V. August, J. Copeland, C. Baker, S.S. Hale, R. W. Latimer. 1996.
Relationships between watershed stressors and sediment contamination in Chesapeake Bay estuaries.
Landscape Ecology 11:307-319.
Davenport, M.L., S.E. Nicholson, 1993. On the relationship between rainfall and the normalized
difference vegetation index for diverse vegetation types in East Africa. International Journal of
Remote Sensing.
EROS Data Center. New York Land Cover Data Set. July 1998. Available:
ftp://edcftp.cr.usgs.gov/pub/edcuser/vogel/states/newyork_readme_ver9807.txt(ll/l/1998).ref).
Featherstone, J. 1996. Conservation in hte Delaware River Basin. Journal of the American
Water Works Association. 88:42.
Finlayson, C.M. 1991. Production and major nutrient composition of three grass species on the
Magela floodplain, Northern Territory, Australia. Aquatic Botany. 38:163-176.
Forman, R.T.T. and L.E. Alexander 1998. Roads and their major ecological effects. Annual Review
of Ecological Systems. 29:207-231.
Graham, R.L., C.T. Hunsaker, R.V. O'Neill, and B.L. Jackson 1991. Ecological risk assessment at
the regional scale. Ecological applications 1:196-206.
Gray, N.F. 1994. Drinking Water Quality: Problems and Solutions. John Wiley & Sons, West
Sussex, England.
Hansen, G. 1996. New York State Water Quality 1996. NYS Department of Environmental
Conservation, Division of Water.
Hecker. G. 1991. The New York City Water Supply: Past, Present & Future. Civil Engineering
Practice: Journal of the Bos. 6:7
Hilsenhoff, W.L. 1982. Using abiotic index to evaluate water quality in streams. Technical Bulletin
No. 132. Department of Natural Resources.
16
-------
Hunsaker, C.T. and D. A. Levine 1995. Hierarchical approaches to the study of water quality in
rivers, BioScience 45:193-203.
Jones, K.B., Riiters, K.H., Wickham, J.D., Tankersly, R.D., O'Neill, R.V., Chaloud, D.J., Smith,
E.R., and Neale, A.C. 1997. An Ecological Assessment of the United States Mid-Atlantic Region.
EPA/600/R-97/130, U.S. Environmental Protection Agency. Office of Research and Development,
Washington DC.
Kuchler, A.W. 1964. Potential Natural Vegetation of the Coterminous United States, Map and
Manual. American Geographical Society, Special Publication 36, New York.
Lee, C.T. and S.E. Marsh. 1995. The use of archival Landsat MSS and ancillary data in a GIS
environment to map historical change in an urban riparian habitat. Photogrammetric Engineering
and Remote Sensing.
Lillesand, T.M. and R.W. Kieffer 1994. Remote Sensing and Image Interpretation. 3rd Ed. John
Wiley and Sons, Inc., New York.
Lowrance, R., J.K. Sharpe and J.M. Sheridan 1986. Long-term sediment deposition in the riparian
zone of a coastal plain watershed. Journal of Soil and Water Conservation. 41:266-271.
Lowrance, R., R. Todd, J. Fail Jr., O. Hendrickson, Jr., RXeonard and L. Asmussen 1984. Riparian
forest as nutrient filters in agricultural watersheds. BioScience 34:374-377.
Lyons, J.G., D. Yuan, R.S. Lunetta, and C.D. Elvidgel998. A change detection experiment using
vegetation indices. Photogrammetric Engineering and Remote Sensing
Malanson, G.P. 1993. Riparian Landscapes. Cambridge University Press, Cambridge, Great Britain.
Madansky, A. 1988. Prescriptions for Working Statistiticians. Springer-Verlag. New York, New
York.
McClave, J.T. and F.H. Dietrich, 1979. Statistics. p31, Dellen Publishing Co. San Francisco,
California.
o
Miller, W.J. 1970. The geological history of New York State. Kennilcat Press. Port Washington,
New York.
MO A, 1997. Watershed Memorandum of Agreement. NYCDEP Office of Watershed
Communications, Kingston, NY
Murphy, S. 1995. Land acquisition for Water Quality Protection: New York City and the
Catskills Watershed System. Water Resources Update. Sumr:60.
Novak, M.A., and R.W. Bode 1992. Percent model affinity: a new measure of macroinvertebrate
17
-------
community composition. J.N. Am. Benthol. Soc. 11:80-85.
Okun, D.A., G.F. Craun, J.B. Rose 1997. New York City: To Filter or not to Filter? Journal of
the American Water Works Association. 89:62.
O'Neill, R.V., J.R. Krummel, R.H. Gardner, G. Sugihara, B. Jackson, D.L. DeAngelis, B.T. Milne,
M.G. Turner, B. Zygmunt, S.W. Christensen, V.H. Dale, and R.L. Graham. 1988. Indices of
landscape pattern. Landscape Ecology 1:153-162.
O'Neill, R.V., B.T. Milne, M.G. Turner, and R.H. Gardner 1988. Resource utilization scales and
landscape pattern. Landscape Ecology 2:63-69.
O'Neill, R.V., C.T. Hunsaker, K.B. Jones, K.H.. Riitters, J.D. Wickham, P.M. Schwartz, LA.
Goodman, B.L. Jackson, and W. S. Baillargeon 1997. Monitoring environmental quality at the
landscape scale. BioScience 47:513-519.
Osborne, L.L. and M.J. Wiley 1988. Empirical relationships between land use/cover and stream
water quality in an agricultural watershed. Journal of Environmental Management 26:9-27.
Rosenberg, D.M. and V. H. Resh 1993. FreshwaterBiomonitoringandBenthicMacroinvertebrates.
Chapman and Hall, New York, NY, USA.
S AS 1990. SAS/SA T User's Guide, Version 6, Fourth Edition, Vol. 2, S AS Institute Inc., Cary, North
Carolina, USA.
Stave, K.A. 1995. Resource conflict in New York City's Catskill watersheds: A case for expanding
the scope of water resource management. p61-68 in: Austin, L.H. (editor, 1995. Water in the 21st
Century: Conservation, Demand and Supply. Proceedings of the AWRA Annual Spring Symposium.
American Water Resources Association, Herndon, Virginia, TPS-95-1, 740 pages.
Taras, M.J., A.E. Greenberg, R.D. Hoak and M.C. Rand 1971 . Standard Methods for the
Examination of Water and Waste-water, 13th Edition. American Public Health Association,
Washington, D.C. USA
United States Department of Agriculture 1994. State Soil Geographic (STATSGO) Data Base: Data
Use Information. U.S. Department of Agriculture Publication #1492.
United States Department of Agriculture 1951. Soil Survey Manual. Agricultural Handbook 18. U.S.
Department of Agriculture, Washington DC
van Valkenburg, N. J. 1996. The Forest Preserve of New York State in the Adirondack and Catskill
Mountains: A Short History, Purple Mountain Press, NY, NY
Wall, G.R., K. Riva-Murray, and P.J. Phillips 1998. Water quality in the Hudson river basin, New
York and adjacent states, 1992-95. U.S. Geological Survey Circular 1165.
18
-------
Weidner, C.H. 1974. Water For a City. Rutgers University, Quinn & Boden Comp, Inc , New
Jersey.
Wickham, J.D. and D.J. Norton, 1994. Mapping and analyzing landscape patterns. Landscape
Ecology. 9:7-23.
Wiens, J.A., and B.T. Milne 1989. Scaling of 'landscapes' in landscape ecology, or landscape
ecology from a beetle's perspective. Landscape Ecology 3:87-96.
19
------- |