ArticlePDF Available

Applying the EPA's Regional Vulnerability Assessment (ReVA) Approach to the Pennypack Creek Watershed

Authors:

Figures

Content may be subject to copyright.
Applying the EPA’s Regional Vulnerability
Assessment (ReVA) Approach to the
Pennypack Creek Watershed
Final Report
Edited by
John A. Sorrentino
Jeffrey Featherstone
Md Mahbubur R Meenar
The Center for Sustainable Communities
Temple University Ambler College
January, 2007
Applying the EPA’s Regional Vulnerability Assessment (ReVA) Approach to the
Pennypack Creek Watershed
Table of Contents
List of Tables ………………………………………………………………………………….…..……. 3
List of Figures ………………………………………………………………………………….….…… 4
Executive Summary …………….………………………………………………………….……….... 5
1. Introduction: Objectives of the Study ………………………………….……………….….…… 6
2. Applications of ReVA Methodology to the Pennypack Creek Watershed (PCW) ….….. 7
2.1. Developing a GIS-Based Watershed Data Inventory ……………………….….... 9
2.1.1. Natural Resources Inventory
2.1.2. Built Environment Inventory
2.2. The State of the Watershed ……………………………………………. ..…………. 13
2.3.Ecological Indicators in the PCW …………………………………………………… 20
2.3.1. Water Volume
2.3.2. Water Quality
2.3.3. Biological Integrity
2.3.4. Impervious Surface
2.4. Two Land-Use Scenarios …………………………………………………….…….… 46
2.4.1. Trend Development
2.4.2. Smart Growth
2.5. Differential Impacts of Land-Use Scenarios ………………………………………. 55
2.5.1. Water Volume
2.5.2. Energy Use, Air and Greenhouse Gas Emissions, Water Quality
and Biological Integrity
2.6. Pennypack Ecological Vulnerability Assessment (PEVA) ……………………… 68
2.7. Relation of PEVA to Pennsylvania’s Sustainability Indicators ………………… 72
3. Potential Applications to Other Watersheds ……………………………………………….… 73
3.1 Data & Metadata
3.2 Models
3.3 Scenario Generation
3.4 Impacts Assessment
4. The Involvement of Local Officials ……………………………………………………..……….. 80
5. Conclusions …………………………………………………………………………….…………… 80
References ……………………………………………………………………………………………… 82
Appendices …………………………………………………………………………...………………… 86
A.1. ReVA Web Page Summaries …………………………… …………………………. 86
A.1.1. Data Preparation
A.1.2. Examples of Variables Used to Estimate Vulnerability
A.1.3. Assessment Questions
A.1.4. Future Vulnerability
A.1.5. Multiple Criteria Decision-making
A.1.6. Demonstrations
A.2. Data & Metadata: Technical Details & Tables ……………………………………. 101
A.2.1. List of Shape Files
A.2.2. Shape File Metadata
A.2.3. List of Databases
A.2.3. DBF File Metadata
A.2.5. Attachment: Data CD
A.3. Ecological Indicators: Technical Details & Tables ……………………………… 167
A.3.1. Graphs and Tables for the Hydrological Modeling
A.3.2. Calculations for the Water Volume Indicator
A.3.3. Calculations for the Water Quality Indicator
A.3.4. Calculations for the Biological Integrity Indicator
A.3.5. Calculations for the Impervious Surface Indicator
2
List of Tables
Table 2.1.1. A List of GIS data layers
Table 2.2.1. Sub-basin Properties
Table 2.2.2. Reach Properties
Table 2.2.3. Rainfall Events Used for Calibration
Table 2.3.1. Interpreting Biological Integrity Scores
Table 2.3.2. Ranges of the IBI Score
Table 2.4.1. Forecasted Population Growth and Housing Unit Construction Needed
Table 2.4.2. Land Development Rates: Trend Development Scenario
Table 2.4.3. Land Development, Municipal Smart Growth Scenario
Table 2.4.4. Regional Smart Growth Scenario
Table 2.5.1. Peak Flows (cfs) for Varying Conditions
Table 2.5.2. Relative Importance of Suitability Criteria
Table 2.5.3. Fuel Economy & Emissions Rates of Baseline Vehicles
Table 2.5.4.Energy Use & Emissions from a Typical Automobile
Table 2.5.5. Energy Use & Emissions in Sprawl & Smart Growth
Table 2.5.6. Water Quality Index
Table 2.5.7. Biological Integrity
Table 2.5.8. Value of Ecosystem Functions/Services
Table 2.5.9. Impacts Summary
Table 2.7.1. PA Sustainability Indicators
Table 4.1. Municipalities Involved in PCW Stormwater Management
3
List of Figures
Figure 2.1.1. Clipping with Specified Buffer Distance
Figure 2.2.1. PCW Land-Use, 2000
Figure 2.2.2. The PCW Hydrology
Figure 2.3.1. PCW Baseflow Geology
Figure 2.3.2. PCW Drainage to the Rhawn St. Gauge
Figure 2.3.3. PCW Upstream of the UM-H STP
Figure 2.3.4. Peak Flows under Current Conditions
Figure 2.3.5. PWD Stations & Associated Sub-Basins: Chemical Data
Figure 2.3.6. %-Alkalinity: Sub-Basin Classification
Figure 2.3.7. %-Fecal Coliform: Sub-Basin Classification
Figure 2.3.8. %-Nitrate: Sub-Basin Classification
Figure 2.3.9. %-Phosphate: Sub-Basin Classification
Figure 2.3.10. %-Suspended Solids: Sub-Basin Classification
Figure 2.3.11. PWD Stations & Associated Sub-Basins: Macroinvertebrate Data
Figure 2.3.12. PWD Stations & Associated Sub-Basins: Fish Data
Figure 2.3.13. Biological Integrity for Macroinvertebrates: Sub-Basin Classification
Figure 2.3.14. Biological Integrity for Fish: Sub-Basin Classification
Figure 2.3.15. %-Impervious Surface: Sub-Basin Classification
Figure 2.4.1. Trend Development Land Use, 2030
Figure 2.4.2. Municipal Smart Growth Land Use, 2030
Figure 2.5.1. Flows from Trend Development
Figure 2.5.2. Sites for Smart Growth: Darker Is More Suitable
Figure 2.5.3. Sites for Sprawl: All but Dark Sites Acceptable
Figure 3.1 Generic Ecological Vulnerability Assessment Workflow
4
Executive Summary
The Center for Sustainable Communities (CSC) at Temple University was
engaged by the Pennsylvania Department of Environmental Protection to
determine whether the ecological modeling approach known as ReVA (Regional
Vulnerability Assessment) developed by the Environmental Protection Agency for
large ecosystems or river basins could be applied at a much smaller scale for
watersheds in Pennsylvania. The CSC chose the 57 square-mile Pennypack
Creek Watershed (PCW) in southeastern Pennsylvania for this application. The
CSC engaged a study team of scientists and researchers from universities,
nonprofit organizations, and consulting firms to conduct this analysis.
This report presents the results of this work. It includes a detailed assessment of
the ReVA methodology and its applicability to the PCW, the preparation of a
comprehensive Geographic Information System (GIS) database, the delineation
of ecological indicators, the preparation of land-use scenarios, and the evaluation
of the differential impacts of those scenarios on various watershed attributes.
The study team has concluded that the ReVA modeling approach can be scaled
down so as to be informative and appropriate for smaller watershed assessments
in the Commonwealth of Pennsylvania. While many larger-scale ReVA
assessment tools and data variables are too coarse for assessment at the local
level, it is feasible to use such processes with more refined local data to
accurately outline the impacts of alternative land-use and resource-allocation
decisions on ecological and other watershed attributes.
The approach developed for the PCW also can be used as a template for other
watersheds. While the PCW is a seriously impaired watershed in a dense urban
setting in the Greater Philadelphia region, the assessment protocol outlined in
this study can be accurately applied for less developed and more pristine
watersheds as well. The report outlines data needs and analytical tools for this
information transfer.
Finally, the ReVA modeling process also can be used to develop watershed
sustainability indicators. The study team outlines four broad indicator categories
that can be applied elsewhere and aggregated to larger regions in Pennsylvania.
While the lack of good time series information makes ecological evaluations
difficult, better models and more accurate digital information on topography and
hydrology allow researchers to more accurately assess many watershed
attributes.
5
1. Introduction: Objectives of the Study
Ecological assessment is a subject of paramount importance as the nation
moves to prevent further degradation of the environment. While many
environmental problems are global or regional in nature, the local level
affects them, and is affected by them. The US Environmental Protection
Agency (EPA) has undertaken regional and national assessments over the
past few decades. The Environmental Monitoring and Assessment Program
(EMAP) is a long-term research effort to enable status and trend
assessments of aquatic ecosystems across the U.S. The Mid-Atlantic
Integrated Assessment (MAIA) project incorporates data from state,
regional and national environmental monitoring programs into regional
assessments for the Mid-Atlantic region. An outgrowth of MAIA is the
Regional Vulnerability Assessment (ReVA) Program. The evolution of this
monitoring and assessment is what the current research has as its focus.
This project was a collaboration among individuals from the federal, state and
local levels of government, local universities, and non-government organizations
(NGOs). The expertise of the research participants spanned the natural and
social sciences. The list of participants is:
Temple University Faculty
A.S.M. Abdul Bari, Center for Sustainable Communities (CSC)
Michel Boufadel, Civil and Environmental Engineering
Jeffrey Featherstone, CSC
Shirley Loveless, CSC
Md Mahbubur R Meenar, CSC
Richard Nalbandian, CSC
Jon Nyquest, Geology
Kurt Paulsen, Community & Regional Planning
John Sorrentino, Economics
Susan Spinella, CSC
Laura Toran, Geology
Temple University Graduate Student Research Assistants (GSRA)
Andreea Ambrus
Dennis Dalbey
Jesse Sherry
Ibrahim Ibrahim
Straso Jovanovski
Melanie Martyn
Grisselle Rodriguez-Herrera
Lilantha Tennekoon
6
Faculty from Other Universities
Kathi Beratan, Duke University
Amy Liu, West Chester University
Peter Petraitis, University of Pennsylvania
Philadelphia Water Department
Jason Cruz, Office of Watersheds
Joanne Dahme, Office of Watersheds
Howard Neukrug, Office of Watersheds
Others
Sean Greene, F.X. Browne, Inc.
David Hart, Acadamy of Natural Sciences
David Robertson, Pennypack Ecological Restoration Trust (PERT)
Puneet Srivastava, Academy of Natural Sciences
Shandor Szalay, F.X. Browne, Inc.
Jim Thorne, Natural Lands Trust
Project Advisors
Donald Brown, Pennsylvania Dept. of Environmental Protection (PA DEP)
Libby Dodson, PA DEP
Deborah Forman, Environmental Protection Agency, Region III
Stanley Laskowski, EPA, Region III
Richard Paste, EPA, Region III
Anita Street, EPA, Headquarters
Betsy Smith, EPA, Research Triangle Park
The objectives of the project were clearly stated at the outset: (1) Determine
whether the ReVA modeling approach can be scaled down so as to be
informative and appropriate for watershed assessment in the Commonwealth of
Pennsylvania; (2) Determine whether the assessment model developed for the
Pennypack can be used as a "template" for assessments in other watersheds in
Pennsylvania and elsewhere; and (3) Determine whether the ReVA modeling
process can be used to develop watershed sustainability goals and indicators.
The Pennypack Creek Watershed (PCW) was used to assess the applicability of
ReVA methods.
Section 2. describes the PCW and applies ReVA methods to some aspects of
the watershed’s ecology. The reader interested in a more detailed account of
ReVA methodology is invited to visit Appendix A.1. and the references cited
there. The methods include data acquisition and generation, metadata creation,
indicator development, scenario generation and impact assessment. These
7
ultimately lead to vulnerability assessment. Section 3. discusses what aspects of
the methodology of Section 3. can and cannot be directly applied to other small
watersheds. Section 4. contains a brief discussion of the potential for informing
and organizing officials from neighboring municipalities with the purpose of
inducing watershed-wide cooperation in avoiding ecological vulnerabilities. Some
Conclusions follow section 4., and References direct the reader to sources of
further information. The Appendices contain more detail on some of the topics in
the text.
2. Applications of ReVA Methodology to the Pennypack Creek Watershed
As noted above, the ReVA program evolved from two EPA programs that began
earlier: EMAP & one of its sub-programs, MAIA. A concise overview of ReVA is
given in the words of the EPA:
The ReVA Program focuses on regional scale integrated assessment with the aim
of assisting decision makers in identifying and locating both environmental
resources and the conditions that are stressing those resources. ReVA
strengthens the decision-making process by identifying the current status and
relationships between stressors and sensitive environmental resources and
estimating the environmental changes that follow from specific actions. (p.1, US
EPA 2006b)
The PCW research team has examined these data and methods with the
charge of determining what aspects of EMAP/MAIA/ReVA methodology are
relevant to watersheds smaller than the USGS hydrologic accounting units
in the Hydrologic Unit Code (HUC) used by the MAIA research team.
(Jones et al. 1997) The application of ReVA methodology in this section will
essentially follow the ReVA assessment questions: (1) What is the overall
condition of the region? (2) What is the relative environmental condition
given all variables or a subset (e.g. those related to water quality)? (3) What
and where are the current most pressing environmental risks for a region?
(4) What and where is the greatest risk in the future likely to be? (5) Where
are the strategic planning or restoration priorities for a region? (US EPA
2006b)
Questions (1) and (2) are answered in the description in Section 2.2. and the
relatively exhaustive data accumulation recounted in Section 2.1. and Appendix
A.2. The answer to Question (3) took the form of assessing the relative values of
indicators measured in Pennypack Creek as described in Section 2.3. The
diagnosis is that certain stressors have put considerable pressure on the Creek’s
resources with the result that the Creek is designated “moderately impaired.” The
future risks referred to in Question (4) took the form of hydrological impacts
resulting from the residential growth scenarios in Section 2.4, and other
environmental impacts discussed in Section 2.5 based on alternative placement
scenarios. Question (5) is dealt with in Section 2.5 and in the Conclusion.
8
2.1. Developing a GIS-Based Watershed Data Inventory
Watershed management and vulnerability assessment require an
interdisciplinary approach to a complex problem. Comprehensive Geographic
Information Systems (GIS) inventories of the natural and built environment
provide watershed managers with the data, tools, and techniques to manage the
complexities. The tools of GIS and spatial analysis allow decision-makers and
citizens to understand and visualize the many features of a watershed, from land-
use patterns to species diversity to flood-hazard areas. The powerful spatial
analytic features of the ArcGIS1 system, combined with increasingly available
high-quality digital data should prove of great benefit to all concerned with
managing the complex ecological, economic, and political systems involved in a
watershed.
The Center for Sustainable Communities (CSC) has developed a comprehensive
inventory of the PCW's natural and built environment. ArcGIS served as the
primary platform for data collection and analysis from a multidisciplinary team of
researchers, including urban planners, landscape architects, geologists, civil
engineers, economists and biologists. Data layers covering the physical
(geology, soil, slope), biological (fish, insect populations), chemical (pollutant
loads, dissolved oxygen), hydrologic (rainfall, runoff, stream flow), demographic,
and land-cover/land-use features were collected for the entire watershed. Table
2.1.1. shows a list of data layers generated for this project.
In order to undertake a more refined assessment of the 56-square-mile PCW, the
study team subdivided the PCW into 49 smaller sub-watersheds or sub-basins,
which would correspond to the size and location of the first-order streams within
its boundaries. Sub-basins were delineated from stream line files based on
stream order and topographic elevation data using the Watershed Modeling
System (WMS) 7.1 and HEC-GeoRAS software.2
In defining sub-basins, the study team sought to create them to allow
aggregation and dis-aggregation from current and past studies. Originally, the
watershed was divided into 10 sub-basins to correspond with the CSC's
hydrologic and hydraulic modeling studies. Those 10 sub-basins provided the
basis for preparation of detailed maps of the watershed's floodways, and 100-
year and 500-year floodplains. Those 10 were further sub-divided into 20
watersheds to correspond with the Philadelphia Water Department's (PWD) 20
water-quality sampling sites. In order to improve the potential richness of the
1 ArcGIS is a GIS software package developed by the Environmental Systems
Research Institute (ESRI) Inc.
2 HEC-GeoRas is an ArcView 3.2 (ESRI software) extension developed by the
U.S. Army Corps of Engineers' Hydrologic Engineering Center (HEC). It prepares
GIS data for import into the HEC River Analysis System (HEC-RAS) and
generates GIS data from RAS output.
9
ecological assessment, the basins were ultimately sub-divided into the 49
smaller, first-order sub-watersheds.
2.1.1. Natural Resources Inventory
Biological data for the watershed were collected from monitoring stations of the
PWD. Data on fish species, insect habitat and macroinvertbrates provided
indicators of water quality and the suitability of the streams and riparian areas to
support a wide variety of species. Sampling data were available at 20 locations
within the watershed. ArcGIS was used to assign those sampling points to the
sub-basin(s) draining to the sampling point. The spatial tools within ArcGIS thus
allow statistical analysis of the relationships between various land-use patterns
and water quality or biological integrity.
Table 2.1.1. A List of GIS data layers
Data Source Year
Biological Data
Fish Philadelphia Water Department (PWD) 2002
Habitat PWD 2002
Macroinvertebrate PWD 2002
Water Related Data
Wetland Delaware Valley Regional Planning Commission
(DVRPC) 1981
Bridge & Culvert Center for Sustainable Communities, Temple
University(CSC) 2005
Dam PWD 1999
Riparian Buffer Heritage Conservancy 2002
Effluent
Concentration PWD 2003
Discharges &
Withdrawals Delaware River Basin Commission (DRBC) 1996
Stream Center for Sustainable Communities, Temple
University(CSC) 2004
Floodplain Federal Emergency Management Agency (FEMA) 1996
Geological Data
Bed Rock Geology DRBC 1998
Soil DRBC Unknown
Base Flow PWD 1998
Demographic Data
Household Density US Census Bureau 1990/2000
Median Household
Income US Census Bureau 1990/2000
Population Density US Census Bureau 1990/2000
Land Features Data
Land use DVRPC 1990/1995/20
00
Land Cover United States Geological Survey (USGS) 2001
Tree Canopy Density USGS 2001
Impervious Surface Pennsylvania State University, Dr. Toby Carlson 1985/2000
10
Slope CSC 2004
Road Density CSC 2005
Forest Fragmentation CSC 2005
An essential feature of the field of "landscape ecology" and the EPA's ReVA
analysis is that landscape patterns – particularly human-influenced landscape
change – affect ecological processes. The ArcGIS tools, combined with freely
available specialized extensions, allows for a detailed understanding of the effect
of landscape change on ecological integrity.
Hawthorne's Analysis Tools for ArcGIS (www.spatialecology.com) is one of a
number of free extensions for the ArcGIS 8 and 9 systems with specialized tools
for landscape measurement. The "Count Points in Polygon" tool, for example,
was used to calculate the number of bridges, culverts, dams, and discharge
points within each sub-basin. The tool, "Sum Line Lengths in Polygons," was
used to calculate the length of roads within buffer-distances from streams (30
and 100 feet), and to calculate the extent of impaired riparian buffers along each
segment of the waterway. Those data can assist in prioritizing stream segments
for mitigation or restoration efforts, as well as indicating the impact of stream
impairments on water quality. The Model Builder feature in ArcGIS was used to
perform many of the repetitive geo-processing steps. Figure 2.1.1. shows an
example of a Model, clipping with buffer distance.
Figure 2.1.1. Clipping with Specified Buffer Distance
11
As part of the hydrologic modeling of the watershed, Curve Numbers (CNs) were
calculated for each sub-basin of the watershed. The CN, developed by US
Department of Agriculture Natural Resource Conservation Service (formerly the
Soil Conservation Service), is a measure of the stormwater runoff potential for a
drainage area. Calculation of the CNs involved using land-use class data, as well
as data in hydrologic soil groups in the ArcCN-Runoff extension. CNs were
calculated for each distinct land-user/soil group type, and aggregated to produce
a composite CN for each sub-basin.
One of the purposes of a comprehensive watershed inventory is flood hazard
mitigation. ArcGIS tools combined with high-resolution digital ortho-photography
provide an important tool in hazard mitigation planning. The Q3 Flood Data from
the Flood Insurance Rate Maps (FIRMS) were collected from FEMA in digital
form. Building footprints for all buildings in the watershed were digitized from
high-resolution (1.5 sq.ft. resolution) digital ortho-photography provided by the
DVRPC. When flood-plain and building-footprint layers were overlain in ArcGIS,
software tools were used to count the number of buildings in the floodplains.
2.1.2. Built Environment Inventory
In order to understand the human influences on watershed, demographic and
land-use data were collected. Demographic data were collected at the Census
Block Group level for 1990 and 2000 from the US Census Bureau website.
Census TIGER/Line boundary files for 1990 and 2000 were also downloaded
from the Census to identify block groups located in the PCW. Data on the
population, number of housing units, median income, population density, and
housing-unit density were summarized for each of the 49 sub-basins. Data show
wide variation in new housing construction across the sub-basins. Demographic
data were combined with land-use change data to produce scenarios of future
land use.
A number of data sources were combined to produce a detailed picture of the
land uses and landscapes within the watershed. The 2001 National Land Cover
Database is a consistent land-cover database for the entire continental US at a
30m resolution, and was downloaded from seamless.usgs.gov. In addition, the
National Land Cover Database provides tree-canopy and impervious-surface
data at the same resolution. For Pennsylvania, higher-resolution impervious-
surface data was estimated by Dr. Toby Carlson of Penn State University and is
available for 1985 and 2000 to show changes in impervious surfaces within a
watershed. The DVRPC made available digital land-use data interpreted from
high-resolution digital ortho-photography. Data from 1990, 1995 and 2000 allow
for an assessment of land-use patterns and land-use change. Although few
areas of the country currently have available consistently-interpreted, high-
resolution land-use data from more than one time period, these data are
becoming increasingly available.
12
ArcGIS tools combined with high-quality land-use and land-cover data from
multiple sources allow for a detailed assessment of landscape patterns. Forests,
for example, play an important role in both species habitat and water quality.
Detailed measures of forest fragmentation can be used to assess ecosystem
threats and prioritize land conservation strategies. Using an ArcGIS extension
written by Kurt Paulsen (author of Section 3.4) and a collaborator, measures of
forest fragmentation are calculated for each sub-basin. Hawthorne's Tool was
used to calculate distances between forest patches, and the proximity of forested
patches to streams.
The next section puts these data and the ArcGIS tools to work in discussing the
hydrology and ecological indicators for the PCW.
2.2. The State of the Watershed
The PCW is a 56 square mile area located in southeastern Pennsylvania that
contains a population of about 300,000. It contains parts of Bucks, Montgomery
and Philadelphia Counties. Besides a section of Northeast Philadelphia, the
watershed resides in all or part of 11 municipalities: Abington, Bryn Athyn,
Hatboro, Horsham, Jenkintown, Lower Moreland, Rockledge, Upper
Southampton, Upper Dublin, Upper Moreland and Warminster. The topography
of the watershed is characterized by gently rolling hills in the headwaters,
moderately sloping valley in the central part of the watershed, and tidal flats
draining to the Delaware River. The elevation over the watershed ranges from
436 feet to less than 10 feet above sea level. The flow regimen in the Creek and
the interrelationships between surface and groundwater within its watershed are
complicated not only by development and other human activities within the basin,
but also by its complex environmental character. In particular, the bedrock
geology is highly diverse and the geologic history spans more than 600 million
years. There are great differences in the physical characteristics of the many
different rock types within the watershed. The Pennypack Creek system contains
roughly 79 miles of surface waters and is classified for the following uses: warm
water fishery, trout stocking fishery, aquatic life, water supply and recreation.
The climate of the region is characterized by warm summers and cold winters
with moderate intermediate seasons. The mean annual temperature is 54oF and
the average annual precipitation is 41.41 inches. (Meenar 2006)
Much of the PCW area was developed as a part of the “inner ring suburbs” of
Philadelphia from the 1950s to the 1980s. As can be seen in Figure 2.2.1., the
area is quite built up and hosts a myriad of land uses that impact energy use and
the air, water and biological integrity.
The water quality in the PCW is significantly influenced by the large Upper
Moreland-Hatboro Sewage Treatment Plant (UM-H STP). Its location is shown in
Figure 2.3.3. The UM-H STP is authorized by the PA DEP through an NPDES
permit and DRBC to discharge treated sewage at an average annual rate of
13
7.173 mgd and a maximum monthly flow of 9.08 mgd. According to the permits,
the STP is required to provide advanced secondary treatment and ultraviolet
disinfection prior to discharge to the Pennypack Creek. Discharge monitoring
reports (DMRs) provided by the plant operator to the DEP indicate that the STP
operates at or near its maximum design capacity. The service area of the STP
includes the Borough of Hatboro and portions of Horsham, Upper Dublin and
Upper Moreland Townships, Montgomery County, PA and a portion of
Warminster Township, Bucks County, PA. The UM-H STP is one of four point
sources of discharge in the watershed. The remaining three are small “package”
plants that do not have significant adverse impacts on water quality.
Air quality is dependent on the mix of land uses in the PCW. Except for Vacant,
Water, & Wooded, the land uses listed in the figure legend entail the use of
human-processed energy & materials for everyday functioning. Where electricity
in not the end-use energy source, fuels are generally burned in stationary or
mobile sources to provide goods and services while generating entropy and
waste materials. Ambient air quality over the PCW is generally good, and
affected mostly by mobile-source emissions.
Figure 2.2.1. PCW Land-Use, 2000
14
2.2.1. Hydrology
The interactions and relationships among the components of the hydrologic cycle
(precipitation, evapo-transpiration, runoff, etc) must be understood in order to
predict a stream’s responses to changes in any of those components, whether
natural or manmade. Rigorous hydrologic modeling was performed in the PCW
to gain such an understanding of the watershed’s hydrologic regimen.
It has been long and widely known that the changes in land cover that
accompany urbanization generally result in decreases in evapo-transpiration and
infiltration (and consequently in baseflow), and increases in surface runoff, (US
EPA 1993). What has also been long, but perhaps less widely realized, is that
the most effective or dominant channel-forming flow of a stream, i.e., that which
transport the largest total amount of sediment over a period of years, is the
bankfull stage (Walman and Miller 1960). This most effective of flows has a
recurrence interval of 1.5 years in the annual maximum flood series 9 and 0.9
years in the partial duration series in a large variety of streams (Dunne and
Leopold 1978).
The CNs used in the hydrologic modeling discussed below are measures of the
influence of the land cover characteristics on infiltration and runoff. CNs are
empirically derived, and depend on a combination of factors: vegetation types
and conditions; impervious cover; land use practices; and hydrologic soil groups
(HSGs). As forest gives way to pasture or cropland, and then the lawns in
subdivisions, as pavements and roofs are introduced, and as soils are
compacted, CNs increase in value, and so too does surface runoff.
Thus, the causal relations are clear – increased CN leads to increased runoff,
which in turn produces increases in the amount of flow in the above-mentioned
bankfull discharge (or conversely, the larger flows have reduced recurrence
intervals). This increased dominant channel-forming flow must then enlarge the
channel to accommodate the increased runoff. In other words, the stream flows
with recurrence intervals of 0.9 to 1.5 years (depending on the series used) are
larger and the channel must adjust itself accordingly. It will do this by eroding its
banks, and, where the gradients permit, by downcutting its bed as well.
During our field reconnaissance, examples of such channel enlargement, recent
and ongoing, were found throughout the watershed – in virtually all tributaries, as
well as the main stem. The most dramatic examples noted were those where the
stream beds of tributaries were founded in bedrock, leading to marked widening
of the channels. Such widespread channel erosion will produce corresponding
increases in sediment transport and thereby in the Total Suspended Solids
(TSS), load of the streams, one of the principal factors cited in the impairment of
stream quality in this and similar watersheds.
15
While the causal and qualitative relationships outlined above are clear, additional
research is needed to quantify the linkages. That is, we can model directly the
runoff increases that will result from specific increases in the CNs, but the
channel responses to those increased bankfull discharges, the consequent
erosion, and the resultant increases in TSS will require more study.
The Hydrologic Modeling
Figure 2.2.2. The PCW Hydrology
For hydrologic modeling, the U.S. Army Corps of Engineers’ (ACOE) software
HEC-HMS was used. The watershed was treated as consisting of 10 subbasins,
whose areas range from 2.6 to 8.3 mile2 with an average of 5.5 mile2. A CN was
computed for each subbasin based on Land Use Land Cover (LULC) data and
soil type data.
The outflows from the subbasins were assumed to pass through six Junctions as
available from HEC-HMS. The junctions are designated by the symbols C in
Figure 2.2.2. (e.g., junction 2C). They were connected to each other and to the
outlet of the watershed by six reaches (designated by the symbol R in Figure
2.2.2., (e.g., 3R). The routing of water flow through the reaches was conducted
16
using the Modified Puls method, which required evaluation of the number of
subreaches. In this work, the following approach was followed for each reach:
1. A Hec-RAS model was developed using multiple cross sections at a
spacing of 30 – 200 feet.
2. Different flow rates, varying from 100cfs to 30,000 cfs were routed.
3. The cumulative volume of water for each reach was recorded and a
storage-outflow table was developed for each reach (see details under the
section “reach properties”).
4. An average travel time was determined for each reach based on the
computational interval of 15 minutes. The number of sub-reaches was
then computed.
5. The number of sub-reaches from 4) was allowed to change by ± 20% in
matching simulated hydrographs to observed hydrographs at the USGS
station (Rhawn Street).
The values of the hydrologic parameters are reported in Table 2.2.1. and 2.2.2.
These values were obtained in a two step process, where the values computed
based on watershed characteristics (such as LULC, soil type, slope) were altered
by calibration of the model to the outflow at the Rhawn Street USGS station in
Philadelphia. In general, the differences between the final and initial values were
less than 5% for CN (Table 2.2.1.) and 20% for the time lags (Table 2.2.1.) and
the travel times in reaches (Table 2.2.2.).
Table 2.2.1. Sub-basin Properties
Basin Area CN Percent
Impervious Time lag
(mile2) (minute)
1B 8.314 80.53 13.34 126
3B 5.9627 77.93 11.64 116
2B 7.9365 80.03 21.32 122
4B 4.9918 74.3 2.37 95
5B 4.1826 77.45 7.41 102
6B 3.9409 77.7 6.8 85
7B 4.7719 74.92 5.14 98
8B 2.6074 74.97 22.39 128
9B 7.1235 73.13 25.49 145
10B 6.0329 76.71 34.97 182
17
Table 2.2.2. Reach Properties
Reach Length Width Slope Manning’s n
ID (feet) (feet) (ft/ft)
1R 17691 15 0.003 0.035
2R 15963 20 0.0017 0.035
3R 1782 25 0.0047 0.035
4R 16502 25 0.0008 0.035
5R 28211 30 0.0018 0.035
6R 18306 30 0.0008 0.035
Hydrologic Data
Topography plays a vital role in the distribution and flux of water and energy in a
watershed. The USGS has prepared 7½° quadrangle topographic maps at a
scale of 1:24,000 for most of the country, and a common contour interval is 10 ft.
This scale is generally considered the minimum scale in hydrologic modeling and
a tighter interval is preferred for accurate and detailed studies. With the help of a
research grant obtained for the Pennypack Creek Watershed Study (Meenar
2006), the CSC consultant, Aero2 Inc., created the digital ortho-photography and
2 ft-resolution elevation data. The aerial mapping was done in the non-growing
season, when foliage was off the trees. Aero 2 has undertaken the following
steps:
Aerial Photography at 1”=660’ negative scale using Airborne GPS
technology flight;
Ground Control Survey, performed by licensed land surveyor;
Analytical Aero-triangulation, which performs image measurements to
achieve interior and exterior image parameters; and
Stereo compilation and creation of new data.
Integration of GIS and hydrologic modeling connects geospatial data with
hydrologic process models describing how water moves through the
environment. GIS is commonly used for watershed delineation, runoff estimation,
hydrologic modeling, and floodplain mapping. In addition to the GIS-based data
inventory described in the previous section, the newly created GIS datasets
include 2003 digital ortho-photographs (1 ft pixel resolution), 2 ft resolution
elevation data such as DEM, Triangulated Irregular Network (TIN), and contour
intervals, updated stream networks, flow-paths, bridges and culverts, dams, and
building foot prints.
Research has indicated that the precipitation values (Herschfield 1961) widely
used in previous studies are no longer valid. These values were used in the
creation of existing Pennypack Flood Insurance Rate Maps (FIRMs) as well. It
has been well established that TP-40 systematically underestimated the extreme
18
precipitation events. This was due to a number of factors: the short average
duration of the precipitation records analyzed; the relatively small number of
weather stations; and the statistical distribution used to analyze the data. The
Temple researchers requested and received permission from FEMA to use more
recent data from the National Oceanic and Atmospheric Administration (NOAA)
Atlas 14, which can be accessed at: http://hdsc.nws.noaa.gov/hdsc/pfds/orb/pa_pfds.html.
This makes a very significant difference in the inputs to the hydrologic and
hydraulic models employed in the re-delineation of the floodplains in the
watershed. For instance, according to TP-40, the precipitation from the 100-year,
24-hour storm in our study area was expected to be 7.2 inches, and this number
has been codified in virtually all local stormwater management ordinances. The
more recent data indicates that the 100-year, 24-hour event is 8.75 inches, a
difference of more than 20%.
Streamflow data used for calibration were obtained from the USGS Station
01467048 located approximately at Rhawn Street. As mentioned previously, in
predicting the runoff resulting from the 100-year storm, the depth of rainfall
estimated by the NOAA Atlas 14 (8.75 inches) was used instead of the older TP-
40 study. This greater rainfall depth along with the new land use data resulted in
runoff peak values and volumes that are larger than those predicted in prior
studies in the PCW. At the location (Lat=40.1; Long=-75.3), the upper 95%
confidence gave 6.28, 7.19, 8.18, 10.81, for the 25, 50, 100, and 500 year
storms. The temporal distribution of rainfall pulses for the design storms for the
area are of the SCS Type II.
Calibration
The hydrologic modeling process entails developing an actual or hypothetical
design storm and then calculating the runoff and peak discharge for the selected
event. Eight storms were used for the calibration. They are listed in Table 2.2.3.
along with the total amount of rainfall and the runoff duration. The automatic
calibration option in HEC-HMS was not used because it provided a different set
of parameters for each storm. We elected to adjust the parameters based on
heuristic arguments and to put a special effort on matching the peak value and
the time-to-peak. This resulted in a unique set of parameters, reported in Tables
2.2.1. and 2.2.2.
19
Table 2.2.3. Rainfall Events Used for Calibration
# Date Total Rainfall
(inch)
1 October 08, 1996 2.00
2 October 16, 1996 3.46
3 September, 1999
(1)
7.03
4 November, 99 1.12
5 December, 99 1.65
6 March, 2002 1.15
7 May, 2002 1.50
8 June, 2002 1.62
(1)
Hurricane Floyd
The graphs in Section A.3.1. of Appendix A.3. show the comparison between
predicted (or simulated) runoff and those observed.
2.3. Ecological Indicators in the PCW
Indicators are used to describe and quantify the status of a system. Changes in
indicators can gauge the outcomes of actions or policies. Indicators are often
used to avoid having to process large amounts of detailed information. Meadows
(1998) presented a list of characteristics that sustainability indicators should
display. They should be, among other things: (1) clear in value; (2) clear in
content, with units that make sense; (3) measurable at reasonable cost; (4)
timely; (5) appropriate in scale; (6) hierarchical; (7) based on physical units rather
than money and prices; and (8) leading, so as to provide information in time to
act on them. These criteria will be discussed below. In the sequel, the terms
variables, environmental metrics and indicators will be used synonymously.
2.3.1. Water Volume
Based on the hydrology discussion above, this indicator seeks to measure the
current amount of water that is consistently available as compared to the amount
of water that would be available if the watershed were in a natural state. As the
area within a watershed is developed, less water is generally available because
impervious surfaces such as buildings, streets, parking lots and driveways cause
water to run immediately into streams rather than infiltrating into the groundwater
system. Groundwater moves slowly through cracks and spaces in the bedrock,
providing water for streams in dry weather as well as drinking water from wells.
New development reduces groundwater supply at the same time that it increases
water demand. This means that as development increases, planning for
adequate, long-term supplies of high-quality water becomes increasingly
important.
20
Data and Methods
Baseflow (often called the dry-weather flow) of a stream, a flow which is supplied
by groundwater. Because of this, baseflow is a good measure of water
availability. Figure 2.3.1. (Map BASEFLOW1) shows the bedrock geology areas
of the Pennypack watershed. The Pennypack baseflow was calculated using
data from the USGS Rhawn St. Stream Gauge in Philadelphia. Figure 2.3.2.
(Map BASEFLOW2) shows the area that drains to the Rhawn St Gauge. The
lack of additional operative stream gauges in the PCW currently makes it
impossible to calculate the baseflow at more than one point along the Creek.
This limitation is considerable, particularly when considering that the discharge
from the UM-H STP falsely inflates the baseflow, and the only functioning stream
gauge is downstream of this plant. The average daily discharge for the UM-H
STP was obtained from the Pennsylvania Department of Environmental
Protection (PA DEP). Figure 2.3.3. (Map BASEFLOW3) shows the area that
drains to the UM-H STP.
The natural baseflow rates used for this discussion came from the Water-Use
Analysis Program for the Neshaminy Creek Basin in Bucks and Montgomery
Counties, prepared by the USGS and the DRBC. This report contains the
baseflow discharge rates for the various geologic formations in PCW.
The 25-year baseflow at the Rhawn St gauge was arrived at by separating the
hydrographs of the stream gauge data using a computer program based on the
local-minimum method. (Sloto and Crouse 1996) Low-flow conditions was
selected for these calculations because it is during dry periods that water supply
presents a problem, and the aim of the indicator is to assess the water supply
that would be consistently available. The 25-year low flow was selected because
it is the most extreme condition that could be accurately modeled with the
available data. The 25-year low-flow condition is the low flow, or drought
condition that is predicted to occur one year out of every 25 years, or has a 4%
chance of occurring in any given year. The amount of the discharge from the UM-
H STP plant that was drawn from outside the PCW was then calculated, and that
amount was subtracted from the baseflow calculated by the separation of
hydrographs. This is the current 25-year baseflow. The natural 25-year baseflow
was calculated by finding the areas of each geologic formation in the PCW.
These areas were then multiplied by the flow rate for each geologic formation.
The sum of these values represents the natural baseflow at the Rhawn St.
Gauge. For more details on these calculations see Appendix A.3.
Three parameters affect the peak flows in watersheds. These are the CN, the
time of concentration, tc, and the percent impervious.
The potential infiltration, S, of a watershed is given by (in inch):
1000 10SCN
=−
(2.3.1.)
21
Where CN is the Curve Number, a number between 0 and 100, but commonly
ranging from 60 to 90. Higher values of CN imply lower potential infiltration, and
thus higher runoff and, most likely, higher peak flows.
The time of concentration represents the time it takes for the water parcel to
propagate from the most hydraulically remote location of the watershed to the
outlet. There is a relation between the time of concentration and the CN. This
relation is:
0.8
0.5
5(1
3 1900
cLS
t)
y
+
= (2.3.2.)
Where L is the length of main channel and y is the average slope of the
watershed (in percent).
Equations (2.3.1.) and (2.3.2.) indicate that the time of concentration is inversely
proportional to the CN, which is intuitive in many aspects; an increase in CN is
commonly a result in an increase in the impervious areas (concrete replacing
open soil) causing water to propagate faster in the watershed. However,
Equation (2.3.1.) should be used with great care, because the quantity length L
and average slope are not pure geomorphic quantities, but they depend rather on
water pathways. Different drainage patterns are expected to affect water flow
differently; dendritic drainage patterns (spatially random) tend to drain
watersheds slower than rectangular patterns. The evaluation of drainage
capability becomes more difficult in urban watersheds, where the geomorphic
significance of the length L and the slope vanishes in favor of hydraulic
significance, mostly due to constructed channels.
The percent impervious that we are referring to herein differs from that would be
incorporated in the CN evaluation. This one relates to impervious areas directly
connected to the channels, and thus their increase causes direct effects
(increase) on peak flows. The US Army Corps of Engineers’ model, HEC-HMS
accounts for these areas explicitly.
22
Results and Interpretation
The calculated current one-year-in-25 annual baseflow at the Rhawn St. gauge
was found to be 6.7292 million gallons per day, while the natural baseflow at that
gauge was found to be 12.8178 millions of gallons per day. Restated as a
percentage, current baseflow is only 52.5% of natural baseflow. This means that
during low-flow conditions, approximately half of the water that would naturally be
available for drinking water, waste assimilation, fish and wildlife habitat, etc. will
not be available. Although this area is water-rich, drought is a regularly recurring
problem, and a decreasingly reliable water supply will only serve to exacerbate
the effects of what might otherwise be minor problems. Additionally, during low-
flow conditions water demand by human beings decreases only slightly. This
means that the proportion of the water supply that is used by human beings
increases greatly, leaving less for the natural communities, particularly aquatic
communities.
Downstream of the sewage treatment plant, the effects of lower baseflow would
appear to be lessened. The stream does have over 90% of its natural baseflow
due to the discharge from the UM-H STP. However, from a regional perspective
the appearance of improvement disappears. The service area for the sewage
treatment plant extends beyond the watershed boundaries, resulting in an inter-
basin transfer of water. Water that is imported to the Pennypack has been
removed from other watersheds, further depleting their water supply. Also, the
water discharged into the stream is not of the same quality as the water already
in the stream.
23
The results for this indicator show that the Pennypack baseflow has decreased
from what it would be in an estimated natural state. However, the lack of stream
Figure 2.3.1. PCW Baseflow Geology
24
Figure 2.3.2. PCW Drainage to the Rhawn St. Gauge
25
Figure 2.3.3. PCW Upstream of the UM-H STP
26
gauges in the watershed imposes serious limitations on the analysis. The effect
of development on the baseflow of the Pennypack Creek is generally clear.
However, continuous and more detailed monitoring is necessary to track carefully
the effects of future development, as well as efforts to remediate the problem.
The peak flows in the PCW for the two-year storm under the current conditions
are shown at the PWD monitoring stations in Figure 2.3.4 and numerically in
Table 2.5.1.
Figure 2.3.4. Peak Flows under Current Conditions
These flows will be compared in Section 2.5.1. with the conditions that result
from the first two residential development scenarios.
27
2.3.2. Water Quality
This indicator seeks to measure the chemical contamination of the water.
Chemicals enter the aquatic system from both point and non-point sources. Point
sources are factories and other facilities that discharge directly into a stream.
These sources are highly regulated. Non-point sources are farms, lawns, parking
lots, etc. Chemicals from these sources are carried by rainwater into streams.
These non-point sources are very difficult to regulate, and so are relatively
uncontrolled.
Regardless of their source, chemicals that contaminate surface and groundwater
impair the water’s ability to act as habitat for wildlife, and make it dangerous for
humans or animals to drink it unless treated. Concentrations of nutrients and
pesticides in streams and shallow groundwater generally increase with
increasing amounts of agricultural and urban land in a watershed. This situation
develops because chemical use increases and less water is available from
undeveloped lands to dilute the chemicals. (USGS 1999)
Data and Methods
The data for this analysis came from the PWD’s Baseline Assessment of PCW
(2002-2003) (Lance et al. 2003). This study assessed the levels of a variety of
contaminants at 20 sampling points along the Pennypack Creek. Figure 2.3.4.
(Map CHEM1) shows these sampling points and the sub-basins to which the
data was attributed. Seven factors that comprise a majority of the impairment in
most streams were chosen to make up this indicator. These factors are
described below.
Alkalinity
Alkalinity is a measure of a stream’s ability to resist changes in pH. It is often
referred to as the buffering capacity of a stream, and is important because it
allows a stream to neutralize acidic pollution or contamination. The target range
for this factor is 100-200 mg CaCO3/L (Lehigh Earth Observatory). The results of
this analysis are on Figure 2.3.5. (Map CHEM2).
Dissolved Oxygen
Dissolved oxygen is absorbed from the atmosphere and from the result of
photosynthesis. If more oxygen is consumed than produced or imported, some
organisms will die. The target for this factor is 6 mg/L (US EPA 2005).
Fecal Coliform
These bacteria are present in the feces and intestinal tracts of humans and other
warm-blooded animals, and can enter water bodies from human and animal
28
waste. If a large number of fecal coliform bacteria (over 200 colonies/100 ml of
water sample) are found in water, it is possible that pathogenic organisms are
also present in the water. High concentrations of the bacteria in water may be
caused by septic tank failure, poor pasture and animal keeping practices, pet
waste, and urban runoff (BASIN 2002). The target for this factor is 200
colonies/100mL (US EPA 2005). The results of this analysis are on Figure 2.3.6.
(Map CHEM3).
Nitrate
Nitrate is the most completely oxidized state of nitrogen commonly found in water
and is the most readily available state utilized for plant growth. High nitrate levels
combined with phosphates cause excessive plant and algae growth, which is a
deteriorating process called eutrophication. Higher concentrations in water are
unsafe to drink due to the possible presence of altered forms of nitrite, which may
cause serious illness to both humans and wildlife. The target for this factor is 1
mg/L (Lehigh Earth Observatory 2005). The results of this analysis are on Figure
2.3.7. (Map CHEM4).
Ortho-phosphate
Ortho-phosphate is the form of phosphate used in fertilizer and applied to
agricultural fields and residential lawns. Like nitrates, phosphates negatively
impact water by causing accelerated rates of eutrophication. The target for this
factor is .03 mg/L (Lehigh Earth Observatory). The results of this analysis are on
Figure 2.3.8. (Map CHEM5).
Suspended Solids
Suspended solids include all particles in water with a diameter of less than 0.45
microns. Typically, suspended solids include items such as soil, algal cells, and
plant particles. High levels of suspended solids smother some aquatic
organisms. The target for this factor is 10 mg/L (Lehigh Earth Observatory). The
results of this analysis are on Figure 2.3.9. (Map CHEM6).
pH
pH is a measure of acidity. Variations in pH affect chemical and biological
processes in water. Low pH increases availability of metals and other toxics for
intake of aquatic life. It is critical to survival, growth, and reproduction of fish and
macro invertebrates to maintain a constant pH. Exposure to very low or high pH
may cause death or reproductive problems for fish and other aquatic life. The
target range for this factor is 6.5-8.5 (US EPA 2005).
The actual values were transformed into percentages of the target. For chemicals
where the actual values were below the target the percentage was calculated by
29
dividing the actual value by the target. For chemicals where exceeding the target
does not have a negative impact, any value exceeding the target was treated as
100%. For chemicals where exceeding the target is negative, the percentage of
the target was calculated by dividing the target by the actual value (an inverse
percentage). Targets for water quality for these contaminants were determined
using information from the State of Massachusetts Division of Water Pollution
Control and Lehigh University (Sherry 2005). For more details on these
calculations see Appendix A.3.
Results and Interpretation
The indicator value for the Pennypack Creek was 47.47%. This means that the
water is less than half of pristine condition. This value would have been much
lower, but both pH and dissolved oxygen were at optimal levels. The values for
Suspended Solids, Ortho-phosphate, and Fecal Coliform were all very far from
the target, in many cases more than ten times the target value. Nitrate values
were also very high. These results implicate the UM-H STP and non-point
sources. No standard for nitrates and phosphates has been set for aquatic life,
but EPA and the New Jersey Dept of Environmental Protection (NJ DEP) have
suggested a level of around 3 mg/L NO3-N . With respect to phosphates,
concentrations exceeding 0.3 mg/L are considered problematic.
The Temple research team evaluated the performance of the STP in removing
nutrients. (Meenar 2006) In September of 2004, automatic samplers were placed
up and downstream of the STP. The upstream sampler was about 2200 ft from
the treatment plant property and the downstream sampler was about 1300 ft from
the STP. The samplers collected water from Pennypack Creek at 10 am every
day for three weeks.
The sampling results highlight the importance of large point sources in
influencing water quality in urban streams. Upstream of the plant, the nitrate
concentrations were steady at 1-2 mg/L, which are typical values for urban
streams as noted by the USGS. Downstream of the plant, however,
concentrations were much higher, typically over 10 mg/L and up to 22 mg/L.
30
Figure 2.3.5. PWD Stations & Associated Sub-Basins: Chemical Data
31
Figure 2.3.6. %-Alkalinity: Sub-Basin Classification
32
Figure 2.3.7. %-Fecal Coliform: Sub-Basin Classification
33
Figure 2.3.8. %-Nitrate: Sub-Basin Classification
34
Figure 2.3.9. %-Phosphate: Sub-Basin Classification
35
Figure 2.3.10. %-Suspended Solids: Sub-Basin Classification
36
Concentrations dipped below 10 mg/L during storm events on 9/8/04 and
9/18/04, and were lower in the hand sample collected after Tropical Storm
Jeanne. With respect to phosphate, high concentrations of this nutrient were also
confirmed by the CSC research team. These concentrations were quite small
above the plant. Downstream, high concentrations, typically 1.5 mg/L, were
found during non-storm events. These sampling results were generally consistent
with previous sampling efforts conducted by the PWD in 2002 at a location near
the CSC’s sampler downstream of the plant.
Non-point sources, however, are responsible for a majority of the impairment in
the Pennypack in terms of geographic context. Non-point-source contamination
comes from many diffuse sources, including fertilizers and pesticides from
agricultural and residential lands, and nutrients from livestock, pet wastes, and
septic systems. Non-point-source contamination is the leading and most
widespread cause of water-quality degradation, and it can have harmful effects
on drinking-water supplies, recreation, fisheries, and wildlife.
2.3.3. Biological Integrity
Data and Methods
This indicator seeks to measure the health of the aquatic ecosystem in the PCW.
Ecosystems rely on the delicate interplay of many different factors including the
availability of food and habitat, the diversity of species, and the chemical
composition, temperature and clarity of the water. This reliance on many factors
makes biological integrity a strong composite indicator of the health of the
watershed. Also, biological criteria are essential to tracking impaired conditions
such as invasive species or loss of biodiversity, that are not directly caused by
chemical factors (Davis, 1995). Upstream land use changes and alterations of
the stream corridors affect the quality of the water delivered to the stream
channel as well as the structure and dynamics of the adjacent riparian
environments (Davis, 1995). The status of the biological community can provide
information about these factors that would not appear in a chemical analysis.
The data for this analysis came exclusively from the PWD Baseline Assessment
(Lance et al. 2003). This study assessed the biological integrity of both the
macroinvertebrate populations and the fish populations in the PCW. The
macroinvertebrates were collected from a period of 4/2/02 to 4/9/02 using Rapid
Bioassessment Protocols (RBPs; Barbour 1999) at 20 locations along
Pennypack Creek. Figure 2.3.10. (Map BIO1) shows these locations and the sub-
basins to which the data were attributed. EPA’s RBPs were designed to provide
cost-effective biological methods for states, and local agencies. The protocols
exist for periphyton, benthic macroinvertebrate, fish, and habitat assessment,
and all of these protocols have been tested in streams in various parts of the
country. The biological integrity at each site was determined by PWD following
RBPs and was compared to a reference site. Table 2.3.1. taken from the PWD
37
study gives a framework for understanding the biological integrity scores. (Lance
et.al. 2003)
Table 2.3.1. Interpreting Biological Integrity Scores
Biological
Integrity
Score
Condition Attributes
> 83% Non-
impaired Comparable to the best situation in an ecoregion.
Balanced trophic structure. Optimum community
structure for stream size and habitat
54-82% Slightly
Impaired Community structure less than expected. Species
composition and dominance lower than expected
due to the loss of some pollution intolerant forms.
Percent contribution of pollution tolerant forms
increases.
21-53% Moderately
Impaired Fewer species due to loss of most of the pollution
intolerant forms.
< 20% Severely
Impaired Few species present. If there are high densities of
organisms then the system is dominated by one or
two taxa.
The fish surveyed in this study were collected in July and August of 2002 at
eleven of their monitoring stations along the Pennypack Creek. Figure 2.3.11.
(Map BIO3) shows these locations and the sub-basins to which the data was
attributed. The fish were collected by electrofishing as described in Barbour
(1999). The biologic integrity of the fish community was then assessed using the
Index of Biological Integrity (IBI) developed by Karr (1981). Table 2.3.2. explains
the meaning of ranges of the IBI score.
A scale of 0 to 100% was established as the common framework for the all of the
indicators in this study. The macroinvertebrate data were already in this
framework. To transform the fish data to match the IBI score, the latter were
multiplied by 2. An average percentage for both fish and macroinvertebrates was
then determined. A weighted average for the two was also calculated in the
interest of having a single score for the indicator. For more details on these
calculations see Appendix A.3.
38
Table 2.3.2. Ranges of the IBI Score
IBI
Score Condition Attributes
45-50 Excellent Comparable to pristine conditions, exceptional
assemblage of species
37-44 Good Decreased species richness, particularly pollution
intolerant species
29-36 Fair Pollution intolerant and sensitive species are absent. The
trophic structure is skewed.
10-28 Poor Top carnivores are absent or rare. Omnivores and
pollution tolerant species dominate.
<10 Very Poor Few species and individuals are present. Pollution
tolerant species are dominant. Diseased fish are
prevalent.
Results and Interpretation
The average biological integrity score of the macroinvertebrate communities
studied in the PCW was 21.67%, which according to the system devised by PWD
is at the lower end of the moderately impaired category. The highest score for
any station was 66.67% - slightly impaired, and this was a station at the extreme
upstream portion of the watershed. Figure 2.3.12. (Map BIO2) presents these
results. In general the stations that monitored tributaries had higher scores than
those stations on the main stem. However, none of the stations monitored had a
non-impaired macroinvertebrate community, and a majority of the communities
were either severely or moderately impaired. These insects are an important link
in the aquatic food web, converting plant and microbial matter into animal tissue
that is then available to fish, and their loss makes it difficult for fish and other
predators to survive. The average biological integrity score for the fish
communities studied in the PCW was 61.20%.This is considerably higher than
the score yielded by the macroinvertebrates.This is at least partially due to the
fact that the fish had greater biodiversity in the tidal areas. The
macroinvertebrates fared poorly there due to the unstable water levels (Lance et.
al. 2003) Figure 2.3.13. (Map BIO4) presents these results.
Combining these scores yielded an overall biological integrity score for the
watershed of 45.39%. This low score strongly suggests that the effects of human
habitation are strongly and negatively impacting the biological communities in the
Pennypack Creek.
39
Figure 2.3.11. PWD Stations & Associated Sub-Basins: Macroinvertebrate Data
40
Figure 2.3.12. PWD Stations & Associated Sub-Basins: Fish Data
41
Figure 2.3.13. Biological Integrity for Macroinvertebrates: Sub-Basin Classification
42
Figure 2.3.14. Biological Integrity for Fish: Sub-Basin Classification
43
2.3.4. Impervious Surface
This indicator seeks to measure the percentage of the land within the watershed
that is impervious. In natural settings, very little annual rainfall is converted to
runoff, and about half of the water that seeps into the underlying soils becomes
groundwater. In urbanized areas, impervious surfaces prevent the rainfall from
seeping into the ground, and as a result the rainfall is converted into runoff. This
water runs directly to the streams and increases stream volume during rainfall
events. Depending on the degree of impervious cover, the annual volume of
stormwater runoff can increase by 2 to 16 times its pre-development rate, with
proportional reductions in groundwater recharge. (Schueler 1994) Impervious
surfaces raise the velocity as well as the volume of stormwater runoff. This
increases erosion and sedimentation, and washes oil and other chemicals from
roadways and parking lots into surface waters. Impervious surfaces also absorb
heat, and often increase stream temperatures during runoff events. These
physical changes are generally accompanied by decreasing water quality and
decreasing biodiversity (Capiella, 2002).
Data and Methods
The Impervious Surface Data originated from The National Land Cover
Database, 2001. The data in this database were generated from satellite
photography. The satellite data provided a percentage of impervious coverage
for each 25 m x 25 m pixel. The raster containing impervious surface data was
clipped to the PCW boundary and then converted to a shapefile. The shapefile
containing the 49 sub-basins was converted to a raster and then back to a
shapefile, to match the transformed raster’s boundary. Using Hawth’s Tool:
Polygon in Polygon Analysis, the average percentage of surface covered by
impervious was calculated for each sub-basin. Then, based on this average of
the impervious surface and the area of each sub-basin, the area covered with
impervious surface (in square miles) was calculated. This can be seen on Table
2.3.14. (Map IMPERVIOUS1). The mean of these percentages provided the
average impervious surface for the PCW. For more details on these calculations
see Appendix A.3.
Results and Interpretation
On average 29.67% of the PCW is covered by impervious surface. Many
different studies from various geographic areas, employing different methods and
concentrating on different variables, have all come to the conclusion that stream
degradation occurs at relatively low levels of imperviousness 10-20% (Schueler,
1994). This suggests that the PCW is experiencing significant impacts to its
water quality, biological integrity, and groundwater/baseflow from the high
percentage of impervious surface in the watershed. The results from the other
three indicators all support this conclusion. The current baseflow levels are at just
over half what they would be if the watershed were not developed. The biological
44
integrity of the aquatic communities is fair at best, and the levels of many of the
pollutants in the stream greatly exceed the acceptable levels, if not the EPA
guidelines. If the trend of development in the watershed continues, then all of
these indicators will continue to worsen unless efforts are put into place to
mitigate the effect of development.
Figure 2.3.15. %-Impervious Surface: Sub-Basin Classification
45
2.4. Two Land-Use Scenarios
In the analysis of the regional environmental future for the Pennypack Creek
Watershed, it is important to project possible future land use scenarios and
futures. In this section, we describe three alternative future land use scenarios.
These scenarios are primarily focused on macro trends in land use change, and
do not reflect site-specific innovations which might occur, such as floodplain
acquisitions or increased use of site-level stormwater management best
practices.
The analysis started with an examination of the demand for land from projected
population growth. The most recent official population forecasts were acquired
from the DVRPC, the designated regional and Metropolitan Planning
Organization (MPO) covering the PCW. We compared the official DVRPC
population forecasts to those utilized by the Montgomery County Planning
Commission and found no significant differences. The official population
forecasts are used by DVRPC for transportation planning and modeling, and
serve as an objective source of population and employment forecasts. The
forecasts have been updated in 2005 to account for changes in municipal
population from 2000-2005. In this analysis, only the projected population growth
rates in the 11 municipalities outside of the city of Philadelphia are examined for
two reasons. First, nearly all the land within the watershed in Philadelphia is
already considered developed. Second, the neighborhoods within Philadelphia in
the PCW are not forecasted to experience any significant population changes,
absent large-scale redevelopment efforts.
The official population forecasts for each of the 11 non-Philadelphia
municipalities which have some or all of their land area in the PCW are then
scaled down to represent the housing and population growth needs of the areas
of the municipalities which lie within the boundaries of the watershed. Using the
land use and demographic databases described in Appendix A.2., it can be
determined how much of a municipality’s population and land area lie within the
PCW using the weighted-average technique. These percentages were used to
apportion future population growth targets to the appropriate watershed. For
example, if a weighted average of 30 percent of a municipality’s housing units
and land area were within the PCW, then 30 percent of the forecasted population
growth rates were apportioned to future growth within the watershed. This
seemed the most reasonable approach. However, it is certainly possible that as
a result of this planning effort, municipalities may choose to allocate a larger
portion of their future housing and employment growth to other watersheds within
their boundaries.
One of the difficulties in watershed planning and land use forecasting within
smaller watersheds in Pennsylvania is that fundamental land use decisions are
made by municipalities and the boundaries of municipalities do not conform to
watershed boundaries. Municipalities are generally required to make adequate
46
provision in their zoning for their projected population growth, but determining
into which watershed the growth will be directed is difficult.
Table 2.4.1. presents the future population needs for PCW municipalities,
representing only those future populations assigned to growth within the
watershed. The first two columns represent the adjusted official population
forecasts for the years 2020 and 2030 for each municipality. Columns 3 and 4
convert population forecasts into an indication of aggregate housing unit needs.
Based on standard practice, these are calculated as future population divided by
average number of persons per occupied housing unit for each municipality in
the most recent Census (2000). Thus, the assumption is made that the average
number of persons per occupied housing unit will remain the same over the 30
year planning horizon. This assumption is the best assumption of household
sizes, even though average household size in the US has been consistently
declining. It would be possible to re-estimate the following scenarios assuming
smaller household sizes, which would only have marginal impacts on residential
land needed. As well, household size will also be a variable somewhat within the
control of municipalities in the zoning policies which control the types and sizes
of housing units constructed. Within the PCW, the average household size is
2.66 persons per household, ranging from a low of 2.2 persons per household in
Jenkintown Borough to a high of 3.5 persons per household in Bryn Athyn
Borough. Housing unit needs were also adjusted upwards by 2 percent to reflect
an estimated average vacancy rate of 2 percent. In the year 2000 within the
watershed, the vacancy rate was 2.4 percent.
Columns 5 and 6 of Table 2.4.1. convert the gross housing unit needs of the
years 2020 and 2030 into the number of new units which need to be constructed
during the planning horizon. These figures are arrived at by subtracting from
housing needs the number of existing units in the year 2020. Overall, the results
of the demographic analysis do not show much growth in the non-Philadelphia
municipalities of the PCW. Population is only expected to grow from a 2000 level
of approximately 100,000 to a 2030 population of 106,000. Only 1855 new
housing units in a 30 year time period would be needed to accommodate this
population growth. As demonstrated below in the land use scenarios, however, if
these housing units are produced at lower densities, the amount of undeveloped
land remaining in the PCW would be significantly reduced.
47
Table 2.4.1. Forecasted Population Growth and Housing Unit Construction Needed
2020
Housing
Unit
Need
2030
Housing
Unit
Need
2020 New
Unit
Construction
2030 New
Unit
Construction
Population
2020 Population
2030
Bucks County
Upper Southampton
Twsp. 4659 4768 1818 1860 157 199
Warminster Twsp. 16056 16803 5923 6198 696 971
Montgomery County
Abingdon Twsp. 28063 28134 11067 11094 -205 -177
Bryn Athyn Borough 1410 1420 401 404 20 23
Hatboro Borough 7470 7500 3134 3147 13 26
Horsham Twsp. 9022 9707 3449 3711 399 661
Jenkintown Borough 523 513 242 238 -10 -14
Lower Moreland Twsp. 9937 10324 3695 3839 73 217
Rockledge Borough 1383 1366 580 573 -28 -35
Upper Dublin Twsp. 892 904 322 327 16 20
Upper Moreland Twsp. 24655 24625 10183 10170 -24 -36
TOTAL PENNYPACK
WATERSHED 104069 106064 40814 41562 1108 1855
2.4.1. Scenario 1: Trend Development
Table 2.4.2. represents the land use analysis associated with Scenario 1: Trend
Development. In this scenario, we assume that each new housing unit will use
the same amount of land as the existing year 2000 average housing unit land
use for each municipality. That is, in this scenario we assume that current
densities (reflecting current zoning and current development practices) predict
future densities. This assumption is still somewhat conservative in terms of land
use, because newer housing units generally are produced at densities lower than
existing average densities.
48
Table 2.4.2. Land Development Rates: Trend Development Scenario
2020
Residential
Need
2030
Residential
Need
2020 Non-
Residential
Need
2030 Non_
Residential
Need
2020
Total
Need
2030
Total
Need
Land Suitable
for
Development
Acres Acres Acres Acres Acres Acres Acres
Bucks County
Upper Southampton
Twsp. 67 85 18 23 84 108 164
Warminster Twsp. 209 292 90 125 299 416 404
Montgomery
County
Abingdon Twsp. -59 -51 -59 -51 432
Bryn Athyn Borough 13 15 3 3 16 18 58
Hatboro Borough 3 5 4 5 6 10 55
Horsham Twsp. 172 284 48 80 220 364 736
Jenkintown Borough -1 -2 -1 -2 0
Lower Moreland
Twsp. 46 135 11 28 56 163 236
Rockledge Borough -4 -5 -4 -5 0
Upper Dublin Twsp. 8 11 2 2 10 13 15
Upper Moreland
Twsp. -6 -10 6 5 0 -5 439
TOTAL
PENNYPACK
WATERSHED 446 758 181 270 627 1029 2539
Using the high-resolution digital land data, gross residential housing unit
densities are determined for each municipality as the number of housing units
divided by land classified as in residential use. (Residential classifications by
DVRPC include streets internal to a development, utility rights-of-way and
stormwater drainage facilities, but do not include commercial facilities or non-
local roads.) Thus, the estimate of gross residential housing unit densities is a
good estimate of the amount of land used per housing unit. Using the figures
from 2000, we project aggregate residential land use in Table 2.4.2., shown in
columns 1 and 2. Development densities across the region range from a low of
1.6 housing units per acre in Bryn Athyn to a high of 7.8 housing units per acre in
Jenkintown.
Estimates of the amount of land needed in non-residential development
(including commercial, industrial, office, utility, and transportation needs) can be
estimated with detailed employment growth forecasts to convert employment
needs into space requirements. Unfortunately, at the small scale of a municipality
or portion of a municipality, employment forecasts are difficult to obtain and of
questionable quality. The alternative approach, common in many planning
applications, is to assume a fixed amount of non-residential land per capita, and
therefore, to assume that non-residential land use needs are driven by local
population growth. In this case, we project that per-capita demand for non-
49
residential land will be approximately 2000 square feet. That is, each person
added will produce a demand for an additional 2000 square feet of non-
residential urban development. These assumptions were parameterized based
on an analysis of aggregate land development uses within the entire Philadelphia
metropolitan area. Based on DVRPC data, we estimate that non-residential
urbanized land uses amount to approximately 2212 square feet per resident in
the region not including Philadelphia and 1833 square feet per resident when
Philadelphia is included. Thus, the assumption of an additional 2000 square feet
of non-residential urban land use per new additional resident seems a
reasonable assumption based on existing development trends. However, in
municipalities which are projected to experience population decline, we assume
that the same amount of urbanized non-residential land is maintained.
The analysis in Table 2.4.2. indicates that, at current trend densities, the PCW
will see an aggregate additional 1029 acres converted to urban development in
the 30 year period between 2000 and 2030. However, the estimate of 1029
acres is aggregated across some municipalities where population growth is
forecasted to be negative. If, as in this scenario, it is assumed that each
municipality must accommodate its own projected growth needs, and that
declining population municipalities retain the same amount of developed land,
then 1092 total additional acres will be needed to service new urban
development by 2030.
For this scenario, in order to apportion future land use growth in the various
scenarios, the suitability and capability of currently undeveloped land to
accommodate land development and growth were analyzed. The first step was to
create a layer of land use which is “potentially developable.” Potentially
developable land was defined as all land currently in use in the agricultural,
wooded, recreation or vacant categories. We then overlaid layers of known
permanently-preserved open space land (state, county and municipal parks,
PERT land, etc.) to remove lands which would not be developed due to
preservation easements or public ownership. All the remaining land is considered
“potentially developable.”
Within the land classified as potentially developable, two criteria were applied to
identify those lands which are most suitable for development. The first criterion
was the absence of environmental constraints and the second criterion was the
availability of sewer infrastructure. For the environmental constraint criteria, we
used data on slopes, streams, floodplains and wetlands. Land was identified as
being environmentally constrained if it were over 15 percent sloped, within a
floodplain, within 100 feet of a stream or within 25 feet of the edge of a wetland.
A data set was used on “sewer service areas” (from DVRPC) to identify those
lands which are within a sewer service area. Thus, those lands which are most
suitable for development are those lands which are within a sewer service area
and which lack environmental constraints. Within the entire PCW (including
Philadelphia), there are nearly 3900 acres considered suitable for development
50
under this methodology, of which over 2500 acres are outside of the city of
Philadelphia.
Of these 3900 acres potentially more suitable for development, over 40 percent
are currently wooded and nearly 13 percent are in agriculture. Thus, one of the
planning challenges facing the watershed is balancing the growth needs with
preserving forested and agricultural landscapes. In this analysis, an area being
classified as potentially suitable for development does not mean that
development of these landscapes is the most appropriate policy choice.
In the future land use development envisioned under this scenario, each
municipality develops land to meet its own projected residential and non-
residential needs. In this scenario, growth needs were compared with lands
designated in the analysis above as suitable for development. As can be seen in
a comparison of the growth needs in columns 5 and 6 of Table 2.4.2. with column
7 (lands suitable for development), nearly every municipality has adequate
suitable land to meet its forecasted growth needs at present densities. The only
exception is Warminster Township, which in this scenario would use 416 acres in
the 30 year planning period, while having only 404 acres available for
development. In the scenarios constructed, it is assumed that all development
for municipalities occurs on lands classified as suitable for development. For
Warminster, the additional 12 acres of development needs are assumed to come
from land designated as un-sewered but not environmentally constrained (of
which 20 acres are available.)
Each municipality accommodating its own projected land development needs in
many ways represents the trend in Pennsylvania land use planning by
municipalities, as each municipality is under an affirmative obligation to
“accommodate reasonable overall community growth, including population and
employment growth” (PDCED; 2005) absent a shared land-use agreement within
a multi-municipal plan.
Thus in the Trend Development scenario, the following rules are used to
accommodate each municipality’s future population growth in the 20 and 30 year
planning horizons. First, if the municipality shows a negative projected population
growth rate, it is assumed that land use patterns in 2020 and 2030 will remain the
same as in 2000. That is, even though population will decline, it is assumed that
the number of housing units and the amount of land in non-residential urban use
will not change. Given the durable nature of infrastructure, housing and urban
development, this is a reasonable assumption. Second, if the municipal growth
can be accommodated on land which has public sewers available and is not
environmentally constrained, all of its development needs were simulated on that
land. As much development as possible was first allocated on land currently
classified as “vacant.” Third, for the one municipality (Warminster) which required
additional land for development, its remaining development needs were allocated
to land which was not environmentally constrained, without public sewers.
51
Figures 2.4.1. shows the projected land use in 2030 under Scenario 1.
Figure 2.4.1. Trend Development Land Use, 2030
Much of the undeveloped land near the various streams of the watershed is
protected in this scenario from development because of the environmental
constraints. Most of the land conversion under this scenario occurs in the
currently less developed townships in the northern portion of the watershed.
2.4.3. Scenario 2: Smart Growth
Municipal Smart Growth
In this scenario, each municipality accommodates its forecasted population
growth needs, but accommodates the residential portion of that growth at
significantly higher gross housing densities and the non-residential portion at
slightly increased intensities. In order to illustrate this scenario, densities of 6
units per gross residential acre were chosen to simulate all new residential
development in the less dense municipalities. Abington Township and Upper
52
Moreland Township were excluded because the analysis in Table 2.4.1. showed
a negative household land-use need.
Depending on the planning decisions of these municipalities accommodating
growth at higher densities in terms of housing mix and design standards (e.g.
cluster subdivisions), some of these housing units could be townhouses and
others would be cluster houses on smaller lots (e.g., 8000 ft2). Further, in this
smart growth scenario, only 1500 square feet of residential land per new resident
was assumed, in that commercial and other uses were developed at higher
intensities. The results are shown in Table 2.4.3.
Table 2.4.3. Land Development, Municipal Smart Growth Scenario
2020
Residential
Need
2030
Residential
Need
2020 Non-
Residential
Need
2030 Non-
Residential
Need
2020
Total
Need
2030
Total
Need
2020
Land
Saved
2030
Land
Saved
Acres Acres Acres Acres Acres Acres Acres Acres
Bucks County
Upper
Southampton
Twsp. 26 33 13 17 39 50 45 58
Warminster
Twsp. 116 68 68 93 184 255 115 161
Montgomery
County
Horsham Twsp. 67 36 36 60 103 170 117 194
Lower
Moreland Twsp. 12 8 8 21 20 57 36 106
Upper Dublin
Twsp. 3 1 1 2 4 5 6 8
TOTALS 224 126 126 193 350 537 320 527
The last column of Table 2.4.3. indicates that, in comparison with the trend
development scenario illustrated in Table 2.4.2., 527 additional acres of forested
and agricultural landscapes would be preserved with accommodation by each
municipality of its future residential needs at reasonably higher densities,
consistent with smart growth. Comparing these figures with the amount of land
which is suitable for development in Table 2.4.2., each municipality has more
than enough land available for this smarter development. Figure 2.4.2. shows the
projected land use futures for 2030 under Scenario 2.
Regional Smart Growth
In this third scenario, the region still accommodates its forecasted population, but
accommodates the development by sharing uses among municipalities. Future
growth needs are targeted to existing vacant (but developed) land as infill
53
development or redevelopment. It is assumed as well that the existing housing
stock of areas forecasted to lose population, are occupied, and therefore, absorb
Figure 2.4.2. Municipal Smart Growth Land Use, 2030
a significant proportion of overall population growth in the region. For example,
the estimates in Table 2.4.1. for Abington Township show an actual population
loss equivalent to 205 households. In this scenario, 205 forecasted new
households from other municipalities would instead move into the existing
housing stock in Abington, reducing the need for new construction. In this
scenario, the number of new households which could be accommodated in
existing housing units was calculated first. For the region as a whole, only 1108
new net housing units need to be constructed to 2020 and 1855 net new housing
units to 2030.
Table 2.4.4. presents an analysis of land available within the region which is
considered “vacant” (not wooded, not agricultural) and also “suitable for
development” (no environmental constraints, public sewer available.)
It should be apparent in examining Table 2.4.4., that all of the PCW’s net new
housing unit development, if accommodated at urban densities, could be easily
accommodated in the currently vacant land in the city of Philadelphia within the
watershed. If the watershed were to undertake something like a regional “transfer
54
of development rights” program, all of the projected growth needs for 30 years
could be accommodated without any additional conversion of currently
undeveloped land to urban development.
Table 2.4.4. Regional Smart Growth Scenario
Vacant Land,
Suitable for
Development (acres)
Bucks County
Upper Southampton Twsp. 44
Warminster Twsp. 77
Montgomery County
Abingdon Twsp. 50
Bryn Athyn Borough 0
Hatboro Borough 14
Horsham Twsp. 62
Jenkintown Borough 0
Lower Moreland Twsp. 23
Rockledge Borough 0
Uper Dublin Twsp. 1
Upper Moreland Twsp. 70
Philadelphia County
Philadelphia 564
TOTAL PENNYPACK
WATERSHED 905
2.5. Differential Impacts of Residential Development Scenarios
Two separate analyses were undertaken to estimate some impacts of traditional
vs. smart residential development. Section 2.5.1. estimates hydrological changes
that result from the trend and municipal smart growth scenarios in Section 2.4
above. Section 2.5.2. assesses impacts based on two slightly different
development scenarios: Smart Growth and Sprawl. In future research, it is hoped
that the analyses in Sections 2.5.1. and 2.5.2. can be unified.
2.5.1. Hydrology
Given the land-use changes forecasted in the development scenarios described
in Section 2.4., Trend Development and Smart Growth, the peak flows were
estimated with the USGS HEC-RAS model. They are shown in Table 2.5.1.
Figures 2.5.1. and 2.5.2.
55
Table 2.5.1. Peak Flows (cfs) for Varying Conditions
Trend Smart Growth
PWD Smart
Growth Flows with Flows with
Station Current Trend tc reduced
10% tc reduced
10% Flows Flows Flows
1 160 235 178 254 193
2 271 403 317 436 343
3 374 518 376 561 408
4 1303 1606 1360 1740 1476
5 403 574 438 622 476
6 1795 2277 1895 2468 2051
7 1900 2394 2010 2594 2180
8 2035 2543 2159 2756 2342
9 253 259 259 280 280
10 2640 3193 2800 3461 3036
11 248 250 250 271 271
12 2691 3250 2856 3524 3098
13 91 115 115 124 124
14 2809 3378 2982 3661 3235
15 3050 3585 3191 3886 3462
16 3141 3671 3277 3980 3556
17 3331 3816 3424 4138 3713
18 3470 3905 3511 4232 3809
19 202 202 202 214 214
20 3690 4087 3697 4436 4016
The Trend Development and Smart Growth flows are shown in Figures 2.5.1.
and 2.5.2.
56
Figure 2.5.1. Flows from Trend Development
2.5.2. Energy Use, Air and Greenhouse Gas Emissions, Water Quality and
Biological Integrity
In this section, the analysis of residential development involved four steps: (1)
desirable attributes of development behavior were transformed into suitability
criteria using GIS software to find areas into which housing can be sustainably
placed; (2) development scenarios were created; (3) the energy and
environmental impacts of the placements were estimated; and (4) the effects on
the value of ecosystem functions were estimated. Housing placed randomly in
57
the most suitable sites represents the Smart Growth scenario. Housing placed
randomly among non-restricted locations in the PCW represents the Sprawl
scenario.
Suitability Analysis for Smart Growth
The criteria used to determine the suitable areas for the Smart Growth scenario
were that development should: (A) be on a suitable site (vacant and low-density
residential land uses, though possible in other areas); (B) not be within the
restricted areas (floodplain boundary; wetland, and other water resources, parks,
open spaces, woodlands, and other protected spaces); (C) minimize the
necessity of new infrastructure and services (minimize distance to existing road
infrastructure, schools, and shopping centers); (D) minimize the impact on air
quality in the airshed (minimize contribution to traffic congestion and minimize
distance to public transit stations); & (E) minimize the impact on water quality in
the PCW (protect the riparian buffer areas, floodplains and open spaces).
More specifically, the criteria were set within the following two broad categories:
(i) Land Features: on lands free from floodplains, wetlands, ponds, and other
water resources; on impervious surface (the goal is to protect pervious surfaces);
on relatively flat land; on suitable land cover; on impermeable soil type
(Impermeable soil should be better for development, as it has minimum impact
on water transmission to soil. Soil types A (high rank) to D (low rank) were
considered.); and (ii) Infrastructure and Facilities: near public transit stops (only
rail station data was used), roads and bike trails, schools, hospitals, and parks
and open spaces. It was assumed for this analysis that the current residential
housing stock stays intact. Though zoning is recognized as an important
determinant of housing placement, it was assumed that it can be changed to
accommodate new development. In addition, the zoning ordinances of the
municipalities in the PCW have different coding systems. This makes it difficult to
create a comprehensive zoning map for this multi-municipal watershed.
The GIS software used for this study was ArcGIS 9.1 (ESRI 2005). The input
data and their sources were:
Land Cover – US Geological Survey (USGS)
Landuse – Delaware Valley Regional Planning Commission (DVRPC)
Floodplains – Center for Sustainable Communities (CSC), Temple
University Ambler
Stream Bank and Buffer – CSC
Slope – CSC
Wetlands, Ponds, and Other Water Resources – CSC
Roads – ESRI
Soil – PA Spatial Data Access (PASDA)
Impervious Surface – PASDA
Southeastern Pennsylvania Transit Authority (SEPTA) Rail Stations –
DVRPC
58
Bike Trails – DVRPC
Municipal Boundaries – DVRPC
Job Centers – DVRPC
Schools ESRI
Hospitals – ESRI
All of these datasets were available in vector format (points, lines and polygons),
except impervious surface data. Vector data were converted to raster (pixels)
and then those layers were reclassified according to the suitability criteria.
After reclassification, the layer Scale Values were determined. The Scale Values
ranged from 10 to 1 – most suitable to least suitable.
Layer: Floodplains, Wetlands, Ponds, Floodway, Wetland, Pond: Restricted
100-Year Floodplain: 1
500-Year Floodplain: 2
Other Areas: 10
Layer: Impervious Surfaces
Lowest 2 categories (in other words, highest impervious) are restricted
Layer: Slope
Lowest 4 categories (in other words, highest slope) are restricted
Layer: Land Cover
Value 0 is restricted land cover
Layer: Hydrologic Soil Group
Group B: 5
Group C: 9
Group D: 10
Layer: Streets
Value 0 is a restricted area
Finally, Layer Influence (user-chosen %s) was determined before the final raster
calculation. These are shown on Table 2.5.2.. The map of suitable sites for the
Smart Growth scenario is shown in Figure 2.5.2.
Suitability Analysis for Sprawl
In this scenario, only the restricted areas were excluded from the total watershed
in order to keep the other areas open for new development or re-development as
much as possible. Restricted areas are those that are restricted by state or local
municipal law, such as state or county parks and open spaces, floodways, and
other protected lands. The output of this analysis is shown in Figure 2.5.3.
59
Table 2.5.2. Relative Importance of Suitability Criteria
Layer Influence
Floodplains, Wetlands, Ponds .30
Impervious Surfaces .15
Slope .05
Land Cover .10
Soil .10
Transit Stops (rail station) .10
Roads .10
Schools .02
Hospitals .02
Parks and Open Space .02
Trails .04
Figure 2.5.2. Sites for Smart Growth: Darker Is More Suitable
60
Figure 2.5.3. Sites for Sprawl: All but Dark Sites Acceptable
Residential Development
The 1855 units in Table 2.4.1. projected to be needed by 2030 were assumed to
be placed in 2006, and to be distributed over the suburban municipalities and the
part of Philadelphia in the watershed. The building units were placed at various
sites for the two scenarios. Hawth’s Tool, an extension of ArcGIS, was used for
generating random points representing the building units. For the Smart Growth
Scenario, 1855 points were randomly generated in the highest suitable areas
(with ranks 10 and 9). The following steps accomplished this task: create a
polygon file from the suitability output raster file; dissolve according to suitability
categories; choose only areas with 10 and 9 and create a new layer; generate
random points for the whole area (using Hawth’s Tool); select only those points
that are inside areas with 10 and 9 ranking, and randomly choose 1855 points
from the selection. Once the building units were placed, distances were
calculated from these points to the nearest commuter rail stations (points) using
Hawth’s Tool. The method was same for the sprawl scenario. The only difference
61
from the Smart Growth selection process is that random points were generated in
areas suitable for the sprawl scenario, i.e., all unrestricted areas.
Energy and Environmental Impacts
As this work deals with residential housing placement, it was assumed that the
same type of housing would be placed in any location chosen. As the goal was to
compare placement in the Sprawl scenario to that in the Smart Growth scenario,
the assumption of the same type of housing obviates the need to specify the
structural, energy-use and emissions characteristics of the houses themselves. It
was also assumed that residents in any location will drive the same class of
automobile. The energy-use and emissions characteristics of the representative
vehicle are given below.
Auto Energy Use, Greenhouse Gases and Air Quality
To calculate energy use and greenhouse gas and criteria air pollutant emissions
of miles driven by residents in the two scenarios, a beta version of the GREET
1.7 program created by the Argonne National Lab’s Center for Transportation
Research (Wang; 2005) was used. Based on fuel type, technology type, market
share and many other parameters chosen by the user, the program can generate
“well-to-pump,” “well-to-tank” and “well-to-wheels” estimates for energy use and
greenhouse gas and criteria air emissions.
The estimates used here are for a car with a conventional spark-ignition (SI)
engine using conventional gasoline (CG) or reformulated gasoline (RFG). Table
2.5.3. shows some parameters used.
Table 2.5.3. Fuel Economy & Emissions Rates of Baseline Vehicles
Items SI Vehicle: CG & RFG
MPG 24.8
Exhaust VOC (all g/mi) .122
Evaporative VOC .058
CO 3.745
NO
x
.141
Exhaust PM10 .0081
Brake & Tire Wear PM10 .0205
CH4 .0146
N2O .012
The results of the GREET run in terms of energy use and emissions are given in
Table 2.5.4.
62
Table 2.5.4.Energy Use & Emissions from a Typical Automobile
Gasoline Vehicle: CG and RFG
Btu/mile or grams/mile
Vehicle
Operation Total Item Feedstock Fuel
Total Energy 188 981 4,912 6,081.497
Fossil Fuels 181 968 4,835 5,983.208
Petroleum 60 470 4,835 5,364.766
CO2 19 73 368 459.778
CH4 0.448 0.085 0.021 0.553
N2O 0.000 0.005 0.012 0.018
GHGs 29 77 372 477.733
VOC: Total 0.017 0.116 0.218 0.351
CO: Total 0.041 0.040 4.917 4.998
NOx: Total 0.117 0.134 0.268 0.518
PM10: Total 0.010 0.038 0.029 0.077
SOx: Total 0.047 0.109 0.006 0.162
VOC: Urban 0.003 0.074 0.135 0.212
CO: Urban 0.001 0.018 3.058 3.078
NOx: Urban 0.006 0.057 0.167 0.229
PM10: Urban 0.000 0.004 0.018 0.022
SOx: Urban 0.004 0.048 0.004 0.057
In both scenarios, automobile energy use and greenhouse gas and criteria air
pollution emissions are computed by multiplying the GREET numbers in Table
2.5.4. by the number of miles residents would drive when they are placed in
chosen locations. Though this could have been done for any or all of the other
types of infrastructure (e.g., roads, schools), distances from the chosen housing
sites to the nearest commuter rail stations were calculated. Assuming two
trips/day, five days/week and forty-nine weeks/year yielded 1,894,340 miles for
Sprawl and 933,940 miles for Smart Growth. The results are shown in Table
2.5.5. Columns 3 and 4 show the products of the items in column 2 and the miles
computed for the two scenarios. The Smart growth miles were 50.7% of the
Sprawl miles.
Suitability Analysis for Sprawl
In this scenario, only the restricted areas were excluded from the total watershed
in order to keep the other areas open for new development or re-development as
much as possible. Restricted areas are those that are restricted by state or local
municipal law, such as state or county parks and open spaces, floodways, and
other protected lands. The output of this analysis is shown in Figure 2.5.3.
63
Table 2.5.5. Energy Use & Emissions in Sprawl & Smart Growth
Gasoline Vehicle: CG and RFG
Miles in Smart
Growth Btu/mile or Miles in Sprawl
Item grams/mile 1894340 933940
Quantities Per Mile Energy Use/Emissions Difference
Total
Energy 6,081.50 11,520,423,027 5,679,753,308 5,840,669,719
Fossil Fuels 5,983.21 11,334,230,243 5,587,957,280 5,746,272,963
Petroleum 5,364.77 10,162,690,824 5,010,369,558 5,152,321,266
CO2 459.778 870,975,857 429,405,065 441,570,791
CH4 0.553 1,047,570 516,469 531,101
N2O 0.018 34,098 16,811 17,287
Total GHGs 477.733 904,988,731 446,173,958 458,814,773
VOC: Total 0.351 664,913 327,813 337,100
CO: Total 4.998 9,467,911 4,667,832 4,800,079
NOx: Total 0.518 981,268 483,781 497,487
PM10: Total 0.077 145,864 71,913 73,951
SOx: Total 0.162 306,883 151,298 155,585
VOC: Urban 0.212 401,600 197,995 203,605
CO: Urban 3.078 5,830,779 2,874,667 2,956,111
NOx: Urban 0.229 433,804 213,872 219,932
PM10:
Urban 0.022 41,675 20,547 21,129
SOx: Urban 0.057 107,977 53,235 54,743
Water Quality and Biological Integrity
As noted in Sections 2.3.2. and 2.3.3. the PWD (2003) provided a chemical
analysis and a biological integrity assessment of the water for each of their 20
monitoring stations located along the Pennypack Creek. For each station, the
water quality index was computed using the components and their target levels.
Each measurement was divided by the target level for that component. Since
some components are negative to water quality (by their nature, or if they exceed
the target), a minus sign was added. The resulting “signed” ratios were added
over the components. Due to the relative magnitudes of the ratios, the sums are
invariably negative. Hence, water quality is highest at those stations with smaller
negative values. Ratios of the observed components of water quality to target
levels are shown in Table A.3.6. The Water Quality Index in Table 2.5.5.. is
simply the sum of the row sum for each station.
The biological integrity values from Tables A.3.7. and A.3.8. are combined and
presented in Table2.5.7. It should be pointed out that the presence of .0000 in
the table implies different things for macroinvertibrates and fish. For
macroinvertibrates, it means that no significant numbers were found at a site. For
fish, it is due to the fact that only 10 stations were sampled. The Total Biological
64
Integrity was gotten by choosing the only non-zero number in the row, or by
averaging the numbers in the row.
Table 2.5.6. Water Quality Index
Water
Quality
Station Index
1 -38.03
2 -36.10
3 -50.82
4 -50.82
5 -40.47
6 -40.47
7 -43.29
8 -36.36
9 -36.36
10 -47.39
11 -23.04
12 -32.31
13 -32.30
14 -63.55
15 -63.55
16 -31.34
17 -79.12
18 -174.51
19 -174.51
20 -22.55
65
Table 2.5.7. Biological Integrity
Macro- Total
Invertebrates Fish Biological
Integrity
(%)
Biological
Integrity (%) Biological
Integrity (%) Station
1 0.0000 0.6800 0.6800
2 0.4000 0.0000 0.4
3 0.0667 0.0000 0.067
4 0.0000 0.7600 0.76
5 0.0000 0.7600 0.76
6 0.0000 0.0000 0
7 0.0000 0.5600 0.56
8 0.4000 0.0000 0.4
9 0.1333 0.6400 0.38665
10 0.2667 0.0000 0.2667
11 0.4667 0.0000
0.4667
12 0.4000 0.0000 0.4
13 0.0000 0.0000 0
14 0.1333 0.5600 0.34665
15 0.2000 0.5200 0.36
16 0.4000 0.0000 0.4
17 0.0000 0.4800 0.48
18 0.4000 0.0000 0.4
19 0.4000 0.4800 0.44
20 0.6667 0.4800 0.57335
Analytical relationships between various land uses and water quality and
biological integrity were not established. To gauge the water-related impacts of
the Smart Growth and Sprawl scenarios, the Water Quality Index and the
Biological Integrity were “weighted” by the placement of the housing units. The
water quality and biological integrity assigned to the stations were then attributed
to the sub-basins. The number of housing units in each sub-basin served as a
weight by which to multiply the sub-basin water quality and biological integrity.
The result was a Weighted Water Quality and Weighted Biological Integrity. The
results of the calculations are summarized in Table 2.5.9.
Reduction in the Value of Ecosystem Services
Absent an explicit inventory created by the PCW research team, the general list
of these services given by deGroot, Wilson and Boumans (2002) was used.
These authors collected a range of values for these services as estimated in
66
other studies. Table 2.5.8. shows the ecosystem services and the midpoint
$/acre values of the ranges provided by the authors.
Table 2.5.8. Value of Ecosystem Functions/Services
Service Midpoint
Category Service ($/acre)
Regulation Gas regulation 55
Climate regulation 63
Disturbance regulation 1465
Water regulation 1102
Water supply 1538
Soil retention 55
Soil formation 2
Nutrient cycling 4287
Waste treatment 1367
Pollination 8
Biological control 16
Habitat Refugium function 309
Nursery function 68
Production Food 560
Raw materials 206
Genetic resources 24
Medicinal resources n.a.
Ornamental resources 30
Information Aesthetic 358
Recreational & tourist 1214
Cultural & artistic n.a.
Spiritual & historic 5
Scientific & Educational n.a.
Total Ecosystem Service Value $12732
Source: Adapted from deGroot, Wilson and Boumans (2002)
Without a model to link water-related attributes to ecosystem services, the latter
were treated in the aggregate. The percentage relative reduction in total
ecosystem service value (TESV) will be taken as an average of the percentages
by which the Sprawl scenario impacts exceed those of the Smart Growth. This
average, giving Sprawl credit for having a better WWQ, is (.507 + .533 -.212)/3 =
.277. If the sum of the midpoints of the value ranges provided by deGroot, Wilson
& Boumans (2002) are updated from the 1st half of 1994 to the 1st half of 2006 by
the all-item, urban CPI, then the average relative reduction in the TESV is
($12732/acre)*(1.36)*(.277)*(1029 acre) = $4,935,495. Table 2.5.9. contains a
finer breakdown.
67
Table 2.5.9. Impacts Summary
Smart
Growth Scenario Sprawl
Total Energy Use (bbl of oil) 2300 1140
Greenhouse Emissions (short
tons/yr) CO2 960.09 473.34
CH4 1.15 0.57
N2O 0.038 0.019
GHG:
Total 997.58 491.82
VOC:
Total Air Emissions (short tons/yr) 0.73 0.36
CO:
Total 10.44 5.15
NOx:
Total 1.08 0.53
PM10 :
Total 0.16 0.08
SOx:
Total 0.34 0.17
Grnhse & Air % Net Reduction in
TESV 50.70%
$9,033,559
Grnhse & Air $$ Net Reduction in TEFSV
Weighted Water Quality "Index" -104075 -124395
Water % Net Reduction in TESV 19.50%
Water $$ Net Reduction in TESV $3,474,446
Biological Integrity 450 684
Bio % Net Reduction in TESV 52.00%
Bio $$ Net Reduction in TESV $9,265,188
Total $$ Net Reduction in TESV $18,298,747 $3,474,445
The results clearly show that the Sprawl Scenario generates more than five times
the ecosystem value reduction than that caused by Smart Growth.
2.6. Pennypack Ecological Vulnerability Assessment (PEVA)
The PEVA team found three particular sources of vulnerability: the main
Philadelphia intake, the PCW-resident sewage treatment plant and PCW
stormwater management.
68
Philadelphia Water Supply
In 2002, the PWD (2002; p. 14) stated that, “The Baxter Water Treatment Plant
provides treated water that comes from the Delaware River. … Particular
tributaries that require special attention to address polluted runoff from
urban/residential areas and agricultural lands include the Pennypack Creek …”
The 600-mgd Baxter Plant is located just up-stream from the confluence of the
Pennypack Creek and the tidal Delaware River, and is the primary source of
Philadelphia water supply. The water supply is vulnerable to PCW impairment
because at flood tides, the Pennypack water moves up river.
The Sewage Treatment Plant
It is apparent that the UM-H STP is severely impacting aquatic life as
concentrations in its discharges far exceed recommended limits. Considering the
land use scenarios conducted as part of this analysis, increasing wastewater
flows and sewage to this plant, which is operating near capacity, will only
complicate nutrient removal and further impair aquatic life and recreation in and
along the Pennypack Creek downstream of the STP. Publicly Owned Treatment
Works (POTWs) are least efficient when operated at capacity because the
hydraulic retention time in primary and secondary clarifiers is reduced, limiting
the ability of plants to settle out solids, including nutrients. Unless the plant is
expanded and outfitted with nutrient-removal technology, water quality would
continue to degrade. As part of its PCW Study (Meenar 2006), the CSC research
team recommended that the Upper Moreland-Hatboro Joint Sewer Authority
conduct a feasibility study to evaluate possible upgrades to significantly improve
the plant’s performance in reducing nutrient levels in its effluent.
Both the Trend and Smart Growth scenarios assume that 1855 new structures
will be built in the watershed. If they are all built within the service area of the
WTP, this would increase flows to the plant by about 0.3 mgd for both scenarios.
(1855 units x 2.66 people per unit x 60 gallons per capita per day of sanitary
water usage = 296,058 gpd). Additional commercial development would also
increase flows and the need for additional treatment capacity." This combined
with a reduction in baseflow stemming from additional ground water withdrawals
in the watershed would further exacerbate the nutrient problems downstream of
the WTP. The receiving stream would have less flow and the larger volume of
effluent would increase loadings of nitrate and phosphate absent a nutrient
removal program. This would further degrade aquatic habitat in the Pennypack
Creek.
Unless the STP improves its nutrient removal, other efforts to improve stream
water quality will only provide nominal improvements. treatment options include
biological removal or chemical additives.
69
Stormwater Management
It was not until the 1970s that requirements for controlling the quantity or quality
of runoff coming from a developed site were a consideration. Water was routed
from a site to the nearest stream in the most expedient manner. This increased
volume of water accelerates erosion and sedimentation and destroys stream
habitat. From the 1970s until recently, stormwater management relied primarily
on the use of detention basins to manage stormwater. While these basins
controlled the peak flows of water, they did not reduce the overall volumes of
runoff and did nothing to address or improve the water quality of runoff.
Stormwater BMPs and regulatory requirements are improving stormwater
management in new developments. However, these actions do little to address
the sins of the past. Water quality and quantity issues in older developments will
require the retrofitting of existing stormwater facilities or the installation of
stormwater controls where none exist in order to reduce the runoff volumes.
Many existing stormwater management facilities in the PCW have become
completely or partially dysfunctional because of poor or no maintenance,
ineffective design, or a combination of such factors. Some of these were
obviously constructed many years ago, as evidenced by their filling by sediments
and debris and the abundant tree growth within them. However, many have been
constructed in recent years, and some have been observed that are quite new,
but which are already evidencing poor performance.
In the summer of 2004 through spring of 2005, a visual assessment of the entire
PCW was performed by Temple researchers in order to get a full picture of what
was actually happening on the ground within the creek’s watershed and its
surrounding riparian corridors. The reconnaissance was conducted mostly on
foot, but often had to resort to “windshield survey” methods, especially in areas
such as residential subdivisions or industrial properties where access was
severely limited or completely prohibited.
This assessment evaluated the condition and functionality of existing stormwater
facilities, assessed the potential for retrofitting such facilities so as to improve
both their environmental and flood control performance, and sought locations for
recommended new stormwater Best Management Practices (BMPs).
Field observations were made at a total of 421 locations within the seven
suburban sub-basins into which the entire watershed study area was divided for
hydrologic and hydraulic modeling purposes. Whatever their age, many
dysfunctional or poorly functioning facilities, whether with respect to management
of discharge rates, volumes, or water quality, have been identified on the
abovementioned maps as having the recommended “Highest Priority” or “High
Priority” for renovation, redesign, and/or retrofitting.
70
Among the 421 observation locations, the CSC has identified that only 73
locations have some form of stormwater management facilities. However, 48 of
these existing facilities are, in the judgment of the study team, either completely
dysfunctional or performing poorly.
One strategy being employed in urbanized watersheds is to retrofit existing
stormwater structures to better control stormwater volume and to improve water
quality. However, public perception and acceptance of stormwater retrofits
cannot be taken for granted. The recent spread of West Nile virus has raised
public concern regarding perceived mosquito breeding sites; basin naturalization
and changes in basin/site hydrology tend to tap into this anxiety despite
abundant evidence that facilities such as wet ponds or constructed wetlands can
provide habitat for insect predators and are actually less likely to harbor such
disease vectors. New stormwater management approaches need to be carefully
explained and resident concerns must be addressed.
The need for education and demonstration of successful stormwater retrofits is
essential to illustrating the effectiveness of such BMPs and alleviating these
public misconceptions, all while improving the conditions within these urbanized
watersheds. Opportunities that municipalities can take advantage of immediately
are those which they can implement on publicly owned lands. Each municipality
owns and/or manages sites that could be potentially “retrofitted” with some form
of BMPs. These sites would not only reduce stormwater impacts, but also serve
as model sites within the region.
The Urban Storm Water Workgroup of the Chesapeake Bay Program compiled
data on the pollutant removal efficiencies of urban storm water management
BMPs. While the actual performance of specific BMP installations varies, the
Workgroup found that practices that could be used in parking areas such as
porous pavement, bioretention areas and infiltration trenches had pollutant
removal efficiencies for Total Suspended Solids (TSS) of 85 to 90% and for Total
Nitrogen (TN) and Total Phosphorous (TP) of 40 to 70%. Conventional detention
basins have pollutant removal efficiencies for TSS of only 10% and for TN and
TP of 5 to 10%. In contrast, practices that might be used to replace or retrofit dry
detention basins had significantly higher pollutant removal efficiencies. For
example, a dry extended detention basin had a pollutant removal efficiency of
60% for TSS, 20% for TN and 30% for TP. A wet pond had an efficiency of 80%
for TSS, 30% for TN and 50% for TP.
The overall watershed stormwater runoff can be controlled through the effective
control of individual sub-basin stormwater runoff. Retrofitting existing stormwater
facilities as well as areas developed prior to the implementation of any
stormwater management controls with BMPs is the key to reducing water quality
and quantity problems within the Pennypack Watershed. The CSC researchers
recommend that each municipality create a stormwater management utility to
provide sufficient revenues to fund such retrofits, to efficiently operate and
71
maintain all stormwater facilities, and to ensure preservation of critical areas that
perform vital stormwater management functions. Furthermore, each municipality
should concentrate its initial efforts on implementing retrofits at the priority sites
within each sub-basin where BMPs can have a significant and cost-effective
impact on controlling stormwater runoff.
2.7. Relation of PEVA to Pennsylvania’s Sustainability Indicators
The Pennsylvania Consortium for Interdisciplinary Environmental Policy (PCIEP;
2004) created the list of environmental indicators shown in Table 2.7.1.
Indicator #1 clearly includes Pennypack Creek. It was found to be moderately
impaired, and a return to a more natural “designated use” will require some
adjustments. Indicator #3. can be related to the Pennypack in at least two ways:
the UM-H STP and landowners may be considered to be using the Creek’s
capacity to receive waste discharges and/or runoff beyond the “sustainable yield”
of the Creek remaining healthy. Indicator #5 applies directly to the PCW – Trend
or Sprawl development will have greater impacts on air, climate and water than
the Smart development. Indicators #6 and #7 are directly affected by increased
peak flows and impairment of the water in the Pennypack. Indicator #9 directly
addresses the impairment of surface water. Besides the STP, the PCW
experiences considerable non-point source pollution. The fuel-use impacts
estimated in Section 2.5.2. directly address Indicator #13. The CSC has shown
that Indicator #14 related to the PCW with respect to property damage and
flooding. It was determined that there are 738 buildings in the 100-year
floodplain. (Meenar 2006) Finally, Indicators #15 and #16 are relevant to the
PCW because the local municipalities are cooperating on sustainable stormwater
management, and the Pennypack Trust under the Directorship of Dr. David
Robertson is spearheading forest, wetlands and Creek restoration.
The research work presented in the previous sections can help a watershed
management group to define PCW sustainability goals and indicators more
refined than the general ones listed in Table 2.7.1. This work remains, and can
be accomplished with future funding.
72
Table 2.7.1. PA Sustainability Indicators
1: # of lakes & surface stream miles supporting
their designated use for aquatic life
GOAL 1:
Sustain, conserve, protect,
enhance, & restore PA’s
environment, natural resources, &
ecological diversity
2: # of designated groundwater sampling points
for each ground water basin located in a
watershed that meets primary drinking water
standards
3: # of water resources being used beyond their
sustainable yield
4: Acres of land by use
5: # of days & # of Pennsylvanians affected when
air quality does not meet health standards
6: Index of ecological diversity
7: Ecosystems or species threatened by
environmental conditions
8: Quantity of waste (by type) generated,
recycled, reused & eliminated
9: Quantity of pollutants released into air, land &
water
GOAL 2:
Reduce harmful effects from
environmental contaminants &
conditions.
10: Annual mean pH of PA precipitation
11: # of public water supply systems meeting all
drinking water standards & maximum contaminant
levels
12. Quantity of waste disposed (by type)
13: Energy use by fuel type
14: Lives lost & property damage from flooding &
mining
15: # of businesses/commercial activities,
government agencies, communities & individuals
implementing sustainable practices
GOAL3:
Engage all Pennsylvanians as
active & informed stewards of the
environment. 16: # of community-based groups performing
activities towards improving their environments
17: Level of environmental literacy of
Pennsylvanians
3. Potential Applications to Other Watersheds
The methodology of the Pennypack Ecological Vulnerability Assessment (PEVA)
study can be easily applied to other watersheds of similar size. The major steps
followed in the PEVA study are: (a) development of a GIS data inventory; (b)
assessment of the state of the watershed; (c) projection of alternate future
landuse scenarios; and (d) assessment of differential impacts of the scenarios.
The summary of the PEVA study work flow has been graphically represented in
Figure 3.1.
73
Figure 3.1 Generic Ecological Vulnerability Assessment Workflow
74
3.1. Data and Metadata
A watershed study team should include researchers from various disciplines,
including environmental and landuse planning, civil and environmental
engineering, ecology, economics, geology, geography, and landscape
architecture. The research team should consider developing a central GIS
database that will include spatial and non-spatial data and metadata. The
database should be available to every researcher.
The database may include basically two types of data: primary and secondary.
Primary datasets can be created using field surveys, digitizing, aero-triangulation,
or LiDAR technologies. Examples of primary datasets can be high resolution
elevation data, including DEM and contour and high resolution ortho-photos. The
PEVA team hired a consultant for creating the high-resolution (2 ft interval)
elevation data and other features, such as stream, bridges and culverts, dams,
and lakes. The building footprint data can be created by digitizing the footprints
from ortho-photos. Primary data can be also collected from a watershed-wide
BMP survey, a water quality survey, and a biological data survey.
The secondary data may be collected from a number of local, regional, state, or
federal agencies, as well as other organizations and online sources. Such types
of data may include demographic data from the U.S. Census Bureau, soil and
geology data from local/regional river basin commissions, landuse and
transportation data from regional planning agencies, parcel and zoning data from
local municipalities, and land cover and tree canopy density data from the US
Geological Survey (USGS). A number of free GIS datasets are also available
from ESRI or other online data provides, statewide GIS data warehouses (e.g.,
the Pennsylvania Spatial Data Access (PASDA), the New York State GIS
Clearinghouse, and the Virginia Geographic Information Network), university
sponsored GIS data warehouses (e.g., the Cornell University Geospatial
Information Repository), and local non-profit agencies and conservancy groups.
Both primary and secondary datasets should be stored in the central GIS
database with required editing and processing. Metadata should be developed
for each type of dataset. Finally, the datasets can be rearranged according to
some broader categories: biological, land feature, and hydrological.
In order to undertake a more refined assessment of the watershed, the study
team may emulate the use by EPA’s ReVA Program of the USGS HUCs . The
watershed can be subdivided into a number of smaller sub-watersheds or sub-
basins, which would correspond to the size and location of the first-order streams
within its boundaries. Sub-basins can be delineated from stream line files based
on stream order and topographic elevation data using the Watershed Modeling
System (WMS) 7.1 and HEC-GeoRas software.
75
Once the sub-basins are delineated, all of the datasets can be analyzed and geo-
processed to re-assign watershed wide data in each of the smaller sub-basins.
The sub-basins should be given unique IDs before this analysis is performed.
Metadata should be updated for each type of data.
The PEVA assessment was done for the following ecological indicators: water
volume, water quality, biological integrity, and impervious surface. Depending on
data availability, more indicators can be assessed in other watersheds. Thorough
analyses should be done for each of the indicators along with results and
interpretations. The GIS maps and tables showing the results in each sub-basin
will help in assessing the overall current state of any watershed.
3.2. Models
In conducting the analyses for the project, the PEVA research team used several
hydrologic models. Two were developed by the Army Corps of Engineers, in
particular its Hydrologic Engineering Center (HEC), and are regularly used for
hydrologic analyses around the United States. They are HEC-HMS and HEC-
RAS. (USACE 2007)
The Hydrologic Modeling System (HEC-HMS) is designed to simulate the
precipitation-runoff processes of complex watershed systems. It can be applied
to a wide range of geographic areas for solving a broad range of problems. This
includes larger river basin flood hydrology, and small urban or natural watershed
runoff. Hydrographs produced by the program are used directly or in conjunction
with other software for studies of water availability, urban drainage, flow
forecasting, future urbanization impact, reservoir spillway design, flood damage
reduction, floodplain regulation, and systems operation. (USACE 2007)
The WMS software can be used in conjunction with the Corps software as it has
the ability to interface with GIS data. WMS is a graphical modeling package to be
used for watershed hydrology and hydraulics. WMS has embedded HEC-HMS
and HEC-RAS.
The HEC-RAS system contains four one-dimensional river analysis components
for: (1) steady flow water surface profile computations; (2) unsteady flow
simulation; (3) movable boundary sediment transport computations; and (4)
water quality analysis. All four components use a common geometric data
representation and common geometric and hydraulic computation routines. In
addition to the four river analysis components, HEC RAS contains several
hydraulic design features that can be invoked once the basic water surface
profiles are computed.
While these models are readily available, they require considerable expertise to
operate. Users who are unfamiliar with such models should take at least one
introductory course before attempting to use them. The HEC offers several
76
courses on a regular basis. In Pennsylvania, Villanova University sponsors short
courses taught to introduce these models to prospective users.
Streamflow and Baseflow Information
The lack of stream gauges can pose a problem for conducting ReVA-type
analyses for watersheds without them. In such cases, researchers must use
hydrologic values from adjacent and similar watersheds to predict runoff, both for
low flows and extreme events.
Stream gauge data with extensive periods of record enable users to calibrate
models to actual recorded events. Once this is done, watersheds can be
disaggregated into many smaller ones and calibrated again. This allows for an
accurate representation of hydrologic values at a small scale.
Accurate baseflow information can be secured by conducting hydrograph
separations of recorded streamflow data. There are numerous methods for do
so, including the local minimum method, which was described earlier. In the
absence of stream gauge data, information can be obtained from the USGS or
PA DEP. As part of the Act 220 State Water Planning process, the PA DEP has
extended the work done by the USGS for the DRBC to the rest of the
Commonwealth. Through use of a software tool called WAVE, baseflow and
water withdrawal information can be obtained for all watershed areas. Users then
can estimate natural baseflows and current flows, the latter by subtracting water
withdrawals and inter-basin transfers. While this is far from precise, it can give
decision makers a better understanding of the impacts of land and water
development on low flow conditions.
In predicting the runoff for extreme events like the 100-year storm, precipitation
data can be obtained from NOAA and its Atlas 14, which is widely available
throughout the Commonwealth. These data are preferred to the data available in
the older Technical Paper (TP) 40 study published by the U.S. Weather Bureau.
(Herschfield 1961) As part of its Pennypack Creek Watershed Study, the CSC
learned that the TP 40 values, which are widely used throughout the U.S., are no
longer valid as they systematically underestimated the extreme precipitation
events. This was due to a number of factors: the short average duration of the
precipitation records analyzed; the relatively small number of weather stations;
and the statistical distribution used to analyze the data. The NOAA precipitation
data vary from place to place within watersheds, but are generally 10-20% higher
than the values contained in TP 40.
3.3. Scenario Generation
Based on existing data and projected population data, a number of alternate
future landuse scenarios can be generated. At a minimum level, two scenarios
should be generated: trend development and smart growth.
77
In preparation of these land use futures, the PEVA research team had a
conversation with Ms. Megan Mehaffey of US EPA ReVA office (June 5, 2006)
in regards to the various approaches to land use change modeling undertaken in
other ReVA studies. Ms. Mehaffey indicated that there is no one standard
approach, and that the techniques and approaches vary with the size of the
region, the environmental focus of the study, and the type of data available.
Many of the modeling approaches in ReVA -type studies have been utilized at,
and are more appropriate to, larger-scale watersheds. One of the purposes of the
present project is to evaluate the suitability of ReVA methodologies in smaller
watersheds. Based on the PEVA project team’s review of previous studies and
the academic literature, it was concluded that standard, off-the-shelf land use
forecasting programs and models were unsuited to smaller watersheds. As well,
many of these standard programs and methodologies would not make good use
of the higher quality land use data collected for this project. In smaller scale
projects, ReVA-type analyses have utilized standard planning support systems
software such as “What If?”, “INDEX”, or “CommunityViz”. The methodology
described below utilized similar techniques, but without requiring any of these
programs.
Land use change models can be simplistically broken down into two approaches:
“demand driven” and “rule driven.” Many of the larger land use change models,
including some used by NASA, USGS, and EPA (such as SLEUTH, Gigapolis,
Clarke Growth Models, and Cellular Automota models) forecast land use change
based mostly on the physical attributes of land (slope, wetlands, etc), and
distance to growth inducers/growth repelers (usually roads). Though often quite
complex and sophisticated, these models simulate urban growth as a series of
algorithms or rules without statistical calibration on previous land use change in a
region. One advantage of these models is that they require relatively few inputs,
usually the easily available USGS Digital Elevation Models (DEM) and the
National Land Cover Data. The output of the models is usually a variable
indicating whether a particular pixel is developed or not, with some models
allowing the intensity of development to vary between high, medium and low.
Demand driven models, more common in the urban planning field, start with
population and economic growth forecasts. Residential uses and jobs are the
prime drivers of urbanized land uses, and therefore, are the “demand” factors in
these models. In these approaches, the population and employment forecasts
are converted into demand for urbanized land of different types and at different
densities/intensities. Users and decision makers can, interactively, vary the
assumptions about the density/intensity of land use to accommodate projected
land use demand. The models then allocate future urban land uses based on any
number of criteria specific to the model or approach, including availability of
infrastructure and underlying soil and physical suitability. Most simulation models
allow the users to specify lands excluded from development (such as near
78
streams, wetlands, steep slopes, etc.) and to direct/prioritize the growth
allocation (such as contiguous with existing development, in areas with existing
infrastructure, such as public water supply, sewers, and roads). The allocation of
land use demand is iterative, and allows decision makers to test the viability and
impacts of various scenarios. However, one of the difficulties of most demand-
based simulation software is that the assumptions and allocation techniques of
many software packages are “black box” approaches where the user is not
certain of allocation criteria.
The approach taken in PEVA study was to utilize the approach and techniques of
the “demand driven” modeling strategy, and to do so with transparent models
and techniques. In part, this decision was driven by the large amount of high-
quality, high-resolution data collected for the Pennypack region. The data
collected would not usually be utilized in some off-the-shelf software packages. A
goal in this work is to demonstrate that ReVA-type analyses could be performed
using standard land use planning analysis techniques and basic GIS software.
3.4. Impacts Assessment
In an ideal world, there would be a wealth of natural science, demographic and
economic data to feed a detailed, comprehensive and integrated ecologic-
economic model of the region under study. It is fortunate that the PCW has been
the subject of considerable data accumulation, but there is no detailed ecologic-
economic model that can generate impacts of actions taken to perturbate the
fundamental variables. In the absence of such a model, impacts must be
assessed in a piecemeal fashion. The hydrologic models can estimate changes
in base and peak flows, and a qualitative description of potential impacts beyond
the physical and chemical changes may serve to paint a rough picture for
decision support. In the analysis above, the hydrological impacts were estimated
using one set of residential development scenarios, and the air- and water-
related impacts from two related, but different scenarios. If the resources are
available, then these should be combined.
The GREET model (Wang 2005) was very useful for energy use and air and
greenhouse gas emissions, though there is some sacrifice in specificity with
respect to the geographic locale. For water quality and biological integrity, the
literature provides significant guidance. The weighted water quality index above
was the ReVA method of simple sum, with the adjustment made for negative
impacts. The biological integrity was fairly standard. Both were constrained by
the paucity of PCW data.
79
4. The Involvement of Local Officials
As noted in Section 1. above, the PCW research team included some individuals
from government and local NGOs. While not explicitly including local municipal
officials in the ReVA-related work, the CSC has been cooperating with a group of
local officials on floodplain realignment and best stormwater management
practices. Table 4.1. gives the list of suburban municipalities and the designated
official who participated in the effort. It is expected that these same officials will
be receptive to decision support regarding ecological vulnerabilities in the PCW
that may affect their constituents. Once feedback is gotten on the research
presented above, the next step for the CSC is to make local officials aware of
what was done.
Table 4.1. Municipalities Involved in PCW Stormwater Management
Township/Borough Name Title
Abington Township Burton T. Conway Manager
Bryn Athyn Borough Vikki Trost Manager
Hatboro Borough James Gardner Manager
Horsham Township Michael J. McGee Manager
Jenkintown Borough Edwin Geissler Manager
Lower Moreland Township Alison D. Rudolf Manager
Rockledge Borough Michael J. Hartey Manager
Upper Dublin Township Paul A. Leonard Manager
Upper Moreland Township David Dodies Manager
Upper Southampton Township Joseph W. Golden Manager
Warminster Township Barbara Sultzbach Assistant Manager
5. Conclusions
The PEVA team has concluded that the ReVA modeling approach can be scaled
down so as to be informative and appropriate for smaller watershed assessments
in the Commonwealth of Pennsylvania. The basic steps are: (a) development of
a GIS data inventory; (b) assessment of the state of the watershed; (c) projection
of alternative future land-use scenarios; and (d) assessment of differential
impacts of the scenarios. While many larger-scale ReVA assessment tools and
data variables are too coarse for assessment at the local level, it is feasible to
use such processes with more refined local data to accurately outline the impacts
of alternative land use and resource allocation decisions on ecological and other
watershed attributes.
The approach developed for the Pennypack Creek Watershed (PCW) also can
be used as a template for other watersheds. While the PCW is a seriously
impaired watershed in a dense urban setting in the Greater Philadelphia region,
the assessment protocol outlined in this study can be accurately applied for less
developed and more pristine watersheds as well. The report outlines data needs
80
and analytical tools for this information transfer. The PEVA team urges others
conducting such studies to create a multi-disciplinary study team and consider
developing a central and accessible GIS database that includes both spatial and
non-spatial data.
Finally, the ReVA modeling process also can be used to develop watershed
sustainability indicators. The PEVA team outlined four broad indicator categories
that can be applied elsewhere and aggregated to larger regions in Pennsylvania.
It should be noted, however, that for many water quality and biological variables,
good time series information is not available and linking land-use change to
changes in water quality and biological diversity is difficult. Fortunately, for other
variables, such as water volume and impervious surface, new models and better
topographic and hydrologic information allow researchers to more accurately
assess them under different future conditions.
81
References
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. (1999). Rapid
Bioassessment Protocols for Use in Streams and Wadeable Rivers: Periphyton,
Benthic Macroinvertebrates and Fish, Second Edition. EPA 841-B-99-002. US
EPA; Office of Water; Washington, DC.
BASIN. (2002). BASIN Water Quality Terminology. Retrieved from
http://bcn.boulder.co.us/basin/natural/wqterms.html on August 10, 2005.
Boulder Area Sustainability Information Network.
Capiella, Karen and Brown, Kenneth (2002). Impervious Cover and Land Use in
the Chesapeake Bay Watershed. Prepared for the U.S. EPA Chesapeake Bay
Program, Unpublished.
Daniels, Tom and Katherine Daniels. (2003). The Environmental Planning
Handbook for Sustainable Communities and Regions. Planners Press, Chicago,
IL.
Davis, W. S. and T. P. Simon (eds) (1995). Biological assessment and criteria:
tools for water resource planning and decision making. Lewis Publishers, Boca
Raton, FL.
de Groot, Rudolf S., Matthew A. Wilson and Roelof M.J. Boumans. (2002).
A typology for the classification, description and valuation of ecosystem
functions, goods and services. Ecological Economics, Vol. 41 (3): 393 – 408.
Dunne, T. and Leopold, L.B. (1978). Water in Environmental Planning, W.H.
Freeman and Co., NY.
Featherstone, Jeffrey P. (1996). Source Water Protection Programs in the
Delaware River Basin, Source Water Protection: An Ounce of Prevention.
American Water Works Association. Blue Bell, PA.
Gunther, O. (1998) Environmental Information Systems, Springer-Verlag, NY.
Hershfield, D.M., 1961: "Rainfall Frequency Atlas of the United States for
Durations from 30 Minutes to 24 Hours and Return Periods From 1 to 100
Years," Technical Paper No. 40, Weather Bureau, US Dept of Commerce,
Washington, DC.
Hitchings, Benjamin. (1998). A typology of regional growth management
system. The Regionalist 3, 1/2: 1-14.
82
Jones, K. Bruce et al. (1997). An Ecological Assessment of the United States
Mid-Atlantic Region: A Landscape Atlas. EPA/600/R-97/130. Office of Research
& Development, US EPA.
Kneese, Allen V. & Blair T. Bower. (1979). The Lower Delaware Valley Case
Study, in Environmental Quality & Residuals Management. Resources for the
Future. Johns Hopkins University Press, Baltimore, MD.
Kepner, William (2003) “Scenario Analysis for the San Pedro River, Analyzing
Hydrological Consequences for a Future Environment.” ReVA-EPA-MAIA
Conference: Using Science to Assess Environmental Vulnerabilities, Valley
Forge, PA, May 13-15, 2003.
Lance H. Butler, et al. (2003). Baseline Assessment of Pennypack Creek
Watershed (2002-2003). Philadelphia Water Dept. , Philadelphia, PA.
Lehigh Earth Observatory. (2005) Lehigh River Water Quality Project. Retrieved
on July 21, 2005, from Lehigh University Earth Observatory Website:
http://www.leo.lehigh.edu/projects/hydroprobe/wqdef.html .
Meadows, D. (1998). Indicators and Information Systems for Sustainable
Development. Sustainability Institute. Last accessed on 5/30/06 at
http://www.sustainabilityinstitute.org/pubs/Indicators&Information.pdf.
Meenar, Md Mahbubur R., ed. (2006). Pennypack Creek Watershed Study.
Center for Sustainable Communities, Temple University Ambler, for the US
Federal Emergency Management Agency.
Myers, M., “Retrofitting Rocky Branch: Applying Natural Features in Urban
Stream Design” Landscape Architecture, Volume 90, Number 3, March, 2000.
National Land Cover Database. (2007). At http://www.epa.gov/mrlc/nlcd.html
Pennsylvania Consortium for Interdisciplinary Environmental Policy (PCIEP;
2004). The State of the Commonwealth: Is Pennsylvania Moving Toward a
Sustainable Future? at http://www.paconsortium.state.pa.us/StateoftheCommonwealth.pdf .
Last accessed on 5/30/06.
Pennsylvania Department of Community and Economic Development. (PDCED;
2005). Pennsylvania Municipalities Planning Code, Article VI. Zoning, cf. 53 P.S.
§ 10604 [5]. at mpc.landuselawinpa.com/MPCode.pdf, last accessed
on January 15, 2007
Philadelphia Water Department (PWD; 2002). Drinking Water Quality Report at
www.phillywater.org/wqr/wqr2002/wqr2002.pdf . Last accessed 1/28/07.
83
Philadelphia Water Department (PWD; 2005). Pennypack Creek Watershed
River Conservation Plan at
http://www.phillywater.org/pennypack/River Conservation Plan/RCP Final Report.htm . Last
accessed 1/28/07.
Schueler, T. (1994). The Importance of Imperviousness. Watershed Protection
Techniques 1(3): 100-111.
Sherry, Jesse (2005) Sustainability Indicators for the Pennypack Creek
Watershed, Master’s Thesis, Department of Community & Regional Planning,
Temple University Ambler.
Sloto, R.A., and Crouse, M.Y. (1996). HYSEP: A computer program for
streamflow hydrograph separation and analysis: U.S. Geological Survey Water-
Resources Investigations Report 96-4040.
Steinitz, Carl (1993). Geographical Information Systems: A Personal Historical
Perspective, The Framework For A Recent Project, (And Some Questions For
The Future); European Conference on Geographic Information Systems,
Genoa, Italy, March 30. (Process available at
http://www.gsd.harvard.edu/studios/brc/maps/fig_5.html).
US Army Corps of Engineers (USACE 2007). HEC-HMS and HEC-RAS
Programs at http://www.hec.usace.army.mil/software/hec-hms and
http://www.hec.usace.army.mil/software/hec-ras . Last accessed 1/30/07/
US EPA. (1993). Guidance Specifying Management Measures for Sources of
Nonpoint Pollution in Coastal Waters. EPA-840-B-93-001c. Washington, DC:
Office of Water.
US EPA 2000a. Guidelines for Preparing Economic Analyses. National Center
for Environmental Economics, EPA 240-R-00-003.
US EPA 2000b. Environmental Planning for Communities: A Guide to the
Environmental Visioning Process Utilizing a Geographical Information System,
Office of R&D, EPA/625/R-98/003. At
http://www.epa.gov/ORD/WebPubs/gis/625r98003.pdf.
US EPA. (2005a) Regional Vulnerability Assessment for the Mid-Atlantic Region:
Evaluation of Integration Methods and Assessments Results. Office of Research
and Development. EPA 09LCB04.RPT.
US EPA. (2005b) Massachusetts Class B and SA Surface Water Quality
Standards. Retrieved from http://www.ipswatch.sr.unh.edu/314cmrclasses.html
on July 21, 2005. Ipswich-Parker Suburban Watershed Channel.
84
US EPA. (2006a) Regional Vulnerability Assessment Program: A Research
Strategy for 2001-2006 (http://www.epa.gov/reva/reva-strategy.pdf) .
US EPA. (2006b) Regional Vulnerability Assessment Program: Toolkit
(http://amethyst.epa.gov/revatoolkit/Welcome.jsp) .
US EPA (2007) Rapid Bioassessment Protocols for Use in Streams and
Wadeable Rivers: Periphyton, Benthic Macroinvertebrates, and Fish. At
http://www.epa.gov/owow/monitoring/rbp/ . Last accessed on 1/18/07.
US EPA. (2007). Rapid Bioassessment Protocols for Use in Streams and
Wadeable Rivers: Periphyton, Benthic Macroinvertebrates, and Fish. At
http://www.epa.gov/owow/monitoring/rbp/ . Last accessed on 1/18/07.
US Geological Survey. (USGS 1996). Water-Use Analysis Program for the
Neshaminy Creek Basin, Bucks and Montgomery Counties, Pennsylvania.
US Geological Survey. (1999). USGS Circular 1225: The Quality of Our Nation’s
Waters, Nutrients and Pesticides. U.S. Department of the Interior and U.S.
Geological Survey, Washington, D.C.
Wang, Michael. (2005). Greenhouse Gases, Regulated Emissions, and Energy
Use in Transportation. Systems Assessment Section, Center for Transportation
Research, Argonne National Laboratory, DuPage County, IL.
Wolman, M. G. and Miller, J. P. (1960). Magnitude and frequency of forces in
geomorphic processes, Journal of Geology, vol. 68. no. 1. pp. 54-74.
85
Appendices
Appendix A.1. ReVA Web Page Summaries
The ReVA Toolkit helps decision-makers evaluate the vulnerability of
ecological goods and services that are valued by society, using several
types of information such as current resource conditions and distributions,
estimated sensitivity of resources to the various stresses, and estimated
spatial distributions of stressors. Of primary importance is the definition of
vulnerability. Vulnerability is the degree to which a system is likely to
experience harm due to exposure to perturbations or stress. Vulnerability
considers both the quality of the valued resources & the intensity of the
stressors. It should be noted that heavily urbanized urban areas are not
considered to be vulnerable because they have already lost most of their
valued natural resources.
The variables used to estimate vulnerability fall into three basic categories.
Resource distribution variables record the current geographic distribution of
important environmental and human resources. Sensitivity variables are
conditions, or changes in conditions, that modify a resource's response to
stress. Stressor variables indicate the distribution of activities or conditions
that combine to cause environmental degradation.
Although the impacts of each individual stressor are important to consider, the
geographic areas most at risk are those that are subject to multiple stresses.
Although the cumulative impacts of multiple stresses across a region cannot yet
be quantified, areas where the possibility of cumulative impacts is high can be
identified.
Species diversity is a good proxy for overall environmental health, since
degraded environmental condition often results in declines in biological
populations. Information about the geographic distribution of cumulative impacts
and resources can be used to identify areas that are most vulnerable to
environmental degradation. For example, areas with high levels of stress and
with high aquatic species diversity might be considered to have the most
vulnerable aquatic populations. Areas with high levels of stress but low numbers
of species are already degraded.
In selecting between alternative scenarios that simulate the future, it is useful to
be able to compare the likely outcomes of each scenario. All land use change
models have some degree of uncertainty. One way to estimate the probability of
a certain type of change is to use a "weight of evidence" approach, as we have
done here. Responsible decision-making requires balancing among multiple
criteria, and various stakeholders value criteria differently. For example,
decisions about land development may involve changes in air quality, water
quality, economic conditions, and native biodiversity. Stakeholders concerned
about environmental conservation will likely prioritize criteria differently from
those concerned about economic development. The EDT allows one to view how
various stakeholders' preferences (values) would affect decision priorities.
86
Stakeholder values can be viewed using only variables important to the individual
or group (e.g. selecting variables relevant to conserving native aquatic species or
for evaluating human health risks) or in multiple weighting combinations to
illustrate trade-offs.
A.1.1. Data Preparation
One basic operation preceding assessment was putting data on a consistent
scale. Integrating the data to get a watershed summary across all variables
involves a few pre-processing steps. To use the variables in a consistent
manner, the data need to be transformed to the same scale. All variables are
scaled so that they were on a 0 (best) to 1 (worst) scale for each variable. To do
this, the following was applied to all variables:
1. (Rescale) Subtract the minimum value for each variable from the data,
setting the minimum value to zero.
2. (Normalize) Divide the data by the maximum value. This sets the maximum
value to one, and sets all values in between to be on the [0, 1] scale.
3. (Directionalize) If the variable's direction dictates that "higher is better", then
take the variable and reverse its normalized direction.
For example, suppose a variable has values of {3, 5, 7, 10, 11}, where the
"direction" for this variable is "-1" (higher is better). Following our steps, the data
becomes:
1. Rescale: {3, 5, 7, 10, 11} -> {0, 2, 4, 7, 8}
2. Normalize: {0, 2, 4, 7, 8} -> {0.00, 0.25, 0.50, 0.875, 1.00}
3. Directionalize: {0.00, 0.25, 0.50, 0.875, 1.00} -> {1.00, 0.75, 0.50, 0.125,
0.00}
Rescaling makes all data range from a set minimum to a set maximum,
normalizing allows comparisons between variables on a consistent scale,
and directionalizing creates a consistent direction for good and bad values
of a variable. Among groups of variables that were correlated to each other,
one variable was chosen to represent the group.
A few of the variables are fairly skewed (most values are near either zero or one
on the normalized scale). This may slightly affect a couple of the integration
methods, such as PCA. Methods that involve ranking, such as the quintile
methods, remain unaffected. The criticality method and clustering should also be
relatively unaffected since they are based on a distance measure.
In terms of types of data, the ReVA methods focus on a few important
categories. The first category is Sensitive Environmental Resources. This
includes ecosystems already stressed, migratory bird stopovers, regions
87
critical to native species, intact interior forest patches, and ecosystems with
low acid-neutralizing capacity. The second category is Current Conditions,
which contains
forest productivity, air quality, groundwater quality, water
quality, aquatic & terrestrial biodiversity, human health, and fiscal health.
The third group is made up of Drivers of Change. The drivers are
r
resource
extraction, land use, non-indigenous species, pollution, and climate change.
Key generic Stressors are given as agricultural runoff, atmospheric
deposition, loss of habitat, and non-point source pollution. With respect to
EPA Region III, the Mid-Atlantic Region, the Stressors are acid deposition,
coal mining, human population, landscape pattern, agricultural nitrogen,
ground-level ozone, and soil erosion / sedimentation.
The methods used to integrate the data into regional overviews and
environmental assessments covered a relatively wide range of analysis.
The best quintile method, the worst quintile method, and the radar summary give
a high-level glance at overall environmental quality. The summation method, in
combination with the Principle Components Analysis / Distance method, can
highlight sensitivity areas. These two methods have different and complementary
sensitivities. Areas that show well or poor in both have the strongest indications
of environmental quality. The state space method, because of its distance
measure, tends to differentiate between the middle (not worst or best)
watersheds. The weighted sum method allows user defined weights for individual
variables or groups. This method allows for the decision-maker to differentiate
based on specific preferences or interests. The stressor/resource matrix method
indicates the most critical stressors and most stressed resources across the
region. The overlap method compares a hypothetical future scenario to a present
environmental condition. The criticality method attempts to highlight areas at risk
for major change. This method uses a distance measure to a pre-defined "natural
state." Cluster analysis and self-organizing maps can be used as planning tools,
as these methods can be tailored for specific planning objectives.
A.1.2. Examples of Variables Used to Estimate Vulnerability
Resource Distribution
Numbers of native aquatic species are a resource of value to society. High
numbers of these species indicate areas that could be a higher priority for
protection over areas with low numbers of species. Numbers of native species
can also tell us where stresses have been acting for some time (low numbers of
native species) or where stress has historically been low (high numbers).
ReVA uses red to indicate lower resource value and green to indicate higher
resource value.
88
Sensitivities
For example, the percent of forest cover that has been defoliated is an
indication of how susceptible to damage a given area of forest may be to
additional stress, and thus is considered to be a sensitivity variable for forest
condition in general. Forests that have been heavily defoliated are more likely
to experience high levels of mortality when an additional stress, such as air
pollution, is also high. High levels of forest mortality can in turn result in
increased sediment loadings (through reduced uptake of water by trees with
higher runoff) and increased nutrients (through decay of trees) in streams.
In a final integration for vulnerability, a user may chose to weight this factor
more highly. Thinking of it in terms of an if-then scenario may be helpful; for
example, research and experience may indicate that the stress on a forest
jumps significantly as percent defoliation rises above some threshold level. In
this case, the cumulative stress displayed for areas above the threshold level
can be amplified by some appropriate factor. In time and with experience, the
best guesses for appropriate threshold and amplification factors may be better
understood.
89
Stressor Distribution
For example, acid rain stresses plants, contributing to poor forest health. Nitrate
(NO3) and Sulfate (SO4) are the major pollutants contributing to acid rain, and
thus estimates of Nitrate wet deposition are commonly used as an indication of
acid rain. The degree to which Nitrate wet deposition becomes a pollutant
depends on the location and the physiochemistry of the forest soils or
waterbody upon which it falls. Some important variables cannot be measured
directly, but their distribution can be estimated using well-established models.
In this example, the spatial distribution of wet nitrate deposition is estimated
from rainfall data (amount) and proximity to sources of NOx air pollution (from
fossil fuel burning). Nitrate from air pollution contributes to nutrient levels in
streams, which affects aquatic habitat (in particular aquatic plants) and native
species' condition.
Here ReVA uses red to indicate higher presence of stressors and green to
indicate lower presence of stressors.
90
A.1.3. Assessment Questions
Resource managers and other decision-makers are frequently required to make
decisions about priorities. A common question is "Given limited resources, what
environmental problems and what geographic areas are most in need of
attention?" Answering this difficult question requires exploration of a series of
focused assessment questions before actions are taken to protect valued
resources, human health, or quality of life. These questions might focus on
information such as:
an evaluation of current overall conditions
risk of future environmental degradation
sustainability of the system
current and future value to society
feasibility of taking some action
All of these assessment questions can be addressed to some degree using
available data and information. Each requires a different integration method (see
the tutorial for more information on this), and each receives a different answer, or
in the case of a visual representation, a different map.
Different assessment questions require different types of variable data and
metrics, and possibly different analysis or integration methods. The selection of
variables and methods is driven by the particular issues of concern and the
questions a user brings to the vulnerability assessment. Further information on
91
selecting data and analysis methods is provided in the tutorial and in the
analytical sections of the tool.
To help you understand how the ReVA EDT can assist in answering these
questions, the next few pages show examples of the types of information that the
web tool can provide.
Cumulative Impacts
This map shows all four stressors: nitrogen, sediment, aquatic exotic species and
nitrate deposition). The individual stressor information has been combined using
an integration method that counts the number of times that a watershed is among
the worst 20% (the Worst Quintiles integration method). All stressors are equally
weighted.
Distribution of Resources
One of the first questions asked in any assessment is "Where are the most
important environmental resources?" It is important to identify areas where
management actions can prevent further damage and improve conditions.
While restoration of degraded areas is valuable, preservation of areas that are
still in good shape is a critical and cost-effective way to maintain a region's
environmental health. Areas with high species diversity have often been less
impacted by stressors while highly impacted areas show little native species
diversity. The duration of stress will also influence condition; unfortunately
historical data on stressor distributions are rarely available.
92