PROTECTING WATER RESOURCES
WITH HIGHER-DENSITY DEVELOPMENT
     United States
     Environmental Protection
     Agency

-------
Acknowledgements
The principal author, Lynn Richards, from the U.S. Environmental Protection Agency's
Development, Community, and Environment Division, would like to recognize people who
contributed insights and comments on this document as it was being developed: Chester
Arnold, University of Connecticut—Non-Point Source Education for Municipal Officials; John
Bailey, Smart Growth America; Deron  Lovaas, Natural Resources Defense Council; Bill
Matuszeski, formerly with EPA Chesapeake Bay Program; Philip Metzger, EPA Office of Water;
Rosemary Monahan, EPA Region 1; Betsy Otto, American Rivers; Joe Persky, University of
Illinois at Chicago; Milt Rhodes, formerly with the North Carolina Department of Environment
and Natural Resources; and William Shuster, EPA Office of Research and Development.
Additional recognition is extended to EPA staff from Office of Water (Robert Goo, Jamal Kadri,
and Stacy Swartwood) as well as staff at EPA's Development, Community, and Environment
Division (Geoffrey Anderson, Mary Kay Bailey, and Megan Susman).

To request additional copies of this report, contact EPA's National Service Center for
Environmental Publications at 800-490-9198 or by email at ncepimal@one.net and ask for
publication number 231-R-06-001.

To access this report online, visit  or .

Front cover photos:
   Left: The Snake River flows outside Jackson, Wyoming. Photo courtesy of USDA NRCS.
   Top right: Rosslyn-Ballston Corridor, Arlington County, Virginia. Arlington County
   Department of Community Planning, Housing, and Development received a 2002
   National Award for Smart Growth Achievement in the Overall Excellence category for
   its planning efforts in the Rosslyn-Ballston Corridor. Photo courtesy of Arlington County.
   Middle right: People gather at Pioneer Square in Portland, Oregon. Photo courtesy
   of US EPA.

Back cover photos:
   Top left: This hillside in Northern California is covered by wildflowers. This open space
   provides habitat to wildlife as well as serving important watershed services. Photo cour-
   tesy of USDA NRCS.
   Middle left: A family enjoys open space in central Iowa. Photo courtesy of USDA NRCS.
   Bottom left: A stream flows through western Maryland. Photo courtesy of USDA NRCS.
   Right: This redevelopment site in Arlington, Virginia, which includes stores, apartments
   townhomes, single family homes, parking garages, and a one-acre public park, was for-
   merly a large department store surrounded by surface parking. Photo courtesy of US EPA.

-------
                UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                              WASHINGTON, D.C. 20460
Dear Colleague:

We are excited to share with you the enclosed report, Protecting Water Resources with
Higher-Density Development. For most of EPA's 35-year history, policymakers have focused
on regulatory and technological approaches to reducing pollution. These efforts have met
with significant success. But, the environmental challenges of the 21st century require new
solutions, and our approach to environmental protection must become more sophisticated.
One approach is to partner with communities to provide them with the tools and informa-
tion necessary to address current environmental challenges. It is our belief that good envi-
ronmental information is necessary to make sound decisions. This report strives to meet
that goal by providing fresh information and perspectives.

Our regions, cities, towns, and neighborhoods are growing.  Every day, new buildings or
houses are proposed, planned, and built. Local governments, working with planners, citizen
groups, and developers, are thinking about where and how this new development can
enhance existing neighborhoods and also protect the community's natural environment.
They are identifying the characteristics of development that can build vibrant neighbor-
hoods, rich in natural and historic assets, with jobs, housing, and amenities for all types of
people. They are directing growth to maintain and improve  the buildings and infrastructure
in which they have already invested.

In addition to enjoying the many benefits of growth, communities are also grappling with
growth's challenges, including development's impact on water resources. In the face of
increasing challenges from non-point source pollution, local governments are looking for,
and using, policies, tools, and information that enhance existing neighborhoods and protect
water resources. This report gives communities a different perspective and set of information
to address the complex interactions between development  and water quality.

Protecting Water Resources with Higher-Density Development  is intended for water quality pro-
fessionals, communities, local governments, and state and regional planners who are grap-
pling with protecting or enhancing their water resources while accommodating growing
populations. We hope that you find this report informative as your community strives to
enjoy the many benefits of growth and development and cleaner water.

For additional free copies, please send an e-mail to ncepimal@one.net or call (800) 490-9198
and request EPA publication 231-R-06-001. If you have any questions concerning this study,
please do not hesitate to contact Lynn Richards at (202) 566-2858.

                                     Sincerely,
Ben Grumbles                          Brian F. Mannix
Assistant Administrator                  Associate Administrator
Office of Water                          Office of Policy, Economics, and
                                       Innovation
                           Internet Address (URL) • http://www.epa.gov
      Recycled/Recyclable .Printed with Vegetable Oil Based Inks on Recycled Paper (Minimum 50% Postconsumer content)

-------
   PROTECTING WATER RESOURCES
WITH HIGHER-DENSITY DEVELOPMENT

-------
Contents
EXECUTIVE SUMMARY	1
INTRODUCTION	2
IMPACTS FROM DEVELOPMENT ON
WATERSHED FUNCTIONS	3
CRITICAL LAND USE COMPONENTS FOR
PROTECTING WATER QUALITY FOR BOTH
Low- AND HIGH-DENSITY DEVELOPMENT	4
LOW-DENSITY DEVELOPMENT—CRITIQUING
CONVENTIONAL WISDOM	7
TESTING THE ALTERNATIVE: CAN COMPACT
DEVELOPMENT MINIMIZE REGIONAL WATER
QUALITY IMPACTS?	9
THE MODEL AND DATA INPUTS	9
SUMMARY OF SCENARIOS	11
RESULTS	13
FINDINGS/DISCUSSION	26
OTHER RESEARCH	31
CONCLUSIONS	32
REFERENCES AND BIBLIOGRAPHY .                                . 34

-------
Exhibits
1:   WATERSHED SERVICES	3
2:   SUMMARY OF SCENARIOS	11
3:   TOTAL AVERAGE ANNUAL STORMWATER RUNOFF
    FOR ALL SCENARIOS	13
4:   EACH SCENARIO ACCOMMODATES EIGHT HOUSES	15
5:   10,000-AcRE WATERSHED ACCOMMODATING
    10,000 HOUSES	17
6:   10,000-AcRE WATERSHED ACCOMMODATING
    DIFFERENT NUMBERS OF HOUSES	19
7:   80,000 HOUSES ACCOMMODATED	21
8:   TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2000	23
9:   TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2020	24
10: TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2040	25
11: SUMMARY OF FINDINGS	27

EXAMPLES
1:   ONE-ACRE LEVEL	13
2:   LOT LEVEL	14
3:   WATERSHED LEVEL	16
4:   REMAINING LAND IN THE WATERSHED DEVELOPED	18
5:   ACCOMMODATING THE SAME NUMBER OF HOUSES	20
6:   TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2000	22
7:   TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2020	23
8:   TIME SERIES BUILD-OUT ANALYSIS:
    BUILD-OUT IN 2040 .                                             . 24

-------
Executive Summary
Growth and development expand communities'opportunities by bringing in new residents,
businesses, and investments. Growth can give a community the resources to revitalize a
downtown, refurbish a main street, build new schools, and develop vibrant places to live,
work, shop, and play. However, with the benefits come challenges. The environmental
impacts of development can  make it more difficult for communities to protect their natural
resources. Where and how communities accommodate growth has a profound impact on the
quality of their streams, rivers, lakes, and beaches. Development that uses land efficiently and
protects undisturbed natural  lands allows a community to grow and still protect its
water resources.

The U.S. Census Bureau projects that the U.S. population will grow by 50 million people, or
approximately 18 percent, between 2000 and 2020. Many communities are asking where and
how they can accommodate this growth while maintaining and improving their water
resources. Some communities have interpreted water-quality research to mean that low-den-
sity development will best protect water resources. However, some water-quality experts
argue that this strategy can backfire and actually harm water resources. Higher-density devel-
opment, they believe, may be a better way to protect water resources. This study intends to
help guide communities through this debate to better understand the impacts of high- and
low-density development on water resources.

To more fully explore this issue, EPA modeled three scenarios of different densities at three
scales—one-acre  level, lot level, and watershed level—and at three different time series
build-out examples to examine the premise that lower-density development is always better
for water quality. EPA examined stormwater runoff from different development densities to
determine the comparative difference between scenarios.This analysis demonstrated:

   • The higher-density scenarios generate less stormwater runoff per house at all scales—
    one acre, lot, and watershed—and time series build-out examples;
   • For the same amount of development, higher-density development produces less
    runoff and less impervious cover than low-density development; and
   • For a given amount of growth, lower-density development impacts more of the
    watershed.
Taken together, these findings indicate that low-density development may not always be the
preferred strategy for protecting water resources. Higher densities may better protect water
quality—especially at the lot  and watershed levels. To accommodate the same number of
houses, denser developments consume less land than lower density developments.
Consuming less land means creating less impervious  cover in the watershed. EPA believes
that increasing development densities is one strategy communities can use to minimize
regional water quality impacts. To fully protect water  resources, communities need to employ
a wide range of land use strategies, based on local factors, including building a range of
development densities, incorporating adequate open space, preserving critical ecological
and buffer areas, and minimizing land disturbance.

-------
Protecting Water Resources with Higher-Density Development
      Introduction

      Growth and development expand communities'opportunities by bringing in new residents,
      businesses, and investments. Growth can give a community the resources to revitalize a
      downtown, refurbish a main street, build new schools, and develop vibrant places to live,
      work, shop, and play. However, with the benefits come challenges. The environmental im-
      pacts of development can make it more difficult for communities to protect their natural
      resources. Where and how communities accommodate growth has a profound impact on the
      quality of their streams, rivers, lakes, and beaches. Development that uses land efficiently and
      protects undisturbed natural lands allows a community to grow and still protect its
      water resources.

      The U.S. Census Bureau projects that the U.S. population    Which is a  better Strategy
      will grow by 50 million people, or approximately 18 per-    tQ protect water quality:
      cent, between 2000 and 2020. Many communities are                         .     •''
      asking where and how they can accommodate this        low~ or high-density
      growth while maintaining and improving their water       development?
      resources. Some communities have interpreted water-
      quality research to mean that low-density development will best protect water resources.
      However, some water-quality experts argue that this strategy can backfire and actually harm
      water resources. Higher-density development, they believe, may be a  better way to protect
      water resources. This study intends to help guide communities through this debate to better
      understand the impacts of high- and low-density development on water  resources.

      Virtually every metropolitan area in the United States has expanded substantially in land area
      in recent decades. According to the U.S. Department of Agriculture's National Resources
      Inventory (NRI), between 1954 and 1997, urban land area almost quadrupled, from 18.6 mil-
      lion acres to about 74 million acres in the contiguous 48 states (USDA, 1997b). From 1982 to
      1997, when population in the contiguous United States
      grew by about 15 percent, developed land increased by    Between 1954 and  1997,
      25 million acres, or 34 percent. Most of this growth is tak-   urban land area  almost
      ing place at the edge of developed areas, on greenfield         ,    ,   ,  c     -nr   .,
       •4    u- u    •   \A  t    *i  A     A        *        quadrupled, from 18.6  mil-
      sites, which can include forestland, meadows, pasture,     ^      ^
      and rangeland (USDA, 1997a). Indeed, in one  analysis of    lion acres to about 74
      building permits in 22 metropolitan areas between 1989    million acres in the COn-
      and 1998, approximately 95 percent of building permits    tidUOUS 48 States
      were on greenfield sites (Farris, 2001).
      According to the American Housing Survey, 35 percent of new housing is built on lots
      between two and five acres, and the median lot size is just under one-half acre (Census,
      2001). Local zoning may encourage building on relatively large lots, in part because local
      governments often believe that it helps protect their water quality. Indeed, research has
      revealed that more impervious cover can degrade water quality. Studies have demonstrated
      that at 10 percent imperviousness, a watershed is likely to become impaired and grows more
      so as imperviousness increases (Arnold, 1996; Schueler, 1994).This research has prompted
      many communities to adopt  low-density zoning and site-level imperviousness limits, e.g.,
      establishing a percentage of the site, such as 10 or 20 percent, that can be covered by

-------
impervious surfaces such as houses, garages, and driveways. These types of zoning and
development ordinances are biased against higher-density development because it has
more impervious cover. But do low-density approaches protect our water resources?
This study examines the assumption that low-density development is always better for water
quality.1 EPA modeled stormwater runoff from different development densities at the site
level and then extrapolated and analyzed these findings at the watershed level. Modeling
results were used to compare stormwater runoff associated with several variations of
residential density.


Impacts from  Development on Watershed Functions

A watershed is a land area that drains to a given body of water.  Precipitation that falls in the
watershed will either infiltrate into the ground, evapotranspirate back into the air, or run off
into streams, lakes, or coastal  waters. This dynamic is described  in Exhibit 1.


                          EXHIBIT 1: Watershed Services
             25% shallow
             infiltration
            Natural Ground Cover
                                  25% deep
                                  infiltration
                       35% evapotranspiration
                     21% deep
                     infiltration
                                              1O%-2O% Impervious Surface


                                                         3O% evapotranspiration
                                                                      	m 	, _,

                                                                      ||||
                                             = 5 :•
             2O% shallow
             infiltration
           35%-5O% Impervious Surface
                                     deep
                                  infiltration
1G% shallow
infiltration
                                              75%-1OO% Impervious Surface
                       deep
                     infiltration
          As land cover changes, so does the amount of precipitation that absorbs into the
          ground, evaporates into the air, or runs off.

A watershed may be large or small. The Mississippi River, for example, drains a one-million-
square-mile watershed made up of thousands of smaller watersheds, such as the drainage
basins of the creeks that flow into tributaries of the Mississippi. In smaller watersheds, a few
acres of land may drain into small streams, which flow into larger streams or rivers; the lands
drained by these streams or rivers make  up a larger watershed. These streams support
1 Stormwater runoff was used as a proxy for overall water quality. In general, the more stormwater runoff a region experiences, the more
associated pollutants, such as total nitrogen, phosphorus, and suspended solids, will enter receiving waterbodies.

-------
Protecting Water Resources with Higher-Density Development
      diverse aquatic communities and perform the vital ecological roles of processing the carbon,
      sediments, and nutrients upon which downstream ecosystems depend. Healthy, functioning
      watersheds naturally filter pollutants and moderate water quality by slowing surface runoff
      and increasing the infiltration of water into soil. The result is less flooding and soil erosion,
      cleaner water downstream, and greater ground water reserves.

      Land development directly affects watershed functions. When development occurs in previ-
      ously undeveloped areas, the resulting alterations to the land can dramatically change how
      water is transported and stored. Residential and commercial development create impervious
      surfaces and compacted soils that filter less water, which increases surface runoff and
      decreases ground water infiltration. These changes can increase the volume and velocity of
      runoff, the frequency and severity of flooding, and peak storm flows.

      Moreover, during construction, exposed sediments and construction materials can be
      washed into storm drains or directly into nearby bodies of water. After construction, develop-
      ment usually replaces native meadows, forested areas, and other natural landscape features
      with compacted lawns, pavement, and rooftops. These largely impervious surfaces generate
      substantial runoff.  For these reasons, limiting or minimizing the amount of land disturbed
      and impervious cover created during development can help protect water quality.


      Critical Land Use Components for Protecting Water
      Quality for Both Low- and  High-Density Development

      What strategies can communities use to continue to grow while protecting their water quality?
      Watershed hydrology suggests that three primary land use strategies can help to ensure ade-
      quate water resource protection:

         •  Preserve large, continuous areas of absorbent open space;
         •  Preserve critical ecological areas, such as wetlands, floodplains,
           and riparian corridors; and
         •  Minimize overall land disturbance and impervious surface associated
           with development.
      These approaches  work because, from a watershed perspective, different land areas have dif-
      ferent levels of ecological value. For example, a nutrient-rich floodplain has a higher ecologi-
      cal value than a grass meadow. Communities should view these strategies as basic steps to
      preserve watershed function and as the framework within which all development occurs.

      PRESERVING OPEN SPACE

      Preserving  open space is critical to maintaining water quality at the regional level. Large, con-
      tinuous areas of open space reduce and slow runoff, absorb sediments, serve as flood control,
      and help maintain aquatic communities. To ensure well-functioning watersheds, regions
      should set aside sufficient amounts of undisturbed, open space to absorb, filter, and store rain-
      water. In most regions, this undeveloped land comprises large portions of a watershed, filtering

-------
out trash, debris, and chemical pollutants before they enter a community's water system. Open
space provides other benefits, including habitat for plants and animals, recreational opportuni-
ties, forest and ranch land, places of natural beauty, and community recreation areas.

To protect these benefits, some communities are preserving undeveloped parcels or regional
swaths of open space. One of the most dramatic examples is the New York City Watershed
Agreement. New York City, New York State, over 70 towns, eight counties, and EPA signed the
agreement to support an enhanced watershed protection program for the New York City
drinking water supply. The city-funded, multi-year, $1.4-billion agreement developed a multi-
faceted land conservation approach, which includes the purchase of 80,000 acres within the
watershed as a buffer around the city's drinking water supply. This plan allows the city to
avoid the construction of filtration facilities estimated to cost six to eight billion dollars (New
York City, 2002).

PRESERVING  ECOLOGICALLY SENSITIVE AREAS

Some types of land perform watershed functions better than others do. Preserving ecologi-
cally important land, such as wetlands, buffer zones, riparian corridors, and floodplains, is crit-
ical for regional water quality. Wetlands are natural filtration plants, slowing  water flow and
allowing sediments to settle and the water to clarify. Trace metals bound to  clay carried in
runoff also drop out and become sequestered in the soils and peat at the bed of the marsh
instead of entering waterbodies, such as streams, lakes, or rivers. Preserving and  maintaining
wetlands are critical to maintain water quality.

                                                  In addition, strips of vegetation along
                                                  streams and around reservoirs are
                                                  important buffers, with wooded
                                                  buffers offering the greatest protec-
                                                  tion. For example, if soil conditions are
                                                  right, a 20- to 30-foot-wide strip of
                                                  woodland removes 90 percent of the
                                                  nitrates in stormwater runoff (Trust for
                                                  Public Land, 1997). These  buffer zones
                                                  decrease the amount of pollution
                                                  entering the water system. Tree and
                                                  shrub  roots hold the  bank in place,
                                                  preventing erosion and its resulting
                                                sedimentation and turbidity. Organic
                                                 matter and grasses slow the flow of
runoff, giving the sediment time to settle and water time to percolate, filter through the soil,
and recharge underlying ground water. Research has shown that wetlands and buffer zones,
by slowing and holding water, increase ground water recharge, which directly reduces the
potential for flooding (Schueler, 1994). By identifying and preserving these critical ecological
areas, communities are actively protecting and enhancing their water quality.
Wetlands, such as this one in Butte County, California, provide
critical watershed services for the region.

-------
Protecting Water Resources with Higher-Density Development
       MINIMIZING LAND DISTURBANCE AND IMPERVIOUS COVER
       Minimizing land disturbance and impervious cover is critical to maintaining watershed
       health.The amount of land that is converted, or"disturbed,"from undeveloped uses, such as
       forests and meadows, to developed uses, such as lawns and playing fields, significantly
       affects watershed health. Research now shows that the volume of runoff from highly com-
       pacted lawns is almost as high as from paved surfaces (Schueler, 1995, 2000; USDA, 2001).
       This research indicates that lawns and other residential landscape features do not function,
       with regard to water, in the same way as nondegraded natural areas. In part, the difference
       arises because developing land in greenfield areas involves wholesale grading of the site and
       removal of topsoil, which can lead to severe erosion during construction, and soil com-
       paction by heavy equipment. However, most communities focus not on total land disturbed,
       but on the amount of impervious cover created.
       Research has revealed a strong rela-
       tionship between impervious cover
       and water quality (Arnold, 1996;
       Schueler, 1994; EPA, 1997). Impervious
       surfaces collect and accumulate pollu-
       tants deposited from the atmosphere,
       leaked from vehicles, or derived from
       other  sources. During storms, accumu-
       lated pollutants are quickly washed off
       and rapidly delivered to aquatic sys-
       tems.  Studies have demonstrated that
       at 10 percent imperviousness,2a
       watershed is likely to become
       impaired (Schueler, 1996; Caraco,  1998;
       Montgomery County, 2000), the
Current construction practices generally disturb the entire
development site, as shown by this site in Des Moines, Iowa.
       stream channel becomes unstable due to increased water volumes and stream bank erosion,
       and water quality and stream biodiversity decrease. At 25 percent imperviousness, a water-
       shed becomes severely impaired, the stream channel can become highly unstable, and water
       quality and stream biodiversity are poor3 (Schueler, 2000).The amount of impervious cover is
       an important indicator of watershed health, and managing the degree to which a watershed is
       developed is critical to maintaining watershed function.
       Although the 10 percent threshold refers to overall imperviousness within the watershed,
       municipalities have applied it to individual sites within the watershed, believing that  lower den-
       sities better protect watershed functions. Indeed, as mentioned earlier, some localities have
       gone so far as to create strong  incentives for, or even require, low densities—with water
       resource protection as an explicit goal.These communities are attempting to minimize hard
       2 The 10 percent figure is not an absolute threshold. Recent studies have indicated that in some watersheds, serious degradation may begin
       well below 10 percent. However, the level at which watershed degradation begins is not the focus of this study. For purposes of our analysis,
       EPA uses the 10 percent threshold as an indicator that water resources might be impacted.
       3 There are different levels of impairment. In general, when the term is used in EPA publications, it usually means that a waterbody is not meet-
       ing its designated water quality standard. However, the term can also imply a decline or absence of biological integrity; for example, the water-
       body can no longer sustain critical indicator species, such as trout or salmon. Further, there is a wide breadth of levels of impairment, from
       waterbodies that are unable to support endangered species to waterbodies that cannot support any of the beneficial-use designations.

-------
surfaces at the site level.They believe that limiting densities within particular development sites
limits regional imperviousness and thus protects regional water quality.The next section exam-
ines this proposition and finds that low-density development can, in fact, harm water quality.

Low-Density Development—Critiquing
Conventional Wisdom

As discussed, studies have demonstrated that watersheds can suffer impairment at 10 percent
impervious cover and that at 25 percent imperviousness, the watershed is typically considered
severely impaired. Communities have often translated these findings into the notion that low-
density development at the site level results in better water quality. Such conclusions often
come from analysis such as: a one-acre site has one or two homes with a driveway and a road
passing by the property. The remainder of the site is lawn. Assuming an average housing foot-
print of 2,265 square feet4 (National Association of Home Builders, 2001), the impervious
cover for this one-acre site is approximately 35 percent (Soil Conservation Service, 1986). By
contrast, a higher-density scenario might have eight to 10 homes per acre and upwards of 85
percent impervious cover (Soil Conservation Service, 1986). The houses'footprints account
for most of the impervious cover. Thus, low-density zoning appears to create less impervious
cover, which ought to protect water quality at the site and regional levels. However, this logic
overlooks several key caveats.

1.  The "pervious" surface left in low-density developmen t often acts like impervious surface.
   In general, impervious surfaces, such as a structure's footprint, driveways, and roads, have
   higher amounts of runoff and associated pollutants than pervious surfaces. However,
   most lawns, though pervious, still contribute to runoff
   because they are compacted. Lawns are thought to     Lawns Still contribute to
   provide "open space"for infiltration of water. However,  runoff because they are
   because of construction practices, the soil becomes
   compacted by heavy equipment and filling of depres-  compacted and disturbed.
   sions (Schueler, 1995, 2000). The effects of this com-
   paction can remain for years and even increase due to mowing and the presence of a
   dense mat of roots. Therefore, a one- or two-acre lawn does not offer the same infiltration
   or other water quality functions as a one- or two-acre undisturbed forest. Minimizing
   impervious surfaces by limiting the number of houses but allowing larger lawns does not
   compensate for the loss of watershed services that the area  provided before develop-
   ment (USDA, 2001).

2.  Density and imperviousness are not equivalent. Depending on the design, two houses may
   actually create as much imperviousness as four houses. The  impervious area per home
   can vary widely due to road infrastructure, housing design (single story or multistory), or
   length and width of driveways. To illustrate, a three-story condominium building of 10
   units on one acre can have less impervious surface than four single-family homes on the
   same acre. Furthermore, treatment of the remaining undeveloped land on that acre can
4 The average house built in 2001 included three or more bedrooms, two and a half baths, and a two-car garage.

-------
Protecting Water Resources with Higher-Density Development
          vary dramatically between housing types. For example, in some dispersed, low-density
          communities, such as Fairfax County, Virginia, some homeowners are paving their front
          lawns to create more parking for their cars (Rein, 2002).
       3.  Low-density developments often mean more off-site impervious infrastructure. Development
          in the watershed is not simply the sum of the sites within it. Rather, total impervious area
          in a watershed is the sum of site developments plus
          the impervious surface associated with infrastructure   Water quality Suffers not
          supporting those sites, such as roads and parking lots,  only from the increase in
          Lower-density development can require substantially   j      vjous surf     but a|SQ
          higher amounts of this infrastructure per house and       r
          per acre than denser developments. Recent research    from tne associated activi-
          has demonstrated that on sites with two homes per    ties: construction, increased
          acre, impervious surfaces attributed to streets, drive-    travel to and from the devel-
          ways, and parking lots can represent upwards of 75    Q      t and extensjon of
          percent of the total site imperviousness (Cappiella,
          2001). That number decreases to 56 percent on sites    infrastructure.
          with eight homes per acre.This research indicates
          that low densities often require more off-site transportation-related impervious infra-
          structure, which is generally not included when calculating impervious cover.
          Furthermore, water quality suffers not only from the increase in impervious surface, but
          also from the associated activities: construction, increased travel to  and from the develop-
          ment, extension of infrastructure, and chemical maintenance of the areas in and sur-
          rounding the development. Oil and other waste products, such as heavy metals, from
          motor vehicles, lawn fertilizers, and other common solvents, combined with the increased
          flow of runoff, contribute substantially to water pollution. As imperviousness increases, so
          do associated activities, thereby increasing the impact on water quality.
       4.  If growth is coming to the region, limiting density on a given site does not eliminate that
          growth. Density limits constrain the amount of development on a site but  have little
          effect on the region's total growth (Pendall, 1999,
          2000). The rest of the growth that was going to come   Growth is Still  coming
          to the region  still comes, regardless of density limits in  to a region, regardless
          a particular place. Forecasting future population       of density limits in a
          growth is a standard task for metropolitan planning        ..    •     •
          organizations as they plan where and how to accom-   ^          ^
          modate growth in their region. They project future
          population growth based on standard regional population modeling practices, where
          wage or amenity differentials, such as climate or culture (Mills,  1994)—and not zoning
          practices such as density limits—account for most of a metropolitan area's population
          gain or loss.5 While estimates of future growth within a particular time frame are rarely
          precise, a region must  use a fixed amount of growth to test the effects  of adopting
       5 The most widely-used such model—the REMI8 Policy Insight™ model—uses an amenity variable. However, even this is implemented as an
       additional change in the wage rate. See Remi Model Structure. .The in-
       house model used by the San Diego Association of Governments is an advanced example of the type used by councils of governments
       around thecountry..

-------
   different growth planning strategies because it still must understand the economic,
   social, and environmental impacts of accommodating a growing population. Absent
   regional coordination and planning, covering a large part of a region with density limits
   will likely drive growth to other parts of the region. Depending on local conditions, water
   quality may be more severely impaired than if the growth had been accommodated at
   higher densities on fewer sites.


Testing  the Alternative: Can  Compact  Development
Minimize Regional Water  Quality Impacts?

To more fully understand the potential water quality impacts of different density levels, this
section compares three hypothetical communities, each accommodating development at
different densities—one house per acre,  four houses per acre, and eight houses per acre.6
To assess regional water quality impacts, EPA modeled the stormwater impacts from different
development densities. In general, the more stormwater runoff generated within a region,
the more associated pollutants, such as total  nitrogen, phosphorus, and suspended solids,
will enter receiving waterbodies.The three density levels capture some of the wide range of
zoning practices in use throughout the country. All of these densities are consistent with sin-
gle-family, detached housing. EPA examined the stormwater impacts from each density sce-
nario at various scales of residential development7—one-acre, lot, and watershed
levels—and through a 40-year time series build-out analysis.


The Model and  Data Inputs

The model used to compare the stormwater impact from the scenarios is the Smart Growth
Water Assessment Tool for Estimating Runoff (SG WATER), which is a peer-reviewed sketch
model that was developed specifically to compare water quantity and quality differences
among different development patterns (EPA, 2002). SG WATER'S methodology is based on the
Natural Resources Conservation Service (NRCS) curve numbers (Soil Conservation Service,
1986), event mean concentrations, and daily rainfall data.8 The model  requires the total num-
ber of acres developed at a certain development density. If density is unknown, total percent
imperviousness can be used.The model was  run using overall percent imperviousness.

EPA believes that the results presented here are conservative. SG WATER uses a general and sim-
ple methodology based on curve numbers. One limitation of curve numbers is that they tend
to underestimate stormwater runoff for smaller storms (less than one inch). This underestimate
6 Densities at one, four, and eight residential units per acre are used here for illustrative purposes only. Many communities now are zoning
for one unit per two acres at the low-density end of the spectrum. Low-density residential zoning exists in places as diverse as Franklin
County, Ohio, which requires no less than two acres per unit ) to Cobb
County, Georgia, outside of Atlanta, which requires between one and two units per acre in its low-density residential districts (). By comparison, some communities are beginning to allow higher densities, upwards
of 20 units per acre. For example, the high-density residential district in Sonoma County, California permits between 12 and 20 units per
acre (), and the city of Raleigh, North Carolina, allows up to 40 units per acre in
planned development districts.
7 This example and others throughout this study compare residential units, but a similar comparison including commercial development could also
be done.
8 Daily time-step rainfall data fora 10-year period (1992-2001, inclusive) were used.

-------
     Protecting Water Resources with Higher-Density Development
            can be significant since the majority of storms are small storms. In addition, the curve num-
            bers tend to overestimate runoff for large storms. However, curve numbers more accurately
            predict runoff in areas with more impervious cover.9 For the analysis here, the runoff from the
            low-density site is underestimated to a larger degree than the runoff from the higher-density
            site because the higher-density site has more impervious cover. Simply put, because of
            methodology, the difference in the numbers presented here is conservative—it is likely that
            the comparative difference in runoff between the sites would  be  greater if more extensive
            modeling were used.
            To isolate the impacts that developing at different densities makes on stormwater runoff, EPA
            made several simplifying assumptions in  the modeling:
               •  EPA modeled only residential growth and not any of the corresponding commercial,
                 retail, or industrial growth that would occur in addition to home building. Moreover, EPA
                 assumed that all the new growth would occur in greenfields (previously undeveloped
                 land). Infill development, brownfield redevelopment, and other types of urban develop-
                 ment were not taken into consideration, nor were multifamily housing, apartments, or
                 accessory dwelling units.10
               •  The modeling did not take  into account any secondary or tertiary impacts, such as addi-
                 tional stormwater benefits, that may be realized by appropriately locating the develop-
                 ment within the watershed. For example, siting development away from headwaters,
                 recharge areas, or riparian corridors  could better protect these sensitive areas. Denser
                 development makes this type of protective siting easier since  less land is developed.
                 However, these impacts are not captured or calculated within the modeling.
               •  Whether developed at one, four, or eight houses per acre, when one acre is developed,
                 EPA assumed the entire acre is disturbed land (e.g., no forest or meadow cover would be
                 preserved), which is consistent with current construction practices.
               •  All the new growth is assumed to be single-family, detached houses.11 Whether
                 developed at one, four, or eight houses per acre, each home has a footprint of 2,265
                 square feet, roughly the current average size for new houses (National Association of
                 Home Builders, 2001).
            9 Most existing stormwater models incorrectly predict flows associated with small rains in urban areas. Most existing urban runoff models
            originated from drainage and flooding evaluation procedures that emphasized very large rains (several inches in depth). These large storms
            contribute only very small portions of the annual average discharges. Moderate storms, occurring several times a year, are responsible for
            the majority of the pollutant discharges. These frequent discharges cause mostly chronic effects, such as contaminated sediment and fre-
            quent high flow rates, and the inter-event periods are not long enough to allow the receiving water conditions to recover.
            10 Single-family, detached housing dominates many low-density residential developments. However, higher-density developments support
            a range of housing types, including townhouses, apartments, and other forms of multifamily housing. These housing types generally have a
            smaller footprint per house than 2,265 square feet. Therefore, a more realistic situation for the higher-density scenarios would either be a
            smaller housing footprint or an increase in the number of homes accommodated on one acre. In either case, including these different hous-
            ing types in the analysis would produce less overall stormwater runoff and less per house runoff for the higher-density scenarios.
            11 It is possible that when additional land uses, such as commercial, transportation, or recreation, are included in the analysis, the low-densi-
            ty scenarios become relatively less dense while the higher-density scenarios become relatively more dense. In general, low-density residen-
            tial development tends to be associated with low-density commercial development, characterized by large retail spaces, wide roads, large
            parking lots, and minimal public transportation. Higher-density residential areas are more likely to have high-density commercial options,
            with smaller retail spaces, mixed land uses, narrower streets, parking garages, on-street parking, and sometimes a well-developed public
            transportation system, which can reduce parking needs.
10

-------
    The same percentage of transportation-associated infrastructure, such as roads, parking
    lots, driveways, and sidewalks, is allocated to each community acre, based on the curve
    number methodology from the NRCS. For example, each scenario has the same width of
    road, but because the higher-density scenario is more compact, it requires fewer miles of
    roads than the lower-density scenarios. So while the same percentage is applied, the
    amounts differ by scenario. Collector roads or arterials that serve the development are
    not included.
    The modeled stormwater runoff quantity for each scenario is assumed to come from one
    hypothetical outfall.
    The model does not take into account wastewater or drinking water infrastructure, slope,
    or other hydrological interactions that the more complex water modeling tools use.
Summary of Scenarios
Example 1 examines the stormwater runoff impacts on a one-acre lot that accommodates one
house (Scenario A), four houses (Scenario B), or eight houses (Scenario C). Example 2 expands
the analysis to examine stormwater runoff impacts within a lot-level development that accom-
modates the same number of houses. Because of different development densities, this growth
requires different amounts of land. Scenario A requires eight acres for eight houses, Scenario B
requires two acres for eight houses, and Scenario C requires one acre for eight houses.

Examples 3,4, and 5 explore the relationship between density and land consumption by build-
ing in a watershed at different densities. Again, different amounts of land are required
to support the same amount of housing. Examples 6, 7, and 8 examine how the hypothetical
community grows over a 40-year timeframe with different development densities.

The scenarios and scales of development are summarized in Exhibit 2. EPA expects to capture
the differences in stormwater runoff associated with different development densities by using
these three scenarios (Scenarios A, B, and C) at four different scales (one acre, lot, watershed,
and build-out).

                        EXHIBIT 2: Summary of Scenarios
   Scale of Analysis
   Example 1: One acre
Scenario A:
One house per
Scenario B:
Four houses
per acre
Scenario C:
Eight houses
per acre
1 house per acre    4 houses per acre    8 houses per acre
   Example 2: Lot—Each deve-
   lopment lot accommodates
   the same number of houses
8 houses built
on 8 acres
8 houses built
on 2 acres
8 houses built
on 1 acre
                                                                                         11

-------
    Protecting Water Resources with Higher-Density Development
             Example 3: Watershed—
             Each 10,000-acre water-
             shed accommodates the
             same number of houses
10,000 houses
built on 10,000
acres
10,000 houses
built on 2,500
acres or 14 of
the watershed
10,000 houses
built on 1,250
acres or V8 of
the watershed
             Example 4: Watershed—
             Each 10,000-acre water-
             shed is fully built out at
             different densities
             Example 5: Watershed—
             Each scenario accommo-
             dates the same number
             of houses
             Example 6: Hypothetical
             build-out in the year 2000
             Example 7: Hypothetical
             build-out in the year
             2020
             Example 8: Hypothetical
             build-out in the year
             2040
10,000 houses
built on 10,000
acres
80,000 houses
consume 8
watersheds
10,000 ho uses
built on 10,000
acres
20,000 houses
built on 20,000
acres, or 2 water-
sheds

40,000 houses
built on 40,000
acres, or 4 water-
sheds
40,000 houses
built on 10,000
acres
80,000 house
consume 2
watersheds
10,000 houses
built on 2,500
acres
20,000 houses
built on 5,000
acres, or 1/2 of 1
watershed

40,000 houses
built on 10,000
acres, or 1
watershed
80,000 houses
built on 10,000
acres
80,000 houses
consume 1
watershed
10,000 houses
built on 1,250
acres

20,000 houses
built on 2,500
acres, or 14 of 1
watershed

40,000 houses
built on 5,000
acres, or 1/2 of 1
watershed
          Before analyzing the impacts of these different scenarios, it is useful to clarify some underly-
          ing premises. This analysis assumes that:
          1.   Metropolitan regions will continue to grow. This assumption is consistent with U.S. Census
              Bureau projections that the U.S. population will grow by roughly 50 million people by
              2020 (Census, 2000). Given this projected population growth, most communities across
              the country are or will be determining where and how to accommodate expected popu-
              lation increases in their regions.
          2.   Housing density affects the distribution of new growth within a given region, not the
              amount of growth. Individual states and regions grow at different rates depending on
              a variety of factors, including macroeconomic trends (e.g., the technology boom in the
              1980s spurring development in the Silicon Valley region in California) and demographic
              shifts. Distribution and density of new development do not significantly affect these factors.
12

-------
3.  The model focuses on the comparative differences in stormwater runoff between scenar-
   ios, not absolute values. As discussed, using the curve number and event mean concen-
   tration approach can underestimate the total quantity of stormwater runoff for smaller
   storm events and in  areas of lower densities. Because of this and other model simplifica-
   tions discussed above, the analysis does not focus on the absolute value of stormwater
   runoff generated for each scenario but instead focuses on the comparative difference, or
   the delta, in runoff between scenarios.


Results

The results from the eight examples for all three scenarios are presented below.
                           EXAMPLE 1: ONE-ACRE LEVEL
   Scale of Analysis
   One Acre
Scenario A
Scenario B
Scenario C
1  house
4 houses
EPA examined one acre developed at three different densities: one house, four houses, and
eight houses. The results are presented in Exhibit 3. As Exhibit 3 demonstrates, the overall
percent imperviousness for Scenario A is approximately 20 percent with one house per acre,
38 percent for Scenario B with four houses per acre, and 65 percent for Scenario C with eight
houses per acre (Soil Conservation Service, 1986).
      EXHIBIT 3: Total Average Annual Stormwater Runoff for All Scenarios
     Scenario A
  Impervious cover = 20%
  Runoff/acre = 18,700 ftVyr
  Runoff/unit = 18,700 ftVyr
      Scenario  B
   Impervious cover = 38%
   Runoff/acre = 24,800 ftVyr
   Runoff/unit = 6,200 ftVyr
              Scenario C
            Impervious cover = 65%
            Runoff/acre = 39,600 ftVyr
            Runoff/unit = 4,950 ftVyr
                                                                                        13

-------
    Protecting Water Resources with Higher-Density Development
          Examining the estimated average annual runoff at the acre level, as illustrated in Exhibit 4,
          the low-density Scenario A, with just one house, produces an average runoff volume of
          18,700 cubic feet per year (ftVyr). Scenario C, with eight houses, produces 39,600 ftVyr, and
          Scenario B falls between Scenarios A and C at 24,800 ftVyr. In short, looking at the compara-
          tive differences between scenarios, runoff roughly doubles as the number of houses increas-
          es from one house per acre to eight houses per acre. Scenario C, with more houses on the
          acre, has the greatest amount of impervious surface cover and thus generates the most
          runoff at the acre level.

          Looking at the comparative difference of how much runoff each individual house produces,
          in Scenario A, one house yields 18,700 ftVyr, the same as the per acre level. In the denser
          Scenario C, however, each house produces 4,950 ftVyr average runoff.The middle scenario,
          Scenario B, produces considerably less runoff—6,200
          ftVyr—per house than Scenario A, but more than
          Scenario C. Each house in Scenario B produces approxi-
          mately 67 percent less runoff than a house in Scenario A,
          and each house in Scenario C produces 74 percent less
          runoff than a house in Scenario A. This is because the
          houses in Scenarios B and C create less impervious sur-
          face per  house than the house in Scenario A. Therefore,
          per house, each home in the higher-density communities
          results in less stormwater runoff.
                           Each house in Scenario B
                           produces approximately
                           67 percent less runoff than
                           a house in Scenario A, and
                           each house in Scenario C
                           produces 74 percent less
                           runoff than a house in
                           Scenario A.
          Modeling at the acre level demonstrates that, in this
          example, when density is quadrupled (from one house
          to four houses), stormwater runoff increases by one-
          third per acre, but decreases by two-thirds per house. Moreover, when density increases by a
          factor of eight—from one house to eight houses—stormwater runoff doubles per acre, but
          decreases by almost three-quarters per house.
          These results indicate when runoff is measured by the acre, limiting density does mini-
          mize water quality impacts compared to the higher-density scenarios. However, when
          measured by the house, higher densities produce less stormwater runoff.


                                         EXAMPLE 2: LOT LEVEL
            Scale of Analysis
Scenario A
Scenario B
Scenario C
                                  8 houses built on
                                  8 acres
                   8 houses built on
                   2 acres
                  8 houses built
                  1 acre
14

-------
           EXHIBIT 4: Each Scenario Accommodates Eight Houses
Impervious cover = 20%
Total runoff (18,700 ftYyrx
 8 acres) = 149,600 ftVyr
Runoff/house =
 18,700 ftVyr
   Scenario B
Impervious cover = 38%
Total runoff (24,800 ftVyr x
  2 acres) = 49,600 ftVyr
Runoff/house
  6,200 ftVyr
   Scenario C
                         ****
Impervious cover = 65%       Total runoff = 39,600 ftVyr
                                Runoff/house
                                 4,950 ftVyr

-------
    Protecting Water Resources with Higher-Density Development
          For each development to accommodate the same num-
          ber of houses, the lower-density scenarios require more
          land to accommodate the same number of houses that
          Scenario C has accommodated on one acre. Specifically,
          Scenario A must develop seven additional acres, or eight
          acres total, to accommodate the same number of houses
          as Scenario C. Scenario B must develop two acres to accommodate the same number of
          houses. Exhibit 4 illustrates.
                   The increase in runoff
                   for Scenario A is due to
                   the additional land
                   consumption.
          With each scenario accommodating the same number of houses, this analysis shows that
          total average runoff in Scenario A is 149,600 ftVyr (18,700 ftVyrx 8 acres), which is a 278 per-
          cent increase from the 39,600 ft3/yr total runoff in Scenario C. Total average runoff from eight
          houses in Scenario B is 49,600 ftVyr (24,800 ftVyr x 2 acres), which is a 25 percent increase in
          runoff from Scenario C. The increase in runoff for Scenario A is due to the additional land con-
          sumption and associated runoff. The impervious cover for Scenario A remains the same at 20
          percent, but now, seven additional  acres have 20 percent impervious cover.

          Examining the comparative difference in runoff between scenarios shows that lower
          densities can create less total impervious cover, but produce more runoff when the
          number of houses is kept consistent between scenarios. Furthermore, the higher-density
          scenario produces less runoff per house and per lot.


                                     EXAMPLE 3: WATERSHED LEVEL
             Scale of Analysis
             Watershed—Each 10,000-acre
             watershed accommodates
             the same number of houses
Scenario A
Scenario B
Scenario C
10,000 houses    10,000 houses     10,000 houses
built on 10,000   built on 2,500     built on 1,250
          Taking the analysis to the watershed level, EPA examined the comparative watershed
          stormwater runoff impacts from accommodating growth at different densities. The water-
          shed used in this analysis is a hypothetical 10,000-acre watershed accommodating only
          houses. As discussed, the modeling does not include retail, business centers, farms, or any
          other land uses typically seen in communities, nor does it take into consideration where the
          development occurs within the watershed. Research has shown that upper sub-watersheds,
          which contain smaller streams, are generally more sensitive to development than  lower
          sub-watersheds (Center for Watershed Protection, 2001).

          Accommodating 10,000 houses at one house per acre in the 10,000-acre watershed would
          fully build out the watershed. At the higher density of four houses per acre, one-quarter of the
          watershed would be developed, and at eight houses per acre, one-eighth of the watershed
          would be developed. Exhibit 5 shows the runoff associated with each of these scenarios.
16

-------
       EXHIBITS: 10,000-Acre Watershed Accommodating 10,000 Houses

     Scenario A
  Scenario B
  Scenario C

    10,000 houses built on
    10,000 acres produce:
    10,000 acres x 1  house
    x18,700 ft3/yr of
    runoff =
    187 million ft3/yr of
    stormwater runoff
    Site: 20% impervious
    cover
    Watershed: 20%
    impervious cover
10,000 houses built on
2,500 acres produce:
2,500 acres x 4 houses
x 6,200 ft3/yr of
runoff =
62 million ft3/yr
of stormwater runoff
Site: 38% impervious
cover
Watershed: 9.5%
impervious cover
10,000 houses built on
1,250 acres produce:
1,250 acres x 8 houses
x 4,950 ft3/yr of
runoff =
49.5 million ft3/yr of
stormwater runoff
Site: 65% impervious
cover
Watershed: 8.1%
impervious cover

As Exhibit 5 illustrates, if development occurs at a lower density, e.g., one house per acre,
the entire watershed will be built out, generating 187 million ft3/yr of stormwater runoff.
Scenario B, at four houses per acre, consumes less land  and produces approximately 62 mil-
lion ft3/yr of stormwater runoff, while Scenario C, at the highest density, consumes the least
amount of land and produces just 49.5 million ft3/yr of stormwater runoff. Looking at the
comparative differences, Scenario A generates approximately three times as much runoff
from development as Scenario B, and approximately four times as much stormwater
runoff as Scenario C.
Exhibit 5 also illustrates that, in this example, overall
impervious cover for the watershed decreases as site den-
sity increases. Scenario C, which has a lot-level impervi-
ousness of 65 percent, has a watershed-level impervious-
ness of only 8.1 percent, which is lower than the 10
                     Overall impervious
                     cover for the water-
                     shed decreases as site
                     density increases.
                                                                                       17

-------
    Protecting Water Resources with Higher- Density Development
          percent threshold discussed earlier. Scenario B, with a density of four houses per acre, has a
          site-level impervious cover of 38 percent, but a watershed imperviousness of 9.5 percent, which
          is still lower than the 10 percent threshold. Finally, Scenario A, at a lot-level imperviousness of
          20 percent, has the same overall imperviousness at the watershed level. Both of the higher-
          density scenarios consume less land and maintain below-the-threshold imperviousness.
          This simplistic illustration demonstrates a basic point of
          this analysis—higher-density developments can minimize
          stormwater impacts because they consume less land than
          their lower-density counterparts. For example, imagine if
          Manhattan, which accommodates 1.54 million people on
          14,720 acres (23 square miles) (Census, 2000), were devel-
          oped not at its current density of 52 houses per acre, but
          at one or four houses per acre. At one house per acre,
          Manhattan would need approximately 750,000 more
          acres, or an additional 1,170 square miles, to accommo-
                       At one house per acre,
                       Manhattan would need
                       approximately 750,000
                       more acres, or an addi-
                       tional  1,170 square miles,
                       to accommodate its current
                       population at two people
                       per household.
          date its current population at two people per household.
          That's approximately the size of Rhode Island. At four houses per acre, Manhattan would
          need approximately 175,000 more acres, or an additional 273 square miles.

          Reducing land consumption is crucial to preserving water quality because, as discussed pre-
          viously, preserving large, continuous areas of open space and sensitive ecological areas is
          critical for maintaining watershed services. In addition, because of their dense development
          pattern, Scenarios B and C may realize additional stormwater benefits if the developed land is
          appropriately sited  in the watershed to protect sensitive ecological areas, such as headwa-
          ters, wetlands, riparian corridors, and floodplains.


                         EXAMPLE 4: REMAINING LAND  IN THE WATERSHED DEVELOPED

          What happens if the remaining undeveloped parts of the watershed in Scenarios B and C are
          developed? Exhibit  6 considers this situation.
             Scale of Analysis
             Watershed—Each 10,000-
             acre watershed is fully built
             out at different densities
Scenario A
10,000 houses
built on 10,000
acres
Scenario B
40,000 houses
built on 10,000
acres
Scenario C
80,000 houses
built on 10,000
acres
uuu
18

-------
     EXHIBIT 6:10,000-Acre Watershed Accommodating Different Numbers of Houses
I
Scenario A
              t
      The watershed is fully
      built out at 1 house per
      acre. 10,000 acres
      accommodates 10,000
      houses, translating to:
      10,000 acres x 1 house x
      18,700 ft3/yr of runoff =
      187 million ft3/yr
      stormwater runoff
      Site: 20% impervious
      cover
      Watershed: 20%
      impervious cover
Scenario B
                         The watershed is fully
                         built out at 4 houses per
                         acre. 10,000 acres
                         accommodates 40,000
                         houses, translating to:
                         10,000 acres x 4 houses
                         x 6,200 ft3/yr of runoff =
                         248 million ft3/yr
                         stormwater runoff
                         Site: 38% impervious
                         cover
                         Watershed: 38%
                         impervious cover
Scenario C
                         The watershed is fully
                         built out at 8 houses per
                         acre. 10,000 acres
                         accommodates 80,000
                         houses, translating to:
                         10,000 acres x 8 houses x
                         4,950 ft3/yr of runoff =
                         396 million ft3/yr
                         stormwater runoff
                         Site: 65% impervious
                         cover
                         Watershed: 65%
                         impervious cover

-------
    Protecting Water Resources with Higher-Density Development
          Each watershed is fully built out, and the watershed
          developed at the highest density (Scenario C) is generat-
          ing approximately double the total stormwater runoff of
          Scenario A. Scenario B is generating approximately one-
          third more runoff than  Scenario A. Similar to the acre-
          level and lot-level results, Scenario C has the highest
          degree of impervious cover at 65 percent, while Scenario
          A maintains the lowest level at 20 percent.
                     Scenarios A and B accom-
                     modate only a small por-
                     tion of the expected
                     growth. The rest will
                     have to  be built in
                     other watersheds.
          The higher densities found in Scenario B and C are degrading their watershed services to a
          greater extent than Scenario A. However, the number of houses accommodated in each commu-
          nity is not the same. Scenario B is accommodating 30,000 more houses (four times the number
          of Scenario A), and Scenario C is accommodating 70,000 more houses (eight times the number
          of Scenario A). Recall that density limits shift growth and do not generally affect the total
          amount of growth in a given time period. Therefore, this is not a fair comparison. Scenarios A
          and B accommodate only one-eighth and one-half, respectively, of the 80,000 houses accommo-
          dated in Scenario C. Where do the other houses, households, and families go? To get a true
          appreciation for the effects of density, Scenarios A and B must also show where those homes
          will be accommodated. It is likely that they would be built in nearby or adjacent watersheds.
          Our hypothetical community that develops at one house per acre (Scenario A) is able to accom-
          modate only 10,000 houses. For the community that develops at that density to accommodate
          the same number of houses that Scenario C contains, it must disturb and develop land  from
          nearby or adjacent watersheds.


                         EXAMPLE 5: ACCOMMODATING THE SAME NUMBER OF HOUSES
             Scale of Analysis
             Watershed—Each scenario
             accommodates the same
             number of houses
Scenario A
1 house per
acre—80,000
houses con-
sume 8
watersheds
Scenario B
4 houses per
acre—80,000
houses con-
sume 2
watersheds
Scenario C
8 houses per
acre—80,000
houses con-
sume 1
watershed
          As discussed, the U.S. population will increase by an estimated 50 million people by 2020.
          Different areas of the country will grow at different rates in the future. Whether a region
          anticipates 1,000 or 80,000 new households to come to the region over the next 10 years,
          comparisons between build-out scenarios must keep the number of homes consistent. In this
          case, if Scenario C is developed so that its entire watershed is built out to 80,000 houses, then
          for a fair comparison, Scenarios A and B must also include 80,000 houses. Exhibit 7 illustrates
          this situation.
20

-------
                  EXHIBIT 7:80,000 Houses Accommodated
  Scenario A
   Scenario B
At 1 house per acre,
80,000 houses require
80,000 acres, or 8 wafer-
sheds, translating to:
80,000 acres x 1  house x
18,700 ft3/yr of runoff =
1.496 billion  ft3/yr of
stormwater runoff
8 watersheds at 20%
impervious cover
   Scenario C
At 4 houses per acre,
80,000 houses require
20,000 acres, or 2 wafer-
sheds, translating to:
20,000 acres x 4 houses x
6,200 ft3/yr of runoff =
496 million ft3/yr of
stormwater runoff
2 watersheds at 38%
impervious cover
At 8 houses per acre,
80,000 houses require
10,000 acres, or 7 wafer-
shed, translating to:
10,000 acres x 8 houses x
4,950 ft3/yr of runoff =
396 million ft3/yr of
stormwater runoff
1 watershed  at 65%
impervious cover

-------
    Protecting Water Resources with Higher-Density Development
          When the number of houses is kept consistent, Scenario A would need to develop an addi-
          tional seven watersheds (assuming the same size watersheds) and Scenario B would need to
          develop one additional watershed to accommodate the same growth found in Scenario C.
          As Exhibit 7 demonstrates, for Scenario A to accommo-
          date the additional 70,000 homes already accommodat-
          ed in Scenario C, it must develop another seven
          watersheds.This generates 1.496 billion ftVyr of
          stormwater runoff. Scenario C, with a development den-
          sity of eight houses per acre, has still developed just one
          watershed and is generating approximately 74 percent
          less stormwater runoff than Scenario A—or 396 million
          ftVyr. Scenario B, at four houses per acre, is generating
          496 million ftVyr runoff, or two-thirds less runoff than
          Scenario A, but 100 million ftVyr more than Scenario C.
                    Scenario A would need to
                    develop an additional seven
                    watersheds and Scenario B
                    would need to develop one
                    additional watershed in
                    order to accommodate
                    the same growth found
                    in Scenario C.
                      EXAMPLE 6: TIME SERIES BUILD-OUT ANALYSIS: BUILD-OUT IN 2000
            Scale of Analysis
             Hypothetical build-out in
             the year 2000
Scenario A
Scenario B
Scenario C
10,000 houses
built on 10,000
acres
10,000 houses
built on 2,500
acres
10,000 houses
built on 1,250
acres
          Another way to examine this issue is to look at what happens to build-out of the three sce-
          narios over time. A basic assumption for EPA's modeling is that growth is coming to the
          hypothetical community, and that growth will be accommodated within a fixed time
          horizon. But what happens to growth in the hypothetical community over several,
          sequential time horizons?
          Given the dynamic nature of population growth, what will build-out look like in the
          hypothetical community in 2000, 2020, and 2040 at different development densities? The
          next several examples examine the amount of land required to accommodate increasing
          populations within a watershed that develops at different densities. The purpose of this
          time series build-out is to examine how much land is consumed as the population grows
          in 20-year increments.

          Starting in the year 2000, the three watersheds each begin with 10,000 homes. The only dif-
          ference between the watersheds is the densities at which the building occurs. In 2000, they
          might look something like Exhibit 8.
22

-------
          EXHIBIT 8: Time Series Build-out Analysis: Build-out in 2000
     Scenario A
 Scenario B
          Scenario C
          t
     10,000 houses on
     10,000 acres at a densi-
     ty of 1 house per acre
     consume 1 entire
     watershed.
10,000 houses on
2,500 acres at a density
of 4 houses per acre
consume 14 of 1
watershed.
          10,000 houses on
          1,250 acres at a density
          of 8 houses per acre
          consume Vsof 1
          watershed.
As previously demonstrated in Example 3, building at higher densities consumes, or converts,
less land within the watershed. Scenario A, developing at one unit per acre, requires the
entire 10,000-acre watershed to accommodate 10,000 houses. Scenario C, on the other hand,
developing at eight units an acre, requires significantly less land to accommodate the same
amount of development.


            EXAMPLE 7: TIME SERIES BUILD-OUT ANALYSIS: BUILD-OUT IN 2020
   Scale of Analysis
   Hypothetical build-out in the
   year 2020
Scenario A
20,000 houses
built on 20,000
acres, or 2
watersheds
Scenario B
20,000 houses
built on 5,000
acres, or V* of 1
watershed
Scenario C
20,000 houses
built on 2,500
acres, or 14 of 1
watershed
Fast-forwarding 20 years, the population in the hypothetical community has doubled from
10,000 houses to 20,000 houses. Each scenario must accommodate this additional growth at
different development densities. Exhibit 9 demonstrates how this development might look.
                                                                                       23

-------
    Protecting Water Resources with Higher-Density Development
                    EXHIBIT 9: Time Series Build-out Analysis: Build-out in 2020
               Scenario A
  Scenario B
           Scenario C
              20,000 houses accom-
              modated on 20,000
              acres at a density of 7
              house per acre will con-
              sume 2 watersheds.
20,000 houses accom-
modated on 5,000
acres at a density of 4
houses per acre will con-
sume !/2 of 1 watershed.
         20,000 houses accom-
         modated on 2,500
         acres at a density of
         eight houses per acre
         will consume 14 of 1
         watershed.
          As Exhibit 9 demonstrates, Scenario A, developing at one house per acre, requires another
          whole watershed to accommodate the additional growth. Scenarios B and C, developing
          at higher densities, can accommodate the additional growth within the same watershed.
          Moreover, by developing at higher densities within the watershed, ample open space or
          otherwise undeveloped land remains to perform critical watershed functions. No such land
          exists in Scenario A, and, as previously discussed, lawns typically associated with one house
          per acre are not able to provide the same type of watershed services as forests, meadows,
          or other types of unconverted land.


                      EXAMPLE 8: TIME SERIES BUILD-OUT ANALYSIS: BUILD-OUT IN 2040
             Scale of Analysis
             Hypothetical build-out in
             the year 2040
 Scenario A
 40,000 houses
 built on 40,000
 acres, or 4
 watersheds
Scenario B
Scenario C
40,000 houses
built on 10,000
acres,  or 1
watershed
40,000 houses
built on 5,000
acres,  or V-i of 1
watershed
24

-------
The hypothetical community continues to grow and, in another 20 years, population has
doubled again, requiring each scenario to accommodate 20,000 more homes at different
development densities. Exhibit 10 demonstrates how this development might look.
         EXHIBIT 10: Time Series Build-out Analysis: Build-out in 2040
     Scenario A
  Scenario B
  Scenario C
    40,000 houses on
    40,000 acres at a den-
    sity of 1 house per acre
    will consume 4
    watersheds.
40,000 houses on
10,000 acres at a den-
sity of 4 houses per
acre will consume 1
watershed.
40,000 houses on
5,000 acres at a density
of 8 houses per acre
will consume V* of 1
watershed.
As Exhibit 10 demonstrates, Scenario A, developing at
one house per acre, must develop land in four water-
sheds, or 40,000 acres, to accommodate all its houses.
Scenario B, developing at a slightly higher density, uses
its remaining land to accommodate the additional
growth. Scenario C is still developing within the same
watershed and still has additional land available to pro-
vide watershed services. Scenario A and B do not. Any
land for watershed services would need to come from
additional watersheds.
                    Lower-density develop-
                    ment always requires
                    more land than higher
                    densities to accommodate
                    the same amount of
                    growth.
This build-out analysis can continue indefinitely with the same result: lower-density
development always requires more land than higher densities to accommodate the same
amount of growth. Because more land is required, more undeveloped land is converted.
                                                                                    25

-------
    Protecting Water Resources with Higher-Density Development
          Findings/Discussion
          The results indicate when runoff is measured by the acre, limiting density does produce less
          stormwater runoff when compared to the higher-density scenarios. However, when meas-
          ured by the house, higher densities produce less stormwater runoff. So, which is the
          appropriate measure?

          Typically, a planning department analyzes the projected stormwater runoff impacts of a
          developer's proposal based on the acreage, not the number of houses being built. Based on
          the results from the one-acre level example, communities might conclude that lower-density
          development would minimize runoff. Runoff from one house on one acre is roughly half the
          runoff from eight houses. However, where did the other houses, and the people who live in
          those houses, go? The answer is almost always that they went somewhere else in that
          region—very often somewhere within the same watershed. Thus, those households still have
          a stormwater impact. To better understand the stormwater runoff impacts from developing
          at low densities, the impacts associated with those houses locating elsewhere need to be
          taken into account. This approach has two advantages:

            • It acknowledges that the choice is not whether to grow by one house or eight but is
             instead where and how to accommodate the eight houses (or whatever number by
             which the region is expected to grow).

            • It emphasizes minimization of total imperviousness and runoff within a region or water-
             shed rather than from particular sites—which is more consistent with the science indicat-
             ing that imperviousness within the watershed is critical.

          To more fully explore this dynamic, EPA modeled scenarios at three scales—one acre, lot, and
          watershed—and at three different time series build-out examples to examine the premise
          that lower-density development better protects water quality. EPA examined stormwater
          runoff from different development densities to determine the comparative difference
          between scenarios. The higher-density scenarios generated less stormwater runoff per house
          at all scales and time series build-out examples. Exhibit 11 summarizes these findings.
26

-------
                         EXHIBIT 11: Summary of Findings
Scenario
                       Number of    Impervious     Total     Runoff
                         Acres         Cover       Runoff    Per Unit
                       Developed       (%)        (ftVyr)    (ftVyr)
                       Savings
                        Over
                     Scenario A:
                      runoff per
                       unit (%)
One-Acre Level: Different densities developed on one acre

                                        20.0       18,700
A: One house/acre
B: Four houses/acre
C: Eight houses/acre
           18,700
                                        38.0

                                        65.0
 24,800      6,200

 39,600      4,950
Lot Level: Eight houses accommodated at different density levels
Scenario A
Scenario B
Scenario C
                                        20.0

                                        38.0

                                        65.0
149,600

49,600

 9,600
Watershed Level: 10,000 houses accommodated In one 10,000-acre watershed

                        10,000          20.0       187M      18,700

                         2,500           9.5        62 M       6,200

                         1,2f
Scenario A
Scenario B
Scenario C
Scenario
49.5 M
4,950
                      Summary of Build-out Examples
Scenario A
Watershed Level: Time Series Build-out Analysis: Build-out in 2000

                      10,000 houses built on 10,000 acres: 1 watershed is consumed

                      10,000 houses built on 2,500 acres: 14 of 1 watershed is consumed

                      10,000 houses built on 1,250 acres: Vs of 1 watershed is consumed
Scenario C
Watershed Level: Time Series Build-out Analysis: Build-out in 2020
Scenario A
Scenario B
Scenario C
                      20,000 houses built on 20,000 acres: 2 watersheds are consumed

                      20,000 houses built on 5,000 acres: 1/2 of 1 watershed is consumed

                      20,000 houses built on 2,500 acres: 14 of 1 watershed is consumed
Scenario A
Watershed Level: Time Series Build-out Analysis: Build-out in 2040

                      40,000 houses built on 40,000 acres: 4 watersheds are consumed

                      40,000 houses built on 10,000 acres: 1 watershed is consumed

                      40,000 houses built on 5,000 acres: 1/2 of 1 watershed is consumed
Scenario C

-------
    Protecting Water Resources with Higher-Density Development
          Specifically, this analysis demonstrates:
            • With more dense development (Scenario C), runoff
             rates per house decrease by approximately 74 per-
             cent from the least dense scenario (Scenario A);
            • For the same amount of development, denser devel-
             opment produces less runoff and less impervious
             cover than low-density development; and
            • For a given amount of growth, lower-density devel-
             opment uses more of the watershed.
EPA found that the higher-
density scenarios generate
less stormwater runoff per
house at all scales—one
acre, lot, watershed—and
time series build-out
examples.
          Taken together, these findings indicate that low-density development may not always be
          the preferred strategy for reducing stormwater runoff. In addition, the findings indicate that
          higher densities may better protect water quality—especially at the lot and watershed levels.
          Higher-density developments consume less land to accommodate the same number of
          houses as lower density. Consuming less land means less impervious cover is created within
          the watershed. To better protect watershed function, communities must preserve large, con-
          tinuous areas of open space and protect sensitive ecological areas, regardless of how densely
          they develop.

          However, while increasing densities on a regional scale can, on the whole, better protect
          water resources at a regional level, higher-density development can have more site-level
          impervious cover, which can exacerbate water quality problems in nearby or adjacent water-
          bodies. To address this increased impervious cover, numerous site-level techniques are avail-
          able to mitigate development impacts. When used in combination with regional techniques,
          these site-level techniques can prevent, treat, and store runoff and associated pollutants.
          Many of these practices incorporate some elements of low-impact development techniques
          (e.g., rain gardens, bioretention areas, and grass swales), although others go further to
          include changing site-design practices, such as reducing parking spaces, narrowing streets,
          and eliminating cul-de-sacs.
          Incorporating these techniques can
          help communities meet their water
          quality goals and create more interest-
          ing and enjoyable neighborhoods.

          A University of Oregon study,
          Measuring Stormwater Impacts of
          Different Neighborhood Development
          Patterns (University of Oregon, 2001),
          supports this conclusion. The study,
          which included a study site near
          Corvallis, Oregon, compared stormwa-
          ter management strategies in three
          common neighborhood development   Thecityof Portland, Oregon, is developing urban stormwater
          patterns. For example, best manage-     strategies, such as these curb extensions that can absorb the
          ment practices, such as disconnecting   street's runoff from large storm events.
28

-------
residential roofs and paved areas from the stormwater system, introducing swales and water
detention ponds into the storm sewer system, and strategically locating open space, consid-
erably reduced peak water runoff and improved infiltration. The study concluded that "some
of the most effective opportunities for reducing stormwater runoff and decreasing peak flow
are at the site scale and depend on strategic integration with other site planning and design
decisions."The study also found that planting strips and narrower streets significantly
reduced the amount of pavement and, as a result, runoff in developed areas.

A development in Tacoma, Washington, demonstrates that increasing densities and address-
ing stormwater at the site level can work effectively. The Salishan Housing District was built
on Tacoma's eastern edge in the 1940s as temporary housing for ship workers. It is currently a
public housing community with 855 units.
Redevelopment of Salishan will increase densities to       Salishan Housinq District
include 1,200 homes (public housing, affordable and mar-  .  rpniarjnn occ ni|hlir
ket rate rentals, and for-sale units), local retail, a farmers          .      .      .
market, a senior housing facility, a daycare center, a        housing units with 1,200
health clinic, commercial  office space, and an expanded    units. Numerous site-level
community center. Among the most important priorities   strategies  Such as inte-
for the redevelopment is  restoring the water quality of         .   ^    narrowj
Swan Creek, which forms the eastern edge of Salishan.     ^     ^    .            .
The creek is a spawning ground for indigenous salmon     *ne Streets, installing rain
populations that feed into the Puyallup River and Puget    gardens, and daylighting a
Sound. The site plan seeks to restore 65 percent of the     Stream, are used to  restore
land to forest and pervious landscape. In addition, the     thfi ^ef      ^  Qf $wan
streets will be narrowed to reduce impervious surfaces                    . '
and also make the neighborhood more inviting for walk-   Creek and revitalize an
ing. Some streets may be eliminated and replaced with     existing neighborhood.
pedestrian paths. The remaining streets will be bordered
by rain gardens that would accept, filter, and evapotranspire runoff. Most existing street sur-
faces would be reused, although  some may be replaced with pervious pavers.

Communities can enjoy a further reduction in runoff if they take advantage of underused
properties, such as infill, brownfield, or greyfield12 sites. For example, an abandoned shop-
ping center (a greyfield property) is often almost completely impervious cover and is already
producing high volumes of runoff (Sobel, 2002). If this property were redeveloped, the net
runoff increase would likely be zero since the property was already predominately impervi-
ous cover. In many cases, redevelopment of these properties breaks up or removes some
portion of the impervious cover, converting it to pervious cover and  allowing for some
stormwater infiltration. In this case, redevelopment of these properties can produce a
net improvement in regional water quality by decreasing total runoff. Exhibit 12
illustrates this opportunity.
12 Greyfield sites generally refer to abandoned or underutilized shopping malls, strip malls, or other areas that have significant paved sur-
face and little or no contamination.
                                                                                           29

-------
    Protecting Water Resources with Higher-Density Development
                         EXHIBIT 12: Redevelopment of a Greyfield Property
          Redevelopment of a former shopping mall in Boca Raton, Florida, provides an example of this
          type of opportunity. The Mizner Park shopping mall was redesigned from its original pattern
          of a large retail structure surrounded by surface parking lots; the 29-acre site now includes
          272 apartments and townhouses, 103,000 square feet of office space, and 156,000 square feet
          of retail space. Most parking is accommodated in four multistory parking garages. Designed
          as a village within a city, the project has a density five times higher than the rest of the city
          and a mix of large and small retailers, restaurants, and entertainment venues (Cooper, 2003).
          Most significantly, the final build-out of Mizner Park decreased overall impervious surface on
          the site by 15 percent through the addition of a central park plaza, flower and tree planters,
          and a large public amphitheater.

          Redeveloping brownfield and greyfield
          sites can reduce regional land con-
          sumption. A recent George Washington
          University study found that for every
          brownfield acre that is redeveloped, 4.5
          acres of open space are preserved
          (Deason, 2001). In addition to redevel-
          oping brownfield sites, regions can
          identify underused properties or land,
          such as infill  or greyfield sites, and tar-
          get those areas for redevelopment. For
          example, a recent analysis by King
          County, Washington, demonstrated
          that property that is vacant and eligible
          for redevelopment in the county's
          growth areas can accommodate
          263,000 new houses—enough for
The redevelopment of Mizner Park, a former shop-
ping mall, decreased impervious cover by 15 per-
cent through the addition of this central plaza.
30

-------
500,000 people (Pryne, 2002). Redeveloping this property   Redeveloping brownfield
is an opportunity to accommodate new growth without    anc| Qrevfield sites can
expanding into other watersheds. As Kurt Zwikl, execu-        .         .          .
tive director of the Pottstown, Pennsylvania-based         reduce regional land
Schuylkill River Greenway Association, said, "Certainly, if we   consumption.
can get redevelopment going in brownfields and old indus-
trial sites in older riverfront boroughs like Pottstown and Norristown, that's a greenfield further
out in the watershed that has been preserved to absorb more stormwater"(Brandt, 2004).


Other Research

Current  research supports the findings of this study. Several site-specific studies have been
conducted across the United States and in Australia that examine stormwater runoff and
associated pollutants in relation to different development patterns and densities. Several
case studies approach the research question with varying levels of complexity. Studies of
Highland Park, Australia; Belle Hall, South Carolina; New Jersey; Chicago, Illinois; and the
Chesapeake Bay each analyze the differences in runoff and associated water pollution from
different types of development patterns.

Queensland University of Technology, Gold Coast City Council, and the Department of Public
Works in Brisbane, Australia, examined the relationship between water quality and six differ-
ent land uses to offer practical guidance in planning future developments. When comparing
monitored runoff and associated  pollutants from six areas, they found the most protective
strategy for water quality was high-density residential development (Goonetilleke, 2005).

The Belle Hall study, by the South Carolina Coastal Conservation League, examined the water
quality impacts of two development alternatives for a 583-acre site in Mount Pleasant, South
Carolina. The town planners used modeling to examine the  potential water quality impacts of
each site design. In the "Sprawl Scenario,"the property was analyzed as if it developed along
a conventional suburban  pattern. The "Town Scenario" incorporated traditional neighbor-
hood patterns. In each  scenario, the overall density and intensity (the number of homes and
the square feet of commercial and retail space) were held constant. The results found that the
"Sprawl  Scenario"consumed eight times more open space and generated 43 percent more
runoff, four times more sediment, almost four times more nitrogen, and three times more
phosphorous than the "Town Scenario" development (South Carolina Coastal Conservation
League, 1995).

These findings hold at a larger, state scale. New Jersey's State Plan calls for increasing densi-
ties in the state by directing development to existing communities and existing infrastruc-
ture. Researchers at Rutgers University analyzed the water quality impacts from current
development trends and compared them to water quality impacts from the proposed com-
pact development. The study found that compact development would generate significantly
less water pollution than current  development patterns, which are mostly characterized by
low-density development, for all categories of pollutants (Rutgers  University, 2000). The
reductions ranged from over 40 percent for phosphorus and nitrogen to 30 percent for
runoff. These conclusions supported  a similar statewide study completed in 1992 that
                                                                                           31

-------
    Protecting Water Resources with Higher-Density Development
          concluded that compact development would result in 30 percent less runoff and 40 percent
          less water pollution than would a lower-density scenario (Burchell, 1995).

          Researchers at Purdue University examined two possible project sites in the Chicago area
          (Harbor, 2000). The first site was in the city; the  second was on the urban fringe. The study
          found that placing a hypothetical low-density development on the urban fringe would pro-
          duce 10 times more runoff than a higher-density development in the urban core.

          Finally, a study published by the Chesapeake Bay Foundation in 1996 comparing conven-
          tional and clustered suburban development on a rural Virginia tract found that clustering
          would convert 75 percent less land, create 42 percent less impervious surface, and produce
          41 percent less stormwater runoff (Pollard, 2001). These studies suggest that a low-density
          approach to development is not always the preferred strategy for protecting water resources.


          Conclusions

          Our regions, cities, towns, and neighborhoods are growing. Every day, new buildings or
          houses are proposed, planned, and built. Local  governments, working with planners, citizen
          groups, and developers, are thinking about where and how this new development can
          enhance existing neighborhoods  and also protect the community's natural environment.
          They are identifying the characteristics of development that can build vibrant neighbor-
          hoods, rich in natural and historic assets, with jobs, housing, and amenities for all types of
          people. They are directing growth to areas that will maintain and improve the buildings and
          infrastructure in which they have  already invested. In addition to enjoying the many benefits
          of growth, communities are also grappling with growth's challenges, including develop-
          ment's impact on water resources.

          Many communities assume that low-density development automatically protects water
          resources. This study has shown that this assumption is flawed and that pursuit of low-density
          development can in fact be counterproductive, contributing to high rates of land conversion
          and stormwater runoff and missing opportunities to preserve valuable land within watersheds.

          The purpose of this study is to explore the effects of development density on stormwater runoff
          and to illustrate the problems with the assumption that low-density development is automati-
          cally a better strategy to protect water quality.To that end, three different development densities
          were modeled at the one-acre, lot, and watershed levels, as well as in the time series build-out
          examples. The modeling results suggest that low-density development is not always the pre-
          ferred strategy for protecting water resources. Furthermore, the results seem to suggest that
          higher-density development could better protect regional water quality because it consumes
          less land to accommodate the same number of homes.

          However, while this study shows that low-density development does not automatically better
          protect water  resources, it does not conclude that high-density development is therefore neces-
          sarily more protective. This study has not considered all factors, such as location of development
          within the watershed, varying  soil types, slope, advanced post-construction controls (and their
          performance over time), and many other factors.  In that sense, this study concludes that there
32

-------
are good reasons to consider higher-density development
as a strategy that can better protect water resources than
lower-density development. However, any bias toward
either is inappropriate from a water perspective. A superior
approach to protect water resources locally is likely to be
some combination of development densities, based on
local factors, incorporating adequate open space, preserv-
ing critical ecological and buffer areas, and
minimizing  land disturbance.
Additional relevant infor-
mation can be found in
these resources:
•  Protecting Water Resources
with Smart Growth, available
at: www.epa.gov/smart-
g rowth/pdf/waterresou rces
_with_sg.pdf.
•  Creating Great Neighbor-
hoods: Density in Your
Community, available at:
www.epa.gov/smart
growth/pdf/density.pdf.
These conclusions have implications for how communities
can enjoy the benefits of growth and development while
also protecting their water quality. Additional relevant infor-
mation can be found in other resources, such as Protecting
Water Resources with Smart Growth and Using Smart Growth
Techniques as Stormwater Best Management Practices.^3 Both
publications draw on the experience of local governments,
which has shown that regional and site-specific strategies are most effective when implemented
together. In addition, Creating Great Neighborhoods: Density in Your Community, by the Local
Government Commission and the National Association of Realtors, can provide
information on some of the other benefits from density that communities can enjoy.

Nationwide, state and local governments are considering the environmental implications of
development patterns. As low-density development and its attendant infrastructure consume
previously undeveloped land and create stretches of impervious cover throughout a region, the
environment is increasingly affected. In turn, these land alterations are not only likely to degrade
the quality of the individual watershed, but are also likely to degrade a larger number of water-
sheds. EPA believes that increasing development densities is one strategy communities can use
to minimize regional water quality impacts.
 5 Forthcoming EPA publication.
                                                                                            33

-------
          References and Bibliography
          American Farmland Trust. 1994. Farming on the Edge: A New Look at the Importance and
              Vulnerability of Agriculture Near American Cities. Washington, DC.

          Arnold, C. L. and C. J. Gibbons. 1996. "Impervious Surface Coverage: The Emergence of a Key
              Environmental Indicator." Journal of the American Planning Association 62.2: 243-258.

          Association of Bay Area Governments. 2002. "Smart Growth Strategy: Shaping the Future of
              the Nine-County Bay Area."Alternatives Report.

          Burchell, R.W. and  D. Listokin. 1995."Land, Infrastructure, Housing Costs and Fiscal Impacts
              Associated with Growth: The Literature on the Impacts of Sprawl Versus Managed
              Growth." New Brunswick, NJ: Rutgers University, Center for Urban Policy Research. As
              summarized in The Technological Reshaping of Metropolitan America, Office of Technology
              Assessment, OTA-ETI-643, Washington, DC.

          Brandt, Evan. 2004. "Schuylkill River Coalition Wins $1.15 Million Grant."
              . Phoenixville, PA. July 24.

          Brown, Patricia Leigh. 2002. "The Chroming of the Front Yard." New York Times. June 13.

          Cappiella, K. and K. Brown. 2001. Impervious Cover and Land Use in the Chesapeake Bay
              Watershed. Ellicott City, MD: Center for Watershed Protection.

          Caraco, Deb. 1998. Rapid Watershed Planning Handbook—A Comprehensive Guide for
              Managing Urban Watersheds. Ellicott City, MD: Center for Watershed Protection.

          Center for Watershed Protection. 2001. Redevelopment Roundtable Consensus Agreement:
              Smart Site Practices for Redevelopment and Infill Projects. Ellicott City, MD.

          Chesapeake Bay Foundation. 2001. State of the Bay Report. Annapolis, MD.

          City of Durham, North Carolina. 2001. Overlay Districts. Zoning Ordinance.
              .

          Cobb County, Georgia. 2003. Community Development. Planning and Zoning. .

          Cooper, Carry. 2003. Projects—Mizner Park. .

          Cote,  M.P., J. Clausen, B. Morton, P. Stacey, S. Zaremba. 2000. Jordan Cove Urban Watershed
              National Monitoring Project. Presented at the National  Conference on Tools for Urban
              Water and Resource Management Paper Protection, February 7-10, 2000, in Chicago, IL.

          Deason, Jonathan, et al. 2001. Public Policies and Private Decisions Affecting the Redevelopment
              of Brownfields: An Analysis of Critical Factors, Relative Weights and Area Differentials.
              Prepared for U.S. EPA, Office of Solid Waste and Emergency Response. Washington, DC:
              The George Washington University.
34

-------
Engdahl, J. 1999. Impacts of Residential Construction on Water Quality and Quantity in
    Connecticut. Storrs, CT: University of Connecticut.

Ewing, Reid. 1999. Pedestrian and Transit-Friendly Design: A Primer for Smart Growth.
    Washington, DC: International City/County Management Association.

Farris, J. Terrence. 2001. "The  Barriers to Using Urban Infill Development to Achieve Smart
    Growth." Housing Policy Debate. 12.1: Fannie Mae Foundation.

Franklin  County, Ohio. 2003. Zoning. Zoning Code, .

Non-Point Education for Municipal Officials. 2002."Impacts of Development on Waterways."
    NEMO Factsheet. Haddam, CT: University of Connecticut.

New Jersey Office of Planning. 2002. .

New York City Department of Environmental Protection. 2002. New York City's Water Supply
    System, .
                                                                                           35

-------
          Orfield, Myron. 1997. Metropolises: A Regional Agenda for Community and Stability.
               Washington, DC: Brookings Institution Press.
          Pendall, Rolf. 1999. "Do Land-Use Controls Cause Sprawl?"Environment and Planning B:
               Planning and Design, 26.4:555.
          Pendall, Rolf. 2000. "Local Land Use Regulation and the Chain of Exclusion." Journal of the
               American Planning Association, 66.2:125.
          Persky, Joe. 2002. Peer review comments to EPA.
          Pollard, Trip. 2001. "Greening the American Dream?"Planning. 67.10:110-116.
          Portland Metro. 2000. The Nature of 2040: The region's 50-year plan for managing growth.
               .
          Pryne, Eric. 2002. "20 Years'Worth of County Land!" Seattle Times. May 20.
          Puget Sound Regional Council. "Overview of Vision 2020." .
          Rein, Lisa and David Cho. 2002. "In Defense of the Front Lawn: Fairfax Attacks Crowding With
               Ban on Oversize Driveways." Washington Post. June 4.
          Rutgers University. 2000. The Costs and Benefits of Alternative Growth Patterns: The Impact
               Assessment of the New Jersey State Plan. New Brunswick, NJ: Center for Urban Policy
               and Research.
          Schueler,Tom. 1994. "The Importance of Imperviousness." Watershed Protection Techniques.
               1.3:100-111.
          Schueler,Tom. 1995. "The Peculiarities of Perviousness." Watershed Protection Techniques. 2.1.
          Schueler,Tom. 2000. "The Compaction of Urban Soil." Techniques for Watershed Protection.
               Ellicott City, MD: Center for Watershed Protection.
          Sobel, Lee and Steven  Bozdin. 2002. Greyfields into Goldfields: Dead Malls Become Living
               Neighborhoods. San Francisco, CA: Congress for New Urbanism.
          Soil Conservation Service. 1986. Urban Hydrology for Small Watersheds. Washington, DC:
               Technical Release No. 55 (TR-55).
          South Carolina Coastal Conservation League, Environmental Protection Agency, National
               Oceanic and Atmospheric Administration, South Carolina Department of Health and
               Environment; Town of Mount Pleasant. 1995. The Belle Hall Study: Sprawl vs. Traditional
               Town: Environmental Implications. South Miami, FL: Dover, Kohl, and Partners.
          Stormwater Managers Resource Center. 2002. .
          St. Paul/Minneapolis Metropolitan Council. "Help Shape a Regional Vision and Strategy for
               How We Will Grow" Blueprint 2030.
36

-------
Trust for Public Land and the National Association of Counties. 2002. Volume 1: Local
    Greenprinting for Growth: Using Land Conservation to Guide Growth and Preserve the
    Character of Our Communities. Washington, DC: Trust for Public Land.

University of Oregon. 2001. "Measuring Stormwater Impacts of Different Neighborhood
    Development Patterns." Neighborhood Lab.

U.S. Census Bureau. 2000. Annual Projections of the Total Resident Population as of July 1: Middle,
    Lowest, Highest, and Zero International Migration Series, 1999 to 2100. Washington, DC.

U.S. Census Bureau. 2001. American Housing Survey for the United States in 2001.
    Washington, DC.

U.S. Congress Office of Technology Assessment. 1995. The Technological Reshaping of
    Metropolitan America. Washington, DC: 1 OTA-ETI-643.

U.S. Department of Agricultural, Economic Research Service, Natural Resources and
    Environmental Division. 1997a."Major Land Use Changes in the Contiguous 48 States."
    Agricultural Resources and Environmental Indicators (AREI) Updates, No. 3. Washington DC.

U.S. Department of Agriculture, Economic Research Service, Natural Resources and
    Environment Division. 1997b. National Resources Inventory.

U.S. Department of Agriculture, Natural Resources Conservation Service. 200']. Impact of Soil
    Disturbance During Construction on Bulk Density and Infiltration in Ocean County, New
    Jersey. Ocean County Soil Conservation District.

U.S. Environmental Protection Agency, Office of Water. 1997. Urbanization and Streams:
    Studies of HydrologicalImpacts. Washington, DC: EPA 841 -R-97-009.

U.S. Environmental Protection Agency. 2000a. National Water Quality Inventory: 1998 Report to
    Congress.

U.S. Environmental Protection Agency, Office of Water. 2000b. Water Quality Conditions in the
    United States: A Profile from the 1998 National Water Quality Inventory Report to Congress.
    Washington, DC: EPA 841-F-00-006.

U.S. Environmental Protection Agency, Development, Community and Environment Division.
    2001. Our Built and Natural Environments: A Technical Review of the Interactions between
    Land Use, Transportation, and Environmental Quality. Washington, DC: EPA 123-R-01 -002.

U.S. Environmental Protection Agency, Development, Community and Environment Division.
    2004. Protecting Water Resources with Smart Growth. Washington, DC: EPA 231 -R-04-002.
                                                                                             37

-------
U.S. Environmental Protection Agency, Development, Community and Environment Division.
    2002. Technical Approach for SG WATER: Smart Growth Water Assessment Tool for
    Estimating Runoff. Unpublished.

U.S. Government Accounting Office. 2001. Water Infrastructure: Information on Federal and
    State Financial Assistance. Washington, DC: GAO-02-134.

Woodworth, James, et al. 2002. Out of the Gutter: Reducing Polluted Runoff in the District of
    Columbia. Washington, DC: Natural Resources Defense Council.
       j Recycled/Recyclable-Printed with vegetable oil based inks on 100% (minimum 50% postconsumer) recycled paper.
                                                                                               38

-------

-------
•* -
    United States
    Environmental Protection Agency
    (1807-T)
    Washington, DC 20460

    Official Business
    Penalty for Private Use $300

    EPA231-R-06-001
    January 2006
    www.epa.gov/smartgrowth

-------