<%BANNER%>

Impervious Surface and Water Quality in Wake County

Permanent Link: http://ufdc.ufl.edu/UFE0024556/00001

Material Information

Title: Impervious Surface and Water Quality in Wake County An Analysis of Recent Development, Impacts, and Opportunities
Physical Description: 1 online resource (93 p.)
Language: english
Creator: Petrosky, Jacob
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: carolina, county, development, green, growth, impervious, north, pavement, pervious, piedmont, planning, quality, raleigh, roof, sprawl, suburban, surface, wake, water
Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Wake County, North Carolina has experienced significant growth over the last twenty years. Between 1996 and 2006 the population of Wake County has grown from 423,380 to 790,007 people. The population is expected to double in the next 30 years. Studies have shown a correlation between urban growth and degraded water quality. This study quantifies impervious surface in selected watersheds within Wake County using Feature Analyst, feature extraction software by Visual Learning Systems, which uses hierarchical learning techniques to extract features based on color, shape, and adjacency. The study determines increases in impervious surface in the study areas and compares these changes with trends in monthly water quality data. Impervious surface rates by land use were produced for the study areas. These rates are compared to impervious surface rates developed in previous studies to evaluate if current development trends remain compatible with these measures. The effectiveness of alternative design techniques, such as green rooftops and pervious pavements are estimated. This study proves that typical suburban development outside of rapidly growing areas in the North Carolina Piedmont can result in a substantial increase in impervious surface. Results of the water quality data analysis, using a Mann-Kendall test for trend, indicate statistically significant trends of decreasing water quality in the study areas. The study estimates that 47% of all new impervious surface constructed in the study areas between 1999 and 2005 could have been mitigated by incorporating green roofs and pervious pavements in new development.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jacob Petrosky.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2009.
Local: Adviser: Steiner, Ruth L.
Local: Co-adviser: Zwick, Paul D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024556:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024556/00001

Material Information

Title: Impervious Surface and Water Quality in Wake County An Analysis of Recent Development, Impacts, and Opportunities
Physical Description: 1 online resource (93 p.)
Language: english
Creator: Petrosky, Jacob
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: carolina, county, development, green, growth, impervious, north, pavement, pervious, piedmont, planning, quality, raleigh, roof, sprawl, suburban, surface, wake, water
Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Wake County, North Carolina has experienced significant growth over the last twenty years. Between 1996 and 2006 the population of Wake County has grown from 423,380 to 790,007 people. The population is expected to double in the next 30 years. Studies have shown a correlation between urban growth and degraded water quality. This study quantifies impervious surface in selected watersheds within Wake County using Feature Analyst, feature extraction software by Visual Learning Systems, which uses hierarchical learning techniques to extract features based on color, shape, and adjacency. The study determines increases in impervious surface in the study areas and compares these changes with trends in monthly water quality data. Impervious surface rates by land use were produced for the study areas. These rates are compared to impervious surface rates developed in previous studies to evaluate if current development trends remain compatible with these measures. The effectiveness of alternative design techniques, such as green rooftops and pervious pavements are estimated. This study proves that typical suburban development outside of rapidly growing areas in the North Carolina Piedmont can result in a substantial increase in impervious surface. Results of the water quality data analysis, using a Mann-Kendall test for trend, indicate statistically significant trends of decreasing water quality in the study areas. The study estimates that 47% of all new impervious surface constructed in the study areas between 1999 and 2005 could have been mitigated by incorporating green roofs and pervious pavements in new development.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jacob Petrosky.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2009.
Local: Adviser: Steiner, Ruth L.
Local: Co-adviser: Zwick, Paul D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024556:00001


This item has the following downloads:


Full Text

PAGE 1

1 IMPERVIOUS SURFACE AND WATER QUALITY IN WAKE COUNTY: AN ANALYSIS OF RECENT DEVELOPMENT, IMPACTS, AND OPPORTUNITIES By JAKE PETROSKY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Jake Petrosky

PAGE 3

3 To Mark and Carrie Petrosky, two of the greatest parents in the world

PAGE 4

4 ACKNOWLEDGMENTS I thank my parents for teaching me a strong work ethic and supporting me in my unlikely decision to go back to school and get a Masters degree. I also would like to acknowledge Dr. Ruth Steiner and Dr. Paul Zwick for their support, guidance, and help in acquiring the software necessary to complete this study.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLE DGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT .......................................................................................................................................... 9 CHAPTER 1 INTRODUCTION ....................................................................................................................... 11 Introducti on ................................................................................................................................. 11 The Built Environment and Urban Land Use Change .............................................................. 11 Study Area ................................................................................................................................... 15 Conceptual Model ....................................................................................................................... 17 Objectives .................................................................................................................................... 18 2 LITERATURE REVIEW ........................................................................................................... 19 Trends .......................................................................................................................................... 19 Urban Growth in the U.S. ........................................................................................................... 19 Urbaniza tion Impacts on Hydrology .......................................................................................... 20 Impervious Surface and Water Quality ...................................................................................... 21 Impervious Surface as an Indicator ............................................................................................ 23 Measurement Techniques ........................................................................................................... 24 Statistical Analysis of Water Quality ......................................................................................... 27 Alternative Stormwater Design .................................................................................................. 28 Low Impact Development ................................................................................................... 29 Green Roofs ......................................................................................................................... 30 Pervious Pavements ............................................................................................................. 31 Headwater Stre ets ................................................................................................................ 32 Summary .............................................................................................................................. 33 3 METHODOLOGY ...................................................................................................................... 35 Methodology Overview .............................................................................................................. 35 Time Frame and Study Area Selection ...................................................................................... 35 Study Area Selection Process ..................................................................................................... 36 Delineating Watersheds .............................................................................................................. 37 Measurement of Variables .......................................................................................................... 38 Impervious Surface in 2005 ................................................................................................ 38 Impervious Surface 1999 .................................................................................................... 41 Analysis Methodology ................................................................................................................ 42

PAGE 6

6 Quantifying Land Use by Watershed ................................................................................. 43 Quantifying Road Density ................................................................................................... 43 Water Quality Analysis ....................................................................................................... 44 Mann -Kendall in R .............................................................................................................. 45 4 DESCRIPTION OF WATERSHEDS ........................................................................................ 52 Over view ...................................................................................................................................... 52 Middle Creek ............................................................................................................................... 52 Perry Creek .................................................................................................................................. 53 Poplar Creek ................................................................................................................................ 54 Smith Creek ................................................................................................................................. 54 5 FINDINGS ................................................................................................................................... 62 Impervious Surface Area ............................................................................................................ 62 Accuracy ............................................................................................................................... 62 Findings ................................................................................................................................ 62 Impervious Surface Growth ........................................................................................................ 63 Impervious Surface Rates by Land Use of Recent Development ............................................ 64 Statistical Analysis of Water Quality Da ta ................................................................................ 65 Results of Mann -Kendall Test ............................................................................................ 65 Annual Trimmed Mean Calculations ................................................................................. 66 Road Density ............................................................................................................................... 66 Effectiveness of Green Roofs and Pervious Pavements ........................................................... 66 6 DISCUSSION .............................................................................................................................. 73 Discussion of Findings and Methods ......................................................................................... 73 Quantifying Impervious Surface Growth ........................................................................... 73 Comparison of Impervious Surface Rates by Land Use ................................................... 73 Benefits of Encouraging Green Roofs and Pervious Pavements ...................................... 74 Roads and Impervious Surface ........................................................................................... 75 Water Quality and Impervious Surface .............................................................................. 75 Feature Extraction Using Feature Analyst ......................................................................... 76 Opportunities for Future Research ............................................................................................. 77 7 CONCLUSION ........................................................................................................................... 80 PARCEL CLASSIFICATION DOCUMENTATION ..................................................................... 83 APA Activity Codes to NRCS Land Use Category Lookup Table ......................................... 83 Parcel Queries .............................................................................................................................. 86 LIST OF REFERENCES ................................................................................................................... 89 BIOGRAPHICAL SKETCH ............................................................................................................. 93

PAGE 7

7 LIST OF TABLES Table page 2 1 Impervious surface percentages by land use ........................................................................ 25 4 1 Middle Creek and Perry Creek land uses, density, and age of housing stock .................... 60 4 2 Poplar Creek and Smith Creek land uses, density and age of housing stock ..................... 61 5 1 Impervious surface coverage by study area in 2005 ............................................................ 63 5 2 Impervious surface growth estimates .................................................................................... 64 5 3 Impervious surface rates by land use comparison ................................................................ 64 5 4 Mann -Kendall trend test results ............................................................................................ 65 5 5 Road density in study watersheds ......................................................................................... 66 5 6 Trimmed mean of Poplar Creek water quality readings ...................................................... 68 5 7 Trimmed mean of Middle Creek water quality readings ..................................................... 68 5 8 Trimmed mean of Smith Creek water quality readings ....................................................... 69 5 10 Trimmed mean analysis results ............................................................................................. 72

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Map of North Carolina's Urban Crescent ............................................................................. 16 1 2 Conceptual model ................................................................................................................... 18 3 1 Study watersheds .................................................................................................................... 46 3 2 Perry Creek training sets ........................................................................................................ 47 3 3 Perry Creek initial multi -class results ................................................................................... 48 3 4 Perry Creek remove clutter training ...................................................................................... 49 3 5 Perry Creek final model results ............................................................................................. 50 3 6 After inputting Poplar Creek conductance into a time series .............................................. 51 3 7 Example Mann -Kendall test results ...................................................................................... 51 4 1 Middle Creek digital elevation model and context map ...................................................... 56 4 2 Perry Creek digital elevation model and context map ......................................................... 57 4 3 Poplar Creek digital elevation model and context map ....................................................... 58 4 4 Smith Creek digital elevation model and context map ........................................................ 59 5 1 Trimmed mean results graph: Conductance ........................................................................ 70 5 2 Trimmed mean results graph: Suspended Residue ............................................................. 70 5 3 Trimmed mean results graph: Total Kjeldahl Nitrogen (TKN) ......................................... 71 5 4 Trimmed mean results graph: Nitrite (NO2) + Nitrate (NO3) ........................................... 71 5 5 Trimmed mean results graph: Total Phosphorus (TP) ........................................................ 72 6 1 Growth in impervious surface and road de nsity ................................................................... 79

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degr ee of Master of Arts in Urban and Regional Planning IMPERVIOUS SURFACE AND WATER QUALITY IN WAKE COUNTY: A N ANALYSIS OF RECENT DEVELOPMENT, IMPACTS, AND OPPORTUNITIES By Jake Petrosky May 2009 Chair: Ruth Steiner Cochair: Paul Zwick Major: Urba n and Regional Planning Wake County, North Carolina has experienced significant growth over the last twenty years. Between 1996 and 2006 the population of Wake County has grown from 423,380 to 790,007 people. The population is expected to double in the next 30 years. Studies have shown a correlation between urban growth and degraded water quality. This study quantifies impervious surface in selected watersheds within Wake County using Feature Analyst, feature extraction software by Visual Learning Syst ems which uses hierarchical learning techniques to extract features based on color, shape, and adjacency. The study determines increases in impervious surface in the study area s and compares these changes with trends in monthly water quality data Imper vious surface rates by land use were produced for the study areas. These rates are compared to impervious surface rates developed in previous studies to evaluate if current development trends remain compatible with these measures. The effectiveness of alternative design techniques, such as green rooftops and pervious pavemen ts are estimated This study proves that typical suburban development outside of rapidly growing areas in the North Carolina Piedmont can result in a substantial increase in imperv ious surface. Results of the water quality data analysis, using a Mann -Kendall test for trend, indicate statistically

PAGE 10

10 significant trends of decreasing water quality in the study areas. The study estimates that 47% of all new impervious surface constructe d in the study areas between 1999 and 2005 could have been mitigated by incorporating green roofs and pervious pavements in new development.

PAGE 11

11 CHAPTER 1 INTRODUCTION Introduction It is essential for planners, elected officials, and the private sector to understand the impacts local land use decisions and development policies have on water quality. Impervious surface has been linked to degraded water quality, however, there is limited research detailing the amo unt of impervious surface growth, trends in the built environment, and water quality effects in the Piedmont region of North Carolina and the Southeastern United States. This paper details the use of Geographic Information Systems (GIS) and remote sensing technologies to quantify impervious surface coverage (ISC) in five watersheds in Wake County, North Carolina. A ctual, measured, impervious surface coverage rates by land use were compared to ISC rates by land use from a variety of sources that are often used to forecast and estimate impervious surface. The result of this analysis provides insight on current development patterns and trends in the built environment. These results may enable more accurate forecasts of ISC. Growth of impervious surface has been quantified in each watershed and has been compared to historical water quality data in order to provide insight on the relationship between urban growth and water quality in urbanizing areas of the southeastern piedmont. The Built Environment and Urban Land Use Change Humans have impacted the natural environment for thousands of years, but never has human induced change occurred at such a scale and at such remarkable speed (Turner and Meyer 1994, p. 4). Between 1982 and 2003 3 5.2 million acres of the United States have been developed for urban uses (National Resources Conservation Service 200 7 p. 5 ). On average over 1.7 million acres is converted from farm and forest every year. Humans a re a primary driver of these changes in land use and land cover.

PAGE 12

12 Land use is an anthropocentric term used to describe how people use a parcel or area of land (Turner & Meyer 1994, p. 5). Land cover is a term generally used by the natural sciences to des cribe the physical state of the land, including, the quantity and type of surface vegetation, water and earth materials (Turner and Meyer 1994, p. 5). Growth and development on the fringe of suburbanizing America is an example of human induced land us e and land cover change. This change has many direct and secondary impacts on the natural environment. Direct effects occur during the land transformation process. These effects include the immediate loss of tree canopy, wetlands, habitat and individual organisms that occurs when land is converted from field to subdivision or forest to shopping center. Secondary impacts occur subsequent to the initial land transformation and are cumulative in nature. Examples of secondary impacts include habitat, wate r quality, and soil degradation in areas adjacent to the initial land cover change (Turner and Meyer 1994, p 6). Urban land use change results in growth of the built environment. The built environment can be defined as, everything humanly created, mod ified, or constructed, humanly made, arranged, or maintained (Bartuska 2007, p 5). This definition aggregates virtually every square meter of Earth into the built environment, including areas modified by humanity such as once logged forests, fields of g rain, and even the atmosphere. For the purpose of this research the built environment will be limited to products and structures that people build, or the built up areas mentioned by Grubler when referring to the area covered by buildings and infrastruc ture (Turner and Meyer 1994, p 323). If this more narrow definition of the built environment is used then the human built environment began with the construction of mud huts in the Fertile Crescent to house a population that was beginning to farm. The built environment has increased in size dramatically since its early beginnings.

PAGE 13

13 Within the last 10,000 years, man has invented an environment which is almost altogether artificial, with the consequence that he now lives in a world of his own devising (Le nihan and Fletcher 1987, p. 125). Settlements begat villages, villages evolved into towns and cities, and cities now connect to form megalopolises urban landscapes that stretch for hundreds of miles. Only relatively recently has man embarked on such audacious endeavors as the construction of freeways, skyscrapers, and the less enthralling, but cumulatively impressive, network of strip development, warehouses, condos and subdivisions that dominate the American landscape It is the nature of typical u rban change and occupation that leads to detrimental effects it is the things that are builtand what people do when they live among them that disrupt natural systems. Impacts are not limited to cities themselves and range from the minute to the incredib le. Land cover change related to urbanization can lead to reductions and fragmentation of wildlife habitat, changes in wind patterns, temperature, climate and hydrology. It is true that i n the places where mans activities are most densely concentrated his settlements the environmental impact is greatest and the risks of environmental damage are most acute (United Nations 1974, p. 12) However, t he impacts are not restricted to local areas, but can impact natural systems downstream, downwind, and acros s the globe. Impacts to hydrological systems can be particularly substantial and can include increases in runoff, erosion, nonpoint source pollution and reduced low flows (Paul and Meyer 2001, pp. 337347). Certain aspects of mans artificial envir onment the rooftops, sidewalks, asphalt, brick and lawns individually and cumulatively provoke changes to natural processes. These tangible, quantifiable facets of the built environment combined with other side effects of modern life (i.e. auto emissions, sewage effluent, nonpoint source pollution, etc.) result in reduced water and air quality.

PAGE 14

14 The hydrological cycle involves evaporation, rainfall, canopy intercepti on, stemflow, overland flow, infiltration and groundwater recharge In u rban areas this cycle is altered. Instead of tree canopy and groundcover, rainfall strikes concrete, brick and asphalt. These textures and surfaces, products of human ingenuity are inventions derived from organic products. Whether these products are natural is deba table. It cannot be debated that these materials and other types of impervious surface are ubiquitous in cities and suburban areas. It has been proven that these man -made surfaces interfere with the hydrological cycle by altering the processes of interception, stemflow and overland flow and increasing the volume and speed of runoff, while reducing rates of infiltration and groundwater recharge. In addition to quantity related effects, g rowth in impervious surface s has bee n correlated with decreases in water quality (Paul and Meyer 2001, Arnold and Gibbons 1996). Increases in runoff and degraded water quality can cause negative economic and public health impacts to urbanizing communities (SCS 1975, p. 1 1). Though imperv ious surface has been directly linked to water quality degradation, planners do not usually measure existing impervious surface or project the impervious surface potentially resulting from the build out of land use plans (Brabec et al. 2002, p. 500). This is unfortunate, as it is essential to quantify changes in and the distribution of, impervious surface in growing areas, in order to better understand how land use decisions result in changes in the built environment that may negative ly impact the natural e nvironment. As noted by Morisawa and LaFlure (1979), stream hydrology and function are dependent on five variables: climate, geology, soils, land use, and vegetation (Brabec et al. 2002, p. 500). Accordingly it is important for planners to understand t he impacts local land use decisions are having on the two variables that they can influence: land use and vegetation.

PAGE 15

15 Urban growth and development are not necessarily the harbinger of environmental degradation. Indeed, there are instances where ecosy stems have been improved in tandem with development. It is not necessarily simply human occupation that leads to significant problems. It is the responsibility of society and those that plan, design, and construct the built environment to reduce impacts on the natural environment. This will become more important if population projections hold true and the cities of the world absorb an additional three billion people. Study Area Land use change and resulting changes in the built environment have been particularly profound in the southeastern United States. According to the U.S. Census Bureau, the South Atlantic division, composed of Deleware, D.C., Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia, has added over 31 million people between 1960 and July of 2007. This increase is larger than any other Census defined division. North Carolina has doubled in population between 1960 and 2007, swelling from 4.5 million people to over 9 million. A disproportionate amount of this growth has occurred in the piedmont region and particularly the metropolitan areas in what has been called North Carolinas Urban Crescent. The Urban Crescent as shown in Figure 1 1, is a number of urbanized counties and metropolitan areas that begins in the city of Charlotte located in Mecklenburg County, and then continues northward along I 85 to Greensboro in Guilford County then eastward a long I 40 to Durham and Raleigh in Wake County. Railroad networks once connected this area of the state These rails were used to transport tobacco, cotton and textiles. The rails remain, but have been augmented by interstates which now move commuters to hubs of economic activity. This area has also been called the Rich Crescent and North Carolinas Ma in Street.

PAGE 16

16 Figure 1 1 Map of North Carolina's Urban Crescent Since 1980, the ten counties that compose the Urban Crescent, Alamance, Cabarrus, Davidson, Durham, Forsyth, Guilford, Mecklenburg, Orange, Rowan, and Wake County, have grown by nearly 1.6 million residents. Fifty percent of North Carolinas population gr owth between 1980 and July of 2007 has occurred in these counties, which comprise only 10% of the states land area. Wake County, which anchors the eastern end of the Urban Crescent, has experienced rapid growth during the last twenty -five years. Wake Co unty also forms the eastern edge of the Research Triangle, composed of Raleigh, Durham and Chapel Hill. Home to a number of major universities, the Research Triangle Park, and a benign climate, the area has experienced significant in -migration. Spurred by a growing economy and residential development in bedroom communities outside of Raleigh, the county h as added 204,466 residents between 1990 and 2000 (Wake County 2005, p. 1). Since 2000, Wake County has grown from 627,846 residents to its present popul ation of 790,007 (Wake County 2005, p. 1 and North Carolina State Demographics 2007).

PAGE 17

17 Conceptual Model Land use decisions are driven by polices of local governments and regional economic realities. Demand, be it for McMansions or donut shops is perce ived and enacted upon by the private sector. Anthropogenic change, resulting from this demand is marked by land cover change and an increase in the built environment. Development in the rapidly urbanizing area on the urban/rural fringe of Wake County is driven by growth in residential subdivisions and is followed by commercial development along major arterials. Suburban growth in the study area has been enabled by the continued public investment in transportation infrastructure primarily focused on incr easing capacity on arterials and reducing commute times through the construction of major highways. Increases in impervious surface, stream crossings and road densities are products of this growth. Currently 63% of the watersheds in Wake County are consi dered impaired or degraded (Wake County 2003, p. ES 6). It is anticipated that drivers of development have led to increases in impervious surface coverage within the study watersheds and has contributed to degraded water quality (See Figure 1 2) Informa tion from this study will provide a better understanding of the relationship between the built environment and the natural environment. Existing rates of average impervious surface coverage, including those available from studies conducted by the Soil Co nservation Service (SCS), are regularly used to estimate and forecast ISC in watersheds. Due to the change s in regional development patterns and trends, these rates may not provide an accurate representation of average ISC rates by land use in Wake County. Accordingly, SCS ISC rates were compared to those measured in the study area. This methodology is suggested as being critical to ensuring the accuracy in impervious surface studies (Brabec, Schulte, and Richards 2002, p. 503). The results of this study will allow for the forecasting of impervious surface in Wake County to occur based on coverage rates based on

PAGE 18

18 local data. Once change in impervious surface coverage is determined in the study areas it can be estimated how much impervious surface cou ld be avoided by using innovative designs such as p ervious pavements and green roofs. Objectives This research has specific aims: 1 Determine how many square feet of impervious surface is attributable to new development in the study areas. 2 Compare establ ished rates of average impervious surface by land use to measured impervious surface rates by land use in development that has occurred between 1999 and 2005. 3 Determine how innovative design techniques can employed and encouraged to decrease impervious surface in new development (i.e. pervious pavement and green roofs). Figure 1 2 Conceptual m odel

PAGE 19

19 CHAPTER 2 LITERATURE REVIEW Trends It has been estimated that there is 1454 million km3 of freshwater in the global hydrosphere (Roger 1994, p. 234). Of this seemingly infinite amount, only 0.6189e8% is readily available for human use ( Ibid., 236). As of 1970, it was estimated that 3,500 km3 of the 9,000 km3 of available freshwater was utilized and 5,800 km3 was polluted by human action ( Ibid., 234). Between 1970 and 2003 the world population has grown from 4.07 billion to 6.465 billion people (United Nations 2005, p. 4). World population is predicted to pea k at 9.22 billion in 2075 (World Pop to 2030 United Nations 2004, p. 1). Levels of use will rise as the developing world adopts more modern lifestyles. Freshwater will become more important as the realities of population growth and consumption trends bec ome apparent On a local level, in Wake County, pending population growth jeopardizes water supplies as well. Falls Lake was added to EPAs list of impaired waters in 2008 due to excessive nutrients and algal growth (Rawlins 2009). Growth and development, combined with wastewater outfalls from municipalities upstream from the lake have contributed to elevated nitrogen and phosphorus levels. This may force municipalities in Wake County to spend more on water treatment. The proposed Little River Reservoi r in Eastern Wake County may be destined to join Falls Lake on the list of impaired reservoirs. Wake County commissioners recently rejected a proposal to limit residential development adjacent to the proposed reservoir to lower densities than currently al lowed adjacent to Falls Lake (Bracken 2009). Urban Growth in the U.S. Urban growth was spurred by the industrial revolution and further enabled by the development of new modes of transportation. The horse and buggy gave way to railroads and

PAGE 20

20 the street car, both of which were replaced with the automobile. These new transportation technologies allowed cities to grow outward. Housing followed expanded transportation networks, strip development of commercial establishments tagged along. The resulting a u to oriented society results in a development pattern that follows arterials with little regard for habitats or hydrological features. Urbanization Impacts on Hydrology As early as the 1960s the United States Geological Survey (USGS) w as studying the hydrologic effects of development ( Rome 2001, p. 192). Scientists working for the agency found that there was a, dramatic increase in flood potential in the wake of suburban growth (Ibid., 194). Studies showed that as land uses changed from undeveloped to developed, the volume of runoff could quadruple and the speed of runoff also increased ( Ibid. ). In addition to increases in flows, urbanization can cause physical changes in the size of channels in st r eams (Hammer 1972, p. 1540). Thomas Hammer studied w atersheds outside of Philadelphia and concluded that land uses within and adjacent to streams can have significant impacts on the structure of stream channels. The study found that the presence of streets, sidewalks, and contiguous impervious surfaces exc eeding one acre can cause the enlargement of stream channels and, thereby cause, a serious reduction in the aesthetic and recreational value of the stream ( Ibid., 1530). In 1975, the Soil Conservation Service (SCS), part of the U.S. Department of Agricu lture detailed the impacts of urbanization on hydrology in Technical Release No. 55 entitled, Urban Hydrology for Small Watersheds. The SCS acknowledged that, As population density and land values increase, the effects of uncontrolled runoff become an economic burden and a serious threat to the health and well -being of a community and its citizens ( Ibid., 1 1). Urbanization reduces the time of concentration, or the time it takes for water to pass through a watershed from point of origin to outlet ( Ibid., 3 1). Increases in

PAGE 21

21 impervious surface lead to higher peaks and lower baseflow (Paul and Meyer 2001, p. 335). This is due in part to channelization and the construction of conveyance structures and partly due to reduced depressional storage from p redevelopment grading. Impervious Surface and Water Quality The construction of new homes, roads, driveways and parking lots increases impervious surface, which, has been shown to serve as a key indicator for the health of aquatic ecosystems (Arnold and Gibbons 1996). Impervious surface affects the hydrology of a watershed, the geomorphology of stream beds, temperature, fish populations, macro invertebrates, microbes, algae, and macrophytes (Paul and Meyer 2001, p. 337347). Impervious surface acts as a conduit through which pollutants are injected into surface water. In a sense, the impervious surface acts as an enabler to non -point source pollution. Non -point source pollution, as opposed to point source pollution, is described by the Environmental P rotection Agency as pollution composed of, natural and human-made pollutants carried by runoff and deposited in bodies of water (EPA 2008). Stormwater runoff can carry sediment from farms and cleared land, chemicals and fertilizers from residential law ns, and toxins from streets or industrial sites to streams, wetlands, lakes, rivers, estuaries and oceans downhill or downstream. Nutrients, toxins and sediment disrupt aquatic ecosystems and contribute to degraded water quality. Non -point source polluti on is the number one threat to water quality in the United States (Arnold and Gibbons 1996, p. 245). The waters of the Southeastern states are in imperiled due to anthropocentric causes. As of 1998, 45% of the rivers assessed in the Southeast were found to be impaired (Potter et al. 2004, p. 62). It has been found that between 1988 and 1998, 70% of stream degradation in the southeast was caused b y nonpoint source pollutants (Ibid. ). The abundance and diversity of f ish and macro invertebrate populatio ns is harmed as ISC increases (Paul and Meyer 2001). Certain species of fish and invertebrates are more susceptible

PAGE 22

22 to increases in urbanization and impervious surface than others. A study in the Tennessee River drainage system in Western North Carolina and Tennessee found that rare endemic fish species decreased in abundance as development increased and forest cover decreased (Scott 200 6 ). The Maryland Department of the Environment found that water temperature of Piedmont streams increased as total perc entage of impervious surface increases (Galli 1990, p. 28). Higher temperatures were determined to be the cause of declining native trout populations throughout headwater streams in the Maryland Piedmon t (Ibid. ). The study found that impacts from impervi ous surface were noticeable even in lightly developed watersheds with total impervious surface coverage of 12% (Ibid. ). For more developed watersheds, the increase in average temperature can be substantial. In the Maryland study researchers found that wi th a watershed with over 60% impervious surface the average temperatures can be 4 15% higher than a similar undeveloped watershed. It should be noted that the extent of riparian vegetation was also found to have a significant impact on temperatures ( Ibid. 44). Impervious surface coverage also has significant impacts on species other than much sought after trout and other gamefish. Invertebrates, the ubiquitous and essential prey of charismatic aquatic denizens, are also negatively impacted by impermeabl e surfaces. The Maryland study found that stoneflies (Plecoptera) and caddisflies (Trichoptera), two favorite foods of trout, were more abundant in habitats that enjoyed cooler waters habitats more likely to occur in watersheds with limited development. A study of 20 catchments in Maine found that species richness of insect and invertebrate prevalence showed an abrupt decline as percent total impervious surface area (PTIA) rose over 6% (Morse et al. 2002, p. 95). Indeed, there are many studies that have shown that urbanization decreases density and diversity of aquatic insects ( Ibid., 96).

PAGE 23

23 The Southern United States has the, fastest rate of urbanization of a ny region in the country (Scott 2005, p. 302). Urbanizations effect on water quality in North Carolinas piedmont is well documented. A study, funded by the North Carolina Division of Environmental Managements Water Quality Section and the United States Geological Surveys Water Resources Division monitored the effect of land use on water quality in three streams in the piedmont. Water quality testing and biological sampling were conducted in streams where land use was mostly forested, agricultural, or urban. The study found that an urban stream, Marsh Creek, in east Raleigh, had the highest lev els of suspended sediment, and the highest recorded levels of metals (Lenate and Crawford 1994, p. 185). The study also found that Marsh Creek had the lowest percentage of gamefish of all of the streams sampled ( Ibid., 191). In addition, the study found that fish and invertebrate species richness, biomass and presence of intolerant species were lowest in the urban stream. Salamanders, considered a species which indicates water quality, have been shown to decrease in prevalence as impervious surface incre ases from 0 to 26% (Miller and Hess 2005, p. 12). The 2004 study of 43 streams in Wake County found that the amount of impervious surface in a watershed had negative effects on larval E. cirrigera, a species of sa lamander ( Ibid. ). Impervious Surface as an Indicator Studies suggest that levels of degradation correspond to ranges of impervious surface ranging from 0 to 10% ( Sensitive ), 1 1 25 % (Impacted) and above 25% ( Non-supporting) (Schueler 2000). This was recently revised from a previous classificatio n range which classified 0 10% as Protected, 10 30% as Im pacted and greater than 30% as D egraded (Arnold and Gibbons 1996). Arnold and Gibbons (1996) argue that years of scientific inquiry has firmly established impervious surface coverage is a reliable and integrative indicator of the impact of

PAGE 24

24 development on water resources (246). This combined with the fact that impervious surface is measurable, makes it an ideal proxy for gauging water quality in an urban environment. Measurement Techniques Imperv ious surface is a simple attribute that is easy to measure and forecast (Barbec et al. 2002, p. 500). Impervious surface measurement techniques can range from in situ measurement at the site level, to interpretation of aerial photos, to interpolation of impervious percentages from satellite imagery or other land use/land cover inventories. Two types of impervious surface area have been measured. Total impervious area is a measure of all impervious surfaces in an area. Effective impervious surface area is a measure of only those areas that are connected hydraulically to drainage channels in the landscape, or connected to other impervious surface which is connected to the drainage network (Alley and Veenhuis 1983, p. 313). Both measures have been found to be related to water quality degradation. Measurement of effective impervious surface has been found to be labor intensive and inherently subjective, accordingly it is not used in this study ( Ibid., 314). Estimation of total impervious surface is possible based on a number of methods. Researchers have developed estimation techniques based on readily available census information, including population, househol ds and employment (Brabec et al. 2002, p. 503). Average impervious surface coverage rates ha ve been measured by land use category in a number of studies (Arnold and Gibbons 1996, 247). The Soil Conservation Service (SCS) was among the first to develop average impervious surface rates by land use. They found that impervious surface in residentia l uses varied significantly based on lot size and that industrial uses typically had the most impervious surface, only exceeded by commercial a nd shopping center land uses (Ibid. ). This study was updated in 1986. A summary of t hese rates and rates from a dditional studies are shown in Table 2 1. Studies in Northern Virginia and Denver found

PAGE 25

25 residential ISC rates generally lower than the SCS study. The ISC rates for commercial and industrial parcels in both the Northern Virginia study and the Denver study were close to those published in the 1986 SCS report. Table 2 1 Impervious s urface p ercentages by land u se Source Land Use Dwelling Units/Acre NVPDC, Northern Virginia (1980) SCS (1986) SCS (1975) Alley and Veenhuis, Denver (1983) Low Density Residential <0.5 6 0.5 12 1 12 20 20 15 Medium Density Residential 2 18 25 25 1526 3 20 30 30 26 4 25 38 38 26 High Density Residential 5 7 35 39 Townhouse >7 3550 65 65 60 Industrial -60 80 72 75 60 Commercial -9095 85 85 88 Shopping Centers -95 -Sources: Stormwater Managers Resource Center 2009, SCS 1975, SCS 1986, Alley and Veenhuis 1983, p. 314 Rates by land use can be used to estimate impervious surface if an accurate land use inventory is available. However, it has been suggested that land use patterns have changed significantly since the development of standard percentages and the collection of new rates based on local data may provide a greater level of accuracy (Brabec et al 2002, p. 503). It is important to note that studies of impervious surface have not been limited to studies of total impervious surface, but also delve into quantifying the types of impervious surfaces in developments. In 1995, the City of Olympia Washington measured impervi ous surface on 11 sites that were classified as high density residential, multi -family residential or commercial. The

PAGE 26

26 study detailed area associated with different elements of the built and natural environment of each site, including sidewalks, streets, p arking/driveways, roofs, lawns/landscaping, and open space ( Arnold & Gibbons 1996, p. 248). The amount of impervious surface attributable to these elements varied based on the type of land use. For instance, in residential development with densities betw een 3 and 7 dwelling units per acre, rooftops accounted for 15% of total impervious area, or 37.5% of the total site area, whereas for residential development with dwelling units densities between 7 and 30 units per acre, rooftops accounted for 17% of the impervious surface, and 35.4% of the total site area. The study found that road related impervious surface accounted for more than 60% of all impervious area ( Ibid. ). Remote sensing technologies have enabled the use of satellite imagery to measure urban sprawl and changes in the built environment. Physical properties of land cover types result in unique spectral signatures that can be classified using various remote se nsing software packages. The availability of higher resolution (>1m) satellite imagery from IKONOS, QuickBird, and Worldview satellites as well as aerial photography has enabled the extraction of features or objects. Until recently this was a manual proc ess that took a great deal of time (ERDAS 2008, 11). The time and expense of manual digitization typically prevented large scale remote sensing studies using aerial photography. Methods of interpreting aerial photography have become more sophisticated over the years. Feature Analyst, an extension for ArcGIS, created by Visual Learning Systems, offers an interface that allows users to manually digitize training samples. The software uses the training samples and hierarchical learning to develop algorithm s that determine the presence of desired features based on the shape, color, and adjacency of the features. The software allows users to refine the methodology to extract features using an iterative process of training and correcting the extraction proced ure. Feature Analyst has been

PAGE 27

27 used to extract impervious surface inventories using high resolution aerials with accuracies over 90% (Miller and Hess 2005, p. 10). It has been shown that feature extraction with Feature Analyst is 5 to 10 times quicker tha n manual digitization (OBrien, p. 1). The availability of up to date high resolution aerial photography combined with the availability of the Feature Analyst extension can potentially enable researchers to accurately quantify impervious surface over larg e watersheds. Statistical Analysis of Water Quality Urban growth, and particularly impervious surface, has been proven to have detrimental effects on certain aquatic species in Wake County (Miller and Hess 2005). This study attempts to clarify whether i mpervious surface growth has had measurable detrimental effects on water quality as measured in the study areas. Statistical analysis of water quality data presents a number of problems. Significant issues analyzing water quality data include seasonal variations in the data, the presence of outliers, or exceptionally high or low values, and missing values (or values below the reporting limit), and the non-parametric nature of the data, which is primarily due to the lack of negative values, positive skewness and outliers (Helsel and Hirsch 2002, p. 2). The mean and median of data are two common methods for measuring the distribution and central tendency of data. The mean, or average, is calculated by summing all data values and then dividing by the samp le size. The mean is heavily influenced by, the presence of, or to changes in the magnitudes of outliers ( Ibid., 5). The median is defined as, the central value of the distribution when the data are ranked in order of magnitude ( Ibid., 5). The media n is much more resistant to outliers or extreme instances of large or small values ( Ibid., 6). A trimmed mean is calculated by removing high and low values in a dataset and calculated the mean based on the remaining values ( Ibid., 7). A common method for calculating a trim med mean, also

PAGE 28

28 known as the 25% trimmed mean is accomplished by calculating the mean after removing 25% of highest values and 25% of the lowest values ( Ibid. ). A study published in 1997 of the River Tweed, located on the border of Eng land and Scotland, used a 5% trimmed mean to exclude outliers in water quality data collected on a variety of nutrients over a ten year period (Robson and Neal 1997, p. 176, 190). Trend detection in water quality data is possible through the use of the Nonparametric Mann -Kendall test. This test for trend was first by Henry B. Mann in 1945 in a paper entitled Nonparametric Tests Against Trend published in the journal Econometrica The test ranks datasets against one another and uses time as the x variable when determining the significance of Kendalls tau (Helsel and Hirsch 2002, p. 326). Mann proposed that a benefit of the test was to allow for the detection of trends even when, quantities cannot be measured, as long as it is possible to rank them (Mann 1945, p. 247). Mann though t it would be helpful in instances where measures of sensory experiences, such as pleasure or pain could not be directly measured (Ibid. ). The tes t has since been used by researchers attempting to determine trends in time series water quality data (Sheng 2001, p. 254). The Seasonal Mann Kendall test, as suggested by Hirsch et al. (1982), improves results of the Mann-Kendall test by accounting for s easonal changes in water quality data. The Seasonal Mann Kendall test was used in this study; however, the results are not reported due to the fact that they were nearly identical to the results of the Mann Kendall test. Alternative Stormwater Design Ne gative impacts on hydrologic systems and water quality can be mitigated using alternative stormwater design techniques. Stormwater has long plagued human settlements. Traditional methods of stormwater engineering sought to herald stormwater offsite as soon as possible (Randolph 2004, p. 434). This was accomplished by the construction of curb and gutter,

PAGE 29

29 and by channelizing existing drainage pathways. It has since been proven economical and more environmentally friendly to retain stormwater, closer to wh ere it falls on the landscape using techniques that include rain gardens, rain barrels, and larger cisterns fed by rainwater falling on roofs and other surfaces (Forman 2005, p. 329). Stormwater management occurring in the headwaters of drainage networks or headwater management, is preferable to the traditional practice of outlet management where costly engineering projects, in the form of large scale retention basins and sewer systems are the norm. An intermediate form of stormwater management kno wn as enroute management is more likely In this form of stormwater management water is detained temporarily in many dispersed basins, stormwater wetlands, and other structures that allow infiltration to occur enroute ( Ibid., 329). New technologies a nd design alternatives at the site and neighborhood level may be aid in reducing impervious surfaces and their effects on water quality and aquatic ecosystems. Among the site specific technologies that promise to make headwater and enroute management of s tormwater more feasible are green roofs and pervious pavements. Low Impact Development Low impact development (LID) is a general term used to describe a development philosophy that seeks to minimize impacts on the natural environment. Stormwater management is a key element of the LID approach. Traditional stormwater management practices focus on controlling runoff for a design storm usually a 1, 10 or 100 year 24 hour storm (Randolph 2004, p. 436). LID attempts to limit runoff volume, discharge frequency and duration to predevelopment conditions for all storms (Ibid., 437) LID practices include preserving vegetative cover, limiting impervious surfaces, preserving multiple drainage pathways, maintaining groundwater recharge and a general shift from outlet management toward integrated

PAGE 30

30 stormwater design. LID is an example of an alternative design technique for neighborhoods that can potentially mitigate negative hydrologic effects of development. Green Roofs Although g reen roofs have been in e xistence for millennia as elements of the vernacular architecture of Scandinavia and parts of the Middle East they have recently grown in popularity in cities all over the globe (Dunnett 2005, p. 15, 21 ). Frank Lloyd Wright, Walter Gropius, and Le Corbus ier were among the first architects to employ roof gardens in the modern age ( Ibid., 14). Germany serves as the epicenter of green roof technology in the modern world ( Ibid., 22). Early research into the merits of green roofs began in the 1950s and conti nued in the 1960s and 1970s ( Ibid., 19). This research combined with a growing interest in environmental and economic benefits of green roofs has led to many governments in Germany and the rest of the world to embrace and encourage the construction of gre en roofs. Green roof s have many environmental benefits, including increasing opportunity for biodiversity and urban habitat, reducing quantity and quality of stormwater runoff, filtering urban air of pollutants, reducing urban heat island effect, and redu cing noise pollution ( Ibid., 4167). Economic benefits of green roof technologies include increased roof life and reduced energy costs through improved insulation ( Ibid., 6876). In addition to economic and environmental benefits, green roofs can provide passive recreation opportunities and provide aesthetic benefits to urban residents often surrounded by concrete and steel. Researchers in the U.S. have noted benefits of rooftop gardens since the 1970s. In a publication in 1975, the SCS noted the benef its of rooftop gardens included reduction in runoff decreased noise levels, benefits to wildlife and an improvement the aesthetics of roofs (SCS 1975, p. 73). The SCS also pointed out disadvantages of rooftop gardens including in increased loadings on t he structure and increased mainten ance and installation costs (Ibid. ).

PAGE 31

31 Green roofs can have a significant impact on reducing the negative hydrological impacts of impervious surface growth. The amount of stormwater retained by green roofs varies between type of roof, climate, and season. In general, studies show that green roofs can reduce runoff between 40 60% ( Dunnett 2005, p. 58). In addition, the quality of stormwater runoff entering aquatic ecosystems can be improved by implementing green roof tech nology. It has been found that green roofs can reduce nitrate loadings in streams and rivers, which can cause algal blooms (Ibid., 61). It has also been found that green roofs can increase the pH of runoff, which may counteract the impacts of acid rain ( Ibid). If employed on a wide scale, green roofs could potentially counteract the increases in runoff due to increases in impervious surfaces. A study of Brussels predicted that runoff could be decreased by 2.7% if 10% of the buildings in the City were retrofitted with vegetated roofs ( Ibid., 60). Although green roofs address some of the negative impacts of impervious surfaces, they are not a panacea to the negative impacts of urban development. Reducing runoff and water pollution from rooftops is benefi cial; however, reduced infiltration and subsequent groundwater recharge resulting from increased impervious surface will not be addressed by constructing and/or retrofitting existing buildings with green roofs. Pervious Pavements A significant portion of the impervious surface in urban areas is due to paved portions of developments including streets, dri veways, sidewalks, and paved plazas As previously mentioned, a study by the City of Olympia, WA found that road related impervious surface accounted f or more than 60% of all impervious area ( Arnold & Gibbons 1996, p. 248). Pervious pavement has been identified as a measure that is effective at mitigating negative hydrological effects of urbanization (SCS 1975, p. 7 2). Pervious pavements can reduce ru noff by increasing the amount of infiltration and alternatively, delay runoff by retaining stormwater in cisterns and

PAGE 32

32 other catchment devices ( Ibid. ). Pervious pavements are created by processes that have been developed whereby concrete and asphalt are cr eated using aggregate that is larger than traditional methods. This combined with a lack of sand provides for a porous material that water can percolate through at relatively high rates ( National Ready Mix Concrete Association 2009) This pervious pavement lacks the strength of traditional pavements; however it can be substituted for traditional pavements in many circumstances where high traffic and heavy loads are not the norm. Headwater Streets Transportation related components of the built environment are ubiquitous. Everywhere humans go, they are accompanied by paved pathways. This is no doubt due to our reliance on, and the adoption of the automobile as our primary mode of transport. Richard Forman, in Road Ecology claims road surfaces make of 10 to 20% of the suburban land cover and over half of the ISC (2003). This is corroborated by the City of Olympia study in 1994 that found that transportation related ISC made up 6370% of total impervious surface (Schueler 1995, p. 19). Disruptions cause d by road construction and maintenance can lead to degraded aquatic ecosystems. Roads have numerous hydrological effects including the direct effect of filling wetlands, changing flow patterns of surface water, increased erosion and deposition, stream cha nnelization and a loss of riffle and pool structure (Forman 2003, p. 229). Roads are one of the largest contributors to non -point source pollution ( Ibid., 248). It has been suggested that reducing the width of residential feeder roads would have a signif icant impact on ISC ( Ibid., 251). A dvocates for narrow headwater streets, or streets that serve small numbers of residential units and have low traffic volumes claim that roads in many residential areas are overbuilt.

PAGE 33

33 Road networks resemble stream systems in many respects. For example, they are connected in a hierarchial network that is quite similar to stream order. Small access streets generate the traffic that is routed to collector streets that in turn connect with arterial roads, that ultima tely feed freeways. Like streams, the capacity and width of roads tend to increase in a downstream direction. And just like headwater streams, local streets comprise the majority of the road length of the entire road network in a community. Recent studi es indicate that they represent between 50 and 65% of the length of the entire road network (Carrol County 1992) (Schueler 1995, p. 129130). AASHTO requires residential streets to be 26 ft minimum width, and ITE recommends 22 35 ft, depending on dwelling unit density, terrain and on street parking (Ibid., 131). Schueler recommends lower design speeds, more limited right -of -way, no curb and gutter, reduced culdes ac radii, and reduced width (Ibid. ). Recommended width is based on anticipated ADT with the following thresholds; >100 ADT 16 ft, 100500 ADT 20 ft, 5003,000 ADT 26 ft, >6du/acre 32 ft ( Ibid. ). Average daily traffic (ADT) was analyzed and it was found that headwater street design could be employed to different degrees depending on traffic volum e. Streets could be 16 ft for roads that serve less than 100 cars per day and 22 to 28 ft if ADT was between 1,000 and 3,000 ( Ibid., 135). The City of Olympia plans to reduce impervious surface in new development by 20%. Reductions will be made by reducing the size of parking areas, reducing street widths required in residential development, retrofitting culdesacs with vegetated depressions that will hold stormwater, constructing narrow sidewalks where there is limited use, building sidewalks on only one side of the street, building taller a nd clustering development Summary More people are going to be living in urban areas. Human activities that result in substantial land cover change such as timber harvesting, agriculture and urban development can b e contribute significantly to the impairment of surface waters. Given this reality, it is imperative that the amount of water that is rendered useless due to anthropogenic pollution be

PAGE 34

34 minimized by implementing policies that limit negative effects on wate r quality. Studies have shown a correlation between impervious surface, and degraded water quality. Accordingly, it is essential to understand the trends in the built environment related to impervious surface growth as well as trends in water quality on a local level. This will provide justification for planners and elected officials who wish to safeguard the water supplies of growing areas by encouraging responsible developments that incorporate low impact development design features and alternative sto rmwater design techniques

PAGE 35

35 CHAPTER 3 METHODOLOGY Methodology Overview This study measures increases in impervious surface in four watersheds in Wake County, NC. The study quantifies the amount of impervious surface by land use classification a nd compares these rates to identified average ISC rates from previous studies. The study estimates the amount of ISC attributable to roads and potential reductions attainable through the implementation of new residential roadway design standards. The st udy estimates the amount of impervious surface from buildings and the potential reduction attainable by retrofitting building s with green roofs. The study seeks to identify the effects new development has on water quality, impervious surface and non -point source pollution. The study will contribute to the understanding of the correlations between land use changes, increases in impervious surface and reductions i n water quality. The findings of the study enable the accurate forecasting of future impervious surface based on adopted land use plans of local governments in the study areas. The study will enable the drafting of recommendations for local governments c onsisting of implementation steps, including ordinances and site design guidelines that will reduce impervious surface and the impervious surface of new development, limiting the ability to conduct pollutants, thereby reducing impacts on water quality. Time Frame and Study Area Selection Land cover change resulting from urbanization, and its derivatives, has significant impacts on water quality throughout the rapidly growing Sunbelt in the southeastern United States. Urbanization in the piedmont region of North Carolina, South Carolina, and Georgia has been especially prolific. Wake County is located in the piedmont region of North Carolina and has experienced rapid suburban growth over the past 26 years. Between 1990 and 2006, Wake

PAGE 36

36 County has grow n fro m 423,380 people to 832,970 people (U.S. Bureau of Census 2009). It has been documented that urbanization can have a negative impact on water quality, invertebrate populations and fish communities ( Lenat and Crawford 1994 Booth and Jackson 1997). Accord ingly, land use change and its effects on water quality in Wake County can be seen as representative of a larger trend occurring on the edges of cities throughout the Piedmont. The study will measure changes in the built environment between 1999 and 2005. This time frame was selected due to the availability of aerials for Wake County. There are 81 watersheds in Wake County identified in the Wake County Watershed Management Plan (Wake County, 2003, p. ES 5). Of these, 30 are classified as healthy, 38 are classified as impacted and 13 are classified as degraded ( Ibid., ES 6). These classifications are based on in stream data and habitat data collected by the North Carolina Division of Water Quality (DWQ) and other agencies. It has been determined that sediment from construction sites and, increased stream bank erosion caused by increased storm water runoff volumes from additional impervious surfaces in developing areas are the two primary causes of degraded water quality (Ibid. ). Study Ar ea Selection Process Within Wake County watersheds or sub -basins were selected based on an attempt to minimize the potential interference of intervening variables by selecting small basins. Preference was given to those basins that beg i n and end in Wake C ounty, due to the variability of aerial photography. Preference was also given to those areas with available water quality data. A shapefile of the river basins and sub-basins (14 digit hydrologic units) for North Carolina was downloaded from the North Carolina Center for Geographic Information & Analysis (NC CGIA) download page on July 12, 2008. This inventory was created through a cooperative effort involving the United States Department of Agriculture Natural Resources

PAGE 37

37 Conservation Service (USDA NRCS ), NC CGIA, and the North Carolina Department of Environment, Health and Natural Resources Division of Water Quality (NC DENR DWQ). ArcGIS 9.1 was used to select the 45 hydrologic units that intersected the boundary of Wake County and export them as a new layer. The watersheds were cross checked with aerials and water quality data inventories of the United States Geological Survey and the Lower Neuse River Basin Association. The size of the catchment, availability of water quality data, and land use mix were evaluated. Four watersheds were selected. Each of the study areas are medium size streams located in urbanizing areas outside the city core. Perry Creek, Poplar Creek, Upper Middle Creek, and S m it h Creek watersheds were chosen. Th e study areas a re shown in F igure 3 1 Delineating Watersheds The watersheds that were selected included parts of the watershed downstream from water quality monitoring stations. In order to ensure that the water quality data were influenced only by impervious surface upstream from the monitoring stations it was nec essary to revise study areas to include only the area upstream from the monitoring stations. This was accomplished using Terrain Analysis Using Digital Elevation Models (TauDEM) and ArcHydro for ArcGIS 9 Version 1.2 Beta. The methodology is explained bel ow. Down loaded 10 meter digital elevation models (DEM) from the USGS's seamless data distribution site, http://seamless.usgs.gov/ Used the Extracted by M ask tool in ArcGIS to extract pixels within the study are a watersheds defined in the EPA watershed layer Filled Pits using Tau DEM to get rid of depressions in DEMs Calculated Flow Direction Grid with ArcHydro Calculated Flow Accumulation Grid with Arc Hydro

PAGE 38

38 Defined Streams based on a stream initiation threshold of 3500 cells which is the equivale nt of 0.202663 square kilometers (based on extent of 24k blueline streams) in Arc Hydro Calculated Stream Segmentation Grid in Arc Hydro Delineated Catchment Grid in Arc Hydro to identify areas draining to each stream link Catchment Polygon Processing in Arc Hydro to create catchment polygons out of the catchment grid Drainage Line Processing in Arc Hydro to create streamlines Adjoint Catchment Processing in Arc Hydro to create simplified catchments Used the Point Deli neation command to delineate the watershed above the location of the water sampling station Measurement of Variables Impervious Surface in 2005 Measurement of independent variables was conducted using Geographic Information Systems (GIS) technology and aer ial photography, tax parcel records and road inventories maintained by Wake County. Feature Analyst, an extension for ArcGIS developed by Visual Learning Systems, was used to extract impervious surface coverage from high resolution aerial photography for 2005. The aerials had a resolution of 6 inches and were flown in April of 2005 and downloaded from Wake Countys website. Each aerial tile was in Mr. Sid format and represented roughly 144 acres. For the study watersheds 357 tiles were processed. Tax parcel records were acquired from Wake County for April of 2005, in order to correspond with the timeframe in which the aerials were taken. The resulting impervious surface polygons were corrected using post processing tools available in the Feature Anal yst software and revised based on manual inspection. The basic process used for impervious surface extraction is outlined below.

PAGE 39

39 1 Create Pervious Surface Training Set/Create Impervious Surface Training Set 2 Prepare Multi -class Input Layer (Vector Tools /Prepare Multi Class Input Layer) a Add pervious_train and impervious_train 3 Set Up Learning a Optimized Results from Learning i Used Bulls Eye 2 9 for the input representation ii. Resample of 1 / Histogram Stretch / Aggregate at 100 pixels 4 Run Extraction Pass a Use Learning to Resolve Ambiguity/Classify Background 5 Split Out Pervious Class (Vector Tools/Split Out Classes) 6 Remove False Positive s From Pervious Features a Hierarchical Learning/Begin Removing Clutter b Select/Digitize correct and incorrectly classified feat ures 7 Set Up Learning a Optimized Results from Learning i Used Manhattan 5 input representation ii. Resample of 4 / Histogram Stretch / Aggregate at 50 pixels 8 Run Extraction Pass 9 Split O u t I mpervious Class (Vector Tools/Split Out Classes) a Select impervious class 10. Remove False Positives From Impervious Features a Hierarchical Learning/Begin Removing Clutter b Select/Digitize correct and incorrectly classified features 11. Set Up Learning a Optimized Results from Learning i Used Bulls Eye 2 9 input representation ii. Resample of 1 /Histogram Stretch/Aggregate at 100 pixels 12. Combine Feature Layers (Vector Tools/Combine Features) a Use Learning to Resolve Ambiguity/ Classify Background b Optimized Results from Learning c Used Manhattan 5 input representation d Resample of 4/Histogram Stretch/A ggregate at 50 pixels

PAGE 40

40 A unique extraction model was built for each study watershed. Originally one model was built and applied to all watersheds, but this resulted in inaccurate results. An alternate method was used in which an extraction model was buil t for each study area. Training images and training sets, as shown in Figure 3 2 were selected in order to account for the variety of land cover found in each study area. Once training sets were produced, the sets were combined into a Multi -class input layer. Then the Set Up Learning options were adjusted. After a number of iterations of experimenting with different learning options it was found that an initial extraction pass based on the Bulls Eye 2 9 input representation resulted in the most accur ate initial extraction. Results from an initial Multi -class extraction are shown in Figure 3 3. After initial Multi -class Results were developed, the impervious and pervious classes were split out and clutter was removed using the hierarchical learning functions within Feature Analyst. Incorrectly classified groups of pixels were selected as well as correctly classified pixels. See Figure 3 4 for an example. After clutter was removed from pervious and impervious classes, an extraction pass was set up. For the impervious class the Bulls Eye 2 9 input representation was used with a resample of 1, approach setting 2, and an aggregation level of 100 pixels. For the pervious class, the Manhattan 5 input representation was used with a resample of 4, approach setting of 1, and an aggregation level of 50 pixels. After the clutter removal results were processed, the resulting layers were combined and learning was used to solve ambiguity between the classes and classify the background. See Figure 3 5 for an example of the results from the final extraction. Once the feature extraction models were devel oped for each watershed, the remaining tiles in the watersheds were processed using the batch classification option. Each tile required 15 20

PAGE 41

41 minutes to process. This process was lengthened due to a Memory Exception error message occasionally received in the middle of processing, which required a reboot and resetting of the batch process. After the tiles were processed the results were merged and the impervious surface was selected and manually edited to remove clutter and misclassified groups of pixels. Corrections included removing false positives created by gray tree trunks and masses of branches in leaf -off deciduous forests, and in bare fields grayish in color. In some instances weathered roofs and red roofs were added to the impervious results. Approximately 8 hours per study area was spent on manually correcting the resulting classification. Impervious Surface 1999 Impervious surface in 1999 was back casted by netting out ISC on parcels that had a built year, of 1999, 2000, 2001, 2002, 2003, 2004, and 2005. All parcel s were selected where "YEAR_BUILT" in ( 2005, 2004, 2003, 2002, 2001, 2000, 1999) and the ISC on these parcels was summed and subtracted from the total ISC measured. Impervious surface associated with roadways was estimated by overlaying the derived impervious surface polygons and road inventor ies developed by Wake County. An archival roadway inventory from Wake County Geographic Information Services was buffered by 12 feet and unioned with the extracted impervious surface polygons. A buffer width of 24 feet (12 feet on each side of road) was chosen based on the average width at a random sampling of 20 points on the road network in the Middle Creek Study Area. This post processing step ensures that roadways ensconced by tree cover are included in the impervious surface totals. Accuracy was d etermined by generating a statistically significant number of random points and visually inspecting the re sults of the feature extraction, after post processing, to determine if the methodology resulted in successfully identifying areas as pervious or impe rvious. Congalton

PAGE 42

42 (1991) recommends at least 50 samples per land use category (43). For this study 152 random points were generated for each class and verified for correct classification. Analysis Methodology Impervious surface coverage rates by land u se were produced for the study areas. This was accomplished by querying the tax parcel information to determine the land use of individual parcels based on American Planning Association Land Based Classification System ( APA LBCS ) codes maintained by the W ake County De partment of Revenue. Impervious surface rates by land use were quantified by merging the impervious coverage output and the parcel information then summarizing by land use type. These were compared to SCS impervious surface coverage rates often used to forecast impervious surface. E stimates of impervious surface coverage attributable to certain facets of the built environment were produced, including square footage of nonpermeable surface from roads, driveways, and rooftops. In the future it may be possible to extract these three features separately by experimenting with hierarchical learning capabilities within Feature Analyst by developing specialized training sets after the initial impervious surface e xtraction has been successfully completed however, this was not done during this research From these totals, estimates will be produced that indicate what percentage of impervious surface can be reduced by incorporating green rooftops and pervious pavem ents in all new developments. The GIS steps taken to quantify ISC by land use are included in the list below. 1 Union ISC extraction with streets 2 Dissolve 3 Calculate ISC by watershed 4 Clip Parcel Classified by SCS LU by Study Areas 5 Recalculate Parcel Ac res

PAGE 43

43 6 Union ISC with Parcel Classified by SCS LU 7 Select ISC and calc Area Acres field 8 Summarize based on PIN in Parcels (summing ISC acres, first AreaInSA, first NR CS) 9 Join with Parcel layer 10. Calculate Percent ISC by Parcel 11. Select where "Per_ISC" >0 AND "BLDG_VAL" >0 and "YEAR_BUILT" in ( 2005, 2004, 2003, 2002, 2001, 2000, 1999) 12. Summarize based on S CS LU 13. Ca lculate Mean ISC by S CS LU Quantifying Land Use by Watershed A Wake County tax parcel database from April 2005 was re ceived from Wake County. Each parcel was classified using a lookup table (See Appendix A) which translated the APA Activity codes to simplified land use types that were used in previous impervious surface studies. The queries used to classify land uses are a lso included in Appendix A Parcels with S CS categories were summed by Study Area. Only those parcels that were completely in the study area were selected. The following queries were used to select and calculated dwelling units built prior to 2000 and between 2000 and 2005: "NRCS_USDA_" in ( 'Multi -family Residential', 'Single Family Residential') AND "YEAR_BUILT"<2000 "NRCS_USDA_" in ( 'Multi -family Residential', 'Single Family Residential') AND "YEAR_BUILT" >= 2000 Quantifying Road Density Road density has been determined to be correlated with impervious surface and water quality degradation (Forman 2005) The number and proximity of roads corresponds with the

PAGE 44

44 intensity of land uses. Road density in 1999 and 2005 was measured in this study. Wa ke County maintains a GIS -based road inventory for emergency services. An archived copy of the Wake County streets shapefile was re ceived from Wake County Geographic Information Services for 1999 and April of 2005. These two files were clipped based on t he study watershed boundaries and an overall road density was calculated by dividing total roadway miles by total square miles of the watersheds. Water Quality Analysis Once increase in impervious surface coverage was quantified, these changes were compa red to changes in water quality in the study areas. It was hypothesized that increases in impervious surface would correlate closely with degraded water quality. Water quality data was collected for Smith Creek, Middle Cree k, and Poplar Creek by the Low er Neuse Basin Association. The data was part of the Ambient Water Quality Monitoring Data records maintained by the North Carolina Division of Water Quality. The data included measurements of a variety of parameters including temperature, PH, dissolved oxygen (DO), conductance, turbidity, suspended residue, Total Kjeldahl Nitrogen (TKN), nitrate (NO3) + nitrite (NO2), and total phosphorus. The data was collected between 1995 and 2006 (1996 and 2006 for the Middle Creek watershed). Prior to clean up th e datasets contained 193 records for Smith Creek 198 records for Poplar Creek and 182 for Middle Creek. There were multiple sampling dates for some months which were averaged to create a single value representative of the conditions for that month. In s ome months where there was not a reading. These missing values were interpolated based on a linear curve. The data for Perry Creek was compiled by the Stormwater Management Division of the City of Raleigh Public Works Department. This dataset consist ed of 30 samples that were taken between 1998 and 2008. The low number of samples prevented statistical analysis.

PAGE 45

45 Water quality indicators that will be measured will include Conductance, Suspended Residue TKN, Nitrate + Nitrite, and Total Phosphorus (TP) A 25% trimmed mean was calculated for each parameter by year except Perry Creek where limited data was available In addition, a Mann -Kendall Trend Test was conducted using the monthly water quality values for all study creek s except Perry Creek, whe re limited data was available. A Mann -Kendall Test for trend was used due to the non -parametric nature of water quality data and its applicability for use with time series water quality data. Results of the Mann -Kendall Test include a value for tau and a 2 -sided p-value. Positive tau values indicate a positive trend; negative values indicate a negative trend. A p -value below 0.05 indicates a statistically significant trend. Mann -Kendall in R R is free software that is essentially a, language and environment for statistical computing and graphics ( The R Foundation for Statistical Computing 2009). R allows for a variety of data processing, statistical analysis and graphics production. Time series analysis is possible within the R environment. Da ta tables can be inputted as a time series layer (See Figure 3 6). The R environment can be extended by downloading and installing a variety of packages. In order to conduct the Mann Kendall trend test it is necessary to download the Kendall package. Fo r this study the Kendall Package, version 2.0 was downloaded and installed. The package was authored by A.I. McLeod For the three study areas with ample reading, the datasets were pre -processed by averaging multiple readings per mon th and filling in mi ssing months based on a linear curve Once the data was pre processed, t he following steps were used to run the Mann Kendall trend test in R. 1 Create comma delimited file for each variable 2 Import variable into R using the command a poplarcond=ts (scan("C:/R/Poplar_cond.csv"), start=1995, frequency=12)

PAGE 46

46 3 Run Mann Kendall Test (See Figure 3 7 for an example of output results) a MannKendall(poplarcond) Figure 3 1 Study w atersheds

PAGE 47

47 Figure 3 2 Perry Creek t raining s ets

PAGE 48

48 Figure 3 3 Perry Creek i nitial m ulti -class r esults

PAGE 49

49 Figure 3 4 Perry Creek r emove c lutter t raining

PAGE 50

50 Figure 3 5 Perry Creek final m odel r esults

PAGE 51

51 Figure 3 6 After inputting Poplar Creek c onductance into a t ime s eries Figure 3 7 Example Mann -Kendall test r esults

PAGE 52

52 CHAPTER 4 DESCRIPTION OF WATER SHEDS Overview The majority of land area in study watersheds is located in Wake County. Smith and Perry creeks are located on the northeastern side of the City of Raleigh, and enter the Neuse River just s outh of the Falls Lake Reservoir. Poplar Creek is on the east side of Raleigh and flows from Knigh t dale south to the Neuse River. Middle Creek is located southeast of Raleigh on the outskirts of Cary, Apex and Holly Springs. Digital elevation and context maps are included as Figure 4 1 through 4 4. Smith Creek, Perry Creek, and Poplar Creek are in the Neuse River Subbasin 02 as defined the North Carolina Department of Environment and Natural Resources Division of Water Q uality Environmental Services Section in the Basinwide Assessment Report for the Neuse River Basin (NCDWQ 2006, p. 29). It has been noted that nonpoint source pollution and agricultural runoff impact water quality in Subbasin 02. There are six large perm itted wastewater discharge sites in the Subbasin, but none discharge into the study creeks. In the Smith Creek Study Area, three minor NPDES dischargers lie within five miles upstream: Whipporwill Valley WWTP, Wake Forest WTP, and Jones Dairy Farm ( Ibid., 32). Middle Creek is located in the Neuse River Subbasin 03 as defined in the Report ( Ibid., 43). Middle Creek The Middle Creek watershed drains 56 square miles in southern Wake County. The Middle Creek Study Area covers only 8.4 square miles of the total watershed and is comprised of land that drains into Middle Creek before it reaches the ambient water quality monitoring station at Holly Springs Road. Middle Creek has its headwaters in downtown Apex and drains rapidly growing residential areas eas t of Apex. The riparian corridor has been fragmented by small

PAGE 53

53 subdivisions with low residential densities. There is some newly constructed industrial development in the northern part of the watershed near US 1. More than a third of the 2,900 homes in th e watershed were built between 2000 and 2005. A detailed summary of land uses, density, and age of housing stock in this watershed is included in Table 4 1 Perry Creek Perry Creek begins in the northern part of Raleigh and drains in a northeasterly dir ection to the Neuse River. It has been described as a, small, sandy, shallow creek approximately 4 meter s in width ( NC DWQ 2006, p. 33). The Perry Creek Study Area is 11 square miles in area. The Study Area is divided into four quadrants by two major t ransportation corridors. US 1, also known as Capital Boulevard travels through the center of the watershed in a north/south alignment and I 540, also known as the Outer Beltline traverses the Study Area in an east to west direction. The section of I 540 west of US 1 was constructed during the study timeframe. The southeast quadrant of the study area is dominated by Triangle Town Center, a regional mall that was constructed between 2000 and 2005. The southwest quadrant is composed of commercial and ind ustrial outparcels near US 1, and residential uses to the west. A number of the residences in this area are part of established low density residential neighborhoods built prior to 1990. The northwest quadrant is composed of medium density residential u ses to the west, between 3 and 5 units per acre, and industrial uses near US 1. The northeast quadrant is composed of residential uses of varying densities and large vacant tracts near the I 540 corridor. Much of the established and new residential devel opment in this watershed is due to high density apartments units. A detailed summary of land uses, density, and age of housing stock in this watershed is included in Table 4 1

PAGE 54

54 Poplar Creek Poplar Creek has its headwaters in the Town of Knightdale, whic h is located just east of Raleigh The creek drains south to the Neuse River. The Poplar Creek Study Area covers nearly all of the 9 square miles of the watershed. Poplar Creek is the most rural of all the study areas. As of 2001 there was a significan t amount of cultivated crops and pasture land. There has been some suburban development in the watershed; the majority of the residential development has been low and medium density development built prior to 2000. A detailed summary of land uses, densit y, and age of housing stock in this watershed is included in Table 4 2 Smith Creek Smith Creek drains an area of 29.2 square miles. T he Smith Creek Study Area comprises only 21.5 square miles due to the fact that 2.1 square miles are located in Frankli n County and necessary aerial photography was not available for the study timeframe and a portion of the watershed drains land that is not hydrologically connected to the water quality sampling point on Burlington Mills Road. The portion of land in the wa tershed which drains Franklin County is rural in character; because of this it was assumed that development in this area does not substantially affect the overall health of the stream. Smith Creek watershed has its headwaters in the Town of Wake Forest and Sanford Creek and smaller streams that drain the north side of the Town of Rolesville. T he watershed is bordered by Main Street in Wake Forest to the northwest and Main Street in Rolesville to the southeast. The creek drains south to its confluence wit h the Neuse River downstream from the Falls Lake Dam. The watershed has experienced significant growth as development has extended northward from Raleigh along US 1 and subdivisions have been built on the outskirts of Wake Forest and Rolesville. Out of t he 4,531 dwelling units, 31.1%, or 1,409 dwelling units were built between 2000 and 2005. The majority of these have

PAGE 55

55 been medium density between 2 to 4 units per acre. A detailed summary of land uses, density, and age of housing stock in this watershed is included in Table 4 2 In the 2006 Neuse Basin Assessment Report, produced by the North Carolina Department of Environment and Natural Resources Division of Water Quality Smith Creek was assigned a Good -Fair classification based on benthic macroinvertebrate data in 2005 ( Ibid., 30). This was one of only two creeks in Subbasin 02 to show improvements in the benthic macroinverterate community. In contrast sampling shows that t he fish community in this creek has declined between 2000 and 2005. In 2000 it received an Excellent rating, in 2005 it received a Fair rating.

PAGE 56

56 Figure 4 1 Middle Creek d igital e levation m odel and c ontext m ap

PAGE 57

57 Figure 4 2 Perry Creek d igital e levation m odel and c ontext m ap

PAGE 58

58 Figure 4 3 Poplar Creek d igital e levation m odel and c ontext m ap

PAGE 59

59 Figure 4 4 Smith Creek d igital e levation m odel and c ontext m ap

PAGE 60

60 Table 4 1 Middle Creek and Perry Creek l and u ses, d ensity, and a ge of h ousing s tock Land Use and Density Middle Creek Perry Creek Parcels Acres Dwelling Units Parcels Acres Dwelling Units Land Use Dwelling Units/Acre Built Before 2000 2000 2005 Total Units Built Before 2000 2000 2005 Total Units Unclassified 861 2,504 636 1642 0 0 Low Density Residential <0.5 44 444 43 4 47 27 511 34 1 35 0.5 300 119 132 168 300 99 128 100 0 100 1 3 00 210 185 150 335 924 606 917 9 926 Medium Density Residential 2 334 132 265 77 342 1236 536 1272 4 1,276 3 3 41 98 212 80 292 941 274 941 21 962 4 288 66 306 134 440 1032 234 1024 22 1,046 High Densit y Residential 5 7 427 75 43 4 47 1093 247 1380 52 1,432 >7 432 107 691 383 1,074 1337 239 1976 927 2,903 Commercial 33 51 79 334 0 0 Industrial 58 257 91 327 0 0 Totals 3,418 4,064 1,877 1,000 2,877 7,495 5,078 7,644 1,036 8,680

PAGE 61

61 Table 4 2 Poplar Creek and Smith Creek l and u ses, d ensity and a ge of h ousing s tock Land Use and Density Poplar Creek Smith Creek Parcels Acres Dwelling Units Parcels Acres Dwelling Units Land Use Dwelling Units/Acre Built Before 2000 2000 2005 Total Units Built Before 2000 2000 2005 Total Units Unclassified 398 1592 0 0 1,740 7,331 Low Density Residential <0.5 93 1558 88 13 101 269 2,011 238 38 276 0.5 223 287 181 43 224 283 401 246 45 291 1 526 3 87 484 46 530 664 484 541 128 669 Medium Density Residential 2 296 129 297 1 298 974 389 709 289 998 3 103 3 0 100 7 107 616 214 341 400 741 4 252 56 242 10 252 375 87 167 213 380 High Den sity Residential 5 7 255 44 229 32 261 238 44 148 104 252 >7 167 27 82 154 236 254 59 732 192 924 Commercial 9 33 0 0 77 199 Industrial 4 7 0 0 37 147 Totals 2,326 4,151 1,703 306 2,009 5,527 11,368 3,122 1,409 4,531

PAGE 62

62 CHAPTER 5 FINDINGS Impervious Surface Area The semi automated impervious surface extraction process utilized Feature Analyst to classify high resolution images, an estimation of roadway impervious surface using a county maintained GIS roadway inventory, and manual cleanup based on visual inspection Accuracy An accuracy assessment was conducted by generating 252 random points using the Generate Random Points Tool within Hawths Analysis Tools for ArcGIS. Half of the points (126) were generated within the pervious extraction layer and half were ge nerated within the impervious extraction layer. Each point was inspected to determine if the resulting classification was correct. It was found that the overall accuracy of this semi automated classification method was 94%. The impervious surface classi fication was found to be correct 90.32% of the time, and the pervious surface classification were found to be correct 97.62% of the time. This resulted in an overall accuracy of 94%. Findings The p ercentage of ISC varie s by study area as shown in Table 5 1 The Poplar Creek watershed had the lowest amount of impervious surface, with 7.56% of the watershed. Perry Creek had the highest level of ISC with 26.51%. The p ercentage of ISC attributable to roads ranged from 18.24% for Perry Creek to 32.84% for Poplar Creek. If the receiving streams in the study areas were classified according to thresholds indicating stream health, Mid dle Creek would be considered I mpacted with over 10% impervious surface, Perry Creek would be considered Non supporting since it exceeds the 25% threshold which indicates a stream degraded to the point that it does not support a variety of

PAGE 63

63 aquatic life Both Poplar Creek and Smith Creek would be considered S ensitive since total ISC in these watersheds is below 10%. Ta ble 5 1 Impervious s urface c overage by s tudy a rea in 2005 Watershed Area of Watershed (Acres) ISC (Acres) % ISC Street ISC (Acres) % Street ISC Middle Creek 5,404.62 740.85 13.71% 166.60 22.49% Perry Creek 7,064.96 1,873.10 26.51% 341.70 18.24% Poplar Creek 5,546.55 419.18 7.56% 137.64 32.84% Smith Creek 14,511.34 1,370.88 9.45% 314.70 22.96% Impervious Surface Growth Impervious surface growth was most pronounced in the Perry Creek Study Area, where substantial commercial growth in and around Triangle Town Center and the completion of a portion of I 540, the outer beltline, increased the amount of impervious surface dramatically. Over five percent of the total land a rea in the Perry Creek watershed was converted to impervious surface between 1999 and 2005. This is a low estimate, as the estimate was based on an assum ption that new roads were 24 ft wide. The outer beltline exceeds this width, but more research needs to be done in order to develop a more accurate methodology to estimate ISC growth attributable to roads. In the Middle Creek Study area ISC growth between 1999 and 2005 resulted in the watershed crossing the threshold from Sensitive to Impacted. Al though, in both the Poplar Creek and Smith Creek study areas impervious surface grew by smaller amounts, the watersheds are approaching 10% impervious surface At the current rate of growth, the Smith Creek watershed will be classified as Impacted and e xceed 10% ISC by 2011. See Table 5 2 for detailed estimates of impervious surface growth between 1999 and 2005 by watershed.

PAGE 64

64 Table 5 2 Impervious s urface g rowth e stimates Watershed Information Estimated ISC Growth 1999 2005 Name Area in Acres On Par cels From Roads Total % of Land A rea Middle Creek 5,404.62 157.76 47.82 205.58 3.80% Perry Creek 7,064.96 325.06 77.60 402.66 5.70% Poplar Creek 5,546.55 30.05 8.78 38.83 0.70% Smith Creek 14,511.34 261.96 87.84 349.80 2.41% Total 32,527.47 774.83 222.04 996.87 3.06% Impervious Surface Rates by Land Use of Recent Development Impervious surface resulting fr om construction on parcels that had a built year of 1999, 2000, 2001, 2002, 2003, 2004, and 2005 was summarized by land use. The average rates of impervious surface attributable to new development in different land use categories in each watershed are shown in Table 5 3 Although the average rates of impervious surface on residential parcels varied between watersheds, the average rates of imperv ious surface in the study as a whole are close to the average rates of ISC by residential densities produced by the SCS The average rates of ISC for commercial and industrial properties in the study (59% and 56%) were much lower than the SCS rates for com mercial and industrial land uses (85% and 72%). Table 5 3 Impervious s urface rates b y l and u se c omparison Land Use Perry Creek Poplar Creek Middle Creek Smi t h Creek N Average SCS <0.5 DU/Acre 24.16% 8.10% 10.21% 8.57% 99 9.34% 0.5 DU/Acre 10.43% 16.65% 14.06% 281 15.12% 12.00% 1 DU/Acre 24.66% 15.51% 22.93% 17.12% 401 19.71% 20.00% 2 DU/Acre 28.38% 12.76% 23.22% 24.50% 474 24.30% 25.00% 3 DU/Acre 27.74% 21.47% 27.93% 31.11% 577 30.13% 30.00% 4 DU/Acre 33.87% 27.50% 31.67% 34.88% 521 33.15% 38.00% 5 7 DU/Acre 39.98% 34.12% 19.19% 40.69% 284 39.00% >7 DU/Acre 58.74% 39.37% 47.67% 72.96% 1,130 55.30% 65.00% Commercial 65.74% 42.54% 52.23% 56.66% 63 59.56% 85.00% Industrial 62.49% 0.00% 57.36% 49.10% 60 56.51% 72.00%

PAGE 65

65 Statistical Analysis of Water Quality Data Results of Mann -Kendall Test The Mann -Kendall trend test showed that there was a statistically significant positive trend in total phosphorus readings in Middle Creek, conductance and total nitrogen readings in Poplar Creek, and conductance and total phosphorus readings in Smith Creek. Interestingly the test showed a statistically significant negative trend in total nitrogen readings in Smith Creek. This could be a result of reduced amounts of fertilizer application due to the abandonment of farming. The majority of the remain ing variables showed positives trends that were not statistically significant ( S ee Table 5 4 ). Overall there were five statistically significant positive trends, indicating increasing pollutant loads, and one statistically significant negative trend, indi cating a decreasing pollutant load. The strongest trends were in conductance values in Poplar Creek and Smith Creek, which had tau values of 0.351 and 0.56, respectively. Table 5 4 Mann -Kendall trend t est r esults Study Area Variable Tau P Value Middle Creek Conductance 0.057 0.33 Sus pended Res idue 0.086 0.15 TKN 0.0753 0.21 NO2/NO3 0.008 0.89 Total Phosphorus 0.149 1.16E 02 Poplar Creek Conductance 0.361 1.77E 10 Suspended Residue 0.0396 0.49 TKN 0.03 0.60 NO2/NO3 0.275 1.05E 06 Total Phosphorus 0.057 0.32 Smith Creek Conductance 0.56 2.22E 16 Sus pended Res idue 0.0411 0.47 TKN 0.053 0.36 NO2/NO3 0.25 9.81E 06 Total Phosphorus 0.125 0.03

PAGE 66

66 Annual Trimmed Mean Calculations A 25% trimmed mean was calculated for each water quality parameter per year. Tables 5 6 through 5 9 and Figures 5 1 through 5 5 show the trimmed means from the study areas for conductance, suspended residue, turbidity, TKN, Nitrite + Nitrate, and Total Phosphorus by year of sample. Annual trimmed means for 2006 were compared to annual trimmed means for the base year. In eight cases the annual trimmed mean increased, indicating decreasing water quality (See Table 5 10) In seven cases the annual trimmed mean decreased, indicating improving water quality (See Table 5 10) Road Density Road density in the study watershed has increased substantially between 1999 and 2005. Smith Creek, Perry Creek, and Middle Creek have experienced increases in excess of 20% (S ee Table 5 5 ). Increases in miles of roadway results in more stream crossings, culverts and, ultimately, disrupted hydrologic systems. Table 5 5 Road d ensity in s tudy w atersheds Study Area Area (square miles) Miles of Roads 1999 Miles of Roads 2005 Road Density 1999 Road Density 2005 Percent Change Smith Creek 22.67 88.53 116.96 3.91 5.16 32.1% Poplar Creek 8.67 51.47 53.48 5.94 6.17 3.9% Perry Creek 11.04 99.89 127.55 9.05 11.55 27.7% Middle Creek 8.45 53.74 67.08 6.36 7.94 24.8% Effectiveness of Green Roofs and Pervious Pavements Using rates of impervious surface for roofs and parking areas/driveways by land use developed by the City of Olympia, WA, estimates can be developed for the amount of ISC that would be eliminated i f new d evelopment were to incorporate green roofs and pervious pavements into site plans. If it is estimated that parking lots and driveways make up 15% of all

PAGE 67

67 ISC in residential development under 7 units per acre, 31.3% of ISC in residential development greater than 7 units per acre, and 61.6% of ISC in commercial and industrial development, then it can be estimated that the parking lots and driveways constructed as components of all new development in the study areas between 1999 and 2005 add up to 178.43 acres of impervious surface. If it is assumed that roofs make up 37.5% of ISC in residential development under 7 units per acre, 35.4% of ISC in residential development greater than 7 units per acre, and 30.2% of ISC in commercial and industrial development, t hen it can be estimated that rooftops make up 130.20 acres of impervious surface built between 1999 and 2005. Together it can be estimated that rooftops, driveways and parking areas accounted for 308.63 acres (47.4%) of new impervious surface built betwee n 1999 and 2005, not including ISC attributable to streets If every house, strip center, and warehouse built included pervious pavements and green roofs, nearly 13.5 million square feet of impervious surface could be eliminated. More realistically enco uraging green roofs and pervious pavements to be utilized during the design and construction of certain types of development can have a significant impact on the amount of ISC. For instance, if new industrial and commercial buildings were required to have green roofs, eco roofs, or some type of vegetated roof, over 4 million square foot of impervious surface would have been eliminated in the Perry Creek watershed alone. Alternatively if residential development was required to have pervious pavement instal led in driveways and parking areas, roughly 400,000 square feet of impervious surface would have been avoided in Perry Creek, Middle Creek and Smith Creek in developments constructed between 1999 and 2005.

PAGE 68

68 Table 5 6 Trimmed mean of Poplar Creek w ater q uality r eadings Year Conductance Suspended Residue Turbidity TKN NO2 + NO3 TP 1995 63.00 7.83 NA 0.42 1.10 0.18 1996 62.50 5.33 NA 0.50 1.04 0.13 1997 103.42 5.83 NA 0.41 1.12 0.14 1998 88.17 8.17 NA 0.41 0.91 0.16 1999 100.33 6.45 6.63 0.36 1.04 0.17 2000 105.00 3.80 5.73 0.37 0.43 0.17 2001 113.92 6.00 8.27 0.48 1.45 0.18 2002 120.67 4.95 7.37 0.45 1.16 0.22 2003 89.25 5.78 9.87 0.50 1.14 0.10 2004 87.42 11.28 16.92 0.47 1.34 0.12 2005 127.67 10.45 12.72 0.41 1.73 0.15 2006 125.42 6.10 7.13 0.46 1.94 0.13 Table 5 7 Trimmed m ean of Middle Creek w ater q uality r eadings Year Conductance Suspended Residue Turbidity TKN NO2 + NO3 TP 1996 136.67 5.83 NA 0.56 0.85 0.19 1997 260.92 6.17 NA 0.70 1.62 0.26 1998 225.50 12.17 NA 0.79 2.66 0.47 1999 246.08 12.12 9.42 1.02 0.57 0.42 2000 265.42 8.50 11.60 0.65 0.08 0.34 2001 276.00 4.83 11.73 0.68 1.38 0.58 2002 366.25 3.48 7.92 0.85 1.68 0.68 2003 111.17 9.98 21.00 0.63 1.08 0.30 2004 113.00 5.62 13.50 0.77 1.44 0.46 2005 169.42 9.10 20.67 0.73 1.30 0.37 2006 203.33 5.33 8.08 0.74 1.06 0.42

PAGE 69

69 Table 5 8 Trimmed m ean of Smith Creek w ater q uality r eadings Year Conductance Suspended Residue Turbidity TKN NO2 + NO3 TP 1995 63.00 7.83 NA 0.42 1.10 0.18 1996 62.50 5.33 NA 0.50 1.04 0.13 1997 103.42 5.83 NA 0.41 1.12 0.14 1998 88.17 8.17 NA 0.41 0.91 0.16 1999 100.33 6.45 6.63 0.36 1.04 0.17 2000 105.00 3.80 5.73 0.37 0.43 0.17 2001 113.92 6.00 8.27 0.48 1.45 0.18 2002 120.67 4.95 7.37 0.45 1.16 0.22 2003 89.25 5.78 9.87 0.50 1.14 0.10 2004 87.42 11.28 16.92 0.47 1.34 0.12 2005 127.67 10.45 12.72 0.41 1.73 0.15 2006 125.42 6.10 7.13 0.46 1.94 0.13

PAGE 70

70 Figure 5 1 Trimmed mean r esults g raph : C onductance Figure 5 2 Trimmed m ean r esults g raph: S uspended R esidue

PAGE 71

71 Figure 5 3 Trimmed m ean r esults g raph: T otal K jeldahl N itrogen (TKN) Figure 5 4 Trimmed m ean r es ults g raph: N itrite (NO2) + N itrate (NO3)

PAGE 72

72 Figure 5 5 Trimmed m ean r esults g raph: T otal P hosphorus (TP) Table 5 10. Trimmed mean analysis r esults Variable Increasing Trimmed Mean Base Year to 2005 Decreasing Trimmed Mean Base Year to 2005 Conductivity Poplar Creek, Middle Creek, Smith Creek None Suspended Residue Poplar Creek, Middle Creek, Smith Creek TKN Poplar Creek, Middle Creek Smith Creek NO3 + NO2 Poplar Creek, Middle Creek Smith Creek

PAGE 73

73 CHAPTER 6 DISCUSSION Discussion of Findings and Methods This study has quantified the growth of impervious surface resulting from development in four watersheds in rapidly urbanizing parts of Wake County, North Carolina. Local information has been collected regarding the average rates of impervious surface by land use in development between 1999 and 2005. In addition, this study has shown that certain facets of the urban environs, including rooftops and pavement for driveways and parking lots are principal contributors to the amount of impervious surface in suburban watersheds. Significant reductions in the amount of impervious surface in new developments are achievable by encouraging the incorporation of pervious pavements and green roofs in new developments. Quantifying Impervious Surface Growth This study found that urban growth in developing areas of Wake County, North Carolina has lead to significant increases in impervious surface. It was found that 13.54%, or 4,404.01 acres of land in the study areas is covered by impervious surface. Of this, it was estimated that at least 996.87 acres of impervious surface was constructed between 1999 and 2005. In the four urbanizing watershed s that were investigated there was a 22.6% growth in impervious surface between 1999 and 2005. Literature suggests that this growth, 43.4 million square feet of impervious surface will have significant impacts on the hydrology of the watersheds and the health of aquatic ecosystems. Comparison of Impervious Surface Rates by Land Use The study also has shown that impervious su rface rates developed by the SCS are accurate estimations of residential impervious surface rates in Wake County, NC Impervious surface rates by land use were quantified for development that took place between 1999 and 2005. For

PAGE 74

74 five residential land us e categories the average impervious surface rates as measured in the study areas were within a few percentage points of the impervious surface rates found by the SCS Only for three classes res idential land use with a density greater than 7 units per acre commercial and industrial were impervious surface rates significantly different. The differences in these rates may be due to the fact that the study areas were suburban in character. Measuring ISC rates in more urban areas may affirm the SCS figures. Also the study areas were primarily residential in nature, with only 63 commercial parcels and 60 industrial parcels. A sampling of more commercial and industrial parcels may result in different impervious surface rates for those land use categories. Benefits of Encouraging Green Roofs and Pervious Pavements Using information collected by the City of Olympia, WA this study estimated the amount of impervious surface growth that can be attributed to rooftops and driveways or parking areas in the study are as. It was found that 308.63 acres of impervious rooftops, driveways and parking areas w ere constructed between 1999 and 2005. This accounts for 7% of all impervious surfaces in the study areas. If all new residential development was required to use perv ious pavement, pavers or concrete for parking areas and driveways, impervious surface growth between 1999 and 2005 would have been reduced by nearly 1.4 million square fee t If just 20% of all rooftops were required to be vegetated, it would have reduced the total amount of impervious surface in the study areas by 1.1 million square feet. This data shows that local governments can have a significant impact in the amount of ru noff generated by new developments by adopting policies that provide incentives to developers that utilize construction methods that have the potential to mitigate increases in impervious surfaces.

PAGE 75

75 Roads and Impervious Surface Impervious surface increase s in tandem with road density (see Figure 6 1 ). In three out of four watersheds road density increases of over 20% corresponded with an increase in impervious surface of over 20%. Roads contribute directly to this growth in IS C due to the face that trans portation related impervious areas can account for 70% of total impervious surface area (Arnold & Gibbons 1996, p. 248). Not only do roads contribute directly to ISC growth, a s roads get built, there is better access to parcels of land which enables more intense development. Larger shopping centers and more dense residential developments are enabled by the reduced travel times resulting from more connectivity. The fact that transportation planners seek to increase connectivity and capacity, in order to r educe congestion, conflicts with the need to limit impervious surfaces in order to protect water quality. T his study did not directly measure the amount of impervious surface that could be saved using alternative design standards such as headwater street s Although additional study is needed to quantify the benefits of adopting alternative design standards it can be assumed that impervious surface attributable to new streets constructed to serve recently constructed residential neighborhoods contributes significantly to the total amount of ISC in a watershed. Accordingly it would be beneficial for growing municipalities in the Piedmont to investigate the adoption of revised street standards that allow for narrower street widths on low volume residential feeder roads Water Quality and Impervious Surface Research has proven that impervious surface is a good indicator of the health of aquatic systems. This study used water quality data collected between 1995 and 2005 to detect trends in readings of key parameters, including conductivity, TKN, Nitrate + Nitrite, suspended residue, and total phosphorus. There were statistically significant positive trends found in total

PAGE 76

76 phosphorus readings in Middle Creek, conductance and total nitrogen readings in Poplar Creek, and conductance readings in Smith Creek. The study found only one statistically significant negative trend. This analysis supports the hypothesis that water quality is negatively impacted by urban growth. As this research was limited to investigating trends in levels of water quality measurements, it does not address impacts to the biotic community. Analysis of invertebrate and fish community sampling data collected by the Division of Water Quality could provide more insight into changes that ma y be detectable only through measuring the subtle effects of increases of impervious surface over a relatively short timeframe. Feature Extraction Using Feature Analyst This study demonstrated that Feature Analyst, an extension to ArcGIS created by Visua l Learning Systems can be used to extract impervious surface from large watersheds. The semi automated extraction process used in the study included automated extraction based on a watershed specific model followed by manual editing. This method enabled the investigation of larger study areas than would be possible if manual digitization was employed. Using one training set to develop a model used to extract features from a number of large watershed s led to a significant amount of clutter. Watershed sp ecific model improved results, however classification still required manual cleanup, which added time to the processing. Even with the added time for manual cleanup, this study illustrates that extraction of impervious surface using Feature Analyst is fas ter than manually digitizing. Classification results could potentially have been improved to the point that manual cleanup was not required by improving training sets, or developing training sets over multiple images. Alternatively, the study watersheds could have been subdivided into catchments and individual model s could have been developed for these catchments Time was a factor due to hardware limitations on the computer

PAGE 77

77 used for model development. Rather lengthy delays (~5 minutes) resulted when se lecting incorrect and correct classification results using the Remove Clutter Tool in Feature Analyst. It was difficult to develop a model that would correctly classify weathered rooftops and red rooftops. These problems were due to the fact that the co lor signatures of these features resembled those found in natural features. Bare fields and leaf off stands of deciduous trees also presented problems due to the fact that these features were shades of gray also found in parking lots, sidewalks and rooftops. These false classifications could potentially have been removed by perfecting training sets. Time constraints did not allow for this to be explored in this study. Opportunities for Future Research Although this study determined that using Feature A nalyst could increase the area in which impervious surface could be quantified, it remains to be seen if extraction models can be produced that result in a high level of accuracy when applied to large watersheds. The mixes of land use and land cover in rural areas of watersheds seem to result in misclassifications. Ideal ly a methodology for developing impervious extraction models for large watersheds could be researched in order to enable the expeditious development and revision of models and subsequent a pplication to new sets of high resolution aerials This study estimated the amount of impervious surface growth due to roads (See Figure 6 1). A vailable datasets may have limited results S patial accuracy of existing road inventories produced and maintained by Wake County may have caused inaccuracies. In addition an underestimation of ISC due to roads may have resulted from the lack of available road width data. A f ocused study of ISC growth resulting from expansion of road networks would be ben eficial. Additional research is needed regarding ISC rates of commercial and industrial properties. This study investigated impervious surface rates on a limited number of commercial and

PAGE 78

78 industrial parcels. A focused study which details the rates of impervious surface on these land uses in different contexts would provide insight on trends and potential differences between industrial and commercial properties in rural, suburban and urban areas. Finally, there is a need for refined p ercentage of impervious surface attributable to different facets of the built environment. This study used estimates produced by the City of Olympia, WA and generalized to match land use categories used by the Soil Conservation Service. Although these figures can be used to estimate percentages of impervious surface in the form of transportation related facilities and rooftops, they were produced based on a small sample size In addition, there may be region al differences in building trends which would prove the use of the estimation technique employed in this study inaccurate. It is possible that a feature extraction software package, such as Feature Analyst could be trained to classify ISC in individual de velopments by distinguishing between rooftops, driveways and streets. This would enable the efficient analysis of building trends in a city or region. Accordingly the percentages of ISC found in the Olympia, WA study, as well as the methodology employed in this study could be validated or refined.

PAGE 79

79 Figure 6 1 Growth in impervious s urface and road d ensity

PAGE 80

80 CHAPTER 7 CONCLUSION This study has analyzed recent development in four watersheds in Wake County. G rowth in impervious surface was measu red using Feature Analyst, water quality trends were measured using a Mann -Kendall test and monthly water quality data, and opportunities to incorporate green roofs and pervious pavements to mitigate negative impacts were identified Suburban development can rapidly increase impervious surface coverage. In a period of seven years, 19992005, impervious surface in the study areas grew by 996 acres. Development in the study areas resulted in as much impervious surface as building a parking lot a mile and a half wide. This study has determined that i mpervious surface rates developed by the Soil Conservation Service can be used to accurately estimate impervious surface for most land uses in existing or proposed development s in the North Carolina Piedmont, pa rticularly in areas on the urban/rural fringe on the edge of North Carolinas Main Street. This finding can be helpful to local and regional governments tasked with estimating current or future impervious surface. The water quality analysis conducted as a component of this thesis found five instances of decreasing water quality and one instance of improving water quality. This supports the initial hypothesis that impervious surface growth negatively impacts water quality. Further studies that investiga te the relationship between impervious surface growth, road density, and biological indicators of the health of aquatic communities in the North Carolina Piedmont would be helpful. Design will be an important determinant of the built environments future effects on aquatic systems. This study has shown that the incorporation of green roofs and pervious pavements in new development can significantly reduce the amount of impervious surface in new development. Local government policy can en courage the inco rporation of green roofs and pervious pavements by providing incentives or amending ordinances to require these types of

PAGE 81

81 low imp act development features. Ideally local governments could require that a certain percentage of pavements in residential neighborhoods be porous pavement or pervious pavers. Sidewalks and driveways rarely bear extreme loads and would be ideal candidates for these features. Commercial and industrial buildings could be required to incorporate green roofs on a certain percentage of buildings. Requirements and incentives could vary based on percentage of impervious surface on site, current health of the watershed, and additional factors that affect the developments potential impacts on stormwater. Implementing such policies may be problematic due to the fact that additional technical expertise and cost may be accrued. However, a departure from existing trends is necessary in order to mitigate the impacts of projected growth. The method of feature extraction employed in this study could be expanded and applied to a larger geographic region, but may require additional time for model development and manual editing One of the drawbacks of using Feature Analyst is the amount of time it requires to create and improve extraction models Depending on the land use mix and the quality of training sets utilized there may be a need for manual editing to remove false positives, as in this study. Using a Mann -Kendall trend test on a larger body of water may require adjustments in order to ac count for intervening variables. A larger drainage area could result in more point source polluters upstream of sampl ing stations. Significant nutrient inputs from point source polluters, including municipal wastewater treatment plants would have to be t aken into account. The methods employed in this study could be applied successfully in many different geographic regions in the United States. Areas with limited tree cover, similar soils, and hydrology would be ideal candidates. Although the effect of tree cover on accuracy of feature extraction was not investigated in this study, a cursory inspection of extraction results indicates increased tree cover may result in reduced accuracy when using Feature Analyst and color aerial

PAGE 82

82 photography. This may be able to be avoided by using color infrared imagery. Th e methodology employed assumed that soils and topography of the study areas result in predictable overland flow, channel initiation and stream formation. It is also assumed that all impervious surfac e is hydrologically connected to stream networks and impacts water quality in the streams sampled. This assumption would not necessarily hold true in a region with different soils and hydrologic patterns. For instance, areas of northern Florida have soil s of a mixture of sand and clay as well as hydrologic systems similar to the North Carolina Piedmont, whereas other areas of Florida, including the west side of Alachua County and other areas of Florida have karst topography and rapidly drained sandy soils which result in limited surface flow. Accordingly, effects of impervious surface on water quality and runoff patterns will be significantly different and methods used to determine the relationship between impervious surface and water quality need to be a djusted accordingly.

PAGE 83

83 APPENDIX A P ARCEL CLASSIFICATION DOCUMENTATION APA Activity Codes to NRCS Land Use Category Lookup Table DESC NRCS_USDA_1986 Residential Activities Residential Household Activities Residential Transient Living Residential Institutional Living Residential Household With Business Use Residential Shopping, Business, Or Trade Activities Commercial Shopping Commercial Goods Oriented Shopping Commercial Service Oriented Shopping Commercial Retail Combination Commercial Restaurant Type Commercial Restaurant Type With Drive Through Commercial Office Activities Office Office Activities With High Turnover Of People Office Office Activities With High Turnover Of Automobiles Office Industrial, Manufacturing, And Waste Related Activities Industrial Plant, Factory, Or Heavy Goods Storage Or Handling Activities Industrial Primarily Plant Or Factory Type Activities Industrial Primarily Goods Storage Or Handling Activities Industrial Laundry Activities Commercial Solid Waste Management Activities Industrial Solid Waste Collection And Storage Industrial Landfilling Or Dumping Other Waste Processing Or Recycling Industrial Construction Activities (Grading, Digging, Etc.) Industrial Research Or Laboratory Activities Office Social, Institutional, Or Infrastructure Related Activities Civic School Or Library Activities Education Classroom Type Activities Education Training Or Instructional Activities Outside Classrooms Education Other Instructional Activities Including Those That Occur In Libraries Education Child Day Care Activities Commercial Emergency Response Or Public Safety Related Activities Civic

PAGE 84

84 DESC NRCS_USDA_1986 Fire And Rescue Related Activities Civic Police, Security, And Protection Related Activities Civic Emergency Or Disaster Response Related Activities Civic Activities Associated With Utilities (Water, Sewer, Power, Etc.) Industrial Water Supply Related Activities Other Sewer Related Control, Monitor, Or Distribution Activities Other Power Generation, Control, Monitor, Or Distribution Activities Industrial Power Generation, Storage, Or Processing Activities Industrial Telecommunications Related Control, Monitor, Or Distribution Activities Industrial Natural Gas Or Fuels Related Control, Monitor, Or Distribution Activities Industrial Mass Storage, Inactive Other Water Storage Other Storage Of Natural Gas, Fuels, Etc. Other Storage Of Chemical, Nuclear, Or Other Materials Other Health Care, Medical, Or Treatment Activities Commercial General Medical Care Commercial Veterinary Medical Care Commercial Interment, Cremation, Or Grave Digging Activities Other Military Base Activities Other Ordnance Storage Other Range And Test Activities Other Travel Or Movement Activities Transportation Pedestrian Movement Transportation Vehicular Movement Transportation Vehicular Parking, Storage, Etc. Transportation Drive In, Drive Through, Stop N Go, Etc. Commercial Truck Transport Terminal Industrial Trains Or Other Rail Movement Transportation Rail Maintenance, Storage, Or Related Activities Industrial Sailing, Boating, And Other Port, Marine And Water Based Activities Transportation Boat Mooring, Docking, Or Servicing Transportation Port, Ship Building, And Related Activities Industrial Aircraft Takeoff, Landing, Taxiing, And Parking Other Spacecraft Launching And Related Activities Other

PAGE 85

85 DESC NRCS_USDA_1986 Mass Assembly Of People Other Passenger Assembly Other Spectator Sports Assembly Recreation and Entertainment Movies, Concerts, Or Entertainment Shows Recreation and Entertainment Gatherings At Fairs And Exhibitions Recreation and Entertainment Mass Training, Drills, Etc. Other Social, Cultural, Or Religious Assembly Other Social Assembly Ot her Religious Assembly Other Gatherings At Galleries, Museums, Aquariums, Zoological Parks, Etc. Commercial Historical Or Cultural Celebrations, Parades, Reenactments, Etc. Recreation and Entertainment Leisure Activities Recreation and Entertainment Active Leisure Sports And Related Activities Recreation and Entertainment Running, Jogging, Bicycling, Aerobics, Exercising, Etc. Other Equestrian Sporting Activities Other Hockey, Ice Skating, Etc. Recreation and Entertainment Skiing, Snowboarding, Et c. Recreation and Entertainment Automobile And Motorbike Racing Recreation and Entertainment Golf Recreation and Entertainment Tennis Recreation and Entertainment Track And Field, Team Sports (Baseball, Basketball, Etc.), Or Other Sports Recreation and Entertainment Passive Leisure Recreation and Entertainment Camping Recreation and Entertainment Gambling Recreation and Entertainment Hunting Recreation and Entertainment Promenading And Other Activities In Parks Recreation and

PAGE 86

86 DESC NRCS_USDA_1986 Entertainment Shooting Recreation and Entertainment Trapping Recreation and Entertainment Flying Or Air Related Sports Recreation and Entertainment Water Sports And Related Leisure Activities Recreation and Entertainment Boating, Sailing, Etc. Recreation and Entertainment Canoeing, Kayaking, Etc. Recreation and Entertainment Swimming, Diving, Etc. Recreation and Entertainment Fishing, Angling, Etc. Recreation and Entertainment Scuba Diving, Snorkeling, Etc. Recreation and Entertainment Water Skiing Recreation and Entertainment Natural Resources Related Activities Recreation and Entertainment Farming, Tilling, Plowing, Harvesting, Or Related Activities Agriculture and Timber Livestock Related Activities Agriculture and Timber Pasturing, Grazing, E tc. Agriculture and Timber Logging Agriculture and Timber Quarrying Or Stone Cutting Extraction Mining Including Surface And Subsurface Strip Mining Extraction Drilling, Dredging, Etc. Extraction No Human Or Unclassifiable Other Not Applicable To This Dimension Other Unclassifiable Other Subsurface Extraction To Be Determined Other To Be Determined Other Parcel Queries The following query was used to extract single family residential land uses from the Wake County parcel layer:

PAGE 87

87 "NRCS_USDA_" = 'Residential' AND "TOTUNITS" = 1 The following query was used to extract multi -family residential land uses from the Wake County parcel layer: "NRCS_USDA_" = 'Residential' AND "TOTUNITS" > 1 The following query was used to extract unclas sified residential (where parcel layer may be incorrect) from the Wake County parcel layer: "NRCS_USDA_" = 'Residential' (where "TOTUNITS" =0) These parcels were filled with "Unclassified Res" To split residential NRCS into specific lot sizes the follo wing queries were used: "NRCS_USDA_" = 'Single Family Residential' and "DU_DEN" < 0.5 "NRCS_USDA_" = 'Single Family Residential' and "DU_DEN" >= 0.5 and "DU_DEN" <1 "NRCS_USDA_" = 'Single Family Residential' and "DU_DEN" >= 1 and "DU_DEN" < 2 "NRCS_USDA _" = 'Single Family Residential' and "DU_DEN" >= 2 and "DU_DEN" < 3 "NRCS_USDA_" = 'Single Family Residential' and "DU_DEN" >= 3 and "DU_DEN" < 4 "NRCS_USDA_" = 'Single Family Residential' and "DU_DEN" >= 4 and "DU_DEN" < 5 "NRCS_USDA_" = 'Single Family Re sidential' and "DU_DEN" >= 5 and "DU_DEN" <= 7 "NRCS_USDA_" = 'Single Family R esidential' and "DU_DEN" > 7 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" < 0.5 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 0.5 and "DU_DEN" <1 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 1 and "DU_DEN" < 2 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 2 and "DU_DEN" < 3 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 3 and "DU_DEN" < 4 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 4 and "DU_DEN" < 5 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >= 5 and "DU_DEN" <= 7

PAGE 88

88 "NRCS_USDA_" = 'Multi -family Residential' and "DU_DEN" >7

PAGE 89

89 LIST OF REFERENCES Arnold Jr., Chester L., & Gibbons, C. James. 1996. Impervious Surface Coverage. Journal of the American Planning Association. 62, 2 : 243258. Bartuska, J. Tom, Editor. 2007. The Built Environment: A Collaborative Inquiry Into Design and Planning. Hoboken, New Je rsey. Wiley, John & Sons, Incorporated. Brabec, Elizabeth, Schulte, Stacey, and Richards, Paul. 2002. Impervious Surfaces and Water Quality: A Review of Current Literature and Its Implications for Watershed Planning. Journal of Planning Literature 16, 4: 499514. Bracken, David. 2008, November 29. New Reservoirs Protection Too Lax, Raleigh Leaders Say. The News & Observer, p. 1B. Congalton, Richard G. 1991. A Review of Assessing the Accuracy of Classifications of Remotely Sensed Data. Re mote Sensing of Environment 37 1: 3546. Dunnet, Nigel & Kingsbury, Noel. 2008. Planting Green Roofs and Living Walls Portland, Oregon. Timber Press. EPA. Polluted Runoff Webpage. http://www. epa.gov/owow/nps/qa.html (a ccessed October 13, 2008). ERDAS. Feature Extraction and Classifications Solutions Paper. http://www.erdas.com/files/whitepapers/Feature_Extrt_Class.p df (a ccessed October 1, 2008) Forman Richard. 2003. Road E cology: Science and Solutions Washington, DC: Island Press. Galli, John. 1990. Thermal Impacts Associated with Urbanization and Stormwater Management Besst Management Practices. Washington, DC. Metropolitan Washington Council of Governments for the Sediment and Stormwater Administration of the Maryland Department of the Envi ronment. Hirsch, R.M., & E.J. Gilroy. 1984. A nonparametric trend test for seasonal data with serial dependence. Water Resources Research 20 5: 727732. Lenihan, John & Fletcher, W. William 1978. Environment and Man Volume Eight. The Built En vironment Chapter 5. James K. Feibleman New York, NY. Academic Press. Lenat, R. David, & Crawford, J. Kent 1994. Effects of land use on water quality and aquatic biota of three North Carolina Piedmont Streams. Hydrobiologia. 294, 3: 185 199. Ma nn, B. Henry. 1945. Nonparametric Tests Against Trend. Econometrica 13, 3: 245259.

PAGE 90

90 Miller, E. Jennifer, & Hess, R. George 2007. Southern TwoLined Salamanders in Urbanizing Watersheds. Urban Ecosystems. 10, 1: 7385. Morse, C. Chandler, Huryn, D. Alexander, & Cronan, Christopher. 2002. Impervious Surface Area as a Predictor of the Effects of Urbanization on Stream Insect Communities in Maine, U.S.A. Environmental Monitoring and Assessment 89 1: 95127, 2003. Netherlands. National Resources Conservation Service 2007 Annual Natural Resources Inventory 2003 http://www.nrcs.usda.gov/technical/NRI/2003/Landuse -mrb.pdf (accessed February 16, 2009) National Ready Mix Concrete Association. http://www.perviouspavement.org/ (a ccessed February 16, 2009). NC Onemap Download Page. www.nconemap.com/default.aspx?tabid=286 (a ccessed July 12, 2008) North Carolina Department of Environment and Natural Resources Division of Water Quality Environmental Services Section. 2006. Neuse River Basinwide Assessment. Raleigh, NC North Carolina State D emographics 2007. County/State Population Estimates. http://demog.state.nc.us/frame_start.html (accesses April 6 2008). OBrien, A. Michael. Feature Extraction with the VLS Feature Analyst Syst em. http://vls inc.com/feature_analyst/publications/reviews/obrien_asprs.pdf (accessed February 22, 2009) Paul, J. Michael, & Meyer, L. Judy 2001. Streams in the urban landscape Annual Review of Ecology and Systematics 32, 1 : 333365. Potter, M. Kevin, Cubbage, W. Fredrick, Blank, B Gary, & Schaberg, H. Rex. 2004. A Watershed -Scale Model for Predicting Non -point Source Pollution Risk in North Carolina. En vironmental Management. 34, 1 : 6274. Randolph, John 2004. Environmental Land Use Planning and Management Washington, D.C. Island Press. Rawlins, Wade. 2009, March 10. Falls Lake Rules Taking a Little Longer. The News & Observer. www.newsobserver.com/news/growth/story/1435747.html (accessed March 19, 2009). Robson, A.J., & Neal, C. 1997. Regional Water Quality of the River Tweed. The Science of the Total Environment 194, 195: 173192. Rogers, Peter. 1994. Hydrology and Water Quality in Changes in Land Use and Land Cover E dited by W. B. Meyer and B. L. Turner III, pp. 231258. University Press, Cambridge.

PAGE 91

91 Rome, Adam. 2001. The Bulldozer in the Countryside: Suburban Sprawl and the Rise of American Environmentalism New York, NY. Cambridge University Press. Schueler, Tom. 2000. Basic Concepts of Watershed Planning. The Practice of Watershed Protection Edited by T. Schueler and H. Holland, pp. 14561. Ellicott City, MD: Center for Watershed Protection. Scott, C. Mark. 2006 Winners and Losers Among Stream Fishes in Relation to Land Use Legacies and Urban Development in the Southeastern U S. Biological Conservation 127 3 : 301309. Soil Conservation Service, U.S. Department of Agriculture. 1975. Urban Hydrology for Small Watersheds. Technical Release No. 55. Washington, D.C. Soil Conservation Service, U.S. Department of Agriculture. 1986. Urban Hydrology for Small Watersheds. Tech nical Release No. 55. Washington, D.C. Stormwater Managers Resource Center. Impervious Cover for Various Land Uses. www.stormwatercen ter.net/monitoring%20and%20assessment/simple%20meth/simple%2 0imp%20table%205.htm (accessed February 16, 2009). The Kendall Package. http://cran.r -project.org/web/packages/Kendall/K endall.pdf (a ccessed February 22, 2009). The R Foundation for Statistical Computing. What is R? http://www.r -project.org/ (accessed February 22, 2009). United Nations Center for Housing, Building and Planning. 1974. Human Settlements: The Environmental Challenge. London, UK Macmillan Press. United Nations Department of Economic and Social Affairs, Population Division. 2005. World Urbanization Prospects The 2004 Revision: Data Tables and Highlights. Un ited Nations, New York. United Nations Department of Economic and Social Affairs, Population Division. 2004. World Population to 2300. United Nations, New York. United States Census Bureau. American FactFinder. http://factfinder.census.gov/servlet/QTTable?_bm=n&_lang=en&qr_name=DEC_1990_ST F1_DP1&ds_name=DEC_1990_STF1_&geo_id=05000US37183 (accessed February 22, 2009) United States Census Bureau. State & County QuickFacts. http://quickfacts.census.gov/qfd/states/37/37183.html (accessed February 22, 2009). Wake County Off ice of Growth Management, Wake Public School System. 2005. Growth and Diversity. Raleigh, NC.

PAGE 92

92 Wake County W atershed Management Plan. 2003. Wake County Water Taskforce. Raleigh, NC Yue, Sheng, Pilon, Paul, & Cavadias, George. 2001. Power of the Mann-Kendall and Spearmans rho tests for detecting monotonic trends in hydrological series. Journal of Hydrology 25 9 1 4 : 254271.

PAGE 93

93 BIOGRAPHICAL SKETCH Jake Petrosky was born in Ft. Lauderd ale, Florida in 1981. He grew up in Raleigh and Knightdale, North Carolina and graduated from East Wake High School with honors in 1999. Jake attended Appalachian State University and graduated in December of 2003 with university and departmental honors and a degree in c ommunity and r egional p lanning a minor in g eography and a minor in s ustainable d evelopment. After graduating Jake worked at the Capital Area MPO as a Planning Technician and GIS Analyst/Programmer conducting multi -modal transportation p lanning, socio-economic forecasting, and environmental analysis for three years prior to attending the University of Florida. As a graduate student at the University of Florida in the Department of Urban and Regional Planning Jake worked at an joint assistantship with the Gainesville Metropolitan Transportation Planning Organization (GMTPO) and the GeoPlan Center. At the GMTPO Jake served as the Efficient Transportation Decisio n Making (ETDM) coordinator and as project manager for the development of the socio -economic forecasts for the upcoming Long Range Transportation Plan update. During his time at the University of Florida Jake was awarded the WRS Infrastructure & Environme nt Inc. Award In Memoriam of Mario Ripol for Outstanding Achievement for Planning Information and Analysis. During his spare time he volunteered for the Conservation Trust for Florida and acted as safety coordinator for the University of Florida Outdoor A dventure Recreation Club (OAR).