An analysis of urban development and its environmental impact on the Tampa Bay watershed

An analysis of urban development and its environmental impact on the Tampa Bay watershed

ARTICLE IN PRESS Journal of Environmental Management 85 (2007) 965–976 www.elsevier.com/locate/jenvman An analysis of urban development and its envi...

2MB Sizes 1 Downloads 32 Views

ARTICLE IN PRESS

Journal of Environmental Management 85 (2007) 965–976 www.elsevier.com/locate/jenvman

An analysis of urban development and its environmental impact on the Tampa Bay watershed George Xiana,, Mike Craneb, Junshan Suc a

SAIC, Center for Earth Resources Observation and Science (EROS), Sioux Falls, SD 57198, USA US Geological Survey (USGS), Center for Earth Resources Observation and Science (EROS), Sioux Falls, SD 57198, USA c Engineering Division, Hillsborough County, Tampa, FL 33601, USA

b

Received 7 January 2006; received in revised form 27 October 2006; accepted 4 November 2006 Available online 8 January 2007

Abstract Urbanization has transformed natural landscapes into anthropogenic impervious surfaces. Urban land use has become a major driving force for land cover and land use change in the Tampa Bay watershed of west-central Florida. This study investigates urban land use change and its impact on the watershed. The spatial and temporal changes, as well as the development density of urban land use are determined by analyzing the impervious surface distribution using Landsat satellite imagery. Population distribution and density are extracted from the 2000 census data. Non-point source pollution parameters used for measuring water quality are analyzed for the subdrainage basins of Hillsborough County. The relationships between 2002 urban land use, population distribution and their environmental influences are explored using regression analysis against various non-point source pollutant loadings in these sub-drainage basins. The results suggest that strong associations existed between most pollutant loadings and the extent of impervious surface within each sub-drainage basin in 2002. Population density also exhibits apparent correlations with loading rates of several pollutants. Spatial variations of selected non-point source pollutant loadings are also assessed. r 2006 Elsevier Ltd. All rights reserved. Keywords: Urbanization; Remote sensing; Water quality; Pollutant loading

1. Introduction Urban development in most metropolitan areas in the United States has grown tremendously in the last 50 years. Associated with this growth are increasing population (US Bureau of Census, 2001) and change of land cover types from permeable land to anthropogenic impervious surfaces. Impervious surface area (ISA) is defined in this study as constructed surfaces—roofs, roads, parking lots, driveways, and sidewalks. Impervious surfaces can alter the natural hydrological condition by increasing the volume and rate of surface runoff and decreasing ground water recharge and base flow (Moscrip and Montgomery, 1997). This eventually leads to larger and more frequent local flooding and reduced water supplies for urban and suburban areas (Harbor, 1994). Other direct environmental Corresponding author. Tel.: +1 605 594 2599; fax: +1 605 594 6529.

E-mail address: [email protected] (G. Xian). 0301-4797/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2006.11.012

impacts of increasing ISA in watersheds include the degradation of water resources and water quality when surface runoff transports non-point source pollutants from their source areas to receiving lakes and streams (Gove et al., 2001; USEPA, 2001). Pollutants either dissolved or suspended in water or associated with sediment, including nutrients, heavy metals, and oil and grease, can accumulate and wash away from ISAs. Impervious surface also has been considered a key environmental indicator of the health of urban watersheds (Schueler, 1994) and as an indicator of non-point source pollution or polluted runoff (Arnold and Gibbons, 1996; Slonecker et al., 2001). Research conducted by Yin et al. (2005) used population density and built-up land surface to investigate the relationship between urbanization patterns and water quality in Shanghai, China. They found strong correlations between developed land, population density, and water quality, where contributions of untreated domestic wastewater and non-point pollution were made to nearby

ARTICLE IN PRESS 966

G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

waterways. Jeng et al. (2005) also suggested that built-up land surface intensified the concentration of indicator organisms in water columns and sediment in an estuarine environment. However, several environmental consequences are associated with runoff in urbanized watersheds, including increases in runoff volume, loss of nutrients, and significant losses of oil and grease and certain heavy metals (Tang et al., 2005). Rapidly built-up land is a threat to surface water quality when total runoff increases, and hydrologic impairment leads to erosion and sedimentation. It is important to map the locations of sources of degraded stormwater runoff quality within drainage basins and identify areas that are hazardous to the beneficial uses of receiving waters (Mitchell, 2005). Using satellite remote sensing information in combination with regression analysis for predicting surface pollutant loadings associated with urban development appears to be an effective approach (Yin et al., 2005). Milesi et al. (2003) suggested that anthropogenic development also produced regional impacts on ecosystem structure and function. Land cover change due to urban development during the 1992–2000 period reduced annual net primary productivity of the southeastern United States by 0.4%. Land use and land cover (LULC) change associated with urban development is considered one of the most disturbing processes because it causes dramatic changes in the natural energy and material cycles of ecosystems and influences mesoscale weather patterns, local climate conditions, biodiversity, and water resources (Miller, 1981; Berry, 1990; Oke, 1989; Pielke et al., 1999; Kalnay and Cai, 2003). These studies reveal the need for a method that is consistent across spatial scales and supports resource management and impact assessment of ecosystem and environmental conditions affected by urban land use. The objective of this study is to investigate the effects of urbanization processes and population distribution on water and ecosystem quality and environmental conditions in the Tampa Bay watershed. The spatial extent of urban development in the region is estimated from Landsat satellite data using sub-pixel ISA as an indicator (Xian and Crane, 2005). The relationship between spatial patterns of urban land use, population distribution, and pollutant loadings in different sub-drainage basins is also investigated. Improved understanding of the relations among these factors will certainly help planning efforts for future water and ecosystem quality and natural resource distribution and management.

Hillsborough, and Manatee Counties and portions of Pasco, Polk, and Sarasota Counties. Four major sources of surface water—the Hillsborough, Alafia, Little Manatee, and Manatee Rivers––flow into the bay. The watershed consists of ten major drainage basins: Hillsborough River, Coastal Old Tampa Bay, Coastal Hillsborough Bay, Alafia River, Coastal Middle Tampa Bay, Little Manatee River, Manatee River, Boca Ciega Bay, Coastal Lower Tampa Bay, and Terra Ceia Bay. Fig. 1 shows the spatial extent of the watershed and sub-drainage basins in Hillsborough County. The largest municipalities within the watershed are Tampa, St. Petersburg, Clearwater, and Bradenton. 2.2. Urban LULC and change measurement Urban development in the Tampa Bay area has been ongoing since the late 1880s. The region’s warm climate, coastal location, and abundant recreational opportunities have attracted people to the area. Many residential areas have developed in the Hillsborough River, Coastal Old Tampa Bay, Costal Hillsborough Bay, Alafia River, Costal Middle Tampa Bay, and Boca Ciega Bay drainage basins. Most of this growth occurred in Pinellas and Hillsborough Counties where the population increased by 148% and 158%, respectively, from 1960 to 2001. More than two

2. Methods and data 2.1. Study area Tampa Bay, located on the Gulf Coast of west-central Florida, has an areal extent of approximately 1030 km2 and is one of the largest open-water estuaries in the southeastern United States. The Tampa Bay watershed covers approximately 6600 km2 and encompasses most of Pinellas,

Fig. 1. The Tampa Bay watershed and 16 sub-drainage basins within Hillsborough County.

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

million people resided in the watershed by 2000. Recent urban land-use development has extended to the northeastern side of Tampa, where large open lands were available (Xian et al., 2005). Accommodating the growing population of the region requires a heightened awareness of environmental conditions associated with urban growth trends. Although the effect of urbanization, population growth, and impervious surface cover on water quality and the health of watersheds has been generally studied over the decades (Slonecker et al., 2001), challenges exist in quantifying the detailed spatial extent and distribution of imperviousness. Urban areas are usually heterogeneous, and most urban image pixels with the resolution of Landsat (30-m) or other similar satellite imagery are comprised of a mixture of different surface types. To measure urban spatial extent and evaluate its environmental influence without losing urban LULC heterogeneity, sub-pixel percent ISA was chosen as an indicator for identifying both spatial extent and intensity of urbanization in the watershed. Sub-pixel percent imperviousness was estimated using high-resolution orthoimagery to create training data sets for sample sites, and medium-resolution Landsat satellite imagery to extrapolate large area ISA through regression models. The ISA information derived from remote sensing imagery provides details of spatial extent and the intensity of urban LULC (Yang et al., 2003; Gillies et al., 2003). To estimate ISA at the sub-pixel level, high-resolution imagery such as 1-m digital orthophoto quarter quadrangles (DOQQ) from aerial photography, IKONOS, QuickBird, and 0.3-m orthoimagery are required to create ISA training data sets. Medium resolution satellite imagery is used for the large area ISA estimation. The general procedure for ISA estimation includes (1) classification of urban and nonurban land use from selected high-resolution images in the study area, (2) calculation of percent imperviousness from classified images and scaling percent imperviousness to 30m resolution for the development of training data in regression modeling, (3) selection of dependent variables for the regression tree models, (4) estimation of large area ISA using regression tree models, and (5) imperviousness change determination and accuracy assessment. The regression tree models obtained from training data contain collections of rules, where each rule has an associated multivariate linear model,

967

dates over the early 1990s to 2000s time period were processed. Detailed procedures can be found in Xian and Crane (2005) and Xian (2007). The accuracy assessment indicated that the mean systematic error (SE) and root mean square error (RMSE) of 2002 ISA estimations are 5.8 and 18.9, respectively. By using the 10% ISA threshold to differentiate urban and rural areas, the spatial extent and intensity of urban land use in the watershed were determined. 2.3. Population distribution One of the direct effects associated with urban growth in the region is the increase of population. To explore population distribution in the watershed, US Census 2000 data were used to display population and calculate population density for the watershed. Population density was calculated as the number of people per unit area in a given region. Fig. 2 shows population density per square kilometer for the entire watershed. High population density centers are associated with the cities of St. Petersburg and Tampa, where densities are greater than 1800 people per square kilometer. Another high population density area occurs on the southeastern side of Tampa. Most highdensity areas appear to be correlated with high-percent

Rules m if ½conditions are true then y ¼ f ðx1 ; x2 ; . . . ; xn Þ, where y is imperviousness and is defined as the dependent variable in the model, and f(x1, x2, y, xn) is a linear combination of multiple independence variables including different spectral bands and other input geographic information. To determine urban land use change over time within the watershed, three Landsat scenes from path 17, rows 40 and 41, and path 16, row 41 acquired from four

Fig. 2. Population density (people/km2) in the Tampa Bay watershed obtained from 2000 Census data.

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

impervious covers. Low-density areas are associated with medium-to-low residential urban land use. 2.4. Surface pollutant loadings in the watershed Anthropogenic inputs of phosphorus can be large for a watershed having large numbers of humans, domestic animals, and livestock (Freedman, 1995). To evaluate water quality in a watershed, land use and non-point pollutant source data are required. In this study, non-point pollutant loadings were provided in a geographic information system (GIS) format by the Engineering Division, Hillsborough County (Hillsborough County, 1999) and were used to investigate the spatial variation of water quality parameters. The pollutant loading data were collected for 16 sub-drainage basins within Hillsborough County, including the Hillsborough River, Alafia River, Little Manatee River, and Coastal Hillsborough Bay subdrainage basins. The City of Tampa is not included in the Hillsborough County drainage basin. The total areal extent of the sixteen sub-drainage basins is approximately 3680 km2 and their locations are depicted in Fig. 1. These data are not acquired on a regular basis. Pollutant loading estimates were produced from the Hillsborough County Pollutant Loading and Removal Model (PLRM). The PLRM was developed by the county’s public works/ stormwater management environmental team and was adopted as the pollutant loading assessment tool for the county. The model involves calculation of gross pollutant loads, estimation of net loads based on existing treatment, and evaluation of water quality level of service based on a comparison of existing loads to the single family residential benchmark. Several water quality parameters that represent domestic (organic), as well as industrial non-point pollutants, including 5-day biological oxygen demand (BOD5), total suspended solids (TSS), total Kjeldahl nitrogen (TKN), total nitrates and nitrites (NO3+NO2), total nitrogen (TN), total dissolved phosphorus (TDP), oil and grease, lead (Pb), and zinc (Zn), were evaluated. The annual average loadings for these variables are presented in Table 1. The Hillsborough River basin has the smallest annual loading for all selected sources, whereas most of the annual loadings are highest in the Alafia River basin. Inspection of land use information in the region showed that most BOD5 and nitrogen loadings in Hillsborough County are concentrated in agricultural and residential areas. Intensive landscape maintenance in residential neighborhoods also increases TKN and TP values. However, the TSS value (Hillsborough County, 1999) is lower

than the United States average and reflects less soil erosion and effective regulations for construction. Lead data for the county are lower than for other locations in Florida and this may indicate decreased emissions due to use of unleaded gasoline.

3. Results and analysis 3.1. Urban land use extent measured by percent ISA Using the 10% threshold of ISA estimated from Landsat imagery, the spatial extent of the built-up surface was calculated. The analysis of urban development during the 1991–2002 period indicated that most of the new growth occurred in the portion of Tampa Bay where vacant lands were available for new development (Xian and Crane, 2005). Fig. 3 shows changes in the extent of urban land use in the watershed from 1991 to 2002. During this period, urban land increased in extent threefold from approximately 600 km2 to approximately 1800 km2 (27% of the total watershed area). The 2002 percent ISA distribution in the watershed is shown in Fig. 4. Most of the high-percent imperviousness cover is located in the cities and along major highways. The dispersed nature of medium-to-lowdensity ISA that represents medium-to-low-density residential neighborhoods indicates many of the areas of new development during the past 10 years. To investigate urban LULC and its environmental impact, percent ISA and its distribution were utilized to characterize built-up surfaces in the 16 sub-drainage basins within Hillsborough County. Fig. 5a presents the spatial 2000 1800 1600 1400 1200 km2

968

1000 800 600 400 200 0 1991

1995

2000

2002

Year

Fig. 3. Change in spatial extent of urban land use from 1991 to 2002 in the Tampa Bay watershed.

Table 1 Average annual non-point source loadings (tons/year) in three sub-drainage basins Drainage basin

BOD5

TSS

TKN

TN

TDP

Oil_Grease

NO3+NO2

Pb

Zn

Hillsborough River Alafia River Little manatee

867 3494 2082

2085 5067 2521

163 524 291

211 629 392

59 123 253

75 191 85

49 99 100

2.5 7.9 2.6

3.3 6.4 3.5

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

Fig. 4. Percent ISA distribution in the Tampa Bay watershed, 2002. High percentage imperviousness is found in the major cities and along main transportation systems.

969

drainage basin, different search radii were used to calculate population density. The 5-km search radius used to map population density for the Tampa Bay watershed proved too coarse for the county-level calculation. Therefore, smaller search radii having narrower search zones were used to produce finer resolution densities. Fig. 5b shows the population density derived from a 1-km search radius within the county watershed. To determine the relationship between population density and urban land use density, percent imperviousness and associated population densities calculated from 500-m and 1-km search radii were compared for each sub-drainage basin (Fig. 6). Correlation results suggest that linear relationships exist between ISA density and population densities obtained from different search sizes. The linear regression for ISA and population densities obtained from 1-km and 500-m search sizes resulted in corresponding r2 values of 0.80 and 0.69, respectively. Further analysis using population densities calculated for smaller search zones suggested that population distribution did not show a closer correlation with patterns of local urban land use. The spatial distribution patterns of population density with a 1-km search zone proved closer to regional urban land use patterns. The linear relationship between ISA density and population density in Hillsborough County suggests that a transition of land use from urban centers to suburbia is associated with a gradual population decrease. Generally, high-density ISAs such as commercial and business complexes, and industrial and utility zones are not composed of large residences. Medium- to high-density ISAs that consist of single or multi-unit housing and apartment units contain large populations. 3.3. Characteristics of water quality parameters, population distribution, and associated urban land use

distribution of percent ISA in 2002. Areas of high-percent imperviousness are primarily found on the north-western side of the county. Several high-percent ISAs are also seen on the south coastal area. The proportion of built-up surface, or ISA density, for each sub-drainage basin was determining by taking the total imperviousness area defined by the 10% imperviousness threshold and dividing by the total area of each basin. Table 2 presents the total area and associated percent ISA coverage for the five subbasins with the highest ISA densities and the three largest sub-basins. Approximately 76.46% of Lower Sweet Basin is comprised of impervious surface, making it the most developed sub-drainage basin in the county. At 76.06% imperviousness, Duck Pond is a close second in urban land cover. The three largest sub-drainage basins—Alafia River, Hillsborough River, and Little Manatee River—all have ISA densities less than 20%. 3.2. Urban land use and population density ISA density generally corresponds to urban land use intensity. To examine the relationship between proportion of impervious surface and population density in each sub-

An overview of annual average non-point source pollutant loadings, including nutrients, heavy metals, and oil and grease, for the 16 sub-drainage basins of Hillsborough County is provided in Table 3. Among the amounts of nine non-point source pollutant loadings, the largest pollutant value is for TSS at 156.52 ton/km2/year. BOD5 has the second largest loading amount of 36.73 ton/ km2/year. TN also shows a relatively large amount of 11.53 ton/km2/year loaded to the watershed. Pb and Zn metals have similar loading values. The spatial distributions of the annual loadings for BOD5, TDP, TSS, oil and grease, TN, TKN, NO3+NO2, Pb, and Zn are displayed in Fig. 7. BOD5 and TDP have similar spatial distribution patterns. Both have larger values in the southern part than those in the northern portion of the county watershed. Much of the eastern and southern portions of Hillsborough County are predominantly agriculture land, e.g., Bullfrog and Little Manatee sub-basins are 48% and 49% agriculture, respectively, and are frequently fertilized and contain higher BOD5 and TDP values. In contrast, Sweet Water, Lower Sweet Water, and Rocky Brushy sub-basins in the northwest part of the county are composed of more

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

970

Fig. 5. Impervious surface (a) and population density (people/km2) distributions (b) within the Hillsborough County watershed.

Table 2 Total area and ISA coverage density for three largest and five highest ISA density basins Drainage basin

Lower sweet

Duck pond

Silver twin

Curiosity

East lake

Hillsborough

Alafia

Little manatee

Total area (km2) ISA density (%)

27.24 76.46

18.07 76.06

3.48 69.1

8.56 64.82

20.53 57.88

970.52 19.98

1084.21 16.27

630.25 14.16

than 58% urban lands. High loadings for TSS and oil and grease also appear in the northwest corner of the watershed where population densities are relatively high. High TSS also follows major highways, including Interstate-4 and Interstate-75, in the area. The spatial distribution patterns of TN and TKN are similar. One high TKN area appears on the eastern edge of the Alafia River sub-basin, although the TN value is not large. Both TN and TKN show larger values in the Sweet Water basin and part of the Rocky Brushy basin, in which only 2% and 9% of the lands are for agriculture, respectively. The magnitude of NO3+NO2 is relatively larger in the northwest corner and southern part of the watershed. High NO3+NO2 in these areas are associated with a greater amount of residential activity. Loadings of metals (Pb and Zn) are seen along major transportation roads, especially on the eastern coast of the bay along Interstate 75. All major non-point pollutant sources exhibit strong regional characteristics from their spatial distribution patterns in the watershed.

To better understand the correlations of these loadings with population and built-up surfaces, polynomial regression analyses were made using pollutant sources, impervious coverage, and population data for each sub-drainage basin within Hillsborough County. Fig. 8 shows the polynomial models with total ISA in each sub-drainage basin as the independent variable and total loadings for nine different parameters as dependent variables. The exponential polynomial regression models and corresponding r2 values indicate strong correlations between ISA and most chemical and biological loadings. Most response functions show large increases when imperviousness exceeds 150 km2. The best regression response functions for ISA and metal loadings are also exponentials with relatively lower r2 values compared with regression results from chemical and biological sources except Zn. BOD, TSS, TDP, and NO3_NO2 have similar correlation patterns. The response functions of TKN, TN, and Oil_Grease are also very similar. The loading of Zn has

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

the highest r2 value. These relationships suggest that nonpoint chemical, biological, and metal loadings tend to be amplified as urbanization intensifies. The correlation patterns between population density and annual non-point source loading intensities were also checked for four parameters. Fig. 9 presents the five best models for the annual areal loading rates of Zn, Cu, Oil_Grease, TSS, and NO3_NO2 associated with population density calculated with a 1-km search radius. Zn and Oil_Grease have the highest r2 value of 0.85 and NO3_NO2 has the lowest r2 value of 0.65 in linear regression models. Cu and TSS have r2 values of 0.75 and 0.82, respectively. High metal, chemical, and biological loading rates are associated with areas of relatively high population density. Other source loading rates do not show a strong correlation with population density.

a 1800 1600

y= 19.623x -194.62 r2 =0.80

Population Density

1400 1200 1000 800 600 400 200 0 10

20

30

40

50

60

70

80

ISA Density

b

1800 1600

y =21.134x -235.39 r2 = 0.69

Population Density

1400 1200 1000 800

971

To examine correlations of ISA density with pollutant loading rates, regression analyses were made for the same five loading parameters versus associated percent of imperviousness cover in each basin. All linear regression models for these five pollutant loading rates and percent ISA are displayed in Fig. 10. The r2 values of linear regression models for all five loadings are lower than the corresponding r2 values for loading rates and population density analyses. Both ISA and population densities are used as the measures of the intensity of urbanization. However, these two variables tend to be different in nature. ISA density is derived from satellite imagery having 30-m spatial resolution and depicts land use patterns at the level of land parcels. Population density is obtained from the census data at sub-district level at a much coarser resolution equivalent to tens of land parcels. Therefore, population density represents the intensity of urbanization from a regional perspective or the background condition for surface pollutant loading. Population density in a 1 km radius is a better predictor than ISA density for pollutant loading estimation. On the other hand, water quality in a watershed is influenced by both regional land use patterns and local conditions. This emphasizes the need to manage water and surrounding environmental conditions through local and regional planning measures, including the removal of pollution sources from waterways, maintaining greenspace along rivers and lake corridors, and building up necessary sewage system and sewage-treatment facilities. Based on analysis of the linear regression results we conclude that the extent of imperviousness exerts considerable influence on total pollutant loadings in Hillsborough County. The percent imperviousness was a good predictor for estimating annual non-point source pollutant loadings. However, population density is a better predicator for the pollutant-loading rate than ISA density, especially for sources of Zn and Oil_Grease. Urbanization and its associated population growth contribute to the increase of non-point pollutant loadings in the watershed and certainly impact the quality of aquatic and terrestrial ecosystems.

600

4. Discussion

400 200 0

10

20

30

40

50

60

70

80

ISA Density

Fig. 6. Population density derived from a 1-km search radius (Fig. 6a) and from a 500-m search radius (Fig. 6b) versus percent ISA coverage.

Percent impervious surface estimated from satellite remote sensing data can be utilized to assess the spatial extent of urban land use, as well as urban development density. Associated with urbanization in the Tampa Bay watershed are increases of impervious cover and expansion of the population. Results of correlation analysis for

Table 3 Average annual non-point source pollutant loadings by type for all sixteen sub-drainage basins

Total loading (ton/km2/year)

BOD5

TSS

TKN

TN

TDP

NO3+NO2

Oil_Grease

Pb

Zn

29.15

141.25

7.40

9.96

2.20

2.50

4.11

0.22

0.23

ARTICLE IN PRESS 972

G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

Fig. 7. Annual BOD5, TDP, TSS, Oil_Grease, TN, TKN, NO3+NO2, Pb, and Zn loading (ton/year) in the 16 sub-drainage basins, Hillsborough County.

Hillsborough County indicate that high population density centers are associated with areas of high imperviousness along the eastern shore of Tampa Bay, the cities of Tampa and St. Petersburg, and major highways. Population and

ISA densities at sub-drainage levels are statistically correlated in a linear relationship. As population and urban LULC growth continue in the Tampa Bay watershed, direct impacts on the aquatic and

ARTICLE IN PRESS 7000

y = 36.118e0.0251x r2 = 0.65

5000

50

100

150

3000

1000

100 0 50

0.0192x

400 200

r2 = 0.63

150 100 50

100

0 0

50

100

150

200

0

50

ISA

150

200

r2 = 0.54

10 Pb

100

8 6 4

50

2

0

0 100

150

200

0

50

ISA

100

150

200

r2 = 0.64

50

100

150

200

150

200

ISA

y = 6.2895e0.0193x

12

150

50

100

y = 3.948e0.019x

0

14

r2 = 0.71

0

180 160 140 120 100 80 60 40 20 0

ISA

y = 6.4601e0.0184x

200

100

Zn

250

50

ISA

NO3_NO2

r2 = 0.72 TDP

TN

0

200

y = 3.1723e0.0217x

200

300

Oil Grease

150

ISA

500

0

100

250

600

300 200

0

800 y = 16.155e

400

2000

ISA

700

r2 = 0.70

500

4000

200

y = 11.214e0.0208x

600

r2 = 0.64

0 0

973

700 y = 224.54e0.0174x

6000

TKN

5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0

TSS

BOD

G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

150

200

ISA

10 9 8 7 6 5 4 3 2 1 0

y = 0.3087e0.0225x r2 = 0.80

0

50

100 ISA

Fig. 8. Polynomial regressions of selected pollutant loadings and associated ISA in each sub-drainage basins. ISA is in km2 and pollutant parameters are in tons/year.

terrestrial ecosystems are inevitable. Possible results include water quality degradation from non-point source pollutants collected and accumulated on impervious surfaces and delivered to water bodies via stormwater runoff. Polynomial regression analyses for the major non-point source pollutant loadings and spatial extent of ISA in Hillsborough County suggest that annual amounts of chemical, biological, and heavy metal loadings are strongly correlated with the aerial extent of impervious surface in sub-drainage basins. The correlation patterns for different chemical, biological, and heavy metal loadings are similar. Heavy metals, such as Zn and Pb loadings, are more closely tied to major highways, whereas TSS, Oil and Grease, and NO3_NO2 loadings are concentrated in urban areas. In contrast, the BOD5 and TDP loadings are apparently from agricultural related activities.

5. Conclusions The intensity and spatial extent of urban development in the Tampa Bay area were characterized using satellite remote sensing data to map anthropogenic impervious surfaces. The study found that the spatial extent of ISA and population densities were closely correlated spatially, and therefore, sub-pixel ISA estimated from satellite remote sensing data provides an excellent estimate for the intensity of urbanization including both land use intensity and population distribution. The spatial distributions of non-point source pollutant loadings in Hillsborough County appear to be related to land use characteristics. To inspect the urbanization impact on water quality, regression analysis was implemented. The correlations for non-point source pollutants that are usually expressed as water quality parameters, population density, and ISA

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

974

30

40 y = 0.0204x + 1.9156

35 30

r2 = 0.75

20 Cu

25 Zn

y = 0.0112x + 2.7478

25

r2 = 0.85

20 15

15 10

10

5

5

0

0 0

500

1000

1500

0

500

Population Density

1000

1500

Population Density

700

25000 y = 0.3241x + 54.79

600

y = 10.994x + 1981.2

r2 = 0.85

r2 = 0.82

20000

400

TSS

Oil Grease

500

300

15000 10000

200

5000

100

0

0 0

500

1000

1500

2000

0

Population Density

500

1000

1500

2000

Population Density 460

y = 0.1511x + 62.037 r2 = 0.65

410 NO3_NO2

360 310 260 210 160 110 60 10 0

500 1000 Population Density

1500

Fig. 9. Polynomial regressions for pollutant loading rates (kg/km2/year) and population density (person/km2) obtained from 1-km search zone.

density were compared. Analysis results indicated that spatial extents of ISA in drainage basins were closely associated with annual non-point source pollutant loadings. However, from inspection of annual areal average loading rates, population density appeared to be a better indicator than ISA density for most non-point source pollutants. The regression analysis modeling technique used for Hillsborough County is well suited to broader scale water quality investigations. Under current urban development

conditions in the Tampa Bay region, the correlation patterns obtained from this study are applicable for estimating the distribution of non-point source pollutant loadings for the entire Tampa Bay watershed. This could be accomplished with water quality data from several additional monitoring sites for use with new satellite imagery and derived ISA information. The results of this study demonstrate a useful tool for monitoring urban development and non-point source surface water loadings in the Tampa Bay watershed. It can be used to help urban

ARTICLE IN PRESS G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976 40

30

y = 0.4196x - 2.8776

35

y = 0.212x + 0.9196

r2 = 0.74

r2 = 0.55

25

30

20

25 20

15

Cu

Zn

975

15

10

10 5

5

0

0 0

20

40

60

80

100

0

20

ISA Density

40

60

80

100

ISA Density 25000

700

y = 223.13x - 466.39

y = 6.2325x -3.0067

600

r2 = 0.70

20000

r2 = 0.65

15000

400

TSS

Oil Grease

500

300

10000

200 5000

100

0

0 0

20

40

60

80

100

0

20

ISA Density

40

60

80

100

ISA Density

460

y = 2.7353x + 42.175 r2 = 0.44

410

NO3_NO2

360 310 260 210 160 110 60 10

0

20

40

60

80

100

ISA Density

Fig. 10. Polynomial regressions for pollutant loading rates (kg/km2/year) and ISA density (ISA/total area).

planners, coastal managers, and decision makers to assess potential long-term influences of urbanization so that appropriate policies can be implemented to mitigate or minimize future impacts. Acknowledgments This research was prepared under contract number 03CRCN0001 between SAIC and the US Geological Survey. We would like to thank two anonymous reviewers

for their constructive comments and suggestions. We also wish to thank Mr. Cory McMahon for preprocessing Landsat imagery and producing part of the impervious surface data. References Arnold Jr., C.A., Gibbons, C.J., 1996. Impervious surface coverage: the emergence of a key urban environmental indicator. Journal of the American Planning Association 62 (2), 243–258.

ARTICLE IN PRESS 976

G. Xian et al. / Journal of Environmental Management 85 (2007) 965–976

Berry, B., 1990. Urbanization. In: Clark, W.C., Kates, R.W., Richards, J.F., Mathews, J.T., Meyer, W.B., Turner II, B.L. (Eds.), The Earth as Transformed by Human Action. Cambridge University Press, Cambridge, England, pp. 103–119. Freedman, B., 1995. Environmental Ecology: The Ecological Effects of Pollution, Disturbance, and other Stresses, second ed. Academic Press, San Diego, CA. Gillies, R.R., Box, J.B., Symanzik, J., Rodemaker, E.J., 2003. Effects of urbanization on the aquatic fauna of the Line Creek watershed, Atlanta—a satellite perspective. Remote Sensing of Environment 86, 411–422. Gove, N.E., Edwards, R.T., Conquest, L.L., 2001. Effects of scale on land use and water quality relationships: A longitudinal basin-wide perspective. Journal of the American Water Resource Association 37 (6), 1721–1734. Hillsborough County, 1999. Watershed Management Plan. Available at URL: /http://www.hillsborough.wateratlas.usf.eduS. Last accessed on August 2, 2005. Harbor, J., 1994. A practical method for estimating the impact of land use change on surface runoff, groundwater recharge and wetland hydrology. Journal of the American Planning Association 60, 91–104. Jeng, H.A., Englande, A.J., Bakeer, R.M., Bradford, H., 2005. Impact of urban stormwater runoff on estuarine environment quality. Estuarine, Coastal and Shelf Science 63, 513–526. Kalnay, E., Cai, M., 2003. Impact of urbanization and land-use change on climate. Nature 423 (29), 528–531. Milesi, C., Elvidge, C.D., Nemani, R.R., Running, S.W., 2003. Assessing the impact of urban land development on net primary productivity in the southeastern United States. Remote Sensing of Environment 86, 401–410. Miller, D., 1981. Energy at the Surface of the Earth–An Introduction to the Energetics of Ecosystems. Academic Press, San Diego, CA. Mitchell, G., 2005. Mapping hazard from urban non-point pollution: A screening model to support sustainable urban drainage planning. Journal of Environmental Management 74, 1–9. Moscrip, A.L., Montgomery, D.R., 1997. Urbanization, flood, frequency, and salmon abundance in Puget Lowland Streams. Journal of the American Water Resources Association 33 (6), 1289–1297.

Oke, T.R., 1989. The micrometeorology of the urban forest. Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences 324, 335–351. Pielke Sr., R.A., Walko, R.L., Steyaert, L.T., Vidale, P.L., Liston, G.E., Lyons, W.A., Chase, T.N., 1999. The influence of anthropogenic landscape changes on weather in south Florida. Monthly Weather Review 127, 1663–1673. Schueler, T.R., 1994. The importance of imperviousness. Watershed Protection Techniques 1 (3), 100–111. Slonecker, E.T., Jennings, D.B., Garofalo, D., 2001. Remote sensing of impervious surface: a review. Remote Sensing Review 20, 227–235. Tang, Z., Engel, B.A., Pijanowski, B.C., Lim, K.J., 2005. Forecasting land use change and its environmental impact at a watershed scale. Journal of Environmental Management 76, 35–45. US Department of Commerce, Bureau of the Census, 2001. Census of population and housing, 2000: Profiles of general demographic characteristics. US Department of Commerce, Bureau of the Census, Washington, DC. USEPA, 2001. Our Built and Natural Environments: A Technical Review of the Interactions between Land Use, Transportation, and Environmental Quality: US Environmental Protection Agency. Development, Community, and Environment, Washington, DC. Xian, G., 2007. Assessing urban growth with sub-pixel impervious surface coverage. In: Weng, Q., Quattrochi, D. (Eds.), Urban Remote Sensing. CRC Press/Taylor & Francis Group, Chapter 9, pp. 179–199. Xian, G., Crane, M., 2005. Assessments of urban growth in the Tampa Bay watershed using remote sensing data. Remote Sensing of Environment 97 (2), 203–215. Xian, G., Crane, M., Steinwand, D., 2005. Dynamic modeling of Tampa Bay urban development using parallel computing. Computers & Geosciences 31, 920–928. Yang, L., Xian, G., Klaver, J.M., Deal, B., 2003. Urban land-cover change detection through sub-pixel imperviousness mapping using remotely sensed data. Photogrammetric Engineering & Remote Sensing 69 (9), 1003–1010. Yin, Z.Y., Walcott, S., Kaplan, B., Cao, J., Lin, W., Chen, M., Liu, D., Ning, Y., 2005. An analysis of the relationship between spatial patterns of water quality and urban development in Shanghai, China. Computer, Environment and Urban Systems 29, 197–221.