Urban Land-Cover Change Analysis in Central Puget Sound

A methodology was developed to interpret and assess land cover change between 1991 and 1999 in Central Puget Sound, Washington at several scales (landscape, sub-basins, and 90 m grid window) relevant to regional and local decision makers. Land cover data are derived from USGS Landsat (Thematic Mapper and Enhanced Thematic Mapper ) images of Central Puget Sound. Landsat data were registered, intercalibrated, and corrected for atmosphere and topography to ensure accuracy of land cover change assessment. We apply a hybrid classification method to each image to address the spectral heterogeneity of urbanizing regions. The method combines a supervised classification approach with a spectral unmixing approach to produce seven classes: 75 percent impervious, 15 to 75 percent impervious, forest, grass, clear cut, bare soil, and water. Land cover change is identified using the direct spatial comparison of classified images derived independently for each time period. We assess that the overall accuracy of each classified image was 91 percent for 1991 and 88 percent for 1999 respectively, which produces an accuracy of 85 percent for the change analysis. Our results show that urban growth over the last decade has produced an overall 6.7 percent increase in paved area. Introduction Remote sensing is a powerful tool for monitoring rapid changes in the landscape resulting from urban development. Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM ) images provide moderate-resolution data sets over large geographic regions. These data are critical to both the natural and social sciences for quantifying urban landscape patterns and testing formal hypotheses about the relationships between urban patterns and various biophysical and ecological processes. Interpretation and analysis of urban landscapes from remote sensing, however, present unique challenges due to the spatial-temporal characteristics of urban land cover change which amplify the spectral heterogeneity of urban surfaces and make it extremely difficult to identify the source of observed change in observed reflectance. The spectral heterogeneity of urban surface areas together with the spatial scale of urban phenomena cause significant subpixel mixing (Foody, 2000; Small, 2002). This is due to the complex mixtures of urban surface materials (i.e., concrete, wood, tiles, asphalt, metal, sand, and stone) and vegetation (grass, scrubs, shrubs, trees, and leaves) within the Landsat Urban Land-Cover Change Analysis in Central Puget Sound Marina Alberti, Robin Weeks, and Stefan Coe pixel resolution. Small (2002) indicates that while urban land cover classes can be distinguished when they occur in homogeneous regions larger than the spatial resolution of the sensor, this is rarely the case for moderate-resolution such as Landsat. The greatest challenge for urban land cover classification is to accurately determine the relative contribution of various materials that make up urban surface reflectance. An additional challenge is to compensate for temporal variability and changes in surface reflectance (such as seasonal variations) that are unrelated to land cover change (Hornstra, et al., 1999; Small, 2002). Previous studies have proposed various strategies to improve classification of urban landscapes. Ancillary data such as population, zoning, housing density and other information are frequently used in preand post-classification procedures (Harris and Ventura, 1995; Mesev, 1998; Vogelmann, et al., 1998; Stuckens, et al., 2000). Since urban areas typically have significant texture resulting from buildings and roads, analysts also use texture analysis in addition to spectral features to characterize urban land cover types (Gong and Howarth, 1990; Berberoglu, et al., 2000; Stuckens, et al., 2000; Stefanov, et al., 2001). More recently, analysts have used data mining techniques, in particular artificial neural networks (Schalkoff, 1992; Paola and Schowengerdt, 1995; Berberoglu, et al., 2000; Foody, 2001), to detect complex spectral urban patterns. Other analysts have emphasized classification strategies using multitemporal images (Martin and Howarth, 1989). An alternative method for detecting urban land cover materials is spectral unmixing (Small, et al., 2002). The Spectral Mixture Analysis (SMA) methodology (Adams, et al., 1986) assumes that radiance from a heterogeneous surface mixes linearly within the Instantaneous Field of View (IFOV) according to the aerial percentages of each pure material present. According to this methodology, if a limited number of distinct endmembers are known apriori, it is possible to define a mixing space within which mixed pixels can be unmixed into percentages of constituent materials using a system of linear equations (Adams, et al., 1986). A spectral unmixing approach is desirable in areas where features on the landscape are smaller than the spatial resolution of the satellite instrument (Small, 2002). This is especially true in urban areas where pixels are highly mixed. While spectral unmixing cannot reliably identify exact urban materials, it can determine relative amounts of vegetation and urban surfaces providing a good estimate for urban intensity. In this study we integrate multiple pre-processing, land cover classification, and change analysis techniques to develop a methodology to interpret and assess land cover change between 1991 and 1999 in Central Puget Sound at P H OTO G R A M M E T R I C E N G I N E E R I N G & R E M OT E S E N S I N G September 2004 1 0 4 3 Marina Alberti is with the Department of Urban Design and Planning, University of Washington, Box 355740, Seattle, WA 98105 ([email protected]). Robin Weeks is with the Department of Geological Sciences, University of Washington, Seattle, WA 98105. Stefan Coe is with the Urban Ecology Research Laboratory, University of Washington, Seattle, WA 98105. Photogrammetric Engineering & Remote Sensing Vol. 70, No. 9, September 2004, pp. 1043–1052. 0099-1112/04/7009–1043/$3.00/0 © 2004 American Society for Photogrammetry and Remote Sensing 99-028.qxd 8/5/04 18:47 Page 1043