Synergistic use of Landsat Multispectral Scanner with GIRAS land-cover data to retrieve impervious surface area for the Potomac River Basin in 1975
For many studies of urbanization, particularly those covering broad spatial extents and thus requiring the perspective afforded by satellite remote sensing, the longest time series of impervious surface area (ISA) initiate with the availability of Landsat Thematic Mapper (TM) data in the mid 1980s. Earlier generation remote sensing data, such as Landsat Multispectral Scanner (MSS) do not generally lead to robust representations of impervious surface area due to poor spectral differentiation between impervious and non-impervious surfaces lacking vegetation. In the 1970s, however, the USGS completed a comprehensive mapping of land cover based on aerial photography called the Geographic Information Retrieval and Analysis System (GIRAS). In this paper we describe a methodology for retrieving ISA estimates through the fusion of GIRAS land cover with historic and contemporary remote sensing data, resulting in a data set that is temporally comparable to more modern data sets (e.g. the National Land Cover Database (NLCD) ISA product). We calculate correlation coefficients between our data set and digitized aerial photography at 4 spatial resolutions, exhibiting a maximum correlation of 0.65 at 120-m and 240-rn pixel sizes. Our highest resolution product (30-m) exhibited 94% accuracy and a kappa coefficient of 36% in rural areas, and 90.7% accuracy and a kappa coefficient of 59.6% in suburban areas. Further, we show that trends in urban area through time derived from these data exhibit accelerating rates of development throughout the Potomac River basin 1975-2001, including increased development in riparian buffer zones and on steeper topography. These results are applicable wherever low-resolution vector-based data sets of urban area are available or could be generated for the pre-Landsat TM era. (C) 2010 Elsevier Inc. All rights reserved.
Keywords: Impervious surface area, GIRAS, Landsat MSS, Spectral mixture analysis, Land-use modeling