Mapping imperviousness using TM data in water resources reservation area of Shanghai

The rapid growth of impervious land covers within urbanizing regions holds many negative implications for environmental quality. The study region is the drinking water conservation areas of Shanghai, which is very important to the megalopolis. Mapping of imperviousness has shown important potentials to acquire such information in great spatial detail but the actual mapping process has been challenged by the heterogeneity of urban and suburb environment and the spatial and spectra capabilities of the sensor. This study focused on mapping the imperviousness fraction using linear spectral unmixing in the area from Landsat satellite remote sensing data. Development of high-quality fraction images depends greatly on the selection of suitable endmembers. A multi-endmemer linear spectral unmixing were evaculated. In the approach, each of the class hold multi-image-endmember representing the heterogeneity of them. The best fraction images were chosen to determine the imperviousness. An unconstrained least-squares solution was used to unmix the MNF components into fraction images. The multi-endmember linear spectral unmixing is then used to map imperviousness fraction for the years of 1987, 1997 and 2006 in upper region of Huangpu River, respectively. In the water resources reservation of Shanghai, the impervious surface area increases approximately 3 times from 1987 to 2002.