Imperviousness Indexes Performance Evaluation for Mapping Urban Areas Using Remote Sensing Data

Urban analysis based on remotely sensed data is commonly performed using high-resolution optical images, acquired with new generation high resolution satellites. This assures a good detail level for cartographic issues, but sometimes these data show not enough spectral richness, preventing the scientific and commercial user from deriving effective classifications or environmental phenomena mapping. Mid-resolution multispectral data, ranging from 20 meters to 100 meters ground pixel size, has proven very effective in environmental analysis, also in areas where the anthropic influence on the natural environment is heavy, as in the case of urban and sub-urban areas. The main topic of this paper is to present a comparison of spectral indexes in order to derive urban areas map with Landsat TM and ETM+ scenes; in particular, an evaluation of performances, in terms of class separability between impervious and non impervious surfaces, was done using vegetation and urbanization indexes well known in the scientific literature (NDVI, SAVI, UI), together with a new imperviousness index specifically introduced and tested for urban mapping purposes: the Soil and Vegetation Index (SVI). Best performances were demonstrated to be gained with this newly introduced index, both for laboratory spectra and satellite data, thus permitting a good mapping of urban features, useful for urban areas analysis and change detection.