Estimating Impervious Surfaces Area of Urban Watersheds Using ASTER Data

Abstract

In this paper we present and compare two approaches in estimating impervious surfaces area for an urban watershed in northern New Jersey, United States using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) imagery. The first approach is to use the spectral mixture analysis (SMA) to separate image pixels into a linear combination of three typical urban land covers: vegetation, impervious surface and soil. The other is to use the normalized difference vegetation index (NDVI) to estimate imperviousness from the same imagery. The accuracies of the estimated imperviousness were assessed using a high-resolution color-infrared orthophoto. In total, 100 polygons with areas between 3 and 6 hectares were randomly selected from five distinct land use/cover categories and the percentage of impervious surface of each polygon was digitized and calculated. The results showed that nearly 90 percent of the variation in actual impervious surfaces in this watershed can be explained by the estimated impervious surfaces by a linear regression model (R 2 = 0.898). The NDVI approach is recommended for urban environments with small proportion of barren soils for its simplicity, while the SMA is suitable for urban environments with approximately equally-distributed vegetation, impervious surfaces and barren soil.