Development of New Indices for Extraction of Built-Up Area & Bare Soil from Landsat Data

This paper presents development of new indices for the identification of built-up area and bare soil from medium resolution remote sensing data, extraction of these two spectrally confusing classes has always been a problem. For nearly three decades extensive research efforts have undergone for the development of various techniques including a number of indices to extract land cover features especially if suitable thresholds are used. These indices include for example NDVI (Normalized Difference Vegetation Index), NBI (Normalized Built up area Index), NDISI (Normalized Difference Impervious Surface Index) & NDWI (Normalized Difference Water Index). All these indices use various combinations of bands in order to extract a particular land cover, and their performance depends upon spectral response of the features. As spectral response of land cover features varies from region to region due to climatic & topographic changes so are their spectral curves. Due to this reason, indices developed for one area can’t be effectively applicable to another area. In this research, a few new indices have been proposed for the extraction of bare soil and built-up area using medium resolution satellite imagery (LandSat 30m). A comparison has been made among newly developed indices and previously used indices. The results show that newly developed indices for built-up area and soil increases accuracy up to 10-13% as compared to previously developed indices.