bers/ccd images evaluation for mapping quartzipsamment bare soils in the southwest

Resumen

The main goal of this work was to assess the performance of different Image enhancement techniques applied to two CBERS II CCD images, acquired during the dry and rainy periods, regarding the discrimination of Quartzipsamment bare soils (sandy spots), in the Ribeirão Sujo watershed, situated in the municipality of Serranópolis/GO. In addition, a literature review, concerning the spectral behavior of sandy soils and degraded areas, was carried out. The three image processing approaches evaluated were: spectral linear mixture modelling, vegetation index (NDVI), and principal component analysis. It was verified that the images of the rainy period favored a better discrimination of the sandy soils in relation to the other land cover types and that supervised classification, based on the CBERS CCD bands 2, 3, and 4 was the best image enhancement approach.