Análise multitemporal do uso da terra e da cobertura vegetal na região de Alfenas - Sul de Minas Gerais, utilizando imagens Landsat-5 TM

Abstract

Several studies have shown that native forest fragments sur rounded by different landuse matrixes undergo different ecological pressures on fauna and flora. In light this, we studied the land-use and land-cover changes in the region of Alfenas, southern Minas Gerais state, aiming the conservation of forest fragments. Landsat-5 Thematic Mapper (TM) images, bands 1 to 5 and 7, from 1987 and 2011, were used. Image classification was achieved using the Geographic Data Mining Analyst (GeoDMA), a toolbox specially addressed for spatial data mining. To carry out this investigation, the following procedures were adopted: image segmentation, spectral and spatial features extraction, sampling, decision tree generation, classification, error edition, and analysis of land-use and land-cover changes by using a change detection matrix. The results showed the importance of complementarity of information available in each band for classifying different land use and land cover types. As for land use change, an increase of sugar cane, coffee and bare soil were registered on previously areas used for pasture. Such information are important, since they may support interpretations of ecological dynamics of forest fragments.