Mapeamento do uso da terra e da cobertura vegetal da sub-bacia hidrográfica do rio são tomé, sul de Minas Gerais, por mineração de dados utilizando imagens IRS-P6/LISS III

Abstract

The objective of this study was to perform a classification of land use and land cover of the São Tomé river watershed located in the south of the Minas Gerais State by using multispectral imagery collected by the Linear Imaging Self-Scanner onboard the Indian Resource Satellite and data mining techniques. The study area lies in a region with great tradition of coffee cultivation whose properties are predominantly of small and medium sizes. To carry out the image classification, we used the Geographic Data Mining Analyst (GeoDMA), a toolbox that has algorithms for segmentation, feature extraction, feature selection and classification. The results showed the great potential of the data mining techniques for remote sensing imagery classification. Some advantages encompass the possibility of incorporate in the classification process a great variety of informations that includes spectral, spatial and topographic attributes, the generation of high classification accuracy, and the simplicity to interpret their classification structure.