Uso de geotecnologias no mapeamento do uso da terra e estudo de fitofisionomias em áreas úmidas na região do Médio Araguaia

Abstract

The main objective of this chapter was to use object-oriented classification of Landsat 5-TM images in the characterization of classes of use and coverage in an area located between the Sao Felix do Araguaia-MT city and Bananal Island-TO. The scene 223/068 acquired on September 5, 2010 was submitted to the stages of pre-processing: radiometric, atmospheric and geometric. Then, NDVI and MNDWI images were produced using spectral band mathematical operations, and used as inputs for the land cover classification. For the segmentation we used the parameters of scale 250, 200, 150, 100, 50 and 30, and the algorithms assign class and nearest neighbor, as well as descriptors of mean, area and relation to border. A confusion matrix was used to evaluate the accuracy of the classification using the Kappa index of agreement. The overall mapping accuracy was 83% with a Kappa coefficient of 0.72 and therefore suitable for the research objective.