COMPARISON OF RESULTS OF THE SUPERVISED CLASSIFICATION OF SOIL COVERAGE BY MAXIMUM LIKELIHOOD OF IMAGES LANDSAT-5 TM AND ALOS AVNIR-2

Abstract

doi: 10.12957/geouerj.2016.19099

Changes in land cover, especially in the tropics, has been occurring at an accelerated rate and with it the largest and most diverse rainforest in the world, the Amazon jungle or the Amazon Biome, also suffers from this process. At present study were compared results obtained from characterization of land cover by supervised classification by maximum likelihood using LANDSAT 5 TM and ALOS-2 AVNIR images. The study area chosen for this evaluation was the area of the Settlement Project Carlinda in Mato Grosso State, located on the so-called Legal Amazon. From the results it was observed that areas of each classes in the two images have similar values. The small variations should be presented due to the difference between the spatial resolutions of the images and the difference of the spectral bands regions involved in the analysis, especially the Green band. Analyzing the classes according spatial representation, in both images, the water bodies class was the smaller, followed by preserved vegetation class and wetland vegetation class. The class of anthropic vegetation / exposed soil was the most represented class in both classifications. Regarding the study area, it was observed that in the region there is a predominance of areas used in various human activities and gradual reduction of areas with preserved vegetation. This fact is diametrically opposite to what one would expect from an area in the Amazon, where, by the law need 80% of its territory be preserved.