Avaliação da classificação da cobertura do solo no bioma caatinga por análise multitemporal e unitemporal.

Abstract

The caatinga is an important Brazilian biome, which covers 11% of its territory. It is characterized by a great diversity of plant species, which present great loss of foliage during the dry season, changing its morphology in a short period of time. The classification of the land cover of this biome is of great importance, since it can help in the study of hydrographic basins, environmental analyzes and observation of the alteration in the soil cover, since this environment has undergone severe alterations throughout the years. Some mechanisms that can be used to carry out its monitoring are the remote sensing and geoprocessing techniques, which, based on machine learning methods and supervised data, make it possible to obtain a classification of the land use cover for this biome. This study was applied in the area of the Sucuru River Basin, in Cariri, Brazil, where the classification of the land cover of this region was made by means of the creation of a computational model based on the Random Forest method, using the R. software. Images of the Landsat 8 and Landsat 7 satellites provided by the USGS, and the vegetation indices EVI, NDVI, SAVI, STI. The 24month NDVI time series proved to be more efficient than the other series for the other indices for land cover classification. In addition, it was more efficient than the traditional instantaneous classification (performing with the bands), thus allowing the identification of a standard for each type of soil cover. The percentage between the temporal series NDVI and classification instantaneous was + 241,03% for the ASMD Arbórea Subarbórea Muito Densa class, which was already classified as an ATSA Arbustiva subabórea Aberta 44.32%. These deficits affect the performance of hydrological and climatological models that use land cover data. The maps covered with soil cover are important to evaluate the changes in the soil area throughout the process of occupancy of the Caatinga areas.