Monitoramento de mudanças na cobertura da terra na região da Floresta Nacional do Tapajós, Amazônia Brasileira, utilizando segmentação de imagens multitemporais

The main purpose of this paper is to present a semi-automated procedure for monitoring land cover changes in Brazilian Amazonia. For this, the region around the Tapajós National Forest in Pará State, Brazilian Amazonia was selected. The study area comprises approximately 10,650 km2 covered by the Landsat Thematic Mapper (TM) images acquired on 22 August 1989; 02 August 1999; and 28 July 2009. According to Radambrasil vegetation map, the study area is primarily covered by dense tropical rain forest (Floresta Ombrófila Densa) with a high number of emergent tree species. Firstly, the image segmentation technique was applied to multitemporal Landsat Thematic Mapper acquired over Tapajos National Forest region to generate the georeferenced polygons. Then the vegetation, soil and shade fraction images derived from TM images using Linear Spectral Mixing Model were used to classify these polygons in the land cover classes for each image date. Finally the Geographic Information System tools are used to map the changes occurred on the land cover and land use in the study site between the dates analysed. Then the results achieved by the proposed methodology were evaluated using the MODIS time series available in the LAF (Remote Sensing Laboratory for Agriculture and Forest applications) website (https://www.dsr.inpe.br/laf/series/en/index.html) at INPE and field data. The proposed methodology presents a potential tool for mapping and monitoring land cover and land use changes in the Brazilian Amazonia.Pages: 7786-779