Análise espacial aplicada à delimitação de áreas úmidas da planície de inundação do Médio Araguaia

Abstract

The demand for water is of great concern in many regions of the planet. In Brazil, particularly in the Araguaia region, water is an abundant resource, but is subject to poor conservation and water use policies. The degradation of the macrosystem of wetlands in the middle Araguaia has been taking place since the 1970s in an accelerated manner. This study aims to provide subsidy in the formulation of environmental policies for monitoring, recovery and conservation, which ensure the ecological functions provided by these important wetlands. In order to achieve that, the knowledge of the boundaries of the area to be studied is necessary. Thus, the main purpose of this work is the delimitation of the maximum surface likely to flood in the middle Araguaia basin, considering seasons of larger floods. Remote sensing techniques were used in multitemporal images from the LANDSAT 5 TM sensor, for the flood season. Through indexes generated by the Tasseled Cap (TC) transformation, areas of interest were obtained from the landscape elements, such as vegetation, soil, water, and wetlands. From training sample sets, a supervised classification was obtained using the Support Vector Machine (SVM) algorithm. In addition, a HAND (Height Above the Nearest Drainage) model was created using the Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM). The classes associated with humidity, resulting from the classification, were correlated to the altimetric classes obtained from the hydrologically consistent HAND-SRTM. Such procedures allowed the delineation of the maximum boundary of areas likely to flood during flood seasons, which cover 88,119 km² - around 23% - of the Médio Araguaia basin.