Identificação e classificação de corpos d’água em imagem digital pela aplicação do método Random Forest

Abstract

Watersheds have a high susceptibility index for floods to happen, as they are associated with climatic and environmental issues of nature. Inundation is an attribution to areas that are submerged in water, a consequence of temporal phenomena that act in an uncontrollable way. For this, climate forecasts are necessary to act as a warning about damages and losses in areas that would possibly be affected by floods. There are several methods that perform this type of analysis in flooded areas, one of them is the Random Forest. The Random Forest works as a classification and regression model, in this case the objective of the work is to carry out the identification of the water mass in the digital images of the Butuí Hydrographic Basin, located in the region of São Borja - RS, obtained by the artificial satellite Landsat 8.