Automatic classification of land cover change associated with the Brazilian sugarcane expansion over the last decade

MODIS time series have provided high quality data which allows retrieving land use and land cover change (LUCC) history related to sugarcane expansion in Brazil. Although visual interpretation is accurate, automatic classification methods for remotely sensed data should be tested mainly due to the extent of the Brazilian sugarcane area. Thus, this work aims at introducing automatic approaches to classify MODIS time series in order to retrieve LUCC history regarding sugarcane expansion over the last decade in Brazil. Results showed that about 65% and 20% of sugarcane expansion was over pasture and annual crop land, respectively. Although partial results indicated that automatic classification is a promising tool to analyze MODIS time series for LUCC studies, several improvements have to be considered before achieving reliable results.