Skip to Main Content
Soya is a widely grown crop in many countries worldwide, being United States, Brazil, Argentina and China the major producers. In Brazil, in the last three decades an explosive growth in production has occurred, nearly thirty times, causing a chain of unprecedent changes of land uses and land cover (LULC) in the territory. Therefore, it becomes highly necessary to develop means to monitor such LULC changes, and remote sensing is one of the more potential tools for this monitoring. Thus, the aim of this research was to develop a methodology for contributing in the mapping and monitoring soybeans planted areas, with time-series of remotely sensed Landsat imagery. Two main techniques were combined: Object Based Image Analysis (OBIA) and Data Mining (DM). OBIA was used to represent the knowledge needed to map soybeans areas, and DM was applied to generate the knowledge model. The study area comprises three municipalities in the northwest of São Paulo State, being this area well representative of the agricultural conditions of the Southern and Southeastern regions of Brazil. Classification accuracy was calculated over a set of 500 points not previously used during the training stage. The statistics indicated an overall accuracy of 95% and a kappa coefficient of 0.88.These results indicate that the combination of OBIA + DM + multitemporal imagery is very efficient and promising for the soybeans mapping process.