Context and objectives
The objectives of this study were: (i) to quantify the forest cover evolution, (ii) to investigate the influence of the nature of the deforestation process on an empirical spatial model of deforestation and (iii) to test a remote sensing-based methodology to define the appropriate spatial entities for modelling deforestation.
Expected scientific results
The definition of the spatial entities used to calibrate a spatial model of deforestation has a significant impact on the model design but not, in this case, on the predictive power of the model. The interpretation of the spatial pattern of forest-nonforest interfaces as detected by remote sensing proved to be efficient in identifying regions affected by different deforestation processes.