Validation spatiale de deux modèles de simulation : l’exemple des plantations industrielles au Chili

Since the end of the 1980s, the commune of San Juan de la Costa (province of Osorno – 10th region of Los Lagos), is experiencing a remarkable expansion of pine and eucalyptus plantations. This afforestation is at the root of many environmental and social disturbances in an area considered as a hotspot for biodiversity. If the Land Use and Land Cover Change (LUCC) processes are widely known, the native forests areas likely to be converted into industrial plantations remain unknown. The following article focuses on the spatial validation of two simulation models in order to suggest prospective interpretations based on a supervised approach (Markov chain coupled with a multi-criteria evaluation) and an automatic approach (multilayer neural network). Both models are built on the knowledge of past dynamics thanks to the diachronic analysis of three satellite images LANDSAT (1986-1999-2008).