A raster-based fuzzy expert system for forestry evolution

GIS users are increasingly turning to decision spatial systems to assist them solving complex spatial problems which are not simply maps overlay. So, the authors have implemented the GEODES (GEOgraphic Decision Expert System) system that allows users to specify easily symbolic knowledge of a problem through a graphic user interface by using, on the one hand, data and knowledge spatial characteristics and, on the other hand, fuzzy knowledge. This problem specification and geographic data are used by the system to provide a problem solution map showing favorable and less favorable areas. GEODES can be used in many domains as geographic, ecology, agriculture, natural resources management, manufacturing, etc. It is a fuzzy expert system using raster data and managing principally several types of rules based on knowledge and data intersection and union. The inference engine uses fuzzy inferences with fuzzy sets trapezoidal representations to decrease computational cost. There is a particular type of rule named gradual rule. Their structure is “If (More/Less) X is A Then (More/Less) Y is B”. This rule can specify particular geographic knowledge as “more the altitude is low then more the forest can grow”. The authors have implemented the system GEODES to modelise the evolution of the forest in a region of the south of France by using several maps. They have used spot maps and several data to obtain the forest map, the DEM and a map of fire frequency with reports of the agriculture ministry. By using these maps, the authors have made some experiments with GEODES to modelise the forest evolution