Treatment of qualitative geographic information in monitoring environmental pollution

The inverse problem of material sources identification, with particular reference to air polluting sources and their detection from experimental data, is a difficult task to solve and it is classified as a typical ill- posed problem. The traditional approach to these problems uses the additional information available on the system in order to limit the number of solutions consistent with the data by means of regularization techniques. In this paper we propose a method that allows to fund reliable solutions by treating qualitative spatial and geographic information. It is shown how to code in a general fuzzy optimization algorithm the experimental knowledge, quantitative models and the a-priori knowledge about territory configuration such as urban areas, lakes, parks and high-traffic motorways. The architecture of an automatic system that allows to handle the coupled qualitative and quantitative knowledge about the system is presented.