Developing a neuro–fuzzy system to classify drainage sub-basins according to erosion processes on the Island of Lefkas, Greece

In this paper we attempt to classify drainage sub-basins according to their erosion risk. We have adopted a multistep procedure to face this problem. The input variables were introduced into a GIS – platform. These variables were the vulnerability of the surface rocks to erosion, topographic variations, vegetation cover, land use and drainage basin characteristics. We constructed a fuzzy inference mechanism to pre-process the input variables. Next we used neural–network technology to process the input variables. The system was trained to ‘learn’ and classify the input data. The output of this procedure was a classification of the sub-drainage basins related to their risk of erosion. This neuro–fuzzy system was applied to the island of Lefkas (Greece).