Evaluating Fuzzy Multi-feature Scenarios for Forest Fire Risk Estimation

Abstract

This paper presents the development and the application of a fuzzy inference intelligent system that performs and evaluates scenarios towards the estimation of a characteristic overall forest fire risk index. The system was built in the integrated environment of the MATLAB platform. It employees fuzzy triangular membership functions to estimate the partial degrees of risk. It also makes use of various fuzzy conjunction operators called T-Norms which are embedded in a fuzzy inference system together with a specially designed Mamdani Ruleset. The main concern was to overcome the problem of combinatorial explosion. The final target was the production of distinct scenarios based on the importance of each involved feature and the determination of the corresponding integrated degrees of risk. Through the execution phase the system assigned even or uneven weights in the involved features giving emphasis either in morphological plus meteorological data or in forest fire history records. Thus the problem was faced under different perspectives. Through its pilot application the system proved its ability to estimate efficiently the partial and the overall risk indices and the results were encouraging.