An Optimized Fuzzy System for Coastal Water Quality Mapping Using Remote Sensing Data

In this paper, we propose coastal water quality mapping process using the combination of in-situ measurements and remote sensing data. Water maps is processed by an hybridization of fuzzy model and genetic algorithm which exploits remotely sensed multispectral reflectances to estimate coastal water quality. The relation between the water parameters and the subsurface reflectances is modeled by a set of fuzzy rules extracted automatically from the data through two steps procedure. First, fuzzy rules are generated by unsupervised fuzzy clustering of the input data. In the second step, genetic algorithm is applied to optimize the rules. Our contribution has focused on the use of several water parameter maps to construct a graphical tool named Pollution Signature Draw (PSD) in order to characterize the water quality. Water characterization is then evaluated by analyzing several types of PSDs related to typical sites selecting the most representatives’ ones. After, the selected PSDs are introduced in a classifier system to generate a pollution map (PM) associated the studied area. The proposed approach was tested on Algiers bay and has highlighted four pollution levels corresponding to High Pollution (HP), Medium Pollution (MP), Low Pollution (LP) and Clear Water (CW).