STUDY OF ECOSYSTEM DYNAMICS IN SAHELIAN FLOODPLAINS USING ENVISAT ASAR C-B AND DATA

The Waza-Logone floodplain in the north of Cameroon is both ecologically and economically of exceptional importance. Envisat ASAR Alternating Polarization images are used for studying the flooded grassland vegetation. Backscatter signature of different vegetation types, varying in species composition, height, density, and flooding condition, were analyzed and interpreted. Feature selection based on the Jeffries-Matusita distance was performed to determine the optimal period of image acquisition and the optimal polarization combination for separating the different vegetation classes. A classification map of the area with relevant vegetation classes was produced. Two methods were compared: the pixel-based maximum likelihood method and object-based nearest neighbor method.