Use of SAR satellites for mapping zonation of vegetation communities in the Amazon floodplain

Radarsat and JERS-1 imagery were used for mapping zonation of vegetation communities in the Amazon floodplain. Imagery analysis indicates that at periods of minimum water level the backscattering values of both C and L bands are the lowest and as the water level rises, so do the backscattering values. JERS-1 imagery exhibits a larger dynamic range of backscattering in response to the ground cover for the two extremes of water level (10 dB) compared to Radarsat imagery. The backscattering differences from different ground cover allowed the use of a region-based classification that produced seasonal maps with accuracies higher than 95% for vegetated areas of the floodplain. These seasonal maps were used to estimate the spatial distribution and time of inundation and the vegetation cover of the floodplain. It was possible to determine that semi-aquatic vegetation, tree-like aquatic plants, and shrub-like trees colonize regions flooded for at least 300 days year(-1) . Secondary colonizers, such as tall well-developed floodplain forest, cover regions flooded for approximately 150 days year(-1) , and floodplain climax forest colonize regions inundated for approximately 60 days year(-1) .