Eco-Hydrological Characterization of Inland Wetlands in Africa Using L-Band SAR

Maps describing the eco-hydrology of inland wetland systems in Africa are needed to identify and implement appropriate adaptive management plans related to land use and land cover. Many African countries lack regional baseline information on the temporal extent, distribution and characteristics of wetlands. This information is provided here in the form of maps which characterize two wetland sites of international importance in Malawi and Mozambique. Multi-temporal L-band Synthetic Aperture Radar (SAR) datasets are combined with Landsat Thematic Mapper and ASTER images, digital elevation models, and vegetation species data to provide information on wetland ecology and hydrology. These data were used as input to a hybrid, Decision Tree classifier and a Principal Components Analysis classification approach to produce maps depicting the spatial distribution of vegetation species and characterizing the wetland dynamics. The maps exhibit classification accuracies of 89% and 84% for the two sites respectively. The L-band SAR datasets have proved to be an essential information source in the production of these maps due to (i) frequent cloud cover/smoke which reduces the temporal coverage of optical data, and (ii) a systematic observation strategy and frequent image acquisition which enables characterization of the flood dynamics at a high temporal resolution.