Assessment of degraded areas in East Africa using ASAR and MERIS data
Klein, Doris¹; Braun, Matthias²; Menz, Gunter^3
1University of Würzburg, GERMANY; ²University of Alaska, UNITED STATES; ³University of Bonn, GERMANY

In East Africa many savannah ecosystems are degraded due to a high land use pressure. Agricultural land is extending into savannas and intense grazing takes place. By using remote sensing such degraded areas can be monitored in a standardized and regular manner. However often the spatial, temporal and spectral resolution of one data set is not sufficient and a combination of the advantages of different systems seems more adequate. ENVISAT MERIS and ASAR on board of the same satellite seem predestined to be used synergistically and to combine the weather independence and higher spatial resolution of ASAR with the high spectral but moderate spatial resolution of MERIS. In this study MERIS and ASAR data are analysed regarding their capability of a straight forward land use classification in a semi-humid to semiarid area in Kenya with a special focus on degraded areas in terms of a reduced vegetation cover.
For this purpose one MERIS data set from 2003 and multitemporal ASAR-data sets from 2003 and 2004 are merged into one data set using three different data fusion methods: Hue –Saturation-Value, Principle –Component and layerstacking. Each single data set and the merged data sets were classified using maximum likelihood and feed forward neural networks. The best overall accuracy was achieved for the layerstack data set, followed by MERIS. Also the accuracies are not very high the benefit of integrating ASAR is clearly visible in the higher spatial resolution of the layerstack classification compared to the single MERIS classification. Additionally the accuracies of the classes showing degraded areas are increasing.