Improving SAR intensity-based land cover classification results by the use of interferometric coherence information and GIS-analysis

The study focuses on land cover detection using ERS C-band SAR data. Project background is the parameterization of distributed hydrological watershed models. The actual land cover type is an important input parameter for the physical and chemical modelling of the watershed hydrology and plays a significant role in the GIS-based delineation of the distributed modelling entities that are used to preserve the spatial heterogeneity of catchments within hydrological models. The study goal was to clarify the feasibility of SAR based land cover detection in a hilly watershed. Besides the frequently performed multitemporal analysis of backscatter intensity and intensity changes, the phase content of complex SAR data provides an additional potential for environmental applications, which is still poor investigated in regional studies. Hence, both backscatter intensity and interferometric coherence information derived from PRI and SLC datasets of the ERS-1/2 tandem mission (1995/96) was used to address the formulated questions. Investigation area is the River Bral catchment. It is part of the Rhine drainage system and located in the Renish Slate Mountains, Central Uplands, Germany. The area covers approx. 216 km² and has a hilly relief with terrain heights varying from 65 to 400 meters. Numerous hydrological field studies have been carried out in the area resulting in a comprehensive hydro-meteorological database available for reference purposes