Multitemporal SAR data from the ERS-1/2 Tandem Mission (1995/96) was used toextract land cover information in a mesoscale basin. The actual land coverplays a key role in the parameterization of distributed, physically-basedwatershed models. For land cover classification both SAR intensity andcoherence information were utilised. Principal Components Analyses(PCA) were used to reduce the strong radiometric terrain effect in theintensity data. It was possible to seperate main land cover classes withaccuracies sufficient for model parameterization. Moreover the resultsshow that coherence information is a valuable additional informationcomplementary to the frequently analyzed SAR backscatter intensity.