The application of simple methods using remote sensing data for the regional validation of a semidistributed hydrological catchment model

Simulation runs of a semidistributed hydrological conceptual catchment model were performed using a spatial data set from a mesoscale catchment located at the moraine landscape of North-East Germany. The simulation quality of the model was estimated by comparing measured daily actual evapotranspiration rates, soil water contents and discharge rates with the corresponding simulated model outputs. Additionally, six Landsat-TM5-subsets covering the catchment were used to calculate the Normalized Difference Vegetation Index (NDVI). These NDVI-distributions were compared with the corresponding simulated regional distributions of actual evapotranspiration (ETr) rates. A visual analysis of the spatial distribution patterns of the NDVI and of the simulated ETr-rates shows some correspondences. However, the spatial variability of the NDVI-patterns was distinctly higher in comparison with the variability of the ETr-rates calculated by the model. We analyzed the coefficients of correlation R between the patterns of the NDVI and the simulated ETr-rates separately for the land cover classes arable land, meadows, coniferous, deciduous and mixed forests. For arable land R ranged within 0.77 and 0.10, for meadows within 0.79 and 0.10, for coniferous forests between 0.73 and 0.10, for deciduous forests between 0.88 and 0.10 as well as for mixed forests between 0.67 and 0.10. The spatial distributions of simulated high and low ETr-rates were mainly correlated with the spatial distributions of forest areas, arable land, water bodies and settlements.