In order to efficiently plan soil conservation measures it is necessary (i) to understand the impact of soil erosion on soil fertility with regard to local land cover classes and (ii) to identify hot spots of soil erosion in a spatially explicit manner. The aim of this study was to combine field observations on incidence of visible signs of soil erosion and soil organic carbon (SOC) content predicted from a soil spectral library in order to determine the state and process of soil degradation for specific land cover classes. Input data consisted of extensive groundtruth, a digital elevation model and Landsat 7 imagery from two different seasons. Soil spectral reflectance readings were taken from soil samples in the laboratory and calibrated to results of SOC chemical analysis using multiple linear regression techniques. The coefficient of determination for the model is promising (R2 1⁄4 0:78). For an area with rugged terrain and small agricultural plots, decision tree models allowed mapping of soil erosion incidence and land cover classes at an acceptable accuracy level for preliminary studies. The various datasets were linked in the hot spot matrix, developed to combine soil erosion incidence information and mean SOC contents for uniform land cover classes in a scatter plot. The quarters of the plot show different stages of degradation, from well conserved land to hot spots of soil degradation. The approach helps to gain a better understanding on the importance of the impact of soil erosion on soil fertility and to identify hot spots. The results show evidence that on cropland seasonal vegetation characteristics are strong explanatory factors for soil erosion incidence and on grazing land fractional vegetation cover in combination with slope.