Spectral, spatial, and geomorphometric variables for the remote sensing of slope processes

A combined spectral, spatial, and geomorphometric variable set was used to separate geomorphic surfaces at three scales, corresponding roughly to process domain (level I), landform (level II), and sublandform (level III). The best results were obtained using the combined data set, demonstrating two concepts: i) With the use of specialized processing techniques, digital variables can separate geomorphic surfaces at a variety of scales, and ii) considerable advantages can be gained by using multisource remote sensing data. Stepwise selection patterns indicated that geomorphometric variables were the most important contributors to discriminant functions at levels I and II. The reduced effectiveness of spectral and spatial measures reflects the highly variable nature of surface cover at the process domain and landform levels. At level III, variable contributions were more balanced, suggesting that high-quality landcover information from spectral and spatial variables are necessary for the consistent separation of sublandform units.