Optimisation of relief classification for different levels of generalisation

Relief plays an important role in the spatial and temporal distribution of soil water and matter transport processes. Each landscape can be segmented into different landform elements based on a digital elevation model. These landforms contain characteristic properties in terms of energy and material balance. Several algorithms are available to classify landscapes at different scales. However, lack of knowledge exists concerning the applicability of relief parameters for landscape stratification for different generalisation levels of underlying data. The objective of this study was to develop a method for agricultural landscapes to classify landform elements across a series of elevation datasets with different spatial resolutions. A non-linear parameter optimisation algorithm was coupled with a relief classification scheme to optimize four classification parameters with regard to environmentally sensitive landforms: shoulder and footslope. Input datasets were based on a LIDAR scan and topographic maps. The magnitude of the optimized relief parameters decreased with decreasing map scale from 1 : 10,000 to 1 : 100,000 or increasing contour line interval. The main conclusion is that if one set of classification rules for a specific landscape was determined for a high-resolution dataset at a small subset, it could be applied for larger areas even if only coarser digital elevation model information were available.